Exercise Intensity and 5-Yr Trajectories in Peak Oxygen Uptake in Older Adults: Analysis from the Generation 100 Study

Abstract

Purpose:

The long-term impact of exercise training on change in cardiorespiratory fitness (CRF) with age remains uncertain. Given that CRF appears to be one of the strongest predictors of current and future health, we examined the differences between high-intensity training (HIT) and moderate-intensity training (MIT) on 5-yr changes in CRF in older adults.

Methods:

In this post hoc observational analysis of the Generation 100 Study, we included 500 participants (52% women; aged 70–77 yr) who consistently reported engaging in HIT or MIT for 5 yr. Participants underwent clinical assessments, including CRF measured as peak oxygen uptake ($\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$) at four timepoints. Linear mixed model was used to examine the longitudinal change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ and differences in trajectories across the 5-yr period.

Results:

Across 5 yr, men reporting HIT and MIT showed $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ declines of −3.1% (P = 0.007) and −7.7% (P < 0.001), respectively. Women reporting HIT showed no change (P = 0.96) in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, whereas those reporting MIT showed a decline of −4.6% (P < 0.001)_. At 5 yr, the difference in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ was 1.31 mL/kg/min (95% confidence interval [CI] 0.17–2.45) and 1.33 mL/kg/min (95% CI 0.28–2.37) in favor HIT in men and women, respectively. In exploratory analyses, higher self-reported exercise intensity was associated with 1.11 (95% CI 0.60–1.62) and 0.64 (95% CI 0.22–1.06) mL/kg/min smaller decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ in men and women, respectively, whereas weekly exercise duration showed weaker and inconsistent associations.

Conclusions:

Among men and women aged 70–77 yr, consistent engagement in aerobic exercise training was associated with more favorable age-related $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectories_, with HIT showing the smallest declines over 5 yr. Exploratory interaction analyses suggested that exercise intensity may become increasingly important over time for preserving $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$.

Cardiorespiratory fitness (CRF) declines markedly with age, typically by 15%–25% per decade after the age of 70 yr (1). Since the landmark study by Blair et al. (2), CRF has consistently been confirmed as a strong, independent predictor of all-cause mortality (3,4), cardiovascular disease (5), and loss of independence (6). Moreover, the magnitude of change in CRF over time is critical for risk prediction (7,8). Kokkinos et al. (9) found that a decline of 1.0 metabolic equivalent of task (METs, 3.5 mL/kg/min) over 6 yr was associated with a 9.0% higher (HR: 1.09, 95% CI 0.99–1.19) risk of all-cause mortality in adults with moderate CRF levels. A decline of >1 METs per decade is expected with aging (10). Given the strong evidence for CRF as a marker of health, maintaining a high age-specific CRF has been emphasized (11).

Exercise training and physical activity (PA) are key strategies for enhancing health in an aging population (12). Current guidelines recommend older adults to do 150–300 min/wk of moderate-intensity aerobic PA, or 75–150 min/wk of high-intensity aerobic PA, or a combination of both (13). Despite well-documented benefits, adherence remains low among older adults. In the United States (14) and Norway (15), only 10.8% and 31.0% of older adults, respectively, meet these PA guidelines. Both moderate- and high-intensity aerobic PA improve CRF in older adults (16), and randomized trials have shown superior effects of high-intensity compared with moderate-intensity aerobic exercise training on improvements in CRF (17,18). While the benefits of PA are well established, its role in slowing CRF decline with age remains debated. Valenzuela et al. (19) reported slower decline in CRF with regular PA and exercise training, whereas Hawkins et al. (20) found limited effects in older adults. A recent review (21) also questioned whether high intensity offers additional benefit over moderate intensity in improving CRF in older adults (>60 yr). Moreover, long-term effects of exercise training on CRF decline in older adults remain insufficiently examined. It is therefore of interest to better characterize the extent to which exercise training is associated with attenuated age-related decline in CRF, and to identify strategies that are associated for preserving it with increasing age.

The aim of this study was to examine how different exercise intensities over a 5-yr period were associated with longitudinal changes in CRF, measured objectively as peak oxygen uptake ($\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$), in older adults. We also explored the independent associations of exercise intensity and time/week with changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. We hypothesized that those exercising at higher intensity had a more favorable $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectory over 5 yr.

