Impact of ICD-10-CM Z-Coded Social Determinants of Health on Transcatheter Aortic Valve Replacement–Related Outcomes

Highlights

• •Z-codes identified transcatheter aortic valve replacement patients associated with higher cardiovascular risk.
• •Mortality was similar, but morbidity burden differed by Z-code status.
• •Findings support integrating social determinants of health data into transcatheter aortic valve replacement risk stratification.

Introduction

Transcatheter aortic valve replacement (TAVR) has become the leading therapy for severe aortic stenosis (AS), with known disparities in diagnosis, treatment, and outcomes.1, 2, 3 In 2015, International Classification of Diseases (ICD)-10 introduced Z-codes to capture social and environmental factors, but their use has remained limited, appearing in fewer than 2% of national claims.^4,5 No prior studies have examined whether Z-codes identify patients at risk for adverse outcomes following TAVR. We therefore evaluated the association between Z-codes and post-TAVR outcomes.

Methods

We conducted a retrospective study using the TriNetX Network, a federated health research platform that aggregates deidentified electronic health records (EHRs) from 107 health care organizations across the United States (∼98%) and internationally (∼2%). We identified adults aged 18 years or older who underwent TAVR between January 1, 2015, and December 31, 2024, with severe AS defined using ICD-10-Clinical Modification (CM) code I35.0. Patients were categorized into two cohorts: first cohort included patients with severe AS who underwent TAVR who also had a recorded diagnosis including ICD-10-CM Z55–Z65, representing social determinants of health (SDoH). Qualifying Z-codes were required to precede the TAVR by at least 1 day and up to 1 year pre-TAVR. Second cohort included patients with TAVR and severe AS who had no Z55–Z65 codes recorded during the same period. Patients with no baseline data using a 1 year lookback period analysis were excluded.

Index date was designated as 21 days post-TAVR to reduce periprocedural confounding and follow-up continued through 1 year. Primary outcomes were all-cause mortality (EHR based death status) and all-cause hospitalization (EHR based observation, inpatient, or critical care coding). Secondary outcomes included acute heart failure (HF) events (defined as HF hospitalizations (ICD-10 coding) or need for intravenous diuretics (RxNorm coding), acute myocardial infarction (AMI [ICD-10 coding]) or percutaneous coronary intervention (Current Procedural Terminology coding), ischemic stroke (ICD-10 coding), cardiac arrest (ICD-10 coding), major adverse cardiovascular events (defined as cardiac arrest, acute HF events, ischemic stroke, or AMI [ICD-10 coding]), bradyarrhythmias (ICD-10 coding), and ventricular tachycardia (ICD-10 coding).

Propensity score matching (PSM) was applied to balance clinical characteristics between cohorts using a 1:1 greedy nearest-neighbor algorithm. Covariates included age, sex, race, comorbidities, pharmacotherapies, and key laboratory data. Hazard ratios (HRs) were estimated using Cox proportional hazard models. Kaplan–Meier survival analysis with log-rank testing was performed for time-to-event outcomes, with censoring applied at the last available encounter. Ethical oversight was not obtained due to the use of publicly available, deidentified aggregate-level data. R software (R Foundation for Statistical Computing, Vienna, Austria) was used for statistical analysis.

Results

Before PSM, a total of 1631 patients were identified to have Z-codes and TAVR, whereas 56,803 patients were identified to have undergone TAVR without documented Z-codes. After PSM with a 1:1 match, a total of 6 and 55,178 patients were excluded from the Z-code + TAVR and the TAVR only cohorts, respectively. Therefore, 1625 patients remained in each cohort. The mean age was similar (78.85 vs. 78.94 years, in Z-coded and non-Z-coded cohorts, respectively), with White patients comprising 74.7 vs. 72.6% and females comprising 45.2 vs. 44.0%. Comorbidities included atrial fibrillation/flutter (45.2 vs. 45.7%), overweight/obesity (50.1 vs. 49.8%), dyslipidemia (89.9 vs. 90.3%), and HF (75.1 vs. 76.0%). Medication use was similar: beta-blockers (81.6 vs. 80.9%) and antilipemic agents (85.8 vs. 86.3%). The mean left ventricular ejection fraction was 55.2 vs. 55.5%, hemoglobin A1c was 6.34% in both groups, and creatinine was 1.38 vs. 1.49 mg/dL. The mean follow-up time was 291.40 (SD 115.45) in the Z-coded cohort and 318.32 (SD 97.43) in the non–Z-coded cohort.

