Assessing a Predictive Model for Wild Type Transthyretin Amyloid Cardiomyopathy

According to a study presented at AMCP 2022, wild-type transthyretin amyloid cardiomyopathy (ATTRwt-CM) is a “frequently misdiagnosed cause of heart failure (HF).” Therapies which are effective in ATTRwt-CM have been developed, but a need now exists to optimize screening for ATTRwt-CM. Lead author of the study, C. Reed, and colleagues evaluated a recent machine learning (ML) approach to identify suspected ATTRwt-CM in patients with HF. They judged that the “ATTRwt-CM predictive model performed well in identifying disease risk in patients with confirmed ATTRwt-CM in a large health plan claims database.”

The retrospective study evaluated the predictive model in a randomized 1:1 sample of patients diagnosed with ATTRwt-CM (n = 119), and patients with non-amyloid HF (n = 119) in a Humana administrative claims database. All patients with non-amyloid HF in the database (n = 266,906) were categorized as having suspected ATTRwt-CM.

According to the article, the ML model had a predictive value, sensitivity, specificity, and accuracy of 71%, 88%, 65%, and 77%, respectively, as well as an area under the receiver operating curve of 0.89. The study’s authors also observed that patients with confirmed ATTRwt-CM had a mean (standard deviation [SD]) time between HF and ATTRwt-CM diagnosis of 751 (528) days. Additionally, 65% of those patients had claims for all-cause hospitalizations before their ATTRwt-CM diagnosis, lowering to 48% after diagnosis. Furthermore, the frequency of comorbid atrial fibrillation was higher in confirmed cases of ATTRwt-CM and in patients with suspected high-risk of ATTRwt-CM when compared to patients without suspected disease (39% and 55% vs. 27%).

Overall, the study’s collaborators concluded that the ML ATTRwt-CM predictive model was effective at identifying risk for ATTRwt-CM, based on its performance in a retrospective health care claims database. In closing, the authors also noted that high ATTRwt-CM risk patients demonstrated high health care resource utilization, “and may benefit from earlier suspicion of ATTRwt-CM.”