Longitudinal changes in aerobic power in older men and women

Liza Stathokostas, Shanthi Jacob-Johnson, Robert J. Petrella, Donald H. Paterson


The purpose of this study was to describe the longitudinal (10 yr) decline in aerobic power [maximal O2 uptake (V̇o2 max)] and anaerobic threshold [ventilatory threshold (Te)] of older adults living independently in the community. Ten years after initial testing, 62 subjects (34 men, mean age 73.5 ± 6.4 yr; 28 women, 72.1 ± 5.3 yr) achieved V̇o2 max criteria during treadmill walking tests to the limit of tolerance, with Te determined in a subset of 45. V̇o2 max in men showed a rate of decline of −0.43 ml·kg−1·min−1·yr−1, and the decline in V̇o2 max was consequent to a lowered maximal heart rate with no change in the maximum O2 pulse. The women showed a slower rate of decline of V̇o2 max of −0.19·ml·kg−1·min−1·yr−1 (P < 0.05), again with a lowered HRmax and unchanged O2 pulse. In this sample, lean body mass was not changed over the 10-yr period. Changes in V̇o2 max were not significantly related to physical activity scores. Te showed a nonsignificant decline in both men and women. Groupings of young-old (65–72 yr at follow-up) vs. old-old (73–90 yr at follow-up) were examined. In men, there were no differences in the rate of V̇o2 max decline. The young-old women showed a significant decline in V̇o2 max, whereas old-old women, initially at a V̇o2 max of 19.4 ± 3.1 ml·kg−1·min−1, showed no loss in V̇o2 max. The longitudinal data, vs. cross-sectional analysis, showed a greater decline for men but similar estimates of the rates of change in women. Thus the 10-yr longitudinal study of the cohort of community-dwelling older adults who remained healthy, ambulatory, and independent showed a 14% decline in V̇o2 max in men, and a smaller decline of 7% in women, with the oldest women showing little change over the 10-yr period.

  • aerobic fitness
  • treadmill testing
  • maximal oxygen uptake
  • aging
  • oxygen pulse
  • physical activity

aerobic fitness in the older population is related to health status (18), all-cause mortality (30), and functional capacity (19, 28, 32). Earlier, our laboratory reported the aerobic fitness [maximal O2 uptake (V̇o2 max) and submaximal aerobic power (ventilatory threshold; Te)] in a relatively large sample of both men and women, aged 55–85 yr, living independently in the community (9, 31). The opportunity arose to retest these individuals in a 10-year follow-up. By natural selection, this sample represents those who remained healthy, ambulatory, living independently, and capable of performing a treadmill test (“successful aging”). Thus the purposes of this study were 1) to examine the longitudinal change in maximal (V̇o2 max) and submaximal (Te) aerobic power of older adults and to compare the changes in men and women, 2) to compare the 10-yr changes in V̇o2 max in young-old (initially 55–62 yr) and old-old (initially 63–80 yr), and 3) to compare the change in V̇o2 max from the longitudinal vs. the cross-sectional data (in the sample drawn from the same population).

The majority of studies of the differences in aerobic fitness measures with age, or “age-related changes,” have been cross-sectional in design. Ideally, longitudinal studies are needed to observe the effects of age. A few longitudinal reports of aerobic fitness with age of older adults have described the rate of decline in “specialized” groups of masters athletes or endurance-trained subjects (15, 25, 26, 35). This information may not be generalizable to the “normal” healthy older population, and the existing data (4, 12, 21, 22, 35) in this regard give an inconsistent representation of the longitudinal rate of decline with age in the older population.

