The Caprini Score in Black and Latino Patients: A Proposed Race- and Ethnicity-Adjusted Model for Perioperative Venous Thromboembolism Risk Assessment

Background: Perioperative venous thromboembolism (VTE) affects up to 50K US pts/yr, causing morbidity and mortality. The Caprini risk assessment model (RAM; 2005), a validated scoring system for assessing perioperative VTE risk, does not account for race (PMID: 15900257)). Moreover, population studies of Black and Latino compared to White individuals indicate variation in VTE incidence and risk of developing provoked VTE (PMID: 19303496). We hypothesized this RAM inaccurately estimates perioperative VTE risk in non-White pts and that a race-specific adjustment is required.

Methods: We retrospectively identified 32,682 adult surgical inpts admitted at an urban academic medical center 2010-2023. VTE events were identified by ICD codes for new acute PE or DVT during hospitalization. Using demographic, hospitalization, laboratory, and surgical procedure data, we calculated Caprini scores and assigned previously validated Caprini risk strata at score cutoffs of 0–2, 3–4, 5–6, 7–8, and >8, with the 0–2 group defined as baseline. We generated separate race/ethnicity-adjusted (RE) Caprini scores by adding or subtracting points for Black race or Latino ethnicity, comparing performance to the original Caprini.

Additionally, using census tract data from American Community Survey, we examined impact of various SDOH measures on perioperative VTE risk; neighborhood poverty rate (% of residents in families with income<federal poverty level; POVR) was selected, based on model performance. Nested multivariate logistic regression models were used to estimate VTE odds ratio (OR), using Caprini risk strata, race/ethnicity, and POVR as predictors.

Models were compared using Akaike Information Criterion. Statistical testing used Breslow-Day and Mantel-Haenszel tests, with analyses on R (v4.4.1).

Results: Median age was 54 y (IQR 42–64 y), 55% female, 47% non-Latino Black (n=15,273), 25% Latino (n=8170), 25% non-Latino White. Overall hospital-associated VTE incidence was 4.4% (n=1427 events), with median Caprini score 5 (IQR 3–7). But, in Black and Latino pts, VTE incidence was 4.8% (n=727) and 3.4% (n=284), with median Caprini scores 5 (IQR 4–7) and 5 (IQR 3–7), respectively; indicating significantly increased (Black; OR 1.12, 95% CI 1.01–1.24, p=0.03), and decreased (Latino; OR 0.75, 95% CI 0.66–0.86, p<0.0001) odds of VTE, despite adjusting for Caprini RAM risk category.

Adding 1 point for Black race optimally corrected race-based discrepancies in the original Caprini RAM (VTE OR 0.95, 95% CI 0.85–1.06, p=0.38), as did subtraction of 2 points for Latino ethnicity (VTE OR 1.04, 95% CI 0.90–1.20, p=0.58), yet appropriate VTE risk stratification was maintained (Black pts: OR 1.4 [Caprini score 3–4; p=0.03], 2.3 [5–6; p<0.001], 2.5 [7–8; p<0.001], 5.2 [>8; p<0.001]), with comparable results in Latino pts.

Of clinical relevance, RE-Caprini upgraded 1326 (8.7%) Black pts into the medium-risk VTE category (3-4), which necessitates use of chemoprophylaxis at our institution, and downgraded 2238 (27.1%) Latino pts, who would no longer require chemoprophylaxis.

Interestingly, adjusting for race/ethnicity and Caprini score, POVR is associated with VTE risk (OR per 10% POVR increment=1.05, 95% CI 1.00–1.10, p=0.05) that on subgroup analysis is significant in Black pts only (OR/10% increment = 1.07, 95% CI 1.01–1.14, p=0.015), and partially accounts for their increased VTE risk (OR 1.06, 95% CI 0.93–1.20, p=0.36). Median POVR for Black and non-Black patients were 25.6 and 19.9%, respectively.

Conclusion: Caprini RAM underestimates perioperative VTE risk in Black and overestimates risk in Latino pts. Simple race-specific point adjustments enhance predictive accuracy, implementation of which results in upgrading of considerable numbers of Black pts into the medium-risk category where use of chemoprophylaxis reduces VTE risk by 57% (PMID: 22315263), and downgrading 1 in 4 Latino pts into the low-risk category to reduce perioperative bleeding risk.

Validation of these results is underway at another large academic site. The contribution of SDOH to increased perioperative VTE risk specifically in Black pts has not been characterized, and may suggest a role for stress-induced inflammation related to allostatic load in VTE pathogenesis, a focus of further studies. In summary, use of race/ethnicity-specific factors in risk assessment may optimize outcomes and mitigate disparities in VTE prevention.