METHODS

Study design and participants

This study used data from the Generation 100 Study, the world largest exercise trial designed to investigate the long-term effects of exercise training on all-cause mortality in older adults (22). The present analysis departs from the original randomization and is based on a post hoc observational reclassification of participants according to their self-reported exercise intensity for each of the 5 intervention years. Recruitment procedures are described elsewhere (23), whereas inclusion and exclusion criteria are presented in Supplemental Box 1, https://links.lww.com/MSS/D418 (Supplemental Digital Content). In 2012, 1567 participants (790 women) aged 70–77 yr from Trondheim, Norway, were randomized 2:1:1 to a control group recommended to follow national PA guidelines, consisting of at least >30 min/d with moderate intensity (24), or into exercise groups of either supervised high-intensity interval training (HIIT) or moderate-intensity continuous training (MICT) 2 d/wk for 5 yr. The supervised HIIT and MICT protocols have been described in detail previously (23). In brief, HIIT intervention consisted of intervals targeting ~90% of peak heart rate, whereas MICT intervention consisted of continuous exercise targeting ~70% of peak heart rate_. Participants underwent repeated clinical testing, and questionnaires covering exercise habits and overall health.

Questionnaires

Information on health status, lifestyle, and exercise training, including cohabitation, smoking habits, was assessed using a standardized self-reported questionnaire (23). Cardiovascular diseases were defined as having a documented history of any of the following: acute myocardial infarction, unstable angina pectoris, atrial fibrillation and or flutter, supraventricular and/or ventricular tachycardia, stroke, coronary revascularization via coronary artery bypass grafting and/or percutaneous coronary intervention, pacemaker, or other invasive cardiovascular procedures. Exercise consistency was evaluated through questions assessing frequency, duration, and intensity, with intensity measured by the Borg scale (6–20,25).

Exercise frequency was determined by the question — How often do you exercise? <1 d/wk [0], 1 d/wk [1], 2 or 3 d/wk [2.5], nearly every day [5]. Duration was assessed by the question — For how long do you exercise each time? <15 min [7.5], 15 to 30 min [22.5], 30 to 60 min [45], >60 min [60]. Exercise time/week was calculated as the product of frequency and duration, using numbers in brackets (26).

We converted the Borg scale to METs using guidelines from the American College of Sports Medicine (27) (e.g., Borg 13 ≈ 4.4 METs). These MET values were used in analysis as a measure of intensity to assess the independent association of exercise intensity on longitudinal changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$_.

Adherence to exercise

Participants from any randomized arm, including controls, were reclassified post hoc and were eligible if they met the criteria for moderate exercise intensity (MIT, ≥ 30 min/wk exercise at a Borg scale ranging from 11 to 14) or high exercise intensity (HIT, ≥ 30 min/wk of exercise at ≥15 on the Borg scale) throughout the 5-yr period, as previously described (22). Because this reclassification was independent of the original randomization, the analytic sample is not expected to correspond to the number remaining in the randomized exercise arms reported previously (22). It should be noted that HIT in this context refers to all high-intensity exercise and does not necessarily imply interval training. Participants reporting any inconsistencies in exercise intensity were excluded from the analysis. Heart rate and rating of perceived exertion were recorded in the supervised exercise sessions. Original randomization is presented in Supplemental Table 1, https://links.lww.com/MSS/D418 (Supplemental Digital Content), and a flowchart of participant selection is shown in Figure 1.

FIGURE 1.

Illustrates the selection of those who reported to exercise with HIT or MIT for 5 yr regardless of original randomization.

