Fitzgerald, Margaret D., Hirofumi Tanaka, Zung V. Tran, and Douglas R. Seals. Age-related declines in maximal aerobic capacity in regularly exercising vs. sedentary women: a meta-analysis.J. Appl. Physiol. 83(1): 160–165, 1997.—Our purpose was to determine the relationship between habitual aerobic exercise status and the rate of decline in maximal aerobic capacity across the adult age range in women. A meta-analytic approach was used in which mean maximal oxygen consumption (V˙o 2 max) values from female subject groups (ages 18–89 yr) were obtained from the published literature. A total of 239 subject groups from 109 studies involving 4,884 subjects met the inclusion criteria and were arbitrarily separated into sedentary (groups = 107; subjects = 2,256), active (groups = 69; subjects = 1,717), and endurance-trained (groups = 63; subjects = 911) populations.V˙o 2 max averaged 29.7 ± 7.8, 38.7 ± 9.2, and 52.0 ± 10.5 ml ⋅ kg−1 ⋅ min−1, respectively, and was inversely related to age within each population (r = −0.82 to −0.87, allP < 0.0001). The rate of decline inV˙o 2 max with increasing subject group age was lowest in sedentary women (−3.5 ml ⋅ kg−1 ⋅ min−1⋅ decade−1), greater in active women (−4.4 ml ⋅ kg−1 ⋅ min−1⋅ decade−1), and greatest in endurance-trained women (−6.2 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1) (all P < 0.001 vs. each other). When expressed as percent decrease from mean levels at age ∼25 yr, the rates of decline inV˙o 2 max were similar in the three populations (−10.0 to −10.9%/decade). There was no obvious relationship between aerobic exercise status and the rate of decline in maximal heart rate with age. The results of this cross-sectional study support the hypothesis that, in contrast to the prevailing view, the rate of decline in maximal aerobic capacity with age is greater, not smaller, in endurance-trained vs. sedentary women. The greater rate of decline inV˙o 2 max in endurance-trained populations may be related to their higher values as young adults (baseline effect) and/or to greater age-related reductions in exercise volume; however, it does not appear to be related to a greater rate of decline in maximal heart rate with age.
- maximal oxygen consumption
maximal aerobic capacity, as measured by maximal oxygen consumption (V˙o 2 max), decreases with advancing age (4, 20). This decrease contributes to the reduction in physiological functional capacity observed with advancing age, which eventually can result in a loss of independence in older adults (5, 9, 12). Moreover, because maximal aerobic capacity is an independent risk factor for cardiovascular and all-cause mortality (2, 3), the age-related decrease may also contribute to premature death in middle-aged and older adults.
In light of the physiological and clinical significance of maximal aerobic capacity, lifestyle factors that may be associated with a reduced rate of decline inV˙o 2 max with advancing age are of considerable public health interest. In this context, it has been reported that the rate of decline inV˙o 2 max with age is smaller in endurance-trained male athletes than in sedentary men (8,11). Based largely on these data in men, the concept has been established and widely promoted that the rate of decline in maximal aerobic capacity with age is attenuated in adults who perform regular aerobic exercise (8, 11, 12).
In contrast to these findings in men, Wells and colleagues (25) and Astrand et al. (1) have reported rates of decline inV˙o 2 max with age in physically active women that are greater than that generally reported for sedentary women. Recently, we (6) observed a rate of decline inV˙o 2 max with age among highly trained female distance runners that was even greater than that reported in these earlier studies (1, 25). However, the relatively small sample sizes, limited age ranges, and lack of sedentary control groups in all of these studies (1, 6, 25) preclude drawing any conclusions concerning this issue.
Accordingly, the primary purpose of the present investigation was to determine the relationship between habitual aerobic exercise status and the rate of decline inV˙o 2 max across the adult age range in women. Our secondary aim was to determine the relationship between aerobic exercise status and the rate of decline in maximal heart rate with age. On the basis of the general prevailing view (8,11, 12), two specific hypotheses were tested:1) women performing regular aerobic exercise demonstrate a slower age-related rate of decline inV˙o 2 max than do sedentary women; and 2) the slower rate of decline inV˙o 2 max in women performing aerobic exercise is associated with a reduced rate of decline in maximal heart rate, an important determinant of the age-related reduction inV˙o 2 max (5, 10,16).
