Tendências na incidência e prognóstico do Acidente Vascular Cerebral

Rui Manuel Cerqueira Magalhães T ENDÊNCIAS NA I NCIDÊNCIA E P ROGNÓSTICO DO A CIDENTE V ASCULAR C EREBRAL Tese de Candidatura ao grau de Doutor em Ciências Biomédicas submetida ao Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto. Orientador Doutora Maria Carolina da Silva Tavares Costa e Silva Professora Associada Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto. Coorientador Doutor Manuel Jorge Maia Pereira Correia Professor Auxiliar Convidado Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto. Agradecimentos - i - Agradecimentos à Professora Doutora Maria Carolina Costa e Silva pela sua imensa paciência e disponibilidade na orientação científica deste trabalho, bem como à amizade demonstrada ao longo de todo estes anos; ao Professor Doutor Manuel Jorge Maia Pereira Correia, meu co-Orientador e mentor do projecto inicial, pelo seu apoio e incentivo durante a elaboração deste trabalho, bem como à amizade que fomos construindo ao longo da última década; ao Professor Trevor Bailey pelo muito que me ensinou sobre análise espacial e pela sua importante contribuição na elaboração de um dos artigos que integram esta dissertação; ao Professor Peter Sandercock e ao Professor William Whiteley pela contribuição que deram na leitura e revisão de alguns dos trabalhos presentes nesta dissertação; ao Dr. Rui Felgueiras e ao Dr. Pedro Abreu, neurologistas, pela sua colaboração nas diferentes fases do projecto ACINrpc e na elaboração de parte dos artigos que compõem esta dissertação; à Dra . Emília Moreira e à Dra . Carla Branco, psicólogas, pelo enorme contributo que deram ao projecto, realizando as avaliações neuropsicológicas dos doentes; à Engª Cláudia Quintas, bolseira da Fundação para a Ciência e Tecnologia, pela persistência e dedicação com que executou muitas das tarefas associadas à realização do último projecto; aos internos de neurologia e a todos os profissionais de saúde, que durante os últimos catorze anos colaboraram, directa ou indirectamente, nos diferentes projectos; à Fundação para a Ciência e Tecnologia, entidade que financiou os dois últimos projectos (“Long-term prognosis of stroke and transient focal symptoms”, POCTI/SAUESP/59885/2004; “Tendências na incidência e prognóstico dos Acidentes Neurológicos: o segundo estudo de base populacional no norte de Portugal”, PIC/IC/82858/2007); aos meus colegas do ICBAS por todo apoio demonstrado ao longo do tempo que levou a concluir este trabalho; à minha família e amigos, pelas palavras de apoio e incentivo que me foram transmitidas bem como pela compreensão pelos momentos que não pudemos partilhar; o meu profundo agradecimento Resumo - iii - Resumo Os principais objectivos desta dissertação foram: (1) estudar o prognóstico a curto e longo prazo dos doentes com um acidente vascular cerebral (AVC); (2) modelar a incidência diária de AVC em função de diferentes alterações nos parâmetros meteorológicos; (3) descrever os aspectos metodológicos utilizados no projecto ACINrpc e, (4) determinar a variação na incidência de AVC no Norte de Portugal. As bases de dados utilizadas neste trabalho pertencem aos dois projectos ACINrpc (1998-00 e 2009-11). Nestes estudos prospectivos de base populacional realizados no Norte de Portugal, todos os doentes com um primeiro AVC na vida ou com um sintoma neurológico focal transitório foram identificados usando métodos de detecção abrangentes, incluindo a referenciação directa pelos médicos das instituições envolvidas no estudo e a revisão dos registos do serviço de urgência e das altas hospitalares. Os profissionais de saúde foram encorajados a referenciar/notificar qualquer doente com um possível episódio deste tipo. Os doentes foram examinados no início do estudo e aos três meses. Os doentes da primeira coorte (1998-2000) também foram examinados aos 12 meses e aos sete anos. Utilizando a mesma metodologia, foi recolhida informação sobre o perfil sóciodemográfico, factores de risco vascular prévios ao evento, diagnóstico e meios de diagnóstico/tratamentos efectuados. A escala de Rankin modificada (mRS) foi utilizada para avaliar a capacidade funcional. Para avaliar a gravidade do AVC foi utilizada, no primeiro estudo a Unified for Neurological Stroke Scale e, no segundo estudo a National Institute of Health Stroke Scale. Para além destes dados, o Instituto de Meteorologia de Portugal forneceu, para o período compreendido entre Setembro de 1998 e Outubro de 2000, informação diária sobre os parâmetros meteorológicos (temperatura, humidade e pressão atmosférica). Utilizou-se a definição de AVC da Organização Mundial de Saúde. Classificou-se o AVC segundo os tipos patológicos definidos Sudlow e Warlow. Para definir os subtipos de AVC Isquémico utilizou-se a classificação clínica OCSP (Oxfordshire Community Stroke Project) e a classificação etiológica TOAST (Trial of Org 10172 in Acute Stroke Treatment) A distribuição de Poisson foi usada para modelar o número de eventos incidentes. No caso do número de eventos que se seguem a uma “quantidade” variável de exposição, como no caso dos valores dos parâmetros meteorológicos, utilizando diferentes períodos de risco, for usada uma regressão de Poisson. No caso de uma “exposição” constante (não relevante para o modelo) e das variáveis explanatórias serem categóricas, como na Tendências na Incidência e Prognóstico do AVC - iv - modelação da incidência por grupo etário, sexo e período do estudo, os dados podem ser sumariados numa tabela de contingência com restrições nas frequências em cada cela, usando-se um modelo linear generalizado com uma função de ligação, um modelo log-linear. Esta distribuição foi também assumida para calcular os intervalos de confiança a 95% para as taxas de incidência brutas e padronizadas para a população portuguesa e europeia. Foi ainda usado um modelo binomial para contrastar o efeito dos parâmetros meteorológicos em subgrupos de doentes. Em relação ao prognóstico, o tempo de sobrevivência foi estimado utilizando o método de Kaplan-Meier, enquanto que o modelo de riscos proporcionais de Cox foi utilizado para identificar os preditores independentes de sobreviver livre de AVC ou de eventos vasculares. Com base no grau de incapacidade aos três meses medido numa escala ordinal variando entre zero e seis foi estimada a possibilidade de um pior prognóstico aos sete anos recorrendo a uma análise Ridit. A diferença na incidência de AVC entre o meio urbano e rural resulta do maior risco de AVC isquémico na população jovem do meio urbano e na população idosa do meio rural. Embora no meio rural os doentes sejam mais idosos, a menor prevalência de factores de risco vascular associada a uma igual gestão do doente com AVC, pode justificar o facto de não existirem diferenças no prognóstico a longo prazo no meio urbano e rural. Por outro lado, uma simplificação na avaliação do grau de incapacidade aos três meses permite avaliar o perfil de risco dos doentes com AVC Isquémico. Este estudo permitiu confirmar que uma ligeira diferença no estado funcional aos 3 meses está associada a um diferença significativa na sobrevivência e estado funcional aos 7 anos, o que tem implicações no planeamento e avaliação económica dos tratamentos para o AVC agudo. Esta investigação permitiu encontrar uma associação entre os parâmetros meteorológicos e a ocorrência de AVC e também com a sua gravidade. A variação do efeito da temperatura ambiental de acordo com o “timing” da exposição, a ausência de associação ao enfarte lacunar e a associação ao enfarte cardioembólico, reflectem a plausibilidade desta associação e podem explicar divergências nos resultados encontrados noutros estudos de base populacional ou hospitalar. Este trabalho destaca também a importância da avaliar os doentes com sintomas focais transitórios para identificar os que têm AVC, sendo um critério metodológico a considerar em estudos futuros. Para o declínio da taxa de mortalidade por AVC em Portugal, contribuiu uma diminuição na incidência de AVC entre 1998 e 2011. Observou-se um efeito de ‘género’ no sentido em que a diminuição do risco de AVC, em particular do AVC hemorrágico ou incapacitante, foi muito superior nas mulheres. Abstract - v - Abstract The main objectives of the present work were: (1) to study shortand long-term prognosis of stroke patients; (2) to model the daily stroke incidence according to shortor long-term weather changes; (3) to describe the methodological aspects of stroke incidence studies used in the ACINrpc project and, (4) to determine changes in stroke incidence in Northern Portugal. The data used is part of the two ACINrpc projects (1998-2000 and 2009-2011). In these two prospective community-based studies implemented in Northern Portugal, all patients with a first-ever-in-lifetime stroke or focal neurologic transient symptoms were ascertained using comprehensive methods, including referrals from physicians working in the study area and data retrieved from emergency/discharge records. Physicians were encouraged to report/notify any patient who might have experienced such a kind of event. Patients were examined at baseline and followed-up at three months. For the first cohort (19982000) patients were also followed at one and seven years. Information about socio-demographic characteristics, prior-to-stroke vascular risk factors, diagnostic and clinical evaluation/treatment and destination after discharge was collected using the same methodology throughout the two study periods. Functional status was assessed with the modified Rankin Scale (mRS) and stroke severity was measured using the Unified for Neurological Stroke Scale (first study) and the National Institute of Health Stroke Scale (second study). For the first study period, an additional dataset with information on daily temperature, humidity and air pressure, between September 1998 and October 2000, was obtained from the National Meteorological Office. Stroke was defined according to the World Health Organization and classified into pathological types according to Sudlow and Warlow standard definitions. The Oxfordshire Community Stroke Project classification and the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria were used to define ischemic stroke subtypes. The Poisson distribution was used for modeling the number of incident events. When the events related to varying amounts of “exposure”, as for values of meteorological parameters using different hazard periods, the Poisson regression was used. When “exposure” is constant (not relevant in the model) and explanatory variables are categorical, as in the model of incidence according to age-group, sex and study period, the data may be summarized in a contingency table with restrictions on cell frequencies. In this case a generalized linear model with a link function was used, a log-linear model. This distribution was also assumed for calculating the 95% confidence intervals for crude Tendências na Incidência e Prognóstico do AVC - vi - and standardized incidence rates for the Portuguese and European population. The binomial model was used to contrast effects of meteorological parameters across subgroups of patients. As far as prognosis is concerned, the Kaplan-Meier estimates for overall survival were calculated and predictors of survival free from stroke or vascular events were determined using Cox proportional hazards models. Ridit analysis was used to estimate the odds of a more serious 7-year outcome according to adjacent values of the 3-month modified Rankin Score (mRS). The age pattern of ischemic stroke incidence marks the difference between rural and urban populations; the youngest urban and the oldest rural residents were at a higher risk. Although patients from rural areas were older, the relatively lower prevalence of simultaneously occurring risk and prognostic factors among them as well as the similar management of rural and urban patients may justify why rurality is not associated with long-term survival. It was also described that a three grade simplified mRS summarises the risk profile and stroke characteristics in 3-month survivors with ischaemic stroke. Moreover we found that modest differences in functional status at 3 months are associated with significant differences in survival and functional status over 7 years, results that have relevant implications for health care planning and economic assessment of treatments for acute stroke. We found that outdoor temperature and related meteorological parameters are associated with stroke occurrence and stroke severity. The different hazard periods for temperature effects, the absence of association with lacunar infarcts and the association with cardioembolic infarcts may explain the heterogeneous effects of weather on stroke occurrence found in community-based and hospital admission studies. This work also highlights the importance of screening all transient focal episodes for identifying patients with stroke and this may be a methodological criterion to be included in future stroke incidence studies. The decline in stroke incidence between 1998 and 2011 contributed for the decline in stroke mortality rates in Portugal. We may add that there was a “gender decline” in the sense that an overall stroke incidence, hemorrhagic stroke incidence and disabling stroke incidence was evidenced in women and not in men. Advances in the quality of inpatient care and primary/secondary prevention in the elderly contributed decisively for the better stroke outcome across the last decade. Publicações e apresentações públicas - vii - Publicações e apresentações públicas As publicações e apresentações públicas que foram editadas ou realizadas no contexto e no decorrer desta dissertação são aqui enumeradas por ordem cronológica. Artigos Moreira E, Correia M, Magalhães R, Silva MC. Stroke awareness in urban and rural populations: Knowledge and action are independent. Neuroepidemiology, 2011; 36:265-273. [T] Magalhães R, Silva MC, Correia M, Bailey T. Are stroke occurrence and outcome related to weather parameters? Results from a community-based study in northern Portugal. Cerebrovascular Diseases, 2011; 32:542-551. [T] Correia M, Magalhães R, Silva MR, Matos I, Silva MC. Stroke types in rural and urban Northern Portugal: incidence and 7-year survival in a community-based study. Cerebrovascular Diseases Extra, 2013; 3:137-149. [T] Moutinho M, Magalhães R, Correia M, Silva MC. [A community-based study of stroke code users in northern Portugal]. Acta Médica Portuguesa, 2013; 26:113-122. [T] Felgueiras R, Magalhães R, Correia M, Silva MC. Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study. International Journal of Stroke and Cerebrovascular Diseases, 2014; 23:2190-2198. [T] Magalhães R, Abreu P, Correia M, Whiteley W, Silva MC, Sandercock P. Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort. Cerebrovascular Diseases, 2014; 38:46-54. [T] Magalhães R, Felgueiras R, Abreu P, Correia M, Silva MC. Decline of stroke incidence and poststroke disability in Porto, Portugal between 1998 and 2011. (to be submitted). Resumos Publicados Tuna A, Correia M, Magalhães R, Silva MC. Long term prognosis of Transient Neurological Attacks in a community-based study. Cerebrovascular Diseases, 27(suppl 6): 72. 2009. [18 th European Stroke Conference. Stockholm, Sweden 2009] Correia M, Tuna A, Magalhães R, Silva MC. Transient Neurological Attacks: incidence and vascular risk factors in Northern Portugal. Cerebrovascular Diseases, 27(suppl 6): 100. 2009. [18 th European Stroke Conference. Stockholm, Sweden 2009] Magalhães R, Marques AI, Correia M, Silva MC. Distribuição espacial da incidência de Acidente Vascular Cerebral na cidade do Porto. Sinapse, 9(1 suppl 1): 45. 2009. [Neuro 2009: Congresso das Sociedades Portuguesas de Neurologia e Neurocirurgia, Albufeira, Maio 2009] Tendências na Incidência e Prognóstico do AVC - xiv - Índice de Figuras - xv - Índice de Figuras [Introdução] Figura 1. Evolução das taxas de mortalidade por doença cerebrovascular .................................. 4 [Capítulo 2. Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study] Figure 1. Ascertainment and inclusion of FELS in northern Portugal and details of the 7year follow-up. ...............................................................................................................15 Figure 2. Kaplan-Meier estimates of the cumulative death risk for all patients (a) and the cumulative death risk in 28-day stroke survivors by stroke type (b) N=Cumulative number of patients. .......................................................................................................20 Figure 3. Joint distribution of standardized IS and intracerebral haemorrhage incidence in community-based studies. The lines represent the median values. .............................22 [Capítulo 3. Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort] Figure 1. Flowchart of inclusion of patients and details of 7-year follow-up ................................33 Figure 2. Distribution of mRS at 7 years according to mRS at 3 months; left side shows the mRS for survivors at 7-years by increasing order of mRS score and the right side shows proportion of deceased. Number of patients in squares ...................................35 Figure 3. Long-term survival from assessment at three-month after an ischemic stroke according to (A) functional status at three months (mRS: modified Rankin Scale), (B) simplified mRS (0-1, 2-3, 4-5) .................................................................................36 [Capítulo 4. Are Stroke Occurrence and Outcome Related to Weather Parameters? Results from a Population-Based Study in Northern Portugal] Figure 1. LOWESS smoothed values of daily incidence of events (per million per day) and meteorological parameters (Tmax, DTR, relative humidity, atmospheric pressure and precipitation) at Porto during the study period. ......................................................49 Figure 2. Fitted number of PICH (black) and IS (grey) (scaled to be a percentage of mean daily strokes) plotted against DTR (a) and Tmin (b); fitted number of fatal (black) and non-fatal (grey) strokes plotted against Tmax (c). (Poisson models for the specific events). Dashed lines represent 95% CI. ........................................................51 [Capítulo 5. Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study] Figure 1. Flowchart of inclusion of patients and details of 7-year follow-up. ...............................62 Tendências na Incidência e Prognóstico do AVC - xvi - Figure 2. Kaplan–Meier survival free from stroke and free from a vascular event according to the number of vascular risk factors (A) and (C) and computed tomography scan results (B) and (D). .............................................................................................. 66 [Capítulo 6. Decline of stroke incidence and poststroke disability in Porto, Portugal between 1998 and 2011] Figure 1. Map of the metropolitan area of Porto, showing areas included in both studies (dark shaded) and added in the second study (light shaded), and main hospitals involved ........................................................................................................................ 77 Figure 2. Age specific annual incidence rates of first-ever stroke in (A) men, (B) women, (C) minor stroke and (D) non-minor stroke in the two study periods. Errors bars are 95% confidence intervals ....................................................................................... 84 Figure 3. Distribution of modified Rankin Scale scores at 28-days, stratified by study period in (A) all patients, and according to (B) age and (C) inpatient care ............................. 87 [Anexo I. Avaliação da Via Verde do Acidente Vascular Cerebral no Norte de Portugal: Caracterização e Prognóstico dos Utilizadores] Figura 1. Metodologia de detecção e inclusão dos casos. ....................................................... 104 Figura 2. Fontes de informação com registos de casos suspeitos. .......................................... 106 Figura 3. Utilização da Via Verde do AVC segundo os critérios de activação (%). .................. 106 [Anexo II. Comunicações] Figure A1. Lowess smoothed values of daily incidence and meteorological parameters at Porto during the study period. .................................................................................... 117 Figure A2. Overall stroke incidence by age-group ...................................................................... 120 Figure A3. Stroke incidence by age-group and gender ............................................................... 120 Figure A4. Evolution of PICH incidence by age-group ................................................................ 123 Figure A5. Evolution of PICH incidence by age-group, according to residence area ................. 123 Figure A6. Evolution of SAH incidence by age-group ................................................................. 126 Índice de Tabelas - xvii - Índice de Tabelas Tabela 1. Taxa de incidência anual e indicadores de prognóstico no estudo decorrido entre 1998 e 2000, em populações urbanas e rurais .............................................................. 2 [Capítulo 2. Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study] Table 1. Ascertainment and patient characteristics (in %) by stroke type in urban and rural áreas .............................................................................................................................17 Table 2. Age-specific annual incidence per 1,000 for stroke types in urban and rural northern Portugal (1998 – 2000) ...................................................................................17 Table 3. HR for the association between factors at presentation and death among 28-da8 survivors ........................................................................................................................21 [Capítulo 3. Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort] Table 1. Distribution of patients’ characteristics at baseline and according to status at three months .................................................................................................................34 Table 2. Distribution of status at 7 years for 3-month survivors and number needed to change to avoid one death at 7 years (NNC) ...............................................................35 Table 3. Cox’s regression models estimates of the hazard ratios for death at seven years according to baseline characteristics for patients alive at three months after a first-ever-in-a-lifetime ischemic stroke ..........................................................................37 [Capítulo 4. Are Stroke Occurrence and Outcome Related to Weather Parameters? Results from a Population-Based Study in Northern Portugal] Table 1. Distribution of patient characteristics and vascular risk factors by types and subtypes of IS ...............................................................................................................47 Table 2. Description of weather parameters, incident events and case fatality by season ........48 Table 3. Association between incident stroke events and meteorological parameters according to exposure period .......................................................................................50 Table 4. Association between incident stroke events and meteorological parameters according to exposure period (multiple-regression models) .........................................51 [Capítulo 5. Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study] Table 1. Ascertainment of patients, characteristics, and diagnostic procedures ........................63 Table 2. Cox proportional hazard rates (HR) for stroke and vascular events in the sevenyear follow-up in patients with transient VS (n=331) ....................................................65 Tendências na Incidência e Prognóstico do AVC - xviii - Table 3. Cox proportional hazard rates (HR) for stroke and vascular events in the sevenyear follow-up for patients with peripheral and unclassified VS (n=300) ..................... 67 [Capítulo 6. Decline of stroke incidence and poststroke disability in Porto, Portugal between 1998 and 2011] Table 1. Annual incidence rates of first-ever stroke per 100,000 in Porto, Portugal over eleven years, by gender and age................................................................................. 82 Table 2. Annual incidence rates of first-ever stroke per 100,000 in Porto, in the two study periods stratified by pathological type and disability in men and women .................... 83 Table 3. Incidence rate ratios (2nd vs. 1st period) by gender and incidence rate ratios (men vs. women) by age in the two study periods for all stroke and non minor stroke; period incidence rate ratio (2nd vs. 1st) and incidence rate ratios (men vs. women) by age for minor stroke .................................................................................. 85 Table 4. Patient’s characteristics, assessment and management in the two study periods ...... 86 Table 5. Coefficients of the linear regression models of the modified Rankin Scale scores on patients and stroke characteristics and management ............................................ 88 Supplemental table. Deviances and goodness of fit test for Poisson Log-linear Models fitted to the expected number of incident strokes according to the Census 2011 population ............................................................................ 95 [Anexo I. Avaliação da Via Verde do Acidente Vascular Cerebral no Norte de Portugal: Caracterização e Prognóstico dos Utilizadores] Tabela 1. Características dos 72 utilizadores da Via Verde externa ......................................... 107 Tabela 2. Caracterização dos doentes com AVC, utilizadores e não utilizadores da Via Verde externa............................................................................................................. 108 Tabela 3. Caracterização e prognóstico dos doentes com critérios de activação da Via Verde do AVC ............................................................................................................ 109 [Anexo II. Comunicações] Table A1. Distribution of vascular risk factors in 78 patients with PICH .................................. 116 Table A2. Estimated percentage of daily variation in Intracerebral haemorrhage by unit variation in the preceding 24 hours of weather parameters .................................... 116 Table A3. Assessment of interaction between DTR and precipitation on overall PICH incidence and in the presence of vascular risk factors............................................ 116 Table A4. Characteristics of patients included......................................................................... 119 Table A5. Evolution of overall stroke incidence (/1,000) ......................................................... 119 Table A6. 28-day case-fatality in rural and urban patients ...................................................... 119 Table A7. Patient's characteristics and case-fatality ............................................................... 122 Table A8. Evolution of PICH incidence (/100,000) .................................................................. 122 Table A9. Patient's characteristics and case-fatality ............................................................... 125 Table A10. Evolution of SAH incidence (/100,000) ..................................................................... 12 - 1 - I IN NT TR RO OD DU UÇ ÇÃ ÃO O Embora as taxas de mortalidade por acidente vascular cerebral (AVC) tenham vindo a diminuir nas últimas décadas, o AVC continua a figurar como uma das principais causas de morte a nível mundial. 1 O AVC é também responsável por um elevado número de hospitalizações, e as suas sequelas pós-AVC têm como consequência uma diminuição da qualidade de vida dos sobreviventes, incapacidade para o trabalho e/ou actividades diárias, e um alto consumo de recursos de saúde. 2 Em Portugal, até finais do século XX, a informação disponível sobre o AVC baseava-se nas taxas de mortalidade publicadas pelo Instituto Nacional de Estatística. Com base nessa informação, Portugal apresentava, no período compreendido entre 1985 e 1994, uma das mais altas taxas de mortalidade por AVC da Europa Ocidental. 3 Para além desta elevada mortalidade, num estudo realizado em 1992, a prevalência estimada do AVC era de 8% com cerca de 20% dos sobreviventes a apresentarem uma incapacidade grave. 