Measuring Frailty Using Self-Report or Automated Tools to Identify Risk of Cardiovascular Events and Mortality: The Million Veteran Program

Abstract

Background: Frailty, a syndrome of physiologic vulnerability, increases cardiovascular disease (CVD) risk. Whether in person or automated frailty tools are ideal for identifying CVD risk remains unclear. We calculated 3 distinct frailty scores and examined their associations with mortality and CVD events in the Million Veteran Program, a prospective cohort of nearly 1 million US veterans.

Methods and results: Veterans aged ≥50 years and enrolled from 2011 to 2018 were included. Two frailty indices (FI) based on the deficit accumulation theory were calculated: the questionnaire-based 36-item Million Veteran Program-FI and 31-item Veterans Affairs-FI using claims data. We calculated Fried physical frailty using the self-reported, 3-item Study of Osteoporotic Fractures. Multivariable-adjusted Cox models examined the association of frailty by each score with primary (all-cause and CVD mortality) and secondary (myocardial infarction, stroke, and heart failure) outcomes. In 190 688 veterans (69±9 years, 94% male, 85% White), 33, 233 (17%) all-cause and 10 115 (5%) CVD deaths occurred. Using Million Veteran Program-FI, 29% were robust, 42% pre-frail, and 29% frail. Frailty prevalence increased by age group (27% in 50-59 to 42% in ≥90 years). Using the Million Veteran Program-FI, over 6±2 years, frail veterans had a higher hazard of all-cause (hazard ratio [HR], 3.05 [95% CI, 2.95-3.16]) and CVD mortality (HR, 3.65 [95% CI, 3.43-3.90]). Findings were concordant for the Veterans Affairs-FI and Study of Osteoporotic Fractures frailty definitions, and remained significant even among younger veterans aged 50-59 years.

Conclusions: Irrespective of frailty measure, frailty is associated with a higher risk of all-cause mortality and adverse CVD events. Further study of frailty in veterans aged <60 years old is warranted.

Keywords: cardiovascular disease; frailty; risk stratification.

Figure 1. Hazards ratios (HR) for the association between frailty group and all‐cause mortality (upper panel) and cardiovascular disease mortality (lower panel), stratified by 10‐year age group.

The P‐value for interaction between frailty group and 10‐year age category was <0.0001 (8 df) for both all‐cause and cardiovascular disease mortality. CVD indicates cardiovascular disease.

Figure 2. Graphical overview of frailty indices (Million Veteran Program‐Frailty Index, Veterans Affairs‐Frailty Index, and Study of Osteoporotic Fractures).

Percentages shown indicate the proportion of total deficits for each frailty index attributed to each domain (health status/function, comorbidities, mood, and cognition). FI indicates frailty index; MVP, Million Veteran Program; SOF, Study of Osteoporotic Fractures; COPD, chronic obstructive lung disease; IADL, instrumental activities of daily living; ADL, activities of daily living; and TIA, transient ischemia attack.