Aging itself is not uniform, either within an individual or between individuals, and it may affect different systems at different onset times and rates (33). Cross-sectional studies do not capture these individual changes, whereas longitudinal data are particularly advantageous in this regard. For example, as opposed to the customary gain in fat mass during middle-age and the importance of changes in body composition to the age-related aerobic fitness (21, 22), older age is often associated with a significant loss of lean body mass (19, 31). Also, exercise behaviors influence the age-related decline in aerobic fitness (21, 22), and across older ages in those remaining ambulatory and independent, it appears that cardiorespiratory fitness declines toward a minimum compatible with daily living (31). These factors might also relate to whether men and women show similar declines in their aerobic fitness in older age. The data of the present study also provided information to assess whether this older age group demonstrates a uniform change or whether there is an age period of maintenance or of accelerated loss. Thus we compared the rates of decline in a young-old group at follow-up of mean age 65–72 yr with the old-old group aged 73–90 yr (i.e., young-old vs. old-old).



The original cohort included 441 men and women aged 55–85 yr from a random sample from the city of London, Ontario (1987, population 280,000) as described by Koval et al. (27). From this group, 298 subjects achieved V̇o2 max criteria on a fatigue-limited treadmill walking test (31). Follow-up subjects were recruited from these 298 by letter, followed with a telephone call.

In the present 10-yr follow-up, 205 of the original 298 subjects who participated in treadmill testing were recruited for retesting (i.e., 69% recruitment rate). Reasons for no follow-up included death, loss of independence, refusal, and no contact. All of the experimental procedures were fully explained (in written and verbal format), and each participant signed an informed consent approved by The Review Board for Research Using Human Subjects at the University of Western Ontario. Inclusion criteria for treadmill retesting were on the basis of assessment by the study physician of the physical and cognitive abilities to walk on a motor-driven treadmill. Exclusion criteria for the present study were adopted from the initial testing and consisted of the following: cardiovascular disease (i.e., unstable angina or uncontrolled severe hypertension), blindness, orthopedic disabilities with inability to walk or confinement to a wheelchair, and the lack of mental and cognitive abilities to understand and execute the activities. Ninety subjects were excluded for medical reasons, health symptoms or signs identified early in the test, a general inability to accommodate to walking on a treadmill due to discomfort or poor balance, or inability to accommodate to the breathing apparatus (mouthpiece). Thus, in the 10-yr follow-up, 115 subjects now aged 65–96 yr maintained an independent lifestyle and were able to exercise on the treadmill.

Measurements and Testing Procedures

Testing procedures and measurements were carried out as closely as possible to the original study. Anthropometric measurements included height, body mass, and four skinfold thicknesses (triceps, biceps, subscapular, and suprailiac) taken with Harpenden calipers on the subject's right side. Body density was estimated from the logarithm of the sum of four skinfold measurements (14), and percent body fat was estimated by using Siri's (37) equation.

Physical activity was assessed by the Minnesota Leisure-Time Activity (LTPA) questionnaire (40). For the present data in older adults, the intensity codes (metabolic rate scores) for different activities, developed for middle-aged subjects, were reduced (proportional to the age-related slowing of self-paced walking) to acknowledge that older subjects would pursue these activities at a slower absolute pace.

The determination of V̇o2 max and associated cardiorespiratory variables was carried out using a treadmill test. Treadmill speed was set at 45.5 m/min (1.7 miles/h), and grade was held at 0% for the first 4 min of walking. This was followed by a grade or velocity increase or combination of velocity and grade increases in specific small increments to elicit a ramplike test. At the end of the test, the subjects walked on the treadmill for 2 min at 0% percent grade and a speed of 45.5 m/min for recovery. The protocol was designed to elicit O2 demand increases each minute, ranging from 1 to 3 ml·kg−1·min−1, and the protocol for each subject was selected to result in a test of 8–12 min. The study physician monitored electrocardiogram, blood pressure, and symptoms (e.g., chest pain, severe shortness of breath) throughout the treadmill tests.

