Statistical Methods to Examine Racial and Ethnic Disparities in the Surgical Literature: A Review and Recommendations for Improvement

Speaker(s)

Sox-Harris A1, Eddington H1, Shah V2, Shwartz M3, Gurewich D3, Rosen AK3, Quinteros B1, Wilcher B4, Nieser K4, Jones G4, Wu J4, Morris A1
1Stanford University, Palo Alto, CA, USA, 2Stanford University, Stanford, CA, USA, 3Boston University, Boston, MA, USA, 4Veterans Affairs Palo Alto Health Care System, Menlo Park, CA, USA

OBJECTIVES: Scores of scientific papers are published each year describing racial and ethnic disparities in surgical access, quality and outcomes. However, the statistical rigor of this literature has never been systematically reviewed. Examinations of racial and ethnic disparities should statistically adjust for social determinants of health (SDOH) and other variables (e.g., age) that might be related to both race/ethnicity and outcomes. Studies using multi-institutional data should additionally account for clustering of patients with hospitals, and distinguish within- and between-hospital effects. Using these criteria, we evaluated the quality of statistical methods for studies of racial and ethnic disparities in published in the surgery literature during 2021-2022.

METHODS: We searched PubMed for surgical and perioperative studies focused on quantifying racial and/or ethnic disparities. Included studies were characterized in terms of three primary methodological criteria: 1) adjusting for SDOH (e.g. insurance status, education, income) and other variables that might be related to both race/ethnicity and outcomes; 2) accounting for clustering of patients with hospitals; and 3) distinguishing within- and between-hospital effects.

RESULTS: We identified 226 relevant papers. Of the 39 single-institution studies, only 24 (62%) adjusted for SDOH and other variables that might be related to race/ethnicity. Of the 187 multi-institution studies, only 36 (19%) adjusted for SDOH and other variables and addressed clustering of patients within hospital. Only 5 (<1%) of multi-institution studies examined how much of overall disparities were driven by within vs between hospital effects.

CONCLUSIONS: Most recently published papers in the surgical literature on racial and ethnic disparities do not meet basic statistical design criteria. Datasets without information on SDOH should not be used in studies of racial or ethnic disparities. For multi-institutional studies, clustering on patients within hospitals, as well as distinguishing between-hospital effects are essential, but cannot be executed in data without hospital identifiers(e.g., ACS-NSQIP).

Code

MSR119

Topic

Health Policy & Regulatory, Methodological & Statistical Research, Study Approaches

Topic Subcategory

Confounding, Selection Bias Correction, Causal Inference, Health Disparities & Equity, Literature Review & Synthesis

Disease

No Additional Disease & Conditions/Specialized Treatment Areas, Surgery