Journal Logo


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


Author Information
Medicine & Science in Sports & Exercise: June 2014 - Volume 46 - Issue 6 - p 1159-1165
doi: 10.1249/MSS.0000000000000219



The authors of “A Simple Nonexercise Model of Cardiorespiratory Fitness Predicts Long-Term Mortality” (1) report an error in Supplemental Digital Content 1 of their paper. The regression equation for age-predicted VO2peak for women in the published supplemental file is 55.6 − (0.328 × age) but should be 51.6 − (0.328 × age). The correct formula is used throughout all analyses, and the error has therefore no implication for any results or interpretation.

The amended supplemental content is provided via the following link:

Medicine & Science in Sports & Exercise. 52(6):1440, June 2020.

Low cardiorespiratory fitness (CRF), as measured by maximal exercise testing, is a potent predictor of cardiovascular disease (CVD) and premature all-cause mortality (12,22,29) and may also be a stronger predictor than traditional risk factors such as overweight, dyslipidemia, high blood pressure, and smoking (7,8,20,24). In addition, a single measurement of CRF considerably improves risk classification beyond traditional risk factors (13). Therefore, the objective assessment of CRF (i.e., by ventilatory gas analysis) is an appealing approach for risk stratification in clinical and public health settings. However, such tests are expensive and time consuming, require trained personnel, and are thus not practically available in most health care settings. Some studies have shown that CRF may be estimated with reasonable accuracy by the so-called nonexercise testing models (17,19,31). These CRF models apply easily accessible variables that could potentially be measured at every clinical consultation (i.e., age, waist circumference, self-reported physical activity, and resting heart rate) and may be used as a proxy for objectively measured CRF (19,31). To gain acceptance as a clinically useful tool in health care settings, such models should also, similar to objective measurements, be able to predict future risk of CVD and all-cause mortality. To date, only one study has examined the predictive ability of a nonexercise test of CRF using pooled data from different British cohorts and approximately 9 yr of follow-up (35).

The aim of the present study was to assess the ability of a nonexercise test (31) to predict long-term (24 yr) all-cause and CVD mortality in a large population of healthy men and women.


Study population

The Nord-Trøndelag health study (the HUNT Study) is a large total population-based study consisting of three surveys conducted in 1984–1986 (HUNT 1), 1995–1997 (HUNT 2), and 2006–2008 (HUNT 3) and has been described extensively elsewhere (23). The present study included participants from HUNT 1. Briefly, the total adult population age 20 yr and above in Nord-Trøndelag County in the middle of Norway was invited. The HUNT population is stable and homogenous with a low net migration (approximately 0.3% per year) and consists of predominantly Caucasians (approximately 97%) (15). Of 75,032 participants (86.8% of the total invited), 55,067 individuals (28,006 women and 27,061 men) were regarded healthy at baseline (i.e., free from self-reported angina pectoris, myocardial infarction, diabetes, stroke, use of blood pressure medication, and any motion impairment or somatic disease resulting in long-term functional impairment) and included in the present study. After exclusion of missing values, 37,112 participants (18,764 women and 18,348 men) were included in the final analysis (Fig. 1). All participants have provided an informed consent; the study was conducted in conformity with the Declaration of Helsinki and approved by the regional committee for medical research ethics.

Flow chart for selection of participants.

Clinical measures and questionnaire-based information

The clinical examination was conducted by trained nurses and consisted of standardized measurements of height, weight, blood pressure, and resting heart rate (23,30). A self-administered questionnaire provided information about leisure time physical activity, smoking habits, alcohol consumption, marital status, family history of disease, and attained education. Physical activity questions were related to frequency, intensity, and duration, respectively (1,31). From these three questions, we constructed a previously published physical activity summary index (PA-I). Weighted values for each response option were multiplied to construct the individual PA-I variable. The question on physical activity frequency was “How often do you exercise?” and had three response categories: “never”, “less than once a week”, “once a week”, “two to three times a week”, and “almost every day”. The intensity question, stated as “How hard do you usually push yourself?”, had three response categories: “no sweating or heavy breathing”, “sweating and heavy breathing”, and “push myself to exhaustion”. The question on average duration per session contained four response options: “less than 15 min”, “between 15 and 30 min”, “between 30 and 60 min”, and “more than 60 min”. The PA-I has been shown to correlate reasonably well with directly measured peak oxygen uptake (31).

