Web-based Intensive Monitoring. a patient based pharmacovigilance tool

Transcription

1 Web-based Intensive Monitoring a patient based pharmacovigilance tool

2 ISBN: The work presented in this thesis was performed at the Netherlands Pharmacovigilance Centre Lareb and the Department of Pharmacotherapy and Pharmaceutical Care, Rijksuniversiteit Groningen. Cover design: Beekhuis&Holthuis, Asten Lay-out inside work: Optima Grafische Communicatie, Rotterdam Printed by: Optima Grafische Communicatie, Rotterdam Financial support by the Nederlands Bijwerkingen Fonds for publication of this thesis is gratefully acknowledged. Linda Härmark, 2012

3 RIJKSUNIVERSITEIT GRONINGEN Web-based Intensive Monitoring a patient based pharmacovigilance tool Proefschrift ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. E. Sterken, in het openbaar te verdedigen op maandag 4 juni 2012 om uur door Linda Veronica Dae-Hee Härmark geboren op 25 januari 1978 te Seoul, Zuid-Korea

4 Promotores: Prof. dr. A.C. van Grootheest Prof. dr. J.J. de Gier Copromotor: Dr. E.P. van Puijenbroek Beoordelingscommissie: Prof. dr. H.G.M. Leufkens Prof. dr. S.A. Shakir Prof. dr. B. Wilffert

5 Table of contents Chapter 1 Introduction 7 Chapter 2 Pharmacovigilance and intensive monitoring Pharmacovigilance: methods, recent developments and future perspectives 19 Chapter 3 Patients as a source of information in web-based intensive monitoring Patients motives for participating in active post-marketing surveillance Non-response in a pharmacy and patient based intensive monitoring system, a 55 quantitative study on non-response bias and reasons for non-response Chapter 4 Representativeness of patients participating in a web-based intensive monitoring system 4.1 Representativeness of diabetes patients participating in a web-based adverse drug reaction monitoring system Chapter 5 Description of the Lareb Intensive Monitoring system using pregabalin and duloxetine as examples 5.1 Intensive monitoring of pregabalin: results from an observational, web-based, prospective cohort study in the Netherlands using the patient as a source of information 5.2 Intensive monitoring of duloxetine, results from a web-based intensive monitoring study 5.3 Longitudinal monitoring of the safety of drugs by using a web-based system: the case of pregabalin Chapter 6 Application of web-based intensive monitoring Monitoring the safety of influenza A (H1N1) vaccine using web-based intensive 133 monitoring 6.2 Web-based intensive monitoring: from passive to active drug surveillance 153 Chapter 7 General discussion 167 Summary Samenvatting Sammanfattning Dankwoord List of publications About the author

6

7 Chapter 1 Introduction

8

9 Introduction 9 1. Introduction Pharmacovigilance 4. During the twentieth century drug development as an industry started to take off. With the 5. discovery of insulin, penicillin and sulphonamides it was possible to treat diseases which 6. had previously been deadly, saving millions of lives. In the early days of drug development 7. there were no regulations regarding a drug s quality, efficacy and safety [1]. In the late 1950 s 8. and beginning of the 1960 s it became evident that drugs were not only saving lives, they 9. could also have a negative impact on it [2]. Even though adverse drug reactions (ADRs) were 10. acknowledged as a problem related to drug use, it was the congenital abnormalities seen 11. in children whose mothers had used thalidomide during pregnancy that acted as a wake 12. up call to start to look more intensively at the negative effects of drugs [3]. Following the 13. thalidomide disaster, authorities all over the world began to set up systems in order to monitor the safety of drugs. These spontaneous reporting systems were based on the collection of reports of ADRs from healthcare professionals. The World Health Organisation (WHO) 16. recognised the need for global drug monitoring, since drugs are seldom registered in one 17. country only, and in 1968 the WHO Pilot Research Project for International Drug Monitoring 18. started its operation with 10 participating countries, with the purpose to develop a system 19. for the detection of previously unknown or poorly understood adverse effects of drugs [4,5]. 20. From there, the practice as well as the science of pharmacovigilance has been developed and 21. today pharmacovigilance is defined by the WHO as the science and activities relating to the 22. detection, assessment, understanding and prevention of adverse effects or any other drug 23. related problem [5] Pharmacovigilance in the 21st century 26. In the last few years there has been renewed interest in pharmacovigilance. It started with the 27. withdrawal of rofecoxib in 2004 [6] followed by the debate about the cardiovascular safety 28. of rosiglitazone [7,8], which ultimately lead to the suspension of the marketing authorisation 29. of the drug in the European Union (EU). These high profile cases lead to a debate about the 30. ability of the current pharmacovigilance system to identify harm [9-12] and forced the pharmacovigilance community to critically evaluate the existing pharmacovigilance systems in place [13,14]. In the European Union the evaluation of the pharmacovigilance system in [14] lead to legislative changes, which were endorsed in September 2010 and will come into 34. force in July 2012 [15,16]. To support the implementation of the new EU pharmacovigilance 35. legislation, the European Medicines Agency (EMA) is developing a new set of guidelines for 36. the conduct of pharmacovigilance. This new guidance on good pharmacovigilance practices 37. (GVP) is organised in 16 different modules [17]. 1

