Reporting quality of statistics in Indian journals: Analysis of articles over a period of two years

Abstract

Statistics has always been ignored by both clinical researchers as well as journal editors. We undertook this study to find out the quality of statistical reporting in some of the Indian journals. A sample of eight Indian medical and surgical journals that publish original articles ‑ Journal of Association of Physicians of India, Indian Pediatrics, Indian Journal of Surgery, Indian Journal of Orthopedics, Indian Journal of Ophthalmology, Neurology India, Indian Journal of Dermatology, National Medical Journal of India were chosen for the study. Only original studies were considered for analysis and were appraised for quality of reporting of descriptive statistics and inferential statistics, sample size calculation, and correction for multiplicity of statistical tests. A total of 415 original articles were identified during the specified period from all the eight journals. Major findings include no mention about sample size calculation (199/415, 48%) and normality check a priori to the use of parametric tests (140/415, 33.7%), use of multiple statistical tests (ranging between 14 and 126 times) without adjusting the P value (116/415, 28%), and no mention of confidence intervals at least for the primary endpoint (85/415, 20.5%). To conclude, we found poor quality of statistical reporting in some of the Indian journals and considering the seriousness of this issue, it is high time for both the researchers and journal editors to consult statistical experts, while the study is being carried out and at the time of publishing the same, respectively.