Data Collection and Analyses

During the test, subjects breathed through a mouthpiece, and inspired and expired gas flows were measured using a low-dead-space (90 ml) bidirectional turbine (model VMM 110, Alpha Technologies) calibrated before each test using a syringe of known volume (0.990 liter). Respired gases were sampled continuously (1 ml/s) at the mouth and analyzed for fractional concentrations of O2, CO2, and N2 using a mass spectrometer (model MGA-1100, Perkin Elmer) calibrated daily against precision-analyzed gas mixtures. Changes in gas concentrations were aligned with gas volumes by measuring the time delay for a square-wave bolus of gas passing the turbine to the resulting changes in fractional gas concentrations as measured by the mass spectrometer. Breath-by-breath alveolar gas exchange data were calculated by using the algorithms of Beaver et al. (6), and data for V̇o2 max and Te were expressed as the 15-s average. Heart rate was recorded continuously by a 12-lead electrocardiogram, and blood pressure was recorded with manual sphygmomanometry. Maximal O2 pulse was calculated from the V̇o2 max and maximal heart rate (HRmax) measurements.

o2 max and Te criteria were replicated from the initial testing. V̇o2 max was considered to have been attained if the following criteria were reached: 1) a feeling of fatigue, and a 15-s plateau in O2 uptake concurrent with an increase in power output, wherein the O2 uptake increase over a 15-s period was less than one-half of the V̇o2 increase per work rate increase during the submaximal period of the test; or 2) a respiratory exchange ratio >1.0 and a heart rate within 5 beats/min of the age-specific HRmax (220 beats/min − age). Te was determined by visual inspection of data using the criteria outlined by Davis et al. (11) of a systematic increase in expired ventilation/O2 uptake (or end-tidal Po2) with no concomitant rise in expired ventilation/CO2 production or a decrease in end-tidal Pco2. Two independent investigators determined Te with discrepancies resolved by conference.

Statistical Analyses

Data analyses were performed with the Statistical Package for the Social Sciences version 11.0. Data collected during initial and follow-up testing were used to determine longitudinal changes in anthropometric, physical activity, and cardiorespiratory variables. Descriptive data are expressed as means ± SD. To compare the follow-up sample with the original random sample, the follow-up subset of men and women was removed from the original sample. Comparisons were made by independent t-tests, with Levene's test used to assess equality of variances. Comparison of original and follow-up data of study variables were performed by using a two-way (initial-10 yr, gender) repeated-measures analysis of variance. Analysis of covariance was used to assess the change in V̇o2 max over time, adjusted for changes in physical activity scores or lean body mass. Additionally, the longitudinal sample was divided into two age groupings (young-old, at follow-up aged 65–72 yr; old-old, aged 73–90 yr) to observe whether the rates of decline differed. Within- and between-group differences were assessed by analysis of variance within each sex. The difference between follow-up and initial values for study variables was used to express absolute and percent longitudinal changes. Significance level was set at P < 0.05.


Subject Participation

Table 1 describes subject participation in the longitudinal follow-up of performance and health-related tests. Average time between tests was 10.1 ± 0.4 yr. In 81 of the 115 subjects, the reason given for stopping the treadmill test was fatigue, and 62 (34 men, 28 women) of these subjects attained V̇o2 max criteria. Te was obtained for 45 subjects (22 men and 23 women).

View this table:
Table 1.

Subject numbers participating in treadmill testing at 10-yr follow-up

The subsets of 34 men and 28 women were removed from, and then compared with, the original larger sample for their representativeness. On the basis of the initial data, those who remained in the longitudinal sample of men were on the initial tests significantly older (73.5 ± 6.4 vs. 69.9 ± 8.1 yr), had a greater sum of four skinfolds (56.8 ± 15.3 vs. 47.5 ± 20.0 mm; P = 0.016), and had significantly greater percent body fat (27.8 ± 4.5 vs. 24.3 ± 4.0%). There were no differences in V̇o2 max. The longitudinal female subset had a greater initial V̇o2 max (1.39 ± 0.27 vs. 1.16 ± 0.23 l/min) and body mass-adjusted V̇o2 max (22.1 ± 4.6 vs. 18.5 ± 3.5 ml·kg−1·min−1; P = 0.026), with a similar percent body fat (36.2 ± 4.5 vs. 35.7 ± 4.4%) compared with the original sample. There was no difference in men or women for baseline activity scores of the follow-up [416 metabolic equivalents (MET)·min−1·day−1] vs. original sample (414 MET· min−1·day−1).

