Racial and ethnic differences in diagnosis, healthcare utilization and 1-year outcomes for patients with significant tricuspid regurgitation

This study used Optum’s de-identified Market Clarity Data (years 2007–2023). The data source is one of the largest repositories of longitudinal data from real-world patients in the US, containing medical and pharmacy claims from a large national payer and additional third-party payers, linked with clinical data from EHRs [13]. Patient data used to perform this analysis were de-identified and accessed in compliance with the Health Insurance Portability and Accountability Act. Institutional review board approval was not required for this study due to its retrospective, observational and anonymized nature. To categorize patients’ race and ethnicity, Market Clarity incorporates patient self-reported values as captured in the EHR.

The analysis was conducted in a population of adults aged 50 years and above with TR who developed sTR between January 2018 and December 2023. As in the related analysis [12], sTR was defined by criteria generated in consultation with expert physicians; patients with either severe, massive or torrential TR in physician notes, or a TR diagnosis (International Classification of Diseases, 10th revision [ICD-10]: I36.1 non-rheumatic tricuspid valve insufficiency) and subsequent development of at least three of seven conditions – ascites, edema, liver disease, chronic kidney disease, extreme fatigue, weight loss and/or right ventricular dysfunction were classified as having sTR. Continuous health plan enrollment was required for the 12 months prior to patients meeting the definition of sTR (year -1) to give adequate time to capture comorbidities and TR-related symptoms. For year -2 and year -3, patients were included if they had continuous enrollment for the full year. Patients were also required to be alive and enrolled for at least 1 day after meeting criteria for sTR. Patients with missing race or ethnicity information, or those from categories with very low counts in the database (e.g., Asian) were excluded from the analysis.

Patients were grouped by race and ethnicity using three mutually exclusive categories: Hispanic (regardless of concurrent race category), White non-Hispanic and Black non-Hispanic. Baseline characteristics were assessed at the time of meeting the sTR definition. Comorbidities were captured using ICD-10 codes as previously described [12]. To give an assessment of overall comorbidity burden, the Elixhauser comorbidity index was computed according to established procedures [14]. Inpatient status at the time of sTR identification was also captured.

The primary outcomes were all-cause death and heart failure hospitalization during the year after identification of sTR. Mortality data are captured from EHR and claims data sources and supplemented with external data sources including the Social Security Death Master File, Centers for Medicare and Medicaid services data and obituaries. Linkages to external mortality data are incorporated into the patient record by the database provider for high degree of confidence matches. In addition to the two main outcomes, we examined measures of healthcare utilization both 3 years before and 1 year after sTR: urgent or emergent hospital admissions, emergency department visits and outpatient cardiac specialist visits.

Black and Hispanic patient groups were compared with the White group on all baseline characteristics using t-tests for continuous variables and chi-square tests for categorical variables. Categorical variables are reported as counts and percentages and continuous variables as means and standard deviations (SD). Survival analysis curves were calculated for 1-year all-cause death (Kaplan–Meier) and heart failure hospitalization (cumulative incidence function for a competing risk of death). Unadjusted and adjusted hazard ratios comparing Black and Hispanic to White patients were calculated for all 1-year events with hazard models (Cox proportional or Fine–Gray subdistribution, as appropriate). Model adjustment included all baseline characteristics included in Table 1. Finally, baseline characteristics were examined as explanatory variables for all-cause death and heart failure hospitalization using Cox proportional and Fine–Gray subdistribution hazard models respectively. Eight explanatory variable models were run – four each for all-cause death and heart failure hospitalization (overall, White, Black, Hispanic). Values of p less than 0.05 were considered significant. All statistics were calculated in R.

Across race/ethnicity groups, the majority of patients were White (79.8%), female (54.4%), Medicare beneficiaries (70%) and from the Midwest (53.7%). Mean patient age was 72.5 years (SD 10.5 years), and mean Elixhauser comorbidity score was 9.5 (SD 3.9). Comorbidity burden increased over time, both in the years prior to the development of sTR and the year after (Supplementary Table 2).

Table 1 displays baseline characteristics by race/ethnicity category. Black and Hispanic patients were younger on average than White patients. Black and Hispanic patients had higher rates of diabetes and kidney disease, while White patients had higher rates of atrial fibrillation and valve diseases other than TR.

Over half (56.8%) of patients had a TR diagnosis 3 years prior to developing sTR. Black patients were less likely to have a prior diagnosis of TR when compared with White (p < 0.001) and Hispanic (p < 0.001) patients. Black patients reached sTR status during inpatient admissions significantly more often than White patients (55 vs 48.3%, p < 0.001), but Hispanic patients (48.2%) did not differ significantly from White patients (p = 0.95).

