articles nature publishing group

inteRVENTION AND pREVENTION

Exercise Intensity Prescription in

Obese Individuals
Bernard M. Pinet1, Denis Prud’homme1,2, Chantal A. Gallant3 and Pierre Boulay4

The main purpose of this study was to evaluate the relationship between different methods proposed by the
American College of Sports Medicine (ACSM) to prescribe exercise intensity using heart rate (HR) and oxygen
uptake (VO2) in obese individuals. Sixty-eight overweight to severely obese adults were divided into three groups
(tertile) based on their BMI. The groups were T1 group (BMI = 30.5 ± 1.5, n = 23), T2 group (BMI = 34.3 ± 1.0, n = 23),
and T3 group (BMI = 40.2 ± 3.7, n = 22). All subjects performed a graded exercise test using a ramp protocol on a
treadmill. Individual linear regressions between %HR reserve (%HRR) and %VO2 reserve (%VO2R), %HRR and
%VO2 peak (%VO2peak), %maximal HR (%HRmax) and %VO2R, and %HRmax and %VO2peak were calculated. When all the
subjects were grouped together, the %HRR–%VO2R mean regression was partially related to the line of identity, while
the %HRR–%VO2peak, %HRmax–%VO2R, and %HRmax–%VO2peak mean regressions were all significantly different than
the line of identity (P < 0.001). The degree of obesity accounted for ~15% of the variation for both %HRR–%VO2R and
%HRR–%VO2peak mean regressions. The %HRmax–%VO2R and %HRmax–%VO2peak mean regressions were not affected
by the degree of obesity but resting HR accounted for 28–37% of the variation. The relationship between the exercise
intensity determined by the %HRR–%VO2R and the %HRR–%VO2peak mean regression seems to be influenced by
the degree of obesity. The degree of obesity does not affect the relationship between exercise intensity generated by
the %HRmax–%VO2R or %HRmax–%VO2peak equations but the resting HR does.

Obesity (2008) 16, 2088–2095. doi:10.1038/oby.2008.272

Introduction
The role of regular physical activity to achieve health benefits
and to increase energy expenditure is well known in the treat-
ment of obesity (1). Therefore, health-care professionals have
to take into account not only the type of exercise but also the
interaction between exercise intensity, duration, and frequency
to determine an efficient exercise prescription according to the
desired objective with obese individuals (1,2).
In accordance with exercise duration and frequency, several
methods have been used in the past to prescribe the inten-
sity at which an individual should exercise for health benefits
(3–6). Oxygen uptake (VO2) is known as the best physiological
measuring
metabolic
parameter to reflect the metabolic stress associated with exer-
stress, cise intensity; however, monitoring exercise VO2 day to day
remains unrealistic (7). Instead, exercise specialists have used
the linear relationship observed between heart rate (HR) and
VO2 to determine and control exercise intensity using a pre-
scribed target HR (2,7,8).
The American College of Sports Medicine (ACSM) recom-
mends three different methods to determine exercise target
HR that best represent the target VO2 associated with any
exercise intensity (2). When VO2 values are available, the
first recommended method to use is to identify the HR asso-
ciated with a defined exercise target VO2 measured during
a graded exercise test. The following two other methods are
used when exercise VO2 values are not available: (i) percent-
age of maximal HR (%HRmax) equation in which target HR =
relative intensity (%) · HRmax and (ii) %HR reserve (%HRR)
equation (9), which takes the resting HR into consideration
as described in the following equation: target HR = rela-
tive intensity (%) · (HRmax – resting HR) + resting HR. To
improve relative exercise intensity equivalency between tar-
get HR and VO2, several studies have demonstrated the supe-
riority of the percentage of VO2 reserve (%VO2R) method
for exercise prescription in many populations (i.e., healthy
adults, cardiac patients, obese and diabetics individuals, and
elite cyclist) rather than simply calculating a certain per-
centage of the peak VO2 (%VO2peak) as it was done for years
(10–15). The %VO2R method takes into account the resting
VO2 to determine the target exercise VO2 as described in
the following equation: target VO2 = relative intensity (%) ·
(VO2peak – r­ esting VO2) + resting VO2, where resting VO2 can

1
School of Human Kinetics, Faculty of Health Sciences, University of Ottawa, Ottawa, Ontario, Canada; 2Behavioural and Metabolic Research Unit, Montfort Hospital,
Ottawa, Ontario, Canada; 3Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada; 4Cardiovascular Health Service, Montfort Hospital, Ottawa, Ontario,
Canada. Correspondence: Pierre Boulay (boulayp@mac.com)
Received 29 September 2007; accepted 6 March 2008; published online 15 May 2008. doi:10.1038/oby.2008.272

