Using Information Systems in Pharmacovigilance. Promotion of Avderse Drug Reaction Reporting

Using Information Systems in Pharmacovigilance PromotionofAdverseDrugReactionReporting Doctoral Program in Clinical and Health Services Research PhD Student Fernanda Inês de Carvalho Pereira Ribeiro Vaz Supervisor Co-supervisors Ricardo João Cruz Correia Cristina Maria Nogueira da Costa Santos Jorge Manuel Junqueira Polónia Professores Catedráticos: Professores Catedráticos Jubilados e Aposentados: Manuel Alberto Coimbra Sobrinho Simões Alexandre Alberto Guerra Sousa Pinto Maria Amélia Duarte Ferreira Álvaro Jerónimo Leal Machado de Aguiar José Agostinho Marques Lopes António Augusto Lopes Vaz Patrício Manuel Vieira Araújo Soares Silva António Carlos de Freitas Ribeiro Saraiva Alberto Manuel Barros da Silva António Carvalho Almeida Coimbra José Manuel Lopes Teixeira Amarante António Fernandes Oliveira Barbosa Ribeiro Braga José Henrique Dias Pinto de Barros António José Pacheco Palha Maria Fátima Machado Henriques Carneiro António Manuel Sampaio de Araújo Teixeira Isabel Maria Amorim Pereira Ramos Belmiro dos Santos Patrício Deolinda Maria Valente Alves Lima Teixeira Cândido Alves Hipólito Reis Maria Dulce Cordeiro Madeira Carlos Rodrigo Magalhães Ramalhão Altamiro Manuel Rodrigues Costa Pereira Cassiano Pena de Abreu e Lima Rui Manuel Almeida Mota Cardoso Daniel Filipe de Lima Moura José Carlos Neves da Cunha Areias Daniel Santos Pinto Serrão Manuel Jesus Falcão Pestana Vasconcelos Eduardo Jorge Cunha Rodrigues Pereira João Francisco Montenegro Andrade Lima Bernardes Fernando Tavarela Veloso Maria Leonor Martins Soares David Henrique José Ferreira Gonçalves Lecour de Menezes Rui Manuel Lopes Nunes Jorge Manuel Mergulhão Castro Tavares José Eduardo Torres Eckenroth Guimarães José Carvalho de Oliveira Francisco Fernando Rocha Gonçalves José Fernando Barros Castro Correia José Manuel Pereira Dias de Castro Lopes José Luís Medina Vieira António Albino Coelho Marques Abrantes Teixeira José Manuel Costa Mesquita Guimarães Joaquim Adelino Correia Ferreira Leite Moreira Levi Eugénio Ribeiro Guerra Raquel Ângela Silva Soares Lino Luís Alberto Martins Gomes de Almeida Manuel António Caldeira Pais Clemente Manu el Augusto Cardoso de Oliveira Manuel Machado Rodrigues Gomes Manuel Maria Paula Barbosa Maria da Conceição Fernandes Marques Magalhães Maria Isabel Amorim de Azevedo Mário José Cerqueira Gomes Braga Serafim Correia Pinto Guimarães Valdemar Miguel Botelho dos Santos Cardoso Walter Friedrich Alfred Osswald i Acknowledgements Reading for a PhD becomes a great adventure in one’s professional and personal life. It would have been impossible to achieve without the valuable help and encouragement of some special individuals to whom I am extremely grateful. I would like to express my deep gratitude to my supervisor, Professor Ricardo Correia, who was the first to encourage me to pursue this endeavour. His support, confidence and guidance enabled me to finish my PhD thesis with contentment. I believe that our conversations and the discussions on the topic of my thesis have been more enlightening than the thousands of scientific papers I have read over the past years. I also would like to thank my co-adviser Professor Cristina Santos for her support and assistance. Together with Professor Ricardo Correia, she always trusted my capacity and ability to successfully conclude these studies. It has been an enriching experience to work side-by-side on this project. In addition, I am grateful to Professor Jorge Polónia, my second co-adviser, for his confidence and dedication to this project, despite his very busy agenda, he always managed to find a moment for me. To Professor Altamiro da Costa Pereira, head of the Health Information and Decision Sciences Department and Director of the Doctoral Program in Clinical and Health Services Research, I would like to thank the opportunity that he provided by allowing me to accomplish this doctorate, during some of my working hours as a Coordinator for the Northern Pharmacovigilance Centre. We can’t do much without the support of our friends. These lines are for them, who were always by my side, supporting me, making me smile when facing and solving problems. I particularly would like to thank Ana Marta Silva, Camila Dias, Daniel Pereira, Patrícia Alves, Pedro Farinha, Pedro Rodrigues, Matilde Monteiro Soares, Madalena Oliveira (always by my side!) and Teresa Alves (as usual, in the same boat, at the same time…that’s our story!). I am also grateful to all my colleagues at the Health Information and Decision Sciences Department for the extraordinary work environment and their daily support. Working with kind people that share a common interest despite their different positions and backgrounds is not only extremely enriching but also makes a big difference. Finally, a big thank you to the most important people in my life: my family. Thank you very much Mum and Dad for your constant support, helping daily with the kids and for your unconditional love. All my love goes to my husband and children, who are my reason for living and my daily inspiration. Thank you for being there and for forgiving me all the absences while completing this PhD study. I love you all very much and day by day even more! ii List of Co-Authors Ricardo João Cruz-Correia 1, 2 Cristina Maria da Costa Santos 1, 2 Jorge Manuel Junqueira Polónia 1, 3 Ana Marta Matos Silva 1, 3 Altamiro Manuel da Costa Pereira 1, 2, 3 Pedro Pereira Rodrigues 1,2 Fabrício Barbosa 4 Domingos Alves 5 1. Centre for Research in Health Technologies and Information Systems (CINTESIS) University of Porto, Portugal. 2. Health Information and Decision Sciences Department (CIDES), Faculty of Medicine, University of Porto, Portugal. 3. Northern Pharmacovigilance Centre, Faculty of Medicine, University of Porto, Portugal. 4. Oswaldo Cruz Foundation (FIOCRUZ), Rio de Janeiro, Brazil. 5. Faculty of Medicine of Ribeirão Preto, University of São Paulo, Brazil. iii Abbreviations ATC - Anatomical Therapeutic Chemical classification system. ADR - Adverse Drug Reaction(s) AERS - Adverse Events Reporting Systems AE - Adverse Events ARIMA - Autoregressive Integrated Moving Average CDSS - Clinical Decision Support Systems CIDES - Health Information and Decision Sciences Department CINTESIS - Centre for Research in Health Technologies and Information Systems CPARA - Catálogo Português de Alergias e Reações Adversas ( Portuguese c atalog of allergies and adverse reactions) EHR - Electronic Health Record(s) EPR - Electronic Patient Record(s) EMA - European Medicines Agency FMUP - Faculdade de Medicina da Universidade do Porto (Faculty of Medicina, University of Porto) FDA - Food and Drug Administration FIOCRUZ - Oswaldo Cruz Foundation GEDII - Grupo de Estudo da Doença Inflamatória Intestinal (Study Group of Inflammatory Bowel Disease) HL7 - Health Level Seven INFARMED - Autoridade Nacional do Medicamento e Produtos de Saúde, IP ( National Authority of Medicines and Health Products, IP) IS - Information Systems ISO - International Organisation for Standardis ation MA - Marketing authorization PPV - Positive Predi c tive Value RAM - Reações Adversas a Medicamentos (adverse drug reactions) SIRAI - Sistema de Informação de Reações Adversas e Incidentes (Information System for Adverse Reactions and Incidents) TTA - Time to causality assessment ULSM - Unidade Local de Saúde de Matosinhos (Health Local Unit of Matosinhos) UFN - Unidade de Farmacovigilância do Norte ( Northern Pharmacovigilance Centre ) USA - United States of America VPP - Valor Positivo Preditivo ( Positive Predi c tive Value ) WHO - World Health O rganis ation 4 6.1 Abstract ............................................................................................................................ 95 6.2 Introduction ..................................................................................................................... 96 6.3 Methods ........................................................................................................................... 96 6.4 Preliminary results ........................................................................................................... 97 Chapter 7 - Ongoing and future work ......................................................................................... 99 7.1 Abstract ............................................................................................................................ 99 7.2 Managing several players at several institutions dealing with the same adverse event. A pilot-study called SIRAI. ............................................................................................................. 100 7.3 Social networks and portals for patients and healthcare professionals ........................ 103 7.3.1 Creation of the personal area on the UFN website ................................................. 104 7.3.2 ADR online reporting form for consumers ............................................................... 106 7.4 App for ADR reporting .................................................................................................... 107 Chapter 8 – General discussion ................................................................................................. 109 References ................................................................................................................................. 111 Annex ........................................................................................................................................ 117 Resumo / Abstract 5 Resumo da tese As decisões na área da segurança do medicamento são tomadas através de revisões regulares, baseadas sobretudo na informação disponibilizada pelos sistemas de farmacovigilância que assentam fundamentalmente na notificação espontânea de Reações Adversas a Medicamentos (RAM). Estas notificações são feitas de forma voluntária pelos profissionais de saúde e pelos utentes. O sistema de notificação espontânea de RAM tem a enorme mais-valia da deteção precoce dos problemas de segurança dos medicamentos, mas sofre da grande limitação decorrente da sub-notificação. Estima-se que apenas cerca de 10% das reações adversas que ocorrem sejam efetivamente notificadas às Autoridades Reguladoras, nos países desenvolvidos. Para combater este problema, os sistemas de farmacovigilância têm desenvolvido várias estratégias, devendo os sistemas de informação ser encarados como uma oportunidade nesta área, uma vez que constituem uma presença central no nosso quotidiano, nomeadamente ao nível das instituições de saúde. O objetivo desta tese é investigar estratégias para promover a notificação de RAM utilizando os sistemas de informação para facilitar o processo. Esta investigação é constituída por seis estudos, com os seguintes objetivos individualizados: 1) estabelecer o estado da arte sobre o uso de sistemas de informação na notificação espontânea de RAM; 2) comparar diferentes estratégias de promoção da notificação de RAM, determinando o seu custo/efetividade; 3) promover a notificação espontânea de RAM entre os profissionais de saúde que exercem atividade em hospitais, através de hyperlinks diretos para o formulário de notificação online; 4) implementar e avaliar o consumo de um webservice por um registo clínico eletrónico habitualmente utilizado por um grupo que participa num estudo multicêntrico na área da gastroenterologia; 5) aperfeiçoar o processo de avaliação de causalidade das notificações de RAM utilizando um modelo de apoio à decisão; 6) criar a área pessoal do notificador no website da unidade de farmacovigilância, tornando a notificação de RAM uma atividade motivadora e informativa. Para estabelecer o estado da arte sobre o tema em estudo, foi elaborada uma revisão sistemática (Capítulo 2) sobre a utilização dos sistemas de informação em farmacovigilância, através das bases de dados bibliográficas das áreas científicas da saúde e das tecnologias de informação. De um total de 3865 artigos selecionados na pesquisa inicial, 33 artigos foram incluídos na análise, descrevendo 29 projetos diferentes. Foi efetuada uma meta-análise com 7 dos 29 projetos, para calcular a medida agregada do aumento da notificação de RAM, Resumo / Abstract 6 obtendo-se o valor de 2.1 (significando que as intervenções duplicaram o número de RAM notificadas). No Capítulo 3, são analisadas e descritas as várias estratégias que têm sido adotadas na Unidade de Farmacovigilância do Norte no sentido de promover a notificação de RAM, no que diz respeito ao número e relevância das notificações de RAM obtidas e aos custos envolvidos. Os custos da notificação de RAM foram calculados adicionando os custos iniciais de implementação da estratégia com os custos de manutenção da mesma (para cada estratégia analisada). Este custo global foi dividido pelo número de notificações de RAM obtidas por cada intervenção, para determinar o seu custo/efetividade. Todas as estratégias aumentaram o número de notificações de RAM. O maior aumento foi verificado com os protocolos estabelecidos com departamentos hospitalares (321 notificações de RAM obtidas, com o custo de 1.96€ cada uma), seguido da intervenção educativa (265 notificações de RAM obtidas, a 20.31€ cada uma) e pela inclusão de hyperlinks (135 notificações de RAM, com o custo de 15.59€ cada uma). Relativamente às RAM graves, os protocolos foram a intervenção mais eficiente (2.29€ cada notificação), seguido da inclusão de hyperlinks (30.28€ cada notificação, sem custos de manutenção). Os protocolos obtiveram o melhor resultado relativamente à notificação de RAM inesperadas (5.12€ cada notificação), seguido da intervenção educativa (38.79€ por notificação). O Capítulo 4 descreve um estudo ecológico desenvolvido durante o período de 2006-2011 na região Norte de Portugal, que consistiu na inclusão de hyperlinks diretos para o formulário online de notificação de RAM nos registos clínicos eletrónicos e/ou nos ambientes de trabalho dos profissionais de saúde dos hospitais desta região. A mediana mensal das notificações de RAM (total e online) e respetivos âmbitos foram analisados antes e depois da intervenção em todos os hospitais neste estudo. Dezasseis centros hospitalares entraram no estudo (27 hospitais), sendo que onze centros (18 hospitais) incluíram o hyperlink. Considerando os hospitais que colocaram o hyperlink no registo clínico eletrónico, a mediana mensal de notificação de RAM aumentou significativamente, de duas (âmbito 0-12) para cinco notificações (âmbito 1-17). A mediana mensal de notificação de RAM através do formulário online também aumentou significativamente, de uma notificação (âmbito 0-5) antes da intervenção para duas notificações (âmbito 1-17) depois da intervenção. Além disso, a notificação de RAM graves aumentou 3 vezes, e a notificação de RAM inesperadas aumentou 4.5 vezes. Nenhum destes aumentos significativos foi observado nos hospitais que não incluíram o hyperlink. Também se observou um aumento significativo das visitas diárias ao Resumo / Abstract 7 website da Unidade de Farmacovigilância do Norte, de 10 visitas diárias antes da intervenção para 27 depois da intervenção (p<0.001). No Capítulo 5 encontra-se detalhado o processo de desenvolvimento de um serviço informático – chamado webservice – que foi implementado num registo clínico eletrónico em utilização por um grupo de gastroenterologistas. Foi feito um estudo entre 2013 e 2015, para analisar a tendência de notificação de RAM e o tipo de RAM notificadas por esse grupo de médicos. De abril de 2013 a fevereiro de 2015, foram enviadas 167 notificações de RAM para a Unidade de Farmacovigilância do Norte, através deste webservice, significando 10% do total de RAM recebidas no mesmo período. Destas 167 notificações, 118 eram casos graves (1 deles colocou a vida do doente em risco). Para medir o impacto da intervenção, foram considerados apenas os médicos do grupo que pertencem à região Norte de Portugal. Estes médicos notificaram 9 RAM durante os 23 meses anteriores à implementação do webservice e 121 RAM nos 23 meses posteriores à sua implementação, significando um aumento de 81%. Para aperfeiçoar o processo de avaliação de causalidade das notificações de RAM, foi desenvolvida uma rede Bayesiana - descrita no Capítulo 6 - baseada na avaliação que tinha sido feita pelo avaliador especialista da Unidade de Farmacovigilância do Norte, ao longo de 12 anos, às notificações de RAM recebidas nesta Unidade. Os resultados desta rede foram comparados com os resultados da introspeção global, com base numa coorte de validação independente, para sensibilidade, valor preditivo positivo (VPP) e tempo de avaliação de causalidade. A causalidade foi classificada, de acordo com os graus de causalidade definidos pela Organização Mundial de Saúde, como: Definitiva, Provável, Possível ou Condicional. A coorte de derivação era constituída por 593 notificações de RAM (10.1% classificadas com o grau Definitiva, 58.4% com o grau Provável, 25.6% com o grau Possível e 5.9% com o grau Condicional) e a coorte de validação por 463 notificações (7.5% classificadas com o grau Definitiva, 79.5% com o grau Provável, 9.5% com o grau Possível e 2.8% com o grau Condicional). Foi obtida elevada exatidão para as notificações com o grau Definitivo (sensibilidade 69.4% e VPP de 71.4%) e com o grau Provável (sensibilidade 91.1% e VPP 87.3%), e mais baixa para as notificações classificadas com o grau de Possível (sensibilidade de 25% e VPP de 28.9%) e Condicional. A rede revelou tendência para sobrestimar a causalidade (96.9% dos erros nos casos classificados pelo avaliador com o grau Possível foram avaliados pela rede com o grau Provável) ou para atribuir o grau imediatamente abaixo (90.8% dos erros nos casos classificados pelo avaliador com o grau Definitivo foram classificados pela rede com o grau Provável; 69.7% dos erros nos casos com o grau Provável foram classificados pela rede com o grau Possível). A mediana e respetivo intervalo interquartil (Q1:Q3) do tempo de Resumo / Abstract 8 avaliação de causalidade foi de 4 dias (2:8) utilizando a rede e de 8 dias (5:14) através da introspeção global. O Capítulo 7 (concretamente, o subcapítulo 7.3.1) descreve o trabalho que continua a ser feito na criação da área pessoal do notificador no website da Unidade de Farmacovigilância do Norte. Atualmente, esta área já apresenta ao utilizador as suas próprias notificações de RAM submetidas online. Pretende-se ainda desenvolver um modelo de comunicação entre a unidade de farmacovigilância e os seus notificadores, fornecendo informação técnica sobre a reação adversa notificada e o medicamento suspeito. Após ter este modelo operacional, pretendemos (como trabalho futuro) avaliá-lo através de uma série de indicadores (por exemplo: aumento das notificações de RAM, aumento das visitas ao site, entre outros). Este trabalho não foi ainda concluído. Muito trabalho pode ser desenvolvido para melhorar as atividades de farmacovigilância através dos sistemas de informação. Soluções promissoras podem passar pela integração das bases de dados de farmacovigilância nos registos clínicos eletrónicos usados habitualmente pelos profissionais de saúde, evitando que estes tenham trabalho adicional para submeter uma notificação de RAM. A inclusão de hyperlinks diretos nos registos clínicos eletrónicos e/ou nos ambientes de trabalho dos computadores dos profissionais de saúde para o formulário de notificação online de RAM é uma forma simples e custo-efetiva de alterar o comportamento dos profissionais de saúde relativamente à notificação de RAM e pode ser facilmente implementado nas instituições de saúde. Os sistemas de informação podem ainda ajudar na melhoria das atividades de rotina em farmacovigilância, melhorando a comunicação entre as Unidades de farmacovigilância e os seus notificadores. É também possível acelerar o processo de imputação de causalidade e subsequente feedback, utilizando métodos Bayesianos. Resumo / Abstract 9 Thesis abstract Decisions in the area of drug safety are made through regular reviews, based on available information from pharmacovigilance systems, which are mostly based on Adverse Drug Reactions (ADR) reports, voluntarily made by healthcare professionals and consumers. The ADR reporting system has the great value of early detection of drug safety problems, but also has a major limitation due to under-reporting. It is estimated that only about 10% of ADR that occur are actually reported to regulatory authorities in developed countries. To counter this problem, pharmacovigilance systems have been instituted to ensure that detected ADR are effectively reported to the regulatory authorities. The use of information systems is currently central in most of our lives, namely in healthcare institutions, and therefore represents an opportunity. The aim of this thesis is to investigate strategies to promote ADR reporting using information systems to facilitate the process. It comprises the following six study aims: 1) to assess the state of the art regarding the utilisation of information systems in spontaneous ADR reporting; 2) to compare different approaches promoting ADR reporting and to determine their cost/effectiveness; 3) to promote spontaneous ADR reporting