Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
PII: S0033-0620(18)30131-2
DOI: doi:10.1016/j.pcad.2018.07.004
Reference: YPCAD 902
To appear in: Progress in Cardiovascular Diseases
Received date: 2 July 2018
Accepted date: 2 July 2018
Please cite this article as: Vaughn W. Barry, Jennifer L. Caputo, Minsoo Kang , The Joint
Association of Fitness and Fatness on Cardiovascular Disease Mortality: A Meta-Analysis.
Ypcad (2018), doi:10.1016/j.pcad.2018.07.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT
Title
The Joint Association of Fitness and Fatness on Cardiovascular Disease Mortality: A Meta-
Analysis
Author Names
T
Vaughn W. Barry1, PhD; Jennifer L. Caputo1, PhD; Minsoo Kang2, PhD
IP
CR
Affiliations
1
Health and Human Performance
2
Exercise Science and Recreation Management
Oxford, Mississippi
CE
AC
ACCEPTED MANUSCRIPT
P. O. Box 96
T
Murfreesboro, TN 37132
IP
Phone: (615) 898-5535
CR
Fax: (615) 898-5550
Vaughn.Barry@mtsu.edu
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
T
CVD = cardiovascular disease
IP
NOS = Newcastle-Ottawa Scale
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
Abstract
The joint association between cardiorespiratory fitness (CRF) and body mass index
(BMI) on cardiovascular disease (CVD) mortality was determined. PubMed and CINAHL were
searched following PRISMA guidelines. Included studies were prospective, had objective
assessments of maximal CRF and BMI, and compared the joint impact of CRF and BMI on CVD
T
mortality risk to normal weight, fit referents. Pooled hazard ratios and 95% confidence intervals
IP
were calculated from eight articles with nine independent groups using a random effects model.
CR
Unfit individuals had two to three times the risk of mortality across all levels of BMI.
Overweight and obese-fit individuals had 25% and 42% increased mortality risk, respectively,
US
compared to normal weight-fit individuals. However, for the obese-fit group, a one study
AN
removed analysis for five studies resulted in non-significant changes in mortality risk.
Researchers, clinicians, and public health officials are encouraged to employ CRF interventions
M
Introduction
overtime (1), CVD mortality remained the leading cause of death in the US in 2015, accounting
for nearly a quarter (i.e. 23.4%) of all deaths nation-wide (2). Because of the significant health
and economic burdens related to CVD death, the need for primary and secondary interventions
T
remains.
IP
Obesity, often illustrated at the population level by body mass index (BMI), is a noted
CR
risk factor for CVD. Based on BMI, 37.7% of US adults are considered obese (3) and annual
heath care costs for obesity-related health issues ranges from $147 - $220 billon (4-5). However,
US
the relationship of BMI to CVD mortality has been challenged in the literature (6-8). Because of
AN
the magnitude of the problem and the dichotomy of research outcomes relating BMI to CVD
Cardiorespiratory fitness (CRF) is also related to CVD mortality (11-12). While this
ED
relationship is not under debate in the literature, the magnitude is being investigated. Several
researchers have shown higher CRF levels associated with lower CVD mortality (3-4, 6, 13-15).
PT
Further, after adjusting for or jointly assessing BMI levels, the relationship between CRF and
CE
The joint association between CRF and BMI relative to all-cause mortality has been
AC
impacted all-cause mortality risk while altering BMI did not (16). While the joint association of
CRF and BMI on CVD mortality has been investigated (3-6, 13-15, 19-20), a summary of
existing literature was not found. Therefore, the purpose of this manuscript was to jointly assess
the relationship of CRF and of BMI on CVD mortality using meta-analytical techniques.