2088 VOLUME 16 NUMBER 9 | SEPTEMBER 2008 | www.obesityjournal.org


be measured or estimated at 1 metabolic equivalent (MET)
(3.5 ml/kg/min) (2,12,13).
Several studies on different populations have shown that the
relationship between the exercise intensities determined by
the %HRR and %VO2R is closely related to the line of identity
using either the measured or estimated resting VO2, whereas
the former %HRR and %VO2peak relationship is not (10–15).
The line of identity is defined as a slope of 1.0 and a Y-intercept
of 0.0 (2). Furthermore, Howley (16) modified the equation
proposed by Londeree and Ames (17) with the purpose of
determining a %HRmax that represents the oxygen cost at spe-
cific exercise intensity. The author used the estimated resting
VO2 (3.5 ml/kg/min) to determine the %VO2R in order to
develop a new equation to determine %HRmax, where %HRmax =
0.731 · %VO2R + 29.95.
The ACSM (2005) has updated its exercise guidelines for
healthy, obese, cardiac, and diabetic individuals accordingly to
the recent equations proposed in the literature (2). It is impor-
tant to note that cardiovascular responses to exercise are dif-
ferent between gender (8,10,11,18,19) and are influenced by
the degree of obesity (20–27). Thus, the purpose of this study
was to determine whether the %HRR − %VO2R, the %HRR −
%VO2peak, the %HRmax − %VO2R, and the %HRmax − %VO2peak
relationships, according to current ACSM’s guidelines, are
affected by gender and/or the degree of obesity.
Methods And Procedures
Subjects
A total of 71 subjects (29 men and 42 women) volunteered to participate
in this study. According to the normal distribution, two men and one
woman had a Z-score value >3.19 for BMI and resting HR, respectively.
Thus, those three subjects were removed from the analyses for a total of
68 subjects (27 men and 41 women). Inclusion criteria to participate in
this study were: (i) overweight or obese (BMI > 27 kg/m2) and weight
stable (±3 kg) for the past 6 months; (ii) sedentary lifestyle according to
the ACSM definition; and (ii) nonsmoker and having an overall healthy
state. Subjects who were using any known medications with an effect on
the metabolism or on the cardiovascular response were automatically
excluded from the study. A complete description of the study was given
to each subject. Thereafter, all participants had to sign a consent form
approved by the Research Ethic Board of the Université de Moncton
before enrolling in the current study.
Anthropometric measures
Body weight and height were measured with a standard physician’s
scale and a stadiometer, respectively when subjects were in a fasting
state before the resting metabolism session. Waist circumference was
measured after a normal expiration under the midline of the subject’s
armpit, at the midpoint between the lower part of the last rib and the
top of the hip. Fat mass, fat-free mass, and percentage of fat (%fat) were
assessed with a bioelectrical impedance apparatus (Bodystat 1500;
Bodystat, Douglas, Isle of Man). This apparatus has been validated in
obese individuals (28). Subjects were advised to be well hydrated and to
limit their physical activity the day before the evaluation.
Aerobic capacity (VO2peak)
Oxygen uptake was continuously measured during an incremental
treadmill exercise test using the BSU/Bruce Ramp protocol (29). The
protocol begins with an intensity of approximately 2 METs and consists
of increments of 0.3 METs every 20 s by increasing the speed and/or the
grade. The speed and the grade of the last 20 s of every 3-min segment of
obesity | VOLUME 16 NUMBER 9 | SEPTEMBER 2008
articles
intervention and Prevention
the protocol are identical to those of the standard Bruce protocol (29).
Ramp protocols are known to be more relevant than step protocols
for determining exercise intensity prescription (30). Respiratory gas
exchange and ventilation were measured continuously using a meta-
bolic cart with a breath-by-breath method (Ergocard; Medi-Soft,
Dinant, Belgium). Both gas analyzer and pneumotach were calibrated
before each test. Subjects were asked to refrain from alcohol and caf-
feine consumption for 24 h before exercise testing. HR was constantly
monitored with a 12-lead electrocardiogram monitor (Marquette Case
16; Marquette Electronics, Milwaukee, WI) and was synchronized
timewise with the metabolic cart. Criteria for achieving VO2peak were:
(i) no further increase in oxygen uptake or an increase of <150 ml O2/
min with an increase in exercise workload (three consecutive 20-s
increment); (ii) a HR at or above the age-predicted maximum (using
the equation 220-age) or no further increase in HR with an increase in
exercise workload; and (iii) a respiratory exchange ratio >1.1. All sub-
jects met at least two of these criteria (31).
Resting measurements
Resting HR was measured after a 15-min rest in a sitting position and in
a quiet environment before the graded exercise test with an electrocar-
diogram (Marquette Case 16; Marquette Electronics). Resting oxygen
uptake assessment was done early in the morning, on a separate occa-
sion, following a 12-h fast and 24-h without consumption of alcohol or
caffeine. Subjects were asked to sleep in their usual environment the
night before the test and to remain physically inactive the morning of
the test. Subjects were positioned in a recumbent position for a period
of 45 min. Resting VO2 was measured using a metabolic cart (Ergocard;
Medi-Soft) using a face mask (Hans Rudolph, Shawnee, KS) with a
breath-by-breath method. Only the last 20 min of the VO2 measure-
ments were included in the analysis.
Statistical methods
The percentages of VO2R, VO2peak, HRR, and HRmax were calculated
at every minute during the graded exercise test using the following
equations:
%VO2R = ((Current VO2 – resting VO2)
× (VO2peak – resting VO2)−1) × 100%.
%VO2peak = ((Current VO2) × (VO2peak)−1) × 100%.
%HRR = ((Current HR – resting HR)
× (HRmax – resting HR)−1) × 100%.
%HRmax = ((Current HR) × (HRmax)−1) × 100%.
Four individuals’ regressions have been calculated for every subject:
method A, %VO2R and %HRR as independent and dependent vari-
ables, respectively; method B, %VO2peak and %HRR as independent
and dependent variables, respectively; method C, %VO2R and %HRmax
as independent and dependent variables, respectively; method D,
%VO2peak and %HRmax as independent and dependent variables, respec-
tively. The mean slopes and mean Y-intercepts of methods A and B
were compared to 1.0 and 0.0, respectively, whereas the mean slope
and mean Y-intercept of methods C and D were compared to 0.731
and 29.95, respectively (2,16) using a one-sample t-test. A slope of 1.0
and a Y-intercept of 0.0 have been used previously by many authors to