among healthcare professionals working in hospitals using hyperlinks to ADR online reporting forms; 4) to implement and evaluate the consumption of a webservice by the usual electronic health record used by a gastroenterologist multicentre study group; 5) to improve the ADR report causality assessment using a decision support tool and 6) to create a personal area in a pharmacovigilance website for each person that submits an ADR, aiming to turn the ADR reporting act into a motivating and informative activity. A systematic review was performed regarding the use of information systems in pharmacovigilance, in bibliographic databases of the scientific fields of healthcare and information technology, as described in Chapter 2. From a total of 3865 articles, 33 articles were included in the analysis, describing 29 different projects. We also performed a metaanalysis on 7 of the 29 projects, to calculate the aggregated measure of increased ADR reporting, with an overall measure of 2.1 (meaning that the interventions doubled the number of ADR reports). Several approaches taken by the Northern Pharmacovigilance Centre to promote ADR reporting are analysed and described in Chapter 3, regarding the number and relevance of ADR reports obtained and the costs involved. The costs of ADR reporting were calculated by Resumo / Abstract 10 adding the initial costs and the running costs of each intervention. These costs were divided by the number of ADR reports obtained with each intervention, to assess its cost/effectiveness. All the approaches increased the number of ADR reports. The biggest increase was noted with protocols in hospital departments (321 ADR reports obtained, costing 1.96€ each), followed by an educational approach (265 ADR reports, 20.31€/report) and a hyperlink approach (136 ADR reports, 15.59€/report). According to serious ADR, the protocol approach was the most efficient, costing 2.29€/report, followed by the hyperlink approach at 30.28€/report (with no running costs). Concerning ADR unexpectedness, the best result was found with the protocol approach (5.12€/report), followed by the educational approach (38.79€/report). Chapter 4 describes an ecological study performed in northern Portuguese hospitals from 2006 to 2011. We included hyperlinks to the online ADR report form either on Electronic Patient Records or on computer desktops. The median of spontaneous ADR reports (total and online) per month and the respective ranges were analysed before and after the intervention in all hospitals in this study. Sixteen hospital centres were involved in the study (27 hospitals). Eleven centres (18 hospitals) included the hyperlinks. Considering the hospitals with hyperlink access to the Electronic Patient Records, the median ADR reports per month increased significantly, from two (range 0-12) to five reports (range 1-17). The median of ADR reports per month using the online form also increased significantly, from one (range 0-5) before the intervention to two (range 1-17) after it. Moreover, serious ADR increased by 3-fold, and nonpreviously described ADR increased by 4.5-fold. None of these significant increases were observed in the hospitals where the hyperlink was not installed. We also found a significant increase in daily UFN website visits, from 10 before the intervention to 27 after it (p < 0.001). Chapter 5 focuses on the creation of a webservice and its implementation in an electronic health record already in use by a group of gastroenterologists. A study was performed between 2013 and 2015 to analyse the trends of ADR reporting and also the type of ADR reported. From April 2013 to February 2015, 167 ADR reports were sent to the Northern Pharmacovigilance Centre through the webservice, meaning 10% of the total of ADR reports received in the same period. Of these 167 ADR, 118 cases were serious (one of them lifethreatening). According to the northern region of Portugal, the studied physicians reported 9 ADR during the 23 months prior to webservice implementation and 121 ADR during the 23 months after webservice implementation, i.e. an increase of 81%. To improve the process of causality assessment, a Bayesian network was developed, as described in Chapter 6. Network development was based on completely filled ADR reports, evaluated by the Portuguese Northern Pharmacovigilance Centre expert over 12 years, and Resumo / Abstract 11 compared with global introspection on an independent validation cohort for sensitivity, positive predictive value (PPV) and time to causality assessment (TTA). Causality was classified as Definitive, Probable, Possible or Conditional, according to the World Health Organisation causality assessment. The derivation cohort included 593 ADR reports (10.1% Definitive, 58.4% Probable, 25.6% Possible and 5.9% Conditional) with a validation cohort including 463 reports (7.5%, 79.5%, 9.5% and 2.8%, respectively). High accuracy was reached for reports with Definitive causality (69.4% sensitivity, 71.4% PPV) and Probable causality (91.1%, 87.3%), but this was lower for reports with Possible (25%, 28.9%) and Conditional (15.4%, 50%) classification. The network tends to overrate causality (96.9% of errors in Possible cases classified as Probable) or give the level immediately below (90.8% of errors in Definitive cases classified as Probable; 69.7% of errors in Probable cases classified as Possible). The median (Q1:Q3) TTA was 4 (2:8) days using the network and 8 (5:14) days using global introspection. The creation of a personal area within the UFN website was created and is described in Chapter 7 (specifically, subchapter 7.3.1). We intend to design a model of communication between the pharmacovigilance centre and its users, providing technical information about the reported adverse reaction and the suspected drug. We expect to have the model in operation on the UFN website and evaluate its use through a set of indicators (e.g. increase in ADR reports, increase of website visits, etc.). This work is ongoing. Much work can be done using information systems to improve pharmacovigilance. The authors see this as a promising solution the integration of pharmacovigilance databases with the habitually used electronic health records, avoiding an extra workload to submit an ADR report. The inclusion of hyperlinks to on-line ADR reporting forms on computer desktops and in Electronic Health Records is an easy and cost-effective way to change healthcare professional behaviours regarding spontaneous ADR reporting and can be easily implemented in healthcare institutions. Information systems can also improve pharmacovigilance routine activities, improving the communication between pharmacovigilance centres and its reporters. It is possible to accelerate the process of causality assessment and subsequent feedback using Bayesian methods. 12 Achievements and background 13 Achievements and background Working in pharmacovigilance for the last 12 years, it was fairly easy to choose this PhD theme, as I have always aimed to change the paradigm of spontaneous adverse drug reaction (ADR) reporting, from the idea of a passive and boring process to something appealing and simple to do. Considering the increasing use of information systems (IS) among healthcare institutions, it became clear that we could use IS to change the way people see the pharmacovigilance system. In this process, the contribution of my supervisor, Prof. Ricardo Correia, was crucial as he who always aroused my curiosity regarding the use of IS to promote spontaneous ADR reporting. The development of this PhD thesis allowed me to face some challenges related to the differences between the structure of healthcare professionals’ knowledge and the specific needs of electronic records. In this sense, I have participated in some discussion groups, both national and internationally (namely the FHIR/OpenEHR [1] and HL7 [2] working groups), aiming to create a standard way in which information should be kept in the ADR registry. In June 2014, I participated in the HL7 pharmacy group meeting, where the International Organisation for Standardisation (ISO) draft of the international standard for ‘Requirements for electronic prescriptions’ was discussed. This document remained under discussion for several months. Contributions were also made to the last OpenEHR discussion, which took place in July 2014. It was focused on ADR definition, its characteristics and what should be kept during the registration of a ADR. At the moment, discussion is ongoing regarding the clinical content of the archetypal adverse reaction (FHIR/OpenEHR). Variables such as substance, status, seriousness, reaction type and certainty are under debate. Nationally, important input was provided by this PhD team to the review of national legislation for the registration of allergies and adverse reactions (the Portuguese Catalogue for Allergies and Adverse Reactions - CPARA, in Portuguese). For this purpose, we created an archetypal Open EHR for the registry of ADR and allergies. Another study developed during the PhD period, but not included in this thesis, was the creation of a list of ‘alert drugs’ to help hospital pharmacies detect ADR. For this purpose, a list of drugs was developed, adapted from Otero et al [3] and Classen et al [4] , that should alert pharmacists to potentials ADR when the drugs are prescribed by physicians. The drugs included were chosen because they are used as specific treatments for ADR or as antidotes. 20 Chapter 2. State of the Art / Pharmacovigilance informatics 21 2.2 Pharmacovigilance Informatics Inês Ribeiro-Vaz, Fabrício Alves, Ana Silva, Domingos Alves, Ricardo Correia Encyclopedia of E-health and Telemedicine 2015 Chapter 2. State of the Art / Pharmacovigilance informatics 22 Chapter 2. State of the Art / Pharmacovigilance informatics 23 2.2 Pharmacovigilance Informatics Chapter 2. State of the Art / Pharmacovigilance informatics 24 Introduction All medicines have adverse effects, most of them unknown until the drug commercialization. As so, it is crucial to keep strategies to monitor the drug safety. Pharmacovigilance is the activity of drug surveillance, after its launch in the market, with the main goal of public health protection, ensuring that the drug benefit outweighs its risks. Worldwide, pharmacovigilance systems are mostly based on spontaneous Adverse Drug Reactions (ADR) reports made by healthcare professionals and consumers. Spontaneous ADR reporting has been described as an essential method to detect drug safety signals; however, underreporting is a major issue undermining the effectiveness of spontaneous reports. Several studies suggest that less than 10% of detected ADR are effectively reported to medicines regulatory authorities [e.g. Food and Drug Administration (FDA), European Medicines Agency (EMA), etc] [7, 10] . Tools used in pharmacovigilance are continually evolving and, worldwide, Information Systems (IS) to promote ADR reporting or to detect ADR occurred in healthcare institutions have been tested and used, such as software that allow voluntary and automated detection of ADR, tools that analyse clinical databases or Web sites that actively inform healthcare professionals [11] . In addition to signal detection, IS can also be used to encourage and facilitate reporting of suspected ADR, such as the creation of on-line reporting forms, development of tools to collect safety data from electronic health records (EHR), among others. In this chapter, it will be described some tools to automatically detect ADR, or encourage ADR spontaneous report. Background Adverse Drug Reactions (ADR) (defined as a response to a medicinal product which is noxious and unintended [6] ) are a well-recognized public health problem worldwide, and a major cause of death and hospitalization in developed countries [12] . It is estimated that about 6,5% of the hospitalizations are related to ADR [13] . Besides, in the USA, about 100.000 people die each year due to ADR [12] , and in Europe this annual mortality rate increases to 197.000 [14] . ADR can be expressed in many ways and with different degrees of seriousness. An anaphylactic shock caused by penicillin is an example of a serious ADR (a serious ADR is any untoward medical occurrence that at any dose: results in death, requires inpatient hospitalisation or prolongation of existing hospitalisation, results in persistent or significant disability/incapacity or is lifethreatening [6] ). Another type of ADR, not always recognized as such, is the drug ineffectiveness, for example, a vaccination failure. This can be (or not) related with a product Chapter 2. State of the Art / Pharmacovigilance informatics 25 quality issue and should be reported when detected in order to allow the regulatory authorities to take appropriate decisions. Rare and long term ADR are difficult to detect during the drug development stage. Only when the drug begins to be used by a large population after marketing authorization it is possible to detect new ADR not previously identified during clinical trials. In reality, it is known that the safety of a new drug cannot be established until it has been on the market for several years [8] . Exceptionally, in a pandemic scenario, drug launch is urgent and, in this particular case, can be justifiable that drug safety profile is not well established. In this scenarios, it is even more important that all the detected ADR are reported (serious or not, expected or not). It is, therefore, essential to keep drugs under close surveillance, after its commercialization, through a pharmacovigilance system, to continuously evaluate their safety profile. In most of the European countries, pharmacovigilance system is based on spontaneous ADR reports, which is passive method, made by healthcare professionals and, since July 2012, also by consumers [15] . These reports can be made using paper, telephone, e-mail or through an online form and consist of a description of an Adverse Event (AE) apparently caused by a medicine. To reverse the problem of underreporting of ADR, which is felt in most developed countries, several strategies have been tested [10, 16, 17] . Particularly, some studies were developed focused on educational interventions to raise awareness on the importance of ADR reporting [18-20] and showed to be very effective increasing the quantity and relevance of spontaneous ADR reports (among health professionals). However, these studies involved a large financial and personal outlay and the authors concluded that the effect was lost after a few months [10, 21] . In a recent American study, the authors developed a signal-detection strategy that combines the Adverse Event Reporting System (AERS) of the regulatory Authority (FDA) and EHR, by requiring signaling in both sources, with promising results [22] . Another study used the unstructured clinical notes included in EHR to detect ADR through a computerized system. The authors concluded that data mining can be used for hypothesis generation and for rapid analysis of suspected AE risk [23] . With a similar aim, a recent study used a physician’s network, created through a mailing list, to send regular emails to doctors, with humorous component attached with informative component, recalling the importance of reporting their suspected ADR [24] . The results showed that this type of intervention has impact on the number of ADR reports made by these professionals (the study did not assess the relevance of reported ADR). In France, it is being Chapter 2. State of the Art / Pharmacovigilance informatics 26 done a work that tries to facilitate the act to ADR reporting, habitually considered a tedious process by health professionals. The authors are using the information contained in EHR to make the semi-automatic filling of ADR notification forms. The objectives of this ongoing study is to increase the rate of ADR report, as well as improving the quality of information submitted to regulatory authorities [25] . Information Systems in Pharmacovigilance Although some authors consider that ADR spontaneous reports suffers from latency and inconsistency, it is still considered as the most valuable method to early detect drugs safety problems. In fact, most of the decisions concerning to drug safety are triggered by daily ADR spontaneous reports. As so, Regulatory Authorities consider as crucial importance to achieve the greatest number of ADR reports possible and with high data quality. The promotion of ADR report among healthcare professionals is a huge task, as it is necessary to regularly recall the importance of ADR reporting and, simultaneously, develop tools to facilitate this duty. Pharmacovigilance centres worldwide develop several strategies to continuously promote the importance of ADR reporting, as workshops, post-graduate courses, and also to make it easier, as development of online reporting forms, inclusion of electronic reporting systems into the hospital Information Systems (HIS), direct hyperlinks to online reporting forms, among others. Along with the promotion of spontaneous ADR reporting, some systems are being tested to detect signals of adverse reactions in large databases, as hospitals databases, epidemiologic databases, or even among social networks. Although the adoption of different strategies, it is consensual the need to obtain the highest quantity and quality of information about the safety of marketed drugs, for the protection of public health. Another important issue in this field is the need to discuss/harmonize what should be recorded during an ADR report. When analyzed the databases from the FDA, EMA and the Portuguese Northern Pharmacovigilance Centre (UFN, in portuguese), all of them with the same purpose of recording ADR reports, it was possible to realize that the variables are different between the 3 databases, as seen in the Table 1. Chapter 2. State of the Art / Pharmacovigilance informatics 27 Table 1. Comparison of the variables included in three pharmacovigilance databases (Food and Drug Administration: FDA; European Medicines Agency: EMA and Portuguese Northern Pharmacovigilance Centre: UFN), grouped in main entities/sections (patients, problems, products and reporters) Section Variable FDA EMA UFN Patient Patient identifier    Age at the time of event    Date of birth    Sex    Weight    Problem What kind of AE, product problem or error did you encounter? - Adverse event - Product use error - Product problem - Problem with Different Manufacturer of the Same Medicine --- --- Outcomes Attributed to AE - Death - Life-threatening - Congenital anomaly/birth defect - Disability or permanent damage - Hospitalization - Required intervention to prevent permanent impairment/damage - Other serious - Fatal - Not recovered - Recovered with sequelae - Recovering - Recovered - Unknown - Fatal - Not recovered - Recovered with sequelae - Recovering - Recovered - Unknown ADR Seriousness --- - Yes - No - Death - Life-threatening - Congenital anomaly - Results in persistent or significant incapacity - Hospitalization - Other serious Date of event    Describe events, problem, or product use error    Relevant test/laboratory data, including dates  --- --- Product Product name    Label strength    Manufacturer    Date of use    Reason for use    Problem went away after use stopped or dose reduced? - Yes - No - Does not apply ---  Problem returned after person started taking or using the product again? - Yes - No - Does not apply ---  Do you still have the product in case we need to evaluate it? - Yes - No - Returned to Manufacturer --- --- Reporter Reporter name  ---  Address    or list of options: Variable exists; ---: Variable does not exist. Chapter 2. State of the Art / Pharmacovigilance informatics 28 For example, the concept of outcome for the FDA form as a similar meaning to the concept of seriousness for the two European databases analyzed. On the other hand, for these two databases, outcome means the evolution of the patient regarding the adverse scenario. Active debate is maintained about this subject, aiming to create a standard to the information that should be kept during the ADR registry. Promotion of Adverse Drug Reaction Reporting Integration of pharmacovigilance system databases with other healthcare IS seems to be an obvious way to improve the knowledge about drugs safety. In fact, healthcare providers insert a lot of information in their EHR about ADR, which is not shared with pharmacovigilance systems. Every approach that promotes drug safety surveillance without increasing the workload of healthcare professionals, one of the main reasons for not reporting ADR, should be considered. The creation and implementation of webservices to collect this information is one of the possible solutions for this problem. This solution is being tested in a Portuguese multi-center research project, in the field of gastrointestinal diseases [26] , with promising preliminary results (See chapter 5). In this case, the physicians insert the usual clinic information during the appointment with the patient, and then, they only have to authorize the transmission of anonymized information about drug-related problems to the pharmacovigilance system. This tool allows ADR reporting without the need to fill the