ACCEPTED MANUSCRIPT
Methods
Literature Search
The PRISMA guidelines were followed during manuscript preparation (21). The
literature was reviewed using PubMed and CINAHL search engines using the following terms:
T
OR “maximal oxygen uptake” OR “stress test” OR “maximal treadmill test”) AND (“Body
IP
composition” OR “BMI” OR “body mass index” OR “obesity” OR “adiposity”) AND (“diabetes
CR
mellitus” OR “chronic disease” OR “cancer” OR “cardiovascular” OR “cerebrovascular” OR
US
“fatality” OR “fatal”)). The search was further limited by three criteria: 1) published between
AN
January 1989 and December 2017, 2) population was adult humans, and 3) written in English
language. During the literature search, articles were included if they were prospective,
M
cardiorespiratory fitness was objectively measured through a maximal exercise test, height and
ED
body mass were measured with a stadiometer and a scale, respectively, and the joint impact of
CRF and BMI on cardiovascular disease mortality or a specific subtype of CVD (i.e. heart
PT
failure) was analyzed. Finally, the reference group for all studies was normal weight, fit
CE
individuals.
During the first stage of the literature search, 1,018 and 543 titles and abstracts from
AC
CINAHL and PubMed were assessed, respectively. Two researchers independently assessed all
of these titles and/or abstracts and any duplicate articles between the lists were removed, leaving
a combined pool of 22 articles. Next, the references pages of these sources were assessed to
document additional titles. Upon further review of the content of each article for the inclusion
ACCEPTED MANUSCRIPT
criteria and an attempt to contact one author for eligibility dependent information, a final pool of
Article Quality
The Newcastle-Ottawa Scale (NOS) for Nonrandomized Studies (22) was used to assess
the quality of the studies in the meta-analysis. This validated scale is used to assess three
T
components of research articles: selection (four questions), comparability (two questions), and
IP
outcome (three questions). A point was available for each question, with a total of nine possible
CR
points. A study was considered to have moderate to high quality if it received ≥ 6 points and low
Data Analysis
US
AN
Data within the eligible articles were categorized into fitness (i.e. fit and unfit) and BMI
(i.e. normal weight, overweight, and obese) categories by the respective study authors.
M
Subgroups within the fitness and the fatness categories were combined into five comparison
ED
groups: unfit and normal weight, fit and overweight, unfit and overweight, fit and obese, and
unfit and obese. Hazard ratios and 95% confidence intervals were recorded from each study. In
PT
two articles, low, moderate, and high fitness levels were included in the analyses (6, 15).
CE
However, a dichotomous fitness variable (i.e. fit and unfit) was used in the meta-analysis.
Therefore, the Hamling method was used to combine the hazard ratios and 95% confidence
AC
intervals for the moderate and high fitness categories from these two studies (23).
Englewood, New Jersey), version 2.2.064, was used to calculate a pooled hazard ratio and 95%
confidence intervals using a random-effects model for each comparison group (i.e. unfit and
normal weight, fit and overweight, unfit and overweight, fit and obese and unfit and obese). An
ACCEPTED MANUSCRIPT
I2 statistic was reported for each analysis to assess heterogeneity among study findings. To assess
publication bias across studies, the Egger test was completed (24). To assess the sensitivity of the
study outcomes, a one study removed analysis was completed. In this analysis, the data were
reanalyzed with each study in the total pool removed from the analysis. This allowed the
determination of the impact of individual studies to the overall analysis. Finally, a moderator
T
analysis was completed to assess the effects of differing study characteristics on study outcomes.
IP
In completing this analysis, length of follow-up (> 14 years), baseline health status (lack of
CR
chronic disease), and confounder control (accounted for ≥ 3 cardiovascular disease risk factors;
(25), were assessed as dichotomous variables (yes or no). Additionally, a meta-regression was
US
completed to assess the effect of age. An α level of 0.05 was used to indicate statistical
AN
significance.
