Comparison of non-exercise cardiorespiratory fitness prediction equations in apparently healthy adults

Abstract

Aims

A recent scientific statement suggests clinicians should routinely assess cardiorespiratory fitness using at least non-exercise prediction equations. However, no study has comprehensively compared the many non-exercise cardiorespiratory fitness prediction equations to directly-measured cardiorespiratory fitness using data from a single cohort. Our purpose was to compare the accuracy of non-exercise prediction equations to directly-measured cardiorespiratory fitness and evaluate their ability to classify an individual's cardiorespiratory fitness.

Methods

The sample included 2529 tests from apparently healthy adults (42% female, aged 45.4 ± 13.1 years (mean±standard deviation). Estimated cardiorespiratory fitness from 28 distinct non-exercise prediction equations was compared with directly-measured cardiorespiratory fitness, determined from a cardiopulmonary exercise test. Analysis included the Benjamini–Hochberg procedure to compare estimated cardiorespiratory fitness with directly-measured cardiorespiratory fitness, Pearson product moment correlations, standard error of estimate values, and the percentage of participants correctly placed into three fitness categories.

Results

All of the estimated cardiorespiratory fitness values from the equations were correlated to directly measured cardiorespiratory fitness (p < 0.001) although the R² values ranged from 0.25–0.70 and the estimated cardiorespiratory fitness values from 27 out of 28 equations were statistically different compared with directly-measured cardiorespiratory fitness. The range of standard error of estimate values was 4.1–6.2 ml·kg⁻¹·min⁻¹. On average, only 52% of participants were correctly classified into the three fitness categories when using estimated cardiorespiratory fitness.

Conclusion

Differences exist between non-exercise prediction equations, which influences the accuracy of estimated cardiorespiratory fitness. The present analysis can assist researchers and clinicians with choosing a non-exercise prediction equation appropriate for epidemiological or population research. However, the error and misclassification associated with estimated cardiorespiratory fitness suggests future research is needed on the clinical utility of estimated cardiorespiratory fitness.

Keywords

Prognosis cardiopulmonary exercise test maximum oxygen consumption exercise test fitness algorithm

Introduction

Numerous studies indicate greater levels of cardiorespiratory fitness (CRF) in apparently healthy adults are associated with decreased mortality risk^1–4 and incidence of many chronic diseases.^3,5 This convincing body of evidence resulted in a recent scientific statement from the American Heart Association that suggested CRF be considered a clinical vital sign that is regularly assessed alongside other established risk factors.³ While the gold standard for directly measuring CRF is the cardiopulmonary exercise test, apparently healthy individuals rarely have the opportunity to perform this test. As a result, CRF is the only major risk factor that is not regularly assessed in the primary care setting. Therefore, to encourage the practice of CRF assessment in apparently healthy adults, recommendations suggest that at a minimum all routine clinical visits should include the determination of estimated CRF (eCRF) using a non-exercise prediction equation.³

Non-exercise prediction equations use the relationship that exists between CRF and various metrics to predict CRF. The most basic prediction equations include age, sex, and anthropometric variables.^6–11 More advanced equations also include measures of body composition.^10,12–18 Physical activity is another variable in many equations^12–21 due to the positive relationship between physical activity and CRF.^3,22 Most of these variables are routinely collected during clinical visits, making it easier to determine eCRF in such settings. Furthermore, previous research has shown eCRF is inversely associated with all-cause mortality.^23–26 However, there are inherent limitations associated with prediction equations. For example, equations are unable to determine the well-known genetic influence on CRF²⁷ and social desirability biases can influence the self-report aspects of some equations (i.e. physical activity levels).²⁸ Nonetheless, determinations of eCRF are still recommended due to the ease of calculation and their relationship with mortality.³

With the recommendation that eCRF be routinely measured within a clinical setting, a comprehensive comparison of prediction equations is needed to determine which equations provide the greatest predictive accuracy. The diversity of metrics used in the various equations could influence both the estimations and the clinical utility of eCRF. Recent reviews have compiled the correlations and errors associated with prediction equations.^3,29–31 However, comparisons between equations is problematic as the various equations were created and validated from different cohorts. Some validation research shows the accuracy of equations can vary from the originally published values,^23,32 yet the generalizability of equations when tested on other cohorts is largely unknown. Thus, the aim of this study was to evaluate the accuracy of non-exercise CRF prediction equations compared with directly-measured CRF using data from a single cohort of apparently healthy adults. To examine the prognostic potential of each equation, we also examined how well each equation categorized participants according to CRF level. We hypothesized that at the level of the individual, non-exercise CRF prediction equations would not accurately determine CRF.