Estimated CRF

A nonexercise prediction model, which was derived and cross-validated in a subsample of healthy participants with a similar age range, in HUNT 3 was used to estimate CRF (31). The sex-specific models consisted of age, body composition (body mass index (BMI) or waist), PA-I, and resting heart rate. In the main model presented in that study, waist circumference was used as a measure of body composition. A secondary model using BMI as a predictor variable yielded only negligible differences regarding SEE and goodness of fit (R2). Because waist circumference was not measured in HUNT 1, the BMI model was used in the present study:

These algorithms were used to predict each individual’s CRF in the HUNT 1 cohort. Furthermore, each participant’s CRF was estimated relative to the expected value for their age, as described in the supplemental material (see document, Supplemental Digital Content 1, Supplemental methods section,, and figure, Supplemental Digital Content 2, Nomogram showing the percentage of age-predicted CRF in 4637 healthy men and women, A nomogram for simple estimation of percentage of age-relative CRF is easily interpretable and may be feasible in primary care settings.

End points and follow-up

The National Cause of Death Register, on the basis of the unique 11-digit person identification number, allowed a virtually complete follow-up. The primary end point was death caused by CVD (International Classification of Diseases, 9th revision, 390–459; 10th revision, I00–I99) or from all causes.

Statistical analysis

We calculated person-days from the date of attendance to the clinical examination in HUNT 1 until the registered date of death or until the end of follow-up, December 31, 2010, whichever came first. Absolute death rate for each category of estimated CRF was calculated per 1000 person-years. Cox regression analyses were used to assess the HR for death associated with estimated CRF, adjusting for potential confounders. We assessed the age-specific effects of estimated CRF on all-cause and CVD deaths and found an evidence of interaction by age at cut-off 60 yr using a likelihood ratio test. HR were calculated per MET increase (MET, approximately 3.5 mL·kg−1·min−1) in estimated CRF. In the basic model, we adjusted for age by entering attained age as the time scale, whereas the multiadjusted model also included smoking, alcohol consumption, education, marital status, and family history of disease. The assumption of proportionality of hazards was examined by Schoenfeld residuals for each covariate. A sensitivity analysis to reduce the likelihood of reverse causation was performed by excluding the first 5 yr of follow-up.

Estimated CRF was also divided into sex-specific quintiles for those above and below 60 yr. Low fitness was defined as the lowest quintile (20%) of CRF, whereas the second and third constituted the medium fitness group (next 40%) and the fourth and fifth, the high fitness groups (highest 40%), as previously recommended (3). The percentage of the age-predicted CRF that was estimated from the model was divided into three sex-specific groups. Those with a CRF below 85% of the age-predicted value constituted the least fit group, as previously suggested (12,21). Those with 85%–100% of the age-predicted CRF constituted the intermediate group, whereas those with a CRF above 100% of the age-predicted value constituted the reference group. The predictive performance of the nonexercise model was assessed by examining measures of discrimination and calibration (14). Discrimination was assessed by calculating the area under the curve for receiver operating characteristic curves (Harrell c-statistic) with 95% confidence intervals (CI) for CRF and each modifiable component. Sensitivity and specificity were assessed using the age- and sex-specific cut points for low CRF. An aggregated score for the components was calculated by summing up the z-scores (z-score = actual value − mean value/SD) for each variable except for age and dividing by three. The z-scores for PA-I, which are considered protective for premature mortality, were multiplied by −1.