assess the equivalency between %HRR, %VO2R, and %VO2peak in adults
with different conditions (10–15). A slope of 0.731 and a Y-intercept
of 29.95 have been suggested by Howley (16) to represent the relation-
ship between %HRmax and %VO2R or %VO2peak. The ACSM used these
slopes and Y-intercepts to establish its latest recommendations for
2089
articles
intervention and Prevention
exercise intensity prescription with healthy adults and patients with an
impaired metabolic and/or cardiac condition (2).
The comparison between the methods A to D with the corresponding
line of identity has been done with all subjects grouped together. Further-
more, BMI was significantly correlated with the slope and Y-­intercept
of methods A and B (r = 0.313, P < 0.01; r = −0.286, P < 0.05; r = 0.314,
P < 0.01; r = −0.294, P < 0.05, respectively). To define a degree of obes-
ity indicator, we categorized our subjects into three groups ­(tertile)
according to their BMI values (T1 group, BMI = 27.1–32.1 kg/m2,
n = 23; T2 group, BMI = 32.9–35.8 kg/m2, n = 23; T3 group, BMI =
36.0–46.6 kg/m2, n = 22) to investigate the effect of the degree of obesity
on exercise intensity determination using different methods.
Two-way ANOVA was performed to analyze the effect of gender
and the degree of obesity on the descriptive variables with a Bonferroni
post hoc test. Levene’s test was performed to evaluate the equivalency
of ­variance between the groups. Two-way ANOVA (gender and degree
of obesity) was also performed originally on the slope and Y-intercept of
each method. However, as no effect of gender was observed for any of the
four methods, we limited our analysis with a one-way ANOVA (degree
of obesity) using a Bonferroni post hoc test to simplify the presentation
of the results. A preliminary analysis has demonstrated that the degree of
obesity defined by BMI, fat mass, %fat, resting HR, and HRR was all indi-
vidually correlated to one of the four methods (P < 0.05). To investigate
any possible confounding variables affecting the individuals’ regressions,
stepwise multiple regressions were performed for all the four methods
using individuals’ slope and Y-intercept as the dependent variables and
the degree of obesity, fat mass, %fat, resting HR, and HRR as independent
variables. Predicted %HRR and %HRmax have been calculated at 40, 50,
60, 70, 80, and 85% of VO2R and VO2peak according to individual simple
regression equation. One-sample t-test was performed to compare the
predicted %HRR and %HRmax with ACSM’s guidelines. One-way ANOVA
with Bonferroni post hoc test was performed to determine the effect of
the degree of obesity. Significance level was set at P < 0.05. All statisti-
cal analyses were performed using SPSS software, version 12.0 (SPSS,
­Chicago, IL), and data manipulation were performed using Microsoft
Office Excel, version 2003 (Microsoft, Redmond, WA).
Results
Subjects’ characteristics
The subjects’ physical and physiological characteristics are pre-
sented in Tables 1 and 2. By study design, the T3 group had a
significantly higher weight, fat mass, and waist circumference
than the T1 and T2 groups. They also presented a lower VO2peak
(ml/kg/min) and VO2R than the two other groups whereas
the HRR was lower only when compared to the T1 group.
Body weight and %fat did not attain the level of significance
Table 1 Physical characteristics of the subjects
Variables T1 group
Gender (male/female) 8/15
Age (years) 45 ± 7 (24–54)
Weight (kg) 85.7 ± 14.0 (60.9–116.2)
Height (m) 1.67 ± 0.11 (1.50–1.91)
BMI (kg/m2) 30.5 ± 1.5 (27.1–32.1)
Waist circumference (cm) 99.6 ± 9.5 (76.0–118.0)
Fat mass (kg) 31.4 ± 4.6 (23.5–40.0)
Fat-free mass (kg) 54.2 ± 13.6 (35.4–80.7)
% of Fat 37.3 ± 6.6 (25.6–45.6)
Data are mean ± s.d. (range).
Significantly different than T1 group: *P < 0.01, **P < 0.001; significantly different than T2 group: †P < 0.05, ††P < 0.01, †††P < 0.001.
2090
between groups T2 and T1, where P = 0.124 and P = 0.240
according to the Bonferroni post hoc test. Men are older and
their weight, waist circumference, fat-free mass, resting HR,
resting VO2, VO2peak, and VO2R are significantly higher than
women. Women have significantly higher fat mass, %fat, and
HRmax (data not shown). BMI and HRR values were statisti-
cally equivalent between men and women. No gender effect
was found for BMI and HRR.
Method A: relationships between %HRR and %VO2R
The mean slope for all subjects was not significantly differ-
ent than 1.0 (1.007 ± 0.139) t(67) = 0.400, P = 0.690, whereas
the mean Y-intercept was significantly lower than 0.0 (−6.30 ±
13.38) t(67) = −3.884, P < 0.001. The mean slope for T1 group
was significantly lower than 1.0 (0.941 ± 0.128) t(22) = −2.209,
P < 0.05, whereas the mean Y-intercept was not significantly
different than 0.0 (−0.46 ± 12.81) t(22) = −0.171, P = 0.865.
The mean slope for T2 group was not significantly different
than 1.0 (1.003 ± 0.126) t(22) = 0.116, P = 0.909, whereas the
mean Y-intercept was significantly lower than 0.0 (−5.83 ±
12.36) t(22) = −2.260, P < 0.05. The mean slope for T3 group
was significantly higher than 1.0 (1.079 ± 0.133) t(21) = 2.785,
P < 0.05, whereas the mean Y-intercept was significantly lower
than 0.0 (−12.91 ± 12.48) t(21) = −4.853, P < 0.001. One-way
ANOVA showed a significant effect of the degree of obesity
on both the method A slope F(2,65) = 6.443, P < 0.01, and
Y-intercept F(2,65) = 5.559, P < 0.01. Bonferroni post hoc
analyses for the degree of obesity showed that T3 group had a
significantly higher slope and a significantly lower Y-intercept
than T1 group (Figure 1a). Stepwise multiple regressions
showed that the degree of obesity accounted for 16.5 and
14.5% of the variance of the slope (β = 0.406, P < 0.01) and
the Y-intercept (β = −0.381, P < 0.01), respectively (Table 3).
The %HRR predicted from the %VO2R for all subjects was
significantly lower than ACSM’s guideline from 40 to 85% of
VO2R (−8.614 ≤ t(67) ≤ −5.777, P < 0.001). When separated
according to the degree of obesity, the predicted %HRR was
similar to ACSM’s guideline only for the T1 group at 40% of
VO2R (t(22) = −1.601, P = 0.124). Otherwise, every other pre-
dicted %HRR was significantly lower than ACSM’s guideline
T2 group T3 group
9/14 10/12
43 ± 7 (32–56) 48 ± 4 (41–54)
†
94.0 ± 9.2 (78.5–111.6) 113.3 ± 16.8**,††† (88.7–146.1)
1.66 ± 0.08 (1.50–1.79) 1.68 ± 0.10 (1.53–1.89)
34.3 ± 1.0** (32.9–35.8) 40.2 ± 3.7**,††† (36.0–46.6)
107.0 ± 10.2 (89.4–131.9) 118.0 ± 12.9**,†† (97.0–145.5)
38.6 ± 7.5* (28.6–61.7) 51.8 ± 9.6**,††† (38.5–71.4)
57.3 ± 12.0 (41.5–81.6) 61.4 ± 14.4 (41.8–85.8)
41.1 ± 7.5 (27.8–56.9) 46.0 ± 7.7** (34.0–58.2)
VOLUME 16 NUMBER 9 | SEPTEMBER 2008 | www.obesityjournal.org
 articles
intervention and Prevention

