Assessment of community perceptions regarding dengue prevention in Madeira Island, Portugal

ASSESSMENT OF COMMUNITY PERCEPTIONS REGARDING DENGUE PREVENTION IN MADEIRA ISLAND, PORTUGAL Teresa Lobo Machado Sousa Nazaré Tese de doutoramento em Biologia Básica e Aplicada 2014 II III Teresa Lobo Machado Sousa Nazaré ASSESSMENT OF COMMUNITY PERCEPTIONS REGARDING DENGUE PREVENTION IN MADEIRA ISLAND, PORTUGAL Tese de Candidatura ao grau de Doutor em Biologia Básica e Aplicada submetida ao Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto. Orientadora - Professora Doutora Rosa Teodósio Professora Auxiliar, Unidade Clínica Tropical, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Portugal Coorientadora - Professora Doutora Carla Alexandra Sousa Professora Auxiliar, Unidade de Parasitologia Médica, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Portugal Coorientadora - Professora Doutora Graça Porto Professora Catedrática Convidada, Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto, Portugal Investigadora Principal, IBMC – Instituto de Biologia Celular e Molecular IV V The research described in this thesis is inserted in the GABBA – Graduate Program in Areas of Basic and Applied Biology (Universidade do Porto) and was performed at the Instituto de Higiene e Medicina Tropical (Universidade Nova de Lisboa). Present work was possible to realize due to collaboration with the Madeira Health Authorities, the Institute of Health Administration and Social Affairs (IASAUDE, IP-RAM) and the financial support from Fundação para a Ciência e Tecnologia, Portugal (Ref. SFRH/BD / 51012 / 2010 and Ref. PTDC/SAUEPI/115853/2009). O trabalho de investigação descrito nesta tese está inserido no Programa Doutoral em Áreas da Biologia Básica e Aplicada – GABBA (Universidade do Porto), e foi realizado no Instituto de Higiene e Medicina Tropical (Universidade Nova de Lisboa). A realização deste trabalho foi possível graças à colaboração com as Autoridades de Saúde da Madeira, o Instituto de Administração da Saúde e Assuntos Sociais (IASAUDE, IP-RAM) e foi financiado pela Fundação para a Ciência e Tecnologia, Portugal (Ref. SFRH/BD/51012/2010 e Ref. PTDC/SAU-EPI/115853/2009). VI VII P REFACE This dissertation assembles data obtained as a result of my PhD research within the framework of the project “Dengue in Madeira archipelago” 1 . This thesis is structured in 4 chapters, preceded by a summary both in Portuguese and English. The first chapter provides an insight on previous knowledge regarding prevention of dengue disease. The objectives of this work are also presented in this section. The second chapter comprises the questionnaire survey performed before the dengue outbreak that occurred in Madeira Island, in 2012. It includes two different sub-chapters. The first presents the results already published (i). Second sub-chapter covers relevant data also collected before the dengue outbreak, but not included in the mentioned paper (i). The third chapter presents the results of main work performed after the dengue outbreak. It comprises two different surveys, separated in two different sub-chapters. One presents a questionnaire survey, the other comprises a focus group survey. Chapter four encloses an overall discussion and conclusions of the studies performed, together with the limitations and future perspectives of the work developed. In the Appendix are included relevant documents mentioned along this manuscript. Both surveys were conducted under my supervision, with collaboration of trained personnel from the project and from local authorities 2 . (i) Nazareth T, Teodósio R, Porto G, Gonçalves L, Seixas G, Silva AC, Sousa CA (2014) Strengthening the perception-assessment tools for dengue prevention: a cross-sectional survey in a temperate region (Madeira, Portugal). BMC Public Health 14: 39. 1 (long title) “Dengue in Madeira archipelago. Risk assessment for the emergence of Aedes aegypti mediated arboviroses and tools for vector control”; Ref. PTDC/SAU-EPI/115853/2009 2 The use of « , “ and ‘ along the document follows the criteria: « - for textual citations; “ – for use a word in its uncommon context; and ‘ – for specific expression of this work. VIII IX A CKNOWLEDGMENTS Tenho a sensação de que nenhuma meta é ganha só com o próprio esforço. Neste trabalho, tenho a certeza de que o que apresento aqui, tudo quanto aprendi, é resultado da interação de muitas pessoas, do encontro de muitos interesses e de muitas vontades, da junção de várias disponibilidades. Obrigada por tanto e a tanta gente. À minha orientadora Professora Doutora Rosa Teodósio, que aceitou o grande desafio de orientar uma Bioquímica (que pouco conhecia) numa tese de Saúde Pública, Obrigada por essa confiança desde início. À minha co-orientadora Professora Doutora Carla Sousa que me integrou no Projeto que coordena sobre o Risco de Dengue na Madeira, permitindome estudar e aprender uma realidade concreta, Obrigada pela oportunidade. À minha co-orientadora Professora Doutora Graça Porto que, muito para além de uma co-orientação institucional proporcionou uma mentoria essencial para o meu crescimento ao longo desta tese, Obrigada pela disponibilidade. Ao Programa GABBA e a todos os alunos e Professores que o compõem, em especial à Professora Doutora Maria de Sousa (fundadora), ao Professor Doutor Alexandre do Carmo (atual Presidente), à Dr.ª Catarina Carona (muito-mais-que-secretária), que proporcionam a jovens (aspirantes a) cientistas uma realidade única de aprendizagem em Ciência, Obrigada por tornarem os valores deste programa uma realidade, mesmo sendo muitas vezes os pioneiros a fazê-lo. Às autoridades de Saúde da Madeira, em particular à Enfermeira Ana Clara Silva, pela abertura de, no meio de um dia-a-dia repleto de “expediente”, XVI breeding sites, assuring its validity as a perception assessment tool. Moreover, the most frequent breeding sites found were compatible with a clean and organized urban environment, presenting an atypical pattern of infestation comparing to dengue endemic regions. After the terminus of the outbreak, a second perception assessment was performed through a randomised block design. Female populations from surveys performed before and after the outbreak were paired ensuring homogeneity in five determinant variables. After the outbreak the population have increased an average of two points in the perception measured (EP-score), but no significant change was observed in the number of residents who approximated to an EP-score=10. The number of myths believed by resident decreased to three out of the twelve myths alleged after the outbreak. Detailed results of the EP-analysis have shown an improvement of the assimilation of the majority of the pre-defined essential concepts and indicated which ones had a greater increase after the experience of a dengue outbreak. Limitations of the quantitative analysis methods are known, in what concerns the assessment of subjective aspects such as perceptions and thus may have a limited ability to identify erroneous or impairing perceptions. Given that the present work has direct implications for policy and practice, the validity of the perceptions assessed was considered to be of particular relevance to the current research. For these reasons, the same perception was assessed also through focus groups sessions and subsequent thematic analysis. This way it was attained an assessment through mixed methods, comprising both quantitative (using questionnaire) and qualitative (using focus group) data collection and analysis. Both quantitative and qualitative data were consistent, in the identification of perceptions that were impairing the community engagement. However, results from the qualitative data analysis have indicated some new perceptions (not identified by EP-analysis) which were mainly, the most difficult to detect, perceptions related to feelings and judgements. Furthermore, it offered an in-depth understanding of how the experience XVII of a dengue outbreak had modulated both cognitive and emotional aspects of the community perception. Overall, the main community perceptions observed as being hardening community engagement were: an inaccurate perception of dengue risk, a disbelief in the domestic control, and a mistrust in governmental entities involvement for the control of this health threat. This work represents the most comprehensive description of community perceptions regarding dengue prevention in short-term dengue communities. Lessons learned will undoubtedly reinforce the efficacy of dengue prevention initiatives in Madeira and also contribute for the preparedness of other dengue risk areas worldwide. XVIII XIX T ABLE OF CONTENTS Preface ……………………………………………………………………………...………. VII Acknowledgments…………………………………………………………………….....IX Resumo………………………………………………………………………………….......XI Summary……………………………………………………………………………….......XV Table of contents……………………………………………………………….….…....XIX List of Figures………………………………………………………………..….…….….XXI List of Tables………………………………………………………………......………..XXIII Abbreviations and Acronyms………………………………………………..….…..XXV CHAPTER I: GENERAL INTRODUCTION ………………………………1 DENGUE – AN ARTHROPOD-BORNE DISEASE……….…...…………………………....3 AEDES AEGYPTI - THE VECTOR……………………………………………………....4 DENGUE - THE DISEASE……………………………… ……………………………….9 DENGUE PREVENTION, CONTROL AND RE-EMERGENCE IN THE PAST………….13 HISTORY OF DENGUE PREVENTION………………………………….…………….13 LESSONS ABOUT RE-EMERGENCE: CLIMATIC OR BEHAVIOURAL CAUSES…...16 HEALTH PROMOTION……………………………………………….………………………18 DEFFINITION AND SCOPE……………………………………………..………….....18 METHODS, METHODOLOGIES, TECHNIQUES AND MODELS……………..…….23 CONTRIBUTIONS TO DENGUE PREVENTION…………………..……………...….28 A IMS………………………………………………………………………………………….…..33 MAIN AIMS ……………………………………………………………………....…….33 SPECIFIC OBJECTIVES…………………………………………………………..……..34 STUDY DESIGN…………………………………………………………………..…….35 CHAPTER II: PRE-OUTBREAK WORK …………….……………………37 II.1 - «STRENGENING THE PERCEPTION-ASSESSMENT TOOLS FOR DENGUE PREVENTION: A CROSS-SECTIONAL SURVEY IN A TEMPERATE REGION (MADEIRA, PORTUGAL)»…………...………………………….……………………..39 II.2 - ASSESSMENT OF PERCEPTION REGARDING BROAD DENGUE PREVENTIVE ISSUES……………………………………………………………………………………67 CHAPTER III: POSTOUTBREAK WORK…...…………………… . 83 III.1 - «IMPACT OF A DENGUE OUTBREAK EXPERIENCE IN A TEMPERATE REGION: FORWARD AND BACKWARD STEPS OF COMMUNITY PERCEPTIONS» ……....85 III.2 - IN-DEPTH PERCEPTION ASSESSMENT AND CONFIRMATORY MYTH IDENTIFICATION…………………………………………………………….…125 CHAPTER IV: FINAL DISCUSSION ………………………………………157 GENERAL DISCUSSION…………………………………………………………………...….159 PERCEPTION ASSESSMENT: QUANTITATIVE AND QUALITATIVE DATA…..…159 LIMITATIONS…………………………………………………………………………...167 CONCLUSIONS………………………………………………………………………………...…….168 LOCAL RECCOMENDATIONS………………………….……………………………..169 FUTURE PERSPECTIVES…………………………………..…………………………….171 References………………………………………………………………………………173 Appendix………………………………………………………………………………...189 XX XXI L IST OF F IGURES FIGURE I.1 – AEDES AEGYPTI PHYSICAL APPEARANCE 6 FIGURE I.2 – DENGUE RISK MAP 7 FIGURE I.3 – AEDES ALBOPICTUS DISTRIBUTION IN EUROPE 13 FIGURE I.4 – GEOGRAPHIC DISTRIBUTION OF AEDES AEGYPTI IN THE AMERICAS IN 1930, 1970, AND 2004 16 FIGURE I.5 – INTEGRATED MODEL OF HEALTH PROMOTION [54] 19 FIGURE I.6 – THE MAIN DETERMINANTS OF HEALTH [57] 20 FIGURE I.7 - PRESENT STUDY DESIGN 36 FIGURE II.1.1– PROPORTION OF RESPONDENTS THAT ACHIEVED EACH EP-SCORE’S LEVELS 49 FIGURE II.1.2 – PROPORTION OF RESPONDENTS THAT ACKNOWLEDGED EACH ESSENTIAL CONCEPT 50 FIGURE II.1.3 – PROPORTION OF RESPONDENTS THAT ‘UNDERSTOOD’, ‘PARTIALLY UNDERSTOOD’ AND ‘DID NOT UNDERSTAND’ EACH ESSENTIAL TOPIC 50 FIGURE II.2.1 – COMMUNITY CONCERN ABOUT MOSQUITOES 71 FIGURE II.2.2 – LEVEL OF CONCERN 71 FIGURE II.2.3 – CAUSES OF CONCERN 72 FIGURE II.2.4 – VECTOR DESCRIPTION 72 FIGURE II.2.5 – OTHER FACTORS OR SITUATIONS THAT PROMOTE MOSQUITO BREEDING 74 FIGURE II.2.6 – MEDIA-BASED STRATEGIES 74 FIGURE II.2.7 – OTHER COMMUNICATION STRATEGIES USED TO DIFFUSE MESSAGES ABOUT DENGUE 75 FIGURE II.2.8 – DENGUE’S MODE OF TRANSMISSION 75 FIGURE II.2.9 – DENGUE SYMPTOMS 76 XXII FIGURE III.1.1: PERCENTAGE OF RESIDENTS THAT ACHIEVED EACH LEVEL OF THE EP-SCORE WITHIN THE FOUR POPULATIONS: TOTAL AND PAIRED IN BOTH PRE-OUTBREAK AND POSTOUTBREAK STUDIES 98 FIGURE III.1.2: PROPORTION OF FEMALE RESIDENTS WHO HAVE ‘ASSIMILATED’ EACH OF THE TEN ESSENTIAL CONCEPTS IN BOTH PRE-OUTBREAK AND POST-OUTBREAK STUDIES. 101 FIGURE III.1.3: PERCENTAGE OF RESIDENTS THAT HAVE ‘UNDERSTOOD’, ‘PARTIALLY UNDERSTOOD’ OR ‘NOT UNDERSTOOD’ EACH OF THE FIVE STUDIED TOPICS. 102 FIGURE III.1.4: COMPARISON OF PERSONAL-SOCIO-DEMOGRAPHIC DATA BETWEEN PRE/POST-OUTBREAK SURVEY POPULATION (TOTAL AND PAIRED) 107 FIGURE III.2.1 – CONCEPTUAL MAP REPRESENTING ‘CONFUSION IN RISK PERCEPTION’ 140 FIGURE III.2.2CONCEPTUAL MAP OF THE PERCEPTION ‘DISBELIEF IN DOMESTIC SOURCE REDUCTION’ 145 FIGURE III.2.3CONCEPTUAL MAP OF THE PERCEPTION ‘MISTRUST IN GOVERNMENTAL ENTITIES’ 149 FIGURE III.2.4 – OVERALL PERCEPTION MODEL – INTEGRATION OF FIGURE III.2.1, FIGURE III.2.2 AND FIGURE III.2.3 151 XXIII L IST OF T ABLES TABLE II.1.1 - SOCIO-DEMOGRAPHIC CHARACTERIZATION OF THE INQUIRED / SCORED POPULATION AND EP-SCORE RESULTS PER SOCIO-DEMOGRAPHIC GROUPS 48 TABLE II.1.2 – LIST OF ALLEGED MYTHS AND FREQUENCY OF RESIDENTS THAT BELIEVED IN EACH OF THEM 51 TABLE II.1.3 – ASSOCIATIONS BETWEEN THE DOMESTIC PRESENCE OF PUTATIVE BREEDING SITES (ANY TYPE) 53 TABLE II.1.4 – ASSOCIATION OF EP-INCOMPLETE SCORES AND PRESENCE OF DOMESTIC BREEDING SITES 54 TABLE II.2.1 – TOPICS COVERED IN THE QUESTIONNAIRE, RESPECTIVE QUESTIONS, FIGURES AND TABLES 70 TABLE II.2.2 – CITED ENDEMIC COUNTRIES 77 TABLE II.2.3 – MOST VISITED ENDEMIC COUNTRIES 78 TABLE III.1.1: LIST OF TEN CONCEPTS DEFINED AS ESSENTIAL WITHIN EP-ANALYSIS 95 TABLE III.1.2: DESCRIPTION OF THE SOCIO-DEMOGRAPHIC FEATURE OF TOTAL AND PAIRED SAMPLES IN BOTH STUDIES 97 TABLE III.1.3: ALLEGED MYTHS IN BOTH PRE AND POST STUDIES AND RESPECTIVE FREQUENCIES 104 TABLE III.1.4: EP-SCORES FROM TOTAL AND PAIRED SAMPLES OF BOTH PRE/POSTOUTBREAK SURVEYS AND ASSOCIATIONS BETWEEN THEM 105 TABLE III.1.5: EVOLUTION OF THE SIZE OF THE RESIDENTS’ GROUPS BEFORE AND AFTER THE OUTBREAK ACCORDING TO THE CUT-OFF: EP-SCORE ≥ 7 105 TABLE III.1.6: COMPARISON BETWEEN BASIC AND ADJUSTED MATCHING CRITERIA AND RESULTS 110 TABLE III.2.1: RESULTS FROM DEDUCTIVE THEMATIC ANALYSIS 132 TABLE III.2.2: RESULTS FROM INDUCTIVE APPROACH ANALYSIS 134 TABLE III.2.3: OVERALL LIST OF THE ERRONEOUS AND IMPAIRING PERCEPTIONS 152 XXIV TABLE IV.1.1: CONSISTENCY IN THE PERCEPTION ASSESSED BY EP-ANALYSIS, OPENQUESTIONS AND FGS 164 TABLE IV.1.2: COMPARISON OF THE EFFECT OF SEVERAL DIFFERENT EXPERIENCES IN EPSCORE 167 XXV A BBREVIATIONS AND A CRONYMS AEGYPTI – study area of the first cross-sectorial survey performed before the outbreak (represents the most infested aegypti area in 2011) AME – admitted mosquito exposure, meaning that admitted to had been bitten by mosquitoes CBPR – Community-based Participatory research COMBI – Communication for Behavioural Impact DDT - dichlorodiphenyltrichloroethane DEC – dengue endemic country(ies). DF – Dengue Fever (classical dengue form) DHF – Dengue Haemorrhagic Fever (severe dengue form) EP – Essential Perception EPPM – External Parallel Process Model Extended-AEGYPTI – study of the second cross-sectorial survey performed after the outbreak (represents the most infested aegypti area in 2012 and the most affected by the dengue outbreak) FGS – Focus Group Session(s) HAPA – Health Action Process Approach KAP – Knowledge-attitude-and-practices (survey) PRE-outbreak survey – EP-analysis applied to results from questionnaire survey performed before the outbreak (Study 1, Part 1) POST-outbreak survey – EP-analysis applied to results from questionnaire survey performed after the outbreak (Study 2) WHO – World Health Organization YF – Yellow fever GENERAL INTRODUCTION (I) 6 FIGURE I.1 – A EDES AEGYPTI PHYSICAL APPEARANCE [11] Invasiveness and geographical spread One of the most alarming A. aegypti’s feature is its high ability to colonize new territories. Some invasive vectors disperse into new habitats by flight or wind, however A. aegypti’s geographic expansion is usually humanmediated. Aedes aegypti perfectly combines the desiccation-resistant eggs with its synanthropic behaviour. This way it not only can travel with humans (either in egg or the adult form), having access to territories miles away from their origin place, but also profit from the highly-moderated microenvironments that human domestic areas provide [6]. This explains the current A. aegypti’s geographic distribution. Originated in West Africa, the subspecies A. aegypti aegypti evolved from a sylvatic ancestor A. aegypti formosus. Its spread most probably started during the Portuguese expeditions which around 1500’s established trading routes from West Africa to Europe, Americas and Asia [12]. Breeding in ship´s water reservoirs or persisting as dessication-resistant eggs, A. aegypti resisted to long maritime travels, being introduced into the almost the rest of the world. Changes in A. aegypti feeding and GENERAL INTRODUCTION (I) 7 breeding habits were undoubtedly crucial for its settlement in urbanized areas. Local reports and studies on genetic variation suggest the A. aegypti’s introduction in Asia did not occur until late in the nineteenth century [7,11]. Currently, A. aegypti is established worldwide in regions located between the northern January and southern July 10 ºC isotherms. All territories positioned between these isotherms, are areas of potential risk for A. aegypti’s infestation (Figure I.2) [6]. F IGURE I.2 – D ENGUE RISK MAP Suitability of dengue transmission is gradually described from high (in red) to low (in dark blue), grey areas are unsuitable or non-endemic [13]. Other Aedes species Besides A. aegypti, several species of the Aedes genus, are described to be vectors of important pathogenic virus, such as A. albopictus, A. atropalpus, A. atlanticus or A. dorsalis. Out of those A. aegypti and A. albopictus are the most remarkable regarding their medical importance. In contrast with A. aegypti’s human preference, A. albopictus females feed upon all kind of mammals, and can also feed on reptiles, birds, GENERAL INTRODUCTION (I) 8 amphibians and most groups of vertebrates (opportunistic and zoonotic feeding habits). Moreover, A. albopictus’s breeding sites are very broad being either natural (such as tree holes) or artificial (water storage containers and other A. aegypti breeding sites). In advantage to A. aegypti, A. albopictus has the ability to adapt to cold temperatures by becoming dormant during the winter of temperate regions. This aptitude favours its invasiveness to northern areas or high-altitudes territories [14]. Medical importance Aedes aegypti is the main vector of three important viral infections: dengue, chikungunya and yellow fevers. Yellow fever (YF) is the original viral haemorrhagic fever known, and was for several years the most fatal, with mortality rates as high as 50% [15]. With the arousal of a highly effective vaccine in 1936, the number of cases decreased abruptly. Financial and logistical vaccination constrains in development countries combined with the growing international travelling had caused an increase of its incidence in last thirty years. Recent estimations count 200 000 persons annually infected by YF in tropical regions of Africa and South America [16]. Chikungunya virus can be transmitted by both A. aegypti and A. albopictus vectors. The first chikungunya’s epidemic outbreak occurred in East Africa around 1950’s [17]. Now-a-days this infection is enzootic/endemic (permanently present) in Asian and African tropical regions. Recently (2007) it has been introduced in Europe and then in the Americas, where only during March 2014 has been over 8000 suspected cases [18,19]. Due to its current huge worldwide dispersion there are not accurate global prevalence estimations. Chikungunya only rarely causes mortality being commonly considered a non-fatal disease [20]. Presently and comparatively to YF and chikungunya, dengue has the higher global prevalence and the larger geographic distribution [21]. Moreover, unlike YF, it still lacks an effective vaccine or specific anti-viral therapy. Consequently out of these three aegypti-transmitted diseases, dengue is currently and undoubtedly the most threatening. GENERAL INTRODUCTION (I) 9 D ENGUE – THE DISEASE Dengue was an important global disease in the 18 th century, and is currently still considered by WHO the major human arboviral infection worldwide [21]. Origin and History The first records of a disease with symptoms compatible with dengue were made in China as early as the period 265-420 AD. Major dengueresembling epidemics have been described in West Indies (in 1635), in Central America (in 1699) and in North America (in 1780 and 1945). Dengue-like clinical illnesses were also described in Australia, Caribbean Islands and in Europe (eventually from returned colonialists from tropical Asia and east Africa). Only after the isolation and characterization of dengue viruses, in early 1900’s, it was possible to attribute past records to dengue infections [7,16]. The uncertain origin of the term «dengue», registered almost contemporaneously as «Ki-dinga pepo» in east Africa, as «dandy» in English colonies and as «dengue» in Cuba and Spain, also confirms its past global spread. Even though the meaning of the dengue term is not perfectly understood, the alternative and previous name «breakbone fever» comes most probably from the dengue-associated bone, muscle and joint pains or from the profound fatigue that it can cause for several weeks after recovery [7,17]. Virus, clinical feature and pathology Now-a-days it is known that dengue is an arbovirose caused by the homonymous virus. Dengue is a single-strand RNA virus member of the genus Flavivirus/family Flaviviridae. There are five different described dengue virus serotypes: DEN-I, DEN-II, DEN-III, DEN-IV and DEN-V, whose single infection results in lifelong immunity to that specific serotype [17,18]. However, cross infections between different serotypes, result in only partial and temporary immunity. Moreover, subsequent dengue infections of different serotypes, even when separated by many years, GENERAL INTRODUCTION (I) 10 increase the probability of developing severe dengue form, denoted as dengue haemorrhagic fever (DHF) or simply severe dengue. Some dengue virulent strains can also cause these severe forms of dengue. Most of the dengue virus infections are asymptomatic and the symptomatic ones present a wide range of clinical manifestations [7,17]. Classic dengue fever (DF) is a flu-like illness characterized by high fever (40°C/ 104°F) accompanied by at least two of the following symptoms: severe headache, pain behind the eyes, muscle, bone or joint pains, and rash [19,20]. Severe dengue (or DHF) also cause abdominal pain, bleeding or breathing difficulty [24]. This occurs usually during two-to-seven days, following three phases, an initial febrile phase, a critical phase when death can happen, and a spontaneous recovery phase. Dengue severity (DHF) is derived from the disorder of the vascular permeability, eventually caused by alterations on the glycocalyx layer of the endothelial cells. This condition, lead to low blood pressure, loss of clotting proteins and platelets. It is thought to be trigged by an antibodydependent enhancement which is coherent with the higher risk of these severe forms to occur in secondary cross infections. However, this is not entirely understood mainly due to the lack of an animal model to study this clinical feature [7,17]. Mainly caused by a huge hypotension, dengue severe forms are sometimes fatal (in 1-10% of the treated cases and in approximately in 30% -50% of the untreated ones). There are no vaccines or specific anti-viral therapy currently available to treat dengue infections. Survival rates increase with prompt clinical diagnosis and appropriate clinical management of patients’ intravenous hydration [21]. After a dengue-infective mosquito bite, the virus replicates in the human host during the so-called intrinsic incubation period. This generally lasts two-to-seven days but it can last until fourteen days. Only then, when human’s viremia is high, transmission can occur from humans to uninfected mosquitoes. The symptoms’ onset usually start simultaneously with this viremia peak. In the mosquito, the virus replicates and reaches the salivary glands in four–to-ten days (extrinsic incubation period). After GENERAL INTRODUCTION (I) 11 being infected mosquito is able to transmit dengue for the rest of its life, continuing the cycle of dengue transmission [7,16,17]. Global current scenario Dengue is present in more than 125 endemic countries distributed throughout all WHO regions (Southeast Asia, Western Pacific, Americas, African, European and Eastern Mediterranean regions). A total of 3.6 billion people are estimated to live in risk of having dengue virus infection [21]. Recent cartographic studies suggests a total of 390 million dengue infections to occur annually worldwide, of which 96 million are severe forms [25]. A total of 20 000 annual deaths are reported but this number is considered to be underestimated. Children are believed to be the most affected by dengue mortality and morbidity [26]. Moreover, dengue also involves a huge economic burden. Studies about the annual aggregate cost in groups of countries are consistent: 2.1 billion American dollars (estimated in all nations of Americas) [27], 950 million American dollars (estimated in 12 southern-east Asian countries) [28], 1.8 billion international dollars (estimated in eight countries either Asian or American) [29]. These costs would be even higher if covering prevention or vector control expenditures, excluded in the previously mentioned studies. Productivity loss was, in fact, the main instalment of dengue costs. The annual global disease burden can also be measured in 700 000 disability adjusted life years (DALYs), which measures the sum of years of potential life lost due to premature mortality and the years of productive life lost due to disability [30]. These recent dengue burden estimations, especially in what concerns its global prevalence, which are more than three times the previous estimations of the World Health Organization, provided a triggering point for a wider discussion about dengue global prevention and control. Moreover, looking at A. aegypti’s previous infested areas and its current climatic distribution it’s clear that A. aegypti is still far from its maximum geographical dispersal [6]. Examples of territories which are at risk of A. GENERAL INTRODUCTION (I) 12 aegypti infestation are Europe, North America and part of Australia that had already been highly-infested in the past [6,11]. Aedes aegypti in Madeira Island (Europe) In 2005, triggered by population complaints about a nuisance mosquito, A. aegypti specimens were found for the first time in Madeira, an Atlantic Portuguese archipelago [27,28]. Despite authorities’ efforts through educational and vector-control strategies, during subsequent years the species thrived in the island, increasing the risk for local dengue emergence [33]. In October 2012, the first dengue outbreak was declared in Madeira Island which was also the first dengue epidemics in Europe after almost 100 years [30,31] . Until March 2013, when the outbreak were considered finished, there were notified 2168 DF cases (DEN-I), zero severe dengue (DHF) cases and 81 imported cases from Madeira to European countries [36]. Madeira Island is the biggest inhabited island of its homonymous archipelago. This European island is characterized by mild temperatures (average temperatures range from 16.1 ºC to 24.7 all over the year). Out of their 263 091 habitants more than 40% live in the major county, Funchal, where the population density is as high as 1433,5 habitants per square kilometres [37]. Currently, Madeira, is at risk of a second dengue outbreak. Also, being a highly touristic destination, Madeira also constitutes an open door for dengue virus introduction into non-endemic albopictus-infested regions such us Europe and North America (Figure I.3) [35,36]. Moreover, according to the European Centre of Disease Prevention and Control (ECDC), if re-introduced in Europe mainland, A. aegypti would most probably find adequate climatic conditions to become widely established [37,38]. GENERAL INTRODUCTION (I) 13 F IGURE I.3 – A EDES ALBOPICTUS DISTRIBUTION IN E UROPE Color Scale: Red – regions where A. albopictus is established; Yellow – regions where A. albopictus was recently introduced; Green - regions where A. albopictus is absent; and Grey - no data regarding A. albopictus’s current vector surveillance is available [42]. DENGUE PREVENTION, CONTROL AND RE-EMERGENCE IN THE PAST H ISTORY OF THE DENGUE PREVENTION Several and crucial issues about dengue current prevention, control and reemergence can be understood looking at how it evolved in the past. Successes (1900-1970) Although until 1970’s there were no dengue specific campaigns, dengue have disappeared from Americas and Europe before that by taking indirect advantage from the vector-control implemented during malaria and yellow fever campaigns in these territories. Until 1900’s dengue was not recognized as a mosquito-borne infection. The discovery that those responsible for causing yellow fever and dengue were filterable agents transmitted by A. aegypti occurred in 1903 [40,41]. GENERAL INTRODUCTION (I) 14 After this discover vector control strategies started to be studied, planned and implemented for the first time by the YF commission in Cuba and by the physician and bacteriologist Oswaldo Cruz in Brazil [40,41]. In this period, mainly due to the absence of the yellow fever vaccine, and the inexistence of multiple dengue serotypes co-circulation, South America YF incidence and mortality rates were much higher than the dengue ones. Vector control strategies were performed to prevent yellow fever epidemics and were based on the elimination of A. aegypti’s breeding sites (source reduction). This led to a dramatic decrease of YF cases in the Americas. Based on the source reduction outcomes in the YC control, the Rockefeller Foundation encouraged a campaign to eradicate it from the western Hemisphere. By 1925 a small coastal Brazilian area was the only recognized YF endemic area remaining in the Americas [43]. However, in 1928 Brazil suffered another big YF epidemic, most probably caused by the decline of the vector control strategies (after almost 20 years of YF control) and the presence of a sylvatic YF virus reservoir. In 1930, with the appearance of an effective YF vaccine, campaigns to control A. aegypti were even more abandoned. After 1945, with the advent of dichlorodiphenyltrichloroethane (DDT), a powerful insecticide A. aegypti control in Americas persisted through an approach alternative to source reduction [44]. The DDT seemed to be a quick and effective way to control A. aegypti, compared with the time-consuming source reduction practices that moreover needed long-term sustainability to be effective. From 1946 until 1970, the Pan American Health Organization (PAHO) coordinated DDT-spraying campaigns which led to A. aegypti’s almost complete disappearance from Americas [45]. The World Health Organization (WHO) also coordinated DDT–based massive campaigns to control malaria, attaining its eradication in North America, Soviet Union, Europe and North Africa between late 1950’s and 1975 [46]. Aedes aegypti had also disappeared from these territories during this post-World War-II period. Even though its eradication in North America and Europe was not planned and it is not well described, it most probably occurred simultaneously to eradication of the malaria vector through the effect of DDT [46]. GENERAL INTRODUCTION (I) 15 Few or none reports describe dengue prevention in Asia and Africa in this period. From 1940 to 1970, Africa saw a vaccine-based effective YF control undertaken by the French and the Rockefeller YF Commission [43]. Inexistent or poor mosquito control activities were performed against A. aegypti and other sylvatic Aedes-species. In Asia, for reasons that are not clear yet, yellow fever has never been recorded and dengue first became an important health threat in late 1950s, when the increased transmission of multiple serotypes resulted in the emergence of DHF epidemics [40,43]. Failures (after 1970) If until 1970’s dengue burden was veiled by other major vector-borne diseases, since this decade it had proved to be a major health threat. In America and also Africa DDT was starting to be implemented, but DDTbased campaigns began to fail. This condition, mainly caused by the emergence of insecticide resistance to DDT within the A. aegypti populations, may have contributed to the re-infestation of Central and South America by this species. Dengue outbreaks became, thus, more frequent and with higher mortality rates. During this decade different dengue serotypes started to co-circulate, arriving from Africa and Asia, causing severe dengue cases. In 1981, the first main DHF outbreak in the Americas occurred in Cuba, with 10.312 DHF cases [47]. Additionally, the increasing international air traffic and the unplanned grow of urban areas have promoted dengue serotypes circulation and its increased transmission during outbreaks. Since A. aegypti control programs had lost political interest with its eradication, they were at this time hard to re-implement. Moreover, when re-introduced, A. aegypti have found breeding-site-enriched environments, consequence of the abandonment of source reduction activities several years ago. Even when implemented, governmental source reduction activities were ineffective due to the mosquito population densities that, in the meanwhile, have increased tremendously. These populations have also established themselves in urbanized areas where their control was much GENERAL INTRODUCTION (I) 22 thus its role in the way people perceive and react to behavioural proposals in a (health) risky context [76-79]. However, few studies explore how past experience influence public perceptions and reactions. Evidence which would be brought by these type of studies would be of great value for those planning behavioural impact campaigns [76]. The role of the community According to WHO, community is a «group of people that may or may not be spatially connected, but who share common interests, concerns or identities». It could be local, national or international, with specific or broad interests, health determinants or socio-demographic features [56]. The community have a dual role in health promotion. First, as its general target since health promotion focuses on changing community behaviours. Second, as a crucial intervenient of health promotion that may actively contribute to health planning in general and to their own behavioural change in particular. In fact, on one hand, through communication and social-marketing methods, health promotion explores attractive and effective ways to transmit messages which promote healthy behaviours (health-seeking messages). On the other hand, health promotion aims to achieve this impact in behaviour through voluntary individual choices rather than through prescriptive impositions. For this purpose, health education is critical in guiding the community in achieving health-literacy and health-empowerment [58]. The former represents «the cognitive and social skills which determine the motivation and ability of individuals to gain access to, understand and use information in ways which promote and maintain good health»; and which strengthens the latter, «a process through which people gain greater control over decisions and actions affecting their health» [56]. Several strategies had been developed in order to promote this aimed community involvement. Community assessment surveys which explore community’s knowledge, perceptions, beliefs and opinions rapidly became a relevant strategy to concretize both community roles. By one side it provide crucial information for guiding the effective communication when GENERAL INTRODUCTION (I) 23 promoting behaviours. By another side, it allows the integration of the community’s views when deciding health measures or planning health interventions [77]. The community’s role was also focused by other approaches and methodologies such as Community-based participatory research (CBPR) and Communication for Behavioural impact (COMBI). These and other approaches /methodologies will be explored in the next section. M ETHODS , METHODOLOGIES , TECHNIQUES OR MODELS Another important theme commonly accepted within all health promotion experts is the relevance of systematic and planned actions [78]. After years of research, several methods, methodologies, techniques or models have been developed in order to: (i) translate health promotion principles into practice (ii) do it in a replicable and evaluable way, and (iii) accurately compare and evaluate interventions. Examples of relevant standardized designs are mentioned in the following paragraphs. The community assessment surveys which, as previously mentioned, is a method for improving community’s involvement, can be performed through quantitative or qualitative methodologies. Conclusions of the former rely on objectivity, validity and reproducibility, while within the latter, knowledge is gained by inter-subjectivity among researchers and the object of the research [79]. These epistemologically opposite approaches have divided researchers. By one hand the deductive