The male and female samples in the 10-yr follow-up (comparison in Tables 2 and 3) did not differ significantly in age. Men were taller with a greater body mass and body mass index (P = 0.023) than the women, whereas the women had a greater sum of skinfolds and percent body fat. Five women (of 28) in the follow-up sample were receiving hormone replacement therapy compared with 15 (of 146) women in the original sample.

View this table:
Table 2.

Comparison of anthropometric and physical activity data for longitudinal sample of men

View this table:
Table 3.

Comparison of anthropometric and physical activity data for longitudinal sample of women

Age-related Longitudinal Changes

Body composition.

In both men (P = 0.042) and women (P = 0.010), the increase in sum of four skinfolds was significant, and in women the increase in estimated percent body fat was also significant (P = 0.023). Nevertheless, there was no significant age-related change over the 10 yr in mean body mass or calculated fat-free mass for either men or women (Tables 2 and 3).

Total LTPA.

LTPA scores showed a significant increase over the 10-yr period (416 ± 216 to 584 ± 271 MET·min−1·day−1; P = 0.001). The increase in activity levels was significant in men (P = 0.005) and approached significance in women (P = 0.063) (Tables 2 and 3). Men vs. women were not different on baseline or 10-yr activity scores.

o2 max.

From the fatigue-limited treadmill test, HRmax averaged 95% of age-predicted maximum, and the mean respiratory exchange ratio was 1.09. The men showed a significant age-related decline in V̇o2 max (l/min) (Table 4), representing a decline of 0.035 l·min−1·yr−1. The change in V̇o2 max in the women (Table 5) was not significant (−0.006 l·min−1·yr−1) over the 10-yr period. In both men and women, there was a significant decrease in body mass-adjusted V̇o2 max. Calculated from the initial and follow-up means (Tables 4 and 5), the men showed a decrease of V̇o2 max of 0.43 ml·kg−1·min−1·yr−1 (P < 0.001) and the women 0.19 ml·kg−1·min−1·yr−1 (P = 0.010), corresponding to declines of 14.7 ± 21.1 and 7.0 ± 15.5% over the decade for men and women, respectively. The rate of decline in men compared with women was not statistically different (P = 0.064). The initial V̇o2 max of men was significantly higher than that of the women, with the rates of decline predicting a similar V̇o2 max in men and women at ∼80 yr of age. In the baseline data, total physical activity was not significantly correlated with V̇o2 max, nor was there a significant correlation with the quantity of activity characterized as heavy intensity. The significant change in V̇o2 max over the 10 yr remained significant after accounting for the change in LTPA as a covariate. In fact, the change in V̇o2 max was not significantly affected by the change in LTPA score (P = 0.232). This also was the case using the “heavy” activities score to represent activity values. From regression analysis of the change in V̇o2 max with the change in LTPA, the R2 was 0.024; that is, only 2.4% of the variance in the change in V̇o2 max was accounted for by the variance in the change in physical activity levels.

View this table:
Table 4.

Comparison of cardiorespiratory fitness data for longitudinal sample of men

View this table:
Table 5.

Comparison of cardiorespiratory fitness data for longitudinal sample of women

Whereas the mean V̇o2 max decreased, this was not an exclusive trend. Figure 1, A and B, shows the individual variability, and notably six men and seven women demonstrated increases in body mass-adjusted V̇o2 max over the 10-yr period. In both the initial and follow-up testing, these 13 subjects were fatigue limited and reached their age-predicted HRmax values (initial, 157 beats/min; follow-up, 145 beats/min), indicating valid results. Of the 13 subjects who demonstrated an increase in V̇o2 max, 10 reported having taken up an exercise training program, and the 13 subjects as a group demonstrated a significant increase in participation of structured exercise as reported in the LTPA (initial, 49.9 ± 98.9 METS·min−1·day−1 vs. follow-up, 53.5 ± 54.7 METS·min−1·day−1; P = 0.04).

Fig. 1.