Table 2 Physiological characteristics of the subjects

Variables T1 group T2 group T3 group
Resting heart rate (bpm) 83 ± 11 (66–110) 80 ± 8 (68–102) 89 ± 11*** (70–110)
Maximal heart rate (bpm) 177 ± 13 (153–198) 172 ± 12 (139–187) 171 ± 8 (156–192)
Heart rate reserve (bpm) 93 ± 14 (63–120) 89 ± 11 (68–108) 81 ± 10** (59–103)
Resting VO2 (ml/min and ml/kg/min) 229 ± 55 2.7 ± 0.5 221 ± 52 2.4 ± 0.4* 255 ± 51 2.3 ± 0.4**
(140–350) (1.7–3.8) (140–330) (1.6–3.1) (160–330) (1.4–3.1)
VO2 peak (ml/kg/min) 27.0 ± 3.4 (22.0–34.2) 25.9 ± 4.2 (19.9–35.2) 22.4 ± 4.1**,*** (14.5–32.8)
VO2 reserve (ml/kg/min) 24.3 ± 3.6 (19.0–32.5) 23.4 ± 4.1 (17.3–32.1) 20.1 ± 4.0**,*** (12.4–30.1)
Data are mean ± s.d. (range).
Significantly different than T1 group: *P < 0.05, **P < 0.01; significantly different than T2 group: ***P < 0.05.

Method B: relationships between %HRR and %VO2peak

a 100
The mean slope for all subjects was significantly higher than
90
1.0 (1.122 ± 0.166) t(67) = 6.043, P < 0.001, whereas the mean
80
Y-intercept was significantly lower than 0.0 (−17.78 ± 15.88)
70 t(67) = −9.235, P < 0.001. The mean slope for T1 group was
% Heart rate reserve

60 not significantly different than 1.0 (1.050 ± 0.150) t(22) =

50 1.613, P = 0.121, whereas the mean Y-intercept was signifi-
40
cantly lower than 0.0 (−11.42 ± 15.14) t(22) = −3.615, P <
0.01. The mean slope for T2 group was significantly higher
30
T1 group Y = 0.941 X − 0.46*
than 1.0 (1.107 ± 0.142) t(22) = 3.612, P < 0.01, whereas the
20 T2 group
T3 group
Y = 1.003 X − 5.83†
Y = 1.079 X − 12.91*†‡
mean Y-intercept was significantly lower than 0.0 (−16.22 ±
10 All subjects
Line of identity
Y = 1.007 X − 6.30†
Y = 1.000 X + 0.00
13.61) t(22) = −5.615, P < 0.001. The mean slope for T3
0 group was significantly higher than 1.0 (1.211 ± 0.170)
0 10 20 30 40 50 60 70 80 90 100
% VO2 reserve
t(21) = 5.819, P < 0.001, whereas the mean Y-intercept was
significantly lower than 0.0 (−26.06 ± 15.85) t(21) = −7.712,
b 100 P < 0.001. One-way ANOVA showed a significant effect of
90
the degree of obesity on both the %HRR and %VO2peak slope
80 F(2,65) = 6.238, P < 0.01, and Y-intercept F(2,65) = 5.636, P <
70 0.01. Bonferroni post hoc analyses for the degree of obesity
% Heart rate reserve