feature of the quantitative approach can be criticized by close the research to unexpected results and not fully detect them. By other hand, the inductive feature of the qualitative one can be censured as positivist and experimental. Although the choice of the research approach is ideally determined by what one is trying to study, the combination of both can take advantage of the potentialities of each approach. This is the basis of the Mixed-methods research whereby quantitative and qualitative data collection analysis strategies are combined, connected or integrated to GENERAL INTRODUCTION (I) 24 provide conclusions regarding the same research question or aim. Standardized quantitative questionnaires are the prime example of the first, focus-group or deep interviews are examples of the second, and mass-media content analysis is a quantitative-qualitative hybrid methodology [77]. Mixed-methods research can follow extensive typologies varying in several aspects such as research design, sampling methodology, methods integration and research function [80]. Their selection should be in accordance with research questions and objectives. Questionnaires (a quantitative methodology for community surveys) A questionnaire survey is a technique to collect data from a particular community (population sample) in order to produce generalizable results. It comprises a list of questions regarding the topic of interest [81]. Questionnaires may be self-completed (in person, through mail, or e-mail) or fulfilled by an interviewer who strictly follows the questions in the inquiry (face-to-face or by telephone). Questions can be closed or open according to whether or not they present previously defined multiple-choice answers. Although the disparate results which may be obtained from these two types of questions, there is no consensus concerning which one generates the most valid results [86,87]. Open questions applied to big representative samples can guide the subsequent construction of multiple answers options for a similar closed question. Considerable literature explores techniques which contribute to the construction of adequate and unbiased surveys apart from the selection of the type of questions [85,87,88]. To achieve the first condition it is fundamental that the questionnaire construction ensures its reliability (ability to produce the same results in equivalent repeated applications) and its validity (ability to measure what is actually intended to be measured). For reaching the second condition, attention should be given to avoid what may empathize the difference between what is answered by GENERAL INTRODUCTION (I) 25 respondents in a particular question and what is their “real” knowledge/perception/opinion regarding that question (bias) [87,88]. Examples of common bias are the tendency of respondents to give social accepted answers or the influence that the order of the questions can have in the answers given in the course of the questionnaire. The analysis of data collected through questionnaire surveys can be as simple as the description of the frequencies of each answer (which implicates the establishment of categories for the open questions) [83]. However questions which intend to measure latent variables (which cannot be directly measured) require, therefore, a more complex analysis. These variables can be indirectly determined by the measurement of several related measurable variables, and by the use of assessment scales as the type of answers for each measurable variable. Examples of latent variable are satisfaction or social attitudes. The latter is frequently measured in public health through the commonly called knowledge-attitude-and-practices surveys (KAP surveys) [84]. Focus group (a qualitative methodology for community surveys) Focus group is a technique to collect data based on discussion sessions within small groups of individuals regarding a topic of research interest [85]. Being qualitative, this technique instead of objective estimations provides not only the identification of values, beliefs, perceptions, judgements and opinions but also their interpretation. The group of individuals participating in a focus group session (FGS) is frequently conveniently selected combining individuals of different socio-demographic characteristics except in at least one variable which should be homogeneous within the group [86]. The group size should range from six to twelve individuals. The FGS are conducted by the moderator who follows a discussion guide while simultaneously GENERAL INTRODUCTION (I) 26 facilitates the debate, drawing attention to all questions and participants [87]. For a complete data collection a minimal of two FGS should be performed or as many as the required until no newer data is obtained. Assessment of complex topics may require until ten FGS [88]. Considerations regarding the physical space selection and organization are of great relevance in order to promote free and equitable participation within the individuals [89,92] . Analysis of focus groups data it is basically the same that is applied to any other qualitative data. Rather than giving percentages to answers, qualitative analysis rely on conceptualization of the data content [93,94]. This process can be incredibly diverse and complex resulting in different methods [91]. Thematic analysis is seen as a foundational method for qualitative data analysis, as it provides core skills that will be useful for conducting many other forms of qualitative analysis. It consists in the identification of patterns (themes) within data and in the analysis of its meanings. Various techniques of identifying themes both manual and computerized, have been described [92]. Examples of other qualitative data analysis are content analysis and grounded theory. The first also “thematizes” data but explores the measurement of the frequency of different categories and themes, only if possible and with caution as a proxy for significance [93]. The second uses the “thematization” for inductively and systematically generate a theoretical explanation of the interest topic [90]. The analysis of focus group data has the advantage of exploring the interaction between research participants when compared with the analysis of qualitative data from other source. Focus group can be used in combination with other methodologies within the same study either in the beginning as a GENERAL INTRODUCTION (I) 27 preliminary / exploratory phase or in the end in order to assess or complement a particular aspect of the study. * Another important recently developed approach is the community-based participatory research (CBPR). This partnership approach has by one side brought the community into the arena of social problems solving among with researchers, governmental or institutional personnel, by other side as a branch of research-action it actively participates in the change of a particular question in an organization whilst conducting research. Also mentioned as community-based research it equitably involves community members, organizational representatives and researchers in the full process of research, and distributes decision-making and responsibilities by all intervenient [94]. Other important principles of CBPR are: its basis on strengths and available resources within the community, its balance between the knowledge production and intervention for improve health outcomes, and its focus on public health problems with local relevance [95]. The CBPR approach is not strict but rather it should be discussed and adjusted according to intervenient, target-community and health-context. Therefore large and diverse examples of participatory/involved/collaborative research have been described in the literature [100-102]. The main benefits described as a result of involving the community through this approach have been a significant gain in knowledge, expertise and skills in the research process and an improvement of research quality, validity, sensitivity and practicability [98]. The recognition of the determinant role of an effective communication to cause behavioural changes, also led to the development of a new methodology – Communication for Behavioural Impact (COMBI). COMBI is methodology which incorporates the many lessons of the past 50 years regarding health education and communication in a behaviour-focused preventive strategy. This process intend to engage individuals in the adoption and the maintenance of recommended healthy behaviours in GENERAL INTRODUCTION (I) 28 different health contexts such as dengue, lymphatic filariasis, malaria or HIV [99]. Finally, health promotion evaluation is also well-accepted among health promotion experts to be a method of extreme relevance, mainly due to the emergent feature of this discipline. Extensive research explores approaches or models which select the methodologies that most achieve health promotion objectives and which enable them to continuously improve its applicability at the local level [58]. Research designs or methodologies for health promotion evaluation are not quite consensual. However, large literature have already explore different ways to perform it. Planning models such as the five phase Preced-Proced model may support the evaluation process [100]. Participatory evaluation is believed to be «a real catalyst for change» by some, but due to its profoundly challenging execution this evaluation approach is still often more consultive than participatory [58]. C ONTRIBUTIONS TO DENGUE PREVENTION The compliance to certain preventive, protective or therapeutic actions is frequently promoted in dengue-preventive campaigns. Due to the “domestic” feature of its main vector, and the lack of an effective vaccine or treatment, community participation is therefore crucial for prevention of dengue fever. With the advent of health promotion dengue-professionals and researchers have detected some weaknesses in the strategies applied in past dengue prevention campaigns, as explored in following paragraphs. The concept of community participation (derived from health promotion models) helped source reduction activities to be cost-effective and longterm sustainable. The community-empowerment ideology also helped community-based strategies to be more effective. In fact, the communitybased educational campaigns which not only have transmitted to the community the key-information, but have also involved the public in the intervention process, were actually more successful in promoting GENERAL INTRODUCTION (I) 29 behaviour changes. Vertical actions (proposed and coordinated by governments in order to reach communities) were not able to ensure dengue-prevention alone, and therefore bottom-up or community-based interventions (in which community is involved since the beginning) were more and more encouraged [7,40]. However, with the development of behaviour models, KAPs were severally criticized in their scope. In fact, they commonly explore knowledge as a measure of practice, even if most of these studies rarely found a correlation between knowledge and practice [106-108]. Recent surveys which integrate the theoretical of behavioural impact models are therefore more focused on perceptions and beliefs of the community, rather than on its knowledge [103]. Consequently new terms (or disciplines) have emerged such as «lay epidemiology», «cultural epidemiology» and/or «epidemiology of beliefs» which look at how community beliefs are formulated [62,110,111]. All this have changed the scope of what is searched when collecting community views, improving the validity and the applicability of community surveys outcomes. In what concerns community surveys analysis, despite the countless studies which suggest methodologies to accurately measure attitudes/beliefs/perceptions/feelings known in the literature, these are still timidly used. Social marketing and communication theories recommend health promotion campaigns to use simple and pragmatic health-messages, contrasting with the common complex or vague behavioural proposals instance such for example: «eliminate or cover all containers on your property» [7]. Furthermore, there are many dengue-related proposed behaviours: some related to mosquito breeding prevention (preventive), some focused on personal protection against mosquito bites (protective), and others dedicated to disease management in order to achieve quick and effective recover (treatment-seeking). To be effective, campaigns should focus on one type behaviour and, if needed, change the type of behaviour focused according to the mosquito seasonality and/or outbreak dynamics [106]. Moreover, when promoting source-reduction practices, health- GENERAL INTRODUCTION (I) 30 messages should clearly explain the way to perform it and empathize the type(s) of breeding site to be eliminated. To achieve this purpose, a prior entomological characterization is advantageous in order to prioritize targeted breeding sites. All these communication lessons are strategically blended within the COMBI methodology. In dengue context COMBI methodology has been materialized into a comprehensive and innovative guide [106]. This guide presents a fifteen step-by-step process illustrated with real-life examples taken from twelve detailed case studies of current worldwide dengue programmes and it is intended for programme managers, NGOs, or researchers interested in integrating biological, chemical, environmental, and communication interventions in dengue prevention [112,113]. Some studies covering COMBI interventions have confirmed its efficacy in impacting behaviour [114,115] . After all these contributions from health promotion, and despite the resultant knowledge gain regarding behavioural impact processes, effective communication and engagement, there are still few records of successful cases of community-based source reduction interventions. Fortunately, the academic and governmental interest have been enabling to evaluate them. The correspondent main lessons learnt reveal [7,114,116]. The difficulty to effectively encourage community participation after the 1970s could be attributed to the long experience of idealistic, infallible, turnkey solutions for mosquito-borne diseases such as DDT-spraying and YF vaccination. Moreover, since up to that period dengue prevention were of the exclusive responsibility of the governmental institutions, when community-based strategies started, its relevance was hardly understood by both official institutions and general public [7]. Furthermore, since health promotion is recent and its achievements are still being analysed, past decades of “bad” communication strategies may have demotivate public and professionals about its efficacy [111]. Moreover, due to its recent implementation, the majority of the current promising community-based and behaviouralfocused interventions, still did not have time to produce visible outcomes [116,118]. Additionally, the time required for community engagement GENERAL INTRODUCTION (I) 31 strategies to be proposed, accepted, planned, implemented, evaluated and finally effectively control A. aegypti populations could also represent a motif of misjudging community-based interventions. Eventually less weight is given to the long period of time required in the development of the most frequently desired alternative tools. Alternative tools for dengue prevention After several decades of stagnation, research is now developing new classes of insecticides able to cause residual activity in the already resistant mosquito populations [119,120]. Nevertheless, attention should be given to the medical impact of mosquito chemical control [115]. Although, the known carcinogen effect of most of the available insecticides, these harmful chemicals are still legal for domestic and largescale use due to the lack of healthier alternatives equally efficient [116]. Opposite to chemical control, biological vector control strategies have no associated medical impact for humans and moreover its resistance process is much slower. Vector control strategies based on the use of Cocepods, Bacillus thuringiensis israelensis (Bti) and Wolbachia are the most efficient ones, all of them with proven outcomes [123,124]. The first two are wellknown larvicides which therefore should be applied in the breeding sites of mosquitoes. The latter is a bacteria which when infecting A. aegypti mosquitoes reduces its virus transmission and is vertically transmitted to the mosquito’ progeny. Other promising biological larvicides are Spinosad a sub-product from Saccharopolyspore spinosa bacteria, and Piriproxyphen a growth inhibitor [125,126]. Mainly due to the monitoring required, this approach is still not widely considered as high-effective. The use of other kind of genetically modified mosquitoes, such as the release of insects with a dominant lethal (RIDL) and the killer rescue-under dominance (KR-UD), are also very promising approaches to control mosquitoes [127,128]. By sterilising mosquito males or by conferring protection against virus infection they seem to have negligible or zero risk for humans, even though ethical issues for its usage are not entirely clarified. 38 39 II.1 - «STRENGTHENING THE PERCEPTION-ASSESSMENT TOOLS FOR DENGUE PREVENTION: A CROSS-SECTIONAL SURVEY IN A TEMPERATE REGION (MADEIRA, PORTUGAL)» 4 (S TUDY 1, P ART 1) 4 This sub-chapter was published as Nazareth T, Teodósio R, Porto G, Gonçalves L, Seixas G, Silva AC, Sousa CA (2014) Strengthening the perception-assessment tools for dengue prevention: a crosssectional survey in a temperate region (Madeira, Portugal). BMC Public Health 14: 39. 