INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE RESTING METABOLIC RATE (RMR) comprises 60-75% of the total daily energy expenditure of sedentary persons (20). Changes in RMR could therefore impact greatly on the regulation of body mass and energy balance. In addition, the agewise decline in absolute energy expenditure for the RMR (12, 32) is reputed to be mainly due to decreases in fat-free mass (FFM), which approximates the respiring tissue mass (31, 35). These FFM losses may be caused by a combination of biological aging and decreasing levels of physical activity. A recent review stated that, although there is a large volume of literature, the chronic effect of exercise on the RMR is still equivocal (15). Furthermore, the only two investigations that, to our knowledge, have been conducted exclusively in groups of nonobese older women studied groups with mean ages of 42-46 yr and also presented conflicting results (26, 34). Notwithstanding the limitation of cross-sectional research designs, which do not control for genetic differences among groups, an RMR comparison between female long-term exercisers (LE) and female long-term nonexercisers (LNE) who are at least 50 yr of age affords the opportunity to probe for the relative effects of decades of strenuous activity and sedentation.

The doubly labeled water (DLW) method yields values for cumulative energy expenditure in the free-living state, but the literature contains data on only 51 women who are 49 yr of age (4, 10, 13, 19, 24, 27). A characteristic of older persons is their marked heterogeneity, which ranges from those being maintained in care facilities to those who live independently and compete in marathons (25). It is therefore vital that the range of energy expenditure in active and inactive older persons be determined accurately because this information has important implications for their nutritional requirements, which at present are largely unknown. Because the DLW method is expensive, it would be interesting to compare energy expenditure values obtained by using this methodology with those derived via the inexpensive procedure of factorial calculations from 7-day activity diaries.

The purposes of this study were therefore to determine 1) the relationship between RMR and aerobic fitness in 49- to 70-yr-old female LE and LNE, 2) the cumulative energy expenditures of the two preceding groups in the free-living state by using DLW, and 3) the validity with which activity diaries estimate energy expenditure.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects. Volunteers were recruited from advertisements placed in a newspaper and on community center and university notice boards. These advertisements called for healthy, mass-stable (±2.0 kg during the last year), nonsmoking 50- to 70-yr-old women who were either LE or LNE. The 12 most sedentary and 12 most active respondents who fulfilled the following additional criteria were selected for the study: a Quetelet's index ranging from 18 to 27 kg/m², no known chronic illness, no taking of medication known to affect energy metabolism, and no clinically significant eating disorder. Their descriptive statistics are contained in Table 1. The 12 LE engaged in 3.5-18.5 h/wk [8.6 ± 4.9 (SD) h/wk] of primarily aerobic exercise that was designed to either maintain or improve fitness. Three of the subjects were triathletes, and the remainder participated in combinations of running, swimming, cycling, aerobics, fitness classes, weight training, gymnastics, tennis, badminton, and netball. Three of them also professed to be lifetime exercisers, whereas the remainder had exercised regularly for the previous 5-41 yr (22 ± 13 yr). One-half of the LNE stated that they had never participated in a regular exercise program; the other six had not engaged in this type of activity for the previous 5-34 yr (16 ± 10 yr). The project was approved by the Flinders Medical Centre's Committee on Clinical Investigation. The purpose, test protocols, possible benefits, and risks were explained to the subjects before their written consent to participate was obtained.

RMR. RMR was measured during one habituation and two experimental trials. The subject controls and methodology have been outlined in a previous publication (33). None of the subjects was premenopausal, so there was no control for the effect of the menstrual cycle on RMR. The most recent data for the reliability of our calorimetry system yielded coefficients of variation of 1.6 and 2.3% for five and six intraday trials, respectively, conducted on two subjects. Interday trials, conducted six and seven times on another two subjects, resulted in coefficients of variation of 1.6 and 3.2%, respectively.