M
Results
ED
All eight articles scored ≥ 6 on the NOS scale, an indication of high quality. There was
no publication bias across studies as assessed by the Egger test, for each of the five comparison
PT
groups (see Table 1). The hazard ratios, 95% confidence intervals, forest plots, and pooled ratios
CE
for the five comparison groups relative to the reference group of normal weight, fit are displayed
in Figures 2, 3, and 4. In addition, the relative weight of each study is indicated within each
AC
figure. The risk of CVD death for normal weight, unfit individuals was twice as high relative to
the reference group, i.e. normal weight, unfit individuals (see Figure 2). A one study removed
sensitivity analysis did not significantly alter the pooled hazard ratio. Furthermore, the three
dichotomous moderator analyses (follow-up, baseline health status, confounder control) did not
significantly affect the main outcome, nor did age influence this relationship.
ACCEPTED MANUSCRIPT
There was an additional increased risk of death due to CVD for the overweight groups
(see Figure 3). Specifically, the risk for those who were overweight and unfit was doubled
whereas the risk for overweight and fit relative to the normal weight fit reference was only 25%
elevated. The sensitivity analysis did not significantly alter the pooled hazard ratios for the
overweight groups. Further, the dichotomous moderator analyses did not significantly alter the
T
outcomes. However, age had a significant inverse effect (slope coefficient [se] = -0.07 [0.03]) on
IP
CVD mortality risk in the overweight, unfit group (model Q = 4.22, df = 1, p = 0.04).
CR
Figure 4 shows the elevated risk of CVD mortality for the obese groups. Again, there was
a higher elevation in risk for the obese unfit group (tripled) than for the obese fit group
US
(increased by 42%). However, the significance of the obese fit data was changed as multiple
AN
analyses were completed with each study removed. This one study removed sensitivity analysis
resulted in a lack of significant increase in the hazard ratio for this group, when five of the
M
articles in the total pool were individually removed, one at a time, and the analyses re-run (4, 6,
ED
13, 15, 19). The one study removed analysis did not significantly affect findings for the obese,
unfit group. The three dichotomous moderator analyses did not affect the main outcomes for the
PT
obese groups, while the age meta-regression showed an inverse relationship (slope coefficient
CE
[se] = -0.09 [0.03]) between age and risk of CVD mortality only in the unfit obese group (model
Q = 8.24, df = 1, p = 0.004).
AC
Discussion
Observational studies, in which the joint impact of CRF and BMI on CVD mortality was
determined, were analyzed using meta-analytical techniques. After assessing publication bias and
moderator analyses and jointly assessing CRF and BMI, both variables (i.e. decreases in CRF
ACCEPTED MANUSCRIPT
and increases in BMI) increased CVD mortality risk. However, the data suggest that low CRF is
a stronger predictor of CVD mortality risk (more than double), than elevated BMI. In fact, a
significant increase in CVD mortality risk occurred, for all unfit groups, in every article included
in the meta-analysis, regardless of BMI. This is compelling evidence of the significant mortality
risk associated with having low CRF. While CVD death risk significantly increased in fit
T
overweight and obese individuals, a one study removed analysis in the obese, fit group reduced
IP
the CVD mortality risk to non-significant levels when each of five articles were removed (4, 6,
CR
13, 15, 19). Therefore, CRF is a powerful predictor of CVD mortality risk.
While there are several pertinent review articles showing strong effects of CRF on
US
mortality while concurrently assessing BMI levels (26, 27, 28, 29), a numerical assessment of
AN
these effects was only available for all-cause mortality (16). These prior all-cause mortality
analyses produced similar findings in that unfit individuals had at least twice the risk of death
M
from all causes compared to fit counterparts. Interestingly, there was a greater risk of CVD
ED
mortality in the current meta-analysis for all fitness and fatness groups, suggesting that fitness
and, to an extent, fatness are greater predictors of CVD mortality than all-cause mortality.