40 PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 41 ABSTRACT Community participation is mandatory in the prevention of Dengue outbreaks. Taking public views into account is crucial to guide more effective planning and quicker community participation in preventing campaigns. This study aims to assess community perceptions of Madeira population in order to explore their involvement in the A. aegypti’s control and reinforce health-educational planning. Due to the lack of accurate methodologies for measuring perception, a new tool to assess the community’s perceptions was built. A cross-sectional survey was performed in the Island’s aegypti-infested area, exploring residents’ perceptions regarding most critical community behaviour: aegypti-source reduction and their domestic aegypti-breeding sites. A novel tool defining five essential topics which underlie the source reduction’s awareness and accession was built, and is here called Essential-Perception (EP) analysis. Of 1276 individuals, 1182 completed the questionnaire (92·6%). EP-Score analysis revealed that community’s perceptions were scarce, inconsistent and possibly incorrect. Most of the population (99·6%) did not completely understood the five essential topics explored. An average of 54·2% of residents only partially understood each essential topic, revealing inconsistencies in their understanding. Each resident apparently believed in an average of four false assumptions/myths. Significant association (p˂0.001) was found between both the EP-Score level and the domestic presence of breeding sites, supporting the validity of this EP-analysis. Aedes aegypti’s breeding sites, consisting of décor/leisure containers, presented an atypical pattern of infestation comparing with dengue prone regions. Essential-Perception seemed to be an accurate tool to assess community’s perceptions regarding a specific behaviour. The studied population was not prepared for being fully engaged in dengue prevention. Moreover, incomplete knowledge may have generated the belief in false assumptions. Evidences suggest that EP-methodology was efficient and accurate in assessing the community perception and its compliance to practices. PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 42 BACKGROUND Aedes aegypti is one of the most competent vectors of dengue, yellow fever and chikungunya viruses. Recent estimations suggest a global impact of 390 million dengue infections annually worldwide [21]. Since there are no vaccines or specific treatments for this arboviral infection, the reduction of vector density is one of the most straightforward strategies for its prevention. Furthermore, recent studies unravel the high cost-effectiveness of an active and continuous vector control as opposed to an answer to dengue outbreaks [126]. According to the World Health Organization (WHO), A. aegypti’s control is mainly achieved by source reduction of the vector through the elimination of the mosquito breeding sites [127]. Due to A. aegypti’s domestic ecological feature, their larvae preferably proliferate in small and artificial water-containers, placed inside or near human houses [128]. Therefore, community contribution is, undoubtedly crucial in dengue prevention and control [118,135] . Educational campaigns that inform and mobilize the local communities are often implemented in the infested areas. In most preventive campaigns, the community is asked to do aegypti-source reduction: to eliminate (cover, empty and/or remove) the most common domestic breeding sites. Abundant literature may be found reporting community-oriented educational interventions and assessments of community knowledge/attitudes/practices/perceptions/beliefs regarding dengue prevention, most of which are performed in tropical regions [58,59,136141]. Even though the relevance of the latter issues is more and more recalled by important entities[142,143], most of the studies emphasize the need of new research approaches to explain and increase their commonly low efficacy [58,59,140,141,144,145]. Consequently, studies that suggest and/or test strategies that more effectively promote community behaviours and more accurately assess community perception, are of great need [77]. The ‘community perception’ term used here means «the collective views of a group of people (...) [perception] involves understanding/misunderstanding and discernment, and it includes a PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 43 choice and action (...) [perception is also] the product of social interaction», as stated by WHO [77]. In the past years, several viruses and vectors have significantly increased their geographic distribution as a result of globalization [43,146]. In 2005, A. aegypti specimens were recorded in Madeira, a temperate European island in the Atlantic, for the first time [34]. Rapidly, the local health authorities promoted educational activities based on television/radio communications, informative flyers/posters distribution and ‘door-to-door’ interventions to achieve community compliance in the domestic control of A. aegypti. In fact, despite these efforts, the mosquito population has thrived. Additionally, entomological studies reported high levels of resistance to DDT and pyrethroids in the local A. aegypti population [138]. In October 2012, less than one year after the beginning of this study, an outbreak of dengue was declared in the Island [139]. Currently, Madeira is at risk of becoming a dengue endemic territory. Also, being a highly touristic destination, it constitutes an open door for A. aegypti and/or dengue virus introduction into other temperate regions [140]. Moreover, non-tropic regions such as Europe and North America host A. albopictus another very competent arboviral vector [36,150,151]. A unique virus introduction into these temperate regions could trigger a disease epidemic [143]. Community-mobilization strategies that effectively reduce A. aegypti’s densities in Madeira Island are thus, mandatory. This study aims to estimate the community’s perceptions of Madeira residents regarding source reduction, and identify the most frequent aegypti-breeding sites present in the domestic environment of this nontropical region. An extensive and in-depth analysis is suggested as a novel tool for community perception assessment and educational planning. METHODS S TUDIED POPULATION The study area was chosen according to the A. aegypti’s distribution area, assessed by an island-wide entomological survey (Additional file II.1.1). PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 44 Based on mosquito abundance levels, a more restrictive zone called ‘AEGYPTI’, was selected. This area includes part of three municipalities 5 : Santa Luzia and São Pedro (both in Funchal county), and Câmara de Lobos (in a Funchal neighbouring county). A representative sample of residents aged 18 years old or over was selected from the electoral system database, using stratified sampling by the municipality. A universe of 13 433 adult subjects lived in the area of study (almost 7% of the Island’s adult total population) [144]. A sample size of 1083 subjects, was required to fulfil the objectives of this study (90% confidence level and 2·5% precision). A prevalence of 50%, regarding good knowledge, was assumed. This sample size was inflated in 20% to account for non-respondents and incomplete interviews. Individuals who were not found or who refused to participate were replaced. Q UEST i ONNAIRE AND ENTOMOLOGICAL INVENTORY A cross-sectional survey was performed through face-to-face interviews. In each interview, both a questionnaire to assess the residents’ perceptions and a domestic breeding site inventory of each household, were fulfilled. The surveys were performed by trained personnel (Health technicians of the local authority-IASAUDE) during October and November 2011. A total of three attempts were undertaken to contact the selected individuals: (i)- on weekdays between 9am and 5pm; (ii)-on weekdays between 5pm and 8pm; and (iii)-on Saturdays between 10am and 7pm. Participants gave oral informed consent prior to data collection. Previous to its application, the questionnaire was pre-tested in an aegypti-infested but non-selected area. The questionnaire comprised 13 questions, addressing five main topics (see criteria in Perceptions Evaluation paragraph): ‘Medical Importance’ (two questions), ‘Local Context’ (two questions), ‘Domestic Attribute’ (three questions), ‘Mosquito Breeding’ (three questions) and ‘Control Measures’ (three questions). The questionnaire also covered sociodemographic characteristics. The breeding site inventory listed 21 types of putative domestic breeding sites present in each househo ld. The study 5 This term is herein used in the sense of the term parish PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 45 was approved by Instituto de Higiene e Medicina Tropical Ethics Committee, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon (reference: 09-2013-TD). P ERCEPTION E VALUATION In order to accomplish accurate and in-depth perception estimation, several analyses were made. The most common answer frequency estimation several analysis were calculated (data not shown). Then, a list of five essential topics regarding source reduction was defined. Topics correspond to variables known to determine behaviour changes, such as, self-efficacy, behavioural expectancies, perceived susceptibility, etc. as mentioned in several models of behavioural change described in the literature [145,154]. According to behavioural change experts, the list of variables/topics were chosen and adapted to dengue context and to the particular Madeira scenario [145,154]. The five selected variables (here called ‘topics') are individually labelled as: (A. aegypti’s) Medical Importance, (its) Local Context, Domestic Attribute (of its vector-control), Mosquito Breeding (process) and finally, (vector)-Control Measures. We established the awareness and the understanding of these five topics as necessary and obligatory for the acceptance and presumed consequent adherence to (and consequent adherence) to domestic source reduction practice. Two concepts were selected to evaluate each of the latter five topics (these are here called ‘essential concepts’). By evaluating the acknowledgement of both concepts, a double-evaluation of the understanding of each of the five topics was done. This allowed for the detection of discrepancies in the way these five topics are understood. Collectively the ten concepts sum-up the awareness of the source reduction. This way, this methodology allows the estimation of the community’s perceptions through three distinct approaches: (i)-score of Essential-Perceptions, (ii)-topic understanding and (iii)-discrepancy detection/myth estimation, all described below. Concepts assimilation and score of essential-perceptions (EP-Score) PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 46 According to the residents’ answers, the acknowledgement of the ten essential concepts was calculated. Each concept corresponds to one or two questions. We obtained the EP-score for each resident assimilated (from 0 to 10), by attributing one point to each perceived essential concept. Thus, EP-score level corresponds to the number of (essential) concepts, out of the ten established that each resident has assimilated. Following EPanalysis’ criteria, only those who achieved an EP-score equal to 10 showed minimal and adequate perceptions to trigger individual compliance in source reduction (see an example in Additional file II.1.2). Respondents who have not answered all the thirteen questions were excluded from score calculation. Topic understanding The understanding of the five covered topics was evaluated according to the knowledge shown in topic-related essential concepts (Figure II.1.1 and II.1.2). Only residents who have acknowledged both topic-related concepts had completely understood the topic. The acknowledgement of only one out of the two topic-related concepts revealed a partial understanding. Residents who did not perceive any of the two topic-related concepts did not understand the topic. False perceptions/myths estimation Partial or absent understanding of one of the five topics could generate false perceptions concerning it (Additional file II.1.3). By analysing the acknowledgement of concepts for each topic and the discrepancies in its understanding, a list of myths (false information that is perceived as true by a part of the population) was estimated and also its supposed frequency in the population (Additional file II.1.4). S TATISTICAL A NALYSIS ( TEST STATISTICS ) All collected information was introduced and records were double-checked. Statistical analysis was performed using Excel (Microsoft Office, Windows Vista) and Statistical Package for Social Sciences 19.0 (SPSS, Inc., Chicago, PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 47 IL, USA). Answers obtained from the questionnaire were re-coded to obtain other categorical variables linked to the above mentioned ten concepts. Determinants of the EP-Score level and predictors of the domestic presence of breeding-sites were also explored. EP-Score percentiles for each sociodemographic group were calculated following Weighted Average method (Table II.1.1 and II.1.4). Comparisons of score medians between sociodemographic groups were made using non parametric tests: Mann-Whitney and Kruskal-Wallis (Table II.1.1, II.1.3 and II.1.4). Associations/differences with the domestic presence of breeding sites were performed using three different approaches: (i)-individual essential concepts: assessed by a chisquare test for categorical variables (Table II.1.3); (ii)-EP-Score: assessed by Weighted Averaged method and Mann-Whitney test (Table II.1.3); (iii)- Incomplete Scores (four combinations of scores covering four out of the five main topics) also assessed by Weighted Averaged method and MannWhitney test (Table II.1.5). In this latter point (iii), by filtering the residents that showed zero points regarding each of the five topics separately, four combinations of incomplete EP-Scores (from 0 to 8 points) were generated. Additionally, logistic regression models were also performed to explore socio-demographic factors that contribute to achieve, or not, an EP-Score equal to or higher than seven. The cut-off would preferably be an EP-Score equal to ten (instead of seven). However, due to the inexistence of a minimum number of individuals that have reached the maximum (EP = 10), the cut-off was adjusted until seven in order to include an enough number of individuals needed to perform the logistic regression. RESULTS A total of 1276 AEGYPTI-residents participated in the study. Out of these, only 92·6% (1182 individuals) answered the thirteen questions and were scored according to the perceptions demonstrated. All individual residences were inventoried to putative breeding sites. Table II.1.1 shows the socio-demographic characteristics of the studied population. PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 54 T ABLE II.1.4 – A SSOCIATION OF EP-I NCOMPLETE S CORES AND PRESENCE OF DOMESTIC BREEDING SITES Incomplete EP-score covered only four out of the five Essential Topics. Essential Topic excluded Residents living in houses WITH breeding-sites n ; median (P 25 -P 75 ) + Residents living in houses WITHOUT breeding-sites n ; median (P 25 -P 75 ) + p - value ‘ Medical Importance 137 ; 2·0 (2·0 – 4·0) 25 ; 3·0 (1·0 – 4·0) 0.615 Local Risk 484 ; 4·0 (3·0 – 5·0) 106 ; 4·0 (3·0 – 5·0) 0.399 Domestic Attribute 267 ; 3·0 (2·0 – 4·0) 60 ; 3·0 (2·0 – 4·0) 0.515 Mosquito Breeding 138 ; 3·0 (2·0 – 4·0) 26 ; 3·0 (1·0 – 3·0) 0.367 Control Measures 155 ; 3·0 (2·0 – 3·0) 29 ; 2·0 (1·0 – 3·0) 0.351 ‘Mann-Whitney test; + Weighted Average method S OCIO - DEMOGRAPHIC CHARACTERISTICS AND PERCEPTION DETERMINANTS All analysed socio-demographic characteristics presented significant differences in EP-scores medians (Table II.1.1). Actually, all males, residents aged 26-35 years old, people that had twelve years or more of education, individuals that live in ‘Santa Luzia’, respondents that have travelled to DEC and those that admitted to have been bitten by mosquitoes, have embraced more essential concepts than their correspondent socio-demographic groups. Following the logistic regression, four socio-demographic characteristics significantly determined a minimum of seven acknowledged essential concepts (EPScore equal to or higher than seven). These were residents’ ‘gender’, ‘municipality’, the eventuality of being ‘bitten by mosquitoes’ and above all ‘educational level’ (Additional file II.1.7). PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 55 DISCUSSION Comparing to other studies, analysis of single concept frequency 8 revealed an (apparent) very good community knowledge [140,141] . For example, almost 80% of the population recognized that «the source reduction is an effective measure for domestic aegypti-control» (Control Measure 1). However, perception evaluation based on EP-score showed that several essential concepts are still unknown by the majority of the population. Regarding topics understanding, only a few respondents completely understood each of the five topics. In all of them, a great discrepancy was found within the knowledge shown in concepts covering the same topic, predicting the presence of alleged myths/erroneous perceptions in most of the AEGYPTI-population. As suggested in Additional file II.1.3, the dissemination of part of the information can promote the advent of myths. To notice, through an anthropological view these myths are considered the real perception of the community [146]. They are here called ‘erroneous perceptions or myths’ since they oppose and contradict what, to date, is considered to be the main community vector-control practice. Sequential educational activities should take into account those myths given that they could be much harder to amend than the lack of awareness itself. Four socio-demographic determinants were described in the logistic regression results. Similarly to other studies, the education level was the most relevant determinant in the EP-Score level above 7, emphasizing the relevance of extensive health education programs to improve the healthliteracy levels [34-37][156-159]. The ‘bitten by mosquitoes’ variable 9 (stating the recognition of having been bitten by mosquitoes) also showed to be a determinant in the level of EP-Score. These suggests that measures that make the problem more ‘visible’ would be of a great impact in community awareness, especially for those who lack the allergic reaction to the bite. Determinants such as, ‘Gender’, and ‘Municipality’ should be considered in the selection of target groups/areas for further campaigns. 