An agenda for research on adverse drug reactions

After a gestation period of 2 years, a new Adverse Effects Methods Group has recently been registered as part of the Cochrane Collaboration (http://www.cochrane.org/contact/mwgfield.htm). This is the culmination of a process that started with a meeting of interested parties in Oxford in August 2005 and was followed by informal meetings of a working party and eventually an open meeting at the Dublin Colloquium of the Collaboration in October 2006. The main objectives of the Group are to develop and implement methods for systematic reviews of the adverse effects of therapeutic interventions. Although adverse drug reactions will form a major thrust of this activity, all therapeutic interventions, including devices and surgical interventions, come within the Group's remit. The co-conveners are Su Golder (spg3@york.ac.uk), Andrew Herxheimer (a.herxheimer@ntlworld.com), and Yoon Loke (y.loke@uea.ac.uk); anyone who wants to participate, either actively or passively, should get in touch with them.

Part of the remit of the working party was to prepare a plan for the work of the Group, including formulation of policy and research proposals. This plan was to relate to systematic reviews, but during our discussions it emerged that it would be useful to formalize an agenda for general types of research on adverse drug reactions. Such an agenda is illustrated diagrammatically in Figure 1 and examples of possible research projects are listed in Table 1[1–9]. I invite comments on this agenda.

Figure 1.

Potential headline topics for research in adverse drug reactions (ADRs); 1 + 2 + 3 = the DoTS classification of adverse drug reactions [1], which applies to ADRs at all stages of detection, not merely during the preclinical phase, as shown

Table 1.

Examples of potential research projects in adverse drug reactions (ADRs)

Heading	Examples
Pharmacology/toxicology	1.Constructing a database of published dose-response curves in ADRs (in vitro and in vivo):
	•in order to analyse their characteristics (e.g. maximal efficacy, slope)
	•the use of in vitro data to predict in vivo outcomes
	2.Defining the time-courses of ADRs
	3.Defining susceptibility factors in patients
	4.Developing and using biomarkers of adverse effects
	5.Effects of overdose and methods of management [1 + 2 + 3 → DoTS classification of adverse drug reactions [1]]
Regulation	How drug regulatory decisions should be made
	•licensing decisions
	•withdrawing drugs after llicensing
	Naming medicines
Marketing	How information about ADRs is disseminated
	•Summaries of Product Characteristics (SPCs)
	•advertising
Communication & education	Methods of communicating with health-care professionals [2] and the public:
	•the nature of risk
	•perceptions of risks [3]
	•the actual risks and their relevance to clinical practice
	•the nature of the balance of benefit and harm
Study design	Reporting methods Which types of design are best for eliciting particular types of ADR (e.g. cohort studies, case-control studies, n-of-1 studies, RCTs)
Analysis	Teleoanalysis – how best to combine information from many different types of evidence (RCTs, observational studies, case series, anecdotes) When different types of evidence are relevant Data mining techniques [4]
Terminology & classification	Developing classification systems [5] Dictionaries – using standard terminology [6]
Publication	Developing CONSORT-like statements [7]
Review	The use of anecdotes and spontaneous reports [8] Systematic review (including narrative review, meta-analysis, and teleoanalysis) Methods of searching for published information [9] Indexing databases

Adverse drug reactions are among the constant preoccupations of clinical pharmacologists and commonly feature in the British Journal of Clinical Pharmacology[10]. This issue is no exception, and papers on adverse reactions that it contains illustrate different aspects of the research agenda – susceptibility factors, mechanisms of adverse reactions, and perceptions of risks and the effects of education.

The need to elucidate individual susceptibility factors is illustrated by a study of dihydropyrimidine dehydrogenase, the rate-limiting enzyme in the catabolism of pyrimidines, such as fluorouracil [11]. Deficiency of dihydropyrimidine dehydrogenase is a susceptibility factor for 5-fluorouracil toxicity. However, it is not known if polymorphisms in the gene for the enzyme influence the risk in individuals with normal enzyme activity. In 131 patients with fluorouracil toxicity dihydropyrimidine dehydrogenase activity was significantly lower than in 185 unselected patients with cancers, but there was extensive overlap between the groups. Furthermore, only two of 93 patients screened carried the IVS14 + 1G > A mutation; although both had had severe toxicity, so had four other patients without the mutation. Since dihydropyrimidine dehydrogenase deficiency explains only part of the susceptibility to fluorouracil, the search for other susceptibility factors will have to continue.

Elucidation of the mechanisms of adverse effects can be important in determining methods of management. Overdosage of venlafaxine can cause cardiac arrhythmias through mechanisms that are unclear. In a retrospective study of 235 young adults described in this issue of the Journal there was a weak but significant relation between the reported dose and the QT_c interval [12]. However, there were only three cases of arrhythmias and none of torsade de pointes. The authors suggested that the effects on heart rate and blood pressure were consistent with noradrenergic potentiation and also speculated about a possible effect of venlafaxine on sodium channels. However, they did not speculate about possible methods of management.

Patients do not always have the same perceptions about the risks of using drugs as health professionals, and there are also differences among the health professionals themselves [3]. This has been illustrated in a study of French medical students reported in this issue of the Journal, as has the effect of education [13]. Before a pharmacology course the students ranked hypnotics as the most dangerous drugs, followed by antidepressants and anticoagulants, similar to the perception of non-health professionals [3]. After the course the order changed to antidepressants, anticoagulants, and hypnotics. Their perceptions of the risks of other drugs also changed. These results reinforce, if reinforcement were necessary, the importance of education in improving prescribing [14].

Finally, a completely different type of adverse effect features in a commentary in this issue of the Journal by Dear & Webb [15]. It is the adverse effect on prescribing of the phenomenon that has been christened ‘disease mongering’, a form of medicalization [16] that was the subject of a series of articles in PLoS Medicine last year. The term seems to have been introduced by the American journalist Lynn Payer [17] and popularized by, among others, the Australian journalist Ray Moynihan [18, 19], although the related term ‘nostrum-monger’ was in use as early as the beginning of the 18^th century. Payer defined disease mongering as ‘trying to convince essentially well people that they are sick, or slightly sick people that they are very ill’. Examples include the conversion of shyness into Social Anxiety Disorder, premenstrual symptoms into Premenstrual Dysphoric Disorder, and various sexual difficulties that some women experience, such as failure to achieve regular orgasms, into Female Sexual Dysfunction. The marketing of these conditions by pharmaceutical companies leads to unnecessary prescribing and potential adverse effects, and Dean & Webb suggest that doctors and patients are complicit in the process.

Several other aspects of research into adverse drug reactions were included in the special issue of the Journal that appeared earlier this year [10]. It is to be hoped that the activities of the new Cochrane Adverse Effects Methods Group will lead to more material for similar issues in the future.