Cardiopulmonary exercise test

Measurements of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ were assessed with a cardiopulmonary exercise test (CPET) using ergospirometry with either Cortex Metamax II (Leipzig, Germany, 95.4% of tests), or the Oxycon Pro (Erich Jaeger, Hoechberg, Germany, 4.6% of tests). Each participant was consistently tested using the same ergospirometry system across assessments (baseline, 1, 3, and 5 yr). All tests began with a warm-up phase, followed by an individualized test protocol in which workload increased approximately every 1.5 min, or earlier if $\dot{\mathrm{V}}{\mathrm{O}}_2$ stabilized. Testing continued until voluntary exhaustion ($\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$), or attainment of maximal oxygen uptake ($\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{m}\mathrm{a}\mathrm{x}}$). A maximal test was considered valid if participants continued until exhaustion and $\dot{\mathrm{V}}{\mathrm{O}}_2$ did not increase by >2 mL/kg/min over two consecutive 30sec epochs (plateau in $\dot{\mathrm{V}}{\mathrm{O}}_2$, despite increased workload), combined with a respiratory exchange ratio of >1.05. In this study, 34.8% did not meet the criteria for $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{m}\mathrm{a}\mathrm{x}}$, and the term $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ will be used throughout the paper.

Statistical analysis

Continuous data is presented as means with standard deviation, and categorical data as percentages. Baseline comparisons between HIT and MIT were performed separately for men and women using the independent sample t-test for continuous variables and Fisher’s exacts test for categorical variables. A linear mixed model was used to assess the difference of self-reported HIT vs. MIT on the 5-yr change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. This model accommodates repeated measurements, as it includes participants with at least one observation and handles missing data under a missing at random assumption, thereby retaining statistical power and reduces potential bias compared with complete-case analyses. Thus, no imputation was required. Participants may have been exposed to the intervention before participation, potentially affecting baseline values. Therefore, no baseline adjustments was applied in our models, in line with methodological recommendations that adjusting for baseline in analyses of change may introduce bias when baseline values already have been influenced by prebaseline exposure (28).

Change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ over 5 yr was analyzed with $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ as the dependent variable, with time, group (HIT vs MIT), and their interaction as fixed factors. Although the age range was restricted (70–77 yr), the models were adjusted for age because of the strong association between age and trajectories in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ (1). Models were also adjusted for cohabitation status (stratification variable in the main trial) and smoking as covariates. We also performed sensitivity analyses to assess the impact of covariate adjustments. Changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ are presented as estimated change in mL/kg/min, alongside the mean estimated between-group difference (group × time interaction) and 95% CI. Within-group changes at each follow-up were obtained directly from the mixed model as estimates of mean change from baseline at each measured time point. In an exploratory analysis, adjusted for the same covariates, a linear mixed model was conducted to examine the independent associations of exercise intensity (computed METs) and total minutes/week of exercise training with changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, using interaction terms (time × intensity and time × time/wk).

Statistical significance was defined as a two-sided P value <0.05. Normality of residuals was visually inspected by Q-Q plots, and all models were carried out separately for each sex. Analysis was conducted using the IBM SPSS Statistics for Windows, version 30.0 (IBM Corp., Armonk, NY).

Participant Involvement

Initial study concepts were presented to a large group of older adults in Trondheim, Norway, approximately 2 yr before study commencement. This was followed by focused discussions with participant representatives 6 months before obtaining ethical approval. The study was developed by the authors in collaboration with user representatives, and public stakeholders, including the Norwegian Directorate of Health. Additionally, a dedicated website was launched to provide weekly updates, and ongoing support was provided via a designated email/phone contact during working hours. Several of the participants have also actively participated in our regular meetings, where we disseminate research findings to the wider community.

Equity, Diversity, and Inclusion Statement

All inhabitants in Trondheim, Norway, aged between 70 and 77 yr were invited to participate in the Generation 100 Study, to ensure equal opportunity for inclusion. The population of central Norway, from which the Generation 100 participants were recruited, is predominantly Caucasian, limiting our ability to examine outcome differences across ethnic groups. However, we included individuals with a range of chronic conditions, which will allow for in depth analysis of outcome effects in these individuals in the future. We also included approximately equal numbers of men and women, allowing for sex-specific analyses of outcomes. Our research and author team consists of both men and women with complementary academic backgrounds.

