Drug treatment episodes in pharmacoepidemiology - antidepressant use as a model

Abstract

In the Netherlands, antidepressants are indicated for treating depression, generalized anxiety disorders, obsessive-compulsive disorders, social phobia, panic disorders, eating disorders, neuropathic pain and nocturnal enuresis. In addition, antidepressants are sometimes used for treating off-label indications such as sleeping disorders, urinary incontinence and headache. The diversity in the nature of these conditions results in a variety of antidepressant treatment patterns. The common use of antidepressants in the general population, in addition to the fact that their treatment pattern does not always represent the traditional episodic nature of depression, makes this particular drug class a suitable model for methodological research on drug exposure. In pharmacoepidemiology the effects of drug use in large populations are studied. Currently, pharmacoepidemiological research often uses prescribing and dispensing moments to construct drug treatment episodes in order to classify drug exposure. Construction of drug treatment episodes involve many methodological aspects, such as defining the start and the end of a treatment episode, which need to be accounted for. The objective of this thesis was to investigate methodological topics in observational research relevant to the construction of antidepressant treatment episode(s). When designing observational studies and deciding which definitions should be used to compose a study cohort and construct treatment episodes, there are some basic considerations that will influence the choice of definitions. Firstly, does the study question require knowledge on disease status of the patient or only drug use as such? Secondly, does the study aim to investigate patients who receive a prescription, regardless of whether they start therapy or not? Are there possible differences in risk profiles between patients who are experiencing their first treatment episode and those who re-initiate treatment? And finally, is the drug intended for short, episodic or long term use and what is its daily dosing scheme? When the aim is to investigate drug exposure in relation to particular outcomes, it is essential that drug treatment episodes are well defined. The methodological choices for study definitions depend on the exposure-outcome relationships under investigation. If definitions are not selected properly, it could influence study outcomes and lead to biased estimates. Nowadays, the majority of pharmacoepidemiologic research uses administrative databases to define drug treatment patterns which are subsequently associated with specific (disease) outcomes. The observed treatment patterns which are used to divide patients into different groups are often based on assumptions of the patients’ specific treatment patterns. Future research should investigate patient behavior in relation to specific treatment patterns seen in administrative databases. Thereby, we would not only describe the treatment patterns and drug taking behavior of the patient but investigate the reasons and understand why the patient exerts a specific treatment pattern. Many factors can influence patient decision making on whether and how they use their medicines and if they comply with therapy or not. These factors can differ between patients and also within a single patient over the course of time. Understanding what kind of patient behavior results in a specific treatment pattern observed in administrative databases may lead to new definitions when investigating adherence in administrative data.