DLW. Total energy expenditure was determined by using DLW, as detailed elsewhere (4). In this investigation, because of financial constraints, only the eight most sedentary and eight most active subjects reported to the laboratory in the morning after an overnight fast for collection of a baseline urine sample and ingestion of the DLW (99.75% ²H₂O, Australian Nuclear Science and Technology Organization, Lucas Heights, NSW, Australia; 10.4% H₂¹⁸O, Isotec, Miamisburg, OH). Each oral dose of 50 mg ²H₂O and 150 mg H₂¹⁸O/kg body mass, respectively, was diluted to ~100 g with water. The doses were mixed in two batches. Samples of these batches and the water used for dilution were retained for subsequent analysis. The doses were drunk through a straw, and the completeness of isotope ingestion was maximized by the subjects also drinking three ~33-ml rinsings of the container. A second urine sample was collected on day 1 after a 3.5-h equilibration period. Subjects thereafter collected single midstream urine samples on days 4, 7, 10, 14, 17, and 21. They were instructed to empty their bladders 30-45 min before the recorded time of collection so that the sample consisted of newly formed urine. Each sample was then labeled and frozen at 5°C.

Body composition. The FFM was estimated by using a four-compartment model, which has been described elsewhere (36). This methodology is an improvement on hydrodensitometry because it controls for interindividual variability in total body water (TBW) and bone mineral (BM).

Submaximal test of aerobic fitness. The subjects were habituated to pedaling a dynamically calibrated Monark model 90652 cycle ergometer (Monark-Crescent, Varberg, Sweden), which was equipped with a lengthened seat pole and adult handlebars. During the second visit to the laboratory, the subjects pedaled the ergometer at three power outputs estimated to elicit ~55, 65, and 75% of a maximal heart rate, which was predicted as 220 minus the subject's age. There was a 1-min active recovery between each work bout, which was of at least 5-min duration. Heart rate was measured electrocardiographically during the last 10 s of each minute, and a "steady state" was regarded to have occurred if consecutive readings differed by <3 beats/min. A fan facilitated heat dissipation during the last two work bouts. Regression analysis was used to predict the power output at 75% of the age-predicted maximal heart rate.

Estimated energy intake and expenditure. The subjects completed 7-day diet and activity diaries in the week immediately after the DLW experiment. They were requested to maintain their normal eating and activity patterns during this period.

Statistical analyses. Independent two-tailed t-tests (P  0.05) were used for comparisons between LE and LNE. Dependent two-tailed t-tests (P  0.05) were used when comparing energy expenditure via DLW with energy expenditure from 7-day activity diaries and energy intake from 7-day diet diaries. Comparisons among the three RMR trials for body mass, RMR [MJ/day and O₂ uptake (O₂) ml/min], respiratory exchange ratio (RER), and resting heart rate (HR_rest) were conducted by using single-factor repeated-measures analysis of variance (ANOVA); none of these data sets violated the assumption of sphericity. Between-trial differences were identified by Tukey post hoc tests in the event of a statistically significant F-ratio (P  0.05).

1

1

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Subject characteristics. The descriptive statistics for the 12 LE and 12 LNE are contained in Table 1. There were no statistically significant differences (P > 0.05) between the groups for age, height, mass, and FFM, but the LE registered lower means than the LNE for Quetelet's index (P = 0.05) and percent body fat (P = 0.0005). Both the power output scores on the cycle ergometer, which are markers of aerobic fitness, were also higher (P = 0.0001) for the LE. Similar trends were exhibited by the data in Table 2 for the 16 subjects who participated in the DLW experiments, except that the difference in body mass was large enough to attain statistical significance (P= 0.03).

RMR. Repeated-measures ANOVA demonstrated that body mass and RER did not differ significantly (P > 0.05) across the three RMR trials; however, significant decreases between trials 1 and 2 were identified for RMR (O₂ = 188 and 183 ml/min; 5.47 and 5.27 MJ/day) and HR_rest (63 and 60 beats/min), respectively. Although there were no differences (P > 0.05) between the means for the second and third trials (O₂ = 183 ml/min; 5.24 MJ/day; HR_rest 60 beats/min), the lowest measured value was regarded as each subject's true RMR (MJ/day). Therefore, the RER and HR_rest data in Table 3 are for this true RMR collection period. It should be noted that the subject with the RER of 1.06 was on the Pritikin diet.

1

1

1

1

Energy intake and energy expenditure. Diary data for the 12 LE and 12 LNE are presented in Table 3. There was no difference (P = 0.71) between the two groups for energy intake estimated from 7-day diet diaries, but the LE achieved higher values for energy expenditure (P = 0.02) and exercise energy expenditure (P = 0.0001).