10 10 Chapter 1 1. With the new legislation a strengthening of post-authorisation regulation of medicines will 2. be implemented, which has two key elements: one related to the process, where it is important that there are clear roles, responsibilities and obligations for the key responsible parties and the other related to the collection of high-quality data relevant to the safety of medicines 5. and patient safety, which is a requirement for the prompt identification of potential risks Spontaneous reporting of ADRs by physicians and pharmacists has been the backbone of 8. data collection in pharmacovigilance and has proven its value in detecting relatively rare and 9. serious ADRs [18]. Within the new legislation, spontaneous reporting will continue to play an 10. important role and the range of possible reporters will be expanded by including patients. 11. Research in the last few years has shown that patients reports are a valuable addition to a 12. spontaneous reporting system [19-23]. Besides spontaneous reporting by healthcare professionals and patients, there is also a need for a different kind of information about ADRs than spontaneous reporting can provide. Spontaneous reporting systems focus on detecting 15. signals of new ADRs and it has proven its strength in detecting previously unknown harm. It 16. has also met criticism; under-reporting and its inability to quantify adverse drug reactions are 17. most often mentioned [18] Waller and Evans [24] have suggested that pharmacovigilance should be less focused on finding harm and more focused on extending knowledge of safety. Increased information about the safety of a drug, e.g. information about the time course of ADRs such as time to onset and 22. duration can help patients to be adherent to their medication. This can be achieved through 23. active, systematic collection of information about adverse drug reactions Intensive Monitoring 26. In an effort to come to terms with some of the shortcomings of spontaneous reporting i.e. 27. under-reporting and the inability to quantify ADRs, a new form of active surveillance was 28. developed. In the late 1970 s the Intensive Monitoring Medicines Programme (IMMP) was 29. established in New Zealand [25] and since the beginning of the 1980 s the Prescription Event 30. Monitoring programme (PEM) has been running in the UK [26]. The basis of these intensive 31. monitoring systems is a non-interventional observational cohort where users of certain drugs 32. are identified on basis of prescription data. The prescriber of the drug is sent a questionnaire 33. and is asked about any adverse events that may occur during the use of the drug being 34. monitored. These data are collected and analysed for new signals [25,26] Intensive monitoring distinguishes itself from spontaneous reporting because the former 37. only monitors selected drugs during a certain period of time. Through its non-interventional character, intensive monitoring provides real-world clinical data involving neither inclusion nor exclusion criteria throughout the collection period. Since it is based on event monitoring

11 Introduction it can identify signals for events that were not necessarily suspected as being ADRs of the 2. drug under study. Since it is a cohort study, it enables the incidence of adverse events to be 3. estimated, thus enabling quantification of ADRs Intensive monitoring also has limitations. As in spontaneous reporting, the proportion of 6. adverse effects that go unreported to doctors is unknown; the studies therefore produce 7. reported event rates rather than true incident rates. This is the same for all studies based on 8. medical record data, including computer databases and record linkage. There is no control 9. group in standard intensive monitoring studies, and the true background incidence for 10. events is therefore not known [25,26] Intensive Monitoring in the Netherlands 13. The Lareb Foundation started in the eighties as a local initiative of pharmacists and general 14. practitioners to collect adverse drug reaction reports with the aim to improve pharmacotherapy [27]. What started as a local initiative grew, and in 1995 the Netherlands Pharmaco vigilance Centre Lareb became responsible for the collection and analysis of adverse drug 17. reaction reports in the Netherlands. In order to be able to do this on a high scientific level, 18. research concerning the core business has always been part of its activities [28]. In recent 19. years, research in the areas of statistical signal detection [29-33] and the contribution of 20. pharmacists [34-38] and patients [39] reports to pharmacovigilance has been conducted in 21. order to develop the spontaneous reporting system further. In the continuous process of 22. trying to further develop its activities, Lareb recognised the need for development of new 23. pharmacovigilance methods that could act as a complement to the spontaneous reporting 24. system In 1996 a study was conducted which investigated if the first prescription signal generated in 27. the pharmacy could be of use in post-marketing surveillance [40]. Patients identified through 28. such a signal were given a questionnaire to complete and it was evaluated if the information provided in the questionnaire could give a clear picture of the patient s experiences with the drug. This study concluded that the first prescription signal in the pharmacy was a 31. good way to identify users of specific drugs in order to follow their experiences with the drug 32. intensively A few years later a second study was conducted in which again the first prescription signal 35. in the pharmacy was used to identify new users of a particular drug, however in this case the 36. questionnaires were not given to the patient but to the prescribing GP. The GP was asked to 37. complete the questionnaire upon the patient s next visit. An evaluation of this study showed that pharmacists and GPs are motivated to participate in an intensive monitoring system [41]. 1

