fbpx Skip to main content

This study highlights the clinical characteristics and healthcare resource use of patients at higher risk for wild-type transthyretin amyloid cardiomyopathy identified using a machine learning model.

Transthyretin amyloid cardiomyopathy (ATTR-CM) can lead to heart failure (HF). If there is a delay in the diagnosis of wild-type ATTRCM (ATTRwt-CM), patients may undergo multiple hospital visits before receiving a diagnosis. Earlier recognition of patients at risk for ATTRwt-CM is the key to prompt diagnosis and treatment. 

A machine learning model has performed well in previous studies for identifying patients with HF showing signs of ATTRwt-CM. A recent study published in the Journal of Managed Care & Specialty Pharmacy further assessed this model’s performance in identifying ATTRwt-CM and analyzed the healthcare resource utilization (HCRU) and clinical characteristics of patients with diagnosed and suspected ATTRwt-CM.

Study Population

This retrospective, observational study comprised 267,025 patients. Out of these, 119 patients had a confirmed diagnosis of ATTRwt-CM based on claims data. The mean time between the HF diagnosis and the diagnosis of ATTRwt-CM was approximately 751 days. The mean follow-up time after the ATTRwt-CM diagnosis was approximately 327 days. Of the 266,906 HF patients without an ATTRwt-CM diagnosis, 119 were randomly assigned as HF controls. In the patients with non-amyloid HF, 4.1%, 25.5%, and 70.3% were deemed high, moderate, and low risk for ATTRwt-CM, respectively.

Performance of the Machine Learning Model

The model correctly identified patients with ATTRwt-CM with 88% sensitivity, 65% specificity, and 77% accuracy. The positive predictive value was 71%.

Characteristics of Patients With Confirmed ATTRwt-CM and At-Risk Groups

More patients in the confirmed ATTRwt-CM group were men (78%) than patients at high, moderate, and low risk (53%, 51%, and 48%, respectively), and fewer patients were White (62%) in this group compared to the high-, moderate-, and low-risk cohorts (80%, 81%, and 79%, respectively). Patients in the confirmed ATTRwt-CM group also had lower comorbidity scores and a higher proportion of heart failure with preserved ejection fraction (HFpEF) than patients in the at-risk groups. The mean age was approximately 76–78 years and was similar across all cohorts. 

You May Also Like::  Study Refutes 'Fat but Healthy' Theory

Healthcare Resource Utilization Patterns Before and After Diagnosis

In the confirmed ATTRwt-CM group, the proportion of patients with claims for most health care resources (except skilled nursing facility use and outpatient visits) was highest in the year before the diagnosis compared to after the confirmation of the diagnosis, or 1 to 2 years before the diagnosis. In patients at high risk for ATTRwt-CM, the proportion of patients with claims for acute care services was lower than that with confirmed ATTRwt-CM before diagnosis but higher than that with confirmed ATTRwt-CM patients after diagnosis.

Factors Affecting Healthcare Resource Utilization

Among patient factors affecting HCRU, age only impacted the frequency of claims for post-acute health care services. Women showed higher odds of hospitalization, emergency visits, and post-acute care services. Black patients had lower odds of hospitalization but higher odds of emergency visits. Certain cardiac conditions, e.g., unstable angina and syncope, increased the risk of HCRU, whereas others, e.g., atrial fibrillation and cardiomyopathy, decreased the risk.


Bruno, M., Sheer, R., Reed, C., Schepart, A., Nair, R., & Simmons, J. D. (2023). Clinical characteristics and health care resource use of patients at risk for wild-type transthyretin amyloid cardiomyopathy identified by machine learning model. Journal of Managed Care & Specialty Pharmacy, 29(5), 530–540. https://doi.org/10.18553/jmcp.2023.29.5.530