To test these hypotheses, we used a meta-analytic approach in which mean V˙o 2 max values of female subject groups across the adult age range were obtained from the published literature. The subject groups were classified according to their aerobic exercise status, and the rate of decline inV˙o 2 max with increasing subject group age was determined for each of these populations. We postulated that new insight into these issues might be gained by using such a large-population approach.
Meta-analysis is a quantitative approach by which mean results from different experimental studies are sorted, classified, and summarized as parametric data (7). It is also the application of research methodology to the characteristics and findings of studies. Meta-analysis in the present study was conducted as described previously (24). Briefly, as an initial step, an extensive literature search was conducted to identify as many English-language studies (1960–1996) as possible in whichV˙o 2 max was measured in women. This was done by using computer searches (via Sport Discus and Medline) using the key words aerobic fitness, maximal oxygen consumption, and women. In addition, extensive hand searching and cross-referencing were done by using bibliographies of already located studies. All mean values from previous studies meeting the following criteria for inclusion were analyzed:1) data on women reported separately; 2) age groups separated;3) at least five subjects per group;4) only the most recently published results used on a particular population;5) subject groups consisted of adult women, i.e., 18–89 yr of age;6)V˙o 2 maxvalues obtained by using objective criteria (23);7) maximal exercise protocols performed either on treadmills or cycle ergometers; and8) only healthy subject populations. A list of papers included in the meta-analysis can be obtained from the authors on request.
Because the studies included in the meta-analysis used different terms to describe the aerobic exercise status of their subject groups, we have separated and analyzed the groups in three arbitrarily defined categories: 1) endurance trained, referring to regular performance (≥3 sessions/wk) of vigorous endurance exercise (e.g., running, cycling, swimming, rowing, cross-country skiing) for ≥1 yr; 2) active, referring to occasional or irregular performance (≤2 sessions/wk) of aerobic exercise (walking, basketball, dancing, stairmaster exercise, etc.); and 3) sedentary, referring to no performance of aerobic exercise.
Subsequently, the important characteristics of all of the relevant studies located in the literature search were classified and coded. To integrate the differing methodologies (subjects, results, and so on), a coding sheet was constructed. Primary variables coded included the following: 1) study characteristics,2) physical characteristics of subjects, 3) exercise program characteristics, and 4)V˙o 2 max and maximal heart rate values.
Data from treadmill and cycle ergometry exercise were evaluated together and separately. There were no differences in results between the two analyses. Therefore, data from both exercise modes were pooled and are presented together. Because we have previously shown that weighted results (by sample size) were not significantly different from unweighted results (24), no weighting scheme was used in the present meta-analysis.
Of the key dependent variables, complete data were available forV˙o 2 max , age, and body weight on all groups. Maximal heart rate values were missing in 10–15% of the subject groups; therefore, for this variable, analyses were performed on the available database only.
Linear regression analyses were performed to determine the association among variables. In all cases, age was used as the predictor variable. Pearson’s product-moment correlation coefficients were used to indicate the magnitude and direction of relations among variables. One-way analysis of variance was used to determine differences in the dependent variables (e.g., percent and absolute decline inV˙o 2 max) among populations (sedentary, active, and endurance-trained). When overall significance was indicated, the Tukey’s honestly significant difference method for multiple comparisons was used to differentiate among the three group means. All data were reported as pooled means ± SD. The statistical significance level was set atP < 0.05 (2-sided tests) for all analyses.
Mean descriptive data for the three subject populations are shown in Table 1. A total of 239 groups (4,884 subjects) and 109 studies met the criteria for inclusion. There were 63 groups (n = 911) in the endurance-trained category, 69 groups (n = 1,717) in the active category, and 107 groups (n = 2,256) in the sedentary category. Mean age was 5–7 yr greater in the sedentary women compared with the two physically active populations. Body mass was ∼4 kg less in the endurance-trained vs. the active and sedentary populations. As would be expected given our subject selection criteria,V˙o 2 max was lowest in the sedentary women, somewhat higher in the active women, and highest in the endurance-trained women.