ADR reporting form and with no additional administrative work for the physician other than the normal registry of clinic patient data. This solution is also available to be included in commercial prescription software. For the implementation of these webservices, it was necessary to map the form entries used by the doctors with the online ADR reporting form developed by the pharmacovigilance system, so that the information is correctly collected. This mapping would be easier if a standard as the one described in the previous section was already in use. An ADR electronic reporting system included into a IS was developed in a Spanish hospital [27] , allowing healthcare professionals (pharmacists, physician and nurses) to report suspected ADR through their usual IS. The biggest advantage of this system is that some data (already included in the IS) appears as default values into the form, which expedites the system and reduces transcription errors [28] . All the reports made by this system are reviewed by a pharmacist, which is responsible to confirm the included data and to report the case to Chapter 2. State of the Art / Pharmacovigilance informatics 29 pharmacovigilance system. This might be a disadvantage, as some cases may be lost and not actually sent to regulatory authorities. Another strategy that can be easily adopted is the inclusion of hyperlinks in the EHR to the online ADR reporting form. This solution was tested between 2006 and 2010, in an ecological study performed in 16 Portuguese hospitals centres [16] . The hyperlinks were included in either EHR or on computer desktops. Considering the hospitals with hyperlink included in the EHR, the median ADR reports per month significantly increased, from two (range 0–12) to five reports (range 1–17). The median of ADR reports per month using the online form also increased significantly, from one (range 0–5) before the intervention to four (range 1–17) after it. Moreover, serious ADR increased 3-fold, and non-previously described ADR increased 4.5fold. None of these significant increases were observed in the hospitals where the hyperlink was not installed. It was also found a significant increase in daily pharmacovigilance centre website visits, from ten before the intervention to 27 after it (p<0.001). The increase in ADR reporting shows that the inclusion of hyperlinks to online ADR reporting forms is an easy way to change health professional behavior with regard to spontaneous ADR reports. Furthermore, this solution seems to be cost effective, when compared with other strategies to increase ADR report, as it has no running costs (after the hyperlink implementation, there are no additional costs to the pharmacovigilance system). (See chapter 4) Chapter 2. State of the Art / Systematic Review and Meta-Analysis 36 Abstract Background: Adverse drug reactions (ADRs) are a well-recognized public health problem and a major cause of death and hospitalization in developed countries. The safety of a new drug cannot be established until it has been on the market for several years. Keeping drug reactions under surveillance through pharmacovigilance systems is indispensable. However, underreporting is a major issue that undermines the effectiveness of spontaneous reports. Our work presents a systematic review on the use of information systems for the promotion of ADR reporting. The aim of this work is to describe the state of the art information systems used to promote adverse drug reaction reporting. Methods: A systematic review was performed with quantitative analysis of studies describing or evaluating the use of information systems to promote adverse drug reaction reporting. Studies with data related to the number of ADRs reported before and after each intervention and the follow-up period were included in the quantitative analysis. Results: From a total of 3865 articles, 33 articles were included in the analysis; these articles described 29 different projects. Most of the projects were on a regional scale (62%) and were performed in a hospital context (52%). A total of 76% performed passive promotion of ADR reporting and used web-based software (55%). A total of 72% targeted healthcare professionals and 24% were oriented to patient ADR reporting. We performed a meta-analysis of 7 of the 29 projects to calculate the aggregated measure of the ADR reporting increase, which had an overall measure of 2.1 (95% IC 1.7-2.6), indicating that the interventions doubled the number of ADRs reported. Conclusions: We found that most of the projects performed passive promotion of ADR reporting (i.e., facilitating the process). They were developed in hospitals and were tailored to healthcare professionals. These interventions doubled the number of ADR reports. We believe that it would be useful to develop systems to assist healthcare professionals with completing ADR reporting within electronic health records because this approach seems to be an efficient method to increase the ADR reporting rate. When this approach is not possible, it is essential to have a tool that is easily accessible on the web to report ADRs. This tool can be promoted by sending emails or through the inclusion of direct hyperlinks on healthcare professionals’ desktops. Chapter 2. State of the Art / Systematic Review and Meta-Analysis 37 Background Adverse drug reactions (ADRs) are a well-recognized public health problem worldwide and a major cause of death and hospitalization in developed countries [12] . Rare and long-term ADRs are difficult to detect during the drug development stage. Detecting new ADRs not previously identified during clinical trials is only possible when the drug begins to be used by a large population after marketing authorization (MA). The safety of a new drug cannot be established until it has been on the market for several years [8] . As such, it is indispensable to keep drug reactions under close surveillance after commercialization through a pharmacovigilance system to continuously evaluate the drug’s safety profile. In most countries, the pharmacovigilance system is based on spontaneous ADR reports made by healthcare professionals and consumers [15] . These reports can be made using paper, telephone, e-mail or through an on-line form and consist of a description of an adverse event apparently caused by a medicine. Spontaneous ADR reporting has been described as an efficient method to detect drug safety signs [39] ; however, underreporting is a major issue that undermines the effectiveness of spontaneous reports. Several studies have suggested that less than 10% of detected ADRs are effectively reported to medicine regulatory authorities [10, 40] . Worldwide, systems using informatics to promote ADR reporting or to detect the occurrence of ADRs in healthcare institutions have been tested and used, such as computer programs that allow voluntary and automated detection of ADR [4, 41] informatics tools created to analyse clinical databases [42] or websites that actively inform healthcare professionals [43] . In addition to signal detection, information and communication technologies can also be used to encourage and facilitate reporting of suspected ADR. In the present work, a systematic review is presented on the use of information systems in pharmacovigilance. Our main goal is to describe the state of the art information systems for the passive or active promotion of adverse drug reaction reporting. Chapter 2. State of the Art / Systematic Review and Meta-Analysis 38 Methods Eligible studies Studies describing or evaluating the use of information systems to promote adverse drug reaction reports were selected. Review team The review team is composed of two pharmacists who are experts in pharmacovigilance (Inês Ribeiro Vaz (IV) and Ana Marta Silva (AS)) and the computer scientist Ricardo Cruz Correia (RC), who is an expert in medical informatics. Search methods Studies were searched in April 2014 in the bibliographic databases. We developed a search query that included the concepts adverse drug reaction, adverse drug reaction reporting system, pharmacovigilance and information system. Only articles written in English, Portuguese or French were included. We did not establish any criteria for the publication date. Four distinct bibliographic databases were searched: Medline (via PubMed); ISI (ISI Web of Knowledge); IEEE (IEEE Xplore) and Scopus. The query search string used in Medline® was ((ADR OR "adverse drug reaction" OR "adverse drug reactions" OR "adverse drug event" OR "adverse drug events" OR "adverse dug effect" OR "adverse drug effects") OR “pharmacovigilance”). A similar query was used in the other databases and was adapted to the search engine. Selection of studies for the review The first selection was based on the study title and abstract (when available). Two reviewers on the review team (IV and AS) were involved in study selection and read all titles/abstracts. The study was considered eligible when at least one of the reviewers decided that the title/abstract mentioned the key concept of using information systems for ADR reporting. In cases of disagreement, a consensus meeting was held with the third reviewer (RC) to decide whether the article should be selected. The second phase of study selection was based on the full text. The team leader (IV) reviewed each full-text article. In this stage, articles were excluded based on the following criteria: (1) the articles were only focused on medication errors; (2) the articles focused on ADR detection; (3) the articles were studies without any information system implemented; (4) the articles were studies concerning data quality; (5) the articles were studies focused on website Chapter 2. State of the Art / Systematic Review and Meta-Analysis 39 usability; (6) the articles were only the authors’ reflections on the theme; (7) the articles were studies only related to incidents that occurred in health institutions; (8) the articles were studies concerning signal detection and (9) the articles were studies concerning electronic transmission between the authority and other institutions (pharmaceutical companies or regional pharmacovigilance centres). The articles remaining after this review were included in the final statistical analysis. These articles were grouped into research projects to avoid the distortion created by multiple papers describing the same project (Figure 1). All statistical analyses were based on the projects and not on the articles. Definition of variables The variables examined in these reviews were related to the projects, papers and information systems described in each project. We used the following data for project identification: (1) project number; (2) Information system name (if any); (3) country; (4) publication date; (5) type of study and (6) reference(s). According to the description of the projects, the following variables were analysed: 1. Area covered by the project (i.e., region, country, or hospital) 2. Type of action promoted by the project (passive promotion of ADR reporting or active promotion of ADR reporting) 3. Type of software (i.e., web-based or mobile) 4. Type of institution (i.e., regulatory authority or universities) 5. Target (healthcare professionals or patients) 6. Type of medicine (all, vaccines, chemotherapy, or others) 7. Type of ADR (all/serious ADRs based on the World Health Organization seriousness criteria [44] ) Statistical analysis The inclusion criteria for the quantitative analysis were the availability of data related to the number of ADRs reported before and after each intervention and a follow-up period. Studies that only disclosed the increased ADR rate and studies that reported zero ADRs before the project implementation were excluded because it was not possible to perform the analysis in these cases. Chapter 2. State of the Art / Systematic Review and Meta-Analysis 40 For each study with available data, the rate of ADRs reporting increase (quotient between ADR reports after and ADR reports before) and the respective 95% confidence intervals were calculated. A rate of ADR reporting increase equal to 2 indicated that the ADR reports doubled after the intervention. Conversely, a rate of ADR reporting increase equal to 1 indicated that the number of ADR reports after the intervention was equal to the number of ADR reports before the intervention. The aggregated rate of the ADR reporting increase was calculated with the inverse variance method using a random effects model and a forest plot was presented. The confidence intervals, aggregated rate of ADRs and forest plot were performed using a Microsoft Excel spreadsheet. The description of the Microsoft Excel spreadsheet and the respective statistical methods used were described by Neyeloff [45] . Results Our search method found 2519 articles in PubMed, 68 in IEEE, 2603 in ISI and 192 in Scopus. After eliminating duplicate articles, 3835 articles were selected. Two reviewers (IV and AS) read all 3835 titles/abstracts. In cases of disagreement, which occurred with 151 articles, a consensus meeting was held with the third reviewer (RC) to decide whether the article should be selected. A total of 643 studies were excluded because they were not related to pharmacovigilance, 85 were excluded because they were not related to information systems and 2993 were excluded for other reasons (mostly because their focus was on data mining in large databases instead of ADR reporting). A total of 114 of the 3835 articles were selected in this first selection based on the title and abstract. The team leader (IV) reviewed each of the 114 full-text articles. After this review, 33 articles remained for the final statistical analysis. At this stage, most of the articles were excluded because: (1) they were only related to medication errors; (2) they were focused on ADR detection; (3) they were studies without any information system implemented; (4) they were studies concerning data quality; (5) they were studies focused on website usability; (6) they were only authors’ reflections on the theme; (7) they were studies only related to incidents that occurred at health institutions; (8) they were studies concerning signal detection or (9) they were studies concerning electronic transmission between the authority and other institutions (pharmaceutical companies or regional pharmacovigilance centres). Chapter 2. State of the Art / Systematic Review and Meta-Analysis 41 These 33 articles were grouped into 29 distinct research projects to avoid the distortion created by multiple papers describing the same project (Figure 1.). All statistical analyses was based on projects and not on articles. Figure 1. Flowchart of study selection Table 2 lists all 29 projects, their country, the number of publications, the publication year and the journal. The country with the most published projects was the USA (11), followed by the United Kingdom (3). Chapter 2. State of the Art / Systematic Review and Meta-Analysis 42 Trends There was an increasing trend in publication, especially after 2009, as seen in Figure 2. Figure 2. Number of publications by year 0 1 2 3 4 5 6 7 8 1992 2001 2002 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Number of publications/year Chapter 2. State of the Art / Systematic Review and Meta-Analysis 43 Table 2. Project identification Project number System name (if any) Country Number of publications Publication date(s) References Journals 4 USA 1 1992 [ 4 ] Hospital pharmacy 28 France 1 2001 [ 46 ] Fundamental & Clinical Pharmacology 11 Japan 1 2002 [ 47 ] Yakugaku Zasshi - Journal of the Pharmaceutical Society of Japan 17 USA 1 2004 [ 48 ] American Journal of Healt h - System Pharmacy 2 USA 3 2005, 2007 [49-51] Journal of Clinical Oncology, Journal of American Medical Information Association 22 USA 2 2005, 2006 [52, 53] Biosecurity and Bioterrorism - Biodefense Strategy Practice and Science, Health Expectations 10 USA 1 2007 [ 54 ] Journal of the American Medical Informatics Association 9 MEADERS USA 2 2007, 2010 [ 55 , 56 ] Annals of Family Medicine, AMIA Annual Symposium proceedings 21 Spain 1 2008 [ 27 ] An nals of Pharmacotherapy 7 Sweden 1 2009 [ 57 ] European Journal of Clinical Pharmacology 5 Canada 1 2010 [ 58 ] International Journal of Medical Informatics 13 Canada 1 2010 [ 59 ] Vaccine 18 USA 1 2010 [ 60 ] Pharmacoepidemiology and Drug Safety 1 9 United Kingdom 1 2010 [ 61 ] Archives of Disease in Childhood 23 ALIAS USA 1 2010 [ 62 ] Contemporary Clinical Trials 27 Taiwan 1 2010 [ 63 ] Value in Health 8 United Kingdom 1 2011 [ 64 ] Journal of Psychiatric and Mental Health Nursing 12 Serbia 1 2011 [ 65 ] Drug Safety 14 France 1 2011 [ 66 ] Therapie 15 USA 1 2011 [ 67 ] Paediatrics 6 United Kingdom 1 2012 [ 68 ] Drug Safety 16 Korea 1 2012 [ 69 ] Yonsei Medical Journal. 20 Portugal 1 2012 [ 16 ] Drug Safety 25 USA 1 2012 [70] 2012 Ni nth International Conference on Information Technology: New Generations 1 Cambodge 1 2013 [ 71 ] Journal of Medical Internet Research 3 Netherlands 1 2013 [ 72 ] Studies in health technology and informatics 24 SALUS France 1 2013 [ 73 ] St udies in health technology and informatics 26 Spain 1 2013 [ 74 ] International Journal of Clinical Pharmacy 29 Denmark 1 2013 [ 75 ] European Journal of Hospital Pharmacy - Science and Practice Chapter 2. State of the Art / Systematic Review and Meta-Analysis 44 Qualitative analysis The qualitative variables analysed in each project are listed in Table 3 and described below. Globally, we found that there was an increase in the publication of projects over the study period, with 4 projects published before 2001, 4 projects between 2005 and 2007, 8 projects between 2008 and 2010 and 13 projects between 2011 and 2013. Geographic area covered by the projects Most of the projects were regional (62%), followed by national projects (34%). We found only 1 international project based on Facebook®. This international project was developed in the last time period (2011-2013). Areas covered by the projects Most of the projects (52%) were developed in hospitals, followed by community projects (21%). A total of 14% covered primary care institutions and 10% (3 projects) were developed for use in any type of healthcare institution. One project was dedicated to a multicentre clinical trial. We also found that all of the projects oriented to the community were developed in the last three years (2011-2013). Types of actions promoted by the projects The majority of the projects passively promoted ADR reporting (76%); the remainder actively promoted reporting (24%). Types of software More than half of the projects (55%) used web-based technology and 41% used electronic health records. Only one project used mobile phone technology. There was an increasing trend in software using web-based technology over all of the time intervals considered. The mobile technology appeared during the last time period. Types of institutions promoting the studies Most of the projects were promoted by hospitals and universities (31% ex aequo). There were 4 projects developed by national institutions (not regulatory) and 5 projects implemented by regulatory authorities. Chapter 2. State of the Art / Systematic Review and Meta-Analysis 45 Targets A total of 72% of the projects were geared to healthcare professionals, 24% to patients and one project was geared to both targets. Most of the projects targeting patient ADR reporting were developed in the last years considered (2011-2013). Types of medicine Most of the projects (72%) covered all medicines, but 17% were specific to vaccines. There were also projects specific to reporting ADRs due to chemotherapy, human albumin and radiopaque agents (1 project for each of these medicines). Types of ADR Only a small percentage of the projects were specific for serious adverse drug reactions. The majority (93%) covered all ADR. Chapter 2. State of the Art / Systematic Review and Meta-Analysis 52 Conclusions Our systematic review allowed us to note some facts about interventions that aim to improve ADR reporting using information systems. According to our aggregation analysis, these interventions doubled the number of ADR reports. We also found that most projects passively promoted ADR reporting (facilitating the reporting process) and the countries involved in this type of project were Northern America countries (USA and Canada), European countries and in a smaller number Far East countries. Authors’ contributions Inês Vaz and Ricardo Correia were responsible for the development of the query. Inês Vaz also performed data collection and was the team leader for article screening and the writing of the manuscript. Ana Silva performed the article screening. Ricardo Correia was involved on the article selection when there was disagreement between the other two reviewers (Inês Vaz and Ana Silva). Cristina Santos performed the quantitative analysis. All four authors were involved in the writing of the manuscript, discussion and the revision of the whole article. Acknowledgements The authors would like to thank Daniel Pereira for the development of the informatics tool used for the abstract/title screening. Competing interests: The authors declare that they have no competing interests. Chapter 3. Promoting ADR reports: Comparison of different approaches 53 Chapter 3 - Promoting ADR reports: Comparison of different approaches Inês Ribeiro-Vaz, Cristina Costa Santos, Ricardo Cruz-Correia Rev Saude Publica. 