DLW. The methodological results are contained in Table 4. The 1.018-1.046 range for the ratios of the isotope dilution spaces is within that of 1.015-1.060, which is recommended by Prentice et al. (22).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Differences between energy expenditure (EE) from 7-day activity diaries (7 DAD) and EE measured via doubly labeled water (DLW). LE and LNE, long-term exercising and long-term nonexercising, respectively.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Differences between energy intake (EI) from 7-day diet diaries (7 DDD) and EE measured via DLW.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The RMR of the twelve 49- to 70-yr-old LE women was significantly greater than that of their sedentary counterparts at the 0.06 level when control was exerted for the effect of FFM via ANCOVA. Table 3 also indicates that the RMR of the LE was significantly greater than that of the LNE when energy expenditure was expressed both absolutely (P = 0.02) and per kilogram of body mass (P = 0.0002). This latter finding is because the body mass of the LE comprised a greater percentage of FFM (P = 0.0005) and lower percentage of fat mass (P = 0.0005) than that of the LNE, and the FFM has a much higher rate of resting energy expenditure (Ref. 7; 124.3 kJ · kg¹ · day¹) than adipose tissue (Ref. 7; 18.8 kJ · kg¹ · day¹). The body composition difference between the groups also translates into a higher power-to-weight ratio for the LE and hence a greater buffer between their physiological capacity and the physiological demands of everyday living. These advantages have implications for increased functional independence and an improved quality of life.

Our statistical power analysis indicated that 12 subjects/group were required to detect a 10% RMR difference (kJ · kg FFM¹ · day¹). ANCOVA of the megajoules per day data with FFM as the covariate is a valid method of probing for differences between the groups because it removes the linear effect of FFM on RMR and does not assume a zero intercept when these variables are graphed. The resultant adjusted means from the ANCOVA were 5.31 and 5.07 MJ/day for the LE and LNE, respectively. Although the mean difference of 4.7% did not quite reach the preset 0.05 alpha level, it is greater than the earlier reported reliability of our calorimetry system, and furthermore there are only 6 chances in 100 that this observed difference occurred by chance. Also, notwithstanding the borderline statistical significance, a real difference of this magnitude is unequivocally physiologically significant from the cumulative point of view. Nevertheless, further studies with larger numbers of subjects need to be conducted.

Most studies that examined the chronic effect of exercise on RMR indexed energy expenditure against a FFM estimated via the two-compartment body composition model of hydrodensitometry. The major problem with this model is that it assumes an overall density of 1.100 g/cm³ (2) for the four FFM components (TBW, protein, BM, nonbone mineral). Individual FFM densities that are greater or lower than this constant will result in respective under- and overestimates of the relative body fat. Our four-compartment model incorporated measurements of body density, TBW, and BM via underwater weighing, isotopic dilution, and dual-energy X-ray absorptiometry, respectively. This model therefore attempts to control for biological variability in two of the four FFM components, thereby enabling the RMR to be indexed against a more valid estimate of the FFM.

Our previous work (33) on 19- to 31-yr-old women also demonstrated a significantly greater (P = 0.047) RMR per kilogam of body mass for a high-fitness group compared with a low-fitness one, but this statistically significant difference disappeared (P = 0.56) when control was exerted for the effect of FFM. This is at variance with our present findings of RMR differences between twelve 49- to 70-yr-old LE and LNE subjects in megajoules per day (P = 0.02), in kilojoules per kilogram per day (P = 0.0002), and when control is exerted for the effect of FFM via ANCOVA (P = 0.06). Toth and Poehlman (34) also confirmed a statistically significant RMR difference (P < 0.01) between younger groups of aerobically trained (n = 12; 42 ± 5 yr; 4.56 ± 0.42 kJ/min) and untrained women (n = 34; 44 ± 4 yr; 3.93 ± 0.42 kJ/min) with identical means for FFM. However, as part of a much larger investigation, Ryan et al. (26) studied age groups similar to those in the study by Toth and Poehlman (34) and reported no RMR differences between 10 female athletes and 10 control subjects. The balance of evidence from cross-sectional studies on older women therefore suggests that aerobic training results in a greater RMR both absolutely and relative to the FFM. Three longitudinal studies (10, 16, 21) also reported pooled data on men and women whose mean ages were approximately one decade greater than those of the subjects used in the present investigation. Meredith et al. (16) found no change in RMR per kilogram of body mass consequent to a 12-wk cycling program with a frequency of 3 times/wk at 70% of heart rate reserve. However, two studies from the same laboratory have demonstrated that 8 wk of cycling progressing up to 85% maximal O₂ increases 1) RMR per kilogram of mineral plus protein (10) and 2) RMR expressed as both kilocalories per minute and kilocalories per kilogram FFM per minute (21). They attributed these effects to the greater exercise intensity of their training program compared with that of Meredith et al. (16). The results of these two longitudinal investigations therefore support the findings of our cross-sectional study.