Methods

Data from the Ball State Adult Fitness Longitudinal Lifestyle Study (BALL ST) cohort was used for this study. The sample included data from apparently healthy participants aged 18–80 years old who performed a comprehensive health and fitness assessment between 1 April 1968–31 March 2018. BALL ST participants were either self-referred to a community-based exercise program or were research participants in studies who provided written informed consent for their data to be used for research. Participants were defined as apparently healthy if they self-reported no lung disease, heart problems, or arterial disease and there was no abnormal electrocardiographic finding from the exercise test. Individuals with obesity, hypertension, diabetes, and/or dyslipidemia,³³ were included in the cohort. Tests were excluded if participants were taking a beta-blocker medication or data were missing for height, weight, or physical activity (a flow chart of inclusion/exclusion criteria is provided in Supplemental Material Figure 1). The protocol for this study was reviewed by the Ball State University Institutional Review Board and determined to be exempt as only de-identified data were used.

Non-exercise prediction equations

A literature search was conducted using the PubMed electronic database and the following terms: ‘predicted VO_2max,’ ‘predicted VO_2peak,’ ‘estimated VO_2max,’ ‘estimated VO_2peak,’ ‘non-exercise testing,’ and ‘non-exercise prediction’ (to 14 June 2019). Additional equations were identified from within previous reviews^3,29–31 and citations of these previous reviews were searched using Google Scholar. A total of 28 non-exercise prediction equations were included based on the following criteria: (a) the equation predicted maximum oxygen consumption (VO_2max) in both males and females; (b) walking or running was the exercise mode used to create the equation; (c) the equation was created from an apparently healthy adult cohort; and (d) variables within the equation were available from the data collected on the BALL ST cohort (e.g. excluded equations requiring variables such as thigh mass and perceived functional ability).

Variables within the equations included: sex, age, height, weight, body mass index (BMI), waist circumference, percentage body fat, smoking, resting heart rate, physical activity, dyslipidemia, hypertension, and diabetes. In the BALL ST cohort, anthropometrics were measured including height, weight, BMI, and waist circumference. Percentage body fat was determined using skin folds.³⁴ Participants also completed a health-history questionnaire, which provided self-reported physical activity status, smoking status, medication use, and medical history. Smoking status was recorded both on a two-level scale (‘yes’/’no’) as well as an eight-level scale that captured smoking habits.¹² When a participant was missing data needed within a prediction equation, no eCRF was calculated for that participant for that equation.

Self-reported physical activity data was collected using the BALL ST scale, which captures both lifestyle and occupational physical activity.¹² For the prediction equations that included measures of self-reported physical activity, various questionnaires and scales were used: BALL ST scale (1–6),¹² five-level Physical Activity Index (0–4),¹⁷ two-level Physical Activity Index (0–1),¹⁷ National Aeronautics and Space Administration (NASA)/Johnson Space Center (JSC) Physical Activity Scale (0–7),³⁵ NASA Physical Activity Status Scale (0–10),³⁶ and the HUNT questionnaire (0–45)¹⁴ (Supplemental Material Table 1). For the equations that did not use the BALL ST scale, physical activity measures were converted from the BALL ST scale to the other scales (Supplemental Material Table 2). Additionally, one prediction equation utilized the results from the International Physical Activity Questionnaire (IPAQ).²¹ For this equation, eCRF was only calculated for the subset of participants who completed an IPAQ.

The equations are provided in Supplemental Material Table 1 and a summary table of the variables included in each equation is provided in Supplemental Material Table 3. To estimate VO_2max, the variables for each equation were matched to those within the BALL ST cohort database. While the majority of equations estimated relative VO_2max (ml·kg⁻¹·min⁻¹), seven equations estimated maximal metabolic equivalents (METs).^15,17 The estimated MET results from these equations were multiplied by 3.5 to determine relative VO_2max.³³

Direct assessment of VO_2max

Participants performed a cardiopulmonary exercise test on a treadmill using a standardized protocol (i.e. Bruce,³³ Ball State University Bruce Ramp,³⁷ modified Balke-Ware,³³ and individualized protocols). The treadmill protocol was selected based on self-reported physical activity level with the goal of participants achieving maximal effort within 8–12 min. Ventilatory expired gas measurements were collected using a TrueOne 2400 computerized indirect calorimetry system (Parvo Medics, Sandy, Utah, USA) starting in 2002 and other systems used before as described previously.¹² The indirect calorimetry systems were calibrated prior to exercise testing according to the manufacturer's instructions. Respiratory data were averaged every 20 or 30 s and VO_2max was determined by averaging the highest 2–3 consecutively measured VO₂ values occurring in the last two minutes of the test. Participants were encouraged to exercise to volitional fatigue and only tests in which participants achieved a respiratory exchange ratio (RER) of ≥1.10 were included in the analysis.