A: individual age-related changes in body mass-adjusted maximal O2 uptake [aerobic power (V̇o2 max)] for male subjects (n = 34). Mean rate of decline was −0.43 ml·kg−1·min−1·yr−1 (14.7% per decade). B: individual age-related changes in body mass-adjusted V̇o2 max for female subjects (n = 28). Mean rate of decline was −0.19 ml·kg−1·min−1·yr−1 (7.0% per decade).


HRmax decreased significantly over the 10-yr period by a mean of 2.4 beats/yr in both men and women (Tables 4 and 5). (Five men were on β-blocker medication, and their data were removed from these heart rate analyses.) The calculated maximal O2 pulse was unchanged over the 10-yr period for men (13.2 ± 2.2 to 13.0 ± 2.9 ml/beat) and women (8.4 ± 1.5 to 9.5 ± 1.8 ml/beat) (Tables 4 and 5). The HRmax showed no difference in men compared with women, either initially or at follow-up; however, the O2 pulse (both initial and follow-up) was significantly higher in the men compared with women.


Te (l/min) was significantly reduced in the men after the 10-yr period, but it was not significantly changed for the women (Tables 4 and 5). Body mass-adjusted Te was not significantly changed for either men or women. Calculated from the initial and follow-up means (Tables 4 and 5), nonsignificant declines of 0.13 and 0.09 ml·kg−1·min−1·yr−1 in Te were observed for men and women, respectively. The Te as a percentage of V̇o2 max was not significantly changed with age in either men (80.2 ± 20.6 to 80.5 ± 11.5%) or women (79.8 ± 12.3 to 83.6 ± 6.7%). The individual variability of the changes in Te is depicted in Fig. 2, A and B.

Fig. 2.

A: individual age-related changes in ventilatory thresold [submaximal aerobic power (Te)] for male subjects (n = 22). Mean rate of decline was −0.14 ml·kg−1·min−1·yr−1 (2.5% per decade). B: individual age-related changes in Te for female subjects (n = 23). Mean rate of decline was −0.11 ml·kg−1·min−1·yr−1 (1.0% per decade).

Age-Related Longitudinal Changes Comparing Age Groups

In the men, only the old-old group showed a significant increase in skinfolds and loss of fat-free mass (−1.9 kg). There was no difference in the rate of decline of V̇o2 max comparing the young-old group (−4.2 ml·kg−1·min−1) vs. the old-old group (−4.4 ml·kg−1·min−1) (Fig. 3A). The reductions in HRmax were different (young-old: 19.1 beats/min; old-old: 30.0 beats/min; P < 0.001). The change in Te was not different for the young-old group (0.1 ml·kg−1·min−1) vs. the old-old group (−1.7 ml·kg−1·min−1). The women showed no between-age-group differences in the change in body composition. The V̇o2 max decline was significant in the young-old group (−3.1 ml·kg−1·min−1) but not in the old-old group (−0.2 ml·kg−1·min−1) (Fig. 3B), and thus there was a significant difference (P = 0.029) in the rates of decline between groups. Both age groups showed a similar reduction in HRmax (24 beats/min). The change in Te was not different for the young-old group (−1.9 ml·kg−1·min−1) vs. the old-old group (0.30 ml·kg−1·min−1).

Fig. 3.

A: individual age-related changes in body mass-adjusted V̇o2 max for male subjects divided by age groups. Left: young-old (from mean age 65.0 to 72.9 yr). Right: old-old (from mean age 73.0 to 90.0 yr). B: individual age-related changes in body mass- adjusted V̇o2 max for female subjects divided by age groups. Left: young-old (from mean age 65.0 to 72.9 yr). Right: young-old (from mean age 73.0 to 90.0 yr). Thick lines, mean data.

Comparison of Longitudinal Changes With Cross-Sectional Data

In men, the aerobic fitness variables in general showed a greater decline with age from the longitudinal data vs. the cross-sectional analysis of the initial data in the same individuals. Specifically, the V̇o2 max decline was −0.43 ml·kg−1·min−1· yr−1 compared with −0.25 ml·kg−1·min−1·yr−1 from the cross-sectional analysis of initial data in the same group (Table 6). In women, the longitudinal and cross-sectional data in general yielded more similar rates of change, with V̇o2 max declines of −0.19 and −0.17 ml·kg−1·min−1·yr−1, respectively (Table 6). Anthropometic data did not show consistent differences in the longitudinal vs. cross-sectional analyses.

