The impact of social determinants of health (SDOH) on completing bariatric surgery at a single academic institution.

Background: Underutilization of bariatric surgery is multifactorial. This study aimed to understand the association of SDOH on not achieving surgery.

Methods: 1081 applications for primary MBS from January-December 2021 were stratified into those that completed surgery (COM; n = 415), in progress > 1-year (IP; n = 107), dropped out (DO; n = 379), and never started (NS; n = 180). Using the American-Community-Survey results (2015-2020) and patient zip-codes, population differences in 4-domains of SDOH (demographic/social/housing/economic) were examined between COM versus the other groups. Additionally, using institutional MBSAQIP and EMR data, patient-specific differences in comorbidities were evaluated for COM versus IP/DO. Univariate analysis using Kruskal-Wallis, chi-squared/Fisher's exact tests were used for continuous and/or categorical variables. For patient-level analysis multinomial logistic regression was used to determine predictors of not achieving surgery. Hypothesis testing was conducted at an overall 5 percent type-I error rate (alpha = 0.05) and Bonferroni's method was used to adjust for multiple comparisons.

Results: Compared to COM, IP-patients resided in zip-codes characterized by fewer married people (43% vs 46%; p = 0.019), lower education levels (49% vs 43%; p = 0.048), more households where rent was > 50% of household income (10% vs 8%, p = 0.002), and households below the poverty line (17.6% vs 14.5%, p = 0.017). At the patient-level, IP were more likely to be male (27.9% vs 14.9%; p = 0.014), publicly insured (44.9% vs 28.4%; p = 0.004), Black (35.5% vs 22.2%; p = 0.006), an active smoker (8.9% vs 2.2%; p = 0.018), have a higher BMI (49.6 vs 47.6; p = 0.01), and coronary intervention (5.8% vs 1.7%, p = 0.034). Comparison of COM vs DO was similar for both phases. Multinomial multivariable logistic regression demonstrated higher BMI (OR = 1.03,[CI]:1.01-1.05, p = 0.001), males (OR = 1.9,[CI]:1.09-3.32, p = 0.024), smoking (OR = 4.58,[CI]:1.74-12.02, p = 0.002), and Medicaid (OR = 2.16,[CI]:1.33-3.49, p = 0.002) independently predicted not achieving surgery.

Conclusion: Patient-level data demonstrated social not clinical factors predicted surgery completion. Given zip-codes characterizing the IP/DO groups had a greater prevalence of social risk, more attention needs to be directed patient-level social risks.