Rate of decline inV˙o2 max with age.
V˙o 2 max was strongly inversely related to age in each of the three populations (r = −0.82 to −0.87, allP < 0.0001; Fig.1). The rate of decline inV˙o 2 max with increasing subject group age was lowest in the sedentary women (−3.5 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1), somewhat greater in the active women (−4.4 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1), and greatest in the endurance-trained women (−6.2 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1) (all P < 0.001 vs. each other, Fig.2, Fig.3 A, Table2). In contrast, the rate of decline in V˙o 2 max when expressed as percent decrease from mean levels at age ∼25 yr was similar among the three subject populations (−10.0 to −10.9%/decade; Table 2, Fig.3 B). Relatively few values forV˙o 2 max were available beyond 65 and 70 yr of age, respectively, for the endurance-trained and active populations (Fig. 1, A andB). To account for this potential influence, the analysis also was performed by using 60 and 70 yr of mean subject group age as end points. These analyses provided the same results as did the original analysis.
Rate of decline in maximal heart rate with age.
Maximal heart rate was strongly inversely related to subject group age in each of the three populations (r = −0.89 to −0.91, all P < 0.001; Fig. 4). In contrast to the absolute declines in V˙o 2 max, however, the corresponding declines in maximal heart rate were not different in the three populations (−7.0 to −7.9 beats ⋅ min−1 ⋅ decade−1). Maximal heart rate was correlated withV˙o 2 max in the each of the three populations (r = 0.75–0.85, all P < 0.0001), explaining 73, 56, and 71% of the variance in the sedentary, active, and endurance-trained subjects, respectively.
At least two important new findings were generated by the present investigation concerning the relationship between habitual aerobic exercise status and the rate of decline in maximal aerobic capacity with adult aging. First, in marked contrast to current theory based primarily on data in men (8, 11, 12), our results indicate that the rate of decline inV˙o 2 max with increasing age is greater in endurance-trained women than in sedentary women. Specifically, we observed a direct, not inverse, relationship between habitual aerobic exercise levels and the rate of decline inV˙o 2 max with age across our three subject populations. Second, the present data indicate that there is no discernible relation between the rate of decline in maximal heart rate with advancing age in healthy women and their aerobic exercise status. Thus both of our working hypotheses were refuted by the present findings.
Rate of decline inV˙o2 max with age.
Only a modest amount of data exists in women concerning the relationship between aerobic exercise levels and the rate of decline inV˙o 2 max with age. In their classic 1987 review onV˙o 2 max and aging, Buskirk and Hodgson (4) summarized the results of several studies in women (their Table 3, p. 1826). No consistent relationship was demonstrated between the habitual exercise status of the population and the associated rate of decline inV˙o 2 max with age.
However, at least four independent lines of evidence support the present findings of a greater rate of decline inV˙o 2 max with age in highly active vs. sedentary women. In a previous cross-sectional study, Wells and colleagues (25) reported a rate of decline inV˙o 2 max with age in female distance runners (aged 35–70 yr) of −4.7 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1, which is much greater than that generally reported for sedentary women (see next paragraph). This is approximately the same high rate of decline that Astrand et al. (1) reported in a longitudinal investigation of active female physical education teachers who were studied ∼20 yr apart. Moreover, in our recent cross-sectional study (6) of highly trained female distance runners (aged 23–56 yr) who were matched for age-adjusted performance,V˙o 2 max was 12.3 ml ⋅ kg−1 ⋅ min−1lower in runners with a mean age of 52 yr vs. those with a mean age of 30 yr, corresponding to a decline of −5.6 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1. Finally, a chapter by Smith and Gilligan in a larger government publication (22) contained a cross-sectional analysis of data onV˙o 2 max women aged ∼18–80 yr taken from published studies up to the year 1987. No methods, results, or specific conclusions were presented. The scatterplot of their data and the accompanying legend (their Fig. 3, p. 297) describe the slopes of the declines inV˙o 2 max with age in “active” vs. “sedentary” women as not significantly different. However, the slope of the regression line for their active women was ∼22% greater than that for the sedentary women, which is similar to the difference observed in the active vs. sedentary populations in the present study.