The nonexercise test model was calibrated by comparing the mean incidence proportion of events for each quintile of predicted risk with the multiadjusted predicted risk obtained by the Cox models. Statistical tests were two sided and considered significant at a P value less than 0.05. All analyses were conducted using STATA version 12.1.


The distribution of clinical and demographic characteristics of the participants is presented in Table 1. The mean CRF was 34.5 mL·kg−1·min−1 (approximately 10 METs) in women and 47.2 mL·kg−1·min−1 (approximately 13 METs) in men. The medians and interquartile ranges were 35.5 mL·kg−1·min−1 (30.4–39.2) and 47.5 mL·kg−1·min−1 (42.5–51.9) in men and women, respectively. During a mean follow-up time of 24 yr (SD, 5.9 yr), 8752 deaths were registered (3453 CVD deaths).

Distribution of baseline data in healthy men and women from HUNT 1 (1984–1986).

CRF and mortality among participants younger than 60 yr at baseline

Higher CRF was consistently associated with lower CVD and all-cause mortality in both men and women younger than 60 yr of age at baseline (Tables 2 and 3). After adjustment for potential confounders, each MET-higher (approximately 3.5 mL·kg−1·min−1) CRF was associated with 21% lower CVD mortality both in men (95% CI, 17%–26%) and women (95% CI, 12%–29%). The corresponding lower risks for all-cause mortality were 15% (95% CI, 12%–17%) for men and 8% (95% CI, 3%–13%) for women.

HR for all-cause mortality and 95% CI.
HR for CVD mortality and 95% CI.

Men with a medium level (20th to 60th percentile) of CRF had 17% (95% CI, 8%–25%) lower risk of all-cause mortality and 25% (95% CI, 12%–37%) lower risk of CVD mortality compared with those in men with a low CRF (<20th percentile) (Tables 2 and 3). Corresponding reduced risks in women were 12% (95% CI, 0%–23%) for all-cause mortality and 26% (95% CI, 3%–43%) for CVD mortality (Tables 2 and 3). These reduced risks were dose dependent across the three CRF categories for both men and women.

CRF and mortality among participants 60 yr and older at baseline

Women 60 yr or older at baseline had 11% (95% CI, 6%–16%) reduced risk of CVD mortality per MET-higher CRF, whereas the corresponding estimates of reduced risk in men was a modest 5% (95% CI, 1%–9%). We observed a 19% (95% CI, 7%–30%) and 29% (95% CI, 17%–39%) reduction in CVD mortality in women with a medium and high level of CRF, respectively, compared with that in the lowest CRF category (Table 3). In men who were 60 yr or older at baseline, a 15% (95% CI, 1%–27%) lower risk of CVD mortality was observed among those with a medium level of CRF, whereas no further risk reduction was observed among those in the high CRF category (Table 3). For all-cause mortality, only modest risk reductions were observed among both men and women with a medium or high level of CRF compared with those within the lowest CRF category (Table 2). Excluding the first 5 yr of follow-up did not change results appreciably in the younger participants but somewhat strengthened the association in the elderly (see Tables, Supplemental Digital Content 3, HR for all-cause mortality and 95% CI after excluding the first 5 yr of follow-up,, and Supplemental Digital Content 4, HR for CVD mortality and 95% CI after excluding the first 5 yr of follow-up,

Age-relative CRF and mortality

Men younger than 60 yr of age at baseline and with a CRF below 85% of the age-predicted value had 82% (95% CI, 52%–118%) higher risk of all-cause mortality and approximately twofold risk of CVD mortality compared with those with a CRF at or above the age-predicted value (see Tables, Supplemental Digital Content 5, HR for all-cause mortality and 95% CI according to age-predicted CRF,, and Supplemental Digital Content 6, HR for CVD mortality and 95% CI according to age-predicted CRF, Men older than 60 yr of age at baseline and with a CRF below 85% of the age-predicted value had a 41% increased risk of all-cause mortality and 63% increased risk of CVD mortality compared with those in the reference group.

