Fitness versus Physical Activity Patterns in Predicting Mortality in Men Jonathan Myers, PhD, Amir Kaykha, MD, Sheela George, MD, Joshua Abella, MD, Naima Zaheer, MD, Scott Lear, PhD, Takuya Yamazaki, MD, Victor Froelicher, MD PURPOSE: To compare the contributions of fitness level and physical activity patterns to all-cause mortality. METHODS: Of 6213 men referred for exercise testing between 1987 and 2000, 842 underwent an assessment of adulthood ac- tivity patterns. The predictive power of exercise capacity and activity patterns, along with clinical and exercise test data, were assessed for all-cause mortality during a mean (SD) follow-up of 5.5  2 years. RESULTS: Expressing the data by age-adjusted quartiles, ex- ercise capacity was a stronger predictor of mortality than was activity pattern (hazard ratio [HR] = 0.56; 95% confidence interval [CI]: 0.38 to 0.83; P  0.001). In a multivariate anal- ysis that considered clinical characteristics, risk factors, ex- ercise test data, and activity patterns, exercise capacity (HR per quartile = 0.62; CI: 0.47 to 0.82; P  0.001) and energy expenditure from adulthood recreational activity (HR per quartile = 0.72; 95% CI: 0.58 to 0.89; P = 0.002) were the only significant predictors of mortality; these two variables were stronger predictors than established risk factors such as smoking, hypertension, obesity, and diabetes. Age-adjusted mortality decreased per quartile increase in exercise capacity (HR for very low capacity = 1.0; HR for low = 0.59; HR for moderate = 0.46; HR for high = 0.28; P  0.001) and phys- ical activity (HR for very low activity = 1.0; HR for low = 0.63; HR for moderate = 0.42; HR for high = 0.38; P  0.001). A 1000-kcal/wk increase in activity was approxi- mately similar to a 1 metabolic equivalent increase in fitness; both conferred a mortality benefit of 20%. CONCLUSION: Exercise capacity determined from exercise testing and energy expenditure from weekly activity outperform other clinical and exercise test variables in predicting all-cause mortality. Am J Med. 2004;117:912–918. ©2004 by Elsevier Inc. I ncreasing evidence of the association between physi- cal inactivity and cardiovascular or all-cause mortal- ity (1) has led health authorities around the world to make physical activity promotion part of broad health care policy goals (1–3). More recent studies (3–6) have observed strong associations between physical fitness, measured by a maximal exercise test, and survival from cardiovascular and noncardiovascular causes. Physical fitness is related to physical activity patterns, and thus current physical activity guidelines generally consider fit- ness a surrogate measure of physical activity. However, other attributes, such as genetics, subclinical disease, and behavioral and environmental factors, determine indi- vidual fitness levels (7). There has been some recent de- bate as to whether daily physical activity patterns largely determine one’s fitness level and therefore its inverse as- sociation with mortality, or whether fitness level predicts mortality independently from activity pattern (8,9). In addition, while these issues have been studied largely in asymptomatic populations (10 –13), less is known about these associations in patients with existing cardiovascular disease. In the present study, we assessed all-cause mortality using fitness measured in subjects referred for exercise testing for clinical reasons, and quantified adulthood physical activity patterns by questionnaire. Our objec- tives were to compare the independent contributions of fitness and physical activity patterns to overall mortality, to determine the predictive power of fitness and activity patterns as compared with other clinical and exercise test variables, and to assess the interaction between fitness and activity in predicting mortality. METHODS Sample The sample was drawn from 6213 consecutive men (mean [SD] age, 59  11 years) who were referred for exercise testing for clinical reasons between April 1987 and July 2000. Of these, a subgroup of 842 subjects un- derwent a detailed evaluation of current and past activity patterns. The subgroup represented a convenience sam- ple tested on a particular day of the week in which re- search assistants were assigned to oversee the data collec- tion. Exercise Testing The exercise laboratory was directed in a consistent fashion by two of the authors (VF and JM). A thorough clinical history, listing of medications, and risk factors (beginning in 1987) were recorded prospectively at the From the Cardiology Division, Veterans Affairs Palo Alto Health Care System, Stanford University School of Medicine, Palo Alto, California. Requests for reprints should be addressed to Jonathan Myers, PhD, Palo Alto VA Health Care System, Cardiology-111C, 3801 Miranda Av- enue, Palo Alto, California 94304, or drj993@aol.com. Manuscript submitted November 24, 2003, and accepted in revised form June 24, 2004. 912 © 2004 by Elsevier Inc. 0002-9343/04/$–see front matter All rights reserved. doi:10.1016/j.amjmed.2004.06.047