View this table:
Table 6.

Comparison of cross-sectional differences across age vs. longitudinal rates of change per year for the same sample of men and women


This longitudinal study examined the relationship between age and aerobic parameters in a sample of independently living older men and women. Past investigations have been either cross-sectional, including large age ranges, or longitudinal but not representative of the general aging population. The present study is unique in the report of the decline of V̇o2 max and Te for a sample of a restricted age range in the older age group (initially 55–86 yr, mean age 72.8 yr) of both men and women over a 10-yr period.

In longitudinal studies, the validity of the data can be compromised, especially when the time between testing intervals is long. In the present study, this issue was minimized by the use of the same exercise test protocol and laboratory equipment setup for both test sessions. Continuous use of the laboratory setup throughout the 10-yr period gave consistent measures of O2 for given cycle work rates. Furthermore, during the time frame of this study, the laboratory achieved the Canadian Society for Exercise Physiology (CSEP) accreditation for cardiorespiratory testing, with measured V̇o2 max on athletes tested in the different laboratories within 5% of the mean standard of other accredited laboratories. Possible error in measurement techniques is acknowledged for the body composition measurements because the study investigators differed for the original and follow-up test sessions. This source of error was minimized by using investigators who were all trained and accredited by CSEP as Certified Fitness Consultants and by the use of carefully documented study protocols. Comparatively, this source of error would be minimal in light of an inherent 20% error in the use of skinfold measurement to determine fat-free mass in older adults (29).

In a descriptive study of the normal population, there also exists the problem of obtaining a representative sample. The original sample, as described by Paterson et al. (31), represented those from a random sample of the healthy older population able to complete a fatigue-limited exercise test. The characteristics of the follow-up sample in this age group are influenced by selective “survivorship” and morbidity, and thus this study necessarily describes the changes over time in those who remained ambulatory and independent (i.e., “successful,” healthy aging). Nevertheless, the initial characteristics of those available for the follow-up were similar to the original sample with only some exceptions. The returning sample of men and women both actually had a greater percent body fat, and the women, but not the men, had a significantly greater V̇o2 max (ml·kg−1·min−1; P = 0.03). Thus the longitudinal sample of men was not overrepresented by leaner and fitter men (as one might have expected), although the female sample may be biased with the fitter subgroup able to participate in the follow-up. Compared with the samples studied by other groups (as detailed in Tables 7 and 8), the V̇o2 max of the men and women of the present study was similar, or somewhat lower, although the age comparisons are not exact. Mean V̇o2 max of the men at age 63.5 yr was 26.5 ml·kg−1·min−1, and at age 73.5 y it was 22.0 ml·kg−1·min−1. Babcock et al. (5) reported a similar V̇o2 max of 23.1 ml·kg−1·min−1 for men of mean age 71.4 yr. Others have reported mean values of 28–29 ml·kg−1·min−1 for samples aged 60+ yr (12, 34, 37). The present values for the V̇o2 max of women were 22.1 ml·kg−1·min−1 at age 62.0 yr and 20.2 ml·kg−1·min−1 at age 72.1 yr. For groups aged 60+ yr, the literature gives mean values of 23.6 (13), 20.3 (34), 22.4 (39), and 23.1 ml·kg−1·min−1 (38). Overall, the sample of the present longitudinal study appears to be a reasonable representation of the healthy older population, for both the men and the women.

View this table:
Table 7.