All-cause mortality occurred in 127 Z-coded patients (7.8%) and 129 non–Z-coded patients (7.9%) (HR 1.07, 95% CI 0.84–1.37) (Figure 1a). All-cause hospitalization was higher in the Z-coded cohort, affecting 830 patients (51.1%) compared with 737 (45.4%) in the non–Z-coded cohort (HR 1.27, 95% CI 1.15–1.40) (Figure 1b). Rates of acute HF were also increased among Z-coded patients (445 [27.4%] vs. 408 [25.1%], HR 1.18, 95% CI 1.03–1.35) (Figure 1c).

Figure 1.

Outcomes. (a) Kaplan–Meier curve representing all-cause mortality-free survival probability, (b) Kaplan–Meier curve representing all-cause hospitalization-free survival probability, and (c) Forest plot representing all hazard ratios.

Abbreviations: AMI, acute myocardial infarction; MACE, major adverse cardiovascular events; PCI, percutaneous coronary intervention; TAVR, transcatheter aortic valve replacement.

AMI or percutaneous coronary intervention occurred in 142 patients (8.7%) vs. 115 (7.1%) in the non–Z-coded group (HR 1.33, 95% CI 1.04–1.70). Ischemic stroke was observed in 148 patients (9.1%) compared with 133 (8.2%) (HR 1.19, 95% CI 0.94–1.51), and cardiac arrest rates were also comparable between groups (31 [1.9%] vs. 28 [1.7%], HR 1.20, 95% CI 0.72–1.99). Major adverse cardiovascular events were higher in the Z-coded cohort, occurring in 606 patients (37.3%) versus 550 (33.8%) in the non–Z-coded cohort (HR 1.21, 95% CI 1.11–1.36).

Bradyarrhythmias occurred in 167 Z-coded patients (10.3%) compared with 149 (9.2%) in non–Z-coded patients (HR 1.21, 95% CI 0.97–1.50). Ventricular tachycardia was similar in the Z-coded cohort compared to the non–Z-coded cohort (95 [5.8%] vs. 122 [7.5%], HR 0.82, 95% CI 0.63–1.07).

Discussion

Our findings highlight that patients with documented Z-codes are associated with higher rates of cardiovascular events following TAVR. These results highlight the potential of Z-codes as a pragmatic, built-in tool for identifying patients associated with higher risk of adverse outcomes beyond conventional clinical factors. Although prior work has focused on disparities in AVR diagnosis and utilization across racial/ethnic and socioeconomic groups, our study extends this evidence by demonstrating that documented SDoH factors, captured through Z-codes, may also be associated with worse postprocedural outcomes. This suggests that barriers to longitudinal follow-up, adherence, or access to supportive resources may drive excess hospitalizations and complications in these populations.⁵

Our study has several limitations. First, reliance on EHRs introduces the possibility of miscoding, incomplete data capture, inherent biases, and skewed follow-up data. Second, observational design precludes establishing causality and may be influenced by residual confounding given that granular clinical details such as coronary or cardiac anatomical features were not available. Finally, because mortality data may be underascertained due to deaths that occur outside participating health systems, our findings may underestimate true event rates.

Conclusions

Z-codes remain underutilized but may offer meaningful insight into SDoH that are associated with outcomes post-TAVR. Broader adoption and integration of these codes into risk stratification may enhance prognostic models.

Ethics Statement

This article adhered to the STrengthening the Reporting of OBservational studies in Epidemiology guidelines.

Funding

This publication was supported by Mayo Clinic Arizona Cardiovascular Clinical Research Center (MCA CV CRC). We are thankful for their generous support. Contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the MCA CV CRC.

Disclosure Statement

The other authors had no conflicts to declare.