Strategies for quantifying the relationship between medications and suicidal behaviour: what has been learned?

In recent years there has been considerable concern that certain classes of drugs, for example antidepressants, may increase the risk of suicide. In this current opinion article, we examine the literature on methodological and statistical approaches to the design and analysis of suicidal event studies. Experimental, ecological and observational studies of the relationship between drugs and suicidal events (thoughts, attempts and completion) are discussed. Areas considered include analysis of spontaneous reporting system data, ecological trends in national and/or small area (e.g. county) suicide rates, meta-analyses of randomized clinical trials, and large-scale medical claims data. New statistical and experimental strategies for investigating possible associations between drugs and suicide are highlighted, and we suggest directions for future statistical/methodological research. To put this into context, we then review the most recent literature on the relationship between drugs (antidepressants, antiepileptics, varenicline, montelukast and antipsychotics) and suicidal events. Overall, there appears to be little evidence that drugs increase the risk of suicide and related behaviour. Numerous lines of evidence in adults clearly demonstrate that inadequate treatment of depression (pharmacotherapy and/or psychotherapy) is associated with increased risk of suicidal behaviour. In children, the results are less clear and further study is required to better delineate which children benefit from treatment and who may be at increased risk as a consequence of treatment. From a statistical and methodological perspective, the field of pharmacoepidemiology is a fertile area for statistical research, both in theory and in application. In general, methods have been adopted from other areas such as general epidemiology, despite the singular nature of many of the problems that are unique to drug safety in general, in particular the study of rare events. Finally, there is considerable debate concerning the communication of risk. For suicide, regulatory action has been taken largely on the basis of evidence suggesting increased risk of suicidal thoughts. However, suicidal thoughts are quite common, particularly among patients with depression, and may have little relationship to suicidal behaviour and/or completion.