Rate of decline in V̇ℴ2 max with age as reported for men in cross-sectional and longitudinal studies

As anticipated, V̇o2 max showed a decline with age in both men and women. The longitudinal rate of decline in men (−0.43 ml·kg−1·min−1·yr−1) was greater than that observed for the cross-sectional data of the original sample at −0.31 ml·kg−1·min−1·yr−1 (31). Other studies have generated the age-related decline from samples of a much broader age range and of mean age considerably younger. Nevertheless, other cross-sectional reports (Tables 7 and 8), have reported similar age-related rates of decline of −0.44 (24), −0.39 (17), −0.42 (42), and −0.46 ml·kg−1·min−1·yr−1 (21). Generally, the reported decline has been lower with values of −0.28 (12), −0.36 (34), −0.28 (5), and −0.33·ml·kg−1·min−1·yr−1 (38). One report gave a decline of −0.52 ml·kg−1· min−1·yr−1 (2). Compared with longitudinal reports (Table 7), the present findings were similar to the report by Åstrand et al. (2) of −0.44 ml·kg−1·min−1·yr−1 in former physical education students and by Rogers et al. (35) of −0.44 ml·kg−1·min−1·yr−1 in a sedentary group; however, they were greater than the decline of −0.32 ml·kg−1· min−1·yr−1 for masters athletes (35) and −0.27 of endurance-trained older men (25) and less than the rate of decline reported for a healthy sample by Dehn and Bruce (12) of −0.94 ml·kg−1·min−1·yr−1.

View this table:
Table 8.

Rate of decline in V̇O2 max with age as reported for women in cross-sectional and longitudinal studies

Contrary to the findings for the men, the longitudinal rate of decline for women of −0.19 ml·kg−1·min−1·yr−1 was slower than the cross-sectional report of −0.25 ml·kg−1·min−1·yr−1 (31). Compared with the decline given in other cross-sectional (−0.25 to 0.62 ml·kg−1·min−1·yr−1) and longitudinal (−0.44 to 0.62 ml·kg−1·min−1·yr−1) studies (Table 8), the present finding for women showed a considerably lower rate of decline. This is likely a reflection of the concept that those remaining independent in old age require a minimum level of aerobic fitness. With their lower V̇o2 max compared with men, only those experiencing a small decline in V̇o2 max would remain mobile and independent. Thus there may be a gender- related selective “survivorship.”

The decline in men was double the rate of that in women. Also, although the men had a greater initial V̇o2 max, the percent decline over the 10 yr was 14.7% in men and 7.0% in women. Previous reports (17, 21, 22, 42) have found a greater decline for V̇o2 max (ml·kg−1·min−1) in men. In the present longitudinal sample, this appears to be the case for men (−0.43 ml·kg−1·min−1) vs. women (−0.19 ml·kg−1·min−1) (P = 0.064).

Previous cross-sectional reports have shown a decrease in Te with age including older samples (5, 8, 31, 34, 41). However, in the present longitudinal data, the change with age was not significant (ml·kg−1·min−1) for either men or women. It appears that the cross-sectional results of the original study (31) closely predicted the longitudinal decline in men (−0.15 vs. −0.13 ml·kg−1·min−1) but overestimated the longitudinal decline in women (−0.14 vs. −0.09 ml·kg−1·min−1). Similar to Paterson et al. (31), the present data indicated no difference between men and women in the age-related rate of decline in Te. However, this is in contrast with Posner et al. (34), who found a greater decline in men vs. women. The present finding of a slower rate of decline in Te (ml·kg−1·min−1) than V̇o2 max with age (one-third to one-half) is consistent with previous reports (5, 10, 31, 34, 38).

For the HRmax, in the longitudinal data, compared with the common predictive equation (HRmax = 220 − age), the initial HRmax exceeded the predicted by 7 beats/min and the follow-up data were 6.6 beats/min below the predicted. Longitudinal analysis of the decline of HRmax yielded a significant reduction of 2.4 beats·min−1·yr−1 in both men and women. The decline for both men and women was greater than cross-sectional reports of a reduction of <1 beats·min−1·yr−1 (21, 31). In both men and women, the decline in V̇o2 max (l/min) was explained by the decline in HRmax such that there was no significant decline in O2 pulse, indicating the decline in V̇o2 max occurred without a reduced maximal stroke volume and/or arteriovenous O2 difference.