2016 Chapter 3. Promoting ADR reports: Comparison of different approaches 54 Chapter 3. Promoting ADR reports: Comparison of different approaches 55 3 Promoting ADR reports: Comparison of different approaches Chapter 3. Promoting ADR reports: Comparison of different approaches 56 3.1 Abstract Objective: To describe different approaches to promote adverse drug reaction reporting among health care professionals, determining their cost/effectiveness. Methods: We analyzed and compared several approaches taken by the Northern Pharmacovigilance Centre (Portugal) to promote adverse drug reaction reporting. Approaches were compared regarding the number and relevance of adverse drug reaction reports obtained and costs involved. Costs by report were estimated by adding the initial costs and the running costs of each intervention. These costs were divided by the number of reports obtained with each intervention, to assess its cost-effectiveness. Results: All the approaches seem to have increased the number of adverse drug reaction reports. We noted the biggest increase with protocols (321 reports, costing 1.96 € each), followed by first educational approach (265 reports, 20.31 €/report) and by the hyperlink approach (136 reports, 15.59 €/report). Regarding the severity of adverse drug reactions, protocols were the most efficient approach, costing 2.29 €/report, followed by hyperlinks (30.28 €/report, having no running costs). Concerning unexpected adverse drug reactions, the best result was obtained with protocols (5.12 €/report), followed by first educational approach (38.79 €/report). Conclusions: We recommend implementing protocols in other pharmacovigilance centers. They seem to be the most efficient intervention, allowing receiving adverse drug reactions reports at lower costs. The increase applied not only to the total number of reports, but also to the severity, unexpectedness and high degree of causality attributed to the adverse drug reactions. Still, hyperlinks have the advantage of not involving running costs, showing the second best performance in cost per adverse drug reactions report. Chapter 3. Promoting ADR reports: Comparison of different approaches 57 3.2 Introduction Adverse drug reactions (ADR) are inherent to medicines use [78] , and most of them can only be detected after the commercialization of the drug [8] . In fact, during clinical trials, rare reactions are hardly detected, as well as the ones associated with chronic utilization of the drug. It is also difficult to predict the drug effect among special populations (pregnant, children, elderly), as they usually are not part of the clinical research. Because of these limitations, post-marketing surveillance is essential, which is why most countries have pharmacovigilance centres for monitoring of detected ADR. The fundamental tool used by these centres is the spontaneous report of ADR, made by healthcare professionals and consumers. This method consists in describing an adverse episode suspected to be caused by one or more drugs, and provides valuable information to the regulatory health authorities, which is important for the decisions about marketed medicines. The biggest problem of this method is the underreporting, ie, ADR are detected but not reported to national regulatory health authorities. Most developed countries suffers from this problem [10, 40] , and Portugal is not an exception [19] . Worldwide, many approaches have been completed to combat the major problem of ADR underreporting, such as regular visits to health professionals [79] , questionnaire studies [80] , educative interventions (including workshops, meetings and presentations) [19, 81, 82] , among others. This study aims to describe several approaches that intended to improve ADR reporting and determine the cost-effectiveness of each one of them. Chapter 3. Promoting ADR reports: Comparison of different approaches 58 3.3 Methods From its creation (in 2000) to 2003, Northern Pharmacovigilance Centre, a Portuguese regional pharmacovigilance centre, had an extremely low rate of ADR reports, about 43 ADR reports per year/1 million of inhabitants. We realize this value is very low, when compared with the World Health Organization (WHO) recommendation for an Optimal National Centre which is, at least, 200 reports per year/1 million of inhabitants 1 . To reach its objectives, in 2004, the Centre established a collaboration protocol (protocol approach) with the Immunoallergology Department of a Central Hospital (located on the same street as the Centre), in order to collect every suspected case of ADR emerged in appointments related to drug allergies. This collaboration includes regular visits of the pharmacovigilance staff to the Immunoallergology department, to collect the detected cases in ADR report forms, under the physician supervision. Then, the form is signed by the physician and it follows the normal course of all the ADR spontaneous reports. This approach was replicated two more times, in 2007 and 2009, in two other immunoalergology departments, one from a specialized hospital (pediatric hospital, located at 6 km from the Centre), and another from a central hospital (located 11 km from the Centre). These three protocols remain active. A study conducted in 2004 provided educational interventions (educational approach) for physicians and pharmacists [19, 20] . Those interventions were based on a previous case-control study that identified the reasons for underreporting [83, 84] . The educational approach includes workshops about pharmacovigilance at health care professionals’ working places. Since the effect of educational interventions decreased over time, the authors of the previously described work promote reinforcement interventions (educational and telephone approach). We started a new study in 2007, also among physicians and pharmacists. This study consisted not only in outreach interventions (workshops), but also in telephone interviews [17, 18] . The phone interviews followed a script about ADR and the importance of reporting. Details are described in a previous publication [18] . Meanwhile, we propose a new approach: the inclusion of a hyperlink (hyperlink approach) to an online ADR reporting form on hospitals' electronic patient records (EPRs). The main aim of this study, performed from 2006 to 2010 was to evaluate the impact of these hyperlinks on the number of spontaneous ADR reports [16] . The inclusion of hyperlinks began in December 2007 and continued over the following five months. 1 Farmacovigilância em Portugal. Lisboa: INFARMED - Autoridade Nacional do Medicamento e Produtos de Saúde, I.P.; 2004. Chapter 3. Promoting ADR reports: Comparison of different approaches 59 The temporal distribution of all these approaches is shown in Figure 4 Figure 4. Timeline of the studied approaches In the present work, we analyzed the number of ADR reports obtained with each one of the described approaches. We know exactly which ADR reports were originated at the three departments participating in the protocol intervention and analyzed them separately. Four physicians were involved. The first educational intervention (in 2004) involved three hospitals, 26 healthcare centres and 73 pharmacies. About 900 healthcare professionals attended these interventions [19] . About 340 healthcare professionals (physicians and pharmacists) attended the second intervention (second workshop + telephone, both in 2007). Five healthcare centres, two hospitals and 40 pharmacies received the telephone intervention, and 16 healthcare centres, 2 hospitals and 23 pharmacies were intervened with the 2 nd educational intervention. For the hyperlinks, we estimated 15,000 health care professionals potentially affected by the intervention, as this is the total number of professionals working at the 12 participating hospital centers (corresponding to 22 hospitals). It was the first exposure to any intervention for eight of these hospital centers. The variables analyzed were: type of approach, ADR relevance, initial costs of the interventions, running costs of the interventions, and costs per ADR report. Each of these variables is described as follows. Type of approach: hyperlink approach, protocol approach, educational and telephone approach. Number of ADR reports obtained with each intervention: the difference between ADR reports received two years after the intervention and ADR reports received two years before the intervention. Chapter 3. Promoting ADR reports: Comparison of different approaches 60 ADR relevance: we adopted the following criteria: (1) ADR seriousness; (2) ADR expectedness; and (3) causality attributed to the ADR report. A serious ADR is any untoward medical occurrence that results in death, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent, or significant disability or incapacity, or is lifethreatening 2 . An unexpected ADR is the one which the nature or severity is not consistent with domestic labeling or market authorization, or expected from characteristics of the drug 2 . According to the causality attributed we considered that an ADR was more relevant if it was evaluated with one of the 2 higher degrees of causality: Definitive/certain or probable (the evaluation of the likelihood that a medicine was the causative agent of an observed adverse reaction) 2 . Interventions initial costs: We consider as initial costs the expenses needed for the establishment of the approach, as educational material and staff working hours. These costs are described in Table 5. Table 5. Estimated values of each approach Approach Initial costs Annual running costs Value Description Value Description protocol approach 150€ pharmacovigilance and clinical services staff working hours 240€ fuel, material and pharmacovigilance and clinical staff working hours hyperlink approach 2120€ pharmacovigilance and software development staff working hours --- --- educational approach 200€ educational material and pharmacovigilance staff working hours 2500€ fuel, material and pharmacovigilance staff working hours telephone approach 400€ telephone calls during the pilot-study and pharmacovigilance staff working hours 800€ telephone calls, material and pharmacovigilance staff working hours Interventions running costs: Annual running costs are the expenses needed for the continuation of the projects, as fuel, material and staff working hours. These costs are described in table 1. In our study, we did not consider the normal (daily) costs of ADR reports processing, as we only meant to compare the costs involved to obtain ADR reports. 2World Health Organization. The Uppsala Monitoring Centre Uppsala WHO Collaborating Centre for International Drug Monitoring [09 Aug 2013]; Available from: http://www.whoumc.org/DynPage.aspx?id=22682. Chapter 3. Promoting ADR reports: Comparison of different approaches 61 Costs per ADR report: Costs for ADR report were calculated adding initial costs and running costs. Initial cost per ADR report were obtained by dividing initial costs by the difference between the ADR reports received two years after the intervention and ADR reports received two years before the intervention (which we consider to have been the number of reports earned with each intervention). Running costs were obtained by dividing the running costs of the two-year intervention by the number of notifications earned with each intervention. To assess the cost/effectiveness of each intervention, we considered these added costs (initial + running costs), as this total means the total cost of each ADR obtained in two years following each intervention. The pharmacovigilance center website uses a web server and has audit trails that read each site visit since 2006. These audit trails are processed using the Webalizer program (www.webalizer.org) to estimate site hits, user logins and visits. ADR reports obtained by these approaches are included in a database. We collect them by selecting the report date and origin. We presented the total number of reports received in each quarter during the period studied. For each health institution, ADR reports made before and after the intervention, if any, were measured. To examine whether each intervention increased the ADR report trend, an interrupted time series analysis using autoregressive integrated moving average (ARIMA) was performed using quarter data of ADR reports, as well as each intervention (first and second educational approach, telephone approach, and hyperlink approach) as dichotomous variables (before and after intervention). We performed an additional analysis with the hyperlinks approach, to consider the institutions exposed to any type of intervention for the first time. With this sub-analysis we intended to isolate the ADR reports obtained with each intervention. This study was approved by the local Ethics Committee of the Faculdade de Medicina of the Universidade do Porto (Process PCEDCSS-FMUP 08/2014, approved in May 7, 2014). Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 68 Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 69 4. Promoting spontaneous Adverse Drug Reaction reporting in Hospitals using a hyperlink to the on-line reporting form: An ecological study in Portugal Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 70 4.1 Abstract Background: Spontaneous Adverse Drug Reactions (ADR) reporting has been described as an efficient method to detect drug safety signs. However, underreporting is a major issue undermining the effectiveness of spontaneous reports. Among hospitalized patients, adverse drug reactions are a particularly serious problem, because these patients are often treated with more than one drug, and with new and aggressive drugs. Objective: In order to promote spontaneous ADR reporting by healthcare professionals working in northern region Portuguese hospitals, we propose the inclusion of a hyperlink to an on-line ADR reporting form on the Electronic Patient Records (EPR) of hospitals. The main aim of this work is to evaluate the impact of these hyperlinks in the number of spontaneous ADR reports to Northern Pharmacovigilance Centre (UFN – Unidade de Farmacovigilância do Norte). We also assess the number of UFN web site daily visits before and after the hyperlinks inclusion. Methods: An ecologic study was performed in the Northern Portuguese Hospitals from 2006 to 2010. The hyperlinks were included either in EPR or computers desktops. The median of spontaneous ADR reports (total and on-line) per month and respective range were presented before and after the intervention in all hospitals. The comparisons were performed using the Mann-Whitney U test. Results: There were 16 hospital centres involved in the study. Eleven centres (18 hospitals) included the hyperlinks. Considering the hospitals with hyperlink, the median ADR reports per month significantly increased, from 2 (range 0-12) to 5 reports (range 1-17). The median of ADR reports using the on-line form per month also increased significantly, from 1 (range 0-5) before the intervention to 4 (range 1-17) after it. Moreover, serious ADR increased 3 fold, and non-previously described ADR increased 4.5 fold. None of these significant increments were observed in the other hospitals without the hyperlink. We also found a significant increase of UFN web site daily visits from 10 before the intervention to 27 after it (p<0.001). Conclusions: The increase in ADR reporting shows that the inclusion of hyperlinks to on-line ADR reporting forms is an easy and cost-effective way to change health professional behaviours on ADR spontaneous report. Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 71 4.2 Introduction Background Adverse Drug Reactions (ADR) are a well-recognized public health problem worldwide, and a major cause of death and hospitalization in developed countries [12] . In fact, rare and long term ADR are difficult to detect during the drug development stage. Only when the drug begins to be used by a large population after Marketing Authorization (MA), it is possible to detect new ADR not previously identified during clinical trials. In reality, it is known that the safety of a new drug cannot be established until it has been on the market for several years [8] . As such, it is indispensable to keep drug reactions under close surveillance, after its commercialization, through a pharmacovigilance system. In Portugal, this system is based on spontaneous ADR reports made by healthcare professionals and, since 2012, also by consumers [15] . These reports can be made using paper, telephone, e-mail or through an on-line form [87] and consist of a description of an adverse event supposedly caused by a medicine. Spontaneous ADR reporting has been described as an efficient method to detect drug safety signs [39] . However, underreporting is a major issue undermining the effectiveness of spontaneous reports. Several studies suggest that less than 10% of detected ADR are effectively reported to medicines regulatory authorities [10, 40] . Besides, spontaneous ADR report rate in Northern Portugal was 90 reports/million inhabitants in 2009, which is highly unsatisfactory according the World Health Organization recommendations (200 reports/million inhabitants [88] ). Worldwide, systems using informatics to promote ADR reporting or to detect ADR occurred in healthcare institutions have been tested and used, such as computer programs that allow voluntary and automated detection of ADR [4, 89] , informatics tools created to analyse clinical databases [90] , or Web sites that actively inform healthcare professionals [43] . Among hospitalized patients, adverse drug reactions are a particularly serious problem. In fact, these patients are often treated with more than one drug, and with new and aggressive drugs. In spite of this, there are no specific systems for monitoring or reporting ADR in Portuguese hospitals. According to the characteristics of Portuguese Healthcare Professionals, we believe that making the reporting system easier would increase considerably the number of ADR reports. Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 72 Intending to promote spontaneous ADR reporting by healthcare professionals working in hospitals, we propose the inclusion of a hyperlink to an on-line ADR reporting form [part of the Northern Pharmacovigilance Centre (UFN) web site] on the Electronic Patient Records (EPR) or on the desktop of hospital computers. With this system, we expected to reach not only the physicians, but also the pharmacists and nurses working at the hospitals. In Portuguese hospitals, pharmacists has an important role in ADR detection and reporting, because physicians discuss with them the adverse events that occurred during medical treatment, asking for alternative drugs available at the pharmacy. Besides, some pharmacists are part of the clinical visit and detect ADR. Aim Our main aim is to evaluate the impact of the hyperlinks implemented in Portuguese hospitals in the number of ADR reported by these hospitals using the hyperlink or not, and in the visits to the UFN web site. 4.3 Methods Intervention Hyperlinks to the ADR on-line reporting UFN form were proposed to the 18 Northern Portuguese Hospitals Centres. The hyperlinks can be included either in healthcare professional specific software (typically EPR or Pharmacy specific applications used by doctors, nurses and pharmacists), or at the computer desktop (see Figure 6 for examples of both situations). It should be noticed that most of the Portuguese EPR are web-based and so the hyperlink opens the UFN form in a new browser window. Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 73 Figure 6. Two examples of hyperlinks in a computer desktop (back image), and on an Electronic Patient Record (front image) The on-line reporting UFN form requires the health professional to login with their personal account. The patient data is collected in an anonymous way (only the initials of the patient name are required aiming to help the health professional identify each case) and the data is secured in an Oracle database with proper access restrictions. In the beginning of October 2007 a letter was sent to the chief physicians of the 18 Northern Portuguese Hospital Centres suggesting the inclusion of the hyperlink. If there was no answer in two weeks, clinical administration boards were reminded by telephone. Thirteen centres forwarded this issue to the respective informatics departments and only one to the pharmaceutical department. Five of them have not answered until the end of 2010. After the approval by the hospital board and the forward to the respective departments, UFN made a third contact in order to explain technical doubts and to send the specific hyperlink of each Hospital. Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 74 Study design and data collection An ecologic study was carried out in the Northern Portuguese Hospitals from 2006 to 2010. The number of spontaneous ADR reports and on-line spontaneous ADR reports originated in hospitals were analysed before and after an intervention without a control group. Five hospital centres implemented the hyperlink on December 2007 and the other 6 implemented it during the next 5 months. We considered 23 months before and the 31 months after each hospital implementation. The UFN web site uses an Apache web-server having the web logs related to the site visits recorded since January 2006. These logs were initially processed using Webalizer software (www.webalizer.org) to calculate site hits, users and visits. Telephonic interviews with the informatics departments of each hospital were performed to collect where each institution putted the hyperlink to the UFN website, and screen-shots illustrating the interventions were taken. Variables The main variables collected for analysis were: • Date – date and time of the ADR report; • Hospital – institutions were the ADR was detected; • Health professional – type of the health professional that reported (doctor, nurse, pharmacist and others); • Seriousness – seriousness of ADR grouped in “Serious” or “Non-serious” according to the WHO criteria; • Previous knowledge – was the ADR previously described on the Summary of Product Characteristics or not. Bias From the initial 18 centres (31 hospitals), we excluded 4 hospitals that established other cooperation protocols with UFN in order to avoid possible confounder’s bias (see Figure 7). For the other 16 hospitals, we believe that there were no external interventions that could potentially explain the observed results. Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 75 Figure 7. Diagram describing the hospitals and hospital centres of the Portuguese northern region included and excluded from the study. Statistical methods The number of spontaneous ADR reports and on-line spontaneous ADR reports per month were compared between the two periods (before and after the intervention). The number of ADR reports per quarter before and after the installation of the hyperlinks was presented graphically (see Figure 8). The number of ADR reports per quarter in the excluded hospitals and in the hospitals that did not participate (did not installed the hyperlinks) was also presented graphically. Median values of number of daily UFN web site visits were reported because of the skewed distribution of data. The number of daily UFN web site visits was compared before and after the intervention using the Mann-Whitney U test. A significance level of 0.05 was used. Northern Portuguese Hospitals 18 centers (31 hospitals) Included Northern Portuguese Hospitals 16 centers (27 hospitals) 2 centers (3 hospitals) and 1 hospital from another center that established cooperation protocols with Northern Pharmacovigilance Center were excluded Included Northern Portuguese Hospitals with the hyperlink 11 centers (18 hospitals) Hyperlink in the Electronic Patient Record 8 centers (12 hospitals) Hyperlink in the Desktop 2 centers (5 hospitals) Hyperlink both in the desktop and in the Electronic Patient Record 1 center (1 hospital) 5 centers (9 hospitals) did not include the hyperlink Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 76 4.4 Results Participants From the 16 centres involved in the study, 11 centres (18 hospitals) included the hyperlinks. Eight centres included the hyperlink only in the EPR, two centres included the hyperlink in the computer desktop and one included in both desktop and EPR (see Figure 7). From the 18 involved hospitals one is a University hospital and three are specialized hospitals. Main results Considering the 16 centres that implemented the project, the median of ADR reports per month, significantly increased after the project implementation. In fact, before the intervention the median of total ADR reports per month was 2, range from 0 to 12, and 31 months after the intervention was 5, range from 1 to 17 (p=0.043). Four months after the project implementation the median number of reports per month was 4. Considering only the reports using the on-line form, before the project implementation the median of total on-line ADR reports per month was 1, range from 0 to 5, and after the intervention was 4, range from 1 to17 (p=0.009). Figure 8 shows the number of ADR reports per quarter before and after the intervention and Figure 9 presents the number of ADR reports per quarter in the excluded hospitals and in the hospitals that did not installed the hyperlinks. Figure 8. Evolution of ADR reports (total and just on-line) in hospitals with the intervention. The quarters were adjusted for the time of intervention. 5 hospitals centres implemented the hyperlink on Dec. 2007, 2 on Jan. 2008, 1 on Feb., 1 on Mar., 1 on Apr. And 1 on May Chapter 4 – Promoting spontaneous ADR reporting in Hospitals using hyperlinks 77 Figure 9. Evolution of ADR reports in the excluded hospitals (with cooperation protocols) and in the hospitals that did not install the hyperlinks The 11 included centres, reported 17 in 23 months serious ADR using the on-line form before the intervention, and 69 in 31 months (increase of 3 fold) after the intervention. Before the intervention these centres reported 7 non-previously described ADR using the on-line form and reported 42 after (increase of 4.5 fold). The hospital with the larger increase of ADR reports submitted on-line was the one that included the hyperlink in both EPR and desktop, with a mean of 3 on-line reports per year before the intervention and 18 after. The three hospitals that included the hyperlink in the desktop (one of them simultaneously with EPR) were in the top five of hospitals with higher increasing of ADR reports submitted on-line. There were no ADR reports sent both by paper and on-line by the same professional. Other analysis There was a significantly increase of UFN web site daily visits after the intervention (p<0.001). Before there was a median of 10 UFN web site daily visits and after increased to 27. 4.5 Discussion Our results show that the inclusion of hyperlinks to an on-line ADR reporting form on the EPR does change health professional behaviours on ADR report. In reality, there was an increase in ADR reporting in the hospitals involved in this project both in the total amount of ADR reports (more than 2 fold), in the amount of ADR reports submitted on-line (4 fold), in serious ADR reported on-line (3 fold) and non-previously described (more then 4 fold). Additionally, daily visits to the UFN website increased about 3 fold after the intervention. Chapter 5 – Using webservices 84 5.1 Abstract Purpose: Adverse drug reaction (ADR) reporting is an efficient method to assess the safety of drugs. However, underreporting is a major issue undermining the effectiveness of this method. Among patients with inflammatory bowel disease (IBD), ADR are a serious problem, because these patients are often treated with new and potent drugs. Electronic registries usually include information on ADR, recorded by physicians. To promote ADR reporting by gastroenterologists working in a multicentre IBD study group, we proposed the utilisation of a webservice in their usual electronic health records (EHR) to collect ADR. The aim of this work was to describe the impact of this intervention on the number of ADR reported to the regulatory authority through a regional pharmacovigilance centre. Methods: A study was performed between 2013 and 2015. A webservice was created and implemented in an EHR in use. We analysed the trends and type of ADR reported through this service. Results: From April 2013 to February 2015, 167 ADR reports were sent to the Northern Pharmacovigilance Centre through the webservice, comprising 10% of the total ADR reports received in the same period. Of these, 118 cases were serious (one was life-threatening). In the northern region of Portugal, GEDII physicians reported 9 ADR during the 23 months previous to webservice implementation and 121 ADR during the 23 months after webservice implementation, i.e. an increase of 81%. Conclusions: This solution allowed for reporting 167 ADR during the first 23 months of implementation, simply by clicking a button included in the usual EHR used by gastroenterologists. These results suggest that information systems (IS) should facilitate the reporting of ADR. . Chapter 5 – Using webservices 85 5.2 Introduction All medicines have adverse effects, some of them unknown until the drug is commercialised. Thus, it is crucial to implement strategies to monitor drug safety. Pharmacovigilance is the activity of drug surveillance, after its launch onto the market, with the main goal of public health protection, ensuring that the drug benefit outweighs its risks. Worldwide, pharmacovigilance systems are mostly based on spontaneous ADR reports made by healthcare professionals and consumers. Spontaneous ADR reporting has been described as an essential method to detect drug safety signals; however, underreporting is a major issue undermining the effectiveness of spontaneous reports. Several studies suggest that fewer than 10% of detected ADR are effectively reported to medicine regulatory authorities (e.g. Food and Drug Administration - FDA, European Medicines Agency - EMA, etc.). [7, 10] It is known that one of the main reasons why healthcare professionals do not report ADR is due to an increase in their workload [48, 54] . So, in order to reduce ADR reporting efforts, information systems (IS) to promote ADR reporting or to detect ADR in healthcare institutions have been tested and used, such as software for voluntary and automated detection of ADR, tools that analyse clinical databases or web-sites that actively inform healthcare professionals [11] . Information and communication technologies can also be used to facilitate and promote ADR reporting, such as the creation of on-line reporting forms and the development of tools to collect safety data from electronic health records (EHR), among others [16, 27] . In Portugal, there is a multi centre research project, in the field of gastric diseases (Study Group of Inflammatory Bowel Disease – GEDII) [26] whose members use the same electronic health record to collect patient information. As these patients are often treated with new and potent drugs (e.g. immunomodulating agents), the EHR has a field related to medication and ADR. Since the group members already fill in this field, it was considered an advantage to create a tool to send the data to the pharmacovigilance system. Aim The aim of this work was to implement and describe the implementation of a webservice in an EHR to collect ADR reports and analyse the number of ADR reports sent to the Portuguese regulatory authority (INFARMED), through a regional pharmacovigilance centre. Chapter 5 – Using webservices 86 5.3 Methods Intervention The multi-centre research project, the Study Group of Inflammatory Bowel Disease – GEDII [26] , has its own electronic health record (EHR) to collect patient information, with a field related to medications and ADR. The GEDII members asked for a collaboration with the Northern Pharmacovigilance Centre (UFN in Portuguese) and the Health Information and Decision Sciences Department (CIDES) of the Faculty of Medicine to develop a tool to allow for easy reporting of ADR included in the EHR to the pharmacovigilance system. The two entities had, at the time, released a new information service (a webservice) to collect suspected adverse reactions directly from prescription and medical records. This system was easily adapted to the EHR used by the GEDII group. The Northern Pharmacovigilance Centre, as part of the Portuguese Pharmacovigilance System, receives ADR reports from healthcare professionals (and, since 2012, from consumers as well) through an on-line form, a paper form, e-mail and by telephone. Since 2013, UFN has also received ADR reports through the webservice. The webservice anonymises patient data (converting the full name of the patient into initials), according to the data protection standards of the Portuguese Pharmacovigilance System. Figure 10 shows the information flow from the electronic health record to the pharmacovigilance system. Figure 10 Information flow Chapter 5 – Using webservices 87 To use this service, there is a button asking the doctor if he/she will allow the information to be sent to the pharmacovigilance database (see Figure 11). If the physician does not allow for this information to be sent to the system, it will only be stored in the health record. Figure 11 Screen shot of the ADR section on the GEDII electronic health record. Study design and data collection The webservice was implemented in the GEDII electronic health record (used only by gastroenterologists) in April 2013. There are 15 hospitals using this information system, covering a total of 4031 patients. The database has 39 registered users, which means that, potentially, 39 physicians could report ADR through this information system [26] . In order to use this webservice, it is necessary to access a specific URL, provided by the Northern Pharmacovigilance Centre and CIDES. The service was incorporated in the GEDII software to collect the ADR already stored by the physicians in the system. Each physician, in the context of the patient, sends the ADR to the Regional Pharmacovigilance Centre. “Do you want this information to be sent to the National Pharmacovigilance System?” (Yes/No) List of adverse drug reactions Drug Patient D emographics Chapter 5 – Using webservices 88 After the ADR report is received by the Regional Pharmacovigilance Centre, their technical staff process the report according to the pharmacovigilance guidelines [14] , sending it to the Portuguese Regulatory Authority (INFARMED). If the doctor does not want to send the information to the pharmacovigilance system, it will only be stored in the medical history. To analyse the trends and type of ADR reported through this service, we performed a descriptive analysis on the number and seriousness of ADR reported. The seriousness was assessed using the World Health Organisation seriousness criterion [6] . According to this criterion, a serious ADR is any untoward medical occurrence that at any dose: results in death, requires inpatient hospitalisation or prolongation of existing hospitalisation, results in persistent or significant disability/incapacity or is life-threatening. 5.4 Results From April 2013 to February 2015, physicians from GEDII reported 167 ADR through the described webservice. Of the 167 reported ADR, 118 (71%) were serious ADR, considering the World Health Organisation seriousness criterion [6] . One of the cases was life-threatening and none were fatal. To calculate the increase in ADR reporting in this period, we used data from the northern region, which is the delimited area of the Northern Pharmacovigilance Centre, as we did not have access to national ADR reporting data. Thus, considering the physicians from the GEDII group that work at the northern region of Portugal, 9 ADR were reported during the 23 months prior to webservice implementation and 121 ADR were reported during the 23 months after webservice implementation, i.e. an increase of 81%. Drugs involved in the reported ADR Most of the cases had one only suspected drug, but in six cases there were two suspected drugs involved. Of the reported ADR, 106 cases (63%) were presumably due to azathioprine, 23 cases (14%) to infliximab, 14 cases (8%) to adalimumab and 9 (5%) cases to methotrexate (Figure 12). Chapter 5 – Using webservices 89 Figure 12 Drugs involved in the reported ADR All of these drugs are classified as antineoplastic and immunomodulating agents, according to the WHO Anatomical Therapeutic Chemical (ATC) classification system [95] . Azathioprine, infliximab and adalimumab are immunosuppressants and methotrexate is an antimetabolite. 0 20 40 60 80 100 120 azathioprine infliximab adalimumab methotrexate Chapter 5 – Using webservices 90 5.5 Discussion Worldwide, many systems use electronic health records to facilitate adverse event reporting, with the same goal as our work, which is to not overload healthcare professionals with additional administrative work to report ADR [27, 55, 64] . These systems can be included in the EHR and be easily completed [64] or can actively help in form filling with the automatic input of certain information already included in the EHR [27] . During the ALIAS (Albumin in Acute Stroke Trial) experience, authors decided to integrate an electronic safety reporting module into the existing web-based system to deal with safety reporting obligations of a multicentre clinical trial [62] . This system pre-populates the reporting form with the existing information, which needs to be completed and validated by clinical staff. The ASTER pilot study also brought an important contribution to the development of these systems, as it triggered ADR within the EHR, collected patient data, populated the ADR reporting form and made the report available to the physician for review [60] . This work is different from ours, because it asks the clinician to provide additional information on the adverse event and then submits the report. All these systems incur extra work on the part of physicians, which is difficult to eliminate. The novelty of our work is the detail of allowing the physician to report ADR without any additional administrative work beyond the usual clinical registry. In addition, the EHR remained the same, without the need to change the physicians’ routine. This system adapts to the routines of the healthcare professionals and not vice versa, which can explain the increasing of ADR reporting among the studied group. It is also important to note that most of the reported ADR were serious. In fact, the studied group deals with innovative and powerful drugs that often cause serious adverse events. This finding reinforces the importance of our work, as Pharmacovigilance Systems seek mainly information about serious (and unexpected) ADR [88] . Our study has some limitations, such that it only included gastroenterologists. However, this webservice is available to be used in several software. Even so, it is currently in use only in two software systems: the one described in this study and an electronic prescription software system. Another issue is that our study does not have a control group, which limits the conclusions drawn. For this reason, the only comparison that the authors were able to perform was for the number of ADR reports made by the same group of gastroenterologists before the webservice implementation. Chapter 5 – Using webservices 91 Conclusions The simple solution described in our work allowed for reporting 167 ADR during the first 23 months of implementation, simply by clicking a button included in the usual electronic health record used by gastroenterologists. Our results suggest that doctors would report more ADR if they do not have to take on any further workload to do it. We propose that IS used to support multicentre studies should be used to report detected ADR. Author contributions The first author was responsible for data collection and writing the paper. The second author conceptualised the intervention and supervised writing of the paper. Conflict of interest The authors declare that they have no conflicts of interest. 92 Chapter 6 – Decision Support Tool 93 Chapter 6 - Decision support tool Bayesian network model to support the causality assessment of adverse drug reactions reports: a regional pharmacovigilance centre experience Inês Ribeiro-Vaz, Pedro Pereira Rodrigues, Ana Silva, Jorge Polónia WHO Winter Meeting. 2015. Chapter 7 – Ongoing and future work 100 7.2 Managing several players at several institutions dealing with the same adverse event. A pilot-study called SIRAI. Introduction In June 2014, this PhD team proposed the development of a pilot study within the local health unit of Matosinhos, Portugal (ULSM in Portuguese) to test some strategies regarding the use of IS on the information exchange about ADR. The first goal of this study was to design an information system (a single database) that will gather all the information about ADR that is disseminated around the institution, allowing for better use by healthcare professionals. Subsequently, the included ADR will be automatically submitted to the regulatory authority (National Authority of Medicines and Health Products; INFARMED, IP) through a regional pharmacovigilance centre (the Northern Pharmacovigilance Centre), with the permission of each healthcare professional. When the database is in full use, an intervention study will be performed, with the following main outcomes: (1) the number of databases integrated into the single database, (2) the number of ADR included in the single database and (3) the number of ADR reports submitted to the regulatory authority by health professionals from the hospital. To measure the submitted ADR reports, the Northern Pharmacovigilance Centre database will be used, which is registered in the Portuguese commission for data protection, and follows the confidentiality and data protection guidelines imposed by the Portuguese pharmacovigilance system. This study was approved by the local ethics committee and the hospital’s board of directors. Chapter 7 – Ongoing and future work 101 Methods In June 2014, a working group was created to design the information system. This expert group was composed of: - two physicians: one an expert in primary care and the other an expert in hospital care who is also responsible for the risk management department at the institution - two pharmacists: one an expert in hospital pharmacy and the other an expert in pharmacovigilance - two nurses - two informaticians: one responsible for the hospital informatics department and the other one from the Faculty of Medicine. From June 2014 to June 2015, this working group met in person 10 times. During these meetings, a discussion took place regarding the type of database that is most appropriate for the institution’s healthcare professionals, and its contents were approved. The screens were designed and improved between meetings, according to the group’s opinions. The content was also developed with the contribution of each participant. Namely, lists of adverse reactions, severity criteria and routes of administration were developed and approved, to facilitate form completion. To meet the needs of health professionals at the institution, we also developed a form to report incidents with medical devices. This form based on the INFARMED paper form. In February 2015, the first version of the prototype was shown, and a name was assigned to the information system: SIRAI 3 . The system is currently being tested. 