12 12 Chapter 1 1. After these two initial studies it was decided to develop an intensive monitoring system. 2. As Lareb had experiences with patient reports [19] and believed in the patient as an important player in pharmacovigilance, it was decided to use patients as a source of information [19,42,43]. In the same period, Lareb had gained experience in using IT-solutions for data 5. processing, so it was chosen to make the system web-based. In 2006 Lareb Intensive Monitoring (LIM), a web-based intensive monitoring system using patients as a source of information was introduced complementary to the spontaneous reporting system In the majority of studies presented in this thesis patients eligible for inclusion were identified using the first dispensation in the pharmacy. However, inclusion is not limited to the pharmacy and one study presented in this thesis uses the general practitioner s office as 12. point of inclusion. At the inclusion point, the patient is informed about the intensive monitoring study and is asked to participate. When registering online, the patient is asked for an address which will be used for further correspondence. In addition, information about 15. patient characteristics and drug use is collected. After registration, the patient receives questionnaires by at specific points in time, allowing longitudinal data collection. In these questionnaires, questions are asked about drug use and possible ADRs. These data are coded 18. and analysed with the purpose of identifying new signals or obtaining information that will 19. extend the knowledge about the safety of the drug under study Objectives and outline of the thesis Web-based intensive monitoring using patients as a source of information was developed 25. with the aim of gathering data about the safety of drugs that more traditional methods such 26. as clinical trials or spontaneous reporting are not able to do. In order for web-based intensive 27. monitoring to be a useful pharmacovigilance method, it has to provide proof of concept and 28. show what kind of information it can capture. In addition, since the method brings in a few 29. relatively new elements such as patients as the source of information and using web-based 30. questionnaires, the method needs to be further characterised The objective of this thesis is to describe a web-based intensive monitoring system using patients 33. as a source of information and its application as a pharmacovigilance tool Chapter 2 provides a background to the field of pharmacovigilance. In addition, current 36. methods used in pharmacovigilance such as clinical trials, spontaneous reporting and intensive monitoring are described in more detail with their strengths and limitations. 37.

13 Introduction In all studies performed with web-based intensive monitoring so far, the patient has been the 2. source of information. Patient participation is increasing in pharmacovigilance but until now 3. this has been limited to spontaneous reporting. Chapter 3 focuses on patient participation in 4. web-based intensive monitoring. Patient participation is essential for the good functioning of 5. the system and therefore patients motivation for participation was investigated. In addition, 6. reasons for non-response were also identified to see what the barriers for participation are Since not all patients who are eligible for inclusion chose to participate in web-based intensive monitoring, it is important to obtain information about the characteristics of the LIM population compared to the whole population using the drug. In Chapter 4, a web-based 11. intensive monitoring population is compared to a reference population to see how they 12. compare on parameters which may influence a patient s susceptibility to develop adverse 13. drug reactions With the theoretical concept of web-based intensive monitoring, it was believed that the 16. system could generate different types of information than for example spontaneous reporting. Chapter 5 provides proof of the theoretical concept. The results from the first web-based intensive monitoring studies, concerning the drugs pregabalin and duloxetine, illustrate the 19. kind of data that can be obtained through longitudinal web-based intensive monitoring The characteristics of web-based intensive monitoring, using a specific inclusion point, 22. letting the patient be the source of information and collecting data through web-based 23. questionnaires can be used to collect data about the safety of drugs in other settings than 24. through community pharmacy. In Chapter 6 this is further elaborated and an example is 25. given how web-based intensive monitoring was used to gather data about the safety of the 26. Influenza A (H1N1 ) pandemic vaccine Chapter 7 comprises a general discussion where web-based intensive monitoring as a 29. methodology will be summarised and future research will be suggested. In addition, since 30. this thesis only shows one application of web-based intensive monitoring, future perspectives about the application of the method and how it can be used to meet society s need for information about adverse drug reactions will be presented

14 14 Chapter References 1. Heath G, Colburn WA. An evolution of drug development and clinical pharmacology during the 20th century. J Clin Pharmacol 2000; 40: van Grootheest K. The dawn of pharmacovigilance: an historical perspective. Int J Pharm Med 2003; 17: McBride WG. Thalidomide and congenital malformations. Lancet 1961; 2: Venulet J, Helling-Borda M. WHO s international drug monitoring--the formative years, : preparatory, pilot and early operational phases. Drug Saf 2010; 33: e1-e The Importance of Pharmacovigilance. WHO Available via Accessed Jan 20, Merck Announces Voluntary Worldwide Withdrawal of VIOXX. Merck & Co Available via Accessed March 1, Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med 2007; 356: Singh S, Loke YK, Furberg CD. Long-term risk of cardiovascular events with rosiglitazone: a metaanalysis. JAMA 2007; 298: Greener M. Drug safety on trial. Last year s withdrawal of the anti-arthritis drug Vioxx triggered a debate about how to better monitor drug safety even after approval. EMBO Rep 2005; 6: Horton R. Vioxx, the implosion of Merck, and aftershocks at the FDA. Lancet 2004; 364: Krumholz HM, Ross JS, Presler AH, et al. What have we learnt from Vioxx? BMJ 2007; 334: Rosen CJ. The rosiglitazone story--lessons from an FDA Advisory Committee meeting. N Engl J Med 2007; 357: Baciu A, Stratton K, Burke SP, editors. Committee on the Assessment of the US Drug Safety System The future of drug safety: promoting and protecting the health of the public. Institute of Medicine Washington DC. 14. Assessment of the European Community System of Pharmacovigilance. European Medicines Agency Available via Accessed March 1, Directive 2010/84/EU. Official Journal of the European Union 2010, Dec 31,L. 348/ Regulation 1235/2010. Official Journal of the European Union 2010, Dec 31,L. 348/ Good Pharmacovigilance Practices. European medicines Agency Available via Accessed March 1, Raine JM. Risk management - a European Regulatory View. In: Mann R, Andrews E (eds) Pharmacovigilance. 2nd edn Wiley, Chichester. 19. de Langen J, van Hunsel F, Passier A, et al. Adverse drug reaction reporting by patients in the Netherlands: three years of experience. Drug Saf 2008; 31: van Hunsel F, Talsma A, van Puijenbroek E, et al. The proportion of patient reports of suspected ADRs to signal detection in the Netherlands: case-control study. Pharmacoepidemiol Drug Saf 2011; 20: Aagaard L, Nielsen LH, Hansen EH. Consumer reporting of adverse drug reactions: a retrospective analysis of the Danish adverse drug reaction database from 2004 to Drug Saf 2009; 32: Anderson C, Krska J, Murphy E, et al. The importance of direct patient reporting of suspected adverse drug reactions: a patient perspective. Br J Clin Pharmacol 2011; 72:

15 Introduction McLernon DJ, Bond CM, Hannaford PC, et al. Adverse drug reaction reporting in the UK: a retrospective observational comparison of yellow card reports submitted by patients and healthcare professionals. Drug Saf 2010; 33: Waller PC, Evans SJ. A model for the future conduct of pharmacovigilance. Pharmacoepidemiol Drug Saf 2003; 12: Harrison-Woolrych M, Coulter DM. PEM in New Zealand. In: Mann R, Andrews E (eds) Pharmacovigilance. 2nd edn Wiley, Chichester. 26. Shakir SAW. PEM in the UK. In: Mann R, Andrews E (eds) Pharmacovigilance. 2nd edn Wiley, Chichester. 27. De Koning, GHP. A Regionalized Spontaneous Surveillance Program for Adverse Drug Reactions as a Tool to Improve Pharmacotherapy, Thesis Utrecht University Meyboom, RHB. Detecting adverse drug reactions, pharmacovigilance in the Netherlands, Thesis Nijmegen University van Puijenbroek EP, Diemont W, van Grootheest K. Application of quantitative signal detection in the Dutch spontaneous reporting system for adverse drug reactions. Drug Saf 2003; 26: van Puijenbroek EP, van Grootheest K, Diemont WL, et al. Determinants of signal selection in a spontaneous reporting system for adverse drug reactions. Br J Clin Pharmacol 2001; 52: van Puijenbroek EP, Bate A, Leufkens HG, et al. A comparison of measures of disproportionality for signal detection in spontaneous reporting systems for adverse drug reactions. Pharmacoepidemiol Drug Saf 2002; 11: van Puijenbroek EP, Egberts, ACG, Meyboom RHB, et al. Signalling possible drug-drug interactions in a spontaneous reporting system: delay of withdrawal bleeding during concomitant use of oral contraceptives and itraconazol. Br J Clin Pharmacol 1999; 47: van Puijenbroek EP. Quantitive Signal Detection in Pharmacovigilance, Thesis Utrecht University van Grootheest AC, van Puijenbroek EP, de Jong-van den Berg LT. Contribution of pharmacists to the reporting of adverse drug reactions. Pharmacoepidemiol Drug Saf 2002; 11: van Grootheest AC, Mes K, de Jong-van den Berg LTW. Attitudes of community pharmacists in the Netherlands towards adverse drug reaction reporting. Int J Pharm Pract 2002; 10: van Grootheest K, Olsson S, Couper M, et al. Pharmacists role in reporting adverse drug reactions in an international perspective. Pharmacoepidemiol Drug Saf 2004; 13: van Grootheest AC, de Jong-van den Berg LTW. The role of hospital and community pharmacists in pharmacovigilance. Res Social Adm Pharm 2005; 1: van Grootheest AC. Improving pharmacovigilance and the role of the pharmacist, Thesis Groningen University van Hunsel F. The contribution of direct patient reporting to pharmacovigilance, Thesis Groningen University van Puijenbroek EP, van Amerongen CA. [Is the first dispensation signal useful in postmarketing surveillance? Results from a pilot study] Pharm Weekbl 1996; 131: van Grootheest AC, Groote JK, de Jong-van den Berg LT. Intensive monitoring of new drugs based on first prescription signals from pharmacists: a pilot study. Pharmacoepidemiol Drug Saf 2003, 12: van Grootheest K, de Graaf L, de Jong-van den Berg LT. Consumer adverse drug reaction reporting: a new step in pharmacovigilance? Drug Saf 2003; 26: van Grootheest K, de Jong-van den Berg LT. Patients role in reporting adverse drug reactions. Expert Opin Drug Saf 2004; 3:

16

17 Chapter 2 Pharmacovigilance and intensive monitoring

18

19 Chapter 2.1 Pharmacovigilance: methods, recent developments and future perspectives Härmark L van Grootheest A.C European Journal of Clinical Pharmacology 2008; 64:

20 20 Chapter Abstract Background 4. Pharmacovigilance, defined by the World Health Organisation as the science and activities 5. relating to the detection, assessment, understanding and prevention of adverse effects or 6. any other drug-related problem plays a key role in ensuring that patients receive safe drugs. 7. Our knowledge of a drug s adverse reactions can be increased by various means, including 8. spontaneous reporting, intensive monitoring and database studies. New processes, both 9. at a regulatory and a scientific level, are being developed with the aim of strengthening 10. pharmacovigilance. On a regulatory level, these include conditional approval and risk management plans; on a scientific level, transparency and increased patient involvement are two important elements Objective 15. To review and discuss various aspects of pharmacovigilance, including new methodological 16. developments