Statistical analyses

Analyses were performed in R version 3.6.0 (R Core Team, Vienna, Austria). Differences between directly-measured CRF and eCRF were examined using the Benjamini–Hochberg procedure: p-values from t-tests of the directly-measured CRF and eCRF were ranked and compared with a critical value with a false discovery rate of 5%. This procedure accounts for multiple comparisons and reduces the risk of false positives.³⁸ The relationship between the different prediction equations and directly-measured CRF was examined by calculating the Pearson product moment correlations and the standard error of estimates (SEE). Bland-Altman plots were created to visualize the relationship between the different prediction equations and directly-measured CRF.

Evidence indicates prediction equations can become less accurate at the upper and lower ends of the CRF spectrum.³ Thus, in addition to examining the prediction equations over the entire sample, subsets of the sample were also examined. Participants were divided into fitness tertiles based on age and sex using the reference standards from the Fitness Registry and the Importance of Exercise National Database (FRIEND).³⁹ In accordance with previous research,¹ participants were classified as having ‘lower’ CRF if they were below the 33rd percentile, ‘average’ CRF if between the 33rd–66th percentile, and ‘higher’ CRF if above the 66th percentile. Additional analysis also examined participants in age tertiles (<40 (‘younger’), 40–60 (‘middle-aged’), > 60 years old (‘older’)). Statistical significance was set at p < 0.05, two-tailed. Data are presented throughout the article as mean ± standard deviation.

Results

Descriptive characteristics of the sample are provided in Table 1. The sample included a total of 2529 tests from participants with an average age of 45.4 ± 13.1 years (42% female). Supplemental Material Table 4 summarizes eCRF from the different equations for the entire sample as well as divided by sex. As expected, the eCRFs from all equations were correlated to directly measured CRF (p < 0.001) when examined over the entire cohort as well as for males and females assessed separately. Within the entire cohort, the R² values had a range of 0.25–0.70 while the SEE values ranged from 4.1–6.2 ml·kg⁻¹·min⁻¹. The range of SEE values was 1.6–5.8 ml·kg⁻¹·min⁻¹ for males and 1.8–5.7 ml·kg⁻¹·min⁻¹ for females. Estimated CRF was statistically different from directly-measured CRF for 27 out of 28 equations when examining the entire cohort, 25 of the equations when examining only males, and 22 of the equations when examining only females. Bland-Altman plots for each equation are provided in Supplemental Material Figure 2.

Table 1.

Descriptive characteristics of the cohort.

	Males (n = 1475)	Females (n = 1054)	All (n = 2529)
Age (years)	46.3 ± 12.9	44.1 ± 13.2^a	45.4 ± 13.1
Height (m)	1.8 ± 0.1	1.6 ± 0.1^a	1.7 ± 0.1
Weight (kg)	89.9 ± 16.8	73.6 ± 17.4^a	83.1 ± 18.9
BMI (kg·m⁻²)	28.1 ± 4.9	27.1 ± 6^a	27.7 ± 5.4
Waist circumference (cm)	96.7 ± 12.8	83.5 ± 14.2^a	91.1 ± 14.9
Body fat (%)	23.5 ± 7.2	34.2 ± 7.6^a	28.4 ± 9.1
Smoker (%)	7.3	4.9^a	6.3
Diabetes (%)	6	4	5
Dyslipidemia (%)	65	40	55
Hypertension (%)	32	19	27
Obesity (%)	36	35	35
Ethnicity (%)
Asian or Pacific Islander	0.3	0.5	0.4
Black, not of Hispanic origin	0	0.4	0.2
Hispanic	0.1	0	<0.1
White, not of Hispanic origin	99.6	99.1	99.4
Resting heart rate (beat·min⁻¹)	66.9 ± 11.3	68.9 ± 9.7^a	67.7 ± 10.7
VO_2max (ml·kg⁻¹·min⁻¹)	36.1 ± 9.6	28.4 ± 8.2^a	32.9 ± 9.8
Maximum heart rate (beat·min⁻¹)	176.3 ± 16.1	176.9 ± 15	176.5 ± 15.6
Maximum RER	1.2 ± 0.1	1.2 ± 0.1	1.2 ± 0.1
Peak RPE	18.4 ± 1.7	18.5 ± 1.7	18.4 ± 1.7
BALL ST Physical Activity Scale (%)
1	30.9%	39.9%	34.7%
2	15.8%	20.1%	17.6%
3	9.7%	7.5%	8.8%
4	0.9%	0.3%	0.6%
5	37.6%	29.2%	34.1%
6	5.2%	2.9%	4.3%
IPAQ vigorous physical activity (MET·min·wk⁻¹)^b	1073.2 ± 1507.1 (n = 199)	736.8 ± 1281^a (n = 246)	887.2 ± 1395.1 (n = 445)