Although the DLW technique does not measure energy expenditure during discrete activities, it is regarded as a valid measurement of both the cumulative energy expenditure in subjects in the free-living state and the energy intake required to maintain body mass (17). A major finding of this study is that the energy expenditure measured via DLW in the LE (12.99 ± 3.58 MJ/day) was significantly greater (P = 0.01) than that of the LNE (9.30 ± 1.15 MJ/day) over the 21-day measurement period, even though there were no significant differences between the groups for age (P = 0.43) and FFM (P = 0.26). This is contrary to the findings of Goran and Poehlman (10), who conducted a longitudinal study on 11 elderly male and female volunteers aged 56-78 yr. They found that an 8-wk endurance training program did not increase total energy expenditure measured via DLW because there was a compensatory decrease in physical activity during the remainder of the day. They proposed that this finding could have been because their elderly volunteers found the 8-wk endurance training program difficult to complete during the last 2 wk, which was when the DLW data were collected. Conversely, Table 5 indicates that there was no difference (P = 0.96) between our groups for the energy expended during activity that was not intended to maintain or improve aerobic fitness. This was despite the LE having less time per day for such activity because of their physical training (1.2 ± 0.7 h/day). Although it can be argued that the greater overall energy expenditure of our LE was because they are more genetically predisposed toward exercise than the LNE, the strength of such cross-sectional designs is that they afford an opportunity to probe for the relative effects of decades of strenuous activity and sedentation by comparing the means of two divergent groups. The identification of intraindividual change and interindividual variability in change via a longitudinal research design with a control group and over a meaningful time period would pose major problems, such as subject retention, funding, continuity of researchers, and improvements in instrumentation.

The significantly greater (P = 0.01) mean difference between our subsamples of eight LE and eight LNE of 3.69 MJ/day in DLW energy expenditure together with the overall coefficient of variation of 28.6% emphasize the wide range of energy expenditure for healthy 49- to 70-yr-old women. This translates into a wide range of energy intake requirements to maintain energy balance. Published data on the DLW energy expenditure of women >49 yr old were located for only 51 subjects. The mean daily energy expenditure of 0.15 MJ/kg for our LNE is similar to those of 0.13-0.16 MJ/kg for these other studies (4, 9, 13, 19, 24, 27 ); the one subject >49 yr old reported by Clark et al. (4) registered a value of 0.18 MJ/kg. Although four of these groups studied subjects with mean ages in the seventies (24, 27) and sixties (9, 19), the 13 older Canadian subjects tested by Martin et al. (13) were very similar to our LNE for age (53.5 ± 2.2 yr) and mass (61.1 ± 7.2 kg). Their means for DLW energy expenditure (9.63 ± 3.73 MJ/day; 0.16 MJ/kg) are also almost identical to those for our LNE (9.30 ± 1.15 MJ/day; 0.15 MJ/kg), but their larger coefficient of variation (38.7 vs. 12.4%) reflects that, by design, our LNE were more homogeneous for energy expenditure. Nevertheless, their coefficient of variation was still larger than that of 28.6% for our overall group. The absolute and relative energy expenditures of our LE (12.99 ± 3.58 MJ/day; 0.23 MJ/kg) are much higher than those that have been reported on younger (4), similarly aged (13), and older subjects (9, 19, 24, 27). The energy expenditures of the eight 49- to 70-yr-old LE are even comparable to the data reported on young distance runners (Ref. 30: 11.84 ± 1.31 MJ/day, 0.23 MJ/kg; Ref. 6: 12.52 ± 1.74 MJ/day, 0.23 MJ/kg). The data on our LE emphasize that aging per se is no excuse for decreasing habitual physical activity. A greater physiological capacity in older persons should translate into an increased functional independence and improved lifestyle.