PT
Age was the only moderator that was statistically significant. In the unfit and fat groups
CE
(those at highest risk), age was inversely related to CVD mortality risk. Younger, unfit,
overweight or obese individuals had a higher CVD mortality risk than older individuals in these
AC
groups. While an inverse relationship seems counterintuitive, specific factors may have
contributed to this outcome: 1) a survivor bias may have occurred in which unfit and overweight
or obese individuals were preselected to survive and 2) the baseline age range within the
There is controversy within the literature surrounding the use of BMI as an indicator of
risk. While other anthropometric measures provide an assessment of body composition, BMI
simply measures body mass standardized to height. With the current anthropometric paradigm, it
seems logical for body fat percentage and possibly waist circumference to be better assessments
of CVD mortality risk. While a direct comparison of these exposure variables on CVD mortality
T
was not found in the literature, a review article showed mortality risk was not elevated in fit and
IP
obese individual, regardless of anthropometric method (30). Furthermore, recent research
CR
demonstrated independent association between fat free mass and mortality in obese individuals
(31). The fact that BMI, as a measure of body mass, includes both fat and fat free mass, may
US
improve its prognostic effect on cardio metabolic risk.
AN
All studies included in this analysis were longitudinal studies with one exposure
assessment at baseline. While the current study design provided powerful information on CVD
M
mortality risk relative to fitness and fatness, changes in the exposure variables (i.e. fitness and
ED
BMI) and their effect on CVD death were not assessed. A recent study by Lee et al. (2011)
assessed the combined association of changes in fitness and BMI and the impact on CVD
PT
mortality. Lee et al. showed fitness gains stabilized CVD mortality risk while fitness loss
CE
increased CVD mortality risk, regardless of BMI changes. This suggests better CVD mortality
prognoses in individuals who improve CRF compared to those who reduce their BMI (32).
AC
It is important to note, the data in the studies in the meta-analysis were drawn from three
sources, the Aerobic Center Longitudinal Study (ACLS) and the Cooper Clinic Longitudinal
Study (CCLS) databases, an independent sample of men from the United States, and an
independent sample of men from Russia. However, the ACLS database name was changed to the
CCLS in 2006. Therefore, these databases were considered the same source for this analysis,
ACCEPTED MANUSCRIPT
resulting in three total sources. This may result in concern of population overlap, specifically
among the ACLS and CCLS studies. However, many of the study samples were from disease-
specific populations within the ACLS and CCLS databases, minimizing this concern. It is also
important to consider how BMI and CRF exposure variables are expressed. As shown in Table 2,
there was variability among studies. Due to this, not all studies were included in every analysis.
T
Also pertinent, is that only eight articles were included in the meta-analysis. However, tens of
IP
thousands of individuals were included in the analyzed samples. In future studies, the joint
CR
association between CRF and BMI on CVD mortality risk in women needs to be investigated as
US
In conclusion, CRF is a powerful predictor of CVD mortality risk. Individuals who are
AN
unfit are at twice the risk of death compared to their fit counterparts, regardless of BMI.
Individuals who are obese and unfit have the highest mortality risk (tripled) compared to normal
M
weight, fit individuals. While an elevated BMI was associated with CVD mortality risk, being fit
ED
nearly eliminated this risk in overweight and obese individuals. Therefore, CRF needs to be a
central component in interventions aimed at reducing the health and economic burdens related to
PT
CVD mortality.