4 No que diz respeito à incidência de AVC e factores de risco associados, os primeiros estudos entretanto realizados apresentavam algumas limitações metodológicas, nomeadamente em termos de representatividade do AVC na comunidade e critérios de diagnóstico diferentes. 5-6 No entanto, os estudos prospectivos na comunidade são os únicos que permitem determinar de forma real a taxa de Tendências na Incidência e Prognóstico do AVC - 2 - incidência de AVC e co-morbilidades associadas. Permitem ainda conhecer melhor a sua etiologia e deste modo desenvolver estratégias mais eficazes para a sua prevenção e tratamento. 7 Tendo como objectivo investigar a elevada taxa de mortalidade por AVC no Norte de Portugal, foi realizado entre Outubro de 1998 e Setembro de 2000, um estudo de incidência de acidentes neurológicos (ACINrpc: Acidentes Neurológicos - registo prospectivo na comunidade), no qual tive a oportunidade de estar envolvido desde a fase inicial. 8 A metodologia adoptada obedeceu aos critérios internacionais estipulados para a realização de um estudo de incidência "ideal”. 9 Este estudo permitiu obter dados fiáveis sobre a taxa de incidência do primeiro acidente neurológico na vida, incluindo o AVC, 10 o Acidente Isquémico Transitório (AIT) 11 e o Acidente Neurológico Transitório não AIT (ANT). 12 Permitiu ainda obter informação sobre a prevalência dos factores de risco vascular (FRV) mais relevantes e determinar o prognóstico a curto prazo, em populações urbanas e rurais (Tabela 1). A partir dos resultados obtidos, na Região Norte de Portugal, foi possível estimar que por ano 28.000 pessoas sofriam um primeiro AVC na vida, sendo este número ainda superior nas regiões rurais. Tabela 1. Taxa de incidência anual e indicadores de prognóstico no estudo decorrido entre 1998 e 2000, em populações urbanas e rurais Urbana Rural Global Evento Indicador taxa IC 95% taxa IC 95% taxa IC 95% AVC Incidência (/100 000) 269 244-293 305 265-344 279 259-300 Padronizada* 173 153-192 202 169-234 181 164-198 28 dias Letalidade (%) 14.6 10.2-19.3 16.9 13.7-20.6 16.1 13.6-19.1 12 Meses Recorrência (%) 7,6 5,5-10,4 11,5 8,0-16,3 8,9 7,0-11,2 Mortalidade (%) 30,1 26,1-34,4 28,8 23,3-35,0 29,7 26,4-33,2 Dependência** (%) 42,7 37,4-48,2 36,0 29,0-43,7 40,4 36,1-44,8 AIT Incidência (/100 000 61 49-73 96 67-133 67 56-78 Padronizada 40 23-69 67 45-104 44 26-73 12 Meses Recorrência (%) 21,9 15,1-30,7 27,8 15,9-44,0 23,4 17,2-31,0 Mortalidade (%) 9,5 5,3-16,6 19,4 9,8-35,0 12,1 7,7-18,5 *População Europeia, **Valor na escala de Rankin modificada >2 Em comparação com outras regiões da Europa Ocidental, 13 a incidência de AVC em Portugal era tendencialmente mais elevada quer em zonas rurais (305/100.000) quer em zonas urbanas (269/100.000). Estes valores de incidência apenas eram superados por países do Leste da Europa. 13-14 No entanto a alta incidência nos mais velhos verificavase em estudos realizados em ambientes rurais ou mistos (rural/urbano) contrastando Introdução - 3 - com estudos realizados em ambiente urbano e, por outro lado, a incidência nos mais novos era superior na cidade do Porto quando comparada com a de outros estudos em meio citadino. 15 Estes resultados foram o ponto de partida para tentar compreender e melhorar a prevenção/tratamento precoce do AVC. Surgiram assim várias questões de investigação que constituíram o tema central desta tese. As primeiras questões que importava responder relacionam-se com a prevalência dos FRV e a incidência e prognóstico do AVC: Será que a população rural em comparação com a urbana tinha uma maior prevalência dos tradicionais factores de risco vascular? Será que por esse facto a incidência dos diferentes tipos de AVC era diferente nas duas populações? Será o seu prognóstico a longo prazo será diferente? Os primeiros resultados do estudo ACINrpc mostraram também que a elevada incidência era acompanhada por uma baixa taxa de letalidade aos 28 dias (14,6% no meio rural e 16,9% no meio urbano), contrariamente ao que se poderia supor a partir das estatísticas oficiais de mortalidade que apontavam para uma taxa de mortalidade padronizada de 154/100.000 habitantes. 16 De notar que os valores encontrados para a taxa de letalidade no estudo eram similares aos de outros estudos de base populacional. 15,17-18 Este facto poderá ser explicado quer pelas altas taxas de incidência nos mais jovens quer pela diferente distribuição de subtipos de enfarte cerebral. 8,10 Surgiu, pois, daqui uma outra questão de investigação abordada no âmbito desta dissertação: Será que a elevada incidência de AVC era acompanhada por uma menor gravidade (incapacidade), especialmente no caso do enfarte cerebral? Tendo como ponto de partida a sazonalidade na ocorrência de AVC encontrada em diferentes estudos, 19-22 encontrou-se, no norte de Portugal, uma diferença ambiental marcada entre o meio urbano (Porto) e o meio rural (Vila Pouca de Aguiar) relacionada com os parâmetros meteorológicos, particularmente com as temperaturas mínimas e máximas em Vila Pouca de Aguiar. 23 No Inverno, registou-se um pico na ocorrência de AVC em Vila Pouca de Aguiar, enquanto no Verão se verificou um menor número de AVCs no Porto. Vários estudos tinham já apontado para a associação entre a temperatura e mortalidade por AVC 22,24-25 mas poucos tinham estudado a associação entre os parâmetros meteorológicos, em particular a temperatura, como despoletadores do AVC. Neste contexto surgiu uma outra questão de investigação: Tendências na Incidência e Prognóstico do AVC - 4 - Será que a exposição transitória com efeito transitório a valores extremos dos parâmetros meteorológicos, nomeadamente a temperatura ambiental, responsável pelo desencadear do AVC, isto é, por variações na incidência, em particular na população idosa? Embora reconhecendo que as estatísticas oficiais de mortalidade por AVC não são isentas de viés, é possível observar um declínio destas taxas entre os anos de 1999 e 2012, tanto em Portugal Continental (de 154,2 para 62,8/100.000), como na Região Norte (de 164,2 para 66,7/100.000) (Figura 1). 16,26-33 Variações na taxa de mortalidade podem resultar de variações na incidência e/ou prognóstico, quer actuando ao nível da prevenção dos FRV quer por ganhos significativos no tratamento precoce da patologia. 75 100 125 150 175 1999 2000 2001 2002 2003 2004 2005 2006 2012 Continente Região Norte* Porto* 75 100 125 150 175 1999 2000 2001 2002 2003 2004 2005 2006 2012 75 100 125 150 175 1999 2000 2001 2002 2003 2004 2005 2006 2012 Continente Região Norte* Porto* Continente Região Norte* Porto* Taxa de mortalidade (/100 000) anos *1999-2005: Região Norte (5 distritos) e distrito do Porto; 2006 e 2012: Região Norte (NUTs III) e Grande Porto Figura 1. Evolução das taxas de mortalidade por doença cerebrovascular Desde a realização do primeiro estudo (1998-2000) verificaram-se avanços consideráveis ao nível da intervenção no AVC com o objectivo de alterar o seu "peso" na comunidade, destacando-se melhorias na prevenção dos principais FRV 36 e a implementação de intervenções terapêuticas, nomeadamente a organização de Unidades de AVC e a utilização do tratamento trombolítico na fase aguda. 34-35 Em resultado dessas estratégias o cidadão comum está mais informado quanto à conveniência de exercer uma vigilância regular dos FRV, em particular da tensão Introdução - 5 - arterial, 37 mas é ainda insuficiente a divulgação e o alerta na população para a emergência do AVC. Após a organização das Unidades de AVC, foi criado um programa - "Via Verde do AVC" - cujo objectivo é minimizar o tempo decorrido entre o aparecimento dos sintomas e o início do tratamento do AVC, esperando-se que tenha reflexos nos indicadores de mortalidade e morbilidade. Este programa requer tanto a organização da emergência pré-hospitalar e hospitalar como o alerta da população, de modo a que a procura de ajuda médica seja a imediata ao início dos sintomas. 38-40 Para monitorizar a eficácia das estratégias de prevenção primária/secundária, é crucial ter informação sobre a evolução do padrão da doença e suas causas. Os dados epidemiológicos sobre tendências temporais na incidência, etiologia e prognóstico dos acidentes neurológicos, fornecem indicadores sobre a eficácia da actuação dos sistemas de saúde na prevenção dos factores de risco modificáveis, partilhados com outras doenças relacionadas com o envelhecimento, e no tratamento dos doentes em ambiente de consulta ou internamento. Assim, uma década após a realização do estudo ACINrpc foi possível desenhar e concretizar, na região norte de Portugal, o projecto ACIN2: “Tendência da incidência e prognóstico dos acidentes neurológicos: o segundo estudo de base populacional no norte de Portugal” (PIC/IC/82858/2007). A implementação do estudo requereu um planeamento cuidadoso e eficaz, adequado à obtenção de dados actuais de forma comparável com o estudo anterior. No decurso deste estudo levantaram-se várias questões, nomeadamente no que diz respeito à comparação dos resultados dos dois estudos: Como planear o estudo de modo semelhante ao primeiro, de forma a obter resultados comparáveis, contemplando: (a) a reorganização do Serviço Nacional de Saúde entretanto ocorrida; (b) a utilização de meios informáticos no tratamento das diferentes fontes de informação, que apresentam diferentes níveis de informatização e, (c) a inclusão de critérios metodológicos adicionais entretanto publicados? Qual a variação na incidência e prognóstico a curto prazo do AVC que ocorreu no espaço de uma década no Norte de Portugal? Estrutura da Tese Para responder às primeiras questões apresenta-se no Capítulo 2 um artigo onde se estudou a incidência e prognóstico a curto e longo prazo (sete anos) do primeiro AVC na vida. Para o efeito, recorreu-se à modelação do tempo de sobrevivência e co-variáveis descritivas dos eventos ocorridos durante o follow-up, nomeadamente utilizando um © 2013 S. Karger AG, Basel 1664–5456/13/0031–0137$38.00/0 Original Paper Cerebrovasc Dis Extra 2013;3:137–149 Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study Manuel Correia a, b Rui Magalhães b Mário Rui Silva c Ilda Matos d Maria Carolina Silva b a Serviço de Neurologia, Hospital de Santo António – Centro Hospitalar do Porto, and b UNIFAI, Instituto de Ciências Biomédicas de Abel Salazar (ICBAS), Universidade do Porto, Porto , c Serviço de Neurologia, Hospital de S. Pedro – Centro Hospitalar de Trás-os-Montes e Alto Douro, Vila Real , and d Serviço de Neurologia, Hospital de Mirandela – Centro Hospitalar do Nordeste, Mirandela , Portugal Key Words Stroke · Incidence · Survival · Prognosis · Rural/urban Portugal Abstract Background/Aim: Differences in stroke incidence and mortality between regions could stem from differences in the incidence of particular stroke types and long-term prognosis. The aim of this study was to investigate whether different risk profiles and stroke types underlie the difference in stroke incidence and patient long-term survival in rural and urban populations. Methods: All suspected first-ever-in-a-lifetime strokes occurring between October 1998 and September 2000 in 37,290 residents of rural municipalities and in 86,023 individuals living in the city of Porto were entered into a population-based registry. Standard definitions of stroke types and overlapping comprehensive sources of information were used for patient identification. Patients were examined by neurologists at 3 months, 1 year and 7 years after the index event. Results: From a total of 688 patients included (226 in rural and 462 in urban areas), 76.2% had an ischaemic stroke (IS; 75.3 vs. 77.9%), 16.1% a primary intracerebral haemorrhage (PICH; 16.3 vs. 14.6%) and 3.3% a subarachnoid haemorrhage (SAH; 2.7 vs. 3.7%); in 4.4% (4.9 vs. 4.1%), the stroke type could not be determined. The annual incidence rate per 1,000 was 2.13 (95% CI, 1.95–2.31), 0.45 (95% CI, 0.37–0.53), 0.09 (95% CI, 0.06–0.14) and 0.12 (95% CI, 0.08–0.17), respectively. The age-specific rural/urban incidence rate ratios for IS in the youngest group (<55 years) was 0.27 (95% CI, 0.11–0.69), increasing to 1.47 (95% CI, 1.07–2.01) for those aged 65–74 years and to 1.87 (95% CI, 1.39–2.52) for those between 75 and 84 years. Rural compared to urban patients with an IS were predominantly men, had a prevalence ratio (PR) of 1.28 (95% CI, 1.05–1.56), were 65 years or older (PR = 1.18; 95% CI, 1.08–1.30) and had Published online: October 18, 2013 EXTRA Rui Magalhães, MSc UNIFAI, Instituto de Ciências Biomédicas de Abel Salazar Universidade do Porto, Rua de Jorge Viterbo Ferreira, 228 PT–4050-313 Porto (Portugal) E-Mail rmag @ icbas.up.pt www.karger.com/cee DOI: 10.1159/000354851 This is an Open Access article licensed under the terms of the Creative Commons AttributionNonCommercial 3.0 Unported license (CC BY-NC) (www.karger.com/OA-license), applicable to the online version of the article only. Distribution permitted for non-commercial purposes only. Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 138 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel in general a lower prevalence of risk factors. There was no evidence of rural/urban differences in 28-day case fatality for the stroke types, although IS tended to be less fatal among urban patients (10.3 vs. 13.1%), whereas PICH (33.3 vs. 24.2%) and SAH (35.3 vs. 16.7%) were less fatal among rural patients. Independently of rural/urban residence, predictors of poor survival after the acute phase (28 days) were age >65 years (HR = 3.57; 95% CI, 2.6–4.9), diabetes (HR = 1.5; 95% CI, 1.2–1.9), ischaemic heart disease (HR = 1.8; 95% CI, 1.3–2.6), atrial fibrillation (HR = 1.5; 95% CI, 1.1–2.0) and smoking habits (HR = 1.6; 95% CI, 1.1–2.3). Conclusions: The age pattern of IS incidence marks the difference between rural and urban populations; the youngest urban and the oldest rural residents were at a higher risk. Although patients from rural areas were older, the relatively lower prevalence of simultaneously occurring risk and prognostic factors among them as well as the similar management of rural and urban patients may justify why rurality is not associated with long-term survival. © 2013 S. Karger AG, Basel Introduction Despite the continuous decrease in mortality rates from stroke in Portugal in the last two decades [1] , disparities still remain in standardized rates among rural (77.1/100,000) and urban (71.8/100,000) areas in northern Portugal [2] . We have shown that this excess mortality in rural areas could be partially explained by a higher incidence of stroke and not by short-term case fatality, i.e., 30.1% in rural areas compared to 27.9% in the city of Porto in the first year following the first-ever-in-a-lifetime stroke (FELS) [3] . Rather than being a single pathological entity, stroke is a disease that includes distinct types having different incidence rates, risk profiles, management guidelines and outcomes that may lead to different disease burdens in different regions. Comparing the incidence of different stroke types as well as the risk profiles and long-term survival of patients with these stroke types in rural and urban populations may add important knowledge about their aetiology, prevention and prognosis. In order to accurately assess the incidence of different stroke types, studies investigating stroke incidences must meet ideal criteria [4, 5] , such as the use of diagnostic brain imaging for the majority (ideally for all) of the patients [6] . In accordance to these criteria, a community-based prospective stroke registry was set up in northern Portugal [3] . The aim of this article is to present data on stroke types regarding incidence, risk profile and long-term survival for understanding the patterns of stroke in rural and urban populations. Population and Methods The ACINrpc was a community-based study of the incidence and outcome of FELS and transient neurological focal symptoms and signs. The overall design of the project has been described in detail elsewhere [3] . In brief, the study population comprised 123,112 individuals registered and identified by a unique health service at five health centres on September 30, 1999 (mid-study period): 37,089 in rural areas and 80,023 in the city of Porto. This population was not significantly different from the corresponding geographic population [3] . Case ascertainment lasted from October 1, 1998, to September 30, 2000, and included both ‘hotand cold-pursuit data collection’ using a variety of overlapping sources of information. A study neurologist examined all suspected cases as soon as possible, and a CT was performed after the event. Medical records from hospitals and/or general practitioners (GP) were checked for details of any previous event and vascular risk factors (VRF). The principal investigator reviewed the information of each patient and classified the type of stroke; Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 139 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel whenever appropriate, the classification was established after consensus between the neurologist who first examined the patient and the principal investigator. All patients were followed up by neurologists at 3 months, 1 year and 7 years after the index event. The long-term follow-up began in September 2005, and every attempt was made to include all patients alive at the 1-year follow-up ( fig. 1 ). The process began by updating the telephone contacts of the patients using health centre/hospital administrative files and all available information on patient medical records. This was followed by a first contact by phone made 15 days before the end of the 7-year period, and, when this failed, two letters were sent explaining the study purpose and suggesting a date for a consultation. Patients who collaborated but were not willing to complete the consultation were contacted by phone, and for those unable to come to the hospital, home visits were scheduled. For patients known to be deceased based on previous information, a family member/caregiver had to give information about the date and circumstances of death; otherwise, a search was done in the computer files held at the Northern Regional Health Administration. In case of death, information about date and circumstances of death was confirmed by manual inspection of written monthly reports at each health centre since current legislation forbids the use of death certificates for research purposes. This information was linked to existing clinical records for assigning the underlying cause of death, determined by a study neurologist. If no contact or information could be obtained, the patient was considered lost to follow-up. Stroke was defined according to the World Health Organization [7] , and stroke types were classified according to Sudlow and Warlow [5] as ischaemic stroke (IS), primary intra1-year follow-up (484 patients) 159 exclusions: 8 – out of study period 39 – out of study population 72 – other diagnosis 40 – prior stroke 688 incident events Identified as 26 TIA 19 TNA Long-term follow-up (265 patients) Face to face (78.9%)209 183 25 4 7 (46.7%) (40.3%) (26.7%) (46.7%) 138 49 8 15 (25.2%) (44.1%) (34.8%) (50.0%) 21 26 9 By phone (7.9%) Clinical records (9.8%) Lost (3.4%) 204 died: Ischaemic stroke Intracerebral haemorrhage Subarachnoid haemorrhage Undetermined stroke type 219 Died: Ischaemic stroke Intracerebral haemorrhage Subarachnoid haemorrhage Undetermined stroke type 802 patients identified Fig. 1. Ascertainment and inclusion of FELS in northern Portugal and details of the 7-year followup. Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 140 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel cerebral haemorrhage (PICH) and subarachnoid haemorrhage (SAH). If no brain CT scan was performed within 30 days and no autopsy, lumbar puncture or angiography (in case of suspected SAH) results were available, stroke was classified as undetermined. For VRF, the following definitions were used: for hypertension, a history of high blood pressure (BP) or an anti-hypertensive treatment or systolic BP >160 mm Hg and/or diastolic BP >95 mm Hg on at least two different measures; for diabetes, a previous diagnosis/treatment of diabetes mellitus with oral anti-diabetic/insulin or fasting glycaemia >126 mg/dl, postprandial glycaemia ≥ 200 mg/dl and/or a glucose tolerance test with values of glycaemia ≥ 200 mg/dl at the second hour; for hypercholesterolaemia, a previous diagnosis/treatment of hypercholesterolaemia or a serum total cholesterol level after 12 h of fasting ≥ 240 mg/dl; and for cardiac disease, a previous diagnosis of angina, myocardial infarction (MI) or atrial fibrillation by EKG confirmation, a previous diagnosis of a transient ischaemic attack (TIA), and smoking, categorized as never smoked, smoked regularly but not in the preceding 12 months (ex-smoker) and current smoker. The Ethics Committee of the Hospital de Santo António, where the study coordination centre was located, approved the study. Informed consent was obtained from each participant, or from the next of kin when appropriate, before any clinical assessment. Since medical records are part of the National Health Service institutions, for follow-up purposes clinical files were used whenever the patient could not be contacted. Statistical Analysis The distribution of patient characteristics at baseline according to stroke type is described. The crude incidence rates age-standardized to the Portuguese [8] and European populations [9] are reported, and the 95% confidence intervals (CI) were calculated by the Poisson distribution. The rural/urban ratios of VRF prevalence, stroke incidence and case fatality were calculated based on cross-tabulation and were used to compare rural and urban patients. The Kaplan-Meier estimates for the cumulative risk of death for stroke types over a period of 7 years after the index event were calculated in rural and urban patients. After checking the assumption of proportional hazards with the Schoenfeld’s test, the rural/urban hazard ratios (HR) were calculated using a Cox model including the baseline risk profiles. Since this assumption failed when considering the time from the index event until death over the 7-year follow-up, this model was restricted to patients surviving the acute phase (28 days). Results Of the 688 FELS (226 in rural and 462 in urban areas), 76.2% were IS (75.3 in rural vs. 77.9% in urban), 16.1% were PICH (16.3 vs. 14.6%), 3.3% SAH (2.7 vs. 3.7%) and 4.4% of undetermined stroke type (4.9 vs. 4.1%). More cases in rural compared to urban areas were ascertained by ‘hot-pursuit’ and sooner after the event. Nearly 56% of the patients were admitted to the hospital, with a similar proportion for IS in both rural and urban areas but a lower proportion of PICH in the rural area (69.7 vs. 96.2%). Overall, a CT scan was performed in 96.9% of the patients (not done in 12 urban and 9 rural patients) and in 70.3% within 24 h following the event. Vascular Risk Profiles and Incidence Although patients were mostly women (58.7%), men predominated in rural areas (48.2 vs. 37.9%) either with an IS or PICH; patients from rural areas were older than patients from urban areas, especially those with an IS ( table 1 ). Hypertension was the most prevalent VRF (60.9%), whereas a previous TIA was seldom registered (8.6%). In general, the prevalence of Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 141 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel Table 1. Ascertainment and patient characteristics (in %) by stroke type in urban and rural areas Urban Rural Ratio rural:urban IS (n = 348) PICH (n = 78) SAH (n = 17) All (n = 462) IS (n = 176) PICH (n = 33) SAH (n = 6) All (n = 226) IS PICH ratio 95% CI ratio 95% CI First source of information ‘Hot pursuit’ 76.4 56.4 23.5 69.5 97.2 97.0 83.3 94.7 1.27 1.19 – 1.35 1.72 1.40 – 2.11 Health centre 5.5 – – 5.6 47.2 60.6 16.7 48.7 Hospital 71.0 56.4 23.5 63.9 50.0 36.4 66.7 46.0 Patient assessment Emergency services 93.7 97.4 100.0 92.0 93.8 97.0 100.0 91.2 1.00 0.95 – 1.05 1.00 0.93 – 1.07 <3 h 36.8 50.0 52.9 39.5 48.8 71.9 66.7 52.9 1.33 1.08 – 1.63 1.45 1.07 – 1.97 In-patient admission 50.3 96.2 100.0 57.8 50.6 69.7 100.0 52.2 1.01 0.84 – 1.20 0.72 0.58 – 0.91 Time between onset and CT scan <24 h 67.2 83.3 70.6 69.1 67.0 66.7 50.0 65.9 1.00 0.88 – 1.13 0.80 0.62 – 1.04 Patient characteristics Male gender 38.2 37.5 29.4 37.9 48.9 51.5 – 48.2 1.28 1.05 – 1.56 1.39 0.89 – 2.15 Mean age ± SD, years 71.0 ± 13.1 67.2 ± 14.6 59.9 ± 19.0 70.3 ± 13.9 73.6 ± 9.4 67.5 ± 12.3 58.8 ± 19.7 72.5 ± 10.9 >65 years 71.0 55.1 52.9 68.2 84.1 60.6 50.0 80.1 1.18 1.08 – 1.30 1.10 0.78 – 1.54 Hypertension 62.9 69.2 52.9 62.3 58.0 69.7 – 58.0 0.92 0.79 – 1.07 1.01 0.77 – 1.32 Hypercholesterolaemia 41.4 28.2 17.6 37.9 28.4 9.1 16.7 24.8 0.69 0.53 – 0.90 0.32 0.10 – 1.00 Diabetes 29.6 24.4 11.8 27.1 21.6 9.1 – 18.6 0.73 0.53 – 1.01 0.37 0.12 – 1.18 Atrial fibrillation 17.2 3.8 – 13.9 17.0 – – 13.3 0.99 0.66 – 1.47 – MI/angina 11.5 10.3 5.9 10.6 5.7 3.0 – 5.3 0.49 0.25 – 0.96 0.30 0.04 – 2.27 TIA 12.4 2.6 – 9.7 7.4 – – 6.2 0.60 0.33 – 1.08 – Smoking habits Current smoker 17.2 12.8 23.5 16.9 8.0 6.1 – 8.0 0.46 0.27 – 0.80 0.47 0.11 – 2.04 Ex-smoker 9.2 5.1 – 7.8 5.7 12.1 – 6.2 0.60 0.30 – 1.19 2.36 0.63 – 8.89 Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 142 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel Table 2. Age-specific annual incidence per 1,000 for stroke types in urban and rural northern Portugal (1998 – 2000) Age group (in years) Personyears IS PICH SAH Undetermined n rate 95% CI n rate 95% CI n rate 95% CI n rate 95% CI Urban 00 – 34 68,706 2 0.03 0.00 – 0.11 2 0.03 0.00 – 0.11 1 0.01 0.00 – 0.08 0 0.00 0.00 – 0.05 35 – 44 24,806 15 0.60 0.34 – 1.00 4 0.16 0.04 – 0.41 3 0.12 0.02 – 0.35 0 0.00 0.00 – 0.15 45 – 54 23,500 24 1.02 0.65 – 1.52 9 0.38 0.18 – 0.73 3 0.13 0.03 – 0.37 2 0.09 0.01 – 0.31 55 – 64 19,584 48 2.45 1.81 – 3.25 17 0.87 0.51 – 1.39 1 0.05 0.00 – 0.28 0 0.00 0.00 – 0.19 65 – 74 19,544 102 5.22 4.21 – 6.23 20 1.02 0.63 – 1.58 6 0.31 0.11 – 0.67 5 0.26 0.08 – 0.60 75 – 84 11,812 108 9.14 7.42 – 10.9 14 1.19 0.65 – 1.99 1 0.08 0.00 – 0.47 6 0.51 0.19 – 1.11 ≥85 4,094 49 11.97 8.85 – 15.8 12 2.93 1.51 – 5.12 2 0.49 0.06 – 1.76 6 1.47 0.54 – 3.19 All 172,046 348 2.02 1.81 – 2.24 78 0.45 0.36 – 0.57 17 0.10 0.06 – 0.16 19 0.11 0.07 – 0.17 ASRP 1.63 1.44 – 1.82 0.38 0.29 – 0.48 0.09 0.05 – 0.14 0.08 0.04 – 0.14 ASRE 1.26 1.10 – 1.44 0.32 0.24 – 0.42 0.08 0.04 – 0.13 0.06 0.03 – 0.11 Rural 00 – 34 33,690 1 0.03 0.00 – 0.17 0 0.00 0.00 – 0.11 1 0.03 0.00 – 0.17 0 0.00 0.00 – 0.11 35 – 44 9,972 2 0.20 0.02 – 0.72 0 0.00 0.00 – 0.37 1 0.10 0.00 – 0.56 1 0.10 0.00 – 0.56 45 – 54 8,360 2 0.24 0.03 – 0.86 7 0.84 0.34 – 1.72 0 0.00 0.00 – 0.44 0 0.00 0.00 – 0.44 55 – 64 8,386 19 2.27 1.36 – 3.54 6 0.72 0.26 – 1.56 1 0.12 0.00 – 0.66 0 0.00 0.00 – 0.44 65 – 74 8,350 64 7.66 5.90 – 9.79 10 1.20 0.57 – 2.20 2 0.24 0.03 – 0.86 3 0.36 0.07 – 1.05 75 – 84 4,152 71 17.10 13.4 – 21.6 8 1.93 0.83 – 3.80 1 0.24 0.01 – 1.34 4 0.96 0.26 – 2.47 ≥85 1,268 17 13.41 7.81 – 21.5 2 1.58 0.19 – 5.69 0 0.00 0.00 – 2.91 3 2.37 0.49 – 6.92 All 74,178 176 2.37 2.02 – 2.72 33 0.44 0.31 – 0.62 6 0.08 0.03 – 0.18 11 0.15 0.07 – 0.27 ASRP 2.12 1.79 – 2.45 0.42 0.28 – 0.59 0.08 0.03 – 0.18 0.13 0.06 – 0.25 ASRE 1.51 1.23 – 1.79 0.35 0.23 – 0.51 0.07 0.02 – 0.16 0.09 0.04 – 0.19 All 00 – 34 102,396 3 0.03 0.01 – 0.09 2 0.02 0.00 – 0.07 2 0.02 0.00 – 0.07 0 0.00 0.00 – 0.04 35 – 44 34,778 17 0.49 0.28 – 0.78 4 0.12 0.03 – 0.29 4 0.12 0.03 – 0.29 1 0.03 0.00 – 0.16 45 – 54 31,860 26 0.82 0.53 – 1.20 16 0.50 0.29 – 0.82 3 0.09 0.02 – 0.28 2 0.06 0.01 – 0.23 55 – 64 27,970 67 2.40 1.86 – 3.04 23 0.82 0.52 – 1.23 2 0.07 0.01 – 0.26 0 0.00 0.00 – 0.13 65 – 74 27,894 166 5.95 5.05 – 6.86 30 1.08 0.73 – 1.54 8 0.29 0.12 – 0.56 8 0.29 0.12 – 0.56 75 – 84 15,964 179 11.21 9.57 – 12.9 22 1.38 0.86 – 2.09 2 0.13 0.02 – 0.45 10 0.63 0.30 – 1.15 ≥85 5,362 66 12.31 9.52 – 15.7 14 2.61 1.43 – 4.38 2 0.37 0.04 – 1.35 9 1.68 0.77 – 3.19 All 246,224 524 2.13 1.95 – 2.31 111 0.45 0.37 – 0.53 23 0.09 0.06 – 0.14 30 0.12 0.08 – 0.17 ASRP 1.76 1.60 – 1.93 0.39 0.32 – 0.48 0.08 0.05 – 0.13 0.10 0.06 – 0.15 ASRE 1.33 1.19 – 1.48 0.33 0.26 – 0.41 0.07 0.04 – 0.12 0.07 0.04 – 0.11 Incidence rate ratios rural/urban 00 – 54 0.27 0.11 – 0.69 1.05 0.43 – 2.57 0.64 0.13 – 3.09 1.12 0.10 – 12.4 55 – 64 0.92 0.54 – 1.57 0.82 0.33 – 2.09 2.34 0.15 – 37.3 – 65 – 74 1.47 1.08 – 2.01 1.17 0.55 – 2.50 0.78 0.16 – 3.86 1.40 0.34 – 5.88 75 – 84 1.87 1.39 – 2.52 1.63 0.68 – 3.87 2.84 0.18 – 45.5 1.90 0.54 – 6.72 ≥85 1.12 0.65 – 1.94 0.54 0.12 – 2.40 – 1.61 0.40 – 6.45 All 1.19 0.95 – 1.50 1.10 0.69 – 1.75 0.89 0.32 – 2.50 1.62 0.62 – 4.27 ASRP = Age-standardized rate for the Portuguese population, 1999; ASRE = age-standardized rate for the European population. Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 143 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel VRF was lower in rural compared to urban patients, namely cardiac disease (IS and ICH), hypercholesterolaemia and smoking habits (IS). The crude overall annual incidence rate per 1,000 population was 2.13 (95% CI, 1.95– 2.31) for IS, 0.45 (95% CI, 0.37–0.53) for PICH, 0.09 (95% CI, 0.06–0.14) for SAH and 0.12 (95% CI, 0.08–0.17) for strokes of undetermined type; adjusted for the European population, these rates were 1.33 (95% CI, 1.19–1.48), 0.33 (95% CI, 0.25–0.41), 0.07 (95% CI, 0.04–0.12) and 0.07 (95% CI, 0.04–0.11), respectively ( table 2 ). The ratio of rates indicates that the incidence of IS in the youngest group (<55 years) was lower in rural compared to urban populations, particularly in men (0.12; 95% CI, 0.02–0.91). The opposite trend was found among those aged 65–84 years (2.19; 95% CI, 1.37–3.49 in men and 1.65; 95% CI, 1.12–2.45 in women; results not shown). For the remaining stroke types, there was no evidence of differences in the age pattern of incidence rates between the rural and urban environment. Shortand Long-Term Survival Of the 688 FELS patients, 204 (29.7%) died during the first year; the 7-year follow-up details for the 484 survivors are described in figure 1 . Nine patients were lost after the 1-year follow-up (1.1–2.4 years), mostly because they had changed residence or went abroad (7 IS, 1 PICH and 1 SAH). Among the 209 (78.9%) patients examined by the neurologist, 21 (10%) were visited at their homes. The follow-up time ranged from 7 to 8.6 years. By day 28, 59 (11.3%), 34 (30.6%), 7 (30.4%) and 11 (36.7%) patients had died after the first IS, PICH, SAH and undetermined stroke, respectively. There was no evidence of rural/ urban differences in 28-day case fatality for the different stroke types, although IS tended to be less fatal among urban patients (10.3 vs. 13.1%), whereas PICH (33.3 vs. 24.2%) and SAH (35.3 vs. 16.7%) were less fatal among rural patients, corresponding to rural/urban ratios of 1.26 (95% CI, 0.77–2.06), 0.73 (95% CI, 0.37–1.44) and 0.47 (95% CI, 0.07–3.16), respectively. The cumulative risk of death at 7 years followed the same pattern: IS was less fatal in urban patients (57.4 vs. 61.5%), and PICH (67.4 vs. 64.6%) or SAH (61.2 vs. 33.3%) were less fatal in rural patients. The proportional mortality from stroke (first or recurrent stroke) was 74.8% (83/111) at 28 days, 48.0% (98/204) during the first year and 30.3% (128/423) at the end of follow-up after 7 years. Figure 2 shows the risk of death at 28 days and by year for ischaemic and haemorrhage stroke (PICH and SAH), indicating a relatively constant yearly risk after 3 years in patients with an IS. Table 3 shows the independent baseline predictors of long-term survival after the acute phase. Besides age, diabetes (HR = 1.48; 95% CI, 1.15– 1.92), MI/angina (HR = 1.80; 95% CI, 1.25–2.58), atrial fibrillation (HR = 1.47; 95% CI, 1.07– 2.00) and being a current smoker (HR = 1.60; 95% CI, 1.13–2.28) increased the risk of death, whereas hypercholesterolaemia was a protective factor (HR = 0.45; 95% CI, 0.34–0.59). Stroke type and rural/urban residence were not associated with survival after the acute phase. Discussion This is the first study to present a comprehensive picture of the burden of stroke among rural and urban populations, looking at the incidence of stroke types as well as at vascular risk profiles and long-term survival of patients. A high proportion of patients were ascertained by ‘hot-pursuit’; almost all underwent a CT soon after the initial symptoms, thus improving the reliability of the results for the incidence of stroke types [5] . Differences in procedural aspects mostly stem from the organization of the National Health Service; health centre services are more readily available than hospital services for rural populations, and in the city this is mainly a question of choice since there are no barriers of distance. Almost all Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 144 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel a 7654 Time (years after stroke) Cumulative risk of death 3210 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Ischaemic stroke Intracerebral haemorrhage Undetermined stroke type Subarachnoid haemorrhage Ischaemic stroke Intracerebral haemorrhage Undetermined stroke type Subarachnoid haemorrhage 7654 Time (years after stroke) 3210 b Cumulative risk of death 0 0.1 0.2 0.3 0.4 0.5 Stroke type 0 – 28 days 29 days to 1 year 1 – 2 years 2 – 3 years 3 – 4 years 4 – 5 years 5 – 6 years 6 – 7 years Ischemic At risk, n 524 465 392 351 331 301 266 242 Death, n (N) 59 (59) 73 (132) 38 (170) 17 (187) 30 (217) 35 (252) 23 (275) 28 (303) Risk, % 11.3 15.7 9.8 4.9 9.1 11.6 8.7 11.6 95% CI 8.8 – 14.3 12.7 – 19.3 7.2 – 13.1 3.1 – 7.7 6.4 – 12.6 8.5 – 15.7 5.9 – 12.7 8.1 – 16.2 Hemorrhagic At risk, n 134 93 77 66 61 61 56 54 Death, n (N) 41 (41) 16 (57) 9 (66) 5 (71) 0 (71) 5 (76) 2 (78) 7 (85) Risk, % 30.6 17.2 12.0 7.6 0.0 8.2 3.6 13.0 95% CI 23.4 – 38.9 10.9 – 26.1 6.4 – 21.3 3.3 – 16.5 0.0 – 5.9 3.6 – 17.8 1.0 – 12.1 6.4 – 24.4 Fig. 2. Kaplan-Meier estimates of the cumulative death risk for all patients ( a ) and the cumulative death risk in 28-day stroke survivors by stroke type ( b ). N = Cumulative number of patients. Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM 145 Cerebrovasc Dis Extra 2013;3:137–149 DOI: 10.1159/000354851 EXTRA Correia et al.: Stroke Types in Rural and Urban Northern Portugal: Incidence and 7-Year Survival in a Community-Based Study www.karger.com/cee © 2013 S. Karger AG, Basel patients (91–92%) were seen at an ‘emergency service’; however, in rural areas it was mostly at the health centre (open 24 h) and this is why these patients are more often observed within 3 h after the stroke. Nevertheless, in case of a PICH, there were fewer hospital admissions of rural compared with urban patients, though they remained in-patients in the health centre. Overall, the health services provided were similar for rural and urban patients, and based on previous studies we know that individuals living in rural areas are more prone to attend the family doctor at the health centre in case of most stroke warning signs than individuals living in urban areas [10] . There was a higher incidence of both IS and PICH in rural than urban areas. Nevertheless, the comparison of standardized rates obscured the differences in the incidence age pattern in the two populations, particularly in IS. On average, the first IS happened almost 3 years earlier in life among the urban population, leading to a higher IS incidence in the youngest group (<55 years), especially in men, whereas for those aged 75–84 years living in rural areas, the average risk is almost twice as high than in the city. Although patients living in rural areas were older, they had, in general, less traditional VRF than patients living in urban areas, in particular cardiac disease and hypercholesterolaemia in patients with IS. The reduced information and awareness of VRF [10] in rural areas and the consequent lack of monitoring probably led to an under-reporting and/or under-diagnosis, mainly by GPs, since by description of ascertainment, health centre services are ‘more accessible’ in rural areas. On the other hand, our results go in the same direction than those from a Dutch study [11] , in which self-reported health problems pointed to a better health in rural areas, although this could not be confirmed by the information available on GP records. Figure 3 shows the joint distribution of IS and PICH incidence across community-based studies with standardized rates (European population) or if data were available for calculation [12–34] . The IS incidence ranged from 57/100,000 in Menorca [29] to 255/100,000 in Table 3. HR for the association between factors at presentation and death among 28-day survivors Univariate Multivariate HR 95% CI HR 95% CI Sociodemographics Rural versus urban 1.18 0.93 – 1.49 1.00 0.79 – 1.29 Men versus women 1.00 0.80 – 1.26 0.94 0.71 – 1.24 Age >65 versus ≤65 years 3.58 2.62 – 4.90*** 3.57 2.58 – 4.95*** Risk factors (yes vs. no) Hypertension 0.81 0.64 – 1.02 0.86 0.68 – 1.09 Diabetes 1.21 0.94 – 1.56 1.48 1.15 – 1.92** Atrial fibrillation 2.01 1.49 – 2.72*** 1.47 1.07 – 2.00* MI/angina 1.58 1.11 – 2.23** 1.80 1.25 – 2.58** TIA 0.92 0.61 – 1.37 0.99 0.66 – 1.47 Hypercholesterolaemia 0.46 0.36 – 0.60*** 0.45 0.34 – 0.59*** Smoking habits Ex-smoker 0.65 0.42 – 1.02 0.80 0.49 – 1.31 Current smoker 1.07 0.80 – 1.43 1.60 1.13 – 2.28** Diagnosis (vs. IS) PICH 0.96 0.69 – 1.34 1.12 0.79 – 1.58 SAH 0.52 0.21 – 1.25 0.63 0.26 – 1.75 Undetermined stroke type 1.15 0.63 – 2.10 1.28 0.69 – 2.36 * p < 0.05; ** p < 0.01; *** p < 0.001. Downloaded by: 193.136.36.134 - 10/23/2013 1:06:47 PM Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort - 29 - Abstract Background: The impact of public health interventions to reduce disability after stroke may be underestimated if only the modest effects on short-term disability are measured. We estimated the impact of differences in short-term functional outcome on long-term functional outcome. Methods: In a prospective community-based study from October 1998 to September 2000, the first-ever ischemic strokes were registered in a population of 95,816 in Northern Portugal. Patients were examined at baseline and followed-up at three months, one and seven years. The Oxfordshire classification and the Unified Neurological Stroke Scale were used to define the stroke type and the severity of neurological impairments. The functional status was assessed with the modified Rankin Scale (mRS). Ridit analysis was used to estimate the odds of a more serious 7-year outcome based on the adjacent values of the 3-month mRS. Cox proportional hazards models were used for estimating the effect of 3-month mRS on survival, adjusting for patients’ characteristics, stroke type and severity. Results: The odds of a more serious 7-year outcome was different among patients with mRS=1 and 2 and also with mRs=3 and 4, defining the no significant (mRS=0-1), moderate (mRS=2-3) and severe disability (mRS=4-5). Of the 380 first-ever strokes, at 3 months, 126 (33.2%) had mRS<2, 114 (30.0%) mRS=2-3, 73 (19.2%) mRS=4-5, and 67 (17.6%) had died. We found linear relations between the 3-month mRS and the patient’s baseline profile, stroke type and severity. The Kaplan-Meier 7-year survival estimates for 3-month survivors with mRS 0-1, 2-3 and 4-5 were 67, 50 and 23%, respectively. For mRS at 3 months of 2-3 versus 0-1 the hazard ratio (HR) for death was 1.61, (95% CI: 1.10-2.38) and for mRS=4-5 versus 2-3 the HR was 2.20 (95% CI: 1.52-3.20); after adjustment the HRs were 1.19 (95% CI: 0.77-1.84) and HR=1.87 (95% CI: 1.18-2.95), respectively. A change in the 3-month mRS from 4-5 to 2-3 would have a “number needed to change” of 9 (95% CI: 6-18) patients to avoid one death in the long run; identical outcome is obtained by shifting the mRS from 2-3 to 0-1 in 27 (95% CI: 15-141) patients. Conclusions: In patients with ischemic stroke who survive to 3 months, a three grade simplified mRS summarizes the patient risk profile and stroke characteristics. These data confirm that modest differences in functional status at 3 months are associated with significant differences in survival and functional status over 7 years follow-up and have implications for health care planning and the health economic assessment of treatments for acute stroke. Tendências na Incidência e Prognóstico do AVC - 30 - Introduction By the year 2015, it is estimated that there will be 67 million stroke survivors worldwide [1], a majority of them with some disability. Treatments for acute ischemic stroke make only modest gains in short-term disability. However, small changes in short-term disability might translate into longer-term improvements in survival and functional status. If differences in functional outcome at three months are associated with differences in longterm survival, or prolonged differences in levels of independence, then such data could materially influence estimates of the cost-effectiveness of many interventions. The modified Rankin Scale (mRS) has been widely used to measure functional outcome of stroke patients as an endpoint either in randomized clinical trials [2] or in the early months after stroke [3]. Further studies have shown its prognostic value for long-term survival in short-term survivors, that is, between 3 and 6 months and seven or more years [4-6]. Yet mortality from stroke has constantly declined in the last two decades as well as the mortality-to-incidence ratio, indicating the success of stroke management in the acute phase and the consequent increase in the number of stroke survivors [7]. Therefore, it is important to discriminate short-term levels of disability (cut-points) that evolve to a worse long-tem mRS score. We aim to optimize mRS cut-points for estimating the impact of functional status three months after a stroke on long-term functional outcome and survival in a prospective, population-based cohort of patients with ischemic stroke.Patients and Methods In a prospective population-based study, all patients with a first-ever-in-lifetime ischemic stroke were registered in four primary healthcare centers (HC) in Northern Portugal (two urban and two rural centers with 58,727 and 37,089 users, respectively) between 1 st October 1998 and 30 th September 2000. The study was designed to meet the criteria of an “ideal” population-based study [8]. We aimed to comprehensively ascertain stroke cases by monitoring the (i) referrals from general practitioners (GP) and other hospital physicians working at the healthcare centers and hospitals; (ii) data retrieved from admission/discharge records; (iii) hospital outpatient records; (iv) nursing homes, private hospitals/practices and seniors residences; and (v) death certificates/autopsy. The study is described in more detailed elsewhere [9]. All patients gave written informed consent. The Ethics Committee of Hospital Santo António, where the study Coordination Centre was located, approved the study. Before any clinical assessment, informed consent was obtained from each participant, or from the next of kin when appropriate. Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort - 31 - Baseline Assessment All patients were assessed by a study neurologist immediately after the index event [9]. Definite ischemic stroke was defined as a focal neurological deficit lasting for more than 24 h with no evidence of hemorrhage on brain imaging [10]. We registered demographic details, vascular risk factors (VRF), radiological and laboratory investigations, treatment patterns and destination details after discharge. CT brain lesions were read by a neuroradiologist blind to clinical symptoms/signs; based on the description of the CT, the neurologist assigned any acute lesion compatible with neurological symptoms and signs. To quantify neurological impairments at baseline, we used the Unified for Neurological Stroke Scale (UNSS), which is a continuous scale of stroke impairment that had good inter-rater agreement [11] and validity [12]. As our investigative resources were limited, and most classification systems available at the time of the inception of the study failed to classify patients into causal groups, we classified the subtype of stroke according to the Oxfordshire Community Stroke Project Classification [13] as: Total Anterior Circulation Infarct (TACI), Partial Anterior Circulation Infarct (PACI), LaCunar Infarct (LACI) or Posterior Circulation Infarct (POCI). The criteria used to define VRF were for hypertension, a history of high blood pressure (BP) or anti-hypertensive treatment or systolic BP >160 mm Hg and/or diastolic BP >95 mm Hg in at least two different measures; for diabetes, a previous diagnosis/treatment of diabetes mellitus with oral antidiabetic/insulin or fasting glycaemia >126 mg/dl, postprandial glycaemia 200 mg/dl and/or glucose tolerance test with values of glycaemia 200 mg/dl at the 2 nd hour; for hypercholesterolemia, a previous diagnosis/treatment of hypercholesterolemia or serum total cholesterol level after 12 h fasting 240 mg/dl; for atrial fibrillation, evidence from the electrocardiogram (ECG) or registration in patient’s record of atrial fibrillation; patients were classified as current smokers if they smoked at the event date or at any time during the preceding 12 months [14]. Long-term Follow-up All patients were followed-up prospectively at three months, one and seven years. The long-term follow-up began by updating telephone contacts using HC/hospital administrative records; first a phone contact was made and when it failed, patients were contacted by mail (at least two consecutive attempts) [14]. For patients known to have died, we asked a family member or caregiver to inform us of the date and circumstances of death; otherwise, we searched the computerized files held at the Northern Regional Health Administration. This information was linked to existing clinical records, and the underlying cause of death was assigned by a study neurologist. A vascular cause of Tendências na Incidência e Prognóstico do AVC - 32 - death was defined as (a) death due to first or recurrent stroke, (b) death due to acute myocardial infarction (MI) whenever the event was confirmed by ECG, myocardial necrosis markers or necropsy and (c) death due to other vascular causes: suspected but not confirmed MI, heart failure, thoracic/abdominal aortic aneurysm rupture or sudden death of presumed vascular origin. Patients were examined at the neurological outpatient clinic and those who were unable to visit the clinic were examined at home. If no contact could be established, information was obtained from the health center or hospital records. Whenever these initiatives failed, the patient was considered lost to follow-up. Functional status at discharge and at 3 months, 1 and 7 years was assessed with the modified Rankin Scale (mRS). Data Analysis Ridit analysis [15] was used to estimate the odds of a more serious outcome (distribution of mRS scores 7 years post-stroke) according to adjacent values of mRS at 3 months, that is, by comparing the outcome between patients with mRS=i and mRS=i-1, for i=1 to 5. On the basis of this analysis, groups of patients with an equally likely long-term outcome were created (a simplified version of the mRS at 3 months). The distribution of patients’ characteristics, VRF and clinical evaluation at baseline across these groups was tested for linear trend. The Kaplan-Meier estimates for overall survival were calculated according to the original and grouped mRS scores in 3-month survivors. Cox proportional hazards models were used for estimating the effect of mRS scores on survival, adjusting for possible confounding variables (gender, age, VRF, pre-stroke mRS, UNSS score and OCSP subtype), after checking the assumption of proportionality with the Schoenfeld’s test. Patients where censored at the time of the last contact with the HC/Hospital if no information on vital status was available at the 7-year follow-up. Using death rates/person-years at 7 years, we estimated the number of patients who would need to have a lower mRS score at 3 months to avoid one death at 7 years. The number needed to change (NNC) to avoid one death at 7 years is given by 1/(mortality rate in the mRS=i group – mortality rate in mRS=i-1 group). Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort - 33 - Results From the 623 patients suspected of stroke registered at the 4 HC, 124 were excluded and among the 499 included, 380 had an ischemic stroke (figure 1). At 3 months 126 (33.2%) had a mRS score <2, 114 (30.0%) a 2-3 mRS score, 73 (19.2%) a 4-5 mRS score, and 67 (17.6%) died (table 1). A worse functional status at 3 months was associated with age, increasing from a median of 70 to 79 years, prevalence of atrial fibrillation, increasing from 9.5% to 34.3% and pre-stroke mRS 2, increasing from 1.6% to 43.8%. A more favourable functional status at 3 months was associated with male gender, the proportion decreasing from 59.5% to 37.3% and the prevalence of hypercholesterolemia, decreasing from 49.2% to 16.4%. All patients had a CT scan; 245 (64.5%) in the first 24 h and 62 (16.3%) between 24 and 48 h after stroke onset; the proportion of patients with a symptomatic lesion on CT ranged from 34.1% in patients with mRS=0-1 to 54.8% when mRS=4-5. Stroke type and severity were also associated with a poor 3-month outcome, the proportion of TACI increasing from 2.4% to 52.2% contrasting with LACI, decreasing from 61.1 to 11.9%. Assessed for eligibility (n=623) Strokes included (n=499) Long-term follow-up modified Rankin Scale 0-1 2-3 4-5 All - Lost 3 6 5 14 - Died 46 58 53 157 - Alive - Face to face 69 39 10 118 - By phone 1 5 2 8 - Clinical records 7 6 3 16 Exclusions (n=124) Logistic reasons - Out of study period (n=8) - Out of study population (n=26) After neurological assessment - Recurrent episode (n=31) - Other diagnosis (n=59) Ischaemic Stroke (n=380) mRS 0-1 (n=126) mRS 2-3 (n=114) mRS 6 (n=67) mRS 4-5 (n=73) Assessed for eligibility (n=623) Assessed for eligibility (n=623) Strokes included (n=499) Strokes included (n=499) Long-term follow-up modified Rankin Scale 0-1 2-3 4-5 All - Lost 3 6 5 14 - Died 46 58 53 157 - Alive - Face to face 69 39 10 118 - By phone 1 5 2 8 - Clinical records 7 6 3 16 Long-term follow-up modified Rankin Scale 0-1 2-3 4-5 All - Lost 3 6 5 14 - Died 46 58 53 157 - Alive - Face to face 69 39 10 118 - By phone 1 5 2 8 - Clinical records 7 6 3 16 Exclusions (n=124) Logistic reasons - Out of study period (n=8) - Out of study population (n=26) After neurological assessment - Recurrent episode (n=31) - Other diagnosis (n=59) Exclusions (n=124) Logistic reasons - Out of study period (n=8) - Out of study population (n=26) After neurological assessment - Recurrent episode (n=31) - Other diagnosis (n=59) Ischaemic Stroke (n=380) Ischaemic Stroke (n=380) mRS 0-1 (n=126) mRS 0-1 (n=126) mRS 2-3 (n=114) mRS 2-3 (n=114) mRS 6 (n=67) mRS 6 (n=67) mRS 4-5 (n=73) mRS 4-5 (n=73) Figure 1. Flowchart of inclusion of patients and details of 7-year follow-up Tendências na Incidência e Prognóstico do AVC - 34 - Of the 313 three-month survivors, 157 (50.2%) died before 7-year follow-up and 14 (4.5%) were lost either because they could not be traced after the first year (9) or had unknown mRS (5). Of the 142 patients alive 7 years post-stroke, 118 (83.1%) were examined by a neurologist of the research team (6 of them at home and the remainder at the hospital), 8 (5.6%) were contacted by phone and for the remaining 16 information was sought at hospital/HC records (figure 1). For the 299 patients with known 7-year status, Ridit analysis indicated that the odds of a worst functional status at 7 years was 1.71 (z=3.1, p<0.001) when comparing a patient with mRS=2 to mRS=1, and 1.55 (z=2.3, p<0.02) for mRS=4 versus mRS=3; this analysis was repeated for the 142 that survived 7 years and the corresponding odds were 1.93 (z=2.7, p<0.007) and 4.06 (z=3.4, p<0.001), respectively. Table 1. Distribution of patients’ characteristics at baseline and according to status at three months Status at three months Alive All (n=380) All (n=313) mRS 0-1 (n=126) mRS 2-3 (n=114) mRS 4-5 (n=73) mRS 6 (n=67) P value linear trend Characteristics N % N % N % N % N % N % Men 170 44.7 145 46.3 75 59.5 46 40.4 24 32.9 25 37.3 <0.001 Median age (IQD) 73.0 (7.5) 72.0 (7.3) 70.0 (7.3) 73.0 (8.1) 77.0 (7.8) 79.0 (6.5) <0.001 Hypertension 229 60.3 36 53.7 76 60.3 71 62.3 46 63.0 36 53.7 0.6 Hypercholesterolemia 136 35.8 11 16.4 54 49.2 50 43.9 21 28.8 11 16.4 <0.001 Diabetes 100 26.3 23 34.3 29 23.0 26 22.8 22 30.1 23 34.3 0.07 Atrial fibrillation 60 15.8 23 34.3 12 9.5 12 10.5 13 17.5 23 34.3 <0.001 Previous TIA 38 10.0 6 9.0 15 11.9 11 9.6 6 8.2 6 9.0 0.4 Current smoker 50 13.2 3 4.5 19 15.1 22 19.3 6 8.2 3 4.5 0.02 Pre-stroke mRS* <0.001 0-1 296 78.5 260 83.1 124 98.4 88 77.2 48 65.8 36 56.3 2-3 71 18.8 46 14.7 2 1.6 25 21.9 19 26.0 25 39.1 4-5 10 2.7 7 2.2 - - 1 0.9 6 8.2 3 4.7 Symptomatic lesion on CT 170 44.7 137 43.8 43 34.1 54 47.4 40 54.8 33 49.3 0.006 Mean UNSS (sd) 23.3 (9.0) 29.2 (3.7) 29.2 (3.7) 26.2 (5.8) 17.3 (8.7) 14.0 (9.5) <0.001 OCSP Classification <0.001 TACI 87 22.9 35 52.2 3 2.4 17 14.9 32 43.8 35 52.2 PACI 70 18.4 12 17.9 20 15.9 23 20.2 15 20,5 12 17.9 LACI 154 40.5 8 11.9 77 61.1 53 46.5 16 21.9 8 11.9 POCI 69 18.2 12 17.9 26 20.6 21 18.4 10 13.7 12 17.9 *Three missings in the mRS=6 group; IQD=Interquartile deviation Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort - 35 - Figure 2 shows the distribution of functional status at 7 years according to the 3-month mRS scores compared in the Ridit analysis; the shading in the left side draws attention to the similar 7-year functional outcome of survivors for patients with mRS=0-1, 2-3 and 4-5 at 3 months, while the right side shows the proportion of patients deceased at 7 years according to mRS at 3 months. The proportions of patients with severe disability (mRS=45) were 8.1, 9.3 and 17.6% according to mRS at 3 months 0-1, 2-3 and 4-5 contrasting with no significant disability (mRS=0-1), 36.6, 13.9 and 0% (table 2). 