RESULTS

Baseline Characteristics

Baseline characteristics of the 500 participants (261 women) are shown in Table 1. Complete $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ data at all four time points were available for 368 participants (130 in HIT and 238 in MIT). Participants mean (standard deviation) age at baseline was 72.2 (2.0) yr, with a $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ of 30.1 (6.2) mL/kg/min, and 87.4% reported good health. Men and women who consistently reported engaging in HIT for 5 yr, had on average, 11.0% (P < 0.001) and 7.0% (P = 0.01) higher $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ compared with counterparts in self-reported MIT at baseline, respectively.

TABLE 1.

Baseline characteristics for the selected participants reporting HIIT or MICT for 5 yr.

		HIT			MIT
	Men n = 100		Women n = 55	Men n = 139		Women n = 206
Age, yr	71.9 (1.8)		72.0 (1.9)	72.2 (2.0)		72.4 (2.0)
BMI, kg/m2	26.0 (2.7)		25.0 (3.6)	25.8 (2.8)		25.0 (3.4)
$\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, mL/kg/min	35.2 (6.4)*		28.8 (5.1)*	31.5 (5.5)		26.9 (4.5)
Cohabitant/married, %	86.0		67.2*	87.0		59.2
Self-reported good health, %	92.0*		83.6*	89.2		84.9
Current smoker, %	4.0*		1.8	7.1		6.7
History of cancer, %	14.0		12.7	17.3		13.1
History of CVD, %	22.0		14.5	14.4		7.8

Continuous data are presented as mean (SD) and categorical data as percentage.

^*Significantly different from individuals (within the same sex) (P < 0.05).

BMI, body mass index; CVD, cardiovascular disease; SD, standard deviation; V̇O_2peak, peak oxygen uptake.

Longitudinal Change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ over a 5-yr Period

After 1 yr, both men and women in self-reported HIT and MIT significantly improved $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. Men who reported HIT increased $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ with 2.27 mL/kg/min (6.4%) and 1.43 mL/kg/min (4.5%) in those with self-reported MIT (both P < 0.001; Fig. 2). For women reporting HIT, the increase in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ were 2.46 mL/kg/min (8.5%) and 1.65 mL/kg/min (6.1%) in MIT (both P < 0.001; Fig. 2). No significant differences in change of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ were observed between the groups for either sex after 1 yr (Table 2). After 3 yr, participants with self-reported HIT, maintained a higher $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ compared with their counterparts reporting MIT (Table 2), with a difference of 2.19 mL/kg/min (95% CI 1.06–3.37) in men and 1.51 mL/kg/min (95% CI 0.47–2.55) in women. After 5 yr, compared with baseline, $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ decreased in both groups. Among men who reported HIT, the reduction was −1.11 mL/kg/min (−3.1%, P = 0.007; Fig. 2) and −2.43 mL/kg/min (−7.7%, P < 0.001; Fig. 2) in MIT. Women reporting MIT had a decrease of −1.26 mL/kg/min (−4.6%, P < 0.001; Fig. 2) in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, while women with self-reported HIT showed no change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ after 5 yr (P = 0.96; Fig. 2). However, after 5 yr, in participants reporting HIT, there was an association toward a higher $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ compared with their counterparts reporting MIT (Table 2), with a difference of 1.31 mL/kg/min (95% CI 0.17–2.45) and 1.33 mL/kg/min (95% CI 0.28–2.37) for men and women, respectively. In sensitivity analyses comparing adjusted and unadjusted models, the inclusion of age, cohabitation status, and smoking did not materially influence the group × time estimates (Table 2). Indicators of effort during CPET were broadly similar between HIT and MIT across time points (Supplemental Table 2, Supplemental Digital Content, https://links.lww.com/MSS/D418).

FIGURE 2.

Change in absolute $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ for men (upper panel) and women (lower panel) from baseline. Data are presented as mean change, with standard error from the mixed model adjusted for age, smoking and cohabitation status. HIT, high intensity training; MIT, moderate intensity training; V̇O_2peak, peak oxygen uptake *Significantly different from MIT (P < 0.05).

TABLE 2.

Five-year changes in VO_2peak for men and women in HIT and MIT.