BMI: body mass index; IPAQ: International Physical Activity Questionnaire; RER: respiratory exchange ratio; RPE: rating of perceived exertion (6–20 Borg scale); VO_2max: maximum oxygen consumption.

Significantly different from males (p<0.05).

Not all BALL ST cohort participants completed the IPAQ.

A summary of the prediction equation results according to age group are presented in Supplemental Material Table 5. Within each age group, the eCRFs from all equations were correlated to directly measured CRF (p < 0.001) with a range of R² values of 0.36–0.68, 0.32–0.67, and 0.26–0.58 for the younger, middle-aged, and older groups, respectively. Estimated CRF was statistically different from directly-measured CRF for 17 out of 28 equations when examining only the younger group, 25 of the equations when examining only the middle-aged group, and 22 of the equations when examining only the older group.

Supplemental Material Table 6 summarizes eCRF from the different equations according to fitness group. Within each fitness group, the eCRFs from all equations were correlated to directly measured CRF (p < 0.001) with a range of R² values of 0.22–0.69, 0.29–0.84, and 0.30–0.82 for the lower, average, and higher fitness groups, respectively. Estimated CRF was statistically different from directly-measured CRF for 26 out of 28 equations when examining only the lower fitness group, 25 of the equations when examining only the average fitness group, and 24 of the equations when examining only the higher fitness group.

Table 2 presents the percentage of participants correctly and incorrectly placed into the three fitness categories when using the prediction equations. On average, the prediction equations correctly categorized 52% of participants. On average, the prediction equations incorrectly placed 30% of participants into a higher fitness category and incorrectly placed 18% into a lower fitness category.

Table 2.

Percentage of participants classified into the three fitness categories according to measured cardiorespiratory fitness (CRF) and estimated CRF. Participants were divided into fitness groups based on age and sex using the reference standards from the Fitness Registry and the Importance of Exercise National Database (FRIEND).³⁹ Participants were classified as having ‘lower’ CRF if they were below the 33rd percentile, ‘average’ CRF if between the 33rd and 66th percentile, and ‘higher’ CRF if above the 66th percentile. Correct classifications are highlighted in bold.