21 Pharmacovigilance and intensive monitoring Introduction The field of drug safety has been receiving a great deal of attention lately. Almost weekly, 4. tabloids as well as scientific journals publish articles on drugs that cause unexpected adverse 5. drug reactions (ADRs). These articles have the unfortunate result of evoking apprehension 6. in both patients and health professionals regarding the use of these drugs. A more serious 7. consequence may be that the patient stops taking the prescribed medication, which may 8. lead to an even more serious situation than the ADR he was initially concerned about. Pharmacovigilance, defined by the World Health Organisation (WHO) as the science and activities relating to the detection, assessment, understanding and prevention of adverse effects or 11. any other drug-related problem [1], plays a vital role in ensuring that doctors, together with 12. the patient, have enough information to make an educated decision when it comes to choosing a drug for treatment. The aim of this review is to provide a summary of the most common methods used in pharmacovigilance to guarantee the safety of a drug. Recent developments 15. in pharmacovigilance as well as future needs are discussed As an introduction to the sort of problems pharmacovigilance has to face, a few examples of 18. recent safety concerns and the action taken are briefly described Safety concerns 21. The withdrawal of rofecoxib directed renewed attention to drug safety. The decision to withdraw rofecoxib was made after the safety monitoring board of the APPROVe trial found an increased risk of cardiovascular (CV) events in patients treated with rofecoxib compared to 24. placebo [2]. The events leading to the withdrawal of rofecoxib, and what have happened since 25. the withdrawal, have been discussed in numerous papers [3-6]. Another association that has 26. been much debated the last year is the association between rosiglitazone and cardiac effects. 27. In June 2007 a meta-analysis was published wherein the use of rosiglitazone was linked to an 28. increased risk of myocardial infarction and death from cardiovascular causes [7]. The results 29. of this one meta-analysis kindled a growing debate on the safety of the drug [8-11], and 30. new studies were rapidly published with the aim of rejecting or confirming the results of 31. the first study [12,13]. Both the US Food and Drug Administration (FDA) and the European 32. Medicines Agency (EMEA) have now concluded that the benefits of rosiglitazone outweigh 33. its risks within the framework of its approved indications [14,15]. However, constant revision/ 34. updating of product information and a continued monitoring of this ADR are necessary A more recent safety concern is the association between aprotinin and increased mortality. In 2006, a study based on observational data was published by Mangano et al. in which 37. these authors questioned the safety of aprotinin [16]. On November 21, 2007, aprotinin was withdrawn from the market in the European Union based on data from the BART clinical 2

22 22 Chapter trial showing increased mortality for patients receiving aprotinin [17]. Table 1 provides an overview of recent major drug safety issues and the evidence that led to their discovery Table 1. Drug safety issues and their evidence in Europe since Drug Trovofloxacin Tolcapone Cisapride Safety concern Hepatoxicity Hepatoxicity QT prolongation Key evidence Spontaneous ADRs Spontaneous ADRs Spontaneous ADRs Regulatory action Withdrawn Suspended Patient registration 11. cardiac arrhythmias licences subsequently cancelled Bupropion Cerivastatin Hormone replace therapy Seizures Drug interaction Rhabdomyolysis CVS risk and cancer long term Spontaneous ADRs Spontaneous ADRs Epidemiological studies Posology change, Warnings Withdrawn Warnings and restriction of indication SSRIs Suicidal behaviour in Clinical trials Warnings accompanied by 20. children clinical guidance COX IIs CVS risk Clinical trials Warnings and clinical guidance Topical macrolide Risk of cancer Spontaneous reports Restriction of use, 25. immunosuppressants Risk management plan SSRI, selective serotonin reuptake inhibitors, CVS, cardiovascular safety, ADR, adverse drug reaction. From Pharmacovigilance; Risk Management- a European Regulatory View. J.M Raine. Copyright Copyright John Wiley & Sons Limited. Reproduced with permission Whenever a drug safety issue occurs, the first reaction is to search for a reason of why such a thing could happen. In the case of rofecoxib, this led to a critical evaluation of the current methods and mechanisms available for safeguarding the safe use of a drug Regulatory action after rofecoxib withdrawal In the aftermath of the withdrawal of rofecoxib, the FDA and the current system of postmarketing surveillance was heavily criticised on a number of points [18-23]. Firstly, the FDA uses only a limited number of data sources (clinical trials, spontaneous reporting) when it comes to assembling information on the safety of a drug. Secondly, the FDA has no control