		HIT		MIT	Adjusted		Unadjusted
	n	Mean (SD)	n	Mean (SD)	Difference (Group × time)		Difference (Group × time)
					Estimate, mL/kg/min (95% CI)	p value	Estimate, mL/kg/min (95% CI)	p value
Men
Baseline	100	35.27 (6.45)	138	31.55 (5.52)
1 yr	95	37.69 (6.29)	132	33.26 (6.49)	0.84 (−0.27 to 1.96)	0.13	0.73 (−0.37 to 1.83)	0.19
3 yr	92	36.37 (6.99)	119	30.52 (6.25)	2.19 (1.06 to 3.37)	<0.001	2.18 (1.05 to 3.31)	<0.001
5 yr	90	34.22 (6.42)	117	29.57 (6.34)	1.31 (0.17 to 2.45)	0.02	1.35 (0.21 to 2.49)	0.02
Women
Baseline	55	28.80 (5.12)	202	26.96 (4.59)
1 yr	54	31.39 (6.25)	196	28.71 (4.95)	0.81 (−0.20 to 1.83)	0.11	0.77 (−0.22 to 1.78)	0.12
3 yr	52	29.86 (6.04)	171	26.73 (4.95)	1.51 (0.47 to 2.55)	0.004	1.43 (0.41 to 2.45)	0.006
5 yr	51	29.15 (5.50)	166	26.12 (4.81)	1.33 (0.28 to 2.37)	0.01	1.19 (0.16 to 2.22)	0.02

Results from the linear mixed model. The primary model was adjusted for age, cohabitation and smoking, and displays the estimated effect as group × time interaction with 95% CIs for HIT compared with MIT after 1, 3, and 5 yr. The unadjusted model included no covariates and is presented as a sensitivity analysis. Mean (SD) for both HIT and MIT are descriptive data.

HIT, self-reported high intensity training; MIT, self-reported moderate intensity training; V̇O_2peak, peak oxygen uptake.

Exercise Training Characteristics

Table 3 shows the self-reported exercise training characteristics. No differences in exercise training frequency or duration were observed when comparing those reporting HIT to those reporting MIT within each sex at baseline. After 1 yr, men reporting HIT had higher time/week than their counterparts in self-reported MIT (P = 0.007), while no differences were observed in women. After 3 yr, men reporting MIT had higher exercise training frequency compared with those with self-reported HIT (P = 0.03). After 5 yr, no differences were observed in time/week or frequency for either gender. During supervised sessions, participants exercised at an average intensity of 91% of peak heart rate in HIT and 75% in MIT, with corresponding Borg ratings of perceived exertion of 16.9 and 13.9.

TABLE 3.

Exercise characteristics of participants who reported HIT or MIT regardless of randomization for 5 yr.

		HIT			MIT
	Men		Women	Men		Women
Baseline
Week/min	141.1 (74.7)		148.1 (69.8)	140.7 (75.2)		139.6 (73.0)
d/wk	2.7 (1.3)		2.9 (1.2)	3.0 (1.4)		2.9 (1.3)
MET/intensity	5.5 (0.8)*		5.0 (0.9)*	4.3 (0.8)		4.4 (0.8)
1 yr
Week/min	175.4 (72.3)*		158.7 (73.9)	151.1 (65.4)		155.4 (72.9)
d/wk	3.2 (1.2)		3.0 (1.1)	3.2 (1.2)		3.3 (1.3)
MET (intensity)	5.9 (0.5)*		5.9 (0.5)*	4.3 (0.7)		4.3 (0.6)
3 yr
Week/min	165.7 (74.9)		156.4 (67.3)	161.0 (73.4)		153.6 (68.9)
d/wk	3.1 (1.2)		3.1 (1.2)	3.5 (1.3)†		3.2 (1.2)
MET (intensity)	5.9 (0.5)*		5.9 (0.9)*	4.3 (0.6)		4.4 (0.6)
5 yr
Week/min	165.9 (73.1)		165.5 (71.7)	158.8 (75.9)		149.0 (70.7)
d/wk	3.1 (1.2)		3.3 (1.2)	3.4 (1.3)		3.1 (1.3)
MET (intensity)	5.9 (0.5)*		5.8 (0.5)*	4.3 (0.7)		4.2 (0.7)

Data are presented as mean (SD) for each group and gender.

^*Significant different from counterparts in MIT (P < 0.05).

^†Significant different from counterparts in HIT (P < 0.05).