8 descriptive analysis 9 also mentioned as AME, reflects those who admitted/not to had been bitten by mosquitoes PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 56 Concerning the entomological survey, only putative breeding sites were inventoried. Due to the un-expected absence of rainfall during the period of the study (carried-out during the beginning of the rainy season), most of the containers were dry (Additional file II.1.8). Nevertheless, this was, to our knowledge, the sole entomological survey in a temperate region describing the most common A. aegypti’s domestic breeding sites. The most inventoried putative breeding sites were housing-components present in any patio, balcony or garden areas. An aegypti-infestation pattern was observed compatible with a clean, organized and well maintained urban environment (as schematized in Additional File II.1.9). These results contrast with the common symbols of mosquito infestation in dengue endemic regions, often related to water supply and waste disposal (tires, water tanks, etc.) [160,161,162]. ‘Santa Luzia’s municipality showed a significantly higher percentage of houses without breeding sites compared to the other two municipalities. This could be explained by a higher conscience of the A. aegypti’s presence in ‘Santa Luzia’ since it was where this mosquito first appeared. Associations found between EP-Score and presence of domestic breeding sites supported the established criteria (Table II.1.3 and II.1.4). The important and most acknowledged concepts: DA2 and CM1, per se did not correlate with the absence of breeding site removal. Yet, the EP-score level is significantly higher in respondents living in households without putative breeding sites (Table II.1.1). These results seem to support that essentialconcepts’ cumulative assimilation is needed for triggering the adoption of the aimed behaviour. Moreover, results from the Incomplete Scores revealed that none of the five topics were dispensable in the improvement of the source reduction compliance. Evidence was provided to use the EPScore analysis as an accurate tool for perception estimation. Furthermore, comparing to the alternative simple analysis of frequencies (see Table II.1.3), this tool provides deeper and more precise results to explore the community involvement. Actually, the major limitation of knowledge/perception assessments is the lack of its correlation with the adoption of proposed practices, frequently observed in similar studies PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 57 (most commonly, knowledge-attitudes-and-practices surveys) [59,112,139,143,156-158]. Methodologies that estimate awareness based on a score were already used in other surveys [59,141]. However, these approaches rarely or never focus on a specific behaviour, and almost never test understanding discrepancies. Since the adoption of different denguerelated practices (preventing, protecting, diagnosing, treatment-seeking practices, etc.) implicates the understanding of distinct concepts, behaviour-oriented approaches are much more useful to prioritize healthmessages and plan campaigns [106]. Analysis of discrepancies in the understanding has been suggested as a way to improve reliability in KAP surveys. Similar studies are now needed to confirm whether this approach is indeed more accurate to assess perceptions and more effective to promote behaviours in the community. CONCLUSIONS AND FUTURE IMPLICATIONS After seven years of coexistence with the A. aegypti, Madeira Island presents an atypical scenario of domestic infestation. Subsequent to several local educational activities, AEGYPTI-community perceptions regarding source reduction were not only insufficient, but also, inconsistent and possibly incorrect. Therefore, future educational activities addressing the essential concepts and the alleged myths may help the community in fully engaging in the proposed behaviour. However, after the experience of a dengue outbreak (2012), local population has probably altered their perception, namely in what concerns the topic ‘Local Context’. Moreover, since, no haemorrhagic clinical cases were detected in the latter outbreak, the real ‘Medical Importance’ of dengue could be still underestimated. These ideas should also be considered by those planning further educational activities on the island. As part of future actions the implementation of another questionnaire, similar to the one carried-out in this study, should be encouraged. In reality, with its recent dengue event, Madeira Island presents an exceptional opportunity to understand the effect of a disease-outbreak in a community’s awareness. Finally, findings PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 58 of this study support the use of EP-Score methodology as a more efficient tool to evaluate the community-perception regarding a specific behaviour. When further tested, this type of tool will probably prove to be of great value for other health problems, far beyond dengue prevention. ACKNOWLEDGEMENTS We thank to Fundação para a Ciência e Tecnologia (FCT) for funding this work (references:PTDC/SAL-EPI/115853/2009 and SFRH/BD/51012/2010), all the survey respondents who participated in the study, and the survey staff who did the interviews. PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 59 ADDITIONAL DATA Additional file II.1.1– A. aegypti’s distribution area (2001). Ovitrap distributions in the two inhabited island of Madeira’s archipelago: Madeira and Porto Santo (2011). Red Points correspond to positive ovitraps, Green Points correspond to negatives ones. Administrative boundaries described as «Municiplities» refer to what in the text is considered «county» or «Municipal Division» PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 60 Additional file II.1.2 – Relevance of cumulative knowledge: Exploring why a ‘higher’ level of knowledge doesn’t necessarily reflect a ‘better’ awareness Additional file II.1.3 – Myth’s appearance: Explaining an example of how a myth can appear from a partial (non-cumulative) understanding Let’s explore the awareness of two hypothetic cases, respondent number 209 and number 344. Number 209 knew that mosquitoes could transmit disease and gave “Dengue” as an example, showing that he/she had a complete notion of the Medical Importance of Mosquitoes; he/she also knew that those mosquitoes are in Santa Luzia (his/her area of residence), and recognized the possibility of a dengue outbreak in Madeira. He/she had only understood two of the five themes assessed (this knowledge corresponds to a Score of EP = 4). Case number 344, had the maximum score for four of the five analysed themes, he/she admitted more concepts than number 209. He/She simply did not admit that mosquitoes could breed inside houses (this knowledge corresponds to a Score of EP=9). Number 344 probably won’t adhere to the domestic control since he/she didn’t understand the real need of control his/her domestic area. Even though case number 344 has more essential perceptions than case number 209, none of them have the sufficient amount of knowledge to be aware of their own involvement in domestic vector control. For instance, let’s explore the meaning of (correctly) admitting that water accumulation leads to the breeding of mosquitoes, but also (erroneously) believe that “food debris can contribute to mosquito breeding”. Food debris on its own (without water accumulation) does not serve as a larvae habitat. Without the mentioning of water accumulation, this belief supports the erroneous idea that “clean places aren’t infested by mosquitoes”. As a result, people who assume their own houses as being “clean” may not f eel implicated in domestic aegypti - control. PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 61 Additional file II.1.4 – False perceptions/myths estimation through the analysis of residents’ topic understanding Concepts acknowledgement comparison Analysis of Topic’s Understanding Community Understanding / Alleged myths Concept 1 Concept 2 Medical Importance   31·9 % (377 ind.) Residents admitted that mosquitoes transmit diseases such as ‘Dengue’ (22·5%), ‘Malaria’ (9·5%), ‘Yellow fever’ (3·1%) or other mosquitoborne diseases (1·2%) or few of the latters. Residents seemed to understand the real medical importance of mosquitoes and, thus the relevance of being involved in the aegypti-control.  X 54·4 % (643 ind.) Even though admitting that mosquitoes can transmit diseases, these residents did not know what kind of diseases mosquitoes transmit. Some residents erroneously referred ‘allergies’ as mosquito-transmitted diseases (6.3%) and 4·5% mentioned other false clinical consequences such us ‘SIDA’, ‘fever’ or ‘cancer’. These residents were not aware of the relevance of being involved in the aegypti-control. Alleged Myth 2: “Mosquitoes only cause mild clinical consequences such as allergies, fever, etc.” X  Not observed X X 13·7% (162 ind.) Residents did not know that mosquitoes can transmit diseases Residents did not understand the medical importance of mosquitoes. Alleged Myth 1: “Mosquitoes do not transmit diseases” Concept 3 Concept 4 % Local Risk   15·4% (182 ind.) Residents recognized that there were mosquitoes that transmit diseases in their residential area, and, also, that there was a risk of a dengue outbreak in Madeira. Residents seemed to understand the local context they are submitted and, thus the urgency of being involved in the aegypti-control.  X 18·8 % (222 ind.) Residents recognized presence of mosquitoes that transmit diseases in their residential area; however they believed that a dengue outbreak will not emerge in the island. Allegedly some made this confusion because they did not recognize dengue as a mosquito-borne disease (20·3%). Eventually some residents could think that Madeira is “protected” since those kind of severe epidemic diseases historically never occurred in temperate countries. These residents were not aware of the urgency of being involved in the aegypti-control. Alleged myths 3 and 4: (i) – “Dengue is not a mosquito-borne disease”; (ii) - “Dengue only occurs in tropical/non-developed countries”. X  15·9 % (188 ind.) Residents did not recognize the presence of mosquitoes, in their residential area, that can transmit diseases; but admitted that a dengue outbreak can emerge in the island. These residents did not have a correct notion of the aegypti’s distribution area. Since 22·2% out of these group referred not be ‘bitten by mosquitoes’, they could believe that they are at lower risk of being infected in an eventual outbreak. Residents did not understand the risk they are subjected to and neither the urgency of being involved in the aegypti-control. Alleged myths 5 and 6: (i) -“Mosquitoes are allocated in a specific area and are not able to spread through the island”; (ii) - “Since I do not feel the byte, I am not a risk of being bitten/infected”. X X 49·9 % (590 ind.) Residents did not recognize mosquitoes that transmit diseases in their residential area neither the possibility of a dengue outbreak in the island. Residents did not understand the risk they are subjected to neither the urgency of being involved in the aegypti-control. Alleged Myth 7 : “Madeira’s residents are not at risk” Concept 7 Concept 8 % Domestic Attribute   20·0 % (236 ind.) Residents know that mosquitoes can breed inside houses and recognized that domestic aegypti-control do have impact in the reduction of aegypti-population. Residents seemed to understand the domestic attribute of the aegypti-control and, thus why community is the key intervenient in the aegypticontrol.  X 6·4% (76 ind.) Residents know that mosquitoes can breed inside houses but they did not believe that the domestic aegypti-control have impact in the reduction of the aegypti’s population. They probably believed that other intervenients have much more impact in the reduction of the aegypti’s population. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypticontrol. Alleged Myth 8: “Local health authorities are the key intervenient in the control of mosquitoes”. PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 62 X  45·9 % (543 ind.) Mosquitoes cannot breed inside houses but domestic aegypti-control does have impact in the reduction of aegypti-population in the neighborhood. Those respondents believed in their role in domestic aegypti-control but did not understood why that control has an impact. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypticontrol. Alleged Myth 9: “Other protective measures can control mosquitoes”. X X 27·7% (327 ind.) Residents do not know that mosquitoes transmit disease, neither that their involvement have an impact in the control of mosquitoes. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypticontrol. Alleged Myth 10: “I am (Community is) not an intervenient in the aegypti-control”. Concept 5 Concept 6 % Mosquito Breeding   27· 6 % (326 ind.) Residents only identified water - containers (and not other false issues) as mosquitoes’ breeding inducers. Residents seemed to understand where do mosquito breed and, thus the need of the aegypticontrol activities.  X 46·5% (550 ind.) Residents identified water-containers but also other false issues (food debris and pets) as mosquitoes’ breeding inducers . These residents did not comprehend what lead to the breeding of new mosquitoes and, thus did not understand the proposed measures to control them. Residents seemed to not understand where mosquitoes breed and neither the need of the aegypti-control activities. Alleged Myths 11 and 12: “Clean houses or houses without pets/animals do not have mosquitoes” or “Clean people did not need to be involved in mosquito control”. X  12·0 (142 ind.) Residents did not identify water - containers neither other false issues (food debris and pets) as mosquitoes’ breeding inducers. These residents did not know where mosquitoes breed. Residents seemed to not understand where mosquitoes breed and neither the need of the aegypti-control activities. X X 13·9 (164 ind.) Residents did not identify water-containers but did identify other false issues (food debris and pets) as mosquitoes’ breeding inducers Residents are completely mistaken regarding mosquitoes breeding, thus did not understand the need of the aegypti-control activities. Alleged Myths 11 and 12: “Clean houses or houses without pets/animals do not have mosquitoes” or “Clean people did not need to be involved in mosquito control”. Concept 9 Concept 10 % Control Measures   13· 0 (154 ind.) Residents only recognized water - containers removal (and not other false measures) as “effective to control mosquitoes” Residents seemed to recognize effective control measures and, thus understand how the domestic aegypti-control should be done.  X 64·1 (758 ind.) Residents recognized water-containers removal and also other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes” Residents seemed to not be focused on effective control measures and, thus did not understand how the domestic aegypti-control should be done. Alleged Myth 13: “Using insecticides or the flyswatter, I am already contributing to control the aegypti-mosquito” X  7·3 (86 ind.) Residents did not recognize water - containers removal neither other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes”. These residents did not know how to control mosquitoes. Residents not recognized effective control measures and, thus did not understand how domestic aegypti-control should be done. X X 15·6 (184 ind.) Residents recognized water-containers removal and also other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes” Residents seemed to not be focused on effective control measures and thus did not understand how the domestic aegypti-control should be done. Alleged Myth 13: “By using protective measures (such as insecticides or flyswatter), I am already contributing to control the aegypti-mosquito” PRE-OUTBREAK WORK: S TUDY 1, P ART 1 (II.1) 63 Additional file II.1.5 – Domestic breeding sites: Percentage (%) of inquired residents living in houses with each type of breeding site (n Total =1276) POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 70 T ABLE II.2.1 – T OPICS COVERED IN THE QUESTIONNAIRE , RESPECTIVE QUESTIONS , FIGURES AND TABLES Subjects Question Results presented Concerns about generic mosquitoes “In your residential area, do mosquitoes worry / concern you in anyway, or not, or maybe you never thought about it? “ Figure 1 – Community concern IF YES “to what level do you worry about mosquitoes?” (subsequent from the previous question) Figure 2 – Level of concern IF YES “Why is it a cause of concern?” (subsequent from the first question of this group) Figure 3 – Causes of concern Local Mosquitoes “How do you describe it?” (subsequent from the previous question) Figure 4 – Vector description “In your opinion, what induces mosquito breeding? e)-Are there other causes?” IF YES, “Which ones?” Figure 5 – Other mosquito breeding inducers Dengue fever “How can one catch Dengue disease?” Figure 6 – Dengue’s mode of transmission “Which are the symptoms of who becomes sick with Dengue?“ Figure 7 – Dengue’s symptoms “Do you know any country in the world where Dengue disease exists?“ Table 1 – Dengue endemic countries Mobility to endemic countries “Have you ever travelled or lived in any country of Africa Americas, Australia or Asia?” Table 2 - Most visited endemic countries (Answers were re-categorized including together people, families or friends that travelled or lived in any endemic country/region) IF YES, “Which ones? When did you come back? (subsequent from the previous question) IF NO, “Do you have family or friends that have travelled or lived in any country of Africa Americas, Australia or Asia?” Mediabased strategies “Have you ever heard or read about “Dengue” disease? IF YES, Where? a) Newspapers or Magazines, b) Flyers, c) Street Posters, d) Television, e) Radio” Figure 8 – Media talking about dengue “Other sources? Which ones?” (subsequent from the previous question) Figure 9 – Other media talking about dengue RESULTS A total of 1276 individuals answered the above described questions. The study sample was represented by 40.6% male and 59.4% female subjects. The average age in this study was 49.9 years (SD=19.04 years, min=18 POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 71 years and max=91 years). In what concerns the respondents’ level of education, 5.9% did not study at all, 38.2% had only completed the fourth grade, 22.5% studied until the ninth grade, 17.7% finished high school or similar and 15.7% graduated or had a master or postgraduate degree. There were 311 individuals (24.4%) that admitted to have already travelled to dengue endemic countries (DEC) and 319 individuals (25.0%) that never travelled to those regions but that have friends or family that did it (out of 1276 in both cases). Most of the respondents revealed to be concerned about mosquitoes, the majority of those declared to be «very» or «greatly» concerned (Figures II.2.1 and II.2.2). The causes of concern mostly pointed out were related to: ‘allergies’, ‘health’ or ‘family’ (Figure II.2.3). F IGURE II.2.1 – C OMMUNITY CONCERN ABOUT MOSQUITOES Proportion of residents that admitted to be or not to be concerned about mosquitoes, and also those that admitted never having thought about it (%, n= 1276) (note: 6 did not answered) F IGURE II.2.2 – L EVEL OF CONCERN Proportion of residents, by level of concern, which admitted to be concerned about mosquitoes (%, n= 770) very POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 72 F IGURE II.2.3 – C AUSES OF CONCERN Proportion of residents that have admitted cause(s) of their concern about mosquitoes (%, n= 797 – those that admitted to be concerned) Out of those who recognized the presence of vectors in their residential area (39.4%, 412/1045) the majority (58.2%, 240/412) either admitted not knowing how to describe A. aegypti or described it incorrectly. Moreover, the remaining, who described A. aegypti appropriately, did so by covering five main characteristics: its «darkness», its «small size», its «long legs», its «white strips», «white dots» or «white legs» and some behavioural features. The most frequently mentioned adjectives were generic ones, such as «small» or «dark», while the more specific feature «white strips» was the less mentioned one (Figure II.2.4). F IGURE II.2.4 – V ECTOR DESCRIPTION Most frequently mentioned vector´s characteristics. Proportion of residents (out of those who recognized A. aegypti presence in their local residence) that stated one or more A. aegypti features (%, n= 412). POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 73 Out of those who did not recognize the presence of vectors and those who believed that mosquitoes could not transmit diseases (who sum-up to a total of 864 individuals), the majority (54.5%), when shown the actual A. aegypti mosquito, admitted to having already seen it. However, out of these majority of 54.5% individuals, very few correctly identified it as «Aedes aegypti» (5.1%, 24/471). The remaining mentioned other broad names such as «mosquito» (31.0%, 146/471) or «Santa Luzia’s mosquito» (12.7%, 60/471). Figure II.2.5 presents the answers given by residents when asked about the eventual existence of other mosquito breeding inducers 13 not mentioned in the multiple choice question. The term “other” refers to mosquito breeding inducers separate from the ones mentioned in the previous question analysed in the sub-chapter II.1 (pets, food debris, plants or water containers). A total of 229 individuals believed that mosquito breeding can be induced by other inducers. According to the residents’ answers and to the categorization applied, these other mosquito breeding inducers were: ‘generic water accumulation’ (referred by 60.4%), ‘absence of hygiene’ (26.9%), ‘weather’ (10.7%), ‘imported species’ (8.6%), or ‘empty houses/common land’ (3.0%), as described in Figure II.2.5. Answers included in the ‘generic water accumulation’ category refer to a range of different meanings such as, «wells», «lakes», «streams», «ponds», «wateraccumulation on deck» and «flower-pot dishes». 