During the development of the present manuscript, the results of a cross-sectional analysis of 409 healthy women aged 20–64 yr by Jackson and colleagues (13) were published. The analysis concerned the relationship between self-report physical activity levels, body fatness, and their interaction on the rate of decline inV˙o 2 max with age. Although the rates of decline inV˙o 2 max across age groups for each of their four self-report physical activity classifications were not presented, calculations based on their mean data (their Table 4, p. 887) indicate similar rates of decline among the four groups at any particular level of body fatness. One explanation for the apparent difference in these results and those of the present study is the high rate of decline inV˙o 2 max with age reported for their sedentary subjects (−5.4 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1). This is much greater than the respective rates for sedentary women reported by Buskirk and Hodgson (4) (range −2.8 to −3.5 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1), by Smith and Gilligan (22) (−3.0 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1), and in the present study (−3.5 ml ⋅ kg−1 ⋅ min−1 ⋅ decade−1) (Figs. 1, 2, 3) and would effectively preclude the ability to show differences from highly active women.
The prevailing concept of a smaller rate of decline in maximal aerobic capacity with age in regularly exercising/endurance-trained compared with sedentary individuals (8, 11, 12) is logical based on our understanding of the physiological adaptations to regular exercise and is certainly attractive from a “preventive gerontology” standpoint. However, an argument also can be made for hypothesizing greater rates of decline inV˙o 2 max with age in endurance exercise-trained adults.
The first argument involves a baseline effect, i.e., the law of initial baseline applied toV˙o 2 max. Stated simply, according to this concept those individuals with the highest levels ofV˙o 2 max as young adults should demonstrate the greatest rates of decline with advancing age. Support for this idea is provided by our data on the “relative” rates of decline inV˙o 2 max. Specifically, when this baseline effect is removed by expressing the data as percent change from mean levels at age ∼25 yr, the rates of decline inV˙o 2 max with age in the endurance-trained, active, and sedentary women are similar. It is interesting (and possibly instructive) that a similar relationship exists in the age-related rates of decline in V˙o 2 max in men vs. women. Men have higher levels of V˙o 2 max as young adults compared with women but demonstrate a greater absolute rate of decline in V˙o 2 max with age compared with women based on cross-sectional data (4, 22). When expressed as percent change, however, gender-related differences are no longer evident.
The second argument involves declines in habitual aerobic exercise levels with advancing age. Studies of male endurance athletes indicate that training volumes of older athletes often are up to 50% lower than those of their young adult colleagues (5, 15, 19). In our recent study (6), female distance runners matched for age-adjusted performance, weekly training mileage was 45% lower in women runners with a mean age of 52 yr vs. that for runners with a mean age of 30 yr; weekly training frequency and running velocity also were lower in the older runners. In the present meta-analysis, weekly running mileage was reported for 25 subject groups in 14 studies. The data are plotted in Fig. 5 and reveal a progressive, significant age-related decline. Based on the regression line from this sample, running mileage would be expected to decline from ∼90 km/wk at age 20 yr to <30 km/wk at age 70 yr, a decrease in excess of 65%. These observations support the view that overall aerobic exercise levels decline markedly with age in endurance-trained adults. Because sedentary young adults, by definition, are not performing regular aerobic exercise, it follows that the magnitude of decline in the intensity, duration, and frequency of aerobic exercise with age, which collectively have a strong influence onV˙o 2 max (17), is much greater in regularly exercising individuals. Thus this greater decrease in the overall exercise stimulus for maintaining maximal aerobic capacity could contribute to a greater rate of decline inV˙o 2 max with age in exercise-trained vs. sedentary adults.
