The recent article by Cramer (1) aroused our full interest. The author reports a systematic literature search performed to identify articles with quantitative data on adherence to oral hypoglycemic agents and insulin and correlations between adherence rates and glycemic control. Abstracts identified by searching Medline, Current Contents, Health & Psychological Instruments, and Cochrane Collaborative databases were screened. The systematic search resulted in 20 publications, i.e., 15 retrospective and 5 prospective studies, with adequate data for synthesis. We would like to suggest that from our experience, the Embase database would have probably added relevant, high-quality articles that do not appear in these databases.

In searching the literature on interventions for improving adherence to treatment recommendations in people with type 2 diabetes (2), we evaluated the input from different databases. Exhaustive searches were performed with peer-reviewed search strategies in the following databases: the Cochrane Library (including the Cochrane Controlled Trials Register, the Database of Reviews of Effectiveness, and the National Health Service Health Economic database), the Metabolic and Endocrine Disorders Group specialized register, Medline, Embase, Psychinfo, Eric, Dissertation and Sociological Abstracts, Cinahl, and the metaregister of controlled trials. Studies were included when interventions aimed at improving the adherence to treatment recommendations in people affected by type 2 diabetes.

Overall, we retrieved 3,210 publications, mainly originating from three core databases: Medline, Embase, and the Cochrane Collaboration databases. Two teams of independent researchers read all abstracts. When insufficient information was available to evaluate the study, the full article was retrieved. A total of 35 prospective studies were included.

We were eager to know which databases were able to identify the majority of the included articles and what the overlap could be between different databases. Of 1,684 Medline hits, 12 articles (0.71%) were included for data extraction. Embase resulted in 21 articles out of 1,165 hits (1.80%), and the Cochrane Library resulted in 17 articles out of 341 hits (4.99%). Five articles were cited by all three databases; 21of 33 articles were cited only once. Of these, 10 were located by Embase, 8 by Cochrane, and only 3 by Medline. If we had only identified articles from Medline, our search would have identified 12 of 33 articles (36.4%). If we had searched only Embase, the outcome would have been 21 (63.6%), and, if we had searched only the Cochrane database, the result would have been 17 (51.5%).

Our key message is that searching in all relevant databases is a must for what is called a systematic literature search. Electronic databases enhance the accessibility of evidence. This supports researchers in being comprehensive. Free access to all databases for research purposes could even improve the outcome.

Being systematic means being exhaustive, which means being as complete as possible, resulting in a harvest of all available research evidence on the topic of interest (3). A partial approach, being systematic without searching all databases, could be synonymous to bias. Is not bias exactly what a researcher would want to avoid in being systematic?

Cramer JA: A systematic review of adherence with medications for diabetes.
Diabetes Care
Vermeire E, Wens J, Van Royen P, Hearnshaw H: Interventions for improving adherence treatment recommendations in people with type 2 diabetes mellitus (protocol for a Cochrane Review). In
Cochrane Database of Systemic Reviews
. Issue 2,
. Available from
Egger M, Juni P, Barlett C, Holenstein F, Sterne J: How important are comprehensive literature searches and the assessment of trial quality in systematic reviews? Empirical study.
Health Technol Assess