P010: An examination of sample size selection in medical record reviews in emergency medicine journals

Abstract

Introduction: Medical record review (MRR) studies are commonly used in Emergency Medicine (EM) research. It is not always clear how sample size calculations are reported, or the methods by which they were derived. This scoping review sought to examine reporting and justification of MRR sample sizes from the EM literature. Methods: Using Web of Science, we identified the top ten journals, based on impact factor rating in 2018, within the field of Emergency Medicine. Journals were excluded if they were not in English or did not include sufficient articles for analysis. Within each of these ten selected journals, we searched for chart reviews and related terms: "medical record", "outpatient record", "inpatient record", "clinical record", and "nursing note". From this search subset, five articles were randomly selected from each journal. Data about sample size and sample size selection were extracted and analyzed by two reviewers independently for each article. Results: Of the 50 articles randomly selected, 48 articles were retrospective MRRs and two articles were prospective MRRs. 78% (39 articles) chose sample size based on availability, 14% (seven articles) chose sample size based on power calculations, 4% (two articles) chose sample size based on a previous study's methodology, and 4% (two articles) did not give details on sample size selection. Conclusion: While some emergency medicine MRRs based sample size selection on power or previous studies, the vast majority are based on availability with study-specific exclusion/inclusion criteria. This may indicate they are using a smaller sample size than necessary to be sufficiently powered to assess their end goal. More work is required to determine the effect of this on outcomes and interpretability of results, as well as which method is most accurate and efficient.