A Simple Nonexercise Model of Cardiorespiratory Fitness Predicts Long-Term Mortality

Abstract

Cardiorespiratory fitness (CRF) is a strong predictor of future health, but measurements of CRF are time consuming and involve costly test procedures. We assessed whether a simple, non-exercise-based test of CRF predicted long-term all-cause and cardiovascular disease (CVD) mortality. In this prospective cohort study, we used a previously published nonexercise test to estimate CRF in healthy men (n = 18,348) and women (n = 18,764) from the first HUNT study (1984-1986) in Norway. We used Cox regression to obtain HR for mortality during a mean follow-up of 24 yr. Assessment of model validity was performed by standard procedures of discrimination and calibration. CRF was inversely associated with all-cause and CVD mortality in men and women below 60 yr of age at baseline, after adjustment for confounders. For each MET-higher CRF (MET, approximately 3.5 mL·kg·min), HR for CVD mortality was 21% lower in both men (95% confidence interval (CI), 17%-26%) and women (95% CI, 12%-29%). HR for all-cause mortality was 15% (95% CI, 12%-17%) lower in men and 8% (95% CI, 3%-3%) lower in women for each MET-higher CRF. The ability of the model to discriminate mortality risk among participants below 60 yr was better for CRF (area under the curve (AUC), 0.70-0.77) compared with that for each variable that constituted the model (AUC, 0.55-0.63) and an aggregated sum of z-scores for each variable (AUC, 0.61-0.65). Comparison of observed and predicted risk indicated good model calibration. This method of assessing CRF is feasible and practically useful in primary care for identification of apparently healthy individuals at increased risk of premature CVD disease and all-cause mortality.