Women younger than 60 yr of age at baseline and with a CRF below 85% of the age-predicted value had 22% (95% CI, 2%–45%) higher all-cause mortality compared with mortality of those at or above the age-predicted value, whereas no increased risk was observed in those 60 yr or older at baseline. For CVD mortality, a modestly increased risk was observed in women with a CRF below 85% of the age-predicted value (≥60 yr: 24%; 95% CI, 6%–46%, and <60 yr: 21%; 95% CI, −15%–71%) compared with the risk of those at or above the age-predicted value (see Table, Supplemental Digital Content 6, HR for CVD mortality and 95% CI according to age-predicted CRF,

Assessment of model performance

Estimated CRF demonstrated better discriminate ability, as judged by the c-statistic or area under the receiver operating characteristic curve, compared with that in each of its components and also an aggregated sum of standardized scores (z-scores) for each modifiable component (Table 4). The sensitivity of the model to predict events occurring in the low CRF category was generally better for CVD mortality (58% for men and women younger than 60 yr at baseline) than that for all-cause mortality (50% and 45% for men and women younger than 60 yr, respectively). A low sensitivity was observed among participants 60 yr or older at baseline (range, 22%–26%). The specificity was generally high for both sex and age groups (range, 82%–85% in below 60 yr and 82%–95% in those above 60 yr). The mean predicted risks of all-cause and CVD mortality were very similar to the observed absolute risk with no systematic over- or underprediction within quintiles of predicted risks (Table 5).

Area under the curve (95% CI) for the estimated CRF algorithm and its modifiable components.
Ratio of predicted to observed risks of all-cause and CVD mortality for men and women across quintiles of estimated risks.


The present study demonstrates that a simple estimation of CRF predicts long-term risk of all-cause and CVD mortality in a general population of men and women who were younger than 60 yr at baseline. The reduced CVD mortality of 20%–22% per each MET-higher (approximately 3.5 mL·kg−1·min−1) CRF and the corresponding 8%–14% reduction in all-cause mortality are similar to estimates obtained in population studies that measured CRF directly by a maximal exercise test (26,27,29). In general, those studies reported risk reductions ranging from 10% to 20% per MET-higher CRF, and a meta-analysis of 33 studies reported a summarized relative risk reduction of 13% for all-cause mortality and 15% for CVD mortality (22).

Our findings among participants younger than 60 yr when CRF was estimated are also consistent with the results of a study that used a comparable nonexercise test of CRF. Stamatakis et al. (35) reported that one SD increase in estimated CRF was associated with approximately 15% reduction in all-cause mortality and 25% reduction in CVD mortality. Given the distribution of estimated CRF in that population, one SD corresponded to approximately 1.6–1.7 METs (or approximately 5–6 mL·kg−1·min−1). In the present study, 5 mL·kg−1·min−1 higher CRF was associated with 20% and 11% reduction in all-cause mortality in men and women younger than 60 yr at CRF assessment, respectively (see Table, Supplemental Digital Content 7, HR and 95% CI for the association between estimated CRF and all-cause and CVD mortality, stratified by age and sex, The corresponding reduction in CVD mortality was 22% in men and 28% in women per 5 mL·kg−1·min−1 higher estimated CRF. Interestingly, we recently showed that 5 mL·kg−1·min−1 lower CRF (measured directly by ergospirometry) was associated with approximately 56% higher odds of CVD risk factor clustering in 4631 healthy men and women age 20–89 yr (2).

The reason why estimated CRF was only weakly associated with mortality in participants who were 60 yr or older at baseline may partly be explained by the fact that some of the components of the CRF model (i.e., resting heart rate and BMI) are poor predictors of mortality in the elderly compared with those in younger persons (11,16,32,36). Furthermore, the misclassification of self-reported physical activity may be greater in the elderly and the questionnaire that was used has not been validated in an elderly cohort.