	Measured lower CRF			Measured average CRF			Measured higher CRF
	Estimated lower CRF	Estimated average CRF	Estimated higher CRF	Estimated lower CRF	Estimated average CRF	Estimated higher CRF	Estimated lower CRF	Estimated average CRF	Estimated higher CRF
No physical activity measure
Riddle et al., 1980⁷	91	8	1	68	29	3	44	42	15
Wasserman et al., 1994⁹	68	31	1	23	72	5	4	74	22
Jang et al., 2012 (Model 1)⁸	2	19	79	0	11	89	0	10	90
Baynard et al., 2016 (BMI)¹⁰	43	49	8	9	63	27	1	42	57
Baynard et al., 2016 (WC)¹⁰	48	44	8	10	63	27	2	38	60
Myers et al., 2017⁶	44	50	6	11	66	23	2	47	51
de Souza et al., 2018¹¹	56	41	3	17	68	16	3	52	45
Includes measure of physical activity
Jackson et al., 1990 (BMI)¹³	78	19	3	49	33	18	17	29	54
Jackson et al., 1990 (%fat)¹³	80	18	2	53	34	13	20	40	40
Heil et al., 1995¹⁸	44	42	14	12	49	38	2	25	73
Whaley et al., 1995 (BMI)¹²	58	36	6	18	52	30	2	27	71
Whaley et al., 1995 (%fat)¹²	65	30	5	30	48	22	8	32	61
Matthews et al., 1999²⁰	49	42	9	11	52	37	2	22	77
Jurca et al., 2005 (ACLS)¹⁵	57	37	6	15	56	29	2	24	74
Jurca et al., 2005 (ADNFS)¹⁵	25	63	13	3	50	47	0	13	87
Jurca et al., 2005 (NASA)¹⁵	64	34	2	22	61	17	4	41	55
Wier et al., 2006 (BMI)¹⁶	71	25	4	32	47	21	8	33	58
Wier et al., 2006 (%fat)¹⁶	72	25	3	42	42	16	13	39	47
Wier et al., 2006 (WC)¹⁶	70	26	4	32	48	20	7	37	55
Nes et al., 2011¹⁴	16	49	34	1	21	78	0	3	97
Schembre and Riebe 2011²¹	3	28	69	0	27	72	0	25	75
Cáceres et al., 2012 (9 Level)¹⁹	54	41	6	20	57	23	7	54	39
Cáceres et al., 2012 (5 Level)¹⁹	43	49	8	12	62	26	3	59	37
Jackson et al., 2012 (BMI, 5 Level PA)¹⁷	32	45	23	3	34	63	0	14	86
Jackson et al., 2012 (%fat, 5 Level PA)¹⁷	21	51	28	3	33	64	0	14	86
Jackson et al., 2012 (BMI, 2 Level PA)¹⁷	32	46	22	3	35	62	0	14	86
Jackson et al., 2012 (%fat, 2 Level PA)¹⁷	21	53	27	3	34	63	0	14	86
Jang et al., 2012 (Model 2)⁸	2	26	72	0	7	93	0	2	98

ACLS: equation based on data from the Aerobics Center Longitudinal Study; ADNFS: Allied Dunbar National Fitness Survey; BMI: body mass index; %fat: percentage body fat; NASA: National Aeronautics and Space Administration; PA: physical activity; WC: waist circumference.

Discussion

A recent recommendation suggests that, at a minimum, clinicians should routinely determine a patient's eCRF using non-exercise prediction equations.³ The present study compared multiple non-exercise prediction equations to directly-measured CRF within a single cohort of apparently healthy adults to determine which equations are the most valid to use. As expected, the prediction equations were all significantly correlated to directly measured CRF although differences in the correlations and SEE values were observed. Some of this variability is explained by the diversity of metrics used in the equations. For example, equations that included measures of physical activity typically resulted in eCRF values that were better correlated to directly-measured CRF. This finding is not surprising given that physical activity is the major modifiable contributor to changes in CRF.³ On the other hand, only small differences were observed between equations that included BMI compared with those including percentage body fat. Although percentage body fat is a better determinant of body composition, these findings suggest BMI is an adequate variable for inclusion in CRF prediction equations. It is important to note that the determinations of percentage body fat were made using skin-fold measurements to conform with the methods of the prediction equations. It is possible that more accurate measures of percentage body fat could improve correlations, but the needed resources for such measurements could also limit the applicability within a research or clinical setting.

While many of the correlations and SEE values observed in the present study are similar or better than those reported in the original publications (Supplemental Material Table 1), the accuracy of the prediction equations varied when the analyses were conducted on the different subgroups. In general, correlations tended to be lower for those in the older age group and those with lower fitness. This is especially concerning as older adults and lower fitness individuals are at the highest risk for adverse health events and could potentially benefit the most from knowledge of CRF status. Additionally, no equation performed accurately across all the different subgroup analyses. For some of these prediction equations though, the mean differences between eCRF and CRF were relatively minor (<1.0 ml·kg⁻¹·min⁻¹), suggesting that some of the prediction equations can reasonably identify group means.