6 modified Rankin Scale at 7 years: 0 1 2 3 4 5 0+2 2 1 2 4 2 6 5 2 4 2 4 2 3 9 1 5 11 7 6 2 7 20 0+8 0+7 9 9 21 25 23 35 28 25 0 1 2 3 4 5 modified Rankin Scale at three months 0 20 40 60 80 100 0 20 40 60 80 % 6 modified Rankin Scale at 7 years: 0 100 11 2 32 33 4 54 5 0+2 2 1 2 4 2 6 5 2 4 2 4 2 3 9 1 5 11 7 6 2 7 20 0+8 0+7 9 9 21 25 23 35 28 25 0 1 2 3 4 5 modified Rankin Scale at three months 0 20 40 60 80 100 0 20 40 60 80 % Figure 2. Distribution of mRS at 7 years according to mRS at 3 months; left side shows the mRS for survivors at 7-years by an increasing order of mRS score and the right side shows proportion of deceased. Number of patients in squares Table 2. Distribution of status at 7 years for 3-month survivors and number needed to change to avoid one death at 7 years (NNC) Status at seven years Status at 3 months mRS 0-1 mRS 2-3 mRS 4-5 mRS 6 Personyears Mortality rate/1000 NNC mRS N % N % N % N % N 95%CI mRS 0-1 45 36.6 22 17.9 10 8.1 46 37.4 725.4 63.4 mRS 2-3 15 13.9 25 23.1 10 9.3 58 53.7 575.1 10.1 27 15-141 mRS 4-5 0 0.0 3 4.4 12 17.6 53 77.9 248.8 21.3 9 6-18 Test for linear trend = 50.1, df=1, p<0.001 For the 3-month survivors, the median survival time was 6.9 years. The proportional 7year survival for patients with 3-month mRS of 0, 1, 2, 3, 4, 5 was 56, 67, 52, 43, 29 and 14%, respectively; for the simplified mRS (0-1, 2-3, 4-5) the values were 77, 50 and 23% (figure 3). The main causes of death were vascular (43.3%); they were a recurrent stroke (20.4%), an acute MI (14.6%), a sudden death (5.1%) or other vascular causes (3.2%). Tendências na Incidência e Prognóstico do AVC - 36 - (A) No at risk (% surviving) mRS 3 months 1st year 2nd year 3rd year 4th year 5th year 6th year 7th year 0 50 (100) 47 (94) 42 (88) 40 (84) 36 (75) 33 (69) 30 (63) 27 (61) 1 76 (100) 75 (99) 71 (93) 70 (92) 68 (90) 59 (78) 53 (74) 50 (72) 2 50 (100) 48 (96) 44 (92) 42 (88) 39 (81) 34 (71) 28 (61) 24 (52) 3 64 (100) 61 (95) 50 (81) 47 (76) 41 (66) 34 (57) 29 (48) 26 (43) 4 43 (100) 39 (91) 30 (76) 26 (68) 21 (55) 18 (47) 15 (39) 11 (29) 5 30 (100) 18 (60) 14 (47) 9 (33) 7 (25) 6 (22) 5 (18) 4 (14) (B) No at risk (% surviving) mRS 3 months 1st year 2nd year 3rd year 4th year 5th year 6th year 7th year 0-1 126 (100) 122 (97) 113 (91) 110 (89) 104 (84) 92 (74) 83 (69) 77 (67) 2-3 114 (100) 109 (96) 94 (86) 89 (81) 80 (73) 68 (63) 57 (54) 50 (50) 4-5 73 (100) 57 (78) 44 (64) 35 (53) 28 (43) 24 (37) 20 (30) 15 (23) Figure 3. Long-term survival from assessment at three-month after an ischemic stroke according to (A) functional status at three months (mRS: modified Rankin Scale), (B) simplified mRS (0-1, 2-3, 4-5) Functional status three months after the first ischaemic stroke is associated with long-term outcome: data from a community-based cohort - 37 - Patients with mRS=2-3 compared to mRS=0-1 at 3 months had a higher death hazard (HR=1.61, 95% CI: 1.10-2.38) as well as those with mRS=4-5 compared to mRS=2-3 (HR=2.20, 95% CI: 1.52-3.20) (table 3). After controlling for pre-stroke mRS, patient’s profile at baseline, presence/absence of symptomatic lesion, UNSS score and OCSP classification, the death hazard for a patient with mRS=4-5 at 3 months compared to mRS=2-3 was 1.87 (95% CI: 1.18-2.95). Age, male gender, diabetes, current smoking, pre-stroke mRS 2 and stroke type (TACI or PACI vs. LACI) are associated with shorter survival times, hypercholesterolemia with longer survivorship and stroke severity (UNSS) is not associated with long-term survival. Table 3. Cox’s regression models estimates of the hazard ratios for death at seven years according to baseline characteristics for patients alive at three months after a first-everin-a-lifetime ischemic stroke Univariate Multivariate Characteristics HR 95%CI P value HR 95%CI P value Men vs. Women 1.29 0.94-1.76 0.1 1.87 1.27-2.74 0.001 Age (10 years) 2.14 1.78-2.58 <0.001 2.18 1.78-2.67 <0.001 Risk factors (yes vs. no) Hypertension 0.79 0.57-1.08 0.1 1.05 0.75-1.47 0.8 Diabetes 1.29 0.91-1.84 0.2 1.90 1.29-2.81 0.001 Hypercholesterolemia 0.52 0.37-0.73 <0.001 0.56 0.38.0.82 0.003 Atrial fibrillation 1.86 1.22-2.84 0.004 0.88 0.54-1.41 0.6 Previous TIA 0.68 0.37-1.26 0.2 1.09 0.56-2.10 0.8 Current smoker 1.59 1.07-2.36 0.021 2.47 1.53-4.00 <0.001 Pré-stroke mRS: 2-5 vs. 0-1 2.87 2.00-4.10 <0.001 2.01 1.31-3.08 0.001 Symptomatic lesion on CT 1.27 0.93-1.74 0.1 1.18 0.84-1.66 0.4 UNSS (for 1 point less) 1.04 1.02-1.06 <0.001 0.98 0.95-1.02 0.4 OCSP Subtype (vs. LACI) (<0.001) (0.011) TACI 2.51 1.66-3.81 <0.001 2.16 1.14-4.11 0.019 PACI 2.22 1.47-3.34 <0.001 1.88 1.21-2.92 0.005 POCI 1.05 0.65-1.69 0.8 0.95 0.57-1.16 0.9 3-month mRS (<0.001) (0.008) 2-3 vs. 0-1 1.61 1.10-2.38 0.015 1.19 0.77-1.84 0.4 4-5 vs. 2-3 2.20 1.52-3.20 <0.001 1.87 1.18-2.95 0.008 For one extra 7-year survivor, a change in the 3-month mRS from 5 to 4 would be needed in 6 (95% CI: 3-19) patients; using the simplified mRS scores the NNC=9 (95% CI: 6-18) in the 3-month mRS from 4-5 to 2-3 or NNC=27 (95% CI: 15-141) in mRS from 2-3 to 0-1 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com Original Paper Cerebrovasc Dis 2011;32:542–551 DOI: 10.1159/000331473 Are Stroke Occurrence and Outcome Related to Weather Parameters? Results from a Population-Based Study in Northern Portugal Rui Magalhães a M. Carolina Silva a Manuel Correia a, b Trevor Bailey c a UNIFAI, Instituto de Ciências Biomédicas de Abel Salazar (ICBAS), Universidade do Porto, e b Serviço de Neurologia, Centro Hospitalar do Porto – Hospital de Santo António, Porto , Portugal; c School of Engineering, Computing and Mathematics, University of Exeter, Exeter , UK patients registered, 19.6% had a primary intracerebral haemorrhage (PICH) and 75.3% an IS. Among patients with IS, 21.6% were total anterior circulation infarcts (TACIs), 19.8% partial anterior circulation infarcts (PACIs), 19.5% posterior circulation infarcts (POCIs) and 39.1% were lacunar infarcts (LACIs). The aetiology of IS was large artery atherosclerosis in 6.9%, cardioembolism in 23.3% and small artery occlusion in 35.6%. The incidence of PICH increased by 11.8% (95% CI: 3.8–20.4%) for each degree drop in the diurnal temperature range in the preceding day. The incidence of IS increased by 3.9% (95% CI: 1.6–6.3%) and cardioembolic IS by 5.0% (95% CI: 0.2–10.1%) for a 1 ° C drop in minimum temperature in the same hazard period. The incidence of TACIs followed the IS pattern while for PACIs and POCIs there were stronger effects of longer hazard periods and no association was found for LACIs. The relative risk of a fatal versus a non-fatal stroke increased by 15.5% (95% CI: 6.1–25.4%) for a 1 ° C drop in maximum temperature over the previous day. Conclusions: Outdoor temperature and related meteorological parameters are associated with stroke occurrence and severity. The different hazard periods for temperature effects and the absence of association with LACIs may explain the heterogeneous effects of weather on stroke occurrence found in community-based and hospital admission studies. Emergency services should be aware that specific weather conditions are more likely to prompt calls for more severe strokes. Copyright © 2011 S. Karger AG, Basel Key Words Epidemiology of stroke ⴢ Seasonal variation ⴢ Poisson model ⴢ Stroke incidence ⴢ Weather Abstract Background: Changes in meteorological parameters have been associated with cardiovascular mortality and stroke. The high incidence of stroke in Portugal may be modelled by shortor long-term weather changes whose effect may be different across stroke types and severity. Methods: Data include all patients with a first-ever-in-a-lifetime stroke registered in a population of 86,023 residents in the city of Porto from October 1998 to September 2000. Specific stroke types were considered and ischaemic stroke (IS) subtype was defined according to the Oxfordshire Community Stroke Projet classification and the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria. Information on daily temperature, humidity and air pressure was obtained from the National Meteorological Office. The Poisson distribution was used to model the daily number of events as a function of each weather parameter measured over different hazard periods, and the binomial model to contrast effects across subgroups. Differential effects of meteorological parameters and hazard periods upon stroke occurrence and outcome were analysed in a stepwise model. Results: Among the 462 Received: March 29, 2011 Accepted: August 3, 2011 Published online: November 18, 2011 Rui Magalhães Instituto de Ciências Biomédicas de Abel Salazar (ICBAS) Largo Professor Abel Salazar, 2 PT–4099-003 Porto (Portugal) Tel. +351 22 206 22 77, E-Mail rmag  @  icbas.up.pt © 2011 S. Karger AG, Basel 1015–9770/11/0326–0542$38.00/0 Accessible online at: www.karger.com/ced Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Weather and Stroke in a Mild-Climate Country Cerebrovasc Dis 2011;32:542–551 543 Introduction The association between temperature and mortality from all causes was described in a cross-country European study, showing that Portugal has the highest coefficient of seasonal variation in mortality among 14 countries despite having the highest mean winter temperature (–3.5 to 13.5  °  C) [1] . Deaths from acute myocardial and cerebral infarction contribute to the excess winter mortality [2–4] , but routine mortality statistics may over-report the number of stroke deaths [5, 6] confounded by poststroke complications. On the other hand, the incidence of stroke in most countries has a seasonal pattern, peaking during winter [7–9] , with a lesser frequency in spring [10, 11] , autumn [12] or summer months [13] . These seasonal/ monthly effects are important to adopt preventive measures and to estimate the overall hospital and/or stroke units’ workload, but fell short of demonstrating meteorological factors underlying and triggering stroke occurrence, in particular a first-in-a-lifetime stroke. Irrespective of a seasonal effect, an association between weather parameters and hospital admissions for stroke was found in several studies [14–18] , pointing out the short-term effects of temperature, atmospheric pressure as well as their short-term variations (24–48 h). In prospective community-based incidence studies, heterogeneous results concerning the effect of outdoor temperature on the incidence of stroke have been reported [12, 19, 20] . The high incidence of stroke in Portugal [5] may in part be explained by exposure to aggressive meteorological conditions. Moreover, it has been suggested that mortality increases to a greater extent during falls in temperature in regions with warm winters and in populations with cooler homes [21] . Portugal, and in particular the city of Porto, with its maritime climate and traditionally unheated homes, represents a ‘natural experimental environment’ to test whether outdoor temperature or other meteorological parameters are associated with the occurrence of stroke. Since different aetiological mechanisms/risk profiles are present in specific stroke types, we may hypothesize that these effects will be different according to stroke type and severity. Moreover, the effects of each parameter may be different according to the hazard period. Materials and Methods Identification and Classification of Stroke Patients All first-ever-in-a-lifetime strokes registered in a population of 86,023 residents in the city of Porto between October 1998 and September 2000 were included. Case ascertainment methods included direct referrals by general practitioners and hospital admissions as well as routine checking contacts with nursing homes, private hospitals/practices and review of death certificates/autopsy findings. Details on methods for identification of patients have been provided elsewhere [5] . All patients were examined by neurologists, and CT scans were performed; for those who died soon after the event or were identified by death certificates, information was given by relatives or an eye witness. Stroke was defined according to the WHO as ‘rapidly developing clinical symptoms and/or signs of focal, and at times global loss of cerebral function (patients in deep coma or cerebral haemorrhage), with symptoms lasting more than 24 h or leading to death, with no apparent cause other than of vascular origin’ [22] . Pathological types of stroke were defined according to Sudlow and Warlow [23] and the ischaemic stroke (IS) subtype by the Oxfordshire Community Stroke Project (OCSP) classification [24] and the aetiology according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria [25] . For patients identified away from the event onset, classification relied primarily on medical record documentation. A stroke was considered to be fatal if death occurred within 28 days. Meteorological Characteristics and Data The city of Porto is situated on the right bank and very close to the mouth of the river Douro, in northern Portugal, and had a population of about 250,000 in the 2001 Census, comprising 15 administrative divisions. The study evolved in 10 of these divisions, the catchment area of the Hospital de Santo António. Porto has a Mediterranean climate, with warm dry summers and mild rainy winters, but unlike the coastal south of Europe, it is often windy and usually cooler in winter with rainy weather for long periods. Daily maximum (T max ) and minimum temperature (T min ) (°  C), relative humidity (%), atmospheric pressure at sea level (hPa), and total precipitation (mm/m 2 ) at the Observatório da Serra do Pilar were obtained from the National Meteorological Office. This weather station provides data representative of the southeast Porto catchment area. Besides the crude values of meteorological data available we have also considered temperature variations within 1 day, i.e. diurnal temperature range (DTR), since it has been described to be associated with acute stroke death [26] . Statistical Analysis The description of stroke types and IS subtypes includes the daily mean of events and the respective 95% confidence interval (95% CI), stratified by season. Case fatality by season is also described. The Poisson distribution was used to model the daily number of events as a function of each weather parameter using a log-link function [27] . Generalized additive Poisson models were used alternatively, to check whether explanatory variables should enter the model as linear terms or smoothed functions with varying degrees of freedom (d.f.), enabling some flexibility in the shape of the function describing the relation. For ascertaining the effect of meteorological variables on daily stroke occurrence, a series of models were considered for exposure at different hazard periods before the event – the previous 24-hour value and the averaged values for the previous 7and 14-day periods. Using a stepwise procedure, the final model for the specific event was built, considering as predictors the values for the three periods, besides the seasonal effect. Finally, the binomial distribution (logistic model) was used to es timate the relative risk of a fatal stroke Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Magalhães  /Silva  /Correia  /Bailey  Cerebrovasc Dis 2011;32:542–551 544 under different weather conditions. The effects (General Linear Model coefficients) are presented as the rate ratio for a unit drop in the temperature parameters (T max , T min , DTR) and unit increase in the other parameters for Poisson models and as the odds ratio for binomial models, with the respective 95% CI. The likelihood ratio ␹ 2 was used for comparing the fitted models against the intercept-only model. All analyses were done with the PASW Statistics 17.0 and by R statistical software Version 2.8.1. For statistical tests, a value ! 0.05 was used to indicate a significant association. For all analyses, we assumed a stable population at risk throughout the study period and so no adjustment for deaths and births were made. R e s u l t s Patients’ Characteristics Based on a detailed clinical examination, including CT scans (97.4% of the patients and 83.5% within the first 48 h) and/or autopsy or lumbar puncture findings, a total of 462 patients were diagnosed with a first stroke. The stroke was fatal in 78 patients (16.9%) and 83.3% of them were 6 65 years old. The distribution according to pathological types and subtypes is described in table1 . Patients with primary intracerebral haemorrhage (PICH) were younger than those with IS ( ␹ 2 = 7.5, d.f. = 1, p = 0.006) and those with a lacunar (LACI) or posterior circulation infarct (POCI) were younger than those with total circulation (TACI) or partial circulation infarct (PACI) ( ␹ 2 = 22.9, d.f. = 3, p ! 0.001). TACIs were more frequent in women compared with the remainder ( ␹ 2 = 16.4, d.f. = 3, p ! 0.001) and cardioembolism was more frequent in the oldest ( ␹ 2 = 24.8, d.f. = 3, p ! 0.001). Seasonal Patterns in Weather Parameters and the Incidence of Stroke Weather parameters varied across seasons as expected, low values of T max , T min and relative humidity and high values of atmospheric pressure in winter contrasting with high temperatures in summer, high relative humidity in summer and autumn and low atmospheric pressure in summer and spring ( table2 ). Precipitation attains the highest values in spring and autumn. There was no seasonality in the overall number of strokes, PICH or IS in spite of the increasing trend from summer to spring in the number of POCI ( ␹ 2 = 8.8, d.f. = 3, p ! 0.04). There was seasonality in case fatality from PICH, higher in summer (62.5%) compared to the remaining seasons ( ␹ 2 = 8.0, d.f. = 3, p ! 0.05). The overall trend in PICH, IS and the incidence of fatal stroke throughout the study period are shown in figure 1 together with the values of the meteorological parameters. T max and T min (not shown) were highly correlated (r = 0.78) as well as relative humidity and DTR (r = –0.63); DTR increased with T max (r = 0.38) and decreased with precipitation (r = –0.45), while relative humidity increased with precipitation (r = 0.31). The Table 1. Distribution of patient characteristics and vascular risk factors by types and subtypes of IS Diagnosis/prognosis All Age ≥65 years W omen n% n% n% All strokes 462 331 71.6 287 62.1 Fatal stroke178 16.9 65 83.3 54 69.2 Pathological types PICH 78 16.9 46 59.0 48 61.5 IS 348 75.3 259 74.4 215 61.8 OCSP classification TACI 75 21.6 66 88.0 59 78.7 PACI 69 19.8 60 87.0 47 68.1 LACI 136 39.1 86 63.2 74 54.4 POCI 68 19.5 47 69.1 35 51.5 TOAST criteria Large artery atherosclerosis 24 6.9 18 75.0 14 58.3 Cardioembolism 81 23.3 76 93.8 56 69.1 Small artery occlusion 124 35.6 78 62.9 69 55.6 Other determined/undetermined 119 34.2 87 73.1 76 63.9 1 D eath within 28 days from onset. Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Weather and Stroke in a Mild-Climate Country Cerebrovasc Dis 2011;32:542–551 545 relative peaks in PICH ( fig.1 ) and fatal strokes closely followed the constant drop and relative trough in T max and DTR. Meteorological Parameters and the Incidence and Outcome of Stroke The incidence of PICH was associated with DTR and precipitation; for 1  °  C drop in DTR over the preceding 24 h, the incidence increased by 11.8%, and for each millimetre of precipitation it increased by 3.1%, reaching 5.7% (95% CI: 0.7–11.1%) when considering the average 14-day period ( table3 ). The incidence of IS, on the other hand, was associated with both Tmax and T min for the three hazard periods considered; for a 1  °  C drop in temperature, the incidence increased between 3.3 and 4.3%. There were nevertheless different hazard periods for the effects of T max and T min according to IS subtype; the incidence of TACI increased by 5.9% for a 1  °  C drop in T max over the preceding 24 h, the incidence of PACI increased by 6.6% after a 1  °  C drop in T min over the previous 24 h or 7-day period and that of POCI increased between 5.8 and 7.4% when T max /T min drops over different hazard periods. The incidence of LACI was not associated with any meteorological parameter. According to aetiology, only the incidence of cardioembolic IS increased by 5.0% (95% CI: 0.2–10.1%) for a 1  °  C drop in T min . In the stepwise models, the most important predictors of PICH and IS were DTR and T min in the previous 24 h, respectively ( table4 ). A 24hour short-term effect of T min and relative humidity was only associated with the incidence of TACI, while for PACI and POCI only the average 7/14 days effect of DTR, relative humidity and T max were included in the model. Despite the effects of relative humidity in the incidence of LACI, the fit was no better than for the intercept-only model. Using the binomial model, the odds of a fatal versus non-fatal stroke increased by 15.5% (6.1–25.4%) after a 24-hour drop in T max , and no significant differences were found across age and gender for IS. The contrasting short-term effects of DTR and T min on the incidence of PICH and IS are displayed in figure 2 a, b, and the effects of T max on fatal and non-fatal strokes are displayed in figure 2 c. Table 2. Description of weather parameters, incident events and case fatality by season Weather and population characteristics All (n = 731) Summer (n = 188) Autumn (n = 178) Winter (n = 181) S pring (n = 184) mean 95% CI mean 95% CI mean 95% CI mean 95% CI mean 95% CI Weather parameters Temperature, ° C Maximum 19.0 18.6–19.3 23.7 23.2–24.1 17.7 17.2–18.2 15.5 14.9–16.0 18.9 18.3–19.5 Minimum 10.3 10.0–10.6 14.7 14.4–15.0 9.1 8.5–9.7 6.2 5.7–6.7 10.9 10.4–11.4 Diurnal range 8.7 8.5–8.9 8.9 8.5–9.4 8.6 8.2–9.0 9.3 8.8–9.7 8.0 7.5–8.4 Relative humidity, % 77.0 76.2–77.9 78.1 76.6–79.5 78.8 77.3–80.3 74.2 72.2–76.2 77.1 75.6–78.6 Precipitation, mm/m23.1 2.6–3.7 2.0 1.1–2.9 3.6 2.4–4.8 2.1 1.4–2.9 4.6 3.4–6.1 Rainy days, % 43.6 40.0–47.2 20.7 14.9–26.6 55.6 48.8–63.5 43.6 36.4–50.9 54.9 47.6–62.2 Atmospheric pressure (above 1,000 hPa) 19.4 18.9–19.9 16.3 15.8–16.9 21.5 20.5–22.5 23.8 22.8–24.9 16.2 15.3–17.1 Daily incident events1 All strokes 7.3 6.6–8.0 5.9 4.8–7.2 7.4 6.0–8.8 7.9 6.5–9.3 8.1 6.7–9.5 PICH 1.3 0.9–1.5 1.1 7.6–1.6 1.3 0.8–2.0 1.2 0.7–1.9 1.5 0.9–2.2 IS 5.6 5.0–6.2 4.3 3.4–5.5 5.8 4.7–7.2 5.9 4.8–7.3 6.0 4.9–7.4 LACI 2.2 1.7–2.6 2.1 1.4–2.9 2.0 1.4–2.9 2.2 1.5–3.0 2.4 1.7–3.3 TACI 1.2 0.9–1.5 0.8 0.7–1.9 1.7 1.2–2.6 0.9 0.6–1.6 1.2 0.7–1.9 PACI 1.1 0.8–1.4 0.8 0.5–1.4 1.2 0.7–1.7 1.5 0.9–2.2 0.9 0.6–1.5 POCI 1.1 0.8–1.4 0.6 0.2–1.1 0.9 0.5–1.5 1.4 0.8–2.1 1.5 0.9–2.2 Case-fatality, % All strokes 16.9 13.5–20.3 21.9 13.6–30.1 14.0 7.7–20.4 14.6 8.4–20.9 17.8 11.2–24.4 PICH 33.3 22.9–43.8 62.5 38.8–86.2 30.0 9.9–50.1 22.2 3.0–41.4 25.0 7.7–42.3 IS 10.9 7.6–14.2 8.6 2.0–15.1 9.0 3.0–14.9 9.7 3.7–15.7 15.6 8.4–22.9 1 I ncidence per 1,000,000 population. Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Magalhães  /Silva  /Correia  /Bailey  Cerebrovasc Dis 2011;32:542–551 546 Discussion This study shows that the high incidence of a firstever-in-a-lifetime stroke in Portugal may be related to adverse effects of meteorological conditions. Moreover, irrespective of seasonal variations, these effects varied according to pathological type. The incidence of PICH was associated with precipitation and DTR while the incidence of IS was associated with temperature (T max /T min ) and relative humidity. Moreover, the relative importance of the hazard period was associated with stroke severity and the OCSP classification. The predominant effects after a 24-hour hazard period were observed in the incidence of PICH, TACI, cardioembolic IS and fatal strokes. The effect of DTR and relative humidity was more important after a 7-day hazard period for PACI and after a 140 70 30 25 20 15 10 5 10 5 0 85 80 75 1,030 1,025 1,020 1,015 1,010 0 Autumn Winter Spring Summer Autumn Winter Spring Summer Daily mean events/1,000,000 1 2 3 4 5 6 7 8 9 10 IS PICH Fatal strokes Tmax DTR Precipitation (mm/m2) Pressure (hPa)Humidity (%)Temperature (°C) Fig. 1. LOWESS smoothed values of daily incidence of events (per million per day) and meteorological parameters (Tmax, DTR, relative humidity, atmospheric pressure and precipitation) at Porto during the study period. Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Weather and Stroke in a Mild-Climate Country Cerebrovasc Dis 2011;32:542–551 547 day period for POCI. No association was found for LACI, irrespective of the hazard period. Despite an overall increase from summer to spring in the incidence of stroke, there was no evidence for a seasonal effect in our region, either for PICH or for IS. This pattern of variation has been previously described in other population-based studies undertaken in England, Italy, France and Russia [12, 19, 20, 28] , while most studies based on registers of hospital admissions found evidence of seasonality [10, 11, 14] . This may reflect the fact of being ‘community-based’ thus including events, some of them reported by general practitioners, that otherwise would be excluded. Moreover, they report only associations for a first-ever-in-a-lifetime event, usually with low proportions of severe cases compared to hospital admission studies or emergency transport events, the latter also being more subject to misclassification bias [29–31] . The lack of seasonality in community-based studies may also result from the different seasonal effects on PICH and OCSP subtypes and their case mix in different populations. The incidence of PICH and TACI, peaking in spring and autumn compared to summer, points to the apparently steepest variation in Tmax/T min in these seasons whilst the incidence of PACI peaks in autumn and winter, pointing to possibly less acute effects of temperature. In addition, there appears to be a seasonal pattern in POCI and complete absence of seasonality in LACI, which represents as much as 39% of IS in this study. Table 3. Association between incident stroke events and meteorological parameters according to exposure period Weather parameter/ hazard period PICH IS all TACI PACI LACI P OCI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI Season Autumn 1.32 0.68–2.55 1.34*0.98–1.84 2.04†1.07–3.88 1.28 0.63–2.60 0.99 0.61–1.62 1.64 0.71–3.80 Winter 1.17 0.60–2.29 1.38†1.01–1.88 1.11 0.54–2.31 1.71 0.88–3.32 1.07 0.66–1.73 2.42†1.11–5.29 Spring 1.53 0.81–2.89 1.40†1.03–1.91 1.39 0.70–2.77 1.10 0.53–2.27 1.18 0.74–1.88 2.73†1.27–5.86 Temperature, f 1° C Maximum Previous 24 h 1.05*1.00–1.10 1.03‡1.01–1.06 1.06†1.01–1.12 1.03 0.98–1.08 1.02 0.98–1.06 1.04 0.98–1.09 Mean 7 days 1.04 0.98–1.10 1.04‡1.01–1.07 1.04 0.98–1.10 1.04 0.98–1.11 1.02 0.98–1.07 1.07†1.01–1.14 Mean 14 days 1.04 0.99–1.11 1.04‡1.01–1.07 1.03 0.97–1.10 1.06*1.00–1.13 1.02 0.98–1.07 1.07†1.00–1.14 Minimum Previous 24 h 1.00 0.95–1.05 1.04§1.02–1.06 1.04 0.99–1.09 1.07†1.01–1.12 1.02 0.98–1.05 1.06†1.01–1.11 Mean 7 days 1.02 0.96–1.08 1.04‡1.01–1.07 1.02 0.97–1.08 1.07†1.01–1.13 1.02 0.98–1.06 1.07†1.01–1.14 Mean 14 days 1.03 0.98–1.09 1.04‡1.02–1.07 1.03 0.97–1.09 1.06*1.00–1.13 1.03 0.99–1.07 1.07†1.00–1.13 Diurnal range Previous 24 h 1.12‡1.04–1.20 0.99 0.95–1.02 1.04 0.96–1.12 0.92†0.85–1.00 1.01 0.96–1.07 0.96 0.88–1.03 Mean 7 days 1.10 0.98–1.23 1.00 0.94–1.05 1.06 0.94–1.20 0.89*0.78–1.01 1.02 0.94–1.12 0.98 0.86–1.11 Mean 14 days 1.07 0.92–1.24 0.98 0.91–1.05 1.02 0.88–1.19 0.97 0.83–1.14 0.95 0.84–1.06 0.99 0.84–1.16 Relative humidity, % Previous 24 h 1.02 0.99–1.04 1.00 0.99–1.01 1.03†1.00–1.05 1.00 0.98–1.02 1.00 0.98–1.01 1.00 0.98–1.02 Mean 7 days 1.01 0.98–1.04 1.01 1.00–1.02 1.01 0.98–1.04 1.01 0.98–1.04 1.01 0.98–1.03 1.01 0.98–1.05 Mean 14 days 1.00 0.96–1.04 1.00 0.99–1.02 1.01 0.97–1.05 1.02 0.98–1.06 0.99 0.96–1.01 1.02 0.98–1.06 Precipitation, mm/m2 Previous 24 h 1.03‡1.01–1.05 1.00 0.99–1.02 0.99 0.96–1.03 1.00 0.97–1.03 1.00 0.98–1.02 1.02 1.00–1.05 Mean 7 days 1.04*1.00–1.08 1.00 0.98–1.03 1.01 0.97–1.06 0.97 0.91–1.03 1.01 0.97–1.04 1.02 0.97–1.07 Mean 14 days 1.06†1.01–1.11 1.00 0.97–1.03 1.02 0.96–1.08 0.97 0.91–1.04 0.98 0.93–1.03 1.03 0.97–1.09 Atmospheric pressure, hPa Previous 24 h 1.00 0.97–1.03 1.01 0.99–1.02 1.02 0.98–1.05 1.02 0.98–1.05 1.00 0.98–1.02 1.01 0.97–1.04 Mean 7 days 1.00 0.97–1.04 1.01 0.99–1.03 1.01 0.98–1.05 1.03 0.99–1.08 1.00 0.98–1.02 1.02 0.98–1.06 Mean 14 days 0.99 0.97–1.01 1.01 0.99–1.02 1.00 0.98–1.02 1.04*0.99–1.08 1.00 0.98–1.02 1.03 0.99–1.07 * p < 0.1, † p < 0.05, ‡ p < 0.01, § p < 0.001, otherwise p > 0.1; f = decrease; RR = rate ratio Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Magalhães  /Silva  /Correia  /Bailey  Cerebrovasc Dis 2011;32:542–551 548 It was hypothesized that the effect of meteorological parameters would be evidenced in specific stroke types since the prevalence of different risk factors varies according to aetiology and clinical type [32, 33] . Despite the reduced number of events, PICH and TACI are more closely related to temperature in shorter hazard periods. Since their prognosis is worst, it is not surprising that a cold diurnal temperature (T max ) is associated with the likelihood of a fatal stroke. The fact that T min (nocturnal) rather than T max is a better predictor of overall and in particular cardioembolic IS, the incidence of IS may be related to the already reported circadian rhythm of IS, peaking in the morning and closely following the morning surge in blood pressure [34, 35] . Other factors that can trigger a stroke after cold exposure, such as activation of coagulation-related factors [36] , haemoconcentration and increased blood viscosity [37] may have a greater impact in cardioembolic IS, the more frequent aetiological mechanism of TACI and PACI. On the other hand, the incidence of PICH is consistently associated with precipitation, increasing with the hazard period (24 h, 7 and 14 days). This is the empirical evidence of a fact already mentioned by neurologists at emergency departments all over the country – why do we see more strokes on rainy days? Our data confirm this hypothesis, but when considering the effect of all-weather parameters, DTR in the previous day seems to explain this effect. This triggering effect of drops in DTR, characteristic of spring months might be explained by the concurrence of other atmo0 0510 DTR, previous 24 h (°C) a 15 20 50 100 150 200 250 300 0 –5 0 5 10 Tmin, previous 24 h (°C)b 15 2520 50 100 150 200 250 300 0 52510 Tmax, previous 24 h (°C)c 15 4020 3530 50 100 150 200 250 300 Daily mean (%) Fig. 2. Fitted number of PICH (black) and IS (grey) (scaled to be a percentage of mean daily strokes) plotted against DTR ( a ) and Tmin ( b ); fitted number of fatal (black) and non-fatal (grey) strokes plotted against Tmax ( c ). (Poisson models for the specific events). Dashed lines represent 95% CI. Table 4. Association between incident stroke events and meteorological parameters according to exposure period (multiple-regression models) Stroke type/weather parameter RR 95% CI p PICH (0.003)1 DTR – previous 24 h 1.12 1.04–1.20 0.003 IS (0.001) Tmin – previous 24 h 1.04 1.02–1.06 0.001 IS subtype TACI (0.011) Tmin – previous 24 h 1.05 1.00–1.11 0.045 RH – previous 24 h 1.03 1.01–1.06 0.018 PACI (0.006) DTR – mean 7 days 0.78 0.66–0.92 0.003 RH – mean 7 days 1.05 1.00–1.10 0.034 Tmax – mean 14 days 1.07 1.00–1.14 0.037 LACI (0.061) RH – mean 7 days 1.04 1.00–1.08 0.029 RH – mean 14 days 0.95 0.91–0.99 0.018 POCI2(0.016) DTR – mean 14 days 0.80 0.64–1.00 0.045 RH – mean 14 days 1.06 1.00–1.12 0.037 Fatal stroke2(0.001) Tmax – previous 24 h 1.17 1.08–1.26 0.0001 Nonfatal stroke2(0.004) RH – mean 7 days 1.02 1.01–1.04 0.010 DTR – mean 14 days 0.88 0.81–0.96 0.005 F or temperature (Tmax, Tmin, DTR), the coefficients are for a 1° C drop. RH = Relative humidity; RR = rate ratio. 1 (value of p for the overall model). 2 Coefficients adjusted for seasonal effect. Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Weather and Stroke in a Mild-Climate Country Cerebrovasc Dis 2011;32:542–551 549 spheric parameters since they coincide with rises in precipitation, relative humidity and steepest drops in T max (inversely correlated with DTR) and atmospheric pressure ( fig.1 ). An inverse relationship with atmospheric pressure was found in Siberia [19] as well as an increase in the incidence associated with a mild ambient temperature. In our study, this effect was found when T max approached T min , characterizing spring time with cold days but not so cold nights. These extreme climate features were apparent in the second spring of the study period and not so marked in the first year, a pattern also found in the incidence of PICH. When dealing with weather effects on health events, it is important to look for unusual combinations of meteorological parameters and sometimes characterize the days based on these combinations, as in the study undertaken in Israel [29] . These triggering exposures were rare and usually concentrated, and their effects might be detected by an adequate data smoothing of events as shown in figure 1 . They may be overlooked when dealing with extended time series data. Indeed, this fact might explain why contradictory findings on the effects of meteorological parameters have been reported in several studies, most of them finding an inverse relation between incidence and temperature, others a direct relation [15, 31] and both direct and indirect relations across different regions within the country [30] . The right answer is probably given by an Australian study [28] reporting an increase in incidence for extreme temperature values. Until now, few community-based studies have examined the relation between first-ever-in-a-lifetime stroke and weather parameters. In England [20] and Italy [28] , only the incidence of specific events increased with falls in temperature, PICH and fatal stroke, respectively, while in Russia [19] the relation was present for both PICH and IS. From previous reports, we know that the incidence of first-ever stroke is higher in Russia and Portugal compared to England. Population characteristics, mainly the endemic level of vascular risk factors already related to weather changes [38, 39] , as well as housing and environmental features, may explain these contradictory findings. Hypertension has a high prevalence in Portugal linked to the excess salt intake, i.e. almost twice as high as that recommended by the WHO [40] as well as a high prevalence of vascular risk factors in general [41] . Therefore, extreme values in environmental temperature, either during daytime or night-time would more likely trigger cardiovascular events. On the other hand, the surrounding conditions for the effect of temperature to be felt cannot be disregarded. The Eurowinter Group, with data from Finland, Germany, the Netherlands, the UK, Italy and Greece [21] , has shown that high indices of coldrelated mortality were associated with high mean winter temperatures, low living-room temperatures, limited bedroom heating, a low proportion of people wearing protective clothes and inactivity. In the region of the city under study, i.e. the old part of the city near the river bank, most of these conditions prevail and thus we may conclude that in Porto we had conditions for an almost ‘experimental’ environmental study, excluding in general possible effect modifiers such as eating/conditioning systems. The major limitation of this study is the reduced number of events, especially when the analysis involves stratification by IS subtypes or overall case fatality. However, the reduced study power for comparing incident strokes according to the OCSP or TOAST classification using the binomial model (that excludes days with no events), had no influence on our conclusion that the effects of cold days on outcome (fatal and non-fatal events) are different. The conclusions of most studies rely on subgroup analysis and although they concluded that there was an effect, some of them did not attempt to verify whether subgroups behave differently [33] . Another important analysis would be to investigate subgroups according to circumstances associated with the onset and time to maximum deficit, as has already been done for myocardial infarction [42] . We have also assumed that meteorological conditions are homogeneous in the study area and the risk to be similar across different environments and circumstances, which might not be true. The data specification included the day when stroke occurred without specifying the hour of the day, and so the 24-hour exposure to meteorological parameters also refers to the preceding calendar day. This means that the value of T min (usually during night-time) might by more distant from the event onset than T max (usually during daytime). It would be rather difficult to know the ‘exact’ event time since patients might not remember the exact time and more importantly this is impossible when symptoms are felt when awakening. Our results point out two major conclusions: stroke type and IS subtypes must be considered when studying the effects of weather on incidence, confirming and corroborating the different aetiological mechanisms of stroke. Moreover, it is not only exposure (intensity) that matters, but also the hazard period involved. Since the trigger effect is associated with severity/outcome, emergency services (either dial emergency number or hospital emergency departments) should be aware that specific Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Magalhães  /Silva  /Correia  /Bailey  Cerebrovasc Dis 2011;32:542–551 550 weather conditions are more likely to prompt calls for more severe strokes. Further studies with larger data sets involving time trends may be useful to show whether the effects remain after all recent developments in stroke prevention and treatment. Acknowledgments This work was supported by grants from FEDER/FCT POCTI/SAU-ESP/54885/2004 and PIC/IC/82858/2007, and the Northern Region Health Authorities. References 1 Healy JD: Excess winter mortality in Europe: a cross country analysis identifying key risk factors. J Epidemiol Community Health 2003; 57: 784–789. 2 Sheth T, Nair C, Muller J, Yusuf S: Increased winter mortality from acute myocardial infarction and stroke: the effect of age. J Am Coll Cardiol 1999; 33: 1916–1919. 3 Eng H, Mercer JB: Mortality from cardiovascular diseases and its relationship to air temperature during the winter months in Dublin and Oslo/Akershus. Int J Circumpolar Health 2000; 59: 176–181. 4 Diaz J, Garcia R, Lopez C, Linares C, Tobias A, Prieto L: Mortality impact of extreme winter temperatures. Int J Biometeorol 2005; 49: 179–183. 5 Correia M, Silva MR, Matos I, Magalhaes R, Lopes JC, Ferro JM, Silva MC: Prospective community-based study of stroke in Northern Portugal: incidence and case fatality in rural and urban populations. Stroke 2004; 35: 2048–2053. 6 Vemmos KN, Bots ML, Tsibouris PK, Zis VP, Grobbee DE, Stranjalis GS, Stamatelopoulos S: Stroke incidence and case fatality in southern Greece: the Arcadia stroke registry. Stroke 1999; 30: 363–370. 7 Wang Y, Levi CR, Attia JR, D’Este CA, Spratt N, Fisher J: Seasonal variation in stroke in the Hunter Region, Australia: a 5-year hospital-based study, 1995–2000. Stroke 2003; 34: 1144–1150. 8 Jakovljevic D, Salomaa V, Sivenius J, Tamminen M, Sarti C, Salmi K, Kaarsalo E, Narva V, Immonen-Raiha P, Torppa J, Tuomilehto J: Seasonal variation in the occurrence of stroke in a Finnish adult population. The FINMONICA Stroke Register. Finnish Monitoring Trends and Determinants in Cardiovascular Disease. Stroke 1996; 27: 1774–1779. 9 Myint PK, Vowler SL, Woodhouse PR, Redmayne O, Fulcher RA: Winter excess in hospital admissions, in-patient mortality and length of acute hospital stay in stroke: a hospital database study over six seasonal years in Norfolk, UK. Neuroepidemiology 2007; 28: 79–85. 10 Turin TC, Kita Y, Murakami Y, Rumana N, Sugihara H, Morita Y, Tomioka N, Okayama A, Nakamura Y, Abbott RD, Ueshima H: Higher stroke incidence in the spring season regardless of conventional risk factors: Takashima Stroke Registry, Japan, 1988–2001. Stroke 2008; 39: 745–752. 11 Karagiannis A, Tziomalos K, Mikhailidis DP, Semertzidis P, Kountana E, Kakafika AI, Pagourelias ED, Athyros VG: Seasonal variation in the occurrence of stroke in Northern Greece: a 10 year study in 8204 patients. Neurol Res 2010; 32: 326–331. 12 Laaidi K, Minier D, Osseby GV, Couvreur G, Besancenot JP, Moreau T, Giroud M: Seasonal variation in strokes incidence and the influence of the meteorological conditions (in French). Rev Neurol (Paris) 2004; 160: 321– 330. 13 Anlar O, Tombul T, Unal O, Kayan M: Seasonal and environmental temperature variation in the occurrence of ischemic strokes and intracerebral hemorrhages in a Turkish adult population. Int J Neurosci 2002; 112: 959–963. 14 Jimenez-Conde J, Ois A, Gomis M, Rodriguez-Campello A, Cuadrado-Godia E, Subirana I, Roquer J: Weather as a trigger of stroke. Daily meteorological factors and incidence of stroke subtypes. Cerebrovasc Dis 2008; 26: 348–354. 15 Dawson J, Weir C, Wright F, Bryden C, Aslanyan S, Lees K, Bird W, Walters M: Associations between meteorological variables and acute stroke hospital admissions in the west of Scotland. Acta Neurol Scand 2008; 117: 85–89. 16 Morabito M, Crisci A, Vallorani R, Modesti PA, Gensini GF, Orlandini S: Innovative approaches helpful to enhance knowledge on weather-related stroke events over a wide geographical area and a large population. Stroke 2011; 42: 593–600. 17 Ohshige K, Hori Y, Tochikubo O, Sugiyama M: Influence of weather on emergency transport events coded as stroke: populationbased study in Japan. Int J Biometeorol 2006; 50: 305–311. 18 Chang CL, Shipley M, Marmot M, Poulter N: Lower ambient temperature was associated with an increased risk of hospitalization for stroke and acute myocardial infarction in young women. J Clin Epidemiol 2004; 57: 749–757. 19 Feigin VL, Nikitin YP, Bots ML, Vinogradova TE, Grobbee DE: A population-based study of the associations of stroke occurrence with weather parameters in Siberia, Russia (1982–92). Eur J Neurol 2000; 7: 171– 178. 20 Rothwell PM, Wroe SJ, Slattery J, Warlow CP: Is stroke incidence related to season or temperature? The Oxfordshire Community Stroke Project. Lancet 1996; 347: 934–936. 21 Keatinge WR, Donaldson GC: Cold exposure and winter mortality from ischaemic heart disease, cerebrovascular disease, respiratory disease, and all causes in warm and cold regions of Europe. The Eurowinter Group. Lancet 1997; 349: 1341–1346. 22 Hatano S: Experience from a multicentre stroke register: a preliminary report. Bull World Health Organ 1976; 54: 541–553. 23 Sudlow CL, Warlow CP: Comparing stroke incidence worldwide: what makes studies comparable? Stroke 1996; 27: 550–558. 24 Bamford J, Sandercock P, Dennis M, Burn J, Warlow C: Classification and natural history of clinically identifiable subtypes of cerebral infarction. Lancet 1991; 337: 1521–1526. 25 Adams HP, Jr., Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EE, III: Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993; 24: 35– 41. 26 Chen G, Zhang Y, Song G, Jiang L, Zhao N, Chen B, Kan H: Is diurnal temperature range a risk factor for acute stroke death? Int J Cardiol 2007; 116: 408–409. 27 Gelman A, Hill J: Data Analysis Using Regression and Multilevel/Hierarchical Models, ed 3. Cambridge, Cambridge University Press, 2007. 28 Carolei A, Marini C, De Matteis G, Di Napoli M, Baldassarre M: Seasonal incidence of stroke. Lancet 1996; 347: 1702–1703. 29 Berginer VM, Goldsmith J, Batz U, Vardi H, Shapiro Y: Clustering of strokes in association with meteorologic factors in the Negev Desert of Israel: 1981–1983. Stroke 1989; 20: 65–69. 30 Ebi KL, Exuzides KA, Lau E, Kelsh M, Barnston A: Weather changes associated with hospitalizations for cardiovascular diseases and stroke in California, 1983–1998. Int J Biometeorol 2004; 49: 48–58. Downloaded by: 89.181.178.32 - 7/8/2013 7:21:18 PM Tendências na Incidência e Prognóstico do AVC - 60 - defined according to the World Health Organization criteria .15 In this cohort, stroke was diagnosed in patients presenting an acute vestibular syndrome associated with ‘‘other’’ central nervous signs, with or without acute symptomatic lesion on CT or magnetic resonance imaging (MRI), and ischemic stroke subtype was defined according to the Oxfordshire Community Stroke Project classification. 16 TIA was defined according to standard criteria, 14 and in this cohort, it was diagnosed in patients with monophasic episodes of VSs associated with central signs, lasting less than 24 hours. In our cohort, patients diagnosed with a peripheral vestibular disorder presented paroxystic vertigo induced by head movement, with typical nystagmus, associated with other otologic manifestations (hypoacusis or tinnitus), in the absence of central nervous signs. Besides isolated VSs other nonfocal symptoms/signs such as blurred vision, bilateral weakness, imbalance without objective ataxia, or decreased consciousness might be present. Laboratory data included a cranial CT scan and triplex scan of carotid and vertebral arteries at entry (either requested by the assistant physician or if considered necessary by the study neurologist); CT brain lesions were read by a neuroradiologist blind to clinical symptoms/signs. Based on the description of the CT, the neurologist assigned any acute lesion compatible with neurologic symptoms and signs as symptomatic; other ischemic lesions (infarctions or lacunas) were considered nonsymptomatic. Triplex scan was considered abnormal in the presence of greater than or equal to 50% stenosis or occlusion of an artery. For the presence of previous vascular risk factors (VRFs), the following definitions were considered: hypertension, previous diagnosis and/or treatment of high blood pressure (BP), systolic BP .160 mm Hg and/or diastolic BP .95 mm Hg in at least 2 different measures; diabetes mellitus, previous diagnosis and/or under treatment with oral antidiabetic/insulin, fasten glycemia.126 mg/dL, postprandial glycemia $200 mg/dL, and/or glucose tolerance test with values of glycemia $200 mg/dL at second hour; hypercholesterolemia, previous diagnosis and/or treatment, serum total cholesterol level after 12 hours of fasting $240 mg/dL; current smoker, smoker at event date or in the last 12 months; atrial fibrillation, evidence from the electrocardiogram (EKG) or registration in patient’s record; acute myocardial infarction (MI), confirmed by an increase in serum cardiac enzymes, EKG abnormalities or diagnosis confirmed by doctor at any time and angor, history of chest pain with or without superior limb pain related to physical exercise or emotion and release by rest. Throughout the follow-up period, the following major vascular events were recorded: stroke, MI (confirmed by EKG, tissular necrosis markers, or necropsy), and death of vascular etiology other than MI or stroke (suspected but not confirmed MI or heart failure, Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study - 61 - thoracic or abdominal aortic aneurysm rupture, or sudden death of presumed vascular origin). Data Analysis The description of patients included according to etiology (ischemic, peripheral, or unclassified) is presented and a logistic regression model was used to estimate the independent predictors of a stroke/TIA at onset of symptoms, considering the sociodemographic profile and relevant VRF in the univariate analysis (P < .3). After checking the assumption of proportional hazards with the Schoenfeld test, Cox models were used to estimate survival free from stroke or a vascular event (stroke, MI, or vascular death) in patients with transient symptoms (excluding those with stroke) according to etiology (ischemic, peripheral, and unclassified), VRF, and laboratory findings, adjusted for age and gender. This analysis was repeated excluding patients with TIA. Patients were censored if death or death from a nonvascular cause occurred before the respective end point. Results During the registration period, 1922 patients were reported as having FNSs. In a first step 759 were excluded, either by logistic reasons (n = 122) or after neurologic assessment, because their symptoms were nonfocal (n = 274), recurrent episodes (n = 198), including 60 who had a previous stroke, or the symptoms were because of infectious/toxic diseases, non-neurologic, or neurologic noncentral diseases (Fig 1). Among the 1163 included, 363 (31.2%) presented a VS—29 (8.0%) had a stroke (a primary intracerebral hemorrhage, 2 partial anterior circulation infarcts, 3 lacunar infarcts, and 23 posterior circulation infarcts), 31 (8.5%) had a TIA, 3 (.8%) other diagnosis (drop attack, functional disorder, and hydrocephalus), 208 (57.3%) had a pVS, and 92 (25.3%) had an uVS. Considering all first-ever incident cerebrovascular events in this population (579 strokes and 141 TIA), 5% of all patients with stroke and 22% of all patients with TIA presented/complained of VS, either isolated or concomitantly with other symptoms. Most patients were referred directly to the study center (60.0%), a higher proportion among those with stroke/TIA (75.0% vs. 57.0%; Table 1). Hospital emergency department (ED) and 24 hours HC service were the most sought (94.7%) and 60.6% of patients were observed in the first 24 hours. The mean age at event onset was 60.1 years (range, 11-93), with a higher proportion of patients with stroke/TIA 65 years of age or older (60.0% vs. 41.0%); 62.8% were women. Hypertension was the most prevalent Tendências na Incidência e Prognóstico do AVC - 62 - VRF (47.8%) followed by hypercholesterolemia (41.9%), diabetes (19.2%), angina/MI (7.8%), and atrial fibrillation (4.2%); 14.7% were current smokers. The logistic regression model indicated that age greater than 65 years (odds ratio [OR] 5 1.91; 95% CI, 1.043.54), male sex (OR 5 2.36; 95% CI, 1.28-4.35), diabetes (OR 5 2.70; 95% CI, 1.405.19), and atrial fibrillation (OR 5 6.30; 95% CI, 1.99-19.9) increased the likelihood of VS of ischemic etiology. This etiology was more frequent in patients that had VS plus FNS (80.9%) or VS plus with other nonfocal neurologic symptoms (22.2% and 22.9%), and more rare in patients with isolated VS (3.5%) or VS plus tinnitus/hypoacusis (2.9%). Assessed for eligibility (n=1922) Exclusions (n=759) Logistic reasons - Out of study period (n=12) - Out of study population (n=110) After neurological assessment - Without focal symptoms/signs (n=274) - Recurrent episode (n=198) - Other diagnosis (n=148) - Refusals (n=17) Total included (n=1163) Long-term follow-up (1) (2) (3) (4) All - Lost 2 0 2 3 7 - Died 6 9 20 21 56 - Alive - Face to face 14 17 59 148 248 - By phone 0 4 9 20 33 - Clinical records 7 1 2 6 16 With vestibular symptoms (n=363) (1) Stroke (n=29) (2) TIA (n=31) (4) Peripheral vertigo (n=208) (3) Unclassified vertigo (n=92) Others causes (n=3) Assessed for eligibility (n=1922) Assessed for eligibility (n=1922) Exclusions (n=759) Logistic reasons - Out of study period (n=12) - Out of study population (n=110) After neurological assessment - Without focal symptoms/signs (n=274) - Recurrent episode (n=198) - Other diagnosis (n=148) - Refusals (n=17) Total included (n=1163) Total included (n=1163) Long-term follow-up (1) (2) (3) (4) All - Lost 2 0 2 3 7 - Died 6 9 20 21 56 - Alive - Face to face 14 17 59 148 248 - By phone 0 4 9 20 33 - Clinical records 7 1 2 6 16 Long-term follow-up (1) (2) (3) (4) All - Lost 2 0 2 3 7 - Died 6 9 20 21 56 - Alive - Face to face 14 17 59 148 248 - By phone 0 4 9 20 33 - Clinical records 7 1 2 6 16 With vestibular symptoms (n=363) With vestibular symptoms (n=363) (1) Stroke (n=29) (1) Stroke (n=29) (2) TIA (n=31) (2) TIA (n=31) (4) Peripheral vertigo (n=208) (4) Peripheral vertigo (n=208) (3) Unclassified vertigo (n=92) (3) Unclassified vertigo (n=92) Others causes (n=3) Others causes (n=3) Figure 1. Flowchart of inclusion of patients and details of 7-year follow-up. Cranial CT/MRI was performed in 228 patients (63.3%) of whom 82 (36.0%) in the first 24 hours; ischemic lesions were detected in 28.9%of all CTs. Only 12 of the 29 patients with stroke showed newrelevant lesions; among the 31 patients with TIA, 28 had cranial CT and in 8 (28.6%) it showed ischemic nonsymptomatic lesions, proportion not significantly different from 21.6 (37 of 171) in patients with other VS. The triplex scan showed signs of stenosis in 61 patients (46.2%) of the 132 (36.7%) investigated. Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study - 63 - Table 1. Ascertainment of patients, characteristics, and diagnostic procedures Other vestibular symptom Stroke/TIA vs. others Stroke/TIA Unclassified Peripheral All All Characteristics (n=60) (n=92) (n=208) (n=300) (n=360) P value Identification, n (%) 1 st source: Direct 45 (75.0) 56 (60.9) 115 (55.3) 171 (57.0) 216 (60.0) .009 Hospital 26 (43.3) 39 (42.4) 83 (39.9) 122 (40.7) 148 (41.1) .08 Health Centre 19 (31.7) 17 (18.5) 32 (15.4) 49 (16.3) 68 (18.9) Assessment Emergency room 59 (98.3) 84 (91.3) 198 (95.2) 282 (94.0) 341 (94.7) .2 Delay from onset <24h 36 (60.0) 50 (54.3) 132 (63.5) 182 (60.7) 218 (60.6) .9 Patients characteristics, n (%) Age, mean (SD) 67.9 (11.9) 61.8 (17.9) 57.1 (15.9) 58.5 (16.7) 60.1 (16.4) Range 38-92 15-93 11-84 11-93 11-93 >65 years 36 (60.0) 45 (48.9) 78 (37.5) 123 (41.0) 159 (44.2) .007 Women 31 (51.7) 59 (64.1) 136 (65.4) 195 (65.0) 226 (62.8) .051 Hypertension 34 (56.7) 45 (48.9) 93 (44.7) 138 (46.0) 172 (47.8) .1 Hypercholesterolemia 27 (45.0) 33 (35.9) 91 (43.8) 124 (41.3) 151 (41.9) .6 Diabetes 21 (35.0) 17 (18.5) 31 (14.9) 48 (16.0) 69 (19.2) <.001 Myocardial infarction/Angina 7 (11.7) 10 (10.9) 11 (5.3) 21 (7.0) 28 (7.8) .2 Atrial fibrillation 8 (13.3) 3 (3.3) 4 (1.9) 7 (2.3) 15 (4.2) <.001 Current smoker 11 (18.3) 16 (17.4) 26 (12.5) 42 (14.0) 53 (14.7) .4 Number of risk factors*, n (%) .001 0 13 (21.7) 41 (44.6) 101 (48.6) 142 (47.3) 155 (43.1) 1 29 (48.3) 32 (34.8) 79 (38.0) 111 (37.0) 140 (38.9) 2 13 (21.7) 15 (16.3) 24 (11.5) 39 (13.0) 52 (14.4) 3+ 5 (8.3) 4 (4.3) 4 (1.9) 8 (2.7) 13 (3.6) Presenting symptoms, n (%) <.001 Vestibular symptom alone 8 (13.3) 59 (64.1) 159 (76.4) 218 (72.7) 226 (62.8) + Focal 38 (63.3) 6 (6.5) 3 (1.4) 9 (3.0) 47 (13.1) + Tinnitus/ Hypoacusis 1 (1.7) 8 (8.7) 25 (12.0) 33 (11.0) 34 (9.4) + Faintness/ Confusion 6 (10.0) 13 (14.1) 8 (3.8) 21 (7.0) 27 (7.5) + Cephalalgias 7 (11.7) 6 (6.5) 13 (6.3) 19 (6.3) 26 (7.2) CT-scan/MRI,† n (%) 57 (95.0) 52 (56.5) 119 (57.2) 171 (57.0) 228 (63.3) <.001 Before 24h 36 (63.2) 15 (28.8) 31 (26.1) 46 (26.9) 82 (36.0) With ischemic lesions 20 (55.6) 7 (46.7) 9 (29.0) 16 (36.8) 36 (43.9) .06 After 24h 21 (36.8) 37 (71.1) 88 (73.9) 125 (73.1) 146 (64.0) With ischemic lesions 9 (42.9) 10 (27.0) 11 (12.5) 21 (16.8) 30 (20.5) .006 Triplex scan, n (%) 24 (40.0) 29 (31.5) 79 (38.0) 108 (36.0) 132 (36.7) .6 Abnormal (with stenosis) 13 (54.2) 14 (48.3) 34 (43.0) 48 (44.4) 61 (46.2) .4 Medication (during 1 st year follow-up), n (%) Antiplatelet therapy 36 (60.0) 33 (35.9) 43 (20.7) 76 (25.3) 112 (31.1) <.001 Antihypertensive therapy‡ 33 (97.1) 39 (86.7) 85 (91.4) 124 (89.9) 157 (91.3) .2 Lypid-lowering therapy‡ 11 (40.7) 8 (24.3) 24 (26.4) 32 (25.4) 43 (28.5) .1 Hypoglycaemic therapy‡ 18 (85.7) 14 (82.4) 24 (77.2) 38 (79.2) 56 (81.2) .5 Multitherapy in patients with >1RF 15 (83.3) 13 (68.4) 20 (71.4) 33 (70.2) 48 (73.8) .3 Abbreviations: MRI, magnetic resonance imaging; SD, standard deviation; TIA, transient i schemic attack; VS, vestibular symptom; RF, risk factor. *includes hypertension, diabetes, angina/myocardial infarction and atrial fibrillation. †only 7 patients with MRI (2 with unclassified VS and 5 with peripheral VS). ‡calculated in patients with hypertension, hypercholesterolemia and diabetes, respectively. Tendências na Incidência e Prognóstico do AVC - 64 - During the first year follow-up 195 patients (95.1%) with VRF were under preventive vascular therapy and 41 (26.5%) of those with no previous VRF began therapy after the episode. Antiplatelet therapy was more prescribed in patients with stroke/TIA and other therapeutic agents equally likely prescribed in all patients with the specific VRF; lipidlowering therapy was seldom prescribed (Table 1). At the 7-year follow-up, 56 patients (15.6%) had died and 7 (1.9%) were lost to follow-up (Fig 1). Most patients alive at the end of follow-up were examined at the outpatient clinic by a research team neurologist (83.5%) and 33 (11.1%) were contacted by phone to inform about details of vascular events that had happened meanwhile. For all patients, clinical records were checked for relevant information. Patients with stroke/TIA had 18 (30%) vascular events: 8 recurrent strokes, 7 first-ever strokes (after a TIA), and 3 other vascular events (2 MI and 1 vascular death). Patients with uVS had 17 (18.5%) vascular events, 7 strokes, 6 MI, and 4 vascular deaths. In the pVS group, there were 23 (11.1%) vascular events, 15 strokes, 3 MI, and 5 vascular deaths. In the univariate analyses, the number of VRF, the presence of silent infarctions, and VS due to TIA shortened survival free from stroke or a vascular event (Table 2 and Fig 2). In the multivariate model including these variables, only in the presence of ischemic lesions on CT scan there was an almost 4-fold risk of stroke in the long-term follow-up (hazard rate [HR] 5 3.96; 95% CI, 1.63-9.60), whereas the longterm risk of a vascular event increased in patients with 2 or more VRFs (HR 5 2.70; 95% CI, 1.25-5.86). After excluding patients with TIA, the multivariate models yielded identical results (Table 3); patients with at least 2 VRFs were at a higher risk of stroke (HR 5 5.75; 95% CI, 1.71-19.4) or of a vascular event (HR 5 5.15; 95% CI, 2.2611.7), whereas the presence of any ischemic lesion on CT scan increased the risk of stroke (HR 5 5.75; 95% CI, 1.71-19.4). Discussion This is the first prospective community-based study to analyze the diagnosis of patients presenting with firstever-in-lifetime VS, and according to this presenting symptom, trying to understand their seriousness in a prolonged long-term follow-up. In most patients, the symptoms at presentation had peripheral etiology (57.3%), but still in 16.5% of them it was the complaining symptom of a first-ever-in-the-lifetime stroke/TIA, indicative in this population of 5% of first-ever strokes and 22% of first-ever TIA. Despite being observed and followed up by neurologists close to onset and at 3 months, in 25.3% of patients the VS remained unclassified. For this relatively high proportion of patients, some discharged with a ‘‘symptomatic diagnosis’’ and others with alternative possible diagnosis (as Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study - 65 - migraine or functional disorder) we may conclude that the presence of silent infarctions on CT and VRFs are the more important predictors of a serious vascular event, such as stroke, MI, or vascular death. Table 2. Cox proportional hazard rates (HR) for stroke and vascular events in the seven-year follow-up in patients with transient VS (n=331) Stroke Vascular events Patients characteristics and diagnostic procedures No. events ‡ Univariate Multivariate No. events ‡ Univariate Multivariate HR 95% CI HR 95% CI HR 95% CI HR 95% CI Socio-demographic Men vs. Women 12/18 1.20 0.58-2.49 1.17 0.55-2.50 22/29 1.36 0.78-2.36 1.55 0.87-2.77 Age (years)* 17/13 1.05 1.02-1.08 1.03 0.99-1.07 35/16 1.06 1.04-1.09 1.06 1.03-1.08 Risk factors (Yes vs. No) Hypertension 20/10 2.39 1.12-5.10 33/18 2.24 1.26-3.97 Hypercholesterolemia 12/18 0.92 0.44-1.91 24/27 1.25 0.72-2.16 Diabetes 9/21 2.09 0.96-4.57 15/36 2.10 1.15-3.83 Angina/Myocardial infarction 6/24 3.36 1.37-8.22 10/41 4.05 1.60-10.2 Atrial fibrillation 4/26 5.19 1.81-14.9 5/46 3.35 1.68-6.69 Current smoker 3/27 0.73 0.22-2.40 6/45 0.86 0.37-2.02 No. risk factors (vs. 0) † 1 14/6 2.94 1.13-7.64 1.89 0.71-5.05 20/12 2.12 1.03-4.33 1.42 0.69-2.93 2+ 10/6 4.85 1.76-13.4 2.43 0.80-7.35 19/12 4.91 2.38-10.1 2.70 1.25-5.86 CT-scan (vs. without ischemic lesions) Not done 8/9 1.07 0.41-2.77 1.08 0.40-2.90 16/19 1.03 0.53-2.00 0.96 0.48-1.91 With ischemic lesions 13/9 5.74 2.45-13.4 3.96 1.63-9.60 16/19 3.40 1.75-6.61 1.95 0.98-3.88 Triplex scan (vs. without stenosis) Not done 19/4 1.64 0.56-4.83 32/8 1.39 0.64-3.02 With stenosis 7/4 2.35 0.69-8.04 11/8 1.85 0.74-4.59 Diagnosis (vs. peripheral VS) Transient ischemic attack 7/15 3.74 1.52-9.17 1.79 0.64-5.06 10/23 3.60 1.71-7.57 1.49 0.65-3.43 Unclassified VS 8/15 1.31 0.56-3.10 0.87 0.36-2.11 18/23 1.99 1.07-3.68 1.36 0.72-2.57 Abbreviations: CI, confidence interval; CT, computed tomography; HR, hazard rates; VS, vestibular symptom. *Number of events according to age in 2 groups:  65, <65 years. † According to univariate analysis includes hypertension, diabetes, angina/myocardial infarction, and atrial fibrillation. ‡ Number of events of each category against reference category. The relatively high proportion of unclassified patients in this study reflects the known difficulty of managing VS patients, especially when they present a first-ever episode. Even so, we have achieved a similar proportion of definitive diagnosis as reported in other studies retrospectively based on ED visits. 