13 situations or occurrences that will promote mosquito development POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 74 F IGURE II.2.5 – O THER FACTORS OR SITUATIONS THAT PROMOTE MOSQUITO BREEDING Resident´s beliefs (one or more) regarding other mosquito breeding inducers stated as an answer to an open question (%, n=197 – all residents that answered to this question). Considering the media-based tools that transmitted dengue-related information, ‘television’ was the most efficient in spreading the term «dengue» within AEGYPTI’s residents, when selected in a multiple-choice question (Figure II.2.6). However, through an open answer question, the ‘web’, ‘travels to foreign countries’ and ‘word-of-mouth’ were also mentioned means of communication in «dengue» dissemination (Figure II.2.7. F IGURE II.2.6 – M EDIA - BASED STRATEGIES Proportion of residents that heard/read about dengue in different mass communication sources (%, n= 770) POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 75 F IGURE II.2.7 – O THER COMMUNICATION STRATEGIES USED TO DIFFUSE MESSAGES ABOUT DENGUE Proportion of residents that heard/read about dengue through other(s) communication sources (%, n= 340 – all residents that answered to this question) In what concerns the dengue disease awareness, 63.1% had already heard about «dengue». Out of those, 78.4% recognized that dengue is transmitted by some kind of mosquito activity and only 0.1% knew that the transmission is done through the byte of an infected mosquito (Figure II.2.8). F IGURE II.2.8 – D ENGUE ’ S MODE OF TRANSMISSION Residents beliefs regarding the dengue mode of transmission (%, n=798 - those that have recognized the term ‘dengue’ were the total inquired about its transmission) When asked about dengue symptoms, approximately half of those inquired mentioned «fever» (54.3%). There were only 5.4% who enumerated the POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 76 more specific dengue symptoms, such as retro-orbital or muscle or bone pain. Moreover, a higher proportion (6.8%) erroneously mentioned «allergic reactions» as dengue symptom (Figure II.2.9). F IGURE II.2.9 – D ENGUE SYMPTOMS Residents’ beliefs regarding symptoms of dengue infection (%, n= 798those that have recognized ‘dengue’ are the total inquired about symptoms) * according to WHO/CDC: headache, joints pain, eye (retro orbital) pain, muscle and/or bone pain, rash Left group - symptoms of DF/DHF typical clinical feature; Left group (darker) – symptom from the DHF typical clinical feature; Central group - symptoms which are not from the typical clinical feature of the dengue/dengue hemorrhagic syndrome; Right group - mosquito bite consequence (not related to DF/DHF When asked to identify DEC, almost half of those who recognized «dengue» recognized Brazil as a DEC (45.5%). Angola was the second most mentioned country (12.8%). A total of 24.1% individuals, more than the double of those that mentioned Venezuela, had generally cited the African continent in the same question (Table II.2.2). Other symptoms from the typical clinical feature* Haemorrhage V omiting D iarrhoea A llergic reaction Fever POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 77 T ABLE II.2.2 – C ITED ENDEMIC COUNTRIES The three most referred endemic countries, their last year of dengue outbreak and correspondent number of cases. Country n % Year of the last outbreak* No. of cases Brazil 363 45.5 2011 [154] 56882 Angola 102 12.8 1980 [44] ? Venezuela 82 10.3 2010 [155] 124931 Angola + Mozambique + ‘Africa’ ** 24.1 n/a n/a * at the date of this survey was performed; n (total) =798 - those that have recognized ‘dengue’ are the total inquired about endemic countries. ** considering altogether residents mentioning “Angola”, “Mozambique” or “Africa”. Brazil and Venezuela have been the first two most visited endemic countries in the past five years, by the inquired individuals or their relatives/friends (Table II.2.3). Angola was the third most visited DEC, but considering those who had also mentioned their returning date, the majority of those who travelled to Angola did it more than 30 years before the study had been performed. An association was found between the respondents who visited Angola and those who have mentioned it as a DEC (p<0.001). POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 78 T ABLE II.2.3 – M OST VISITED ENDEMIC COUNTRIES Left side –endemic countries most visited in any period of time, travels done by the respondent, their family or friends (n =630 – all that have travelled or that have family/friends that have travelled to those countries) ; Right side - Most visited endemic countries in the last five years regarding travels done by the respondent, their family or friends. (n=119 – individuals who have reported the returning date of travels to endemic countries). All returning dates In the last 5 years Country n % Country n % Venezuela 307 48.7 Brazil 47 39.5 Brazil 194 30.8 Venezuela 47 37.8 Angola 69 11.0 Others* 32 26.8 USA 54 8.6 USA 10 8.4 Australia 52 8.3 Cape verde 9 7.6 Mozambique 49 7.8 Angola 4 3.4 * Argentina, Australia, Burkina Faso, Cape Verde, Caribbean archipelago, China, Colombia, Hawaii archipelago, India, Macau, Malaysia, Mexico, Mozambique, São Tome and Príncipe, Thailand,Vietnam DISCUSSION AND CONCLUSIONS The community’s perceptions assessed revealed an incipient knowledge and awareness regarding the explored subjects: A. aegypti’s physical feature, mosquito breeding inducers, modes of dengue transmission and dengue-syndrome symptoms. Even though some of these questions had a high percentage of correct answers, only generic perceptions (not directly related to DF or A. aegypti) were revealed. These results could be explained taking into account that this region did not have any historical experience of autochthonous dengue cases. Moreover, the last European dengue outbreak occurred 84 years ago [35]. Residents who describe A. aegypti as «dark» or «small» most probably are not able to identify it or distinguish it from other mosquitoes in their daily routine. Very few respondents revealed to be ready for recognizing dengue symptoms. The most mentioned dengue symptom, fever, can also be a symptom of flu, cold, or any other infection. Those who mentioned «allergic reactions» reinforced the misperception of dengue health risks (myth 2, described in Table II.1.2). As a matter of fact, 96.3% of those who mentioned «allergic reaction» did not mention any other dengue-specific POST-OUTBREAK WORK: S TUDY 1, P ART 2 (II.2) 79 symptom. Dengue implicates a mosquito bite which may provoke an allergic reaction. Even though, if one does not mention other dengue specific symptoms, one is probably referring to the nuisance that a mosquito bite could bring even when uninfected, thus undervaluing the clinical impact of a dengue-infective bite. Findings about the dengue mode of transmission support the latter consideration. In fact, the majority of people associated dengue with mosquito, but not with an infection, also corroborating the presence of myths 1 and 2 (Table II.1.2). As expected from an open question, mosquito breeding inducer results (Figure II.2.5) did not describe the weight of each answer, but rather suggested ideas or beliefs that had not been mentioned in the close multiple-choice answers. ‘Absence of hygiene’, ‘weather’ and ‘empty houses/common lands’, are examples of categories of the answers mentioned by the community as situations that promote mosquito breeding, these ideas or beliefs. By believing in the existence of these inducers residents may lose compliance to the removal of actual aegyptiinducers (water accumulation in domestic containers). In fact, the absence of hygiene could be an inducer of A. aegypti mosquito proliferation, since rubbish and garbage could accumulate water, especially when spread in outdoor areas. However, this is not the most common case of Madeira, as shown in the findings of the entomological characterization (Additional File II.1.5). The breeding site ‘rubbish (on the floor)’ was found ten times less than the most frequently found ‘flower pot dishes’ presented in 52.7% of the infested area houses. Therefore, the belief in the absence of hygiene as an inducer of A. aegypti proliferation in Madeira is not entirely correct, and can mislead those who believe in it, by suggesting the erroneous ideas that «clean houses do not have mosquitoes» and «people living in clean houses have nothing to do concerning mosquitoes» (myth 11 and 12, described in Table II.1.2). The role of temperature, humidity and rainfall on mosquito survival and ecology is well known. Therefore, the belief that the ‘weather’ is a mosquito breeding inducer is partially true. However, it is also known that other non-climate factors, such as the domestic environment and the close 86 POST-OUTBREAK WORK: S TUDY 2 (III.1) 87 ABSTRACT The ability to effectively promote behaviours is more and more relevant to attain and maintain a good individual and collective health status. For the last years several models and theories have been proposed to explain behaviour-change, covering two main approaches for (healthier) decisionmaking: one analytical/logical and one experiential/emotional/intuitive. According to the former approach, community perception assessments frequently measure cognitive issues. However, few studies explore how past experiences impact public views in particular health-risk contexts, impeding an integral and dynamic understanding of the behaviour change process. Before Madeira’s first dengue outbreak (in 2012), community perceptions regarding domestic source reduction, were assessed. This offered a unique opportunity to assess and compare community perceptions before and after the experience of the dengue outbreak, and this was the aim of this study. A cross-sectional survey was performed within female residents at the most aegypti-infested areas. Perceptions regarding the main denguepreventive behaviour were assessed according to the Essential Perception (EP)-analysis tool. A matching process (or randomised block design) was used in order to pair individuals from studies performed before and after the outbreak, ensuring homogeneity in 6 determinants variables. After the outbreak, there were more female residents that assimilated the concepts considered to be essential to understand the proposed behaviour. Nevertheless, no significant difference was observed in the number of female residents who achieved the defined ‘minimal understanding’. Moreover, as observed before the outbreak, most of the population (95.5%) believed at least in one of the identified myths. Even though, the number and frequency of myths did not change significantly, some myths disappeared and others, which were absent before the outbreak, had appeared. In the present study we were able to quantify how the experience of a risk event influenced the perception regarding a dengue-preventive behaviour. POST-OUTBREAK WORK: S TUDY 2 (III.1) 88 Results have shown that the assimilation of several relevant concepts improved. This experience have also surprisingly led to the appearance of new myths within the population, apart from the general decrease of the previous myths’ frequency. Monitoring public perceptions is therefore crucial to make preventing dengue campaigns updated and, thus, worthy. Lessons from this work can be useful not only for local authorities but also for dengue-related professionals and researchers in public-health, decision-making or experience-integration. BACKGROUND Most of the worldwide major causes of death (MCD) in 2011, rely on behaviour changes for their prevention [158]. In fact, increasing physical activity, fruits/vegetables intake, hand-washing, use of condoms and decreasing not only fat, salt and sugar intake but also smoking habits, are crucial in the control of heart disease (1 st MCD), stroke (2 nd MCD), chronic obstructive lung disease (4 th MCD), diarrhoea (5 th MCD), HIV (6 th MCD), or diabetes (8 th MCD). Behaviour changes are increasingly relevant to attain and maintain a good health status, especially when facing health threats for which there is no efficient or timely treatment. This is the case for dengue fever that, such as other mosquito-borne diseases, requires a good compliance to certain preventive, protective or therapeutic actions. Moreover, since there is no vaccine nor treatment for dengue fever, neither 100% effective insecticides, community behaviours have a huge impact on its prevention and control [21]. It is still not widely understood how to effectively promote behaviour changes [106]. In fact, during several decades many behaviour impact campaigns have shown to be fruitless. In the last 50 years, extensive literature, presented theoretical models that tried to clarify cognitive ways for (healthier) behaviour acquisition [63,76]. More recently, the concept of ‘past experiences’ has been stated as also being crucial in determining (healthier) decision-making. Countless authors claim that due to the type of emotions and intuition that they produce, ‘past experiences’ can stoutly POST-OUTBREAK WORK: S TUDY 2 (III.1) 89 encourage or discourage a particular action [10-16]. Altogether, these contributions seem to present two different approaches by which humans perceive decision-making and then make decisions: one analytical and one experiential [14,15]. In order to improve the efficacy of the behaviourpromoting messages, these authors strongly suggest that messages should be meaningful as well as affectively adequate for the targeted community. This way, the assessment of community’s cognitive and emotional perceptions, is hence useful in the guiding of effective healthseeking messages. However, few studies explore emotional experiencedriven perceptions but rather frequently only focus on the assessment of the cognitive ones [71]. Some evidence has suggested that experience can influence public perceptions and reactions in two ways [72,73]. On one hand, it can overestimate the risk perception [17,18] (i.e. alert-feeling, also mentioned as ‘availability bias’ [163]) and consequently, promote protective/preventive actions. On the other hand, it can underestimate the risk perception [2022] (i.e. habituation effect also mentioned as ‘gambler’s fallacy’ [163]) and hence, discourage protective/preventive actions. However, very few studies have explored this issue in real situations. Surveys that explore in what way and how much past experiences impact public views or actions in particular health-risk contexts are of great need [103]. Besides the scientific interest of scrutinizing the complex process of (healthier) decision-making, the monitoring of public perceptions and behaviours contributes to the continuous and adequate update of the behaviourpromoting messages concerning their (rational and emotional) content. This is the case of any chronic and endemic disease, where the (health) risk is maintained during time such as dengue risky, endemic and endemic areas [103]. Madeira archipelago was colonized by a dengue vector species, Aedes aegypti, in 2005 and suffered the first dengue outbreak in 2012 [139]. Community perception regarding preventive behaviours (domestic source reduction) was assessed and described in detail before the outbreak had been declared (presented in sub-chapter II.1) [166]. With the end of the POST-OUTBREAK WORK: S TUDY 2 (III.1) 90 outbreak, a unique opportunity to explore and compare community perception before and after the outbreak had appeared. This was the aim of this study, which constitute the first report on the effect of an outbreak experience on community perceptions regarding a specific vector-borne disease. METHODS To our knowledge, the results presented here constitute the first report on the effect of a disease outbreak experience on community perceptions. In order to ensure an accurate comparison between public perceptions before and after the dengue outbreak in Madeira Island, present survey tried to follow as much as possible the methodology used in the prior-tothe outbreak survey (herein mentioned as PRE-outbreak study) [166]. Therefore, the tool used in the assessment of the community perceptions was maintained, i.e., an ‘Essential-Perception analysis’ (described in the homonymous sub-section). However, due to ethic, time and logistic constrains implicit in the preparation and implementation of this survey during the outbreak and just after it, adjustments in the size of the studied area and in the sampling methodology, were introduced as explained in ‘Studied population’ sub-section. In order to overcome those constrains and guarantee an unbiased comparison, populations that were surveyed in both PRE/POST-outbreak studies were matched according to critical sociodemographic variables, as described in sub-section ‘Matching Process’. The cross-sectional survey performed after the outbreak will be subsequently mentioned as POST-outbreak study. S TUDIED POPULATION Due to the mentioned limitations, the studied area in the POST-outbreak study was not the same as the in PRE-outbreak one (Additional file 1 a) , b) ). Exclusively the urban part of the most aegypti-infested area was selected 15 . 