Because there is no generally accepted clinical categorization of CRF, an assessment relative to age-based reference values is an appealing approach. However, there is no general agreement on whether absolute or age-relative cut-off values provide the better prognostic power. An external validation study concluded that age- and sex-based nomograms with simple cut-off values (i.e., 85% of expected CRF for age) are superior to categorical descriptors of CRF (21). Others have found that categorical cut-offs provide better accuracy in terms of predicted survival than percentage of age-predicted values. (29) In the present study, both an estimated CRF below the 20th percentile and an estimated CRF below 85% of the age-predicted value were good predictors of long-term mortality.

An important feature of estimated CRF in a clinical setting is the ability to correctly classify events in the low fitness category because current evidence suggest a nonproportional association between exercise-tested CRF and mortality risk where the largest risk reduction is observed between low fitness and medium fitness groups (28,29). Because a large proportion of events in an otherwise healthy sample will occur in medium- or low-risk participants (frequently called the “prevention paradox”) (34), the obtained sensitivity of nearly 50% may be considered satisfactory. Although we emphasize that the aim of the current study was not to develop a new risk assessment model (4) but rather to validate the nonexercise test against prospective hard end points, it is intriguing that the overall discriminative ability of the nonexercise test was comparable with that of widely used risk models such as the “Framingham Risk Score” and the European “SCORE” algorithm (9,10). These scoring-systems are broadly recommended for risk prediction in clinical settings in the United States and European countries (33). Somewhat surprisingly, however, none of these established risk models have included estimates of fitness or physical activity in the scoring systems. However, studies have shown that exercise-tested CRF in combination with a risk model considerably improves risk prediction (13,25). We propose that an estimation of CRF through a nonexercise model may also provide a useful supplement to regularly measured risk factors and to currently recommended risk models. Moreover, CRF estimation provides a method that would place CRF and physical activity in a central position in individual patient counseling.

Strengths and limitations

The main strengths of the present study include a large population sample with a wide age range and virtually complete follow-up. The HUNT cohort is also recognized for having valid data and an extraordinary high participation rate (23). The cohort is quite homogenous with respect to ethnicity and socio-economic status, and a low net migration consolidates the high internal validity but may be a threat to generalizability. Hence, the nonexercise test should be validated and the ability to predict future risk evaluated in a separate population. Other limitations include the self-reported measure of physical activity included in the model, which is inherently prone to misclassification bias. Although self-report questionnaires may be the only practically useful method of incorporating physical activity in prediction models, the inclusion of objective measures of PA has been proposed to increase precision (5,6). Lastly, because it was derived using cross-sectional data, the nonexercise model has not been validated for the capability of capturing temporal changes of CRF, which would strengthen the clinical use in primary care. Recently, a new nonexercise test algorithm was derived from longitudinal CRF data from the Aerobics Center Longitudinal Study (18). This model may be more precise in capturing changes in CRF over time. However, given that this nonexercise test was not validated to hard end points, future studies should examine whether improvement in estimated CRF translates into better prognosis.


The simple nonexercise test of CRF that was used in this study could predict long-term risk of premature all-cause and CVD mortality with an accuracy that was similar to what has been obtained using directly measured CRF. The ability to discriminate risk was better for the CRF test as a whole compared with that in its components. Our findings are in accordance with previous findings and provide further evidence for the applicability of nonexercise testing models of CRF in clinical settings.

The authors are supported by grants from the K. G. Jebsen Foundation, Norwegian Research Council (U. W.), Norwegian Council on Cardiovascular Disease (B. N.), the Liaison Committee between the Central Norway Regional Health Authority and the Norwegian University of Science and Technology (I. J. and U. W.), and the Swedish Research Council and the Swedish Council of Working Life and Social Research (I. J.).

