Comprehensive Unified Regimen for Eliminating Undiagnosed/Untreated Aortic Valve Stenosis: Algorithm Validation for Identifying Aortic Stenosis and Treatment Disparities

TL;DR

A new algorithm accurately identifies severe aortic stenosis from echocardiograms and reveals treatment disparities among older adults, women, and those with nonprivate insurance.

Contribution

Development and validation of an echocardiogram-based rules engine for identifying untreated severe aortic stenosis and analyzing treatment disparities.

Findings

The rules engine achieved 100% sensitivity and 95.4% specificity in identifying severe aortic stenosis.

Untreated severe aortic stenosis was found in 11% of patients.

Disparities in treatment were observed for women, older adults, and those with nonprivate insurance.

Abstract

Severe aortic stenosis (sAS) leads to high morbidity and mortality when left untreated. We sought to develop and validate an algorithm-based rules engine to identify patients with untreated sAS and to evaluate differences between those who did and did not subsequently receive guideline-concordant treatment with aortic valve replacement (AVR).

We curated discrete and nondiscrete data from our echocardiography system, then created a rules engine to identify and grade aortic stenosis. We assessed sensitivity and specificity of the rules engine to identify sAS using manual adjudication. We additionally conducted a retrospective cohort analysis to identify demographic and socioeconomic factors associated with receipt of guideline-concordant AVR treatment for sAS.

The rules engine demonstrated 100% sensitivity and 95.4% specificity for identifying sAS across n ​= ​2162 echocardiographic studies from unique patients. Univariate analyses revealed patients with untreated sAS were more likely to be older and female, with no appreciated differences by race, ethnicity, insurance status, or neighborhood-level socioeconomic scores. In multivariable analyses, older individuals, women, and those with Medicare/Medicare advantage were less likely to undergo AVR. Among treated patients, those who underwent surgical AVR were more likely to be younger, male, and have lower socioeconomic neighborhood scores.

Untreated sAS is prevalent but can be accurately identified at scale using an echocardiogram report-based rules engine. Disparities in the receipt of AVR persist, particularly among women, older adults, and patients with nonprivate insurance coverage. The systematic use of automated algorithmic protocols may facilitate valvular heart disease identification and reduction of treatment disparities.

•Development and validation of an echocardiogram-based rules engine identified a high prevalence (11%) of untreated severe aortic stenosis.•The developed rules engine demonstrated high sensitivity (100%) and specificity (95%) for categorizing severe aortic stenosis.•Disparities in the receipt of guideline-concordant aortic valve replacement persist, particularly among women, older adults, and patients with nonprivate insurance coverage.

Development and validation of an echocardiogram-based rules engine identified a high prevalence (11%) of untreated severe aortic stenosis.

The developed rules engine demonstrated high sensitivity (100%) and specificity (95%) for categorizing severe aortic stenosis.

Disparities in the receipt of guideline-concordant aortic valve replacement persist, particularly among women, older adults, and patients with nonprivate insurance coverage.

Figures