15 PRE-outbreak study area was divided in two zones according to population density and economic activities. The first, with almost half of the demographic population of the other and a predominance of POST-OUTBREAK WORK: S TUDY 2 (III.1) 91 Facing the impossibility of including both urban and rural parts, the urban part (Funchal District) was preferred based on two main reasons: (i) it corresponded to the area of maximum prevalence of dengue-cases during the outbreak (see Additional file III.1.1 c) ); and (ii) it includes the capital city of the archipelago, Funchal, and thus an important point of aegyptidispersion. In order to decrease the period of data collection, residents of part of ‘Sé’, ‘Santa Maria Maior’ and ‘Imaculado Coração de Maria’ municipalities were also included, besides those considered in the PREoutbreak study (‘São Pedro’ and ‘Santa Luzia’). The studied area limits were defined according to 2012 most aegypti-infested area in order to ensure a homogeneous level of natural exposure to the A. aegypti among the studied residents (Additional File III.1.1 d) ). Analysis of the demographic data of the extra-included areas confirm that there are no significant differences between these and the previously studied, in what concerns the two critical socio-demographic determinants: age groups and education levels (Additional File III.1.2). The geographic area covered in the present study will be mentioned as ‘Extended-AEGYPTI area’ and consists of five Funchal’s municipalities that belong to the most aegypti-infested area (Additional file III.1.1 b)). An intentional sample of female subjects, residents in the ExtendedAEGYPTI area, aged 18 years old or over, and who didn’t integrate the previous PRE-outbreak survey was selected from customers of central hairdressers and pharmacies, placed in the selected area. All women who entered in the establishment and who met the inclusion criteria were invited to participate. Three reasons relied on the women preference: (i) before the outbreak they were significantly less aware to domestic source reduction than men (Additional file III.1.3); (ii) they are the majority within the studied population [144]; (iii) women above 15 years-old were the age/gender-group more affected by the disease during the outbreak [167]; and (iv) culturally, in Madeira Island, women are more related to the main dengue-preventive behaviour proposed than men do (see details about the fishing activities is herein called as rural part (Câmara de Lobos) and the other is considered the urban one (Funchal). REF POST-OUTBREAK WORK: S TUDY 2 (III.1) 92 behaviour proposed in ‘Essential-Perception’ subsection). The type of establishment were chosen in order to allow the study to cover the most possible heterogeneous women sample, in what concerns their age groups, education levels and socio-economic background. Two establishments of each service were chosen to participate in study, placed in the east and west boundaries of the studied area to promote participation of women from all the included municipalities. A sample size of 157 subjects was required to detect a difference of one point in the level of perception, fulfilling the objectives of this study - 95% confidence level and 80% of power (Additional file III.1.4 a) ) [168]. A maximum variation of the score, 0-10 was assumed, based on what was observed in the PRE-outbreak study [166]. This sample size was inflated in 30% to account for incomplete interviews. Q UESTIONNAIRE A cross-sectional survey was performed to assess residents’ perceptions through face-to-face interviews. During the interview, a questionnaire comprising 21 questions was applied, covering dengue-preventive issues and personal-socio-demographic characteristics. In agreement with what was inquired in the PRE-outbreak study, questionnaire covered five main topics: ‘Medical Importance’ (two questions), ‘Local Risk’ (two questions), ‘Domestic Attribute’ (three questions), ‘Mosquito Breeding’ (three questions) and ‘Control Measures’ (three questions) [166]. The survey was performed by trained personnel from the local authority-IASAUDE, from 22 nd of March until 16 th of April, 2013. In each establishment (pharmacies/hairdressers), interviews were performed during a Monday-toSaturday week, between 9am and 7pm (according to establishments’ opening hours). Before data collection, establishments’ managers/participants gave their written/oral informed consent respectively. Previous to the beginning of this survey, the questionnaire was pre-tested in a non-selected establishment placed in the selected area. The study was approved by Instituto de Higiene e Medicina Tropical Ethics POST-OUTBREAK WORK: S TUDY 2 (III.1) 93 Committee, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon (reference: 09-2013-TD). M ATCHING P ROCESS Populations studied in both PRE/POST-outbreak surveys were matched into pairs, ensuring homogeneity in six critical socio-demographic variables. Resulting pairs of individuals were equal in (or “blocked” on) gender, education level, age group, municipal division, travels to DEC and admitted mosquito exposure (AME) variables, which were already shown to be determinants to the individual perception [166]. This sampling methodology can also be called as randomized block design, and the latter variables as blocking factors [169]. Since no significant differences were observed in the perception of consecutive ten-year age group and municipalities from Funchal, matchings were adjusted (Additional File III.1.5). Age groups covered individuals with a maximum difference of age equal to 20 years old, and residents of different municipalities of a unique district, were considered as belonging to the same ‘municipal division’. For comparative purposes, an alternative matching without adjustments were also performed (basic matching). The matching models used (basic and adjusted) were built in Excel (Microsoft Office, Windows 8), and guaranteed that individuals were randomly selected within those that were personalsocio-demographically equivalent. Moreover, it was optimized in order to re-include all the non-selected individuals in the subsequent matching rounds. E SSENTIAL - P ERCEPTION ANALYSIS (P ERCEPTION E VALUATION ) The assessment of the community perception was performed using the Essential-Perception analysis (EP-analysis), as described in the PREoutbreak study [166]. Essential-Perception analysis assesses community perception regarding a particular behaviour proposal: the domestic A. aegypti’s source reduction, considered the most critical dengue-preventive practice by the World Health Organization [124]. Basically it corresponds to the elimination (emptying, covering or removing) of water-containers POST-OUTBREAK WORK: S TUDY 2 (III.1) 94 present inside or around residential buildings. In EP-analysis’ theoretical, there are ten essential concepts which assimilation by individuals is needed to ensure the minimal understanding of the proposed behaviour (Table III.1.1). Essential-Perception analysis allows the characterization and estimation of the community’s perceptions through four different approaches, all of them used here: (i) score of Essential-Perception, (ii) concept assimilation, (iii) topic understanding and (iv) myth identification and estimation. The first measures the number of concepts that were correctly assimilated (out of those defined to be ‘Essential’) by each individual, and how far is the studied population from the complete ‘Essential Perception’ (EP-Score = 10). The second describes how much those ‘essential’ concepts were assimilated or not-assimilated by the community. The third, organizes the ‘essential concepts’ in topics and describes how topics are/not being understood. Residents who have acknowledged both topic-related concepts are considered as having completely understood the topic, the acknowledgement of only one out of the two topic-related concepts is considered as a partial understanding of the topic, and residents who did not perceive any of the two topic-related concepts are considered as not having understood the topic. Finally the fourth, by analysing the concept assimilation, identifies erroneous beliefs, that may persist in the community, herein mentioned as ‘myths’, and estimate their putative frequency in the studied population. POST-OUTBREAK WORK: S TUDY 2 (III.1) 95 T ABLE III.1.1: L IST OF TEN CONCEPTS DEFINED AS ESSENTIAL WITHIN EPANALYSIS Essential Topic Essential Concepts Medical Importance (MI) MI1-conceptTransmission of disease through mosquitoes (bite) MI2-concept – Example of mosquito-borne diseases Local Risk (LR) LR1-concept - Presence of vector-mosquitoes in their own residential area LR2-concept - High possibility of a dengue outbreak in Madeira Domestic Attribute (DA) DA1-concept - Eventuality of indoor mosquito-breeding DA2-concept - Impact of domestic vector control Mosquito Breeding (MB) MB1-concept - Role of water-containers as breeding contributors MB2-concept – False role of ‘pets’ or ‘food debris’ as breeding contributors Control Measures (CM) CM1-concept – Source reduction as an effective domestic aegypti-control measure CM2-concept - ‘Insecticide application’ or ‘use of a flyswatter’ as an erroneous measure for the domestic aegypti-control’) S TATISTICAL A NALYSIS (T EST STATISTICS ) All collected information was introduced and records were double-checked. Statistical analysis was performed using Excel (Microsoft Office, Windows 8) and Statistical Package for Social Sciences 19.0 (SPSS, Inc., Chicago, IL, USA). Answers obtained from the questionnaire were re-coded to obtain other categorical variables implicit in the EP-analysis. Comparisons of EPscore medians between populations from PRE/POST-outbreak studies were made using the non-parametric Wilcoxon Test, after ensuring its normal distribution through Kolmogrov-Smirnov test; (Table III.1.3). Additionally, the number of individuals that achieved an EP-score equal to or higher than seven (EP-score≥7) was compared between both studies. Differences were tested using the McNemar Test (Table III.1.4). This cut-off was chosen due to the lack of subjects that achieved an EP-score equal to ten (EP-score = 10). In order to confirm the methodology used during the matching process, comparisons between total and paired samples (in both PRE/POST-outbreak studies) were performed concerning their EP-score and POST-OUTBREAK WORK: S TUDY 2 (III.1) 102 F IGURE III.1.3: P ERCENTAGE OF RESIDENTS THAT HAVE ‘ UNDERSTOOD ’, ‘ PARTIALLY UNDERSTOOD ’ OR ‘ NOT UNDERSTOOD ’ EACH OF THE FIVE STUDIED TOPICS . For Figure simplification, topics were abbreviated to their name initials: Medical Importance (MI), Local Risk (LC), Domestic Attribute (DA), Mosquito Breeding (MB), Control Measures (CM). ‘PRE’ and ‘POST’ represents PRE-outbreak study and POST-outbreak study. Myth identification and estimation Based on the thirteen myths/beliefs that were identified in the PREoutbreak study, an updated list is suggested in Table III.1.3, with myths identified after the outbreak. The frequency of each believed myth were (re-)calculated in Additional file III.1.6 and are also presented in Table III.1.3. Out of the thirteen alleged myths identified in the PRE-outbreak study, some had most likely disappeared after the outbreak. This was what happened with the myths: «dengue is not a mosquito-borne disease» or «dengue only occur in tropical/non-developed countries» (Table III.1.3). However, new beliefs emerged after the end of the outbreak, such as the idea that Madeira is protected from a second dengue outbreak (alleged myths 3 and 4). This is suggested to be believed by the majority of the female community (54.6%). According to the myth analysis, after the outbreak each female resident believed, on average, in three out of the twelve myths, less than the four myths out of thirteen believed by the average of the residents before the outbreak. Most of them believed at POST-OUTBREAK WORK: S TUDY 2 (III.1) 103 least in one myth either before or after the outbreak (99.5% and 95.5%, respectively). After the outbreak, the most disseminated alleged myth, found in 62.5% of the paired sample, was that «clean houses or houses without animals do not have mosquitoes» or «people living in these houses have nothing to do concerning the control of mosquitoes». POST-OUTBREAK WORK: S TUDY 2 (III.1) 104 T ABLE III.1.3: ALLEGED MYTHS IN BOTH PRE AND POST STUDIES AND RESPECTIVE FREQUENCIES Myths derived from PRE-outbreak study were renumbered PREOUTBREAK STUDY POSTOUTBREAK STUDY D IFFERENCE E SSENTIAL T OPICS Old / New No. A LLEGED M YTHS n (%) n (%) M EDICAL I MPORTANCE M YTH 1 ‘Mosquitoes do not transmit diseases’ 11 (12.5) 5 (5.7) ↓ M YTH 2 ‘Mosquitoes only cause mild clinical consequences such as allergies, fever, etc.’ 45 (51.1) 21 (23.9) ↓↓ L OCAL R ISK M YTHS 3 AND 4 ‘Dengue is not a mosquito-borne disease’ and/or “Dengue only occur in tropical/non-developed countries” 14 (15.9) (disappear ed) M YTH 3 ‘Dengue will not occur again in Madeira, it is very not likely‘ 32 (36.4) (new) M YTHS 5 AND 6 (i) ‘Since I do not feel the byte, I am not at risk of being bitten/infected’; (ii) “Mosquitoes are allocated in a specific area and are not able to spread through my municipality’ 16 (18.2) 8 (9.1) ↓ M YTHS 4 AND 5 M YTH 7 ‘Madeira’s residents are not at risk‘ 39 (44.3) (disappear ed) M YTH 6 ‘Dengue/A. aegypti was, finally, eradicated‘ 16 (18.2) (new) D OMESTIC A TTRIBUTE M YTH 7 ( MYTH 8) ‘Local health authorities are the key intervenient in the control of mosquitoes‘ 7 (8.0) 11 (12.5) ↑ M YTH 8* ( MYTH 9) ‘Insecticides or other protective measures can control mosquitoes‘ 36 (40.9)* 17 (19.3)* ↓↓ M YTH 9 ( MYTH 10) ‘I am (Community is) not an intervenient in the aegypti-control‘ 24 (27.3) 11 (12.5) ↓↓↓ M OSQUITO B REEDING M YTHS 10 AND 11 ( MYTHS 11 AND 12) ‘Clean houses or houses without animals do not have mosquitoes‘ and/or ‘Clean people have nothing to do concerning the control of mosquitoes‘ 53 (60.2) 54 (61.4) = C ONTROL M EASURES M YTH 12* ( MYTH 13) ‘By the usage of insecticides and/or flyswatter, I am already contributing to the aegypti-control‘ 71 (80.7)* 54 (61.4)* ↓ Myths 8 and 12 cover the same idea and if added, reveal a total of 67.0 % of the residents feeling that by the usage of protective measures, they are already contributing to the aegypti-control. (↑) Differences of 5-20 percentage points (↑↑) Differences of 20-30 percentage points (↑↑↑) Differences of more than 30 percentage points (=) Differences of less than 5 percentage points S TATISTICAL A NALYSIS (T EST STATISTICS ) Statistical tests were performed in order to explore the differences between medians of populations from both PRE/POST-outbreak studies, POST-OUTBREAK WORK: S TUDY 2 (III.1) 105 confirming a significant increase in the EP-Score median of the POST-study population (p<0.001, Table III.1.4). T ABLE III.1.4: EP-S CORES FROM TOTAL AND PAIRED SAMPLES OF BOTH PRE/POSTOUTBREAK SURVEYS AND ASSOCIATIONS BETWEEN THEM n Total ⱡ (matching compatible) EP - Score medians (P 25 -P 75 ) + n Paired EP - Score medians (P 25 -P 75 ) + p value PRE - outbreak survey 1145 • 5.0 (3.0 – 6.0) 88 5.0 (4.0 – 7.0) 0.245 ′ ↕ POST - outbreak survey 154 7.0 (5.0 – 8.0) 88 7.0 (6.0 – 8.0) 0.350 ′ p value <0.001 * + Weighted Average method; * Wilcoxon test ; ‘ Mann-Whitney test; ⱡ number of individuals compatible for matching. i.e. individuals who were scored regarding the 13 questions for perception assessment and who also have answered to the socio-demographic data and thus were punctuated for the matching process. • Out of the 1182 individuals that were scored in the PRE-study, 37 subjects were not included in the matching process, since they lack critical socio-demographic data; An increase of the number of individuals who achieved an EP-score equal to or higher than seven (EP-score≥7) in the POST-study population, was also statistically confirmed (p<0.001, Table III.1.5). T ABLE III.1.5: E VOLUTION OF THE SIZE OF THE RESIDENTS ’ GROUPS BEFORE AND AFTER THE OUTBREAK ACCORDING TO THE CUT - OFF : EP-S CORE ≥ 7 POST - outbreak survey (paired population) p value EP-Score < 7 EP-Score ≥7 PRE-outbreak survey (paired population) EP-Score < 7 27 (=) 38 (↑) <0.001 + EP-Score ≥ 7 4 (↓) 19 (=) nTOTAL=88 pairs (=) Number of individuals that did not change the EP-Score level compared with its pair (↓) Number of individuals that have increased the EP-Score level compared with its pair (↑) Number of individuals that have decreased the EP-Score level compared with its pair + McNemar test POST-OUTBREAK WORK: S TUDY 2 (III.1) 106 Confirming validity of the ‘Matching Process’ The validity of the model used in the matching process was also statistically established. As shown in Table III.1.3 the difference between the EP-score from the total and paired populations (in both PRE/POSToutbreak studies) did not change significantly (p>0.05 in both cases). In what concerns the personal-socio-demographic feature, total and paired populations also did not differ expressively. Slight differences are detected in proportions of age groups and in high education levels (Figure III.1.4). POST-OUTBREAK WORK: S TUDY 2 (III.1) 107 F IGURE III.1.4: C OMPARISON OF PERSONAL - SOCIO - DEMOGRAPHIC DATA BETWEEN PRE/POSTOUTBREAK SURVEY POPULATION ( TOTAL AND PAIRED ) Age group, Education level, Travels to DEC, and AME (bitten by mosquitoes) variables are presented. Since Gender and Municipal Division are fixed within matching pairs (only female Funchal residents are matched) these variables are not presented in these Figures. Figure III.1.4 a) – Data from Total Sample of the PRE-outbreak study Figure III.1.4 b) – Data from Total Sample of the POST-outbreak study POST-OUTBREAK WORK: S TUDY 2 (III.1) 108 Figure III.1.4 c) – Data from Total Paired of the PRE-outbreak study Figure III.1.4 d) – Data from Total Paired of the POST-outbreak study Comparison between basic and adjusted matching models The basic matching resulted in 65 pairs of individuals with equivalent personal-socio-demographic characteristics, being the pairs derived from individuals of each of the PRE/POST-outbreaks studies performed. The assessed differences in the perception of those surveyed before and after the outbreak, were equivalent to the previously described, obtained from the comparison of perception of the pairs derived from the adjusted matching. In fact not only the distribution and mean of the EP-score, but POST-OUTBREAK WORK: S TUDY 2 (III.1) 109 also the concept assimilation and even the topic understanding observed for the 65 pairs selected by the basic matching, were approximately the same that the ones measured for the 88 pairs resulted from the adjusted matching. Moreover, and as observed within the pairs derived from the adjusted matching, in the pairs derived from basic matching the POSToutbreak EP-score median has also significantly increased when compared to the PRE-outbreak one (p<0.001). Additionally, a significant increase of the number of individuals that achieved an EP-score equal to or higher than seven (EP-score≥7) in the POST-study paired population, was also observed when looking at pairs derived from the basic matching (p<0.001). Table III.1.6 summarizes the main results obtained from both matchings, including the power values which were equal in both cases. POST-OUTBREAK WORK: S TUDY 2 (III.1) 110 T ABLE III.1.6: C OMPARISON BETWEEN BASIC AND ADJUSTED MATCHING CRITERIA AND RESULTS Basic Matching Adjusted Matching Criteria* ten-year ‘age group’ twenty-year ‘age group’ ‘Municipalities’ in the same county represent different demographic group ‘Municipalities’ in the same county represent the same demographic group (‘Municipal Division’) n of individuals in the POST Total sample (matching compatible)** 93 154 No. of pairs PRE/POST (POST Paired sample) 65 88 Differences in EP - score medians and percentiles (POST-PRE) 2.0 (7.0-5.0) 2.0 (7.0-5.0) Power of the Wilcoxon test (used in the comparison between POST/ PRE EPScore median) ~1.000 ~1.000 * No significant differences observed (Additional file III.1.5) ** Individuals that were scored regarding the 13 questions for perception assessment and that also have answered to the sociodemographic data and thus were punctuated for the matching process. DISCUSSION In general, the community perception regarding preventive domestic practices increased within female residents of most aegypti-infested areas in Madeira Island after they experienced a dengue outbreak. By analysing how and how much assimilation of each 'Essential-concept’ has changed, crucial information can be retrieved regarding people´s perceptions about this experience and their future role in its prevention. For many Madeira residents, the experience of a dengue outbreak, the first in almost a hundred years in Europe, was probably the first contact with a mosquito borne disease [172]. This can explain the increase in the POST-OUTBREAK WORK: S TUDY 2 (III.1) 111 assimilation of the idea that ‘mosquitoes can transmit diseases’ (MI1concept). Moreover, before experiencing the outbreak, the community's worst incident with mosquitoes was allergic reactions, which could be considered as the sole health consequence of mosquito bites. After the outbreak, it was not surprising that the percentage of residents that were aware of ‘the kind of diseases that mosquitoes can transmit (such as dengue, yellow fever and malaria)’ (MI2-concept) almost doubled. Therefore, in the POST-outbreak study there were a higher percentage of people who rightly appraised the impact of mosquitoes in health. Since no fatal cases occurred during the dengue outbreak, some beliefs such as, ‘dengue disease does not kill’ and ‘dengue in Madeira is less aggressive’ may be present in the community. These questions should be considered in the case that a different virus serotype reaches the Madeira territory, increasing the risk of dengue haemorrhagic cases. Even though assimilation of both ‘Local Risk’ concepts increased after the outbreak, the majority of residents still ignored that ‘there is a high possibility for a (second) dengue outbreak in Madeira’ (LC2-concept). The acknowledgement of this concept was expected to increase after the outbreak, assuming that the previous identified myth which states that ‘Madeira were not at risk of have dengue’ would be opposed with the experience of a dengue outbreak. However, its assimilation merely increased 5%. Even though people had probably realized that Madeira was at risk and that several dengue cases occurred, two erroneous interpretations could explain this 5% result. Firstly, the false belief that the ‘dengue outbreak have ended due to the eradication of the disease or the mosquito’ (alleged myth 6, Table III.1.3). Secondly, the invalid belief that when something happens more frequently than normal during a period of time, the probability of happening again in the future decreases (gambler’s fallacy) (alleged myth 3, Table III.1.3) [163]. People who believe in these alleged myths underestimate the probability of another dengue epidemics occur in Madeira Island. Improvements in DA1-concept, DA2-concept, LC1-concept and MB1concept can be attributed to the “boom” of educational information POST-OUTBREAK WORK: S TUDY 2 (III.1) 118 Additional File III.1.1 d ) – Aedes aegypti’s distribution in Madeira Island (2012) – resulted from an Island-wide longitudinal entomological survey using ovitraps performed from tenth week of 2012 (end of February) until ninth week of 2013 (end of February) 17 . Due to their different type of surveys and ovitraps position it cannot be directly compared with its Additional File III.1.1 a). Even though, it reveals an expansion of A. aegypti’s mosquito considering the PRE-outbreak study confirming the need to expand it in the POST-oubreak study. Administrative boundaries described as «Municiplities» refer to what in the text is considered «county» or «Municipal Division» 17 This period represents what local authorites is consider the mosquito’s year calendar POST-OUTBREAK WORK: S TUDY 2 (III.1) 119 Additional File III.1.2: Socio-demographic characterization of Funchal’s Municipalities: Santa Luzia (SL), São Pedro (SP), Sé, Imaculado Coração de Maria (ICM) e Santa Maria Maior (SMM). Differences between proportions of those included in the PRE-study (SP and SL, in green) and those that were added in the POST-study (Sé, ICM and SMM in orange) are presented (in grey). % Women resident Women between 15&19 years Women between 20&24 years Women between 20&64 years Women between 25&64 years Women more than 64 years SL 55.7 2.5 2.4 31.6 29.2 15.3 SP 55.4 2.4 2.6 32.6 30.0 13.0 SL+SP 55.6 2.5 2.5 32.0 29.5 14.5 Sé 58.2 2.8 2.1 32.5 30.4 16.7 ICM 54.2 5.1 5.2 34.6 31.6 12.4 SMM 54.3 2.5 2.8 31.3 28.4 14.6 Sé+ICM+SMM 55.1 3.2 3.3 32.4 29.6 14.4 Difference -0.5 0.7 0.8 0.4 0.1 0.0 % Never studied 4 th grade 6 th grade 9 th grade 12 th grade Post - secondary school Degree SL 2.1 20.0 11.2 16.9 17.1 0.9 20.5 SP 3.5 24.6 11.9 16.4 13.5 1.4 15.5 SL+SP 2.6 21.7 11.4 16.8 15.8 1.1 18.6 Sé 2.1 17.7 9.5 15.7 15.6 1.0 27.3 ICM 7.3 26.6 15.2 17.7 14.9 3.4 15.4 SMM 3.3 23.6 12.8 16.6 15.6 1.0 16.2 Sé+ICM+SMM 4.1 23.2 12.8 16.7 15.4 1.6 18.2 Difference 1.5 1.5 1.3 -0.1 -0.3 0.5 -0.5 POST-OUTBREAK WORK: S TUDY 2 (III.1) 120 Additional File III.1.3: EP-score median differences regarding Gender using MannWhitney test (Output from Statistical Package for Social Sciences 19.0) Additional File III.1.4: Sample size estimation for PRE/POST pairs EP-scores comparison (Outputs from EPI-tools results for comparison of means with equal size and variance) [168] a) Sample size for detecting a 1-value difference within the PRE/POST pairs EPscore means b) Sample size for detecting a 2-value difference within the PRE/POST pairs EPscore means POST-OUTBREAK WORK: S TUDY 2 (III.1) 121 Additional File III.1.5: Analysis on the basis of the matching process adjustment. Part a) represent the absence of significant differences between Municipalities, if comparing individuals of the same Education level (as happens in the matching). Part b) represents the absence of significant differences between consecutive ten-year age group. Output from S Statistical Package for Social Sciences 19.0 (SPSS, Inc., Chicago, IL, USA) a) Comparisons of EP-Score medians between ‘Municipalities’ according to their Education level (results presented in an ascending order from Educational level = 1 to Educational level =5) b) Comparison of EP-Scores medians between the eight Age groups (black strips represent non-significant differences, yellow strips represent significant differences) POST-OUTBREAK WORK: S TUDY 2 (III.1) 122 Pair-wise comparisons Test statistics Signifi cance Adj Sig POST-OUTBREAK WORK: S TUDY 2 (III.1) 123 Additional File III.1.6: Discrepant concepts assimilation analysis POST-outbreak survey Concepts acknowledgement comparison Analysis of Topic’s Understanding Community Understanding / Alleged myths MI1- (concept) MI2- (concept) Medical Importance   70.5 % (62 ind.) Residents admitted that mosquitoes transmit diseases such as ‘Dengue’ (22·5%), ‘Malaria’ (9·5%), ‘Yellow fever’ (3·1%) or other mosquitoborne diseases (1·2%) or few of the latters. Residents seemed to understand the real medical importance of mosquitoes and, thus the relevance of being involved in the aegypti-control.  X 23.9 % (21 ind.) Even though admitting that mosquitoes can transmit diseases, these residents did not know what kind of diseases mosquitoes transmit. Some residents erroneously referred ‘allergies’ as mosquito-transmitted diseases (6.3%) and 4·5% mentioned other false clinical consequences such us ‘SIDA’, ‘fever’ or ‘cancer’. These residents were not aware of the relevance of being involved in the aegypti-control. Alleged Myth 2 : “Mosquitoes only cause mild clinical consequences such as allergies, fever, etc.” X  Not observed X X 5.7 % (5 ind.) Residents did not know that mosquitoes can transmit diseases Residents did not understand the medical importance of mosquitoes. Alleged Myth 1 : “Mosquitoes do not transmit diseases” LC1 LC2* Local Context   36.4 % (32 ind.) Residents recognized that there were mosquitoes that transmit diseases in their residential area, and, also, that there was a risk of a dengue outbreak in Madeira. Residents seemed to understand the local risk they are submitted and, thus the urgency of being involved in the aegypti-control.  X 36.4 % (32 ind.) Residents recognized the presence of mosquitoes that transmit diseases in their residential area; however they believed that a dengue outbreak will not re-emerge in the island. Eventually some residents could think that Madeira is now “protected”. Since a dengue outbreak has just occurred, there is a current very low probability of another dengue outbreak to emerge (gambler’s fallacy). Other possibility is that some residents could think that the end of the outbreak occurred when the mosquito/disease was eradicated from the island, and thus, now it won’t occur anymore. These residents were not aware of the urgency of being involved in the aegypti-control. Alleged myth 3: “‘Dengue will not occur again in Madeira, it is very not likely‘”; X  9.1 % (8 ind.) Residents did not recognize the presence of mosquitoes that can transmit diseases in their residential area; but admitted that a dengue outbreak can emerge in the island. These residents did not have a correct notion of the aegypti’s distribution area. Residents did not understand the risk they are subjected to and neither the urgency of being involved in the aegypti-control. Alleged myths 4 and 5: (i) -“Since I do not feel the byte, I am not at risk of being bitten/infected”. (ii) - “Mosquitoes are allocated in a specific area and are not able to spread to my municipality”; X X 18.2 % (16 ind.) Residents did not recognize mosquitoes that transmit diseases in their residential area neither the possibility of a dengue outbreak in the island. Residents did not understand the risk they are subjected to neither the urgency of being involved in the aegypti-control. Alleged Myth 6: “Dengue/A. aegypti was, finally, eradicated”. DA1 DA2 % Domestic Attribute   55.7 % (49 ind.) Residents know that mosquitoes can breed inside houses and recognized that domestic aegypti-control do have impact in the reduction of aegypti-population. Residents seemed to understand the domestic attribute of the aegypticontrol and, thus why community is the key intervenient in the aegypticontrol.  X 12.5 % (11 ind.) Residents know that mosquitoes can breed inside houses but they did not believe that the domestic aegypti-control have impact in the reduction of the aegypti’s population. They probably believed that other intervenients have much more impact in the reduction of the aegypti’s population. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypti-control. Alleged Myth 7: “Local health authorities are the key intervenient in the control of mosquitoes”. POST-OUTBREAK WORK: S TUDY 2 (III.1) 124 X  19.3 % (17 ind.) Mosquitoes cannot breed inside houses but domestic aegypti-control does have impact in the reduction of aegypti-population in the neighborhood. Those respondents believed in their role in domestic aegypti-control but did not understood why that control has an impact, probably by avoided them to enter in the house. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypti-control. Alleged Myth 8: “Other protective measures can control mosquitoes”. X X 12.5 % (11 ind.) Residents do not know that mosquitoes can breed inside houses, neither that their involvement have an impact in the control of mosquitoes. Residents did not understand the domestic attribute of the aegypti-control, neither why community is the key intervenient in the aegypti-control. Alleged Myth 9: “I am not/Community is not an intervenient in the aegypti-control”. MB1 MB2 % Mosquito Breeding   36.4 % (32 ind.) Residents only identified water - containers (and not other false issues) as mosquitoes’ breeding inducers. Residents seemed to understand where do mosquito breed and, thus the need of the aegypti-control activities.  X 591 % (52 ind.) Residents identified water-containers but also other false issues (food debris and pets) as mosquitoes’ breeding inducers These residents did not comprehend what lead to the breeding of new mosquitoes and, thus did not understand the proposed measures to control them. Residents seemed to not understand where mosquitoes breed and neither the need of the aegypti-control activities. Alleged Myths 10 and 11: “Clean houses or houses without pets/animals do not have mosquitoes” or “Clean people did not need to be involved in mosquito control”. X  2.3 % (2 ind.) Residents did not identify water - containers neither other false issues (food debris and pets) as mosquitoes’ breeding inducers. These residents did not know where do mosquitoes breed or believe in other false breeding sites. Residents seemed to not understand where mosquitoes breed and neither the need of the aegypti-control activities. X X 2.3 % (2 ind.) Residents did not identify water-containers but did identify other false issues (food debris and pets) as mosquitoes’ breeding inducers Residents are completely mistaken regarding mosquitoes breeding and, thus did not understand the need of the aegypti-control activities. Alleged Myths 10 and 11: “Clean houses or houses without pets/animals do not have mosquitoes” or “Clean people did not need to be involved in mosquito control”. CM1 CM2 % Control Measures   37.5 % (33 ind.) Residents only recognized water - containers removal (and not other false measures) as “effective to control mosquitoes” Residents seemed to recognize effective control measures and, thus understand how the domestic aegypti-control should be done.  X 60·2 % (53 ind.) Residents recognized water-containers removal and also other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes” Residents seemed to not be focused on effective control measures and, thus did not understand how the domestic aegypti-control should be done. Alleged Myth 12: “Using insecticides or the flyswatter, I am already contributing to control the aegypti-mosquito” X  1.1 % (1 ind.) Residents did not recognize water - containers removal neither other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes”. These residents did not know how to control mosquitoes. Residents not recognized effective control measures and, thus did not understand how domestic aegypticontrol should be done. X X 1.1 (1 ind.) Residents recognized water-containers removal and also other false measures (such as insecticide indoor application and flyswatter use) as “effective to control mosquitoes” Residents seemed to not be focused on effective control measures and thus did not understan d how the domestic aegypti-control should be done. Alleged Myth 13: “By using protective measures (such as insecticides or the flyswatter), I am already contributing to control the aegypti-mosquito” 125 II.2 IN-DEPTH PERCEPTION ASSESSMENT AND CONFIRMATORY MYTH IDENTIFICATION POST-OUTBREAK QUALITATIVE ANALYSIS – AN INDEPTH PERCEPTION ASSESSMENT AND A CONFIRMATORY M (S TUDY 3) 126 POST-OUTBREAK WORK: S TUDY 3 (III.2) 127 BACKGROUND Results from subchapter II.1 suggested that even after experiencing a dengue outbreak, community believed in several myths regarding domestic dengue prevention. Myths (i.e. erroneous perceptions) are mentioned as such since they oppose and contradict the procedure which, to date, is considered to be the most effective in aegypti-control. However, it is important to notice that these results were provided by quantitative analysis, which can be limited for exploring perceptions, feelings, and beliefs. Qualitative research, which has greater focus on individuals, is thus much able to examine how people integrate their experiences and their deep beliefs [174]. These approaches are in fact opposite in their epistemological basis. Quantitative research conclusions rely on objectivity, validity and reproducibility, while within the qualitative approach, knowledge is gained by inter-subjectivity among researchers and the object of the research [79]. This polarity had divided researchers. By one hand the deductive feature of the quantitative analysis can be criticized by close the research to unexpected results and not fully detect them. By other hand, the inductive feature of the qualitative can be censured as positivist and experimental. The combination of both can take advantage of the potentialities of each approach. This is the basis of the mixed methods data analysis whereby quantitative and qualitative data analysis strategies are combined, connected or integrated to provide conclusions regarding the same research question or aim [174]. Given the direct applicability of current research scope into policy and practice of dengue prevention campaigns, the validity of the perception assessed has particular relevance. In effect, the myths identified by quantitative analysis could have been derived from answers given by chance, (and not by the assumed erroneous perception). In this case, myth identification results would not be not valid in practice. Moreover, questions could be raised regarding the possibility of other (erroneous) perceptions or beliefs which were not detected by the quantitative analysis exist in the community.