Methods
Literature Search and Selection Criteria
A systematic search for relevant publications was conducted in bibliographic databases (Medline via EMBASE, EMBASE via EMBASE, CENTRAL via Cochrane Library) in March, 2013. A search strategy for each database was developed using medical subject headings and key words for adherence, pharmacist interventions and type 2 diabetes mellitus. The full search strategies are provided in Additional file 1 http://www.biomedcentral.com/1472-6823/14/53/additional. Randomized controlled/cluster-randomized controlled trials as full-text publications investigating pharmacist interventions in which a pharmacist is involved in the provision of the intervention to improve adherence, defined as the degree to which a patient follows the medical prescription in terms of interval and dose of a dosing regimen, to oral medication in type 2 diabetes mellitus were eligible for inclusion. If the type 2 diabetes mellitus medication could not be clearly classified as oral medication (e.g. metformin, alpha-glucosidase inhibitors, thiazolidinediones) the study was excluded. Moreover, the examined population had to consist of adult patients (≥18 years) and adherence to the oral medication in type 2 diabetes mellitus had to be measured. No limitation regarding the language or publication year of the studies was made.
Study Selection
Two independent reviewers screened the titles and abstracts of the identified publications according to the pre-defined criteria. After obtaining the full-texts of the potentially relevant publications two independent reviewers screened them and determined their eligibility for further analysis. If discrepancies regarding study inclusion could not be solved by discussion a third reviewer was involved.
Data Extraction
The results and study characteristics of each included study were then extracted and a second reviewer checked for accuracy and completeness. For this, standardized tables were used. These contained information on the first author, the publication year, the study type, the country and setting the trial took place, the study population size, age and sex as well as the content and length of the intervention and control intervention, the definition of adherence, the adherence measures and the adherence rate at baseline and last follow-up.
Risk of Bias
The risk of bias in the included studies was assessed by two independent reviewers according to pre-defined criteria based on the Cochrane risk of bias tool. However, the criteria related to blinding of participants and personnel were not applicable. In adherence-enhancing interventions participants and the personnel delivering the intervention cannot be blinded due to the nature of the interventions. Thus, the respective criteria were not considered. Consequently, the criteria implemented to assess the methodological quality of the included studies consisted of questions related to the random sequence generation, the allocation concealment, blinding of outcome assessment, the analysis according to intention-to-treat, selective reporting and other sources of bias. If discrepancies regarding quality assessment could not be solved by discussion a third reviewer was involved. It was decided to rate each risk of bias item only as "yes" and "no" and not as "unclear" as recent research suggests that rating as "unclear" "becomes the default for the risk of bias tool assessments (RoB) regarding reliability". Moreover, it could be shown that a "significant difference in effect sizes […} between studies with a high or unclear risk of bias and those with a low risk of bias" exists.