60
showed that T3 group had a significantly higher slope and a
50
significantly lower Y-intercept than T1 group (Figure 1b).
Stepwise multiple regressions showed that the degree of
40
obesity accounted for 15.7 and 14.2% of the variance of the
30
slope (β = 0.396, P < 0.01) and the Y-intercept (β = −0.377,
20 T1 group
T2 group
Y = 1.050 X − 11.42†
Y = 1.107 X − 16.22*† P < 0.01), respectively (Table 3). The %HRR predicted from
10
T3 group
All subjects
Y = 1.211 X − 26.06*†‡
Y = 1.122 X − 17.78*†
the %VO2peak for all subjects were significantly lower than
0
Line of identity Y = 1.000 X + 0.00
ACSM’s guideline from 40 to 85% of VO2peak (−11.967 ≤ t(67)
0 10 20 30 40 50 60 70 80 90 100
≤ −10.740, P < 0.001). Furthermore, one-way ANOVAs have
% VO2 peak
shown a significant degree of obesity effect for the predicted
Figure 1 Mean individual’s regression of (a) %HRR predicted from %HRR at 40 and 50% of VO2peak F(2,65) = 4.465, P < 0.05 and
%VO2R (method A) and (b) %HRR predicted from %VO2peak (method F(2,65) = 3.824, P < 0.05, respectively. No variation between
B) for all subjects and according to the degree of obesity. Significantly groups was observed from 60 to 85% of VO2peak (P > 0.062)
different than the slope of the line of identity: *P < 0.05; significantly (Table 4).
different than the Y-intercept of the line of identity: †P < 0.05; significantly
different than T1 group (both slope and Y-intercept): ‡P < 0.05. %HRR,
% of heart rate reserve; %VO2R, % of VO2 reserve with the measured Method C: relationships between %HRmax and %VO2R
resting VO2; %VO2peak, % of peak VO2. The mean slope for all subjects was significantly lower than
0.731 (0.517 ± 0.082) t(67) = −21.546, P < 0.001, whereas
(−6.249 ≤ t(21 or 22) ≤ −2.182, P < 0.05). One-way ANOVAs the mean Y-intercept was significantly higher than 29.95
have shown a significant degree of obesity effect for the pre- (45.25 ± 8.22) t(67) = 15.354, P < 0.001. The mean slope for
dicted %HRR at 40 and 50% of VO2R only F(2,65) = 4.008, P < T1 group was significantly lower than 0.731 (0.503 ± 0.101)
0.05 and F(2,65) = 3.244, P < 0.05, respectively. No variation t(22) = −10.806, P < 0.001, whereas the mean Y-intercept was
between groups was observed from 60 to 85% of VO2R (P > ­significantly higher than 29.95 (46.42 ± 10.02) t(22) = 7.880,
0.108) (Table 4). P < 0.001. The mean slope for T2 group was significantly

obesity | VOLUME 16 NUMBER 9 | SEPTEMBER 2008 2091

articles
intervention and Prevention

Table 3 Effects of degree of obesity and resting heart rate on four different HR–VO2 relationships during treadmill-graded
exercise testing
Method A Method B Method C Method D
(%HRR vs. %VO2R) (%HRR vs. %VO2peak) (%HRmax vs. %VO2R) (%HRmax vs. %VO2peak)
Independent variable Degree of obesity Degree of obesity Resting heart rate Resting heart rate
Dependent variable β R2 P β R2 P β R2 P β R2 P
Slope 0.406 0.165 <0.01 0.396 0.157 <0.01 −0.570 0.325 <0.001 −0.530 0.281 <0.001
Y-intercept −0.381 0.145 <0.01 −0.377 0.142 <0.01 0.605 0.366 <0.001 0.567 0.321 <0.001
%HRmax, percentage of maximal heart rate; %HRR, percentage of heart rate reserve; %VO2R, percentage of VO2 reserve; %VO2peak, percentage of peak VO2.
Results are from a stepwise multiple regression analyses. Independent variables included in the stepwise multiple regression were: degree of obesity, fat mass, %fat,
BMI, VO2peak (ml/kg/min), VO2R, HRR, and resting HR. Only the degree of obesity was a significant predictor for methods A and B, whereas resting heart rate was a
major and significant predictor of methods C and D.

Table 4 Predicted values for %HRR and %HRmax from the %VO2R and %VO2peak compared with the ACSM’s guideline and between
degrees of obesity
%HRR %HRmax
Predicted Predicted
Relative from from
intensity ACSM guideline Predicted from %VO2R Predicted from %VO2peak %VO2R %VO2peak
% %HRR %HRmax All T1 group T2 group T3 group All T1 group T2 group T3 group All All
40 40 64 34.0 ± 37.2 ± 34.3 ± 30.3 ± 27.1 30.6 ± 28.1 ± 22.4 ± 65.9 ± 5.3* 62.4 ± 5.9*
8.6* 8.4 8.4* 7.9*,** ± 9.9* 9.8* 8.9* 9.7*,**
50 50 71 44.0 ± 46.6 ± 44.3 ± 41.1 ± 38.3 41.1 ± 39.2 ± 34.5 ± 71.1 ± 4.6 68.2 ± 5.1*
7.6* 7.5* 7.6* 7.0*,** ± 8.6* 8.5* 7.9* 8.3*,**
60 60 77 54.1 ± 56.0 ± 54.4 ± 51.9 ± 49.5 51.6 ± 50.2 ± 46.6 ± 76.3 ± 4.0 73.9 ± 4.3*
6.7* 6.6* 6.9* 6.1* ± 7.4* 7.4* 7.1* 7.0*
70 70 84 64.2 ± 65.4 ± 64.4 ± 62.6 ± 60.7 62.1 ± 61.3 ± 58.7 ± 81.5 ± 3.4* 79.7 ± 3.7*
6.0* 5.9* 6.4* 5.5* ± 6.4* 6.5* 6.4* 6.0*
80 80 91 74.2 ± 74.9 ± 74.4 ± 73.4 ± 71.9 72.6 ± 72.4 ± 70.8 ± 86.6 ± 3.0* 85.5 ± 3.1*
5.6* 5.5* 6.2* 5.2* ± 5.7* 5.7* 6.1* 5.3*
85 85 94 79.3 ± 79.5 ± 79.4 ± 78.8 ± 77.6 77.9 ± 77.9 ± 76.9 ± 89.2 ± 2.9* 88.3 ± 3.0*
5.5* 5.3* 6.1* 5.1* ± 5.5* 5.5* 6.0* 5.2*
ACSM, American College of Sports Medicine; %HRmax, percentage of maximal heart rate; %HRR, percentage of heart rate reserve; %VO2R, percentage
of VO2 reserve; %VO2peak, percentage of peak VO2.
*Significantly different than ACSM’s guidelines P <0.05; **significantly different than T1 group P <0.05.