23 Pharmacovigilance and intensive monitoring over the performance of post-marketing safety studies. The majority of post-marketing study 2. commitments are never initiated, and the proportion of post-marketing safety studies (phase 3. 4 studies) that were completed declined from 62% between 1970 and 1984 to 24% between and Thirdly, the FDA has no authority to take direct legal action against companies that do not fulfill their post-marketing commitments [24]. Some critics also claim that the FDA has become too close to the industry that they are supposed to regulate and that a 7. separation between regulatory duties and the post-marketing surveillance activities is desirable [21]. In response to the criticism, the Centre for Drug Administration (CDER) at the FDA asked the Institute of Medicine (IOM) to assess the US drug safety system. In September 2006, 10. the IOM released the committee s findings and recommendations in a report The future of 11. drug safety: promoting and protecting the health of the public [25]. The main message in 12. this report is that the FDA needs to follow the safety of a drug during its whole life cycle. This 13. life-cycle approach includes identifying safety signals, designing studies to confirm them, 14. evaluating benefits as well as risks, using risk-benefit assessments to integrate study results 15. and communicating key findings to patients and physicians [24,26] In Europe the withdrawal of rofecoxib led to an assessment of the pharmacovigilance system 18. in the different European Union member states, which was published in March The 19. report Assessment of the European Community System of Pharmacovigilance highlighted 20. the strengths and weaknesses of the European pharmacovigilance system. The report s 21. recommendations focused on the breadth and variety of data sources, the pro-active use 22. of registration, the speed of decision-making, the impact of regulatory action and communication, compliance by marketing authorisation holders and general principles of quality management and continuous quality improvement [27,28] Methods used in pharmacovigilance The activities undertaken in the name of pharmacovigilance can be roughly divided into 30. three groups: regulatory, industry, and academia. Regulatory pharmacovigilance is driven 31. by the aim to provide drugs with a positive benefit-harm profile to the public. Some of the 32. problems related to regulatory post-marketing surveillance will be discussed in this context, 33. followed by a description of the methods used to detect new ADRs and a discussion of the 34. pros and cons of each method Clinical trial data insufficient to evaluate drug risk 37. The main method currently used to gather information on a drug in the pre-marketing phase is to conduct a clinical trial. Pre-marketing clinical trials can be divided into three phases. Phase III studies are often double-blind randomised controlled trials; these are considered 2

24 24 Chapter to be the most rigorous approach to determining whether a cause-effect relationship exists 2. between a treatment and an outcome. However, when it comes to monitoring the safety of a 3. drug, this study design is not optimal. Due to the limited number of patients participating, it 4. is generally not possible to identify ADRs that occur only rarely. The relatively short duration 5. of clinical trials makes it difficult to detect ADRs with a long latency. Another limitation of 6. clinical trials is the population in which a drug is tested. The characteristics of the participants 7. do not always correspond to the characteristics of the population in which it will later be 8. used; consequently, it may be difficult to extrapolate the results obtained from clinical trials 9. to the population at large [29]. This is especially true for the elderly, for women or for people 10. belonging to a minority ethnic group [30,31]. In order to study rare ADRs, ADRs with a long latency and ADRs in specific populations, careful monitoring of the drug in the post-marketing phase is essential Post-marketing studies can be descriptive or analytical. Descriptive studies generate hypotheses and attempt to describe the occurrence of events related to drug toxicity and efficacy Analytical studies test hypotheses and seek to determine associations or causal connections 17. between observed effects and particular drugs, and to measure the size of these effects. 18. Descriptive studies are widely used in post-marketing surveillance because they are able to 19. generate hypotheses that will become starting points for analytical studies [32]. Two forms of 20. descriptive studies, spontaneous reporting and intensive monitoring, will be discussed here. 21. Analytical studies can be conducted using a variety of approaches, including case-control 22. studies, cohort studies and clinical trials. In order to be able to conduct retrospective cohort 23. and case-control studies, data which have been collected in a reliable and routine manner 24. needs to be available. To provide an example of such studies, we describe here two European 25. databases frequently used for analytical studies, the General Practitioners Research Database 26. (GPRD) in the UK and the PHARMO Record Linkage System in the Netherlands Spontaneous reporting 29. In 1961, a letter from the Australian physician WG McBride was published in Lancet. In this 30. letter, he shared his observation that babies whose mothers had used thalidomide during 31. pregnancy were born with congenital abnormalities more often than babies who had not 32. been exposed to thalidomide in utero [33]. In the years to come it became evident that 33. thousands of babies had been born with limb malformations due to the maternal use of 34. thalidomide. In order to prevent a similar disaster from occurring, systems were set up all over 35. the world with the aim of regulating and monitoring the safety of drugs Spontaneous reporting systems (SRS) were created, and these have become the primary method of collecting post-marketing information on the safety of drugs. The main function of SRS is the early detection of signals of new, rare and serious ADRs. A spontaneous reporting

