Cohort | Year published | Heart failure definition | Risk-window (years) | Risk factors included in base model | Summary findings | Major strengths | Selected limitations |
---|---|---|---|---|---|---|---|
Framingham Heart Study3 | 1999 | Symptoms, physical exam features Radiography | 4 | Age, ECG LVH, heart rate, systolic BP, CHD, valvular disease, diabetes, with and without radiographic cardiomegaly and vital capacity | Publication of regression coefficients Articulation of risk profile with point scoring | Broad age range (45–94 years) Long-term follow-up (38 years) Pooled regression model Heart failure adjudication | Racially homogenous cohort Absence of biomarker or imaging assessments No discrimination or reclassification metrics Generalisability (CHD, hypertension, valvular heart disease) |
Health ABC5 | 2008 | Overnight hospitalisation | 5 | Age, CHD, ECG LVH, heart rate, systolic BP, albumin, creatine, glucose, smoking | C-statistic 0.73 Framingham Heart Study score C-statistic <0.70 | Older adults Biracial (black and white race) Internal validation | Heart failure determination by hospitalisation only Misclassification of baseline covariates Cohort size (n=2853) Generalisability (cohort is black and white race) |
Atherosclerosis Risk in Communities Study6 | 2012 | ICD-9 coding, hospitalisation or death certificates | 10 | Age, sex, race, heart rate, systolic BP, CHD, antihypertensive medication, diabetes, BMI, smoking, log NT-BNP | C-statistic 0.80 (without NT-BNP) C-statistic 0.81 (with inclusion of NT-BNP) | Inclusion of BNP Biracial (black and white race) 15.5-year follow-up Covariate range | Heart failure ascertainment by hospitalisation, death certificate Generalisability (cohort is black and white race) |
Multi-Ethnic Study of Atherosclerosis7 | 2014 | Self-report and medical record | 5 | Age, sex, race and ethnicity, heart rate, systolic BP, diabetes, BMI, smoking, cholesterol, creatine, with and without log NT-BNP and LVMI | C-statistic 0.80 (without NT-BNP) C-statistic 0.87 (with inclusion of NT-BNP) Significant reclassification (25%–37%) with addition of LVMI or BNP | Multiethnic design Inclusion of BNP and LVMI Covariate range | Follow-up duration (median 4.7 years) with low heart failure incidence Younger-aged cohort Limited events, precluding race-specific risk score development |
BMI, body mass index; BP, blood pressure; CHD, coronary heart disease; Health ABC, Dynamics of Health, Aging and Body Composition; ICD-9, International Classification of Diseases, Ninth Revision; LVH, LV hypertrophy; LVMI, LV mass index; NT-BNP, N-terminal B-type natriuretic peptide.