lower than 0.731 (0.533 ± 0.075) t(22) = −12.717, P < 0.001,
whereas the mean Y-intercept was significantly higher
than 29.95 (43.47 ± 7.41) t(22) = 8.756, P < 0.001. The
mean slope for T3 group was significantly lower than 0.731
(0.516 ± 0.065) t(22) = −15.474, P < 0.001, whereas the mean
Y-intercept was significantly higher than 29.95 (45.90 ± 6.89)
t(21) = 10.860, P < 0.001. One-way ANOVA showed no sig-
nificant effect of the degree of obesity on the slope F(2,65) =
0.794, P = 0.456, and Y-intercept F(2,65) = 0.835, P = 0.438
(Figure 2a). Stepwise multiple regression showed that resting
HR accounted for 32.5 and 36.6% of the variance of the slope
(β = −0.570, P < 0.001) and the Y-intercept (β = 0.605, P <
0.001), respectively (Table 3). The %HRmax predicted from
the %VO2R for all subjects was significantly higher at 40%
of VO2R t(67) = 3.040, P < 0.01, identical at 50 and 60% of
VO2R (P > 0.142), and significantly lower from 70 to 85% of
VO2R t(67) < −6.077, P < 0.001 when compared to ACSM’s
guideline. One-way ANOVAs have shown no degree of obe-
sity effect on the predicted HRR from 40 to 85% of VO2R
F(2,65) < 0.830, P > 0.441 (Table 4).
Method D: relationships between %HRmax and %VO2peak
The mean slope for all subjects was significantly lower than
0.731 (0.576 ± 0.094) t(67) = −13.672, P < 0.001, whereas the
mean Y-intercept was significantly higher than 29.95 (39.38 ±
9.35) t(67) = 8.310, P < 0.001. The mean slope for T1 group was
significantly lower than 0.731 (0.560 ± 0.112) t(22) = −7.326,
P < 0.001, whereas the mean Y-intercept was significantly
higher than 29.95 (40.65 ± 11.10) t(22) = 4.625, P < 0.001. The
mean slope for T2 group was significantly lower than 0.731
(0.589 ± 0.085) t(22) = −8.010, P < 0.001, whereas the mean
Y-intercept was significantly higher than 29.95 (37.89 ± 8.27)
t(22) = 4.608, P < 0.001. The mean slope for T3 group was sig-
nificantly lower than 0.731 (0.579 ± 0.084) t(22) = −8.446, P <
0.001, whereas the mean Y-intercept was significantly higher
than 29.95 (39.59 ± 8.62) t(21) = 5.247, P < 0.001. One-way
ANOVA showed no significant effect of the degree of obe-
sity on the slope F(2,65) = 0.560, P = 0.574, and Y-intercept
F(2,65) = 0.502, P = 0.608 (Figure 2b). Stepwise multiple
regression showed that resting HR accounted for 28.1 and
32.1% of the variance of the slope (β = −0.530, P < 0.001) and

2092 VOLUME 16 NUMBER 9 | SEPTEMBER 2008 | www.obesityjournal.org

 articles
intervention and Prevention

a 100 first one that demonstrates an effect of the degree of obesity

90 on the %HRR and %VO2R relationship. A stepwise multiple
80
regression has shown that the degree of obesity was the only
variable associated to method A and accounted for ~15% of
70
the variation. Furthermore, the most significant effect of the
60
degree of obesity was observed with the T3 group at rela-
% HRmax

50 tive exercise intensities of 40 and 50% of VO2R, where the

40 %HRR equaled 30.3 and 41.1%, respectively. This finding has
30 its importance because relatively low-to-moderate exercise
20
T1 group Y = 0.503 X + 46.42*† intensities are the ones primarily selected in obese individuals,
T2 group Y = 0.533 X + 43.47*†
T3 group Y = 0.516 X + 45.90*† especially for those who present several risk factors of coro-
10 All subjects Y = 0.517 X + 45.25*†
Line of identity Y = 0.731 X + 29.95 nary artery disease and/or other chronic diseases (1,2,32–34).
0
0 10 20 30 40 50 60 70 80 90 100
Results from this study suggest that using the HRR method
% VO2 reserve at exercise intensities of 40 and 50% could overestimate the
exercise metabolic rate normally observed at 40 and 50% of
b 100 VO2R in severely obese individuals. An overestimation of the
90
exercise metabolic rate in this population could affect the pre-
80
dicted energy expenditure associated to exercise when those
70 intensities are performed for longer period (i.e., ≥1 h) which
60 could have a clinical effect when a certain volume of exercise is
% HRmax