25 Pharmacovigilance and intensive monitoring system enables physicians and, increasingly more often, pharmacists and patients to report 2. suspected ADRs to a pharmacovigilance centre [34-36]. The task of the pharmacovigilance 3. centre is to collect and analyse the reports and to inform stakeholders of the potential risk 4. when signals of new ADRs arise. Spontaneous reporting is also used by the pharmaceutical 5. industry to collect information about their drugs. By means of a SRS it is possible to monitor 6. all drugs on the market throughout their entire life cycle at a relatively low cost. The main 7. criticism of this approach is the potential for selective reporting and underreporting [37]. In a 8. review article, Hazell and Shakir investigated the magnitude of underreporting in SRS and determined that more than 94% of all ADRs remain unreported [38]. Underreporting can lead to the false conclusion that a real risk is absent, while selected reporting of suspected risks may 11. give a false impression of a risk that does not exist. However, underreporting and selective 12. reporting can also been seen as advantages. Because only the most severe and unexpected 13. cases are reported, it is easier to detect new signals of ADRs because the person reporting the 14. reaction has already pinpointed what may be a new safety issue. Against this background, 15. the system should perhaps be called concerned reporting instead of spontaneous reporting, 16. seeing as those reporting the issues are highly selective of what they are reporting [39]. With 17. a SRS, it is not possible to establish cause-effect relationships or accurate incidence rates; it 18. is also not possible to understand risk factors or elucidate patterns of use. Although critics 19. say that spontaneous reporting is not the ideal method for monitoring the safety of drugs, 20. it has proven its value throughout the years. Eleven products were withdrawn from the UK 21. and U.S. markets between 1999 and Randomised trial evidence was cited for two products (18%) and comparative observational studies for two products (18%). Evidence from spontaneous reports supported the withdrawal of eight products (73%), with four products 24. (36%) apparently withdrawn on the basis of spontaneous reports only. For two products, 25. the evidence used to support their withdrawal could not be found in any of the identified 26. documentation [40]. Of nine recent significant drug safety issues handled in the European 27. Union since 1995, six were detected by spontaneous reports, Table 1, which demonstrates 28. the strength of spontaneous reporting in detecting new safety issues [28] Data mining in spontaneous reporting 31. In the past, signal detection in spontaneous reporting has mainly occurred on the basis of 32. case-by-case analyses of reports. In recent years, however, data mining techniques have 33. become more important. The term data mining refers to the principle of analysing data 34. from different perspectives and extracting the relevant information. Algorithms are often 35. used to determine hidden patterns of associations or unexpected occurrences, i.e. signals, 36. in large databases. Although the methodology of the various data mining methods applied 37. in pharmacovigilance differ, they all share the characteristic that they express to what extent the number of observed cases differs from the number of expected cases [41]. 2

26 26 Chapter Several approaches of data mining are currently in use. Proportional reporting ratios (PPRs), 2. compare the proportion of reports for a specific ADR reported for a drug with the proportion 3. for that ADR in all other drugs. The calculation is analogous to that of relative risk. Using the 4. same information, it is also possible to calculate a reporting odds ratio [42] The Bayesian confidence propagation neural network (BCPNN) method is used to highlight 7. dependencies in a data set. This approach uses Bayesian statistics implemented in a neural 8. network architecture to analyse all reported ADR combinations. Quantitatively unexpectedly 9. strong relationships in the data are highlighted relative to general reporting of suspected 10. adverse effects. The WHO Collaborating Centre for International Drug Monitoring uses this 11. method for data mining [43]. A related approach is the Multi-Item Gamma Poisson Shrinker 12. (MGPS) used by the FDA for data mining of their spontaneous report s database. The MGPS 13. algorithm computes signal scores for pairs, and for higher-order (e.g. triplet, quadruplet) 14. combinations of drugs and events that are significantly more frequent than their pair-wise 15. associations would predict [44]. All data-mining approaches currently cannot distinguish 16. between associations that are already known and new associations. Moreover, clinical information described in the case reports is not taken into account; consequently, there is still the need for a reviewer to analyse these events Intensive monitoring 21. In the late 1970s and early 1980s a new form of active surveillance was developed in New 22. Zealand (Intensive Medicines Monitoring Programme) and the UK (Prescription Event 23. Monitoring). These intensive monitoring systems use prescription data to identify users of a 24. certain drug. The prescriber of the drug is asked about any adverse event occurring during 25. the use of the drug being monitored. These data are collected and analysed for new signals. 26. The methodology of these intensive monitoring systems have been described in depth 27. elsewhere [45-48] The basis of intensive monitoring is a non-interventional observational cohort, which distinguishes it from spontaneous reporting because the former only monitors selected drugs dur ing a certain period of time. Through its non-interventional character, intensive monitoring 32. provides real world clinical data involving neither inclusion nor exclusion criteria throughout 33. the collection period. It is unaffected by the kind of selection and exclusion criteria that characterise clinical trials, thereby eliminating selection bias. Another strength of the methodol ogy is that it is based upon event monitoring and is therefore capable of identifying signals 36. for events that were not necessarily suspected as being ADRs of the drug being studied. 37. Intensive monitoring programmes also enable the incidence of adverse events to be estimated, thus enabling quantification of the risk of certain ADRs. This approach, however, also has recognised limitations. The proportion of adverse effects that go unreported to doctors is