The Nord-Trøndelag health study (HUNT) is a collaboration between the HUNT Research Centre (Faculty of Medicine, Norwegian University of Science and Technology), the Nord-Trøndelag County Council, and the Norwegian Institute of Public Health. We are indebted to the participants of the HUNT study.

There are no further disclosures and no conflicts of interest to report.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Aspenes ST, Nauman J, Nilsen TI, Vatten LJ, Wisløff U. Physical activity as a long-term predictor of peak oxygen uptake: the HUNT Study. Med Sci Sports Exerc. 2011; 43 (9): 1675–9.
2. Aspenes ST, Nilsen TI, Skaug EA, et al. Peak oxygen uptake and cardiovascular risk factors in 4631 healthy women and men. Med Sci Sports Exerc. 2011; 43 (8): 1465–73.
3. Blair SN, Kohl HW 3rd, Paffenbarger RS Jr, Clark DG, Cooper KH, Gibbons LW. Physical fitness and all-cause mortality. A prospective study of healthy men and women. JAMA. 1989; 262 (17): 2395–402.
4. Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart. 2006; 92 (12): 1752–9.
5. Cao ZB, Miyatake N, Higuchi M, Ishikawa-Takata K, Miyachi M, Tabata I. Prediction of V˙O2max with daily step counts for Japanese adult women. Eur J Appl Physiol. 2009; 105 (2): 289–96.
6. Cao ZB, Miyatake N, Higuchi M, Miyachi M, Ishikawa-Takata K, Tabata I. Predicting V˙O2max with an objectively measured physical activity in Japanese women. Med Sci Sports Exerc. 2010; 42 (1): 179–86.
7. Church TS, Cheng YJ, Earnest CP, et al. Exercise capacity and body composition as predictors of mortality among men with diabetes. Diabetes Care. 2004; 27: 83–8.
8. Church T, Kampert JB, Gibbons LW, Barlow CE, Blair SN. Usefulness of cardiorespiratory fitness as a predictor of all-cause and cardiovascular disease mortality in men with systemic hypertension. Am J Cardiol. 2001; 88 (6): 651–6.
9. Conroy RM, Pyörälä K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003; 24 (11): 987–1003.
10. D’Agostino RB Sr, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008; 117 (6): 743–53.
11. Greenland P, Daviglus ML, Dyer AR, et al. Resting heart rate is a risk factor for cardiovascular and noncardiovascular mortality: the Chicago Heart Association Detection Project in Industry. Am J Epidemiol. 1999; 149 (9): 853–62.
12. Gulati M, Black HR, Shaw LJ, et al. The prognostic value of a nomogram for exercise capacity in women. N Engl J Med. 2005; 353 (5): 468–75.
13. Gupta S, Rohatgi A, Ayers CR, et al. Cardiorespiratory fitness and classification of risk of cardiovascular disease mortality. Circulation. 2011; 123 (13): 1377–83.
14. Hlatky MA, Greenland P, Arnett DK, et al. Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association. Circulation. 2009; 119 (17): 2408–16.
15. Holmen J, Midthjell K, Kruger O, Langhammer A, Holmen T, Bratberg G. The Nord-Trondelag Health Study 1995–97 (HUNT2): objectives, contents, methods and participation. Nor J Epidemiol. 2003; 13 (1): 19–32.
16. Hsia J, Larson JC, Ockene JK, et al. Resting heart rate as a low tech predictor of coronary events in women: prospective cohort study. BMJ. 2009; 338:b219.
17. Jackson AS, Blair SN, Mahar MT, Wier LT, Ross RM, Stuteville JE. Prediction of functional aerobic capacity without exercise testing. Med Sci Sports Exerc. 1990; 22 (6): 863–70.
18. Jackson AS, Sui X, O’Connor DP, et al. Longitudinal cardiorespiratory fitness algorithms for clinical settings. Am J Prev Med. 2012; 43 (5): 512–9.