50 being prescribed with the purpose to achieve a specific energy

40
30
T1 group
20 T2 group
T3 group
10 All subjects
Line of identity
0 significantly different than the line of identity (slope = 1 and
0 10 20 30 40 50 60 70
% VO2 peak
Figure 2 Mean individual’s regression of (a) %HRmax predicted from
%VO2R (method C) and (b) %HRmax predicted from %VO2peak (method D)
for all the subjects and according to the degree of obesity. Significantly
different than the slope of the line of identity: *P < 0.05; significantly
different than the Y-intercept of the line of identity: †P < 0.05. %HRmax =
% of maximal heart rate; %VO2R, % of VO2 reserve with the measured
resting VO2; %VO2peak, % of peak VO2.
the Y-intercept (β = 0.567, P < 0.001), respectively (Table 3).
The %HRmax predicted from the %VO2peak for all subjects was
significantly lower than ACSM’s guideline from 40 to 85% of
VO2peak −2.216 ≥ t(67) ≥ −15.769, P < 0.05. One-way ANOVAs
have shown no degree of obesity effect on the predicted HRR
from 40 to 85% of VO2peak F(2,65) < 0.468, P > 0.628 (Table 4).
Discussion
Method A: relationships between %HRR and %VO2R
Results from this study showed that method A had a slope sta-
tistically undistinguishable from the line of identity defined
by the ACSM in a group of overweight to severely obese indi-
viduals (2). However, the %HRR was significantly lower (6%)
than the %VO2R at any given exercise intensity. Therefore,
our results are partially in agreement with previous work per-
formed on obese individuals who presented a wide BMI range
(13), on athletes (10,15), on cardiac patients (12), and on dia-
betic patients (13), where the %HRR − %VO2R mean regres-
sion was entirely statistically undistinguishable from the line
of identity. Interestingly, this study is, to our knowledge, the
Y = 0.560 X + 40.65*†
Y = 0.589 X + 37.89*†
Y = 0.579 X + 39.59*†
Y = 0.576 X + 39.38*†
Y = 0.731 X + 29.95
80 90 100
expenditure goal (i.e., weight loss).
Method B: relationships between %HRR and %VO2peak
Similar to several other studies performed in different popula-
tions (10–13,15), our results showed that the method B was
Y-intercept = 0) in our group of overweight to severely obese
individuals. Thus, these results suggest that assuming equiva-
lencies in relative exercise intensities between HRR and VO2peak
methods would result in an underestimation of the O2 cost
associated to exercise when using the HRR methods, especially
at lower relative exercise intensities. This finding is supported
by the work previously performed by Byrne and Hills (13) in a
similar population.
Even though the nonequivalency of the %HRR and %VO2peak
relationship with the line of identity has been demonstrated
many times (10–13,15) this study is the first one, to our knowl-
edge, that clearly demonstrate an effect of the degree of obe-
sity on the %HRR and %VO2peak relationship. In fact, method
B was significantly shifted to the right of the line of identity for
all three groups and with the T3 group mean regression being
significantly different than the T1 group. Moreover, similar
to method A, the degree of obesity was the only descriptive
variable associated to method B where it accounted for ~15%
of the variation. Therefore, the largest difference between the
HRR and VO2peak method was observed with the T3 group at
lower exercise intensities where 40 and 50% of VO2peak equaled
to 22.4 and 34.5% of HRR, respectively. Interestingly, method B
presented the same pattern when separated in degree of obe-
sity as we observed with method A with the only difference of
being shifted more toward the right of the line of identity.
Method C: relationships between %HRmax and %VO2R
Strong significant differences between method C and the line
of identity defined by Howley (16) were found either when

obesity | VOLUME 16 NUMBER 9 | SEPTEMBER 2008 2093

articles
intervention and Prevention
all subjects were grouped together or for any group defined
by the degree of obesity. Both one-way ANOVA and stepwise
multiple regression have shown no effect of the degree of
obesity on method C. On the other hand, very few studies
have directly challenged the %HRmax and %VO2R relation-
ship against the latest line of identity defined by Howley (16)
and adopted by the ACSM (2). Lounana et al. (15) recently
regressed the %HRmax against %VO2R in a group of elite
cyclist but compared their results with the previous line of
identity defined by the ACSM (4). Interestingly, the stepwise
multiple regression showed that resting HR was the strongest
predictor of method C where it accounted for ~35–39% of the
variation. These findings support the rational of taking into
account the resting HR to determine the exercise target HR
in this population. Thus, our results suggest the importance
of using the HRR method for exercise intensity prescription
purposes as opposed to the HRmax method for overweight
to severely obese individuals when oxygen uptake measure-
ments are not available.
Method D: relationships between %HRmax and %VO2peak
As observed previously with method C, there was a strong
variation between method D and the line of identity defined
by Howley (16) when all the subjects were grouped together
or separated by degree of obesity. Again, no effect of the
degree of obesity has been observed statistically. A stepwise
multiple regression also showed that resting HR accounted
for ~28–32% of the variation, which strongly suggests the
importance of considering the resting HR when prescrib-
ing exercise target HR in overweight to severely obese indi-
viduals when O2 measurement is not available. Furthermore,
Howley in 2001 (16) proposed a table with different %VO2peak
equivalencies to a similar %HRmax according to one’s func-
tional capacity in METs following the equation: “%VO2peak =
((100% − %VO2R)/MET) + %VO2R.” Even though this table
has been done to improve the equivalency of relative exer-
cise intensity with the %VO2peak as opposed to other method,
the usage of such table remains too cumbersome for clini-
cal purposes. Moreover, having the resting HR associated to
approximately one-third of method D reinforces the fact of
using the HRR method when O2 uptake data are not available
in obese and sedentary individuals.
In summary, the main finding of this study is that both
%HRR − %VO2R and %HRR − %VO2peak relationships are sig-
nificantly affected by the degree of obesity, especially at low
to moderate exercise intensities. However, this study did not
permit the researchers to identify the possible physiological
factor associated to the degree of obesity that might trigger
the variation in both %HRR − %VO2R and %HRR − %VO2peak
relationships. Furthermore, the %HRmax − %VO2R and %HRmax
− %VO2peak relationships were affected by the resting HR inde-
pendently of the degree of obesity, which suggests the prefer-
ence of using the HRR method when O2 uptake measurements
are not available. These findings have considerable importance
on the application of the current ACSM’s guidelines regarding
the prescription of exercise intensity in obese individuals.
2094
Acknowledgment
We sincerely acknowledge the work of Monique Dufour Doiron and Mathieu
Bélanger in the data collection process of this study.
Disclosure
The authors declared no conflict of interest.
© 2008 The Obesity Society
References
1. Jakicic JM, Clark K, Coleman E et al. American College of Sports Medicine
position stand. Appropriate intervention strategies for weight loss and
prevention of weight regain for adults. Med Sci Sports Exerc 2001;33:
2145–2156.
2. American College of Sports Medicine. ACSM’s Guidelines for Exercise
Testing and Prescriptions, 7th edn. Lippincott Williams & Wilkins:
Philadelphia, PA, 2005.
3. American College of Sports Medicine position stand. The recommended
quantity and quality of exercise for developing and maintaining fitness in
healthy adults. Med Sci Sports Exerc 1978;10:vii–x.
4. American College of Sports Medicine position stand. The recommended
quantity and quality of exercise for developing and maintaining
cardiorespiratory and muscular fitness in healthy adults. Med Sci Sports
Exerc 1990;22:265–274.
5. American College of Sports Medicine position stand. The recommended
quantity and quality of exercise for developing and maintaining
cardiorespiratory and muscular fitness, and flexibility in healthy adults. Med
Sci Sports Exerc 1998;30:975–991.
6. American College of Sports Medicine. ACSM’s Guidelines for Exercise
Testing and Prescriptions, 6th edn. Lippincott Williams & Wilkins:
Philadelphia, PA, 2000.
7. Achten J, Jeukendrup AE. Heart rate monitoring: applications and
limitations. Sports Med 2003;33:517–538.
8. Skinner JS et al. Heart rate versus %VO2max: age, sex, race, initial fitness,
and training response—HERITAGE. Med Sci Sports Exerc 2003;35:
1908–1913.
9. Karvonen MJ, Kentala E, Mustala O. The effects of training on heart rate.
Ann Med Exp Biol Fenn 1957;35:307–315.
10. Swain DP, Leutholtz BC. Heart rate reserve is equivalent to %VO2 reserve,
not to %VO2max. Med Sci Sports Exerc 1997;29:410–414.
11. Swain DP, Leutholtz BC, King ME, Haas LA, Branch JD. Relationship
between % heart rate reserve and %VO2 reserve in treadmill exercise. Med
Sci Sports Exerc 1998;30:318–321.
12. Brawner CA, Keteyian SJ, Ehrman JK. The relationship of heart rate
reserve to VO2 reserve in patients with heart disease. Med Sci Sports Exerc
2002;34:418–422.
13. Byrne NM, Hills AP. Relationships between HR and VO2 in the obese. Med
Sci Sports Exerc 2002;34:1419–1427.
14. Colberg SR, Swain DP, Vinik AI. Use of heart rate reserve and rating of
perceived exertion to prescribe exercise intensity in diabetic autonomic
neuropathy. Diabetes Care 2003;26:986–990.
15. Lounana J, Campion F, Noakes TD, Medelli J. Relationship between
%HRmax, %HR reserve, %VO2max, and %VO2 reserve in elite cyclists. Med
Sci Sports Exerc 2007;39:350–357.
16. Howley ET. Type of activity: resistance, aerobic and leisure versus
occupational physical activity. Med Sci Sports Exerc 2001;33(Suppl 6):
S364–S369.
17. Londeree BR, Ames SA. Trend analysis of the %VO2 max—HR regression.
Med Sci Sports 1976;8:122–125.
18. Panton LB et al. Relative heart rate, heart rate reserve, and VO2 during
submaximal exercise in the elderly. J Gerontol 1996;51A:M165–M171.
19. Byrne NM, Hills AP, Hunter GR, Weinsier RL, Schutz Y. Metabolic equivalent:
one size does not fit all. J Appl Physiol 2005;99:1112–1119.
20. Gustafson AB, Farrell PA, Kalkhoff RK. Impaired plasma catecholamine
response to submaximal treadmill exercise in obese women. Metabolism
1990;39:410–417.
21. Salvadori A, Fanari P, Mazza P, Agosti R, Longhini E. Work capacity and
cardiopulmonary adaptation of obese subject during exercise testing. Chest
1992;101:674–679.
22. Miller WC, Wallace JP, Eggert KE. Predicting max HR and the HR-VO2
relationship for exercise prescription in obesity. Med Sci Sports Exerc
1993;25:1077–1081.
VOLUME 16 NUMBER 9 | SEPTEMBER 2008 | www.obesityjournal.org