17 The high incidence of stroke/TIA among patients with VS maybe justified by the fact that our study is community-based and mainly because we excluded many patients with only nonFNSs such as presyncope Tendências na Incidência e Prognóstico do AVC - 66 - or other toxic/infectious etiologies. Moreover, by including only the first-ever-in-life episode, many patients with usually benign and highly prevalent chronic or recurrent etiologies, such as paroxysmal positional vertigo or Meniere disease, were excluded. This was a more restrictive approach because we were mainly interested in symptoms that could be attributed to a vascular cause. More and less restrictive approaches may justify the wide range of central causes of vertigo/dizziness found in other studies, such as .5% using a National Health Insurance database, 18 and proportions between 3.2% and 9.2% 3,17,19-21 based on patients seen at EDs or 42% in patients seen at na emergency neurologic consultation. 22 (A) (C) (B) (D) Figure 2. Kaplan–Meier survival free from stroke and free from a vascular event according to the number of vascular risk factors (A) and (C) and computed tomography scan results (B) and (D). Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study - 67 - Table 3. Cox proportional hazard rates (HR) for stroke and vascular events in the seven-year follow-up for patients with peripheral and unclassified VS (n=300) Patients characteristics and diagnostic procedures No. events ‡ Stroke No. events ‡ Vascular events HR 95% CI HR 95% CI Socio-demographic Men vs. Women 9/14 1.21 0.50-2.92 18/23 1.68 0.88-3.20 Age (years)* 13/10 1.03 0.99-1.07 28/13 1.06 1.03-1.09 No. risk factors (vs. 0)† 1 10/4 2.10 0.64-6.89 15/9 1.49 0.64-3.45 2+ 9/4 5.75 1.71-19.4 17/9 5.15 2.26-11.7 CT-scan (vs. without ischemic lesions) Not done 8/5 1.35 0.43-4.15 14/14 0.82 0.39-1.73 With ischemic lesions 10/5 5.75 1.86-17.7 13/14 2.20 1.00-4.83 Diagnosis Unclassified vs. peripheral VS 8/15 0.76 0.31-1.87 18/23 1.32 0.70-2.50 Abbreviations: CI, confidence interval; CT , computed tomography; HR, hazard rates; VS, vestibular symptom. *Number of events according to age in 2 groups:  65, <65 years. † According to univariate analysis includes hypertension, diabetes, angina/myocardial infarction and atrial fibrillation. ‡ Number of events of each category against reference category. Considering that most patients were notified by physicians at ED, the proportion found is relatively high, reaching 3.7% (8 of 228) in patients whose complaint was isolated vertigo, again much higher than reported in patients admitted to hospital ED (.7%). 3 As it would be expected the diagnosis of stroke/TIA was more frequent among patients complaining or presenting any focal abnormality on examination (80.9%), but other presentations in which vertigo was accompanied by headache, faintness/confusion were also common in patients with stroke/TIA. Moreover, the likelihood of vertigo with central cause was higher in men and increased with age and the number of risk factors, namely atrial fibrillation and diabetes. These results corroborated previous findings in ED patients with a triage complaint of vertigo 20 and in studies comparing patients with and without vertigo. 3,18 As it might be expected, patients with TIA had a higher long-term risk of stroke or other vascular events than patients with a peripheral etiology, although after adjusting for age, gender, VRFs, and the presence of nonsymptomatic (silent) ischemic lesions on CT, the diagnosis/etiology lost importance. For patients without definite diagnosis, the long-term risk of vascular events is somewhere between those with known etiologies. Probably the recognition and description of symptoms by these patients was not so accurate to allow a better interpretation by the physician. After excluding patients with a TIA from the analysis Tendências na Incidência e Prognóstico do AVC - 68 - of long-term prognosis, the high risk of upcoming vascular events in patients with 2 or more VRFs compared with those with no risk factors suffers a remarkable increase from 2.7 to 5.2, indicating that in this reduced group VRFs are even more important. It is possible that a referral bias exists because patients with VRFs maybe more prone to attend medical care and thus to be notified by their physicians. Identical findings were reported in patients hospitalized with a principal diagnosis of vertigo (excluding central vertigo). 23 Another interesting finding is the fact that age (.65 years) is not a prognostic factor when the outcome is stroke, contrary to a vascular event. This maybe explained by the fact that patients presenting VS from central cause are a younger group compared with what is found in the general population, 13-14 and their risk profile, excluding diabetes and atrial fibrillation, is similar to that of patients with pVS or uVS. This may also explain the diagnosis difficulties and the ‘‘useless’’ of the CT scan for diagnostic purposes in patients with vertigo as remarked in other studies. 21 This inadequacy of CT scan might be even more pronounced because in some patients (36%) it was performed in the first 24 hours after the first-ever episode of VS, and it was not repeated few days later, making it difficult to identify eventual symptomatic lesions and a consequent change in diagnosis. 24 On the other hand and independently from diagnosis, CT scan results, namely, the presence of nonsymptomatic ischemic lesions of any type/location (mostly lacunas) increases significantly the long-term risk of stroke, both including (HR 5 3.96) or excluding patients with TIA (HR 5 5.75). Again, we cannot rule out a possible bias because CT/MRI might had been more likely performed in patients for whom an ischemic etiology was admitted. Nevertheless the risk of stroke/vascular event in patients without CT/MRI was similar to that of patients showing no ischemic lesions. In spite of the low short-term risk of stroke of patients with an acute VS when compared with other transient focal symptoms/signs, 25 VRF and the presence of silent ischemic lesions deserve better secondary prevention for avoiding vascular events in the long run. Our results illustrate more closely the spectrum of a first-ever VS at community level and their prognosis than other retrospective studies based only on ED visits, relying mostly on ,QWHUQDWLRQDO&ODVVLILFDWLRQRI'LVHDVHV codes and so disregarding subtle neurologic findings that might indicate a central nervous system dysfunction. 20,23,26 Even so, we may have missed episodes that do not come to medical attention, although less likely first episodes, because in the public perception whenever these symptoms are present the person seeks medical care 27 providing a good opportunity for risk factors surveillance. We recognize 2 major limitations in this study. The first is the lack of useful imaging data, namely, MRI in the acute setting or a later repeated CT scan that could have helped in the identification of TIA/stroke and eventually diminish the number of patients with Long-term Prognosis of Patients Presenting First-ever Vestibular Symptoms in a Community-based Study - 69 - undetermined etiology. The second is the lack of systematic data about medication used before and after the acute event. Even so, during the first year follow-up, a considerable number of patients were not under an adequate therapy according to the risk factors presented, particularly antiplatelet and lipid-lowering therapy. Moreover, these secondary preventive measures were more likely undertaken in patients with a TIA rather than in those with uVS, which might partially explain why the diagnosis was not an important prognostic factor. Conclusions More than 8% of patients with a first-ever stroke/TIA present to the assistant physician with a first-ever complaint of VS. About a quarter of all patients presenting with VS remain with no definitive diagnosis. However, independently of diagnosis, the prognosis is mostly dependent on the number of VRF and on the presence of old, nonsymptomatic ischemic lesions on the acute CT scan. To prevent future vascular events, we purpose that those patients with VRF and/or old asymptomatic ischemic lesions should have a vascular workup similar to those diagnosed with stroke/TIA and a strict control of their VRF. Acknowledgment: Author contributions: R.F. drafted the article and was responsible for data validation; R.M. was responsible for data management and statistical analysis; M.C. conceived and designed the community-based study and was responsible for neurologic evaluation; M.C.S. contributed to analysis and interpretation of data and critical revision of the article. All authors read and approved the final article. References 1. Sloane PD. Dizziness in primary care. Results from the National Ambulatory Medical Care Survey. J Fam Pract 1989;29:33-38. 2. Burt CW, Schappert SM. Ambulatory care visits to physician offices, hospital outpatient departments, and emergency departments: United States, 1999–2000. Vital Health Stat 2004;13:1-70. 3. Kerber KA, Brown DL, Lisabeth LD, et al. Stroke among patients with dizziness, vertigo, and imbalance in the emergency department: a population-based study. Stroke 2006;37:24842487. 4. Drachman DA, Hart CW. An approach to the dizzy patient. Neurology 1972;22:323-334. 5. Edlow JA, Newman-Toker DE, Savitz SI. Diagnosis and initial management of cerebellar infarction. Lancet Neurol 2008;7:951-964. Tendências na Incidência e Prognóstico do AVC - 76 - One decade ago Portugal had one of the highest stroke incidence rates among Western European countries, 305 per 100,000 in rural and 269 per 100,000 in urban populations, and a low case-fatality at 28 days, 14.6% in rural and 16.9% in urban areas. 1 This high incidence could explain why stroke was the leading cause of death in Portugal. However, mortality from stroke declined from 154.2 to 91.6 per 100,000 between 1999 and 2005 and from 164.2 to 93.1 in the Northern Region of Portugal, 2-3 and these changes may be associated with a decline in incidence and/or changes in survivorship. Conclusions from the WHO MONICA project on populations aged 35 to 64 years indicated that changes in stroke mortality were mainly attributable to changes in case fatality rather than in event rates, 4 but the study fell short from demonstrating that they resulted from disease severity and management. Nevertheless disability-adjusted life-years lost from stroke in Portugal had a 39.4% reduction in the 1990-2005 period, 5 indicating that changes in stroke severity might be implied. Soon after the first study was carried out, several health measures were advanced by the National Health Authorities, resulting from recognizing that stroke was the leading cause of death and hospital inpatient care calling to integrated primary, secondary and tertiary prevention actions. 6 Among others, it was purposed to intensify population-based campaigns about the disease and risk factors 7 and to identify VRF carriers using a periodical medical examination; the organization of stroke units and implementation of Stroke Code pathways 8 were the measures directed to patients. Using the updated methodology for ‘ideal’ population-based studies, 9 a second incidence study was undertaken in the northern region of Portugal. In this article we focused in urban populations living in the region of Porto, and we are mainly interested in knowing whether the reduction in mortality stems from a decrease in incidence rates, particularly a decrease in short-term stroke severity and measures implemented meanwhile. Subjects and Methods The ACIN2 (prospective community registry of Neurological Attacks) is the second population-based study in the North of Portugal for studying the incidence and outcome of stroke and transient focal neurologic attacks (TNA) which could be attributed to a dysfunction of the central nervous system. We used the criteria for epidemiological population-based stroke research proposed by Sudlow and Warlow 10 and updated by Feigin and Carter 9 for achieving complete case ascertainment. Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 77 - Study Population Following the reorganization of the Portuguese National Health Service (NHS) in 2008, the metropolitan area of Porto (Figure 1) was divided in 9 major health divisions, two of them, the Western and Eastern Porto ACES (association of health centers) for persons living in the city of Porto. The study population comprised all individuals registered in the Western Porto ACES (WPACES), involving approximately 57% of the city population residing in 12 administrative regions. This health unit aggregates five health centers (HC) and 105 family doctors, including the three HC involved in the study undertaken in 19982000, and the Centro Hospitalar do Porto, which includes the Hospital Santo António (HSA) that receives all patients from the WPACES who are referred for emergency/specialized care. The national health number (NHN) database from the WPACES was used to define the study population and served as reference to collect medical information. The NHN provides a unique patient identifier for every resident in Portugal after registration at a particular health unit/family doctor, for accessing the NHS network. H H H H 3RUWR N H Hospital HMunicipality H H H H 3RUWR H H H H 3RUWR3RUWR NN H Hospital HMunicipality H Hospital HMunicipality Figure 1. Map of the metropolitan area of Porto, showing areas included in both studies (dark shaded) and added in the second study (light shaded), and main hospitals involved Case ascertainment and follow-up To make studies comparable we used identical sources of information. (1) All family doctors were informed about the project and were asked to report, as soon as possible Tendências na Incidência e Prognóstico do AVC - 78 - after the event, any patient suspected of a stroke or TNA either by using the project website (www.acin2.com) or by other means (mail, fax, phone, electronic hospital outpatient clinic booking). This dedicated website was created on purpose for the ACIN2 and was used by all physicians at the HSA for reporting patients admitted at Emergency Department (ED), hospital ward or examined at the ultrasound/radiology laboratories. This topic was addressed every day by the neurology team at the emergency room and at weekly meetings of the neurology department. This webpage, identified by the patient NHN, contained the essential information for each event/patient: details of place, time and clinical characteristics, diagnostic procedures, and proposed diagnosis. (2) While in the first study hospital discharge registers, ED records and a sample of imaging department records were reviewed, in the present study a record-linkage methodology based on the NHN was implemented. For this purpose, several computer-generated lists were provided by the entities involved in the study: ED admissions, inpatient discharges and brain/cervical arteries imaging procedures. (3) In the first study death certificates could be scrutinized, but the new legislation issued in Portugal forbids access to this “individual” information for research purposes; to overcome this issue we screened the HSA and the National Network for integrated long-term care death lists. Autopsies performed at HSA pathology department or at the Medical Forensic Institute in Porto (covering the Northern Region) whose death cause was stroke were reviewed in both studies. (4) Regular contacts with private hospitals and nursing homes were made in both studies to capture patients not attending the NHS. In the present study we used two additional case-ascertainment methods recently implemented in the NHS. The pre-hospital stroke code activation lists and the computergenerated list of all primary care patients (HCs) with a diagnostic code of stroke/TIA. The reorganization of the NHS in the city of Porto centralized the emergency care out-ofhospital in a single dedicated unit (Service for Urgent Situations), open daily from 8h-24h. This unit receives approximately 4,300 patients per month and the clinical information is hand-written. To scrutinize all records would be an exceedingly time-consuming task and so we selected a two month sample to estimate an eventual loss of cases. Prospective methods included direct “individual” referrals and daily check of ED admissions at HSA; at other hospitals in the region of Porto (Figure 1) this was done on a monthly basis. Retrospective methods included a monthly, quarterly or yearly based review of hospital diagnostic coding data (International Classification of Disease, Ninth revision codes 430 to 438, 342 and 781) and the overall Northern Region pre-hospital stroke code activation records, for identifying cross-boundary flow of patients. At the end of the study period the Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 79 - computerized register of ED visits of the remaining 21 hospitals in the Northern Region were checked. Between October 1, 2009 and September 30, 2011, all possible strokes occurred in patients registered at Western Porto unit were recorded. Surveillance of all sources of information continued for a further three months to ensure full registration. Patients were examined by a neurologist as soon as possible after the event at ED, during their hospital stay or at a special study outpatient clinic and, followed up at three months. As in the first study, the principal investigator (M.C.) reviewed the medical history of each patient to ascertain the first-ever-in-a-lifetime stroke (FELS) and its pathological type. If a patient died soon after the event, we attempted to obtain additional information from an eyewitness and clinical records. For patients unable to communicate we interviewed close relatives or other suitable informants. The objectives and field work planning were presented to the Northern Region Health Authorities for granting permission to perform the study. The study was approved by the Porto Hospital Center Ethics Committee. Informed consent was obtained from the prospectively included patients or from next of kin, when appropriate, before any clinical assessment. The Portuguese Data Protection Authority approved all procedures implemented. Definitions As in the first study, 1 stroke was defined according to the World Health Organization as ‘rapidly developing clinical signs of focal (or global) disturbance of cerebral function, with symptoms lasting 24 h or longer or leading to death, with no apparent cause other than of vascular origin’. 11 Pathological types were classified according to Sudlow and Warlow as ischemic stroke, primary intracerebral hemorrhage (PICH) and subarachnoid hemorrhage (SAH). 12 Patients without brain CT scan performed within 30 days, no brain MRI, no postmortem confirmation, and no lumbar puncture or angiography in case of suspected SAH, were classified as stroke of undetermined type. Information on vascular risk factors (VRF) was collected using the same methodology throughout the two study periods. The criteria used for hypertension was a history of known hypertension or antihypertensive treatment; for diabetes, a previous diagnosis/treatment of diabetes mellitus with oral anti-diabetic/insulin or fasting glycaemia>126mg/dl, postprandial glycaemia200mg/dl and/or glucose tolerance test with values of glycaemia200mg/dl at the 2 nd hour; for hypercholesterolemia, a previous diagnosis/treatment of hypercholesterolemia; for atrial fibrillation, evidence from the Tendências na Incidência e Prognóstico do AVC - 80 - electrocardiogram or documented in patient’s record; patients were classified as current smoker if they smoked at the event date or at any time during the preceding 12 months. The pre-stroke and post-stroke (approximately 1 month after the event) modified Rankin Scale (mRS) scores were recorded for every patient and were used for classifying disability resulting from stroke as follows: a minor (non-disabling) stroke if the post-stroke score was less than 2 or the mRS score did not change after the event; otherwise they were considered non-minor (disabling). Statistical methods Following identical criteria of the first study, incidence rates were calculated using the population registered in the WPACES at the middle of the study period (September 30, 2010) and the confidence intervals (CI) were calculated based on the Poisson distribution. These rates were standardized to the European population. 13 Specific incidence rates and ratios (IRR) and the respective 95%CI were calculated for pathological stroke type and severity using as standard the Portuguese population of the 2011 Census. Poisson regression models were used to analyze the relationship between study period, gender, age group (<55, 55-64, 65-74, 75-84 and ≥85 years), and stroke risk. The outcome was the number of strokes within each age/gender/period strata, with the population counts from the 2011 Portuguese Census for each strata serving as the offset. We examined all models with the various combinations of effects: main effects for age, gender and, period and effect modification terms: age-gender, age-period, and genderperiod using the deviance that follows approximately a chi-squared distribution with degrees of freedom equal to the number of cells minus the number of parameters. The preferred model was that with fewer parameters that fitted the data (p>0.05), unless a model with an extra effect provided a significantly improved fit (p<0.05) over the simplest model evaluated by the difference in chi-square values for the two models with the difference in degrees of freedom. Models were tested for all strokes combined and a sensitivity analysis was performed for minor and non-minor stroke separately. Proportions and median values of baseline characteristics were compared in the two periods using the chi-square test and the median test following also a chi-square distribution. Setwise regression models were used to ascertain a period effect in short-term disability (28-day mRS score). The first model included the set of baseline and stroke characteristics and the effect of period, followed by all two-way interaction effects with period entered in the equation using a stepwise procedure; in the following model the set of management variables was added and again in a stepwise manner their interaction effects with period Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 81 - (final model). A robust Poisson model 14 was used to estimate the prevalence ratio of prestroke risk factors and the ratio of case-fatality rates by stroke type adjusted for age and sex, using the values of the first period as reference. Results The population of the second study comprised 189968 individuals registered at WPACES on September 2010, from whom 1020 were notified as stroke patients. Diagnosis was confirmed in 942 (92.4%) and 78 were excluded (31 with incorrect diagnosis, 36 out of study population and 11 for other reasons). A FELS occurred in 721 patients, 674 from the above and 47 (6.5%) first notified as TNA (compared to 3.2% in the first study). From prospective methods we identified 643 (89.2%) patients, from discharge lists 41 (5.7%), from HCs list and imaging department records more 28 (3.9%) cases and the remaining 9 (1.2%) by checking ED lists of hospitals outside metropolitan area; in the first study 69.5% patients were found by prospective methods and 19.3% by manual searching of ED lists at HSA and emergency transportation calls (equivalent in the present study to patients identified by prospective methods), 4.8% in discharge lists, 0.6% from imaging records and 5.6% by death certificates. Incidence Eleven years apart the incidence of FELS decreased from 269 to 190 per 100,000 and from 173 to 125 after standardized to the European population (Table 1). The overall reduction was 25%, 16% in men and 33% in women (Table 2). PICH incidence rate reduced 44% and that of ischemic stroke 16%; this reduction was higher for women compared to men. The incidence rate of fatal stroke declined 54%, 28% in men and 66% in women. The incidence in strokes that resulted in none or slight disability (mRS 0-1) declined 29%. Figures 2(A) to (D) display the age-sex specific incidence for all strokes and the age-specific incidence for minor and non-minor stroke in the two study periods. The Poisson models that fitted data for all strokes and both minor and non-minor strokes includes the interaction between gender and age (Supplemental table), indicating a higher incidence in men compared to women in the youngest and no difference in the oldest. While in minor strokes there is a constant effect of period for all age-gender strata, IRR=0.79 (95%CI, 0.65-0.96), for all strokes and non-minor strokes the effect of period depends on gender; the overall decrease in all strokes for the 2 nd period, IRR=0.75 (95%CI, 0.67-0.84), was significant in women, IRR=0.67 (95%CI, 0.57-0.79) and not in men, IRR=0.85 (95%CI, 0.72-1.02) (Table 3); for non-minor strokes the overall IRR=0.73 Tendências na Incidência e Prognóstico do AVC - 82 - (95%CI, 0.63-0.84), IRR=0.62 (95%CI, 0.51-0.75) for women and, IRR=0.89 (95%CI, 0.71-1.12) for men. Table 1. Annual incidence rates of first-ever stroke per 100,000 in Porto, Portugal over eleven years, by gender and age Men Women Total Age, years at Risk n rate 95% CI at Risk n rate 95% CI at Risk n rate 95% CI 1998-2000 <35 16178 2 6 1-22 18175 3 8 2-24 34353 5 7 2-17 35-44 5276 10 95 45-174 7127 12 84 43-147 12403 22 89 56-134 45-54 5168 17 164 96-263 6582 21 160 99-244 11750 38 162 114-222 55-64 4202 34 405 280-565 5590 32 286 196-404 9792 66 337 261-429 65-74 3916 57 728 551-943 5856 76 649 511-812 9772 133 681 565-796 75-84 1991 42 1055 760-1426 3915 87 1111 890-1370 5906 129 1092 904-1281 85 519 13 1252 667-2142 1528 56 1832 1384-2380 2047 69 1685 1311-2133 Total 37250 175 235 200-270 48773 287 294 260-328 86023 462 269 244-293 ASRE 179 148-209 167 141-193 173 153-192 2009-2011 <35 37690 8 11 5-21 37774 2 3 0-10 75464 10 7 3-12 35-44 14530 18 62 37-98 15702 12 38 20-067 30232 30 50 33-71 45-54 12690 42 165 119-224 14624 21 72 44-110 27314 63 115 89-148 55-64 10482 70 334 260-422 12976 45 173 126-232 23458 115 245 200-290 65-74 7052 73 518 406-651 9506 70 368 287-465 16558 143 432 361-503 75-84 4264 80 938 744-1168 7790 136 873 726-1020 12054 216 896 776-1015 85 1296 36 1389 973-1923 3592 108 1503 1220-1787 4888 144 1473 1232-1714 Total 88004 327 186 166-206 101964 394 193 174-212 189968 721 190 176-204 ASRE 151 133-169 102 89-116 125 113-136 ASRE indicates age-standardized rate for the European population Patient characteristics and assessment In both cohorts, the included patients were more often women, with a lower proportion in the second cohort (p<0.02). Median age in the second cohort was 2 years higher than in the first cohort (p<0.04), and 4 years higher in women (p<0.005) (Table 4). The proportion of patients independent before stroke decreased in the second study but the PR=1.02 (95%CI, 0.95-1.09) after adjustment. In general, the prevalence of VRF increased in the 2009-2011 period; after adjustment the prevalence ratio for hypertension was 1.20 (95%CI, 1.08-1.33), 1.45 (95%CI, 1.09-1.93) for atrial fibrillation, 1.29 (95%CI, 1.06-1.56) for hypercholesterolemia, and 2.12 (95%CI, 1.80-2.49) for former smokers; the prevalence of diabetes and myocardial infarction/angina remained stable. The pattern of Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 83 - stroke has changed with a reduction in the proportion of hemorrhagic stroke from 20.6% to 15.3%. Table 2. Annual incidence rates of first-ever stroke per 100,000 in Porto, in the two study periods stratified by pathological type and disability in men and women 1998-2000 2009-2011 rate 95% CI rate 95% CI IRR* 95% CI All first stroke 260 236-285 195 181-209 0.75 0.67-0.84 Men 239 204-274 202 181-223 0.84 0.71-1.01 Women 281 248-314 188 169-207 0.67 0.57-0.83 Pathological type Ischemic stroke 196 176-217 164 151-177 0.84 0.73-0.95 Men 181 151-212 164 145-183 0.91 0.74-1.11 Women 211 182-240 165 147-182 0.78 0.66-0.93 Primary intracerebral haemorrhage 44 35-55 25 20-30 0.56 0.41-0.76 Men 41 28-59 31 24-41 0.79 0.51-1.23 Women 47 35-63 18 13-25 0.38 0.25-0.59 Subarachnoid haemorrhage 9 5-15 6 3-8 0.59 0.31-1.14 Men 7 2-16 6 3-10 0.85 0.29-2.48 Women 12 6-20 5 3-10 0.48 0.21-1.10 Rankin score at 28 days mRS 0-1 72 59-85 51 44-58 0.71 0.57-0.89 Men 86 66-110 59 47-70 0.69 0.50-0.94 Women 59 45-77 44 35-54 0.74 0.53-1.02 mRS 2-3 66 54-78 