Should CINAHL be used as one of the main databases for evidence synthesis of health services intervention?

Introduction

CINAHL is not listed as one of the minimum databases for systematic review (SR) of interventions in the Methodological Expectations of the Cochrane Intervention Review.

Objective

To determine additional studies uniquely identified from the CINAHL search in SR of health services interventions (HSI).

Methods

We searched PubMed from inception to October 1, 2022 to identify SRs of HSI that determined clinical or humanistic outcomes of HSI and used CINAHL. Out of 5655 Systematic reviews identified, we randomly selected 374 SRs and extracted all primary studies included. We then explored the bibliographic databases in which the journals of those studies were indexed. The outcome of interest was the number of studies uniquely available in CINHAL. We also performed a subgroup analysis based on the type of HSI. We performed descriptive statistics to report the study outcomes using Excel (Microsoft 365).

Results

A total of 7550 primary studies were identified from the 374 Systematic reviews that met the inclusion criteria. Of these studies, 7380 were journal publications that have been indexed in MEDLINE/PubMed (75.1%), Scopus (74.5%), Sciences Citation Index, SCI (54.7%), Embase (48.1%), and CINAHL (34.9%). Only 83 out of 7380 (1.1%) studies were published in journals that were uniquely indexed in CINAHL. The percentage of studies that were only available in other databases was 9.7% (Scopus), 4.3% (MEDLINE/PubMed), 1.6% (SCI), and 0.3% (Embase). The number of studies that were unique to CINAHL in specific types of HSI were: 24/1570 (1.5%) for community health services, 20/1520 (1.3%) for preventive health services, 45/3624 (1.2%) for patient care, 8/1173 (0.7%) for mental health services, and 18/2804 (0.6%) for rehabilitation.

Conclusion

The gain of CINAHL to identify unique primary studies for SR of HSI appears minimal. The impact of missing studies uniquely available in CINAHL on SR summary or magnitude of effect estimates from meta-analysis requires further investigation.