19. Jurca R, Jackson AS, LaMonte MJ, et al. Assessing cardiorespiratory fitness without performing exercise testing. Am J Prev Med. 2005; 29 (3): 185–93.
20. Katzmarzyk PT, Church TS, Blair SN. Cardiorespiratory fitness attenuates the effects of the metabolic syndrome on all-cause and cardiovascular disease mortality in men. Arch Intern Med. 2004; 164 (10): 1092–7.
21. Kim ES, Ishwaran H, Blackstone E, Lauer MS. External prognostic validations and comparisons of age- and gender-adjusted exercise capacity predictions. J Am Coll Cardiol. 2007; 50 (19): 1867–75.
22. Kodama S, Saito K, Tanaka S, et al. Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis. JAMA. 2009; 301 (19): 2024–35.
23. Krokstad S, Langhammer A, Hveem K, et al. Cohort profile: the HUNT Study, Norway. Int J Epidemiol. 2013; 42: 968–77.
24. Laukkanen JA, Kurl S, Salonen R, Rauramaa R, Salonen JT. The predictive value of cardiorespiratory fitness for cardiovascular events in men with various risk profiles: a prospective population-based cohort study. Eur Heart J. 2004; 25 (16): 1428–37.
25. Laukkanen JA, Rauramaa R, Kurl S. Exercise workload, coronary risk evaluation and the risk of cardiovascular and all-cause death in middle-aged men. Eur J Cardiovasc Prev Rehabil. 2008; 15 (3): 285–92.
26. Laukkanen JA, Rauramaa R, Salonen JT, Kurl S. The predictive value of cardiorespiratory fitness combined with coronary risk evaluation and the risk of cardiovascular and all-cause death. J Int Med. 2007; 262 (2): 263–72.
27. Lee DC, Sui X, Artero EG, et al. Long-term effects of changes in cardiorespiratory fitness and body mass index on all-cause and cardiovascular disease mortality in men. Circulation. 2011; 124 (23): 2483–90.
28. Lee DC, Sui X, Ortega FB, et al. Comparisons of leisure-time physical activity and cardiorespiratory fitness as predictors of all-cause mortality in men and women. Br J Sports Med. 2011; 45 (6): 504–10.
29. Myers J, Prakash M, Froelicher V, Do D, Partington S, Atwood JE. Exercise capacity and mortality among men referred for exercise testing. N Engl J Med. 2002; 346 (11): 793–801.
30. Nauman J, Janszky I, Vatten LJ, Wisløff U. Temporal changes in resting heart rate and deaths from ischemic heart disease. JAMA. 2011; 306 (23): 2579–87.
31. Nes BM, Janszky I, Vatten LJ, Nilsen TI, Aspenes S, Wisløff U. Estimating V˙O2peak from a nonexercise prediction model; the HUNT Study, Norway. Med Sci Sports Exerc. 2011; 43 (11): 2024–30.
32. Okamura T, Hayakawa T, Kadowaki T, et al. Resting heart rate and cause-specific death in a 16.5-year cohort study of the Japanese general population. Am Heart J. 2004; 147 (6): 1024–32.
33. Perk J, De Backer G, Gohlke H, et al. European Guidelines on cardiovascular disease prevention in clinical practice (version 2012): the Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J. 2012; 33 (13): 1635–701.
34. Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001; 30 (3): 427–32.
35. Stamatakis E, Hamer M, O’Donovan G, Batty GD, Kivimaki M. A non-exercise testing method for estimating cardiorespiratory fitness: associations with all-cause and cardiovascular mortality in a pooled analysis of eight population-based cohorts. Eur Heart J. 2012; 34 (10): 750–8.
36. Vatten LJ, Nilsen TI, Romundstad PR, Drøyvold WB, Holmen J. Adiposity and physical activity as predictors of cardiovascular mortality. Eur J Cardiovasc Prev Rehabil. 2006; 13 (6): 909–15.


Supplemental Digital Content

© 2014 American College of Sports Medicine