77 68-86 1.16 0.94-1.44 Men 58 42-78 84 70-97 1.40 1.00-1.97 Women 73 57-92 71 60-83 0.99 0.75-1.31 mRS 4-5 80 67-94 47 40-53 0.58 0.46-0.73 Men 61 44-81 37 28-46 0.60 0.41-0.88 Women 97 79-119 55 45-66 0.57 0.43-0.75 mRS 6 44 34-55 20 16-25 0.46 0.34-0.64 Men 33 22-50 23 17-32 0.72 0.44-1.20 Women 53 39-69 18 12-24 0.34 0.22-1.52 Minor stroke (non-disabling)† 92 78-106 72 64-81 0.79 0.65-0.96 Men 98 77-123 77 64-90 0.78 0.59-1.04 Women 86 69-107 68 57-79 0.80 0.61-1.04 Non-minor stroke (disabling) 169 149-188 122 111-134 0.73 0.63-0.84 Men 140 113-166 125 109-142 0.89 0.71-1.13 Women 195 167-222 120 106-136 0.62 0.51-0.75 *IRR indicates incidence rate ratio (2009-2011 versus 1998-2000), standardized to the 2011 Census population of Portugal; †excluding 2 patients in 1998-2000 Tendências na Incidência e Prognóstico do AVC - 84 - 0 200 400 600 800 1000 1200 1400 1600 1800 2000 <35 35-44 45-54 55-64 65-74 75-84 85+ Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 (A) Men 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400 1600 1800 2000 <35 35-44 45-54 55-64 65-74 75-84 85+ Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 1998-20001998-2000 2009-20112009-2011 (A) Men 0 200 400 600 800 1000 1200 1400 1600 1800 2000 <35 35-44 45-54 55-64 65-74 75-84 85+ Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 (B) Women 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400 1600 1800 2000 <35 35-44 45-54 55-64 65-74 75-84 85+ Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 1998-20001998-2000 2009-20112009-2011 (B) Women <35 0 200 400 600 800 1000 1200 1400 1600 1800 35-44 45-54 55-64 65-74 75-84 85+ (C) Minor stroke Incidence rate per 100 000 1998-2000 2009-2011 Age (years) <35 0 200 400 600 800 1000 1200 1400 1600 1800 35-44 45-54 55-64 65-74 75-84 85+ (C) Minor stroke Incidence rate per 100 000 1998-2000 2009-2011 1998-20001998-2000 2009-20112009-2011 Age (years) 0 200 400 600 800 1000 1200 1400 1600 1800 <35 35-44 45-54 55-64 65-74 75-84 85+ (D) Non-minor stroke Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 0 200 400 600 800 1000 1200 1400 1600 1800 <35 35-44 45-54 55-64 65-74 75-84 85+ (D) Non-minor stroke Incidence rate per 100 000 Age (years) 1998-2000 2009-2011 1998-20001998-2000 2009-20112009-2011 Figure 2. Age specific annual incidence rates of first-ever stroke in (A) men, (B) women, (C) minor stroke and (D) non-minor stroke in the two study periods. Errors bars are 95% confidence intervals Overall, 19 patients in the first period and 8 in the second did not attend ED services; the proportion of patients assessed in the first 24h after stroke onset was similar in both cohorts, while brain imaging (CT/MRI) performed in the first 24h hours in these patients increased from 84.5% to 91.9% (p<0.001). In-patient admissions increased from 57.8% to 64.2% (p<0.03), including 18 patients in the first study and 42 in the second study who had a stroke while in hospital for another reason. The degree of handicap has decreased in the second period, mainly shifting from severe (mRS>4) to moderate disability (1<mRS<5) (Figure 3(A)). Stroke case-fatality in the first 28 days decreased 37.9%, from 16.9% to 10.5%, and after adjustment 40% (95%CI, 2353%). Overall there was a significant mRS decrease in the second period after adjusting for patient and stroke characteristics, that lost importance when the interaction of period with age was included in the model, indicating that the better outcome was achieved in Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 85 - patients over 74 years compared to others (Table 5); moreover the overall higher level of disability of inpatients decreased in the second period. Figure 3(B) and (C) resumes these effects. Given there was non random missing data in pre-stroke disability in the first period (38 patients excluded from the previous analyses), we repeated the models using several inputting methods and the relative importance of these factors was not altered. Table 3. Incidence rate ratios (2nd vs. 1st period) by gender and incidence rate ratios (men vs. women) by age in the two study periods for all stroke and non-minor stroke; period incidence rate ratio (2nd vs. 1st) and incidence rate ratios (men vs. women) by age for minor stroke All strokes (PG + GA) Non minor stroke (PG + GA) Minor stroke (P + GA) Gender IRR 2 nd vs. 1 st 95% CI P IRR 2 nd vs. 1 st 95% CI P IRR 2 nd vs. 1 st 95% CI P Men 0.85 0.72-1.02 0.076 0.89 0.71-1.12 0.327 Women 0.67 0.57-0.79 <0.001 0.62 0.51-0.75 <0.001 All 0.79 0.65-0.96 0.017 Study period Agegroup IRR* M vs. W 95% CI P IRR* M vs. W 95% CI P IRR M vs. W 95% CI P 1st period < 55 1.32 0.94-1.86 0.112 1.74 1.06-2.88 0.030 55-64 1.48 1.07-2.05 0.018 1.27 0.83-1.94 0.279 65-74 1.10 0.83-1.44 0.520 0.76 0.54-1.09 0.134 75-84 0.88 0.68-1.14 0.328 0.86 0.63-1.18 0.363 85+ 0.72 0.51-1.02 0.068 0.69 0.46-1.02 0.065 2nd period < 55 1.69 1.22-2.32 0.001 2.53 1.57-4.08 <0.001 55-64 1.89 1.39-2.56 <0.001 1.84 1.23-2.74 0.003 65-74 1.40 1.09-1.80 0.009 1.11 0.80-1.53 0.536 75-84 1.12 0.90-1.41 0.314 1.25 0.95-1.66 0.111 85+ 0.92 0.66-1.28 0.623 1.00 0.69-1.44 0.993 All < 55 1.09 0.72-1.68 0.676 55-64 1.87 1.20-2.92 0.006 65-74 2.04 1.39-3.01 <0.001 75-84 0.92 0.64-1.33 0.656 85+ 0.62 0.28-1.37 0.242 Incidence rate ratio standardized to the 2011 Census population of Portugal *The IRR for the second period are obtained from the values of the first multiplied by the ratio of the period IRR in men and women (0.85/0.67 for all stroke and 0.89/0.62 for non-minor stroke) Tendências na Incidência e Prognóstico do AVC - 92 - References 1. Correia M, Silva MR, Matos I, et al. Prospective community-based study of stroke in Northern Portugal: incidence and case fatality in rural and urban populations. Stroke. 2004;35:2048-53. 2. [Risk of dying in Portugal, 1999]. Lisboa, Portugal: Direcção-Geral da Saúde 2001. 3. [Risk of dying in Portugal, 2005]. Lisboa, Portugal: Direcção-Geral da Saúde 2007. 4. Sarti C, Stegmayr B, Tolonen H, et al. Are changes in mortality from stroke caused by changes in stroke event rates or case fatality? Results from the WHO MONICA Project. Stroke. 2003;34:1833-40. 5. Feigin VL, Forouzanfar MH, Krishnamurthi R, et al. Global and regional burden of stroke during 1990-2010: findings from the Global Burden of Disease Study 2010. Lancet. 2014;383:245-54. 6. [National Program for Prevention and Control of Cardiovascular Diseases]. Despacho nº. 16415/2003 (II Série) - Diário da República nº. 193 de 22 de Agosto. Directorate-General of Health - Ministry of Health; 2003. 7. Moreira E, Correia M, Magalhaes R, et al. Stroke awareness in urban and rural populations from northern Portugal: knowledge and action are independent. Neuroepidemiology. 2011;36:265-73. 8. Moutinho M, Magalhaes R, Correia M, et al. [A community-based study of stroke code users in northern Portugal]. Acta Med Port. 2013;26:113-22. 9. Feigin VL, Carter K. Editorial comment--Stroke incidence studies one step closer to the elusive gold standard? Stroke. 2004;35:2045-7. 10. Sudlow CL, Warlow CP. Comparing stroke incidence worldwide: what makes studies comparable? Stroke. 1996;27:550-8. 11. Aho K, Harmsen P, Hatano S, et al. Cerebrovascular disease in the community: results of a WHO collaborative study. Bull World Health Organ. 1980;58:113-30. 12. Sudlow CL, Warlow CP. Comparable studies of the incidence of stroke and its pathological types: results from an international collaboration. International Stroke Incidence Collaboration. Stroke. 1997;28:491-9. 13. Waterhouse J, Muir C, Shanmugarathan K, et al. Cancer incidence in five Continents. Lyon, France: IARC Scientific Publishers; 1982: 673. 14. Deddens JA, Petersen MR. Approaches for estimating prevalence ratios. Occup Environ Med. 2008;65:481, 501-6. 15. [Risk of dying in Portugal, 2006]. Lisboa, Portugal: Direcção-Geral da Saúde 2009. Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 93 - 16. [Risk of dying in Portugal, 2012]. Lisboa, Portugal: Instituto Nacional de Estatística/DirecçãoGeral da Saúde 2014. 17. Lewsey JD, Gillies M, Jhund PS, et al. Sex differences in incidence, mortality, and survival in individuals with stroke in Scotland, 1986 to 2005. Stroke. 2009;40:1038-43. 18. Sealy-Jefferson S, Wing JJ, Sanchez BN, et al. Ageand ethnic-specific sex differences in stroke risk. Gend Med. 2012;9:121-8. 19. Petrea RE, Beiser AS, Seshadri S, et al. Gender differences in stroke incidence and poststroke disability in the Framingham heart study. Stroke. 2009;40:1032-7. 20. Appelros P, Stegmayr B, Terent A. Sex differences in stroke epidemiology: a systematic review. Stroke. 2009;40:1082-90. 21. Lofmark U, Hammarstrom A. Evidence for age-dependent education-related differences in men and women with first-ever stroke. Results from a community-based incidence study in northern Sweden. Neuroepidemiology. 2007;28:135-41. 22. Ahmad OB, Boschi-Pinto C, Lopez A, et al. Age standardization of rates: A new WHO standard. GPE Discussion Paper Series: No 31. Geneva: World Health Organization2001. 23. Feigin VL, Lawes CM, Bennett DA, et al. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review. Lancet Neurol. 2009;8:355-69. 24. Cabral NL, Goncalves AR, Longo AL, et al. Trends in stroke incidence, mortality and case fatality rates in Joinville, Brazil: 1995-2006. J Neurol Neurosurg Psychiatry. 2009;80:749-54. 25. Johansson B, Norrving B, Lindgren A. Increased stroke incidence in Lund-Orup, Sweden, between 1983 to 1985 and 1993 to 1995. Stroke. 2000;31:481-6. 26. Hallstrom B, Jonsson AC, Nerbrand C, et al. Stroke incidence and survival in the beginning of the 21st century in southern Sweden: comparisons with the late 20th century and projections into the future. Stroke. 2008;39:10-5. 27. Rothwell PM, Coull AJ, Giles MF, et al. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet. 2004;363:1925-33. 28. Kita Y, Turin TC, Ichikawa M, et al. Trend of stroke incidence in a Japanese population: Takashima stroke registry, 1990-2001. Int J Stroke. 2009;4:241-9. 29. Vibo R, Korv J, Roose M. The Third Stroke Registry in Tartu, Estonia: decline of stroke incidence and 28-day case-fatality rate since 1991. Stroke. 2005;36:2544-8. 30. D'Alessandro G, Bottacchi E, Di Giovanni M, et al. Temporal trends of stroke in Valle d'Aosta, Italy. Incidence and 30-day fatality rates. Neurol Sci. 2000;21:13-8. Tendências na Incidência e Prognóstico do AVC - 94 - 31. Corso G, Bottacchi E, Giardini G, et al. Community-based study of stroke incidence in the Valley of Aosta, Italy. CARe-cerebrovascular Aosta Registry: years 2004-2005. Neuroepidemiology. 2009;32:186-95. 32. Bejot Y, Mehta Z, Giroud M, et al. Impact of completeness of ascertainment of minor stroke on stroke incidence: implications for ideal study methods. Stroke. 2013;44:1796-802. 33. Polonia J, Martins L, Pinto F, et al. Prevalence, awareness, treatment and control of hypertension and salt intake in Portugal: changes over a decade. The PHYSA study. J Hypertens. 2014;32:1211-21. 34. Costa J, Oliveira E, David C, et al. Prevalence of hypercholesterolemia in Portugal and Europe: the same reality? Rev Port Cardiol. 2003;22:967-74. 35. Perdigão C, Duarte JS, Santos a. Prevalência e caracterização da hipercolesterolemia em Portugal. Estudo HIPOCRATES. Revista Factores de Risco. 2010;17:12-9. 36. Precioso J, Calheiros J, Pereira D, et al. [Prevalence and smoking trends in Portugal and Europe]. Acta Med Port. 2009;22:335-48. 37. Wieberdink RG, Ikram MA, Hofman A, et al. Trends in stroke incidence rates and stroke risk factors in Rotterdam, the Netherlands from 1990 to 2008. Eur J Epidemiol. 2012;27:287-95. 38. Alhusban A, Fagan SC. Secondary prevention of stroke in the elderly: a review of the evidence. Am J Geriatr Pharmacother. 2011;9:143-52. Decline of stroke incidence, disability and case fatality in Porto, Portugal between 1998-2000 and 2009-2011 - 95 - Supplemental data Supplemental table. Deviances and goodness of fit test for Poisson Log-linear Models fitted to the expected number of incident strokes according to the Census 2011 population Degrees of freedom All strokes Non minor stroke Minor stroke Model Deviance Goodness of fit P Deviance Goodness of fit P Deviance Goodness of fit P Null 19 1928.2 <0.001 1472.7 <0.001 533.8 <0.001 One-factor Models Period 18 1905.7 <0.001 1454.9 <0.001 528.3 <0.001 Gender 18 1928.1 <0.001 1470.8 <0.001 532.5 <0.001 Age 15 62.6 <0.001 47.5 <0.001 43.0 <0.001 Two-factor Models Period + Gender 17 1905.6 <0.001 1453.0 <0.001 526.9 <0.001 PG 16 1901.5 <0.001 1447.0 <0.001 526.9 <0.001 Period + Age 14 40.1 <0.001 29.7 0.008 37.5 0.001 PA 10 35.6 <0.001 28.0 0.002 29.4 0.001 Gender + Age 14 51.8 <0.001 43.7 <0.001 35.5 0.001 GA 10 35.7 <0.001 28.8 0.001 19.6 0.033 Three-factor Models P + G + A 13 29.3 0.006 25.9 0.018 29.9 0.005 PG + A 12 25.2 0.014 19.8 0.071 (1) 29.9 0.003 PA + G 9 24.8 0.003 22.4 0.008 GA + P 9 13.2 0.154 (1) 14.1 0.120 (1) PG + PA 8 17.3 0.027 PG + GA 8 9.1 0.337 (2) 4.9 0.767 (2) 14.1 0.080 PA + GA 5 8.7 0.123 6.5 0.261 (2) PG + PA + GA 4 2.6 0.633 (3) 1.3 0.856 (3) 6.1 0.191 (3) All strokes: (1) The simplest model that provided an adequate fit to the data is (GA + P); (2) this hierarchic model (PG + GA) improved significantly the fit of the previous model (chi-square=4.1, df=1, p<0.05) and the model (3) (PG + PA + GA) did not improve the fit (chi-square=10.6, df=5, p>0.05) - Final model PG + GA. Non minor stroke: (1) The simplest model that provided an adequate fit to the data is (PG + A); (2) this hierarchic model (PG + GA) improved significantly the fit (chi-square=14.9, df=4, p<0.005) and the more complex model (3) (PG + PA + GA) also improved significantly the fit (chi-square=18.5, df=8, p<0.02) - Final model PG + GA (the simplest that improved the fit). Tendências na Incidência e Prognóstico do AVC - 96 - Conclusions: Both models are of partial independence, i.e., the interaction between Gender and Age (higher incidence rates for youngest men compared to women and no differences in the oldest age-groups) is similar in both study periods and the decline in incidence from 1998 to 2011 was higher in women compared to men, irrespective of age-group (Table 3). Minor stroke: (1) The simplest model that provided an adequate fit to the data is (GA + P); (2) this hierarchic model (PA + GA) did not improve significantly the fit of the previous model (chisquare=7.6, df=4, p>0.1) as well as model (3) (PG + PA + GA), (chi-square=8.0, df=5, p>0.1) - Final model GA + P. Conclusions: The interaction between Gender and Age (higher incidence rates for men compared to women for those aged 55 to 74 years and no differences in the other age-groups) is similar in both study periods and there was an overall decline in incidence from 1998 to 2011 (Table 3). - 97 - C CO ON NC CL LU US SÕ ÕE ES S Nos capítulos anteriores (2-6) são apresentados cinco artigos que, naturalmente, apresentam uma discussão própria, onde se comparam os resultados encontrados com os descritos na literatura, discutem-se os pontos fortes e as limitações da investigação bem como as possíveis implicações em termos de saúde pública. Neste capítulo serão abordadas as principais conclusões resultantes desses artigos. A comparação das taxas de incidência de AVC em diferentes regiões e países permite aumentar o conhecimento dos vários mecanismos etiológicos e dos meios de prevenção. Para possibilitar a comparação é necessário que os diferentes estudos utilizem as mesmas definições, a mesma metodologia e que apresentem essa informação de forma semelhante. Com o segundo projecto ACINrpc pretendeu-se registar todos os primeiros acidentes neurológicos na vida ocorridos no período compreendido entre 1 Outubro de 2009 e 30 de Setembro de 2011. Tal como no primeiro projecto, a metodologia adoptada seguiu de perto os critérios ideais definidos para a realização de estudos de incidência. Para além disso a metodologia foi adaptada à evolução na quantidade e armazenamento de informação clínica em bases de dados informatizadas. No segundo projecto, a utilização de meios informáticos para efectuar a revisão das diferentes fontes de informação, garantiu uma maior fiabilidade na identificação dos Tendências na Incidência e Prognóstico do AVC - 98 - eventos ocorridos na população, ao contrário do primeiro, onde a revisão das fontes de informação decorreu manualmente, quase que exclusivamente, com recurso a fontes de informação em suporte físico (processos clínicos de doentes, boletins do serviço de urgência escritos, etc). Para além disso, a qualidade da informação clínica informatizada permitiu aferir, com maior clareza, se os eventos identificados correspondiam ao primeiro evento na vida, bem como confirmar muitos dos aspectos relacionados com os antecedentes clínicos do doente, principalmente a presença de factores de risco vascular. Por outro lado, ambos os projectos ACINrpc permitiram demonstrar a importância da inclusão e seguimento dos episódios neurológicos transitórios nos estudos de incidência de AVC. O primeiro estudo permitiu concluir que o padrão etário na incidência de AVC marca a diferença entre populações rurais e urbanas, com um maior risco para os mais novos no meio urbano e para os mais velhos no meio rural. No entanto, a ruralidade não está associada com o prognóstico a longo prazo, o que pode ser explicado pela menor prevalência dos factores risco associados ao AVC no meio rural e pelo facto de não haver diferenças no tratamento e gestão dos doentes. Com o seguimento a longo prazo, foi também possível concluir que o nível de incapacidade aos três meses é um bom indicador do prognóstico a longo prazo dos doentes com AVC Isquémico, quer em termos de sobrevivência quer em termos de incapacidade. Este conhecimento, para além de permitir informar melhor os doentes com AVC e os seus familiares, tem implicações no planeamento dos serviços de saúde, nomeadamente na avaliação custo/beneficio dos tratamentos na fase aguda e na organização de unidades de reabilitação dirigidas para o doente com AVC. O conhecimento dos factores que aumentam o risco de sofrer um AVC é amplo, ao contrário dos factores que o podem desencadear (despoletar). Com o estudo realizado no Porto, foi possível concluir que existe uma associação entre a temperatura ambiental e a ocorrência de AVC e sua gravidade. Ressalta também da análise a importância do tipo patológico de AVC, bem como o subtipo de AVC Isquémico, na interpretação destas associações, pois os mecanismos etiológicos subjacentes são diferentes. Para além disso, não é só a intensidade da exposição que interessa mas também a duração da mesma. Os serviços de emergência devem ter presente que determinadas condições meteorológicas podem contribuir quer para um aumento do número de episódios quer para uma maior gravidade dos mesmos. Com o estudo realizado na cidade do Porto, podemos concluir que o declínio da mortalidade por AVC pode ser explicado, em parte, pela diminuição na incidência e pela melhoria na incapacidade pós-AVC decorrente das gestão dos doentes mais velhos e Conclusões - 99 - tratamento durante o internamento. Para esta diminuição da incidência contribuiu predominantemente a diminuição do risco nas mulheres, em particular, do risco do AVC incapacitante e de AVC hemorrágico. Como referido noutros estudos, parece estar subjacente a este facto as mudanças socioculturais ocorridas entretanto na população portuguesa. Os avanços na qualidade dos cuidados no internamento (unidades de AVC) e a prevenção primária/secundária nos mais velhos contribuíram para uma melhoria no prognóstico destes doentes nos últimos anos. Perspectivas de investigação futura Os registos organizados no âmbito dos projectos ACINrpc constituem uma valiosa base de informação quer para a investigação actual quer para investigações futuras no âmbito do AVC ou dos acidentes neurológicos. Num futuro próximo, com o objectivo de comparar a evolução das taxas de incidência entre o meio urbano e o rural, será estudada a evolução das taxas de incidência no meio rural, recorrendo a uma metodologia de análise análoga à utilizada para medir o declínio da incidência no meio urbano. De igual modo, e dada a importância do AIT, será estudada a evolução da incidência destes eventos, comparando o meio urbano com o meio rural. Com o propósito de verificar se as associações com os parâmetros meteorológicos se mantém, após todos os desenvolvimentos recentes na prevenção e tratamento do AVC, será possível replicar o estudo realizado, utilizando para o efeito a informação recolhida durante os dois períodos (1998-2000 e 2009-2011). Em função da maior dimensão amostral, o estudo poderá ser alargado no sentido de compreender como é que estas associações variam com as diferentes etiologias do AVC isquémico. Antes da utilização da trombólise no tratamento do AVC na fase aguda, colocava-se pouco ênfase na rapidez e precisão de um diagnóstico de AVC. Mas para a trombólise ser eficaz é necessário que o doente procure e chegue atempadamente à instituição de saúde após o início dos sintomas. Com a informação do segundo estudo ACINrpc será possível estudar o prognóstico dos doentes com AVC em função do percurso que este realiza nas instituições de saúde. Em particular, será possível estender a caracterização dos doentes que utilizaram a Via Verde do AVC aos dois anos do estudo. A médio prazo, com o seguimento dos doentes do segundo estudo, será possível comparar o prognóstico a longo prazo dos dois coortes de doentes, para analisar até que ponto as melhorias observadas no curto prazo se repercutem no longo prazo. Tendências na Incidência e Prognóstico do AVC - 100 - - 101 - A AN NE EX XO OI I A AV VA AL LI IA AÇ ÇÃ ÃO OD DA AV VI IA AV VE ER RD DE ED DO OA AC CI ID DE EN NT TE EV VA AS SC CU UL LA AR R C CE ER RE EB BR RA AL LN NO ON NO OR RT TE ED DE EP PO OR RT TU UG GA AL L: : C CA AR RA AC CT TE ER RI IZ ZA AÇ ÇÃ ÃO OE EP PR RO OG GN NÓ ÓS ST TI IC CO OD DO OS S U UT TI IL LI IZ ZA AD DO OR RE ES S Mariana Moutinho 1 Rui Magalhães 2 Manuel Correia 3 Maria Carolina Silva 2 1 Serviço de Angiologia e Cirurgia Vascular, Centro Hospitalar Lisboa Norte. Lisboa, Portugal; 2 UNIFAI, Departamento de estudo de Populações, Instituto de Ciências Biomédicas de Abel Salazar. Universidade do Porto. Porto, Portugal; 3 Serviço de Neurologia, Centro Hospitalar do Porto. Instituto de Ciências Biomédicas de Abel Salazar. Universidade do Porto. Porto, Portugal. Acta Medica Portuguesa, 2013;26:113-122 118 Revista Científica da Ordem dos Médicos www.actamedicaportuguesa.com Moutinho M, et al. Avaliação da via verde do acidente vascular cerebral no norte de Portugal, Acta Med Port 2013 Mar-Apr;26(2):113-122 ARTIGO ORIGINAL da VVE. A maioria dos doentes que acederam à VVE estavam assintomáticos ou com sintomas não incapacitantes antes do episódio, em contraste com os restantes (85,4% vs. 58,2%). Globalmente, o transporte por ambulância foi usado por 50% dos doentes; 85,4% dos utilizadores da VVE chegaram dentro da janela terapêutica, descendo esta proporção para 34,8% nos restantes. A discriminação entre utilizadores da VVE e restantes é manifesta na proporção que apresenta cumulativamente os três sinais de AVC, 47,9% vs. 13,6%, e mais de 95% apresentam os sinais mais característicos - falta de força e/ou boca ao lado. Globalmente, 72,3% dos doentes tiveram o primeiro AVC na vida, sendo a proporção de AVCs isquémicos ligeiramente superior nestes (83,4% vs. 81,3%), independentemente do acesso ser a VVE. A proporção de utentes da VVE internados é muito superior (93,8%) à dos restantes (57,2%), particularmente quando o AVC é isquémico (94,1% vs. 53,1%). De notar que 26,3% dos doentes internados utilizaram o acesso pela Via Verde (45 pela VVE e 50 pela VVI). Os determinantes de utilização da VVE entre as características sócio-demográficas e sintomatologia, foram a idade e o número de sintomas, diminuindo 4% por um aumento unitário na idade (RP = 0,96; IC 95%: 0,94 - 0,98) e quintuplicando por cada sintoma adicional (RP = 5,0; IC 95%: 3,0 - 8,4). (Tabela 2). Na Fig. 3 está descrito o acesso aos cuidados de saúde após um AVC, nomeadamente a activação da VVE ou VVI, serviço de urgência sem VVAVC ou outro. Considerando cumulativamente os quatro critérios de utilização da VVAVC, só 95 (15,9%) doentes a poderiam utilizar e, dos que utilizaram a VVE (8%), apenas 56,3% satisfaziam esses critérios. Globalmente a janela terapêutica (112 / 574 = Tabela 2 - Caracterização dos doentes com AVC, utilizadores e não utilizadores da Via Verde externa Via Verde externa Outros Total (n = 48) (n = 552) (n = 600) P Idade média (dp), anos 68,2 (11,8) 73,2 (13,8) 72,8 (13,7) 0,014 n%n%n % Homens 27 56,3 257 46,6 284 47,3 0,2 Meio urbano 41 85,4 427 77,4 468 78,0 0,2 Rankin prévio 0,005 Assintomático 34 70,8 227 43,0 261 45,3 Sintomas não incapacitantes 7 14,6 80 15,2 87 15,1 Incapacidade ligeira 4 8,3 83 15,7 87 15,1 Incapacidade moderada 12,1 93 17,6 94 16,3 Incapacidade moderadamente grave 12,1 32 6,1 33 5,7 Incapacidade grave 12,1 13 2,5 14 2,4 Desconhecido 24 Transporte em ambulância 48 100,0 252 45,7 300 50,0 Tempo entre episódio e SU < 3h 41 85,4 192 34,8 233 38,9 0,001 Sinais “falta de força num dos membros” 46 95,8 323 58,5 369 61,5 0,001 “boca ao lado” 47 97,9 312 56,5 359 59,9 0,001 “dificuldade em falar” 24 50,0 116 21,0 140 23,3 0,001 Com os três sinais 23 47,9 75 13,6 98 16,3 0,001 1º AVC na vida 40 83,3 394 71,4 434 72,3 0,08† Isquémico 29 72,5 333 84,5 362 83,4 0,08‡ Hemorrágico 11 27,5 58 14,7 69 15,9 Desconhecido - 30,8 3 0,7 AVC recorrente 8 16,7 158 28,6 166 27,7 Isquémico 5 62,5 130 82,3 135 81,3 0,2‡ Hemorrágico 225,0 24 15,2 26 15,7 Desconhecido 112,5 4 2,5 5 3,0 Internamentos 45 93,8 316 57,2 361 60,2 0,001 AVC isquémico§32 94,1 246 53,1 278 55,9 0,001 AVC hemorrágico§12 92,3 66 80,5 78 82,1 0,3 †Média e desvio padrão; ‡comparação do 1º AVC na vida com AVC recorrente; §comparação do diagnóstico; ¶percentagem sobre o total de AVC desse tipo. Revista Cie ntífi ca da Ordem dos Médic os ww w.actame dica portuguesa.co m 119 Moutinho M, et al. Avaliação da via verde do acidente vascular cerebral no norte de Portugal, Acta Med Port 2013 Mar-Apr;26(2):113-122 ARTIGO ORIGINAL 19,5%) e a idade superior a 80 anos (42 / 574 = 7,3%) são os critérios que isoladamente impediriam mais frequentemente a utilização da VVAVC. Dos não utilizadores da VVAVC, 426 (89,5%) não reuniam critérios de activação e destes 97 (22,8%) não seriam eventuais candidatos apenas porque não chegariam dentro da janela terapêutica e 21 (4,9%) apenas pelo critério da idade. Restringindo a comparação aos doentes com critérios de activação da VVAVC, pode observar-se na Tabela 3 que as características sócio-demográficas, período de atendimento e tempo médio de acesso não são significativamente diferentes nos utilizadores da VVE, VVI ou não utilizadores. A falta de força é o sintoma mais frequente (87,4%), seguido da boca ao lado (83,2%) e dificuldade em falar (31,6%). Nenhum doente na VVI apresenta cumulativamente os três sintomas, em comparação com 22,0% dos não utilizadores e 44,4% dos utilizadores da VVE. Cerca de 76,8% dos doentes são internados, proporção muito superior (96,3%) na VVE, decrescendo na VVI (83,3%) e não utilizadores (64,0%). Esta tendência é também manifesta nos doentes com AVC isquémico que fazem fibrinólise (77,3%, 36,4% Tabela 3 - Caracterização e prognóstico dos doentes com critérios de activação da Via Verde do AVC Via Verde Externa Via Verde Interna Não Via Verde Total (n = 27) (n = 18) (n = 50) (n = 95) P Caracterização Idade média (dp), anos 64,2 (9,6) 61,9 (11,6) 64,4 (9,9) 63,9 (10,1) 0,7 Tempo médio (dp) de acesso SU, minutos 82 (44) 93 (43) 91 (47) 89 (45) 0,7 n % n % n % n % Homens 16 59,3 12 66,7 29 58,0 57 60,0 0,8 Área urbano 23 85,2 16 88,9 38 76,0 77 81,1 0,4 Transporte em ambulância 27 100,0 13 72,2 28 56,0 68 71,6 Período 0,4 00 - 08h 311,1 15,6 7 14,0 11 11,6 08 - 16h 15 55,6 7 38,9 28 56,0 50 52,6 16 - 24h 9 33,3 10 55,6 15 30,0 3 35,8 Sintomas “falta de força num dos membros” 26 96,3 15 83,3 42 84,0 83 87,4 0,3 “boca ao lado” 26 96,3 14 77,8 39 78,0 79 83,2 0,1 “dificuldade em falar” 12 44,4 316,7 15 30,0 30 31,6 0,1 Número de sintomas 0,004 1 2 7,4 4 22,2 15 30,0 21 22,1 2 13 48,1 14 77,8 24 48,0 51 53,7 3 12 44,4 - 11 22,0 23 2,2 1º AVC na vida 21 77,8 14 77,8 38 76,0 73 76,8 1,0 Tipo de AVC: Isquémico 20 71,4 14 77,8 41 82,0 75 78,9 0,7 Tratamento e prognóstico Internados 26 96,3 15 83,3 32 64,0 73 76,8 0,005 Isquémicos 19 73,1 11 73,3 23 71,9 53 72,6 1,0 Fibrinólise 14 77,3 4 36,4 4 17,4 22 41,5 0,001 Rankin pós episódio 0,027† Sintomas não incapacitantes - 211,1 12,0 3 3,2 Incapacidade ligeira 311,1 15,6 10 20,0 14 14,7 Incapacidade moderada 5 18,5 8 44,4 22 44,0 35 36,8 Incapacidade moderadamente grave 8 29,6 5 27,8 10 20,0 23 2,2 Incapacidade