Abstract
BACKGROUND: The frailty index (FI) is an established predictor of all-cause mortality among older adults, but less is known with regard to cause-specific mortality, and whether the predictive power of the FI varies between men and women and by socio-economic position. METHODS: We assessed all-cause and cause-specific mortality during 8 years of follow-up (median = 7 years) among the population-representative sample of older adults (65 + , n = 2,561) from the European Health Interview Survey in Austria (ATHIS 2014). A FI at baseline was constructed from 41 health deficits. Official cause of death information from Statistics Austria was linked with the survey data by the Austrian Micro Data Center (AMDC). Next to all-cause mortality, we differentiated between mortality from cardiovascular diseases (CVD), cancer, and other causes. Cox proportional hazard models adjusted for socio-demographic variables and causes of death as competing risks were used to assess mortality prediction. RESULTS: Among the participants, 43.5% were robust (FI < 0.10), 37.7% pre-frail (FI = 0.10-0.21), and 18.7% were frail (FI > 0.21). 405 (15.8%) participants died during follow-up. Among the deceased, 148 (36.5%) died from CVD, 127 (31.4%) died from cancer, and 130 (32.1%) died from other causes of death. The FI predicted all-cause (hazard ratio, HR = 1.33 per 0.1 FI and HR = 2.4 for frail compared to robust older adults) and cause-specific mortality risk (HR(CVD) = 1.25/2.46, HR(cancer) = 1.19/1.47, HR(other) = 1.49/3.59). Area under the curve (AUC) values were acceptable for CVD mortality (0.78) and other causes of death (0.74), and poor for cancer mortality (0.64). CONCLUSIONS: The FI predicts all-cause and cause-specific mortality (CVD, other causes) well, which points to its relevance as a potential screening tool for risk stratification among community-dwelling older adults.