3 The acronym means Sistema de Informação de Reações Adversas e Incidentes (Information System for Adverse Reactions and Incidents). Chapter 7 – Ongoing and future work 102 Workflow The project aims to promote ADR reports identified in the ULSM to the regulatory authority, through the following tasks: Task #1: Design and implementation of an information system for the management of information on drug safety. Task #2: Integration in a single database of all the ADR detected at the institution (ULSM). Task #2.1: Identification of databases that could include information about ADR detected at ULSM. Task #3: Submission of the ADR collected in the database referred in #2 to the regulatory authority through the Northern Pharmacovigilance Centre. Task #4: Feedback on ADR reports will be send to the reporters. Expected results As the main outcome, we have developed a functional information system on drug safety that is being used at the institution. The intervention impact will be measured on the number of ADR submitted to the regulatory authority. In the last four years, healthcare professionals from ULSM reported about 22-43 ADR each ear, without a regular trend. As we strongly believe that, during daily activities, many more ADR are detected by these healthcare professionals, so we expect that the number of ADR submitted will increase by about 50% each year. In addition to the number of ADR reports, an increase in relevance is also expected. Established relevance criteria include ADR seriousness, ADR expectedness and the causality degree attributed to the ADR. Chapter 7 – Ongoing and future work 103 7.3 Social networks and portals for patients and healthcare professionals Internet users are increasing and many of them use the web for issues related to health. According to Cybercitizen Health® Europe [97] , in 2012, 72% of European online adult consumers (ages over 18) were social health users (which means individuals that “have conducted any of the following activities online for health within the past 12 months: used a community, group or social networking website, or conducted any social-related activity online such as reading or posting on health blogs, message boards or health ratings websites”). Additionally, 44% of European online consumers use social networking for issues related to health, 33% read or posted patient testimonials and 34% used health ratings or reviews. To take advantage of this reality, it is important to adopt strategies to use the internet in the promotion of pharmacovigilance and ADR reporting, using for this purpose social networks and portals for patients and healthcare professionals. Social networks are excellent media and can be used to disseminate information about pharmacovigilance and drug safety issues. In 2011, Knezevic et al. studied the use of social networks, such as Facebook®, to increase spontaneous ADR reporting. This study tested if the creation of a group in Facebook® increased ADR detection and reporting by its members [65] . For this purpose, an open group (available to the general public) was created where regular information about ADR was posted. During the experience period (seven months), 21 ADR were reported, by 2% of the total group members (n=1034). Among the 1034 members, 370 provided their educational profile (88% had a university degree and 12% a high school degree). For those with a university degree, 67% had a degree in medicine, dentistry or pharmacy. None of the 21 ADR reported were serious or unexpected. Based on their results, the authors stated that Facebook® can be useful for improving spontaneous ADR reporting. In line with this, UFN created its own Facebook® page in October 2013, and currently stands at 876 “likes” (in april 2016). On this public page, we publish information about drug safety and also about our events (open classes and post-graduate courses, among others). We also publish images and messages promoting spontaneous ADR reporting and emphasising the importance of this subject. In the future, we expect that this will be a source of debate about ADR reports, with the creation of an online forum and maybe part of a study similar to the one performed by Knezevic et al. [65] . Chapter 7 – Ongoing and future work 104 7.3.1 Creation of the personal area on the UFN website Websites of pharmacovigilance centres and regulatory authorities are important for providing information about drug safety and to promote ADR reporting, both by patients and healthcare professionals. Some of these websites also provide online ADR reporting forms. In these cases, it is essential that the provided form is simple, intuitive, quick to fill in and with few mandatory fields. Otherwise, the experience of reporting may seem too complicated for the users and discourage them [43] . Thus, we aimed to create a model of communication between our pharmacovigilance centre and its reporters (health professionals) through our website. With this model, we wanted to turn ADR reporting into an informative and motivating activity. To develop this project, the first step was to redesign the former Northern Pharmacovigilance Centre website, in order to make it more appealing and user-friendly (Figure 14 and Figure 15): Figure 14. UFN website 2005-2016 Figure 15. UFN website 2016 The new website was launched on 18 January 2016. On the new website, we created a reporter area dedicated to healthcare professionals. This area compiles all the ADR reports submitted online by each healthcare professionals, and are only available after authentication, to ensure the confidentiality. Our aim was to provide our users also with feedback after submitting their ADR report. Each healthcare professional will have access to a report that compiles the research results generated by the ADR report. The search will be conducted using selected information sources, each time a healthcare professional reports an ADR through the ADR online form. The results of the searches will be presented in the form of ADR frequency, ADR absolute value, ADR expectedness and percentage of ADR by age and gender. Chapter 7 – Ongoing and future work 105 We selected the following information sources to perform the searches: • The European database of suspected ADR 4 , which provides information on absolute number and percentage of ADR by sex, age and other criteria. • The WHO Global Database of Individual Case Safety Reports (Vigilyze®) 5 , which provides information on the number of cases reported, describing each case individually. • Medscape® 6 , which provides information on the frequency (%) of ADR. • The Medicines & Healthcare Product Regulatory Agency 7 (MHRA), which provides information on the number of cases reported. We are still working, together with software development staff, to conclude the creation of the personal area on our website. 4 www.adrreports.eu 5 www.vigilyze.who-umc.org 6 http://reference.medscape.com/medscapetoday 7 https://www.gov.uk/government/organisations/medicines-and-healthcare-products-regulatory-agency Chapter 7 – Ongoing and future work 106 7.3.2 ADR online reporting form for consumers In Portugal, consumers have been allowed to directly report their own suspected ADR since July 2012 [76] . To facilitate the ADR reporting process among consumers, we have started to develop an ADR report online form for consumers in Portugal, taking advantage of our website redesign. To develop the consumer’s online form, we used the printable form as a standard. The form was carefully developed and adjusted for patient reporting. A heuristic evaluation of the form was performed to obtain the final version, which is already finished (see Figure 16, designed by Natacha Oliveira) and available on the Northern Pharmacovigilance Centre website [5] . Figure 16. Screen shot of the online patients reporting form (ADR screen). Chapter 7 – Ongoing and future work 107 7.4 App for ADR reporting During this study, the idea arose of developing a smartphone application for ADR reporting. For this purpose, the user interface has been designed (as seen in Figure 17 and Figure 18, both designed by Natacha Oliveira), consistent with the new image of the UFN website. This project is in the implementation phase. Figure 17. User interface of the ADR reporting app (first screen). Figure 18. User interface of the ADR reporting app (suspected drug screen). 108 Chapter 8 – Discussion 109 Chapter 8 – General discussion These studies have allowed us to realise that much work can be done using informatics to improve pharmacovigilance. Worldwide underreporting of ADR is a major concern, and many institutions are aware that it is feasible to use IS to improve ADR reporting. Currently, the most commonly used technological platform is web-based and shows an increasing trend, but interventions within electronic health records also have the potential to improve pharmacovigilance activities, particularly ADR reporting. ADR directly reported by consumers (direct ADR reporting) is being increasingly taken into account when the aim is to improve information on drug safety [98] . Most international studies that aimed to promote ADR reporting were developed in hospitals and tailored to healthcare professionals. In fact, most serious ADR are detected in hospitals and reported by healthcare professionals [72] . Additionally, in Europe, direct reporting (ADR reported by patients) has only been allowed in all countries since 2012 [76] . This may also explain why most of the projects that target direct ADR reporting have only been recently developed [65, 67, 71, 72] . As with most of international groups, we developed systems for the passive promotion of ADR reporting because most busy healthcare professionals only submit their suspected ADR if it does not increase their workload [55, 77] . Active promotion of ADR reporting is difficult and not always ethically acceptable because no material reward can be given to the reporters. Thus, projects have aimed to actively promote ADR reporting involving teaching sessions [50] or emails containing ADR information [46, 57, 59, 72] . Based on our research, we believe that it would be useful to adopt systems to assist healthcare professionals with completing ADR reporting within electronic health records because this approach seems to be an efficient method to increase the ADR reporting rate [75] . When this approach is not possible, it is essential to have tools that are easily accessible to healthcare professionals (e.g. on the web) to report ADR. These tools can be promoted by sending emails or through the inclusion of direct hyperlinks on healthcare professionals’ desktops, which is a simple and cost-effective way to change the behaviour of healthcare professionals regarding spontaneous ADR reporting, as seen in Chapters 3 and 4, and can be easily implemented and disseminated in healthcare institutions [16] . References 116 time series analysis in Spain. Drug safety : an international journal of medical toxicology and drug experience. 2009;32(1):77-83. 94. Christensen E, Curbera F, Meredith G, Weerawarana S. Web services description languade (WSDL) 1.1 2001. Available from: http://www.w3.org/TR/wsdl. 95. World Health Organization. WHO Collaborating Centre for Drug Statistics Methodology 2014 [cited 2015 03-2015]. Available from: http://www.whocc.no/. 96. Agbabiaka TB, Savovic J, Ernst E. Methods for causality assessment of adverse drug reactions - A systematic review. Drug Safety. 2008;31(1):21-37. 97. manhattanRESEARCH. manhattanRESEARCH 2011 [cited 2014]. Available from: http://manhattanresearch.com/. 98. Ribeiro-Vaz I, Silva AM, Costa Santos C, Cruz-Correia R. How to promote adverse drug reaction reports using information systems - a systematic review and meta-analysis. BMC medical informatics and decision making. 2016;16(1):27. 117 Annex 118 6 Annex This annex contains reprints of all the published papers included in this thesis, as follows: 1. Inês Ribeiro-Vaz, Fabrício Alves Barbosa Silva, Ana-Marta Matos Silva, Domingos Alves, Cruz-Correia R. Pharmacovigilance informatics. In: Maria Manuela Cruz-Cunha, Isabel Maria Miranda, Ricardo Martinho, Rijo R, editors. Encyclopedia of E-health and telemedicine: IGI Global; 2016. p. 299-315. 2. Ribeiro-Vaz I, Silva AM, Costa Santos C, Cruz-Correia R. How to promote adverse drug reaction reports using information systems - a systematic review and meta-analysis. BMC Medical Informatics and Decision Making. 2016;16(1):27. 3. Ribeiro-Vaz I, Santos CC, Cruz-Correia R. Promoting adverse drug reaction reporting: comparison of different approaches. Rev Saude Publica. 2016;50:14. 4. Ribeiro-Vaz I, Santos C, Da Costa-Pereira A, Cruz-Correia R. Promoting spontaneous adverse drug reaction reporting in hospitals using a hyperlink to the online reporting form: An ecological study in Portugal. Drug Safety. 2012;35(5):387-94. 5. Ribeiro-Vaz I, Correia RJC (2016) Using Web Services to Link Electronic Clinical Records to Pharmacovigilance Databases. Int J Pharmacovigil 1(1): 4. This annex also contains the approval document of the local Ethics Committee of the Faculty of Medicine of the University of Porto (Process PCEDCSS-FMUP 08/2014, approved in May 7, 2014). Encyclopedia of E-Health and Telemedicine Maria Manuela Cruz-Cunha Polytechnic Institute of Cávado and Ave, Portugal & Algoritmi Research Centre, Portugal Isabel Maria Miranda Câmara Municipal de Guimarães, Portugal Ricardo Martinho Polytechnic Institute of Leiria, Portugal & CINTESIS - Center for Research in Health Technologies and Information Systems, Portugal Rui Rijo Polytechnic Institute of Leiria, Portugal & INESCC - Institute for Systems and Computers Engineering at Coimbra, Portugal & CINTESIS - Center for Research in Health Technologies and Information Systems, Portugal Published in the United States of America by Medical Information Science Reference (an imprint of IGI Global) 701 E. Chocolate Avenue Hershey PA, USA 17033 Tel: 717-533-8845 Fax: 717-533-8661 E-mail: [email protected] Web site: http://www.igi-global.com Copyright © 2016 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher. Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark. Library of Congress Cataloging-in-Publication Data British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library. All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the authors, but not necessarily of the publisher. For electronic access to this publication, please contact: [email protected]. Names: Cruz-Cunha, Maria Manuela, 1964editor. | Miranda, Isabel Maria, 1954editor. | Martinho, Ricardo, 1974editor. | Rijo, Rui, editor. Title: Encyclopedia of E-health and telemedicine / Maria Manuela Cruz-Cunha, Isabel Maria Miranda, Ricardo Martinho, and Rui Rijo, editors. Description: Hershey, PA : Medical Information Science Reference, 2016. | Includes bibliographical references and index. Identifiers: LCCN 2015051069| ISBN 9781466699786 (hardcover) | ISBN 9781466699793 (ebook) Subjects: LCSH: Medical care--Technological innovations--Encyclopedias. | Medical informatics--Encyclopedias. Classification: LCC R858 .E518 2016 | DDC 610.28503--dc23 LC record available at http://lccn.loc.gov/2015051069 D 299 Copyright © 2016, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Category: Developments and Solutions DOI: 10.4018/978-1-4666-9978-6.ch025 INTRODUCTION TO PHARMACOVIGILANCE INFORMATICS All medicines have adverse effects, most of them unknown until the drug commercialization. As so, it is crucial to keep strategies to monitor the drug safety. Pharmacovigilance is the activity of drug surveillance, after its launch in the market, with the main goal of public health protection, ensuring that the drug benefit outweighs its risks. Worldwide, pharmacovigilance systems are mostly based on spontaneous Adverse Drug Reactions (ADR) reports made by healthcare professionals and consumers. Spontaneous ADR reporting has been described as an essential method to detect drug safety signals; however, underreporting is a major issue undermining the effectiveness of spontaneous reports. Several studies suggest that less than 10% of detected ADR are effectively reported to medicines regulatory authorities [e.g. Food and Drug Administration (FDA), European Medicines Agency (EMA), etc] (Hazell & Shakir, 2006; McGettigan, Golden, Conroy, Arthur, & Feely, 1997). Tools used in pharmacovigilance are continually evolving and, worldwide, Information Systems (IS) to promote ADR reporting or to detect ADR occurred in healthcare institutions have been tested and used, such as software that allow voluntary and automated detection of ADR, tools that analyse clinical databases or Web sites that actively inform healthcare professionals (Molokhia, Tanna, & Bell, 2009). In addition to the signal detection, ICT can also be used to encourage and facilitate reporting of suspected ADR, such as the creation of on-line reporting forms, development of tools to collect safety data from electronic health records (EHR), among others. In this chapter, it will be described some tools to automatically detect ADR, or encourage ADR spontaneous report. BACKGROUND ON PHARMACOVIGILANCE Adverse Drug Reactions (ADR) defined as a response to a medicinal product which is noxious and unintended (WHO) are a well-recognized public health problem worldwide, and a major cause of death Pharmacovigilance Informatics Inês Ribeiro-Vaz Northern Pharmacovigilance Centre, Portugal Fabrício Alves Barbosa Silva Oswaldo Cruz Foundation, Brazil Ana-Marta Matos Silva Northern Pharmacovigilance Centre, Portugal Domingos Alves Faculty of Medicine of Ribeirão Preto (USP), Brazil Ricardo Cruz-Correia Centre for Health Technology and Services Research, Portugal Pharmacovigilance Informatics 300 and hospitalization in developed countries(Lazarou, Pomeranz, & Corey, 1998). It is estimated that about 6,5% of the hospitalizations are related to ADR(Pirmohamed et al., 2004). Besides, in the USA, about 100.000 people die each year due to ADR(Lazarou et al., 1998), and in Europe this annual mortality rate increases to 197.000(European Medicines Agency, 2014). ADR can be expressed in many ways and with different degrees of seriousness. An anaphylactic shock caused by penicillin is an example of a serious ADR (a serious ADR is any untoward medical occurrence that at any dose: results in death, requires inpatient hospitalisation or prolongation of existing hospitalisation, results in persistent or significant disability/incapacity or is life-threatening(WHO). Another type of ADR, not always recognized as such, is the drug ineffectiveness, for example, a vaccination failure. This can be (or not) related with a product quality issue and should be reported when detected in order to allow the regulatory authorities to take appropriate decisions. Rare and long term ADR are difficult to detect during the drug development stage. Only when the drug begins to be used by a large population after marketing authorization it is possible to detect new ADR not previously identified during clinical trials. In reality, it is known that the safety of a new drug cannot be established until it has been on the market for several years(Lasser et al., 2002). Exceptionally, in a pandemic scenario, drug launch is urgent and, in this particular case, can be justifiable that drug safety profile is not well established. In this scenarios, it is even more important that all the detected ADR are reported (serious or not, expected or not). It is, therefore, essential to keep drugs under close surveillance, after its commercialization, through a pharmacovigilance system, to continuously evaluate their safety profile. In most of the European countries, pharmacovigilance system is based on spontaneous ADR reports, which is passive method, made by healthcare professionals and, since July 2012, also by consumers (Ministério da Saúde, 2006). These reports can be made using paper, telephone, e-mail or through an on-line form