Black patients had more urgent/emergent admissions and emergency department visits than White patients for all 3 years prior (year -3, year -2 and year -1) to the development of sTR (Figure 2). Black patients also had more urgent/emergent admissions than Hispanic patients for all prior years, and more emergency department visits in year -2 and year -1. Hispanic patients had more emergency department visits than White patients for all three prior years, but only differed on urgent/emergent admissions in year -1. White patients had more outpatient cardiac specialist visits than Black and Hispanic patients for all years prior to developing sTR. Hispanic patients also had more outpatient cardiac specialist visits than Black patients in year -3 but not in year -2 or year -1.

Table 2 shows the adjusted hazard ratios (HRs) for healthcare utilization measures 1-year after the development of sTR. Trends were similar to those seen prior to meeting sTR criteria. After adjustment, Black patients had higher rates of urgent/emergent admissions (p < 0.001) and emergency department visits (p < 0.001) but fewer cardiac specialist visits (p < 0.01) compared with White patients. Hispanic patients did not differ from White patients for urgent/emergent admissions or outpatient cardiac specialist visits, but did have higher rates of emergency department visits (p < 0.001). See Supplementary Table 3 for unadjusted hazard ratios.

Figure 3 displays the Kaplan–Meier curve for death in the year after development of sTR by race/ethnicity. After adjustment for baseline characteristics (Table 1), Black and Hispanic patients both had a lower risk of death compared with White patients (p < 0.001 and p < 0.01 respectively; Table 2).

Figure 4 displays the cumulative incidence function for heart failure hospitalization in the year after development of sTR by race/ethnicity. After adjustment for baseline characteristics (Table 1), Black and Hispanic patients both had a greater risk of heart failure hospitalization compared with White patients (p < 0.001 and p < 0.05 respectively; Table 2).

Figure 5 displays the HRs for potential explanatory variables for death (all baseline characteristics from Table 1), both overall and for each race/ethnicity category. Inpatient status at the time of meeting sTR criteria was the strongest predictor of death overall (HR: 2.39, 95% CI: 2.29–2.48, p < 0.001) and for each race/ethnicity category (Supplementary Table 4).

Figure 6 displays the HRs for the potential explanatory variables for heart failure hospitalization both overall and for each race/ethnicity category. Pre-existing congestive heart failure at baseline was the strongest predictor of heart failure hospitalization for all groups (HR: 3.60, 95% CI: 3.41–3.81; Supplementary Table 5).

Overall, older age, male sex, higher Elixhauser score, atrial fibrillation, congestive heart failure, pulmonary hypertension, kidney disease, oxygen dependence and identification of sTR during an inpatient encounter were all associated with greater risk of death and heart failure hospitalization. Receiving prior intervention on other (i.e., non-tricuspid) valves was associated with lower hazard ratios for both death and heart failure hospitalization. Diabetes and a prior pacemaker were both associated with lower risk of death only. Peripheral artery disease, stage 5 chronic kidney disease and chronic obstructive pulmonary disease were associated with higher risk of heart failure hospitalization only. Medicaid enrollment, other valve diseases, coronary artery disease, hypertension and outpatient cardiac specialist visits in the year prior to sTR were all associated with lower risk of death but higher risk of heart failure hospitalization. Stroke and cancer were associated with higher risk of death and lower risk of heart failure hospitalization. Some associations such as census region, hypertension and diabetes varied by race/ethnicity (Supplemental Tables 4 & 5).

Our study utilized a large database that included a diverse payer mix, including commercial insurance, Medicaid and Medicare throughout the US The data source incorporated EHR data and physician notes as well as administrative claims, allowing for longitudinal analysis of TR progression and healthcare utilization. Despite these strengths, the findings from this analysis should be interpreted with the following limitations in mind. First, this was an observational study. Proxy definitions for clinical symptomatology were used to define our sTR population and only a small percentage of patients entered the cohort directly from a physician note of severe or greater TR, a point in contrast to prior literature where severity was known. Granular clinical details are missing from our analysis, including patient baseline laboratory results and echocardiographic parameters such as left ventricular ejection fraction, and severity ratings are not well populated in the physician notes. In addition, the cohort was restricted to patients with non-rheumatic TR, and TR etiology was not captured. Due to limitations in the data, concomitant medical therapy was not assessed nor were other important potential confounding variables, including lifestyle factors such as diet and exercise or socioeconomic characteristics aside from health insurance payor (which may be considered at best a proxy for socioeconomic status). Clustering of patients at the hospital level was not adjusted for in associations between race/ethnicity and outcomes. Finally, race and ethnicity categories were simplified to ensure mutually exclusive groups for analysis; if patients identified as Black or White and also identified as Hispanic, they were classified as Hispanic within the cohort.