Assessing cardiorespiratory fitness without performing exercise testing

Background: Low cardiorespiratory fitness (CRF) is associated with increased risk of chronic diseases and mortality; however, CRF assessment is usually not performed in many healthcare settings. The purpose of this study is to extend previous work on a non-exercise test model to predict CRF from health indicators that are easily obtained. Methods: Participants were men and women aged 20 to 70 years whose CRF level was quantified with a maximal or submaximal exercise test as part of the National Aeronautics and Space Administration/ Johnson Space Center (NASA, n=1863), Aerobics Center Longitudinal Study (ACLS, n=46,190), or Allied Dunbar National Fitness Survey (ADNFS, n=1706). Other variables included gender, age, body mass index, resting heart rate, and self-reported physical activity levels. Results: All variables used in the multiple linear regression models were independently related to the CRF in each of the study cohorts. The multiple correlation coefficients obtained within NASA, ACLS, and ADNFS participants, respectively, were 0.81, 0.77, and 0.76. The standard error of estimate (SEE) was 1.45, 1.50, and 1.97 metabolic equivalents (METs) (1 MET=3.5 ml O₂ uptake·kilograms of body mass ^-1·minutes^-1), respectively, for the NASA, ACLS, and ADNFS regression models. All regression models demonstrated a high level of cross-validity (0.72<R<0.80). The highest cross-validation coefficients were seen when the NASA regression model was applied to the ACLS and ADNFS cohorts (R=0.76 and R=0.75, respectively). Conclusions: This study suggests that CRF may be accurately estimated in adults from a non-exercise test model including gender, age, body mass index, resting heart rate, and self-reported physical activity.