Purpose: We aimed to review systematically the published evidence regarding the effect of questionnaire design on the recall of pharmacological treatments.
Methods: The electronic databases Pubmed, EMBASE, and Cochrane Library were searched from inception to October 2007, using the following search terms: drug utilization, pharmaceutical preparations, pharmacoepidemiology, validation studies, methods, epidemiologic methods, interviews, data collection, and questionnaires. Drug utilization studies comparing different types of questionnaire or methods of questionnaire administration were included. Backward and forward citation tracking were also conducted.
Results: Eight studies were included in the systematic review, comparing questions asking for specific drugs or indications with open-ended questions (n = 5), evaluating the use of memory aids (n = 1), or studying the influence of response order on recall (n = 2). The studies were heterogeneous, namely regarding the populations evaluated (e.g., pregnant women, hypertensive patients, general population), mode of questionnaire administration (e.g., personal or telephone interview, self-administered), recall period (e.g., current use, 1 week, previous episode of a disease), or drugs evaluated (e.g., analgesics, antimalarials, all medicines). Despite the lack of standardization in presentation of results, the prevalence of drug use may vary between 5 and 40% when drug names and indications or pictures are used as memory aids, or as a result of primacy effects in self-administered questionnaires. The yielding of the questionnaires depended on the pharmacological groups evaluated.
Conclusions: Scientific work regarding methods for drug utilization data collection is scarce. The available evidence highlights the importance of knowing the questionnaire characteristics for a proper interpretation of results from drug utilization studies.
(c) 2009 John Wiley & Sons, Ltd.