Heart failure model predicted risk well overall in US veterans
The model scored 0.81 for distinguishing veterans who developed heart failure from those who did not, and closely matched overall risk over 10 years.

Our read
Biomarkers- Studied:
- Cohort study of 1,990,002 US veterans without known heart failure or ASCVD, 10 years
- What changed:
- Predicted heart failure risk closely matched observed risk
- The catch:
- Heart failure outcomes relied on diagnosis codes in health records
- In the record:
- The 4th paper on PREVENT-HF with results in people; 13 on record in all.
Editor of record: Boris Djordjevic · AgingResearch.News
Among 1,990,002 US veterans aged 30–79 without known heart failure or cardiovascular disease caused by fatty buildup in arteries, researchers tested a heart failure risk equation in a retrospective observational study. Women made up 8.8% of the cohort. The analysis used Veterans Health Administration and Medicare records.
Over 10 years, 160,400 veterans received new heart failure diagnoses. The American Heart Association’s equation closely matched overall observed risk and performed well at distinguishing who developed heart failure, while accounting for deaths. Performance was similar in urban and rural veterans. However, it underestimated risk in female veterans and those without diabetes. It overestimated risk in Asian and Hispanic veterans.
Why it matters
Estimating future heart failure risk can help prioritize prevention as people age. The authors emphasized validating risk models in diverse populations before using them to guide prevention.
Caveats
Heart failure outcomes relied on diagnosis codes in health records. These conference-abstract findings do not cover people with existing heart failure or cardiovascular disease caused by fatty buildup in arteries.
The paper
Validation of the PREVENT-HF Risk Prediction Equation in the Veterans Health Administration
Stanford University; University of California, Los Angeles
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 4672



Comments
Be the first to comment.