The overall mean of 11.80 MJ/day for energy expenditure via 7-day activity diaries was similar to that of 11.14 MJ/day (P = 0.33) for the expensive criterion method using DLW. The former therefore appears to be a less expensive alternative for estimating the mean energy expenditure of a group. However, Fig. 1 highlights large intermethod differences, which were primarily for the LE, but their overestimates tended to be balanced by underestimates. Interestingly, the statistically significant difference for the LNE was due to the homogeneity of their deviations. The range of intermethod differences (5.20 to 3.60 MJ/day) and mean of absolute differences of 2.27 MJ/day therefore emphasize problems with the 7-day activity diary at the individual level. Major limitations include obtaining an accurate record, estimating the intensity of the various listed tasks/activities because energy expenditure is directly proportional to intensity, and then assuming that everyone has the same mechanical efficiency.

The body mass changes (0.086 ± 0.389 kg) in our subjects over the 21 days of the DLW experiment were within the range of diurnal variation, and the corresponding fluctuations in the estimates of FM via TBW were all less than the smallest change that can be detected by such a method (11). If we therefore assume that our subjects were in energy balance, then the 7-day diet diaries underestimated the DLW energy expenditures for our LE, LNE, and combined group by 40.8, 17.0, and 30.8%, respectively. This greater disparity for endurance athletes, which is emphasized in Fig. 2, has been noted in a recent review (28), but the magnitude of our difference was not significantly correlated (r = 0.45, P < 0.10) with relative body fat via the four-compartment model. The differences for our 16 subjects ranged from 9.72 to 1.60 MJ/day, which represented respective under- and overestimates of 53.6 and 21.0%, but Schoeller (28) has emphasized that individual underestimates of 50% are not uncommon. Our overall mean underestimate is comparable to those reported by other investigators for older (9, 24) and similarly aged women (13). The 40.8% underestimate for our LE is of particular concern when one considers that DLW is the criterion method for measuring cumulative energy expenditure in the free-living state and that the subjects appeared to be in energy balance. The study design did not enable us to determine the relative contribution of selective or across-the-board underreporting, undereating during the 7-day measurement period, and inaccurate descriptions of food eaten.

Dividing the total daily energy expenditure determined via either DLW or activity diaries by the RMR yields the PAL ratio, which is analogous to the exercise physiology practice of expressing the intensity of discrete activities in metabolic equivalents or multiples of the RMR. A reference PAL of 1.4 has been quoted for persons whose only physical activity is light and occupational (5). The PAL of all 16 of our subjects were above this value. The highest mean of 2.48 logically occurred for the LE, and this was significantly greater (P = 0.03) than that of 1.87 for the LNE. The overall coefficient of variation of 27.2% yet again emphasizes the wide range of physical activity levels in a heterogeneous group of 49- to 70-yr-old women. Furthermore, although the PAL ratio should be independent of mass, Norgan (18) has demonstrated that it is elevated by 12.5% during standardized walking as the body mass increases from 45 to 90 kg. A further problem is that, if physical activity and body mass remain constant, then the agewise decrease in RMR, which is associated with a decrement in FFM, would elevate the PAL ratio. Clearly, the PAL ratio needs to be treated with caution before it can be universally regarded as an index of activity for comparison between groups, and further research is warranted.

In summary, our cross-sectional data on 49- to 70-yr-old women suggest that 1) aerobic-type training results in a greater RMR per unit time, per kilogram body mass, and also when control is exerted for the effect of the metabolically active FFM; 2) there is a large range in total daily energy expenditure estimated via DLW in weight-stable subjects; hence, there is wide variation in the energy intake necessary to maintain energy balance; 3) DLW experiments demonstrated that LE expended significantly more energy than LNE, but there was no difference between the groups for the thermic effect of activity that was not designed to either improve or maintain fitness. It would therefore appear that aerobic training does not result in a compensatory decrease in energy expenditure during the remainder of the day; 4) comparisons with DLW data indicate that 7-day activity diaries yield meaningful group means, but there are problems at the individual level; and 5) 7-day diet diaries underestimate energy requirements, and this error is significantly greater for LE than LNE.