27 Pharmacovigilance and intensive monitoring unknown. The studies also produce reported event rates rather than true incident rates. This 2. is the same for all studies based on medical record data, including computer databases and 3. record linkage. There is no control group in standard intensive monitoring studies, and the 4. true background incidence for events is therefore not known [49] Although the intensive monitoring methodology was developed more than 20 years ago, 7. this methodology has received renewed interest in the last years. In the European Commission consultation Strategy to better protect public health by strengthening and rationalising EU pharmacovigilance intensive monitoring is mentioned as one tool that can improve the 10. pharmacovigilance system [50] Database studies 13. In order to test a hypothesis, a study has to be performed. The study can be conducted using a variety of methods, including case-control studies and cohort studies. The limitations of these methods include power considerations and study design. In order to be able to 16. conduct retrospective cohort and case-control studies, data which have been collected in a 17. reliable and routine fashion needs to be available. The General Practice Research Database 18. (GPRD) and the PHARMO Record Linkage System, which will be described in further detail 19. in the following sections, were chosen here because they represent two different types of 20. European databases. Other database and record linkage systems are available for research 21. purposes in both Europe and in North America [51] General Practice Research Database 24. Virtually all patient care in the UK is coordinated by the general practitioner (GP), and data 25. from this source provide an almost complete picture of a patient, his illnesses and treatment. 26. In any given year, GPs, who are members of the GPRD, collect data from about 3 million 27. patients (about 5% of the UK population). These patients are broadly representative of the 28. general UK population in terms of age, sex and geographic distribution. The data collected 29. include demographics (age and sex), medical diagnoses that are part of routine care or 30. resulting from hospitalisations, consultations or emergency care, along with the date and 31. location of the event. There is also an option of adding free text, referral to hospitals and 32. specialists, all prescriptions, including date of prescription, formulation strength, quantity 33. and dosing instructions, indication for treatment for all new prescriptions and events leading 34. to withdrawal of a drug or a treatment. Data on vaccinations and miscellaneous information, 35. such as smoking, height, weight, immunisations, pregnancy, birth, death, date entering the 36. practice, date leaving the practice and laboratory results, are also collected. 37. A recent review of protocols using GPRD data showed that the database is used for pharmacoepidemiology (56%), disease epidemiology (30%) and, to a lesser degree, drug utilisation, 2

28 28 Chapter pharmacoeconomics and environmental hazards. There have been over 250 publications in peer-reviewed journals using the GPRD [52-54]. PHARMO In the early 1990s, the PHARMO system of record linkage was developed in The Netherlands. PHARMO links community pharmacy and hospital data within a specific region on the basis of patient birth date, gender and GP code. The system now includes drug-dispensing records from community pharmacies and hospital discharge records of about 2 million people in the Netherlands. The data collection is longitudinal and goes back to More recently, PHARMO has also been linked to other data, such as primary care data, population surveys, laboratory and genetic data, cancer and accident registries, mortality data and economic outcomes. The system has well-defined denominator information that allows incidence and prevalence estimates and is relatively cheap because existing databases are used and linked. The PHARMO database is used for follow-up studies, case-control studies and other analytical epidemiological studies for evaluating drug-induced effects. In the past the database has been used for studies on drug utilisation, persistence with treatment, economic impact and ADRs [55,56]. Developments Pharmacovigilance and the methods used need to continue to develop in order to keep up with the demands of society. In recent years, three publications have been of utmost importance in terms of providing guidance on the future of pharmacovigilance. The first is the Erice Declaration on transparency, which was published in 1997 [57]. In this declaration, pharmacovigilance experts from all over the world, representing different sectors, emphasise the role of communication in drug safety with the following statements: Drug safety information must serve the health of the public Education in the appropriate use of drugs, including interpretation of safety information, is essential for the public at large, as well as for health care providers All the evidence needed to assess and understand risks and benefits must be openly available Every country needs a system with independent expertise to ensure that safety information on all available drugs is adequately collected, impartially evaluated and made accessible to all Innovation in drug safety monitoring needs to ensure that emerging problems are promptly recognised and efficiently dealt with, and that information and solutions are effectively communicated

29 Pharmacovigilance and intensive monitoring It is believed that these factors will help risks and benefits to be assessed, explained and acted upon openly and in a spirit that promotes general confidence and trust. This declaration was followed in 2007 by the Erice Manifesto for global reform of the safety of medicines in patient care [58]. The Erice Manifesto specifies the challenges which must be addressed to ensure the continuing development and usefulness of the science, in particular: The active involvement of patients and the public in the core debate about the risks and benefits of medicines, and in decisions about their own treatment and health The development of new ways of collecting, analyzing and communicating information about the safety and effectiveness of medicines; open discussion about it and the decisions which arise from it The pursuit of learning from other disciplines about how phamacovigilance methods can be improved, alongside wide-ranging professional, official and public collaboration The creation of purposeful, coordinated, worldwide support amongst politicians, officials, scientists, clinicians, patients and the general public, based on the demonstrable benefits of pharmacovigilance to public The third article that has had a profound impact on how pharmacovigilance should work in the future is the article published in 2002 by Waller and Evans in which they give their view on the future conduct of pharmacovigilance. The key values that should underpin pharmacovigilance are excellence (defined as the best possible result), the scientific method and transparency. The paper defines five elements that are considered to be essential for achieving excellence. Three of these are: process-oriented best evidence, robust scientific decision-making and effective tools to deliver protection of public health. The other two elements, scientific development and audit, underpin these processes, recognising that excellence cannot be achieved merely by process [59] International developments In the past, pharmacovigilance has been most concerned with finding new ADRs, but Waller and Evans suggest that pharmacovigilance should be less focused on finding harm and more focused on extending knowledge of safety [59]. In recent years, regulatory agencies have been reforming their systems in order to keep pace with the developments in pharmacovigilance, with the focus on being more pro-active Europe In 2005, a document was drafted by the Heads of the Medicines Agencies called Implementation of the Action Plan to Further Progress the European Risk Management Strategy. In July 2007, the EMEA published a document in which they discussed the achievements booked 2