A Mixed-methods Study to Explore Statistical Quality and Utilization of Research Articles Published in Myanmar Health Sciences Research Journal (2010-2019)

Abstract

Aim: The study was carried out to collate all evidences in the study design and statistical analyses used in the research articles in a peer-reviewed health research journal widely published in Myanmar. The perspectives of the end-users were also studied to improve dissemination of the research findings.

Study Design: Mixed Methods Research Design.

Methodology: A standardised checklist was used to examine the study designs and statistical methodologies utilised in all research papers (n=391) published in the Myanmar Health Sciences Research Journal (MHSRJ) from 2010 to 2019. In-depth interviews were carried out on the end-users, to determine the application of the research findings and to find ways to improve the accessibility, and distribution mechanisms.

Results: Among the publications, commonly used study designs were descriptive (cross-sectional) studies (n=191,49%) and basic science (laboratory-based) studies (n=129, 33%). The remaining were analytical observational studies (n=49, 12%) and clinical trials (n=22, 6%). In 206 studies, parametric data analysis methods were applied (64 percent). Basic statistics, contingency table analyses, epidemiological statistics, t-test, correlation, and regression analysis were the most commonly employed statistical methods, accounting for 77 percent of the statistical tests. Although the sampling unit and study area/site were completely stated in over 92 percent (n=363) of the papers, just a few (16 percent, n=57) indicated the sample size computation. Utilization of improper study design and/or statistical methods, failures to mention the underlying assumption for using statistical methods were noted. It was also found that although the research findings were useful, the circulation of the Journal was limited.

Conclusion: According to our research, a reader with knowledge on descriptive statistics, analysis of contingency tables, epidemiological methods, t - tests and regression analysis will be able to understand 77% of the published articles in the Journal. For the quality of papers to improve, flaws in the research design and statistical analysis must be addressed. Use of social media networks could expand the dissemination of the research findings published in the Journal.