Social Vulnerability Subtheme Analysis Improves Perioperative Risk Stratification in Hepatopancreatic Surgery

There has been increased interest in understanding how social determinants of health (SDH) may affect care both in the medical and surgical setting. We sought to define the impact of various aspects of social vulnerability on the ability of patients to achieve a "textbook outcome" (TO) following hepatopancreatic surgery.

Medicare beneficiaries who underwent hepatopancreatic resection between 2013 and 2017 were identified using the Medicare database. Social vulnerability was defined using the Centers for Disease Control Social Vulnerability Index (SVI), which is comprised of four subthemes: socioeconomic (SE), household composition and disability (HCD), minority status and language (MSL), and housing type and transportation (HTT). TO was defined as the composite endpoint: absence of 90-day mortality or readmission, absence of an extended length of stay (LOS), and no complications during the index admission. Cluster analysis was used to identify vulnerability cohorts, and multivariable logistic regression was utilized to assess the impact of these SVI subthemes on the likelihood to achieve a textbook outcome.

Among 37,707 Medicare beneficiaries, 64.9% (n = 24,462) of patients underwent pancreatic resection while 35.1% (n = 13,245) underwent hepatic resection. Median patient age was 72 years (IQR: 68-77), just over one-half were male (51.9%; n = 19,558), and the median CCI was 3 (IQR: 2-8). Cluster analysis revealed five distinct SVI profiles with wide variability in the distribution of SVI subthemes, ranging from 15 (profile 1 IQR: 7-26) to 83 (profile 5 IQR: 66-93). The five profiles were grouped into 3 categories based on median composite SVI: "low vulnerability" (profile 1), "average vulnerability" (profiles 2 and 3), or "high vulnerability" (profiles 4 and 5). The rate of TO ranged from 44.6% in profile 5 (n = 4022) to 49.2% in profile 1 (n = 4836). Multivariable analyses comparing patients categorized into the two average SVI profiles revealed that despite having similar composite SVI scores, the risk of adverse postoperative outcomes was not similar. Specifically, patients from profile 5 had lower odds of achieving a TO (OR 0.89, 95%CI: 0.83-0.95) and higher odds of 90-day mortality (OR 1.29, 95%CI: 1.15-1.44) versus patients in profile 4.

Distinct profiles of SVI subtheme characteristics were independently associated with postoperative outcomes among Medicare beneficiaries undergoing HP surgery, even among patients with similar overall composite SVI scores.