The rate of decline of V̇o2 max has also been related to changes in body composition (23) and reductions in activity levels (1, 21, 42). Toth et al. (42) in the study of men and women across ages 17–81 yr found that, when controlling for both fat mass and fat-free mass, the rate of decline of V̇o2 max was decreased by 50%. In the present study of older adults, however, there was no increase in percent body fat for men over the decade, the women showed only a moderate increase in body fat, and the fat-free mass was maintained across the 10-yr period for both men and women. For the present sample, the changes in percent body fat and fat-free mass were not significantly correlated with the decline of V̇o2 max with age. Analysis of covariance showed no significant effect on the change in V̇o2 max with the change of lean body mass as a covariate. The maintenance of fat-free mass contrasts to results of a 12-yr longitudinal study of nine similarly healthy male older subjects (19) who showed a 15–17% decline of muscle mass. It is possible that fat-free mass becomes overestimated with increasing age and that we could not detect the loss of fat-free mass or a relationship with the decline in V̇o2 max. In addition, the conflicting results may be sample dependent. That is, our returning sample was highly mobile and may not demonstrate the same rate of decline in lean body mass as in more frail samples described in strength studies. The quality of the muscle component of fat-free mass can also be compromised with age. Conley et al. (7) reported a decline in quadriceps oxidative capacity with age resultant from reductions in both muscle volume and oxidative capacity. However, reduced quality of oxidative capacity might not have occurred in the present subjects, who maintained their estimated lean body mass.

Cross-sectional analysis of physical activity scores with age showed no significant decline for this age group, which is in contrast to studies of a much greater age span (21, 22, 42). In fact, in this successful aging group, physical activity scores were increased over the 10-yr period. However, total LTPA (and also total score in activities characterized as heavy intensity) was not correlated with V̇o2 max, nor was the variance in the change in V̇o2 max affected by differences in physical activity; <3% of the variance in change in V̇o2 max could be accounted for by activity. Thus, contrary to findings that particularly relate to middle-age (21, 22, 42), in the older age groups, the present data showed that regular leisure time daily activities did not influence V̇o2 max. This finding warrants further investigation.

The observation of a decline in V̇o2 max over the 10-yr period was not an exclusive one. Although it has been shown that V̇o2 max declines with age, it has also been shown that increases in V̇o2 max after a training stimulus can occur in older adults (10). In the present study, checking the activity questionnaire at 10 yr revealed that 10 of the 13 subjects who demonstrated an increase in aerobic power were involved in a structured exercise program. Furthermore, analysis of the physical activity questionnaire data reflecting participation in structured exercise indicated a significant (P = 0.04) increase in participation by the 13 subjects who increased their V̇o2 max. Clearly, this is one factor related to the variability in age-related changes.


The purpose of the present investigation was to examine the longitudinal decline in V̇o2 max and Te in a general population of community-dwelling older adults and to analyze the data relative to gender and effects of body mass and composition and physical activity. Returning subjects were reasonably representative of the original larger sample. There was no significant loss of body mass or fat-free mass over the 10-yr period. The present 10-yr longitudinal description of older adults remaining independent showed a loss of body mass-adjusted V̇o2 max of 14.7% per decade for men and of 7.0% per decade for women. No significant decline in Te was observed for men or women or in the Te as a percentage of V̇o2 max, confirming that Te declines at a slower rate than V̇o2 max. The variance in the decline was not affected by the variability in LTPA scores. Comparison of cross-sectional and longitudinal rates of V̇o2 max decline derived from the same sample concur with previous reports of generally greater rates of decline obtained by longitudinal vs. cross-sectional study methods for the men; however, this was not the case for women. The older subgroup of women showed a maintenance of V̇o2 max over the 10 yr. Further research into the longitudinal contribution of various factors in the decline of V̇o2 max with age will assist in identifying potentially modifiable factors and thereby assist in the maintenance of independence throughout adult life.


This work was supported by grants from the Ontario Ministry of Health, Canadian Fitness and Lifestyle Research Institute, and Natural Sciences and Engineering Research Council of Canada.


View Abstract