IntroductionPrevention of heart failure (HF) is achievable with aggressive risk factor management. While the updated 2017 ACC/AHA/HFSA guidelines recommend identification of individuals at risk for HF to target with intensive risk factor modification, recommended strategies for quantitative risk assessment of HF are lacking. Therefore, we sought to evaluate the performance of the recently developed the Pooled Cohort equations to Prevent Heart Failure (PCP-HF) within a large, contemporary, electronic health record (EHR) based cohort to support rapid translation into clinical practice.HypothesisThe sex- and race-specific PCP-HF tool will have good discrimination and calibration when applied within a “real-world” clinical cohort derived in the EHR.MethodsIn this observational longitudinal study, patients were identified between January 2005 and October 2018 using the Northwestern Medicine Enterprise Data Warehouse, an integrated health care system with over 200 care locations. Patients included were between 30-79 years of age, free of cardiovascular disease at baseline, had available data on HF risk factors (age, sex, smoking status, systolic blood pressure, hypertension treatment, body mass index, total cholesterol, high density lipoprotein cholesterol, glucose, and diabetes treatment), and at least 5 years of follow-up. We applied the PCP-HF model to estimate 10-year risk of incident HF. We assessed model performance comparing predicted and observed rates for incident HF defined by established criteria using International Classification of Diseases codes.ResultsAmong 31,256 eligible adults, mean age was 51.4 years, 57% were women, and 11% black. Incident HF occurred in 568 patients (1.8%) with 161 cases in white males (1.5%), 186 cases in white females (1.4%), 58 cases in black males (6.9%), and 106 cases in black females (4.3%). The PCP-HF tool had excellent discrimination in white men (c-statistics 0.82) and women (0.82) and modest discrimination in black men (0.69) and women (0.69). Calibration of the model was reasonable with a χ2 value less than 20 in all race-sex subgroups (Figure 1).ConclusionsA novel risk score predicts incident HF in a “real-world” EHR-based cohort. Integration of HF risk into the EHR may allow for risk-based discussion and opportunities for sequential screening (biomarkers, echocardiography) and targeted preventive intervention to reduce HF risk. Prevention of heart failure (HF) is achievable with aggressive risk factor management. While the updated 2017 ACC/AHA/HFSA guidelines recommend identification of individuals at risk for HF to target with intensive risk factor modification, recommended strategies for quantitative risk assessment of HF are lacking. Therefore, we sought to evaluate the performance of the recently developed the Pooled Cohort equations to Prevent Heart Failure (PCP-HF) within a large, contemporary, electronic health record (EHR) based cohort to support rapid translation into clinical practice. The sex- and race-specific PCP-HF tool will have good discrimination and calibration when applied within a “real-world” clinical cohort derived in the EHR. In this observational longitudinal study, patients were identified between January 2005 and October 2018 using the Northwestern Medicine Enterprise Data Warehouse, an integrated health care system with over 200 care locations. Patients included were between 30-79 years of age, free of cardiovascular disease at baseline, had available data on HF risk factors (age, sex, smoking status, systolic blood pressure, hypertension treatment, body mass index, total cholesterol, high density lipoprotein cholesterol, glucose, and diabetes treatment), and at least 5 years of follow-up. We applied the PCP-HF model to estimate 10-year risk of incident HF. We assessed model performance comparing predicted and observed rates for incident HF defined by established criteria using International Classification of Diseases codes. Among 31,256 eligible adults, mean age was 51.4 years, 57% were women, and 11% black. Incident HF occurred in 568 patients (1.8%) with 161 cases in white males (1.5%), 186 cases in white females (1.4%), 58 cases in black males (6.9%), and 106 cases in black females (4.3%). The PCP-HF tool had excellent discrimination in white men (c-statistics 0.82) and women (0.82) and modest discrimination in black men (0.69) and women (0.69). Calibration of the model was reasonable with a χ2 value less than 20 in all race-sex subgroups (Figure 1). A novel risk score predicts incident HF in a “real-world” EHR-based cohort. Integration of HF risk into the EHR may allow for risk-based discussion and opportunities for sequential screening (biomarkers, echocardiography) and targeted preventive intervention to reduce HF risk.
Read full abstract
Oct 23, 2024
Maj-Britt Jensen + 10
Open Access