Finally, because V˙o 2 maxis traditionally expressed in units corrected for differences in body weight, it is possible that greater increases in body weight with age in the exercising groups contributed to their greater rates of decline in maximal aerobic capacity in the present study. This would appear to be a reasonable possibility in that young adults who perform aerobic exercise on a regular basis demonstrate lower levels of body weight due to lower body fatness compared with their sedentary peers and, therefore, might tend to gain more weight with age as they undergo a greater decline in exercise-related energy expenditure. This does not appear, however, to have played a role in the results of the present study. Body weight increased with age within each of our populations (allP < 0.01), but the slopes of regression lines representing the respective rates of increases were not different from each other.
Age-related rate of decline in maximal heart rate.
Because maximal heart rate is considered to be an important determinant of age-related decline inV˙o 2 max (10,16, 19), the question has been raised as to whether the greatest declines in V˙o 2 max with age are associated with the largest reductions in maximal heart rate (8, 10, 11). The present findings indicate that there is no obvious relationship between age-related declines inV˙o 2 max and maximal heart rate in women differing widely in habitual aerobic exercise status. The data from the aforementioned analysis of Smith and Gilligan (22), as well as the earlier findings of Astrand et al. (1), support this observation. Taken together, these findings suggest that other factors (e.g., declines in maximal stroke volume or skeletal muscle oxidative capacity) were responsible for differences in the absolute rates of decline inV˙o 2 max observed in the exercising vs. sedentary populations in the present study.
Physiological significance and study limitations.
The present finding that the absolute rate of decline inV˙o 2 max with advancing age is directly related to aerobic exercise status suggests that the rate of loss of physiological functional capacity with age, at least that which depends on maximal aerobic power, would be greater in endurance-trained vs. sedentary women. Thus, from their high levels of functional capacity as young adults, exercise-trained women may experience greater absolute reductions over the normal life span.
We wish to emphasize, however, that despite their apparent greater rate of decline in V˙o 2 max at any age, women who regularly engage in aerobic exercise demonstrate significantly higher absolute levels of maximal aerobic capacity than do their sedentary peers. Therefore, on average, exercise-trained women enjoy a higher level of physiological functional capacity throughout adult aging compared with sedentary women. This means that endurance-trained females are able to perform physical tasks that cannot be performed by their sedentary peers (or would require a much greater effort and percentage of their physiological reserve).
The major limitation of the present investigation is its cross-sectional design. A longitudinal approach in which the same subjects are studied over their entire adult life span, theoretically, would provide more definitive insight into this issue. Saltin (21) has demonstrated that the average rate of decline inV˙o 2 max with advancing age in endurance athletes is similar whether they are studied cross-sectionally or longitudinally. Despite this finding, however, it is important to appreciate that the cross-sectional nature of the present study could have produced results that are different from those that would have been obtained from a longitudinal approach.
Finally, we recognize that it has been demonstrated that maximal aerobic capacity can be maintained over 10- to 20-yr periods in middle-aged men who are able to sustain their levels of aerobic exercise training (14, 18). The present findings are not in conflict with these previous observations in any way. Rather, our results support the hypothesis that on average the absolute rate of decline inV˙o 2 max in endurance-trained women as a whole is greater than that observed in less active women when viewed over the normal adult life span. As discussed previously, this may be due in part to a progressive and, ultimately, substantial reduction in their exercise habits with advancing age.
The results of the present study provide experimental support for the idea that the rate of decline in maximal aerobic capacity with age in healthy women is greatest in endurance-trained women and least in sedentary women, contrary to the currently accepted view. The greater rate of decline inV˙o 2 max in highly active women may be due to their high baseline levels as young adults and/or to greater reductions in their exercise levels with advancing age compared with sedentary women. Our findings also fail to provide evidence that the level of habitual aerobic exercise is related to the rate of decline in maximal heart rate with age.
This study was supported by National Institutes of Health (NIH) RO1 Awards AG-06537, AG-13038, and HL-39966. H. Tanaka was supported by NIH Individual National Research Service Award AG-05717.
Address for reprint requests: D. R. Seals, Univ. of Colorado, Dept. of Kinesiology, Campus Box 354, Boulder, CO 80309 (E-mail:).
- Copyright © 1997 the American Physiological Society