23. Notarius CF, Rhode B, MacLean LD, Magder S. Exercise capacity and
energy expenditure of morbidly obese and previously obese subjects. Clin
Invest Med 1998;21:79–87.
24. Hulens M, Vansant G, Lysens R, Claessens AL, Muls E. Exercise capacity in
lean versus obese women. Scand J Med Sci Sports 2001;11:305–309.
25. Salvadori A et al. Kinetics of catecholamines and potassium, and heart rate
during exercise testing in obese subjects. Heart rate regulation in obesity
during exercise. Eur J Nutr 2003;42:181–187.
26. Mittendorfer B, Fields DA, Klein S. Excess body fat in men decreases
plasma fatty acid availability and oxidation during endurance exercise. Am J
Physiol 2004;286:E354–E362.
27. Ozcelik O, Aslan M, Ayar A, Kelestimur H. Effect of body mass index on
maximal work production capacity and aerobic fitness during incremental
exercise. Physiol Res 2004;53:165–170.
28. Fuller NJ, Sawyer MB, Elia M. Comparative evaluation of body composition
methods and predictions, and calculation of density and hydration
obesity | VOLUME 16 NUMBER 9 | SEPTEMBER 2008
articles
intervention and Prevention
fraction of fat-free mass, in obese women. Int J Obes Relat Metab Disord
1994;18:501–512.
29. Kaminsky LA, Whaley MH. Evaluation of a new standardized ramp
protocol: the BSU/Bruce Ramp protocol. J Cardiopulm Rehabil
1998;18:438–444.
30. Myers J, Bellin D. Ramp exercise protocols for clinical and cardiopulmonary
exercise testing. Sports Med 2000;30:23–29.
31. Howley ET, Bassett DR Jr, Welch HG. Criteria for maximal oxygen
uptake: review and commentary. Med Sci Sports Exerc 1995;29:
1292–1301.
32. American College of Sports Medicine position stand. Exercise and type 2
diabetes. Med Sci Sports Exerc 2000;32:1345–1360.
33. American College of Sports Medicine position stand. Exercise and
hypertension. Med Sci Sports Exerc 2004;36:533–553.
34. Jakicic JM, Otto AD. Physical activity recommendations in the treatment of
obesity. Psychiatr Clin North Am 2005;28:141–150.
2095