and consist of a description of an Adverse Event (AE) apparently caused by a medicine. To reverse the problem of underreporting of ADR, which is felt in most developed countries, several strategies have been tested(M. T. Herdeiro et al., 2012; McGettigan et al., 1997; Ribeiro-Vaz, Santos, da Costa-Pereira, & Cruz-Correia, 2012). Particularly, some studies were developed focused on educational interventions to raise awareness on the importance of ADR reporting(Figueiras A., Herdeiro T, Polonia J, & JJ, 2006; M. T. Herdeiro, Polonia, Gestal-Otero, & Figueiras, 2008; Ribeiro-Vaz, Herdeiro, Polonia, & Figueiras, 2011) and showed to be very effective increasing the quantity and relevance of spontaneous ADR reports (among health professionals). However, these studies involved a large financial and personal outlay and the authors concluded that the effect was lost after a few months(McGettigan et al., 1997; Nazario, Feliu, & Rivera, 1994). In a recent American study, the authors developed a signal-detection strategy that combines the Adverse Event Reporting System (AERS) of the regulatory Authority (FDA) and EHR, by requiring signaling in both sources, with promising results(Harpaz et al., 2013). Another study used the unstructured clinical notes included in EHR to detect ADR through a computerized system. The authors concluded that data mining can be used for hypothesis generation and for rapid analysis of suspected AE risk(LePendu et al., 2013). With a similar aim, a recent study used a physician’s network, created through a mailing list, to send regular emails to doctors, with humorous component attached with informative component, recalling the importance of reporting their suspected ADR(Goldstein, Berlin, Saliba, Elias, & Berkovitch, 2013). The results showed that this type of intervention has impact on the number of ADR reports made by these professionals (the study did not assess the relevance of reported ADR). In France, it is being done a work that tries to facilitate the act to ADR report, habitually considered a tedious process by health D Category: Developments and Solutions 301 professionals. The authors are using the information contained in EHR to make the semi-automatic filling of ADR notification forms. The objectives of this ongoing study is to increase the rate of ADR report, as well as improving the quality of information submitted to regulatory authorities(Pares, Declerck, Hussain, Ng, & Jaulent, 2013). INFORMATION SYSTEMS IN PHARMACOVIGILANCE Issues, Controversies, and Problems Although some authors consider that ADR spontaneous reports suffers from latency and inconsistency, it is still considered as the most valuable method to early detect drugs safety problems. In fact, most of the decisions concerning to drug safety are triggered by daily ADR spontaneous reports. As so, Regulatory Authorities consider as crucial importance to achieve the greatest number of ADR reports possible and with high data quality. The promotion of ADR report among healthcare professionals is a huge task, as it is necessary to regularly recall the importance of ADR reporting and, simultaneously, develop tools to facilitate this duty. Pharmacovigilance centres worldwide develop several strategies to continuously promote the importance of ADR reporting, as workshops, post-graduate courses, and also to make it easier, as development of online reporting forms, inclusion of electronic reporting systems into the Hospital Information Systems (HIS), direct hyperlinks to online reporting forms, among others. Along with the promotion of spontaneous ADR reporting, some systems are being tested to detect signals of adverse reactions in large databases, as hospitals databases, epidemiologic databases, or even among social networks. Although the adoption of different strategies, it is consensual the need to obtain the highest quantity and quality of information about the safety of marketed drugs, for the protection of public health. Another important issue in this field is the need to discuss/harmonize what should be recorded during an ADR report. When analyzed the databases from the FDA, EMA and the Portuguese Northern Pharmacovigilance Centre (UFN), all of them with the same purpose of recording ADR reports, it was possible to realize that the variables are different between the 3 databases, as seen in the Table 1. For example, the concept of outcome for the FDA form as a similar meaning to the concept of seriousness for the two European databases analyzed. On the other hand, for these two databases, outcome means the evolution of the patient regarding the adverse scenario. Active debate is maintained about this subject. The authors participate in some discussion groups, both national and internationally [namely the FHIR/OpenEHR(OpenEHR, 2014) and HL7(Health Level Seven, 2014) working groups], aiming to create a standard to the information that should be kept during the ADR registry. Authors also participate, in June 2014, in the HL7 pharmacy group meeting, where was discussed the International Organization for Standardization (ISO) draft of international standard for ‘Requirements for electronic prescriptions’. This document is still under discussion until February 2015. The last OpenEHR discussion took place in July 2014 and was focused in the ADR definition, its characteristics and what should be kept for the registration of a suspected ADR. At the moment, it is in discussion the clinical content of the Archetype Adverse Reaction (FHIR/OpenEHR). Variables as substance, status, seriousness, reaction type, certainty are under debate. Pharmacovigilance Informatics 302 Table 1. Comparison of the variables included in 3 pharmacovigilance databases (Food and Drug Administration: FDA; European Medicines Agency: EMA and Portuguese Northern Pharmacovigilance Centre: UFN), grouped in main entities/sections (patients, problems, products and reporters) Section Variable FDA EMA UFN Patient Patient identifier ✓ ✓ ✓ Age at the time of event ✓ ✓ ✓ Date of birth ✓ ✓ ✓ Sex ✓ ✓ ✓ Weight ✓ ✓ ✓ Problem What kind of AE, product problem or error did you encounter? • Adverse event • Product use error • Product problem • Problem with Different Manufacturer of the Same Medicine --- --- Outcomes Attributed to AE • Death • Life-threatening • Congenital anomaly/birth defect • Disability or permanent damage • Hospitalization • Required intervention to prevent permanent impairment/damage • Other serious • Fatal • Not recovered • Recovered with sequelae • Recovering • Recovered • Unknown • Fatal • Not recovered • Recovered with sequelae • Recovering • Recovered • Unknown ADR Seriousness --- • Yes • No • Death • Life-threatening • Congenital anomaly • Results in persistent or significant incapacity • Hospitalization • Other serious Date of event ✓ ✓ ✓ Describe events, problem, or product use error ✓ ✓ ✓ Relevant test/laboratory data, including dates ✓ --- --- Product Product name ✓ ✓ ✓ Label strength ✓ ✓ ✓ Manufacturer ✓ ✓ ✓ Date of use ✓ ✓ ✓ Reason for use ✓ ✓ ✓ Problem went away after use stopped or dose reduced? • Yes • No • Does not apply --- ✓ Problem returned after person started taking or using the product again? • Yes • No • Does not apply --- ✓ Do you still have the product in case we need to evaluate it? • Yes • No • Returned to Manufacturer on --- --- Reporter Reporter name ✓ --- ✓ Address ✓ ✓ ✓ ✓or list of options: Variable exists; ---: Variable does not exist D Category: Developments and Solutions 303 PROPOSED SOLUTIONS AND RECOMMENDATIONS Promotion of Adverse Drug Reaction Reporting Integration with IS (Patient Records, Multi-Center Research Projects) Integration of pharmacovigilance system databases with other healthcare IS seems to be an obvious way to improve the knowledge about drugs safety. In fact, healthcare providers insert a lot of information in their EHR about ADR, which is not shared with pharmacovigilance systems. Every approach that promotes drug safety surveillance without increasing the workload of healthcare professionals, one of the main reasons for not reporting ADR, should be considered. The creation and implementation of webservices to collect this information is one of the possible solutions for this problem. This solution is being tested in a Portuguese multi-center research project, in the field of gastric diseases(Study Group of Inflammatory Bowel Disease, 2014), with promising preliminary results (5% of all of the ADR reports received in the 1st semester of 2014 were collected by webservice). In this case, the physicians insert the usual clinic information during the appointment with the patient, and then, they only have to authorize the transmission of anonymized information about drug-related problems to the pharmacovigilance system. This tool allows ADR reporting without the need to fill the ADR reporting form and with no additional administrative work for the physician other than the normal registry of clinic patient data. This solution is also available to be included in commercial prescription software. For the implementation of these webservices, it was necessary to map the form entries used by the doctors with the online ADR reporting form developed by the pharmacovigilance system, so that the information is correctly collected. This mapping would be easier if a standard as the one described in the previous section was already in use. An ADR electronic reporting system included into a HIS was developed in a Spanish hospital(Ortega et al., 2008), allowing healthcare professionals (pharmacists, physician and nurses) to report suspected ADR through their usual IS. The biggest advantage of this system is that some data (already included in the IS) appears as default values into the form, which expedites the system and reduces transcription errors(Cruz-Correia et al., 2009). All the reports made by this system are reviewed by a pharmacist, which is responsible to confirm the included data and to report the case to pharmacovigilance system. This might be a disadvantage, as some cases may be lost and not actually sent to regulatory authorities. Another strategy that can be easily adopted is the inclusion of hyperlinks in the EHR to the online ADR reporting form. This solution was tested between 2006 and 2010, in an ecological study performed in 16 Portuguese hospitals centres(Ribeiro-Vaz et al., 2012). The hyperlinks were included in either EHR or on computer desktops. Considering the hospitals with hyperlink included in the EHR, the median ADR reports per month significantly increased, from two (range 0–12) to five reports (range 1–17). The median of ADR reports per month using the online form also increased significantly, from one (range 0–5) before the intervention to four (range 1–17) after it. Moreover, serious ADR increased 3-fold, and non-previously described ADR increased 4.5-fold. None of these significant increases were observed in the hospitals where the hyperlink was not installed. It was also found a significant increase in daily pharmacovigilance centre website visits, from ten before the intervention to 27 after it (p<0.001). The increase in ADR reporting shows that the inclusion of hyperlinks to online ADR reporting forms is an easy way to change health professional behavior with regard to spontaneous ADR reports. Furthermore, this solution seems to be cost effective, when compared with other strategies to increase ADR report, Pharmacovigilance Informatics 310 • Use websites and social networks to disseminate information on drug safety and promote ADR reporting. • Create data mining processes to search ADR on websites, social networks and EHR. FUTURE RESEARCH DIRECTIONS The authors believe that much work can be done to use informatics in improving pharmacovigilance. As future work, pharmacovigilance websites can be explored to create a personal area for each person that submits an ADR, aiming to turn the ADR reporting act in a motivating and informative activity. It should provide technical information about the reported adverse reaction and the suspected drug, during the ADR report act, and give feedback. Another research plan is to use a pharmacovigilance facebook page to encourage ADR report by its members, conducting research work similar to those performed by Knezevic et al. It should also be carefully developed ADR reporting forms adjusted to patients report. These forms should be simple and easy to fulfill, with accessible language (not too technical). Authors see as a promising solution the integration in a single database all the suspected ADR detected within a healthcare institution. These ADR will then be sent to the Authority through regional pharmacovigilance centres. These centres will provide feedback related to the reported ADR. An emerging research topic in pharmacovigilance is that the combination of information from multiple data sources may lead to more effective and accurate discovery of ADR. Depending on the data sources used, and how they are combined, it is believed that the resulting system could raise the statistical significance of results or facilitate new discoveries that are not possible using a single data source. It should be establish a database as a reference (gold) standard for adverse reactions, from the systematic exploration of heterogeneous data sources and more comprehensive than is recommended by Harpaz and colleagues. CONCLUSION Pharmacovigilance should use informatics to promote and complement spontaneous ADR report. Integration of pharmacovigilance databases with other healthcare IS seems to be an easy way to promote ADR reports among healthcare professionals. In addition, the trend of use of social networks has been exploited by pharmacovigilance with mechanism to detect adverse reactions among these networks. ACKNOWLEDGMENT This work was partially funded by project “NORTE-01-0145-FEDER-000016” that is financed by the North Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, and through the European Regional Development Fund (ERDF). D Category: Developments and Solutions 311 REFERENCES Abookire, S. A., Teich, J. M., Sandige, H., Paterno, M. D., Martin, M. T., Kuperman, G. J., & Bates, D. W. (2000). Improving allergy alerting in a computerized physician order entry system. Proc AMIA Symp, 2–6. PMID:11080034 Bate, A., Lindquist, M., Edwards, I. R., Olsson, S., Orre, R., Lansner, A., & De Freitas, R. M. (1998). A Bayesian neural network method for adverse drug reaction signal generation. European Journal of Clinical Pharmacology, 54(4), 315–321. doi:10.1007/s002280050466 PMID:9696956 Bodenreider, O. (2004). The unified medical language system (UMLS): Integrating biomedical terminology. Nucleic Acids Research, 32(90001suppl 1), D267–D270. doi:10.1093/nar/gkh061 PMID:14681409 Butler, D. (2013). When Google got flu wrong. Nature, 494(7436), 155–156. doi:10.1038/494155a PMID:23407515 Chan, E. H., Sahai, V., Conrad, C., & Brownstein, J. S. (2011). Using web search query data to monitor dengue epidemics: A new model for neglected tropical disease surveillance. PLoS Neglected Tropical Diseases, 5(5), e1206. doi:10.1371/journal.pntd.0001206 PMID:21647308 Classen, D. C., Pestotnik, S. L., Evans, R. S., & Burke, J. P. (1992). Description of a computerized adverse drug event monitor using a hospital information system. Hospital Pharmacy, 27(9), 774, 776–779, 783. PMID:10121426 Cruz-Correia, R., Rodrigues, P. P., Freitas, A., Almeida, F. C., Chen, R., & Costa-Pereira, A. (2009). Data Quality and Integration Issues in Electronic Health Records. In V. Hristidis (Ed.), Information Discovery on Electronic Health Records (pp. 55–96). doi:10.1201/9781420090413-c4 DuMouchel, W. (1999). Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system. The American Statistician, 53(3), 177–190. Duval, F., Caffarena, E. R., Cruz, O. G., & Silva, F. A. B. (2014). Mining for adverse drug events on twitter. Paper presented at the 6th International Conference on Knowledge Discovery, and Information Retrieval (KDIR), Roma, Italy. doi:10.5220/0005135203540359 Estatuto do Medicamento, DL nº 176/2006 C.F.R. (2006). European Medicines Agency. (2014). European Medicines Agency. Retrieved 05-09-2014, 2014, from www.ema.europa.eu Evans, S. J., Waller, P. C., & Davis, S. (2001). Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports. Pharmacoepidemiology and Drug Safety, 10(6), 483–486. doi:10.1002/pds.677 PMID:11828828 Figueiras, A., Herdeiro, T., Polonia, J., & Gestal-Otero, J. J. (2006). An educational intervention to improve physician reporting of adverse drug reactions: A cluster-randomized controlled trial. Journal of the American Medical Association, 296(9), 1086–1093. doi:10.1001/jama.296.9.1086 PMID:16954488 Ginsberg, J., Mohebbi, M. H., Patel, R. S., Brammer, L., Smolinski, M. S., & Brilliant, L. (2008). Detecting influenza epidemics using search engine query data. Nature, 457(7232), 1012–1014. doi:10.1038/ nature07634 PMID:19020500 Pharmacovigilance Informatics 312 Goldstein, L. H., Berlin, M., Saliba, W., Elias, M., & Berkovitch, M. (2013). Founding an adverse drug reaction (ADR) network: A method for improving doctors spontaneous ADR reporting in a general hospital. Journal of Clinical Pharmacology, n/a. doi:10.1002/jcph.149 PMID:23852627 Harpaz, R., DuMouchel, W., Shah, N. H., Madigan, D., Ryan, P., & Friedman, C. (2012). Novel datamining methodologies for adverse drug event discovery and analysis. Clinical Pharmacology and Therapeutics, 91(6), 1010–1021. doi:10.1038/clpt.2012.50 PMID:22549283 Harpaz, R., Vilar, S., Dumouchel, W., Salmasian, H., Haerian, K., Shah, N. H.,... Friedman, C. (2013). Combing signals from spontaneous reports and electronic health records for detection of adverse drug reactions. J Am Med Inform Assoc, 20(3), 413-419. doi: 10.1136/amiajnl-2012-000930 Hauben, M. (2004). Trimethoprim-induced hyperkalaemia -- lessons in data mining. British Journal of Clinical Pharmacology, 58(3), 338–339. doi:10.1111/j.1365-2125.2004.02153.x PMID:15327598 Hauben, M., & Reich, L. (2004). Safety related drug-labelling changes: Findings from two data mining algorithms. Drug Safety, 27(10), 735–744. doi:10.2165/00002018-200427100-00004 PMID:15350157 Hauben, M., & Zhou, X. (2003). Quantitative methods in pharmacovigilance: Focus on signal detection. Drug Safety, 26(3), 159–186. doi:10.2165/00002018-200326030-00003 PMID:12580646 Hay, S. I., George, D. B., Moyes, C. L., & Brownstein, J. S. (2013). Big data opportunities for global infectious disease surveillance. PLoS Medicine, 10(4), e1001413. doi:10.1371/journal.pmed.1001413 PMID:23565065 Hazell, L., & Shakir, S. A. (2006). Under-reporting of adverse drug reactions: A systematic review. Drug Safety, 29(5), 385–396. doi:10.2165/00002018-200629050-00003 PMID:16689555 Health Level Seven. (2014). Retrieved from www.hl7.org Herdeiro, M. T., Polonia, J., Gestal-Otero, J. J., & Figueiras, A. (2008). Improving the reporting of adverse drug reactions: A cluster-randomized trial among pharmacists in portugal. Drug Safety, 31(4), 335–344. doi:10.2165/00002018-200831040-00007 PMID:18366244 Herdeiro, M. T., Ribeiro-Vaz, I., Ferreira, M., Polonia, J., Falcao, A., & Figueiras, A. (2012). Workshopand telephone-based interventions to improve adverse drug reaction reporting: a cluster-randomized trial in Portugal. Drug Saf, 35(8), 655-665. doi: 10.2165/11599750-000000000-00000 Hey & Trefethen. (2003). The data deluge: An e-science perspective. Grid Computing: Making the Global Infrastructure a Reality, 809-824. Huang, L., Guo, T., Zalkikar, J. N., & Tiwari, R. C. (2014). A Review of Statistical Methods for Safety Surveillance. Therapeutic Innovation & Regulatory Science, 48(1), 98–108. doi:10.1177/2168479013514236 Kilbridge, P. M., & Classen, D. C. (2008). The informatics opportunities at the intersection of patient safety and clinical informatics. Journal of the American Medical Informatics Association, 15(4), 397–407. doi:10.1197/jamia.M2735 PMID:18436896 Kilbridge, P. M., Noirot, L. A., Reichley, R. M., Berchelmann, K. M., Schneider, C., Heard, K. M., & Bailey, T. C. etal. (2009). Computerized surveillance for adverse drug events in a pediatric hospital. Journal of the American Medical Informatics Association, 16(5), 607–612. doi:10.1197/jamia.M3167 PMID:19567791 D Category: Developments and Solutions 313 Lampos, V., & Cristianini, N. (2012). Nowcasting events from the social web with statistical learning. ACM Transactions on Intelligent Systems and Technology, 3(4), 72. doi:10.1145/2337542.2337557 Lasser, K., Allen, P., Woolhandler, S., Himmlstein, F., Wolfe, S., & Bor, D. (2002). Timing of New Black Box Warnings and Withdrawals for Prescription Medications. Journal of the American Medical Association, 287(17), 2215–2220. doi:10.1001/jama.287.17.2215 PMID:11980521 Lazarou, J., Pomeranz, B. H., & Corey, P. N. (1998). Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA, 279(15), 1200-1205. LePendu, P., Iyer, S. V., Bauer-Mehren, A., Harpaz, R., Mortensen, J. M., Podchiyska, T.,... Shah, N. H. (2013). Pharmacovigilance using clinical notes. Clin Pharmacol Ther, 93(6), 547-555. doi: 10.1038/ clpt.2013.47 Levy, M., Azaz-Livshits, T., Sadan, B., Shalit, M., Geisslinger, G., & Brune, K. (1999). Computerized surveillance of adverse drug reactions in hospital: Implementation. European Journal of Clinical Pharmacology, 54(11), 887–892. doi:10.1007/s002280050571 PMID:10027665 McGettigan, P., Golden, J., Conroy, R. M., Arthur, N., & Feely, J. (1997). Reporting of adverse drug reactions by hospital doctors and the response to intervention. British Journal of Clinical Pharmacology, 44(1), 98–100. doi:10.1046/j.1365-2125.1997.00616.x PMID:9241104 Molokhia, M., Tanna, S., & Bell, D. (2009). Improving reporting of adverse drug reactions: Systematic review. Clin Epidemiol, 1, 75–92. doi:10.2147/CLEP.S4775 PMID:20865089 Nazario, M., Feliu, J. F., & Rivera, G. C. (1994). Adverse drug reactions: The San Juan Department of Veterans Affairs Medical Center experience. Hospital Pharmacy, 29(3), 244–246, 249–250. PMID:10171808 OpenEHR. (2014). Retrieved 09-2014, from www.openehr.org Ortega, A., Aguinagalde, A., Lacasa, C., Aquerreta, I., Fernandez-Benitez, M., & Fernandez, L. M. (2008). Efficacy of an adverse drug reaction electronic reporting system integrated into a hospital information system. The Annals of Pharmacotherapy, 42(10), 1491–1496. doi:10.1345/aph.1L130 PMID:18780808 Osheroff, J. A., Teich, J. M., Middleton, B., Steen, E. B., Wright, A., & Detmer, D. E. (2007). A roadmap for national action on clinical decision support. Journal of the American Medical Informatics Association, 14(2), 141–145. doi:10.1197/jamia.M2334 PMID:17213487 Otero, M.J., & Domínguez-Gil, A. (2000). Acontecimientos Adversos por Medicamentos: Una Patología Emergente. Farmacia hospitalaria: órgano oficial de expresión científica de la Sociedad Española de Farmacia Hospitalaria, 24(4), 258-266. Pares, Y., Declerck, G., Hussain, S., Ng, R., & Jaulent, M. C. (2013). Building a time-saving and adaptable tool to report adverse drug events. Studies in Health Technology and Informatics, 192, 903–907. PMID:23920689 Phansalkar, S., Desai, A., Choksi, A., Yoshida, E., Doole, J., Czochanski, M., & Bates, D. W. etal. (2013). Criteria for assessing high-priority drug-drug interactions for clinical decision support in electronic health records. BMC Medical Informatics and Decision Making, 13(1), 65. doi:10.1186/1472-6947-1365 PMID:23763856 Pharmacovigilance Informatics 314 Pirmohamed, M., James, S., Meakin, S., Green, C., Scott, A. K., Walley, T. J.,... Breckenridge, A. M. (2004). Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ, 329(7456), 15-19. doi: 10.1136/bmj.329.7456.15 Reshef, Reshef, Finucane, Grossman, McVean, Turnbaugh,... Sabeti. (2011). Detecting novel associations in large data sets. Science, 334(6062), 1518-1524. Ribeiro-Vaz, I., Herdeiro, M. T., Polonia, J., & Figueiras, A. (2011). Strategies to increase the sensitivity of pharmacovigilance in Portugal. Rev Saude Publica, 45(1), 129-135. Ribeiro-Vaz, I., Santos, C., da Costa-Pereira, A., & Cruz-Correia, R. (2012). Promoting spontaneous adverse drug reaction reporting in hospitals using a hyperlink to the online reporting form: An ecological study in Portugal. Drug Safety, 35(5), 387–394. doi:10.2165/11597190-000000000-00000 PMID:22468615 Rommers, M. K., Teepe-Twiss, I. M., & Guchelaar, H. J. (2007). Preventing adverse drug events in hospital practice: An overview. Pharmacoepidemiology and Drug Safety, 16(10), 1129–1135. doi:10.1002/ pds.1440 PMID:17610221 Rommers, M. K., Teepe-Twiss, I. M., & Guchelaar, H. J. (2011). A computerized adverse drug event alerting system using clinical rules: A retrospective and prospective comparison with conventional medication surveillance in the Netherlands. Drug Safety, 34(3), 233–242. doi:10.2165/11536500000000000-00000 PMID:21332247 Rommers, M. K., Zwaveling, J., Guchelaar, H. J., & Teepe-Twiss, I. M. (2013). Evaluation of rule effectiveness and positive predictive value of clinical rules in a Dutch clinical decision support system in daily hospital pharmacy practice. Artificial Intelligence in Medicine, 59(1), 15–21. doi:10.1016/j. artmed.2013.04.001 PMID:23664455 Saxena, K., Lung, B. R., & Becker, J. R. (2011). Improving patient safety by modifying provider ordering behavior using alerts (CDSS) in CPOE system. AMIA... Annual Symposium Proceedings / AMIA Symposium. AMIA Symposium, 2011, 1207–1216. PMID:22195181 Signorini, A., Segre, A. M., & Polgreen, P. M. (2011). The use of Twitter to track levels of disease activity and public concern in the US during the influenza A H1N1 pandemic. PLoS ONE, 6(5), e19467. doi:10.1371/journal.pone.0019467 PMID:21573238 Stephenson, W. P., & Hauben, M. (2007). Data mining for signals in spontaneous reporting databases: Proceed with caution. Pharmacoepidemiology and Drug Safety, 16(4), 359–365. doi:10.1002/pds.1323 PMID:17019675 Study Group of Inflammatory Bowel Disease. (2014). Grupo de Estudos da Doença Inflamatória Intestinal (GEDII). Retrieved from www.gedii.pt Szarfman, A., Machado, S. G., & O’Neill, R. T. (2002). Use of screening algorithms and computer systems to efficiently signal higher-than-expected combinations of drugs and events in the US FDA’s spontaneous reports database. Drug Safety, 25(6), 381–392. doi:10.2165/00002018-200225060-00001 PMID:12071774 White, R. W., Tatonetti, N. P., Shah, N. H., Altman, R. B., & Horvitz, E. (2013). Web-scale pharmacovigilance: Listening to signals from the crowd. Journal of the American Medical Informatics Association, 20(3), 404–408. doi:10.1136/amiajnl-2012-001482 PMID:23467469 D Category: Developments and Solutions 315 WHO. (n.d.). The Uppsala Monitoring Centre. Retrieved 09 Aug 2013, from http://www.who-umc.org/ DynPage.aspx?id=22682 Zorych, I., Madigan, D., Ryan, P., & Bate, A. (2013). Disproportionality methods for pharmacovigilance in longitudinal observational databases. Statistical Methods in Medical Research, 22(1), 39–56. doi:10.1177/0962280211403602 PMID:21878461 RESEARCH ARTICLE Open Access How to promote adverse drug reaction reports using information systems –a systematic review and meta-analysis Inês Ribeiro-Vaz 1,2* , Ana-Marta Silva 1,2 , Cristina Costa Santos 2,3 and Ricardo Cruz-Correia 2,3 Abstract Background: Adverse drug reactions (ADRs) are a well-recognized public health problem and a major cause of death and hospitalization in developed countries. The safety of a new drug cannot be established until it has been on the market for several years. Keeping drug reactions under surveillance through pharmacovigilance systems is indispensable. However, underreporting is a major issue that undermines the effectiveness of spontaneous reports. Our work presents a systematic review on the use of information systems for the promotion of ADR reporting. The aim of this work is to describe the state of the art information systems used to promote adverse drug reaction reporting. Methods: A systematic review was performed with quantitative analysis of studies describing or evaluating the use of information systems to promote adverse drug reaction reporting. Studies with data related to the number of ADRs reported before and after each intervention and the follow-up period were included in the quantitative analysis. Results: From a total of 3865 articles, 33 articles were included in the analysis; these articles described 29 different projects. Most of the projects were on a regional scale (62 %) and were performed in a hospital context (52 %). A total of 76 % performed passive promotion of ADR reporting and used web-based software (55 %). A total of 72 % targeted healthcare professionals and 24 % were oriented to patient ADR reporting. We performed a meta-analysis of 7 of the 29 projects to calculate the aggregated measure of the ADR reporting increase, which had an overall measure of 2.1 (indicating that the interventions doubled the number of ADRs reported). Conclusions: We found that most of the projects performed passive promotion of ADR reporting (i.e., facilitating the process). They were developed in hospitals and were tailored to healthcare professionals. These interventions doubled the number of ADR reports. We believe that it would be useful to develop systems to assist healthcare professionals with completing ADR reporting within electronic health records because this approach seems to be an efficient method to increase the ADR reporting rate. When this approach is not possible, it is essential to have a tool that is easily accessible on the web to report ADRs. This tool can be promoted by sending emails or through the inclusion of direct hyperlinks on healthcare professionals’desktops. Keywords: Adverse drug reactions report, Information systems, Pharmacovigilance * Correspondence: [email protected] 1 Northern Pharmacovigilance Centre, Faculty of Medicine, University of Porto, Rua Doutor Plácido da Costa, 4200-450 Porto, Portugal 2 Center for Health Technology and Service Research (CINTESIS), Faculty of Medicine of the University of Porto, Porto, Portugal Full list of author information is available at the end of the article © 2016 Ribeiro-Vaz et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ribeiro-Vaz et al. BMC Medical Informatics and Decision Making (2016) 16:27 DOI 10.1186/s12911-016-0265-8 Background Adverse drug reactions (ADRs) are a well-recognized public health problem worldwide and a major cause of death and hospitalization in developed countries [1]. Rare and long-term ADRs are difficult to detect during the drug development stage. Detecting new ADRs not previously identified during clinical trials is only possible when the drug begins to be used by a large population after marketing authorization (MA). The safety of a new drug cannot be established until it has been on the market for several years [2]. As such, it is indispensable to keep drug reactions under close surveillance after commercialization through a pharmacovigilance system to continuously evaluate the drug’s safety profile. In most countries, the pharmacovigilance system is based on spontaneous ADR reports made by healthcare professionals and consumers [3]. These reports can be made using paper, telephone, e-mail or through an on-line form and consist of a description of an adverse event apparently caused by a medicine. Spontaneous ADR reporting has been described as an efficient method to detect drug safety signs [4]; however, underreporting is a major issue that undermines the effectiveness of spontaneous reports. Several studies have suggested that less than 10 % of detected ADRs are effectively reported to medicine regulatory authorities [5, 6]. Worldwide, systems using informatics to promote ADR reporting or to detect the occurrence of ADRs in healthcare institutions have been tested and used, such as computer programs that allow voluntary and automated detection of ADR [7, 8] informatics tools created to analyse clinical databases [9] or websites that actively inform healthcare professionals [10]. In addition to signal detection, information and communication technologies can also be used to encourage and facilitate reporting of suspected ADR. In the present work, a systematic review is presented on the use of information systems in pharmacovigilance. Our main goal is to describe the state of the art information systems for the passive or active promotion of adverse drug reaction reporting. Methods Eligible studies Studies describing or evaluating the use of information systems to promote adverse drug reaction reports were selected. Review team The review team is composed of two pharmacists who are experts in pharmacovigilance (Inês Ribeiro Vaz (IV) and Ana Marta Silva (AS)) and the computer scientist Ricardo Cruz Correia (RC), who is an expert in medical informatics. Search methods Studies were searched in April 2014 in the bibliographic databases. We developed a search query that included the concepts adverse drug reaction, adverse drug reaction reporting system, pharmacovigilance and information system. Only articles written in English, Portuguese or French were included. We did not establish any criteria for the publication date. Four distinct bibliographic databases were searched: Medline(viaPubMed);ISI(ISIWebofKnowledge);IEEE (IEEE Xplore) and Scopus. The query search string used in Medline® was ((ADR OR “adverse drug reaction”OR “adverse drug reactions”OR “adverse drug event”OR “adversedrugevents”OR “adverse dug effect”OR “adverse drug effects”)OR“pharmacovigilance”). A similar query was used in the other databases and was adapted to the search engine. Selection of studies for the review The first selection was based on the study title and abstract (when available). Two reviewers on the review team (IV and AS) were involved in study selection and read all titles/abstracts. The study was considered eligible when at least one of the reviewers decided that the title/abstract mentioned the key concept of using information systems for ADR reporting. In cases of disagreement, a consensus meeting was held with the third reviewer (RC) to decide whether the article should be selected. The second phase of study selection was based on the full text. The team leader (IV) reviewed each fulltext article. In this stage, articles were excluded based on the following criteria: (1) the articles were only focused on medication errors; (2) the articles focused on ADR detection; (3) the articles were studies without any information system implemented; (4) the articles were studies concerning data quality; (5) the articles were studies focused on website usability; (6) the articles were only the authors’reflections on the theme; (7) the articles were studies only related to incidents that occurred in health institutions; (8) the articles were studies concerning signal detection and (9) the articles were studies concerning electronic transmission between the authority and other institutions (pharmaceutical companies or regional pharmacovigilance centres). The articles remaining after this review were included in the final statistical analysis. These articles were grouped into research projects to avoid the distortion created by multiple papers describing Ribeiro-Vaz et al. BMC Medical Informatics and Decision Making (2016) 16:27 Page 2 of 10 the same project (Fig. 1). All statistical analyses were based on the projects and not on the articles. Definition of variables The variables examined in these reviews were related to the projects, papers and information systems described in each project. We used the following data for project identification: (1) project number; (2) Information system name (if any); (3) country; (4) publication date; (5) type of study and (6) reference(s). According to the description of the projects, the following variables were analysed: 1. Area covered by the project (i.e., region, country, or hospital) 2. Type of action promoted by the project (passive promotion of ADR reporting or active promotion of ADR reporting) 3. Type of software (i.e., web-based or mobile) 4. Type of institution (i.e., regulatory authority or universities) 5. Target (healthcare professionals or patients) 6. Type of medicine (all, vaccines, chemotherapy, or others) 7. Type of ADR (all/serious ADRs based on the World Health Organization seriousness criteria [11]) Statistical analysis The inclusion criteria for the quantitative analysis were the availability of data related to the number of ADRs reported before and after each intervention and a follow-up period. Studies that only disclosed the increased ADR rate and studies that reported zero ADRs before the project implementation were excluded because it was not possible to perform the analysis in these cases. For each study with available data, the rate of ADRs reporting increase (quotient between ADR reports after and ADR reports before) and the respective 95 % confidence intervals were calculated. A rate of ADR reporting increase equal to 2 indicated that the ADR reports doubled after the intervention. Conversely, a rate of ADR reporting increase equal to 1 Records identified through database searching (n=5382) Records (after duplicates removed) screened based on title and abstract (n=3835) Full-text articles assessed for eligibility (n=114) Medline: 2519 papers IEEE: 68 papers ISI web of knowledge: 2603 papers Scopus: 192 papers Studies included in qualitative synthesis (n=33 articles / 29 projects) Records excluded (n=3721) - Not related with pharmacovigilance (n=643) - Not related with information system (n=85) - Other (n=2993), mostly because the focus was data mining on big databases, instead of ADR report. Duplicates removed (n= 1547) Full-text articles excluded (n=81) Studies included in quantitative synthesis (n=7 projects) Fig. 1 Flowchart of the study selection Ribeiro-Vaz et al. BMC Medical Informatics and Decision Making (2016) 16:27 Page 3 of 10 indicated that the number of ADR reports after the intervention was equal to the number of ADR reports before the intervention. The aggregated rate of the ADR reporting increase was calculated with the inverse variance method using a random effects model and a forest plot was presented. The confidence intervals, aggregated rate of ADRs and forest plot were performed using a Microsoft Excel spreadsheet. The description of the Microsoft Excel spreadsheet and the respective statistical methods used were described by Neyeloff [12]. Results Our search method found 2519 articles in PubMed, 68 in IEEE, 2603 in ISI and 192 in Scopus. After eliminating duplicate articles, 3835 articles were selected. Two reviewers (IV and AS) read all 3835 titles/abstracts. In cases of disagreement, which occurred with 151 articles, a consensus meeting was held with the third reviewer (RC) to decide whether the article should be selected. A total of 643 studies were excluded because they were not related to pharmacovigilance, 85 were excluded because they were not related to information systems and 2993 were excluded for other reasons (mostly because their focus was on data mining in large databases instead of ADR reporting). A total of 114 of the 3835 articles were selected in this first selection based on the title and abstract. The team leader (IV) reviewed each of the 114 fulltext articles. After this review, 33 articles remained for the final statistical analysis. At this stage, most of the articles were excluded because: (1) they were only related to medication errors; (2) they were focused on ADR detection; (3) they were studies without any information system implemented; (4) they were studies concerning data quality; (5) they were studies focused on website usability; (6) they were only authors’reflections on the theme; (7) they were studies only related to incidents that occurred at health institutions; (8) they were studies concerning signal detection or (9) they were studies concerning electronic transmission between the authority and other institutions (pharmaceutical companies or regional pharmacovigilance centres). These 33 articles were grouped into 29 distinct research projects to avoid the distortion created by multiple papers describing the same project (Fig. 1.). All statistical analyses was based on projects and not on articles. Table 1 lists all 29 projects, their country, the number of publications, the publication year and the journal. The country with the most published projects was the USA (11), followed by the United Kingdom (3). Trends There was an increasing trend in publication, especially after 2009 (Fig. 2). Qualitative analysis The qualitative variables analysed in each project are listed in Table 2 and described below. Globally, we found that there was an increase in the publication of projects over the study period, with 4 projects published before 2001, 4 projects between 2005 and 2007, 8 projects between 2008 and 2010 and 13 projects between 2011 and 2013. Geographic area covered by the projects Most of the projects were regional (62 %), followed by national projects (34 %). We found only 1 international project based on Facebook®. This international project was developed in the last time period (2011–2013). Areas covered by the projects Most of the projects (52 %) were developed in hospitals, followed by community projects (21 %). A total of 14 % covered primary care institutions and 10 % (3 projects) were developed for use in any type of healthcare institution. One project was dedicated to a multicentre clinical trial. We also found that all of the projects oriented to the community were developed in the last 3 years (2011–2013). Types of actions promoted by the projects The majority of the projects passively promoted ADR reporting (76 %); the remainder actively promoted reporting (24 %). Types of software More than half of the projects (55 %) used web-based technology and 41 % used electronic health records. Only one project used mobile phone technology. There was an increasing trend in software using web-based technology over all of the time intervals considered. The mobile technology appeared during the last time period. Types of institutions promoting the studies Most of the projects were promoted by hospitals and universities (31 % ex aequo). There were 4 projects developed by national institutions (not regulatory) and 5 projects implemented by regulatory authorities. Targets A total of 72 % of the projects were geared to healthcare professionals, 24 % to patients and one project was geared to both targets. Most of the projects targeting Ribeiro-Vaz et al. BMC Medical Informatics and Decision Making (2016) 16:27 Page 4 of 10