Prediction equations that accurately estimate a group or population mean can be useful in a research setting. However, when using prediction equations in a clinical setting, the accurate estimation of individual values is more important. Therefore, in the clinical setting, the measures of error surrounding eCRF are of greater consequence as errors in eCRF can impact diagnosis and treatment plans for the individual. In the overall cohort analysis, the range of SEE values from all of the prediction equations was 4.1–6.2 ml·kg⁻¹·min⁻¹, which is equivalent to 1.2–1.8 METs. Previous research has shown a one MET improvement significantly reduces mortality,^1,3 so a degree of error greater than one MET could impact risk classification and treatment. Accordingly, when using eCRF to categorize participants into fitness groups, there was a considerable degree of misclassification. The correct classification of participants into fitness groups averaged only 52% with a range of 34–62%. This level of misclassification is similar to the findings for the Nes et al. equation which reported correctly classifying ∼54% of participants¹⁴ and the Jackson et al. two-level BMI equation¹⁷ which correctly classified 58% of participants.²³

Misclassification at the individual level is worrisome as the suggested reason for measuring eCRF in the clinical setting is that it can improve risk categorization and patient management. Previous research has shown low levels of eCRF are associated with increased risk for mortality,^23–26 yet many prediction equations have a limited ability to correctly identify these individuals. On average, the prediction equations were only able to correctly classify 47% of participants as having lower CRF. The Riddle et al. equation⁷ was notable in that it correctly classified 91% of participants with lower CRF, yet this equation also misclassified a large number of participants into the lower CRF category. This suggests that non-exercise prediction equations may not improve patient risk categorization and if eCRF is to be used in a clinical setting as recommended, more accurate prediction equations are needed.

There are a number of factors to consider for the creation of future prediction equations. The average age of the cohorts used to create the current prediction equations was typically 40–50 years old.^{6,10–14,16,17,20} This could limit the accuracy of these equations in younger or older individuals. While age is included in the prediction equations, age group specific equations may improve eCRF accuracy. Additionally, ethnicity may also play a role in the accuracy of prediction equations. The ethnicity in the present study was predominantly white and the poorer performing Jang et al. equations⁸ were developed using an Asian cohort, which suggests there is potential for ethnicity differences in prediction equation accuracy. Future research should include a variety of ethnicities as many of the current prediction equations were created from a predominantly white cohort^15–17,20 and limited validation has occurred in other ethnicities. Lastly, a significant determinant of CRF is genetics²⁷ and the non-exercise prediction equations do not account for this important variable. With an increasing prevalence of genetic testing, future research should explore how to include this variable and whether it is feasible to include within a CRF prediction equation.

The strengths of the present study are that 28 prediction equations were compared within a single cohort, which included a diverse range of ages and fitness levels. However, there were limitations. The study compared only prediction equations developed for treadmill testing. Cycling is another common mode of exercise and often results in a different CRF value compared with exercise on a treadmill.³³ As such, future research should compare prediction equations using cycling. Also, data from some of the participants in the present cohort were part of the datasets used to create six of the prediction equations.^6,10–12 Additionally, different indirect calorimetry systems were used in the studies that developed the prediction equations as well as the present study. The degree of variance between systems is not well-known but could influence the relationships between eCRF and directly-measured CRF. The present study also excluded those with lung disease, heart problems, or arterial disease and the accuracy of prediction equations may be different for individuals with these conditions. It is also possible that the conversion of physical activity scales during analysis may have increased the error associated with certain equations. However, the correlations found in the present study were similar to many of those reported in the original publications. Furthermore, different results were not observed with the Whaley et al. equations,¹² which used the original physical activity coding (59% of participants correctly categorized into the fitness groups).

In conclusion, all of the non-exercise prediction equations determined eCRFs that were significantly correlated to directly-measured CRF, yet variability was observed between the equations. Depending on whether analyses occurred over the entire cohort or subgroups of the cohort, varying levels of accuracy were also observed. Although some equations accurately estimated the CRF group mean, each equation had a low level of accuracy for placing participants into fitness categories. Thus, while prediction equations are useful for epidemiological or population research, further work is needed to determine the prognostic utility of eCRF within a clinical setting.

Supplemental Material

CPR881242 Supplemental Material - Supplemental material for Comparison of non-exercise cardiorespiratory fitness prediction equations in apparently healthy adults

Supplemental material, CPR881242 Supplemental Material for Comparison of non-exercise cardiorespiratory fitness prediction equations in apparently healthy adults by James E Peterman, Mitchell H Whaley, Matthew P Harber, Bradley S Fleenor, Mary T Imboden, Jonathan Myers, Ross Arena and Leonard A Kaminsky in European Journal of Preventive Cardiology

Footnotes

Acknowledgements

The authors would like to acknowledge the efforts of Leroy ‘Bud’ Getchell who established the Adult Physical Fitness Program at Ball State University and began the data collection for the BALL ST cohort.