CE
AC
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
References
1. Benjamin EJ, Virani SS, Callaway CW, et al: Heart disease and stroke statistics – 2018
2. National Center for Health Statistics. Health, United States, 2016: With Chartbook on
T
IP
3. Flegal KM, Kruszon-Moran D, Carroll MD, et al: Trends in obesity among adults in the
CR
4. Finkelstein EA, Trogdon JG, Cohen JW, et al: Annual medical spending attributable to
US
obesity: Payer-and service-specific estimates. Health Affairs 28:w822–w831, 2009
6. McAuley PA, Artero EG, Sui X, et al: Fitness, fatness, and survival in adults with
7. Wei M, Kampert JB, Barlow CE, et al: Relationship between low cardiorespiratory
PT
fitness and mortality in normal-weight, overweight, and obese men. JAMA 282:1547-53,
1999
CE
8. Church TS, LaMonte MJ, Barlow CE, et al: Cardiorespiratory fitness and body mass
AC
Index as predictors of cardiovascular disease mortality among men with diabetes. Arch
9. McAuley PA, Artero EG, Sui X, et al: The obesity paradox, cardiorespiratory fitness, and
10. McGee DA: Body mass index and mortality: A meta-analysis based on person-level data
11. Blair SN, Kampert JB, Kohl HW, et al: Influences of cardiorespiratory fitness and other
precursors on cardiovascular disease and all-cause mortality in men and women. JAMA
276:205-2010, 1996
12. Evenson KR, Stevens J, Thomas R, et al: Effect of cardiorespiratory fitness on mortality
among hypertensive and normotensive women and men. Epidemiol 15:565-572, 2004
T
13. Farrell SW, Finley CE, Radford NB, et al: Cardiorespiratory fitness, body mass index,
IP
and heart failure mortality in men: Cooper Center Longitudinal Study. Circ Heart Fail
CR
6:898-905, 2013
14. Lee DC, Artero EG, Sui X, et al: Mortality trends in the general population: The
US
importance of cardiorespiratory fitness. J Psychopharmacol 24:27-35, 2010
AN
15. McAuley PA, Sui X, Church TS, et al: The joint effects of cardiorespiratory fitness and
2009
ED
16. Barry VW, Baruth MB, Beets MW, et al: Fitness vs. fatness on all-cause mortality: A
17. Folgelholm M: Physical activity, fitness and fatness: relations to mortality morbidity and
CE
18. Pedersen BK: Body mass index-independent effect of fitness and physical activity for all-
AC
19. Stevens J, Evenson KR, Thomas O, et al: Associations of fitness and fatness with
mortality in Russian and American men in the lipids research clinics study. Int J Obes
20. Lee CD, Jackson AS, Blair SN: US weight guidelines: Is it also important to consider
21. Moher D, Liberati A, Tetzlaff J, et al: Preferred reporting items for systematic reviews
22. Wells GA, Shea B, O’Connell D, et al: The Newcastle-Ottawa Scale (NOS) for assessing
T
the quality of nonrandomized studies in meta-analyses. http://www.ohri.ca/programs/
IP
clinical_epidemiology/oxford.asp
CR
23. Hamling J, Lee P, Weitkunat R, et al: Facilitating meta-analyses by deriving relative
effect and precision estimates for alternative comparisons from a set of estimates
US
presented by exposure level or disease category. Stat Med 27:954-970, 2008
AN
24. Egger M, Smith GD, Schneider M, et al: Bias in meta-analysis by a simple, graphical test.
25. Riebe D, Ehrman JK, Liguori G, et al (eds): ACSM’s guidelines for exercise testing and
ED
26. Kennedy AB, Lavie CJ, Blair SN: Fitness or fatness: which is more important? JAMA
PT
319:231-232, 2018
CE
27. Lavie CJ, Sharma A, Alpert MA, et al: Update on obesity and obesity paradox in heart
28. Lavie CJ, Schutter AD, Parto P, et al: Obesity and prevalence of cardiovascular diseases
and prognosis – the obesity paradox updated. Prog Cardiovasc Dis 58:537-547, 2016
29. Oktay AA, Lavie CJ, Kokkinos PF, et al: The interaction of cardiorespiratory fitness with
obesity and the obesity paradox in cardiovascular disease. Prog Cardiovasc Dis 60:30-44,
2017
ACCEPTED MANUSCRIPT
30. Ortega FB, Lavie CJ, Blair SN: Obesity and cardiovascular disease. Circ Res 118:352-
382, 2016
31. Ortega FB, Sui X, Lavie CJ, et al: Body mass index, the most widely used but also
widely criticized index: would a gold-standard measure of total body fat be a better
T
32. Lee D, Sui X, Artero EG, et al: Long-term effects of changes in cardiorespiratory fitness
IP
and body mass index on all-cause and cardiovascular disease mortality in men: the
CR
aerobics center longitudinal study. Circulation 124:2483-2490, 2011
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
Figure 1. Selection of Articles for Meta-analysis; CVD, Cardiovascular disease; BMI, body
mass index; CRF, Cardiorespiratory fitness; HR, Hazard ratio; CI, Confidence interval.
