gms | German Medical Science

Background/research question: For evidence synthesis, electronic database retrieval is a core component of systematic and comprehensive literature searching as indicated by current methodological guidance. Such databases cover different health topics and references but also show overlaps. With regard to evidence synthesis in dementia care research, it is unclear which combination of databases might ensure optimal research efficiency (i.e. retrieving most of the eligible references by using a minimum number of databases). We aimed therefore to identify the optimal database combination for evidence synthesis using a dementia-specific research question.

Methods: Based on a previously conducted scoping review on facilitators and barriers to implementing nurse-led interventions in dementia care using a comprehensive literature search in eight databases (CENTRAL, CINAHL, Embase, Emcare, MEDLINE, Ovid Nursing Database, PsycINFO, and Web of Science Core Collection) and citation tracking, we retrospectively analysed database overlap and unique references using cross tables and descriptive statistics. Additionally, we analysed similarity of databases using a multiple correspondence analysis (MCA). MCA is a descriptive data analysis technique that simplifies the presentation of complex data.

Results: Our search in eight databases and citation tracking of included studies retrieved 10,527 references. Of these, 6,944 were duplicates and 3,583 were unique references. Searching CINAHL, MEDLINE, and Web of Science as well as conducting backward and forward citation tracking was necessary to identify all the included studies of our final sample. MCA resulted in considerable overlaps in some databases using our search strategies such as CINAHL, PsycINFO, and Emcare.

Conclusion: Searching CINAHL, MEDLINE, and Web of Science and to use citation tracking might be essential to identify studies for evidence synthesis in dementia care research. Additional screening of studies retrieved by other databases might therefore be avoided. Since our study represents one case, the generalisability of the results is limited. Further methodological research on database coverage and overlaps is required to confirm our findings, e.g. as part of dementia-related systematic reviews. Thus, more certainty about optimal database retrieval in dementia care research could be achieved which might be helpful to wisely choose databases for efficient evidence synthesis methods.

Competing interests: All authors declare that there is no competing interest.