Author contribution

JP, MHW, LK, and MH contributed to the conception or design of the work. All authors contributed to the acquisition, analysis, or interpretation of data for the work, drafted and critically revised the manuscript, gave final approval, and agree to be accountable for all aspects of work ensuring integrity and accuracy.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support for this project was provided, in part, from an American Heart Association Award #AIREA33930023 (M. Harber, PI).

Supplemental Material

Supplemental material for this article is available online.

References

Imboden

Harber

Whaley

, et al. cardiorespiratory fitness and mortality in healthy men and women. J Am Coll Cardiol 2018; 72: 2283–2292.

Blair

Kohl

3rd Paffenbarger

Jr , et al. Physical fitness and all-cause mortality. A prospective study of healthy men and women. JAMA 1989; 262: 2395–2401.

Ross

Blair

Arena

, et al. Importance of assessing cardiorespiratory fitness in clinical practice: A case for fitness as a clinical vital sign: A scientific statement from the American Heart Association. Circulation 2016; 134: e653–e699.

Laukkanen

Araujo

CGS

Kurl

, et al. Relative peak exercise oxygen pulse is related to sudden cardiac death, cardiovascular and all-cause mortality in middle-aged men. Eur J Prev Cardiol 2018; 25: 772–782.

Ekblom-Bak

Ekblom

Fagman

, et al. Fitness attenuates the prevalence of increased coronary artery calcium in individuals with metabolic syndrome. Eur J Prev Cardiol 2018; 25: 309–316.

Myers

Kaminsky

Lima

, et al. A reference equation for normal standards for VO_2 max: Analysis from the Fitness Registry and the Importance of Exercise National Database (FRIEND registry). Prog Cardiovasc Dis 2017; 60: 21–29.

Riddle

Younes

Remmers

, et al. Graphical analysis of patient performance in the pulmonary function laboratory. Proc Annu Symp Comput Appl Med Care 1980; 1: 283–290.

Jang

T-W

Park

S-G

Kim

H-R

, et al. Estimation of maximal oxygen uptake without exercise testing in Korean healthy adult workers. Tohoku J Exp Med 2012; 227: 313–319.

Wasserman

Hansen

Sue

, et al. Principles of exercise testing and interpretation, 2nd ed. Malvern, PA, USA: Lea and Febiger, 1994. .

10.

Baynard

Arena

Myers

, et al. The role of body habitus in predicting cardiorespiratory fitness: The FRIEND registry. Int J Sports Med 2016; 37: 863–869.

11.

de Souza

ESCG

Kaminsky

Arena

, et al. A reference equation for maximal aerobic power for treadmill and cycle ergometer exercise testing: Analysis from the FRIEND registry. Eur J Prev Cardiol 2018; 25: 742–750.

12.

Whaley

Kaminsky

Dwyer

, et al. Failure of predicted VO_2peak to discriminate physical fitness in epidemiological studies. Med Sci Sports Exerc 1995; 27: 85–91.

13.

Jackson

Blair

Mahar

, et al. Prediction of functional aerobic capacity without exercise testing. Med Sci Sports Exerc 1990; 22: 863–870.

14.

Nes

Janszky

Vatten

, et al. Estimating VO_2peak from a nonexercise prediction model: The HUNT study, Norway. Med Sci Sports Exerc 2011; 43: 2024–2030.

15.

Jurca

Jackson

LaMonte

, et al. Assessing cardiorespiratory fitness without performing exercise testing. Am J Prev Med 2005; 29: 185–193.

16.

Wier

Jackson

Ayers

, et al. Nonexercise models for estimating VO_2max with waist girth, percent fat, or BMI. Med Sci Sports Exerc 2006; 38: 555–561.

17.

Jackson

Sui

O'Connor

, et al. Longitudinal cardiorespiratory fitness algorithms for clinical settings. Am J Prev Med 2012; 43: 512–519.

18.

Heil

Freedson

Ahlquist

, et al. Nonexercise regression models to estimate peak oxygen consumption. Med Sci Sports Exerc 1995; 27: 599–606.

19.

Cáceres

Ulbrich

Panigas

, et al. Equações de predição da aptidão cardiorrespiratória de adultos sem teste de exercícios físicos. Rev Bras Cineantropom Desempenho Hum 2012; 14: 287–2. .

20.

Matthews

Heil

Freedson

, et al. Classification of cardiorespiratory fitness without exercise testing. Med Sci Sports Exerc 1999; 31: 486–493.

21.