Figure 2. Meta-analysis of CVD mortality for normal weight unfit individuals compared to
normal weight fit individuals; CVD, Cardiovascular disease; r, Russian males; u, United States
males.
T
Figure 3. Meta-analysis of CVD mortality for unfit and fit overweight individuals compared to
IP
normal weight fit individuals; CVD, Cardiovascular disease.
CR
Figure 4. Meta-analysis of CVD mortality for unfit and fit obese individuals compared to normal
weight fit individuals; CVD, Cardiovascular Disease; r, Russian males; u, United States males.
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
T
IP
CR
US
AN
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
Follow-
Exposures
up Outcome
Age Baseline Study
Men (mean health data- (mean (No. of
2
Article N (%) years) status base years) deaths) CRF BMI (kg/m )
Church et al. 2,316 100 49.3 Diabetes ACLS 15.9 179 F: upper 80% NW: 18.5-24.9
2005
T
UF: lower 20% OW: 25-29.9
IP
OB: 30-34.9
CR
a
Farrell et al. 44,674 100 45 - CCLS 19.8 153 F: upper 80% NW: 18.5-24.9
2013
US
UF: lower 20% OW: 25-29.9
AN
OB: ≥30
a
Lee et al. 1998 21,865 100 43.9 - ACLS 8.1 144 F: upper 80% NW: 19-24.9
M
OB: ≥27.8
PT
McAuley et al. 13,155 100 62.4 HTN ACLS 12 355 F: upper 80% NW: 18.5-24.9
2009
CE
OB: ≥30
AC
McAuley et al. 9,563 100 43.9 CHD ACLS 13.4 348 F: upper 66.6% NW: 18.5-24.9
2012
UF: lower 33.3% OW: 25-29.9
OB: 30-34.9
ACCEPTED MANUSCRIPT
Follow-
Exposures
up Outcome
Age Baseline Study
Men (mean health data- (mean (No. of
2
Article N (%) years) status base years) deaths) CRF BMI (kg/m )
McAuley et al. 17,044 89 47.4 Pre- ACLS 13.9 246 F: upper 66.6% NW: 18.5-24.9
2014
T
diabetes
UF: lower 33.3% OW: 25-29.9
IP
OB: ≥30
CR
a b
Stevens et al. 1,359 100 47.7 - LRCS 17.6 - F: upper 80% Non-OB: lower 80%
2004r
US
UF: lower 20% OB: upper 20%
AN
a b
Stevens et al. 1,716 100 57.3 - LRCS 17.6 - F: upper 80% Non-OB: lower 80%
2004u
UF: lower 20% OB: upper 20%
M
a
Wei et al. 1999 25,714 100 64.4 - ACLS 10.1 439 F: upper 80% NW: 18.5-24.9
ED
OB: ≥30
CE
a b
: no specific baseline disease reported; :outcome not provided; r, Russian sample; u, United States sample; ACLS, Aerobic Center
Longitudinal Study; CCLS, Cooper Clinic Longitudinal Study; LRCS, Lipids Research Clinics Study; CRF, Cardiorespiratory Fitness;
BMI, Body Mass Index; F, Fit; UF, Unfit; NW, Normal Weight; OW, Overweight; OB, Obese. Note: The ACLS database is a subset of
AC