grave 11 40,7 211,1 7 14,0 20 21,1 Taxa de letalidade aos 28 dias 311,1 211,1 36,0 8 8,4 0,7 †Média e desvio padrão; ‡Teste exacto de Fisher 120 Revista Científica da Ordem dos Médicos www.actamedicaportuguesa.com e 17,4%, respectivamente). No total de utilizadores da VVE, pode calcular-se uma (sub)estimativa da proporção de doentes com AVC isquémico submetidos a fibrinólise de 41,2% (14 em 34). O Rankin pós-AVC é mais grave nos utilizadores da VVE, 70,3% com dependência grave ou moderadamente grave, em comparação com 38,9% na VVI e 34,0% nos não utilizadores. Ajustando para a idade, sexo e número de sintomas, o risco de incapacidade grave pós- -AVC não está significativamente aumentado (RP = 2,9; IC 95%: 0,8 - 10,2) no acesso pela VVE, aumenta no entanto nos homens (RP = 4,2; IC 95%: 1,1 - 15,6) e com o número de sintomas (RP = 24,4; IC 95%: 5,2 - 114). DISCUSSÃO Tendo por base um registo prospectivo comunitário decorrido no Norte de Portugal entre Outubro de 2009 e Setembro de 2011, este é o primeiro estudo de base populacional a abordar a utilização da VVAVC, permitindo deste modo conhecer as condicionantes e efeito na população em geral de um programa destinado a minimizar as consequências do AVC. Neste registo foram usadas fontes de informação compreensivas de acordo com os critérios estabelecidos internacionalmente,14 salientando-se a informação recebida do INEM referente à utilização da VVE. De um modo geral pode concluir-se que a incidência de AVC está a diminuir, de 245 / 100 000 em 19991 para 180 / 100 000 após dez anos. A taxa de activação da VVE foi de 30 / 100 000 habitantes, correspondendo a 8% do total de AVC’s ocorridos na população, embora apenas 56,3% destes cumprissem os critérios de activação. A restrição imposta pelos actuais critérios faria com que apenas 95 (15,8%) dos casos de AVC pudessem ter um acesso regulamentar à VVAVC. Mesmo atendendo a este facto, apenas 27 (28,4% dos que cumpriam critérios) foram conduzidos ao hospital pela VVE. Tem sido descrito como factor importante associado à VVE a chegada breve ao serviço de urgência (SU) e subsequente tratamento, sendo este um dos seus objectivos principais.7 Neste estudo, o tempo decorrido entre o aparecimento dos sintomas e a chegada ao SU foi, em média, 82 minutos, menos nove do que verificou Quain et al15 num estudo na Austrália, mas ainda elevado. Podem apontar- -se alguns factores responsáveis por esta demora, como a falta de conhecimento dos sintomas mais comuns do AVC e dos procedimentos para solicitar uma ajuda mais imediata (112). Estes foram já descritos em vários estudos internacionais,16-18 referindo o papel preponderante desempenhado pelas campanhas de alerta da população e também de uma adequada interacção/comunicação entre o Centro de Orientação de Doentes Urgentes (CODU) e o INEM. No SU, 68% dos episódios foram classificados como AVC/AIT, valor semelhante aos 70% de uma triagem correcta objectivados para a VVE,7 assim como noutros estudos internacionais.19 Por outro lado, a proporção de falsos positivos neste estudo é quase seis vezes superior à encontrada por Robert et al20 num estudo semelhante realizado em Barcelona. Provavelmente, este facto deve-se ao receio de atrasar o acesso em caso de possível AVC e, mais uma vez, a dificuldades na realização da triagem pré-hospitalar. A sintomatologia mais frequente na activação da VVE foi a falta de força num dos membros, seguida da dificuldade em falar e boca ao lado, idêntico ao relatado noutros estudos.21,22 Uma explicação é ser uma alteração mais perceptível para quem vê primeiro o doente e contacta o CODU. No entanto, quando nos restringimos aos episódios de AVC, a dificuldade em falar é mais frequente nos utilizadores da VVE23 comparados com os restantes, talvez por ser um sintoma mais reconhecido pelos pacientes, uma vez que é característico da afectação do lobo esquerdo, permitindo, teoricamente, ao doente reconhecer melhor os seu défices (em comparação com o lobo direito) e, desta forma, pedir mais rapidamente ajuda. Por outro lado a falta de força num dos membros parece ser desta forma um sintoma mais comum no diagnóstico diferencial de AVC. A VVE teve uma taxa de activação de 8%, inferior aos 17,9% de Robert et al20 Este valor mais elevado foi, no entanto, encontrado numa situação mais restritiva, nomeadamente após exclusão dos doentes em que o episódio ocorreu no próprio hospital e dos transferidos de outros hospitais. Por outro lado, no relatório da VVAVC24 é mencionado um valor de 26% a nível nacional em 2010 para a percentagem de doentes internados em Unidades de AVC em que a VV (externa ou interna) foi activada e, neste estudo, o valor foi semelhante (26,3%). Como seria de esperar, o Rankin prévio foi mais baixo quando foi utilizada a VVE, o que pode ser explicado pelo critério de activação da independência prévia. A maioria dos utilizadores da VVE chegaram dentro da janela terapêutica, contrastando com pouco mais de um terço dos não utilizadores, padrão também verificado noutros estudos,23,25,26 sendo esta percentagem superior à verificada por Derex et al, em França,27 assim como por Kleindorfer et al, nos EUA.22 A discriminação entre utilizadores da VVE e restantes foi manifesta na proporção que apresenta cumulativamente os três sinais de AVC, o que poderá estar relacionado com a maior exuberância do quadro clínico traduzida, na subsequente chamada do 112. O facto de a idade média dos utilizadores da VVE ser inferior aos restantes pode novamente estar relacionado com o critério do limite da idade, mas também com o facto dos doentes mais velhos terem tendencialmente mais comorbilidade e, por isso, um crescente grau de dependência, o que restringe novamente a activação da VVAVC. Resultados semelhantes foram encontrados no estudo de Robert et al.20 Embora se esperasse que doentes com um AVC recorrente reconhecessem melhor os sintomas e por isso ligassem mais frequentemente para o 112, a VVE foi mais utilizada por doentes com o primeiro AVC na vida. Esta tendência foi também relatada por outros autores21 e pode ser devida ao facto de subsistirem alterações cognitivas sequelares ao primeiro AVC ou a uma dependência para as actividades de vida diária que não permita aos doentes com AVC recorrente preencher os critérios de activação da VVE. Nem um quinto de todos os AVC’s satisfizeram os critérios da VVE e, mesmo nos utilizadores da VVE, pouco mais Moutinho M, et al. Avaliação da via verde do acidente vascular cerebral no norte de Portugal, Acta Med Port 2013 Mar-Apr;26(2):113-122 ARTIGO ORIGINAL Revista Cie ntífi ca da Ordem dos Médic os ww w.actame dica portuguesa.co m 121 Moutinho M, et al. Avaliação da via verde do acidente vascular cerebral no norte de Portugal, Acta Med Port 2013 Mar-Apr;26(2):113-122 ARTIGO ORIGINAL de metade satisfaziam os critérios diminuindo este valor para aproximadamente um terço na VVI. O actual protocolo da VVAVC abrange assim uma percentagem reduzida de todos os episódios de AVC registados na população em geral, não atingindo o valor mencionado nos relatórios da VVAVC7 pois neste o denominador só inclui os doentes internados. A janela terapêutica e a idade superior a 80 anos foram os critérios que isoladamente impediriam com mais frequência a utilização da VVAVC. Pode aqui referir-se que a campanha de alerta da população Seja mais rápido que o AVC não está ainda a resultar num efeito desejado, ou porque a população não reconhece ainda os principais sinais/ sintomas de AVC ou porque, embora reconhecendo, não adopta uma acção rápida e adequada. Este ponto é também focado num estudo realizado no distrito de Viana do Castelo, chamando a atenção para o facto da idade avançada e/ou escolaridade baixa serem barreiras para que este alerta funcione adequadamente.28 Mesmo nos doentes que preenchiam os critérios da VVAVC a taxa de activação (externa) foi de apenas 28,4%. Em pouco mais de metade destes doentes não foi activada nem a VVE nem a intra- -hospitalar, com a agravante de 56,0% destes terem usado o transporte em ambulância para se deslocarem ao SU. Carecemos de uma informação importante referente aos doentes com AVC que recorreram ao CODU para os quais não foi activada a VVE, que poderá eventualmente vir a ser alvo de futura análise com inquérito aos próprios utentes. Mas, por outro lado, há ainda a questão de não ter sido activada a VVI, possivelmente devido ao facto do doente não preencher os critérios clínicos mais restritivos para a realização da fibrinólise, à falta de segurança/experiência de alguns profissionais para a realização deste tratamento ou ainda à falta de formação da equipa de realização da triagem hospitalar. Nos doentes com critérios de acesso à VVAVC, o tempo médio entre o início dos sintomas e entrada no SU não foi significativamente diferente nos utilizadores da VVE, VVI ou não utilizadores, resultados semelhantes aos encontrados por Robert et al.20 A proporção de doentes internados e de doentes com AVC isquémico que fazem fibrinólise vindos pela VV justifica amplamente a operacionalização da VVAVC, pois é muito superior nestes doentes. Resultados semelhantes foram encontrados noutros estudos, mas esta proporção é superior à encontrada em alguns deles, sobretudo nos realizados em Espanha.15,20,22,25,29,30 Em termos de prognóstico a curto prazo, o Rankin pós-AVC foi mais grave nos utilizadores da VVE, a exemplo de outros estudos.20,31 Note-se, no entanto, que, após ajustamento para as características sócio-demográficas e sinais/sintomas, este facto é explicado não pelo acesso ter sido pela VVE mas por estes doentes terem um maior número de sintomas e serem do sexo masculino. Como limitação deste estudo podemos apontar o facto de não terem sido analisados em detalhe outros critérios constantes das recomendações da VVE,7 entre outros, o tempo entre o acesso ao SU e a avaliação por um neurologista, realização de TAC, assim como de fibrinólise. Dado não haver diferenças no tempo decorrido entre os sintomas e a entrada no SU nos utilizadores (com critérios) da VVE em comparação com os restantes tal como em outros estudos,20 a exemplo destes, esta análise mais detalhada poderia ser indicativa de intervalos de tempo intrahospitalares mais baixos e adequados quando a VVAVC é utilizada. Um possivel viés de informação pode surgir do facto de nem sempre ter sido registava no episódio de urgência a utilização da VVI. No entanto, é pouco provável que isso tenha acontecido porque o número de activações da VVI é superior ao da VVE e além disso foi até mesmo mencionada em mais doentes que não obedeciam aos critérios de activação (64,0% na VVI vs 43,7%). Será necessário realizar mais estudos de base populacional para definir os riscos reais do uso de fibrinólise em doentes com mais de 80 anos de idade, bem como sobre a eventual extensão da janela terapêutica.32 Outro aspecto importante é o facto de a VVE estar a ser mais frequentemente utilizada nos AVC’s mais graves, impedindo um eventual tratamento fibrinolítico por contra-indicações clínicas. Atendendo a que existe uma alta taxa de ocorrência de AVC após um AIT,1 também o critério referente ao tipo de sintomas poderia ser mais abrangente, seguindo, por exemplo, a campanha americana Suddens,33 acrescentando nomeadamente a perda súbita de visão, tal como acontece no Código Ictus em Espanha.34 Deste modo, poderá ser possível aumentar ou mesmo eliminar o cut-off da idade, como sugerido noutros estudos,32,34 ou aumentar a janela terapêutica pelo menos para quatro horas e meia, considerando os resultados de estudos mais recentes.32,35-37 Poderá também sugerir-se a intensificação das campanhas populacionais, uma vez que a sua eficácia foi já positivamente avaliada por vários estudos.16,18,38,39 Devem ser dirigidas a públicos diversos utilizando meios e argumentos compreensivos, pelo menos focando os sintomas e activação da VVAVC. Uma abordagem educacional alternativa, já verificada em alguns estudos,40-42 seria a dos profissionais de saúde responsáveis pela triagem quer em relação ao quadro clínico objectivado pela VVAVC quer em relação à adequada comunicação entre os responsáveis. CONCLUSÃO Embora durante uma década a incidência de AVC tenha diminuído, este estudo mostra que os efeitos de alguns programas nacionais para diminuir as repercussões do AVC a nível comunitário, nomeadamente a VVAVC, poderiam ser mais abrangentes quanto aos critérios de acessibilidade, cobrindo actualmente apenas uma proporção estimada em 16% de todos os casos. Considerando os doentes que cumprem estes critérios, pode concluir-se que os utilizadores da VV têm um espectro sintomatológico/clínico mais grave e são também mais frequentemente internados e tratados por fibrinólise em comparação com os restantes. Mesmo assim, após ajustamento para estas características, a incapacidade pós-AVC não difere significativamente nos utilizadores e não utilizadores da VV. 122 Revista Científica da Ordem dos Médicos www.actamedicaportuguesa.com Moutinho M, et al. Avaliação da via verde do acidente vascular cerebral no norte de Portugal, Acta Med Port 2013 Mar-Apr;26(2):113-122 ARTIGO ORIGINAL CONFLITO DE INTERESSES Os autores declaram não existir qualquer conflito de interesses relativamente ao presente artigo. FONTES DE FINANCIAMENTO A informação usada neste artigo é parte integrante do Projecto PIC/IC/82858/2007 financiado pela Fundação para a Ciência e a Tecnologia. REFERÊNCIAS 1. Correia M, Silva MR, Matos I, Magalhães R, Lopes JC, Ferro JM, et al. Prospective community-based study of stroke in Northern Portugal: incidence and case fatality in rural and urban populations. Stroke. 2004;35:2048-53. 2. Sudlow CL, Warlow CP. Comparing stroke incidence worldwide: what makes studies comparable? Stroke. 1996;27:550-8. 3. Direcção Geral de Saúde. Risco de Morrer em Portugal. 1999. Lisboa: DGS; 2001. 4. Redon J, Olsen MH, Cooper RS, Zurriaga O, Martinez-Beneito MA, Laurent S, et al. Stroke mortality and trends from 1990 to 2006 in 39 countries from Europe and Central Asia: implications for control of high blood pressure. Eur Heart J. 2011;32:1424-31. 5. Johnston SC, Mendis S, Mathers CD. Global variation in stroke burden and mortality: estimates from monitoring, surveillance, and modelling. Lancet Neurol. 2009;8:345-54. 6. Direcção Geral da Saúde. Risco de Morrer em Portugal, 2006. Lisboa:DGS; 2009. 7. Administrações Regionais de Saúde, Instituto Nacional de Emergência Médica, Coordenação Nacional para as Doenças Cardiovasculares. Documento Orientador sobre Vias Verdes do Enfarte Agudo do Miocárdio (EAM) e do Acidente Vascular Cerebral (AVC). Lisboa: INEM; 2007. 8. Alto Comissariado da Saúde, Ministério da Saúde. Campanha nacional sobre enfarte e AVC. [Acedido em 21 de Novembro de 2012].Available from: http://www.min-saude.pt/portal/conteudos/a+saude+em+portugal/ noticias/arquivo/2008/1/campanha+avc.htm. 9. Feigin VL, Carter K. Editorial comment-Stroke incidence studies one step closer to the elusive gold standard? Stroke. 2004;35:2045-7. 10. Hatano S. Experience from a multicentre stroke register: a preliminary report. Bull World Health Organ. 1976;54:541-53. 11. Sudlow CL, Warlow CP. Comparable studies of the incidence of stroke and its pathological types: results from an international collaboration. International Stroke Incidence Collaboration. Stroke. 1997;28:491-9. 12. Rankin J. Cerebral vascular accidents in patients over the age of 60. II. Prognosis. Scott Med J. 1957;2:200-15. 13. Bamford JM, Sandercock PA, Warlow CP, Slattery J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1989;20:828. 14. Coull AJ, Silver LE, Bull LM, Giles MF, Rothwell PM. Direct assessment of completeness of ascertainment in a stroke incidence study. Stroke. 2004;35:2041-5. 15. Quain DA, Parsons MW, Loudfoot AR, Spratt NJ, Evans MK, Russell ML, et al. Improving access to acute stroke therapies: a controlled trial of organised pre-hospital and emergency care. Med J Aust. 2008;189:42933. 16. Alberts MJ, Perry A, Dawson DV, Bertels C. Effects of public and professional education on reducing the delay in presentation and referral of stroke patients. Stroke. 1992;23:352-6. 17. Alberts MJ, Bertels C, Dawson DV. An analysis of time of presentation after stroke. JAMA. 1990;263:65-8. 18. Segura T, Vega G, Lopez S, Rubio F, Castillo J. Public perception of stroke in Spain. Cerebrovasc Dis. 2003;16:21-6. 19. Kothari R, Barsan W, Brott T, Broderick J, Ashbrock S. Frequency and accuracy of prehospital diagnosis of acute stroke. Stroke. 1995;26:93741. 20. Belvís R, Cocho D, Martí-Fàbregas J, Pagonabarraga J, Aleu A, García- -Bargo MD, et al. Benefits of a prehospital stroke code system. Feasibility and efficacy in the first year of clinical practice in Barcelona, Spain. Cerebrovasc Dis. 2005;19:96-101. 21. Agyeman O, Nedeltchev K, Arnold M, Fischer U, Remonda L, Isenegger J, et al. Time to admission in acute ischemic stroke and transient ischemic attack. Stroke. 2006;37:963-6. 22. Kleindorfer D, Lindsell CJ, Moomaw CJ, Alwell K, Woo D, Flaherty ML, et al. Which stroke symptoms prompt a 911 call? A population-based study. Am J Emerg Med. 2010;28:607-12. 23. Palomino-Garcia A, Moniche-Alvarez F, De La Torre-Laviana FJ, Cayuela-Dominguez A, Vigil E, Jimenez-Hernandez MD. Factors that affect time delays to fibrinolytic treatment in ischaemic stroke. Rev Neurol. 2010;51:714-20. 24. Coordenação Nacional para as Doenças Cardiovasculares. Vias Verdes Coronária e do Acidente Vascular Cerebral: Indicadores de actividade. Lisboa: CNDC; 2010. 25. Geffner-Sclarsky D, Soriano-Soriano C, Vilar C, Vilar-Ventura RM, Belenguer-Benavides A, Claramonte B, et al. Provincial stroke code: characteristics and impact on health care. Rev Neurol. 2011;52:457-64. 26. Schroeder EB, Rosamond WD, Morris DL, Evenson KR, Hinn AR. Determinants of use of emergency medical services in a population with stroke symptoms: the Second Delay in Accessing Stroke Healthcare (DASH II) Study. Stroke. 2000;31:2591-6. 27. Derex L, Adeleine P, Nighoghossian N, Honnorat J, Trouillas P. Factors influencing early admission in a French stroke unit. Stroke. 2002;33:1539. 28. Moreira E, Correia M, Magalhaes R, Silva MC. Stroke awareness in urban and rural populations from northern portugal: knowledge and action are independent. Neuroepidemiology. 2011;36:265-73. 29. Riopelle RJ, Howse DC, Bolton C, Elson S, Groll DL, Holtom D, et al. Regional access to acute ischemic stroke intervention. Stroke. 2001;32:652-5. 30. de la Ossa NP, Sánchez-Ojanguren J, Palomeras E, Millán M, Arenillas JF, Dorado L, et al. Influence of the stroke code activation source on the outcome of acute ischemic stroke patients. Neurology. 2008;70:123843. 31. Adeoye O, Lindsell C, Broderick J, Alwell K, Jauch E, Moomaw CJ, et al. Emergency medical services use by stroke patients: a population-based study. Am J Emerg Med. 2009;27:141-5. 32. Sandercock P, Wardlaw JM, Lindley RI, Dennis M, Cohen G, Murray G, et al. The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (the third international stroke trial [IST-3]): a randomised controlled trial. Lancet. 2012;379:2352-63. 33. Kleindorfer DO, Miller R, Moomaw CJ, Alwell K, Broderick JP, Khoury J, et al. Designing a message for public education regarding stroke: does FAST capture enough stroke? Stroke. 2007;38:2864-8. 34. Masjuan J, Alvarez-Sabín J, Arenillas J, Calleja S, Castillo J, Dávalos A, et al. Stroke health care plan (ICTUS II. 2010). Neurologia. 2011;26:38396. 35. Hacke W, Kaste M, Bluhmki E, Brozman M, Dávalos A, Guidetti D, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med. 2008;359:1317-29. 36. Bluhmki E, Chamorro A, Dávalos A, Machnig T, Sauce C, Wahlgren N, et al. Stroke treatment with alteplase given 3.0-4.5 h after onset of acute ischaemic stroke (ECASS III): additional outcomes and subgroup analysis of a randomised controlled trial. Lancet Neurol. 2009;8:1095-102. 37. Lansberg MG, Bluhmki E, Thijs VN. Efficacy and safety of tissue plasminogen activator 3 to 4.5 hours after acute ischemic stroke: a metaanalysis. Stroke. 2009;40:2438-41. 38. Barsan WG, Brott TG, Broderick JP, Haley EC Jr., Levy DE, Marler JR. Urgent therapy for acute stroke. Effects of a stroke trial on untreated patients. Stroke. 1994;25:2132-7. 39. Pancioli AM, Broderick J, Kothari R, Brott T, Tuchfarber A, Miller R, et al. Public perception of stroke warning signs and knowledge of potential risk factors. JAMA. 1998;279:1288-92. 40. Kothari RU, Brott T, Broderick JP, Hamilton CA. Emergency physicians. Accuracy in the diagnosis of stroke. Stroke. 1995;26:2238-41. 41. Libman RB, Wirkowski E, Alvir J, Rao TH. Conditions that mimic stroke in the emergency department. Implications for acute stroke trials. Arch Neurol. 1995;52:1119-22. 42. Morgenstern LB, Staub L, Chan W, Wein TH, Bartholomew LK, King M, et al. Improving delivery of acute stroke therapy: The TLL Temple Foundation Stroke Project. Stroke. 2002;33:160-6. - 113 - A AN NE EX XO OI II I C CO OM MU UN NI IC CA AÇ ÇÕ ÕE ES S Anexo II - 115 - Effects of outdoor temperature and rain on the risk of hemorrhagic stroke Rui Magalhães 1 Manuel Correia 2 Maria Carolina Silva 1 1 Departamento de Estudo de Populações, ICBAS, Universidade do Porto; 2 Serviço de Neurologia, Centro Hospital do Porto, Porto, Portugal XX IEA World Congress of Epidemiology (2011) [Poster] Introduction: Changes in meteorological parameters have been associated with stroke occurrence. The incidence of primary intracerebral haemorrhages (PICH) seems to increase in days with cold/mild outdoor temperature. In Portugal, neurologists forward the hypothesis that the incidence of PICH increases in rainy days and not particularly low temperatures. This study aims to study the association between occurrence of PICH and weather parameters. Methods: Data from ACINrpc-project, involving 78 patients suffering a first-ever-in-the-lifetime PICH over a 2-year period in the city of Porto was used. Information on daily weather parameters was obtained from the National Meteorological Office. A Poisson model was used to estimate the association between weather parameters and PICH incidence. Using a conditional logistic regression model, a case-crossover design was then used to estimate the risk of PICH following specific exposures associated with PICH incidence: low diurnal temperature range (DTR) and rainy days. For each subject, the case period was matched with 4 control periods, the same weekday in the previous four weeks. Results: PICH incidence increases by 11.8% (95%CI: 3.8-20.4%) for 1ºC drop in DTR and 3.1% (95%CI: 1.1-5.1%) for a 1mm/m² in precipitation. Following a day with a DTR<4ºC the odds ratio is 2.9 (95%CI: 1.4-5.8), increasing to 8.8 (95%CI: 1.7-44.8) after a 48h exposure. Following days with low DTR and rain, the odds ratio is 3.2 (95%CI: 1.3-8.1) and 9.5 (95%CI: 1.1-88.9) for a precipitation>10mm/m² and 40mm/m², respectively. Conclusion: Precipitation by itself is not associated with PICH incidence, nevertheless has a synergistic effect in low DTR days. Anexo II - 116 - Table A1. Distribution of vascular risk factors in 78 patients with PICH Vascular risk factor N % Age  65y 46 59.0 Women 48 61.5 High blood pressure (HBP) 54 69.2 Cardiac disease (CARD) 23 29.5 Diabetes 19 24.4 Smoking habits 14 17.9 Table A2. Estimated percentage of daily variation in Intracerebral haemorrhage by unit variation in the preceding 24 hours of weather parameters Season/Weather parameter % 95%CI Season vs. Summer Autumn 32.0 -31.6 to 54.8 Winter 16.9 -40.4 to 29.1 Spring 53.3 -18.6 to 88.5 Temperature (decrease 1ºC) Maximum 4.7 * -0.4 to 10.1 Minimum -0.3 -5.1 to 4.7 Diurnal range 11.8 ‡ 3.8 to 20.4 Relative humidity (%) 1.6 -0.6 to 3.8 Precipitation (mm/m 2 ) 3.1 ‡ 1.1 to 5.1 Atmospheric pressure (hPa) 0.1 -3.0 to 3.4 *p<0.05; ‡ p<0.01 Table A3. Assessment of interaction between DTR and precipitation on overall PICH incidence and in the presence of vascular risk factors DTR (ºC) Precipitation (mm 2 ) Cases Controls OR 95%CI All patients  4 < 10 56 259 1.0  4  10 6 27 1.0 0.4-2.5 < 4 < 10 7 13 2.6 0.9-7.1 < 4  10 9 13 3.2 1.3-8.1 Expected OR: Additive model: 2.6 + 1.0 – 1.0 = 2.6 Multiplicative model: 1.0 x 2.6 = 2.6 Oldest ( 65 years) < 4  10 7 6 5.0 1.6-16 Expected OR: Additive model: 0.3 + 1.1 – 1.0 = 0.4 Multiplicative model: 0.3 x 1.1 = 0.3 with high blood pressure < 4  10 6 7 4.1 1.3-13 Expected OR: Additive model: 1.3 + 3.2 – 1.0 = 3.5 Multiplicative model: 1.3 x 3.2 = 4.2 All patients  4 < 40 61 282 1.0  4  40 1 4 1.3 0.1-2.2 < 4 < 40 12 22 2.5 1.1-5.3 < 4  40 4 4 9.5 1.1-89 Expected OR: Additive model: 2.8 Multiplicative model: 3.2 Anexo II - 117 - 0 30 25 20 15 10 5 10 5 0 0 Autumn Winter Spring Summer Autumn Winter Spring Summer Temperature (ºC) Precipitation (mm/m 2 ) Daily mean of events / 1 000 000 1 2 3 4 5 Primary Intracerebral haemorrhage Maximum temperature Diurnal temperature range Figure A1. Lowess smoothed values of daily incidence and meteorological parameters at Porto during the study period. Anexo II - 124 - Change in incidence of subaracnoid haemorrhage from 1999 to 2011 in the northern region of Portugal Rui Felgueiras 1 Rui Magalhães 2,3 Maria Carolina Silva 1,2 Mário Rui Silva 4 Ilda Matos 5 Carla Branco 1 Miguel Veloso 6 Marta Freijo 5 Jorge Poço 5 Manuel Correia 1 on behalf of ACIN2 Investigators group. 1 Serviço de Neurologia, Hospital de Santo António, Centro Hospital do Porto; 2 Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto; 3 UNIFAI, Universidade do Porto; 4 Serviço de Neurologia, Hospital de São Pedro, Centro Hospital de Trás-os-Montes e Alto Douro, Vila Real; 5 Serviço de Neurologia, Hospital de Mirandela, Centro Hospital do Nordeste, Mirandela; 6 Serviço de Neurologia, Hospital Santos Silva, Centro Hospital Vila Nova de Gaia/Espinho. Cerebrovascular Diseases, 35(suppl 3): 620. 2013. 22 nd European Stroke Conference (2013) [Poster] Background: One decade after the first community-based prospective incidence study, we intend to describe current trends in the incidence and short-term prognosis of subarachnoid haemorrhage (SAH) in northern Portugal. Methods: Data from the second prospective community-based study are presented and compared to the first one. All suspect first-ever-in-a-lifetime stroke occurring between October 2009 and September 2011 in about 240,000 residents were entered into a stroke registry. Based on standard definitions, both hot and cold pursuit sources of information were used for case ascertainment. Patients were observed at onset and at three months. Results: A total of 30 patients were included, half of them were man compared with 78.3% in the first study (p<0.04) and the average age increased from 60 to 63 years (p>0.5). The overall crude annual incidence rate decreased from 9.3/100000 (95% CI, 5.9-14.0) to 6.3/100000 (95% CI, 4.28.9), and after standardization to the European population from 7.3 (95% CI, 4.3-11.6) to 4.5 (95% CI, 2.9-6.9). The age-specific incidence rates were stable or lower than in the first study, with the previous peak at 65-74 years (29/100000) shifting for the 74-85 years (21/100000). The highest values for those 85 years or older were stable (37.3 and 36.6/100000), but there was an increase in the oldest women, from 26 to 41/100000. Case-fatality at 28 days decreased from 30.4% (95% CI, 15.6-50.9) to 23.3% (95% CI, 11.8-40.9). Conclusion: Compared to other studies in European countries the incidence of SAH in Portugal is one of the lowest after the decrease in the last decade. The age-sex incidence pattern is changing, mainly by the increase in incidence in the oldest women. Despite being now more common among the oldest, the short-term prognosis remained stable in the last decade. Anexo II - 125 - Table A9. Patient's characteristics and case-fatality 1998-2000 2009-2011 Patients 95% CI 95% CI Mean age 59.7 51.5-67.8 62.8 55.4-70.1 Rural 58.8 38.1-79.5 63.3 49.0-77.5 Urban 59.9 50.2-69.7 62.6 53.3-71.9 Gender: % Male 21.7 9.7-41.9 50.0 33.2-66.8 Rural 0.0 0.0-39.0 50.0 21.5-78.5 Urban 29.4 13.3-53.1 50.0 30.7-69.3 Case-fatality, % 30.4 15.6-50.9 23.3 11.8-40.9 Rural 16.7 3.0-56.4 25.0 7.2-59.1 Urban 35.3 17.3-61.4 22.7 10.1-43.4 Table A10. Evolution of SAH incidence (/100,000) 1998-2000 2009-2011 n rate 95% CI n rate 95% CI Residence area Rural 6 8.1 3.0-17.6 8 8.6 3.7-16.9 Urban 17 9.9 5.8-15.8 22 5.7 3.6-8.6 Gender Male 5 4.5 1.5-10.6 15 6.7 3.8-11.1 Female 18 13.2 7.9-20.9 15 5.9 3.3-9.7 All 23 9.3 5.9-14.0 30 6.3 4.2-8.9 ASR Europe 7.3 4.3-11.6 4.5 2.9-6.9 ASR – Indicates age-standardized rate Anexo II - 126 - 0 20 40 60 80 100 120 140 <35 35-44 45-54 55-64 65-74 75-84 85+ 1998-2000 2009-2011 Incidence rate /100,000 Age (years) 0 20 40 60 80 100 120 140 <35 35-44 45-54 55-64 65-74 75-84 85+ 1998-2000 2009-2011 1998-2000 2009-2011 Incidence rate /100,000 Age (years) Figure A6. Evolution of SAH incidence by age-group