Schembre

Riebe

. Non-exercise estimation of VO(2)_max using the International Physical Activity Questionnaire. Meas Phys Educ Exerc Sci 2011; 15: 168–181.

22.

Garber

Blissmer

Deschenes

, et al. American College of Sports Medicine position stand. Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: Guidance for prescribing exercise. Med Sci Sports Exerc 2011; 43: 1334–1359.

23.

Artero

Jackson

Sui

, et al. Longitudinal algorithms to estimate cardiorespiratory fitness: Associations with nonfatal cardiovascular disease and disease-specific mortality. J Am Coll Cardiol 2014; 63: 2289–2296.

24.

Nes

Vatten

Nauman

, et al. A simple nonexercise model of cardiorespiratory fitness predicts long-term mortality. Med Sci Sports Exerc 2014; 46: 1159–1165.

25.

Wang

Chen

Zhang

, et al. Nonexercise estimated cardiorespiratory fitness and all-cancer mortality: The NHANES III study. Mayo Clin Proc 2018; 93: 848–856.

26.

Shigdel

Dalen

Sui

, et al. Cardiorespiratory fitness and the risk of first acute myocardial infarction: The HUNT study. J Am Heart Assoc 2019; 8: e010293.

27.

Bouchard

Daw

Rice

, et al. Familial resemblance for VO_2max in the sedentary state: The HERITAGE family study. Med Sci Sports Exerc 1998; 30: 252–258.

28.

Sallis

Saelens

. Assessment of physical activity by self-report: Status, limitations, and future directions. Res Q Exerc Sport 2000; 71: S1–S14.

29.

Maranhão Neto

Lourenço

PMC

Farinatti

PTV

. Non-exercise models for prediction of aerobic fitness and applicability on epidemiological studies: Descriptive review and analysis of the studies. Rev Bras Med Esporte 2003; 9: 304–314. .

30.

Maranhão Neto

GdA

Lourenço

PMC

Farinatti

PdTV

. Equações de predição da aptidão cardiorrespiratória sem testes de exercício e sua aplicabilidade em estudos epidemiológicos: Uma revisão sistemática. Cad Saude Publica 2004; 20: 48–56.

31.

Paap

Takken

. Reference values for cardiopulmonary exercise testing in healthy adults: A systematic review. Expert Rev Cardiovasc Ther 2014; 12: 1439–1453.

32.

Sanada

Midorikawa

Yasuda

, et al. Development of nonexercise prediction models of maximal oxygen uptake in healthy Japanese young men. Eur J Appl Physiol 2007; 99: 143–148.

33.

Riebe D, Ehrman JK, Liguori G, et al. ACSM's guidelines for exercise testing and prescription. 10th ed. Philadelphia, PA: Wolters Kluwer, 2018.

34.

Siri WE. Body composition from fluid spaces and density: Analysis of methods in technologies for measuring body composition. Washington DC: National Academy of Science, National Research Council, 1961.

35.

Ross RM and Jackson AS. Exercise concepts, calculations, and computer applications. Carmel, IN: Benchmark Press, 1990.

36.

Wier

Ayers

Jackson

, et al. Determining the amount of physical activity needed for long-term weight control. Int J Obes Relat Metab Disord 2001; 25: 613–621.

37.

Kaminsky

Whaley

. Evaluation of a new standardized ramp protocol: The BSU/Bruce Ramp protocol. J Cardiopulm Rehabil 1998; 18: 438–444.

38.

Benjamini

Hochberg

. Controlling the false discovery rate – a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 1995; 57: 289–300.

39.

Kaminsky

Arena

Myers

. Reference standards for cardiorespiratory fitness measured with cardiopulmonary exercise testing: Data from the Fitness Registry and the Importance of Exercise National Database. Mayo Clin Proc 2015; 90: 1515–1523.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.35 MB

Comparison of non-exercise cardiorespiratory fitness prediction equations in apparently healthy adults

Abstract

Aims

Methods

Results

Conclusion

Keywords

Introduction

Methods

Non-exercise prediction equations

Direct assessment of VO2max

Statistical analyses

Results

Discussion

Supplemental Material

CPR881242 Supplemental Material - Supplemental material for Comparison of non-exercise cardiorespiratory fitness prediction equations in apparently healthy adults

Footnotes

Acknowledgements

Author contribution

Declaration of conflicting interests

Funding

Supplemental Material

References

Supplementary Material

Direct assessment of VO_2max