HIT, self-reported high intensity training; MET, metabolic equivalent; MIT, self-reported moderate intensity training.

Association of Exercise Training Characteristics on 5-yr Change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$

The interaction time × time/week, as well as time × intensity in relation to changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ among men and women are shown in Table 4. In both genders, higher self-reported exercise intensity as reported with MET was significantly associated with a reduced decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ over the follow-up period. For men, the estimated interaction effect increased from 0.77 mL/kg/min (95% CI 0.25–1.30) at 1 yr to 1.11 mL/kg/min (95% CI 0.60–1.62) after 5 yr. A similar pattern was observed in women, as the interaction increased from 0.59 mL/kg/min (95% CI 0.17–1.02) after 1 yr to 0.64 mL/kg/min (95% CI 0.22–1.06) after 5 yr. In contrast, the interaction between time and time/week showed no significant association in men at any timepoint. Among women, there were small but statistically significant effects after 1 and 3 yr, which diminished after 5 yr.

TABLE 4.

Effects of exercise time/week and intensity on $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ change for men and women.

		Interaction (time*week/min)		Interaction (time*MET)
	n	Estimate, mL/kg/min (95%CI)	p value	Estimate, mL/kg/min (95%CI)	p value
Men
Baseline	221
1 yr	239	0.000 (−0.006 to 0.007)	0.890	0.77 (0.25 to 1.30)	0.004
3 yr	239	0.003 (−0.004 to 0.010)	0.432	1.26 (0.70 to 1.81)	<0.001
5 yr	239	0.001 (−0.006 to 0.007)	0.868	1.11 (0.60 to 1.62)	<0.001
Women
Baseline	241
1 yr	261	0.006 (0.001 to 0.010)	0.024	0.59 (0.17 to 1.02)	0.006
3 yr	261	0.007 (0.002 to 0.013)	0.007	0.88 (0.43 to 1.33)	<0.001
5 yr	261	0.005 (−0.001 to 0.010)	0.084	0.64 (0.22 to 1.06)	0.003

Results from the linear mixed model adjusted for age, cohabitation, and baseline smoking. Displayed as estimated effects of time*weekly minutes and intensity of exercise with 95% CIs, for men and women separately.

CI, confidence interval; MET, metabolic equivalent of task.

DISCUSSION

In this study of older adults aged 70–77 yr, we found associations suggesting that older adults who consistently reported either HIT or MIT for 5 yr appeared to attenuate the expected age-related decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. Notably, HIT was associated with a smaller decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ when compared with MIT, for both sexes, and higher self-reported exercise intensity in METs showed a stronger association with preserved $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ than total time/week, with the association for exercise intensity appearing to strengthen over time. Together, these associations suggest that exercise intensity may play an important role in long-term preservation of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ in older adults.

Longitudinal Change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$

The observation that individuals reporting both HIT and MIT improved $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ after 1 yr align with previous literature, showing that both exercise modalities improve $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ in older adults (29). Contrary to the main Generation 100 Study (22), no between-group differences in terms of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ were observed here after 1 yr. This discrepancy may be explained by differences in analytic approach and sample definition. The main Generation 100 Study compared randomized groups with the intention to treat approach, whereas the present analysis reclassified participants based on consistent self-reported exercise intensity, thereby selecting a subgroup of participants that were more likely to be healthier. The highly selected nature of the present study sample is also a reflected by the fact that participants showed baseline $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ values comparable to or even exceeding those in the Generation 100 reference data (30). Nevertheless, our findings confirms that $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ can be improved even in older adults beyond the age of 70 yr (16).

The $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectory appeared to be nonlinear, with an initial improvement after 1 yr followed by a decline between 1 and 3 yr. This pattern likely reflects an early exercise response combined with the expected age-related decline in aerobic capacity, potentially also compounded by challenges in sustaining a sufficient training stimulus over time. Importantly, despite this decline at 1 yr, participants reporting HIT maintained a significantly higher $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ than those reporting MIT at 3 and 5 yr, suggesting that higher exercise intensity may be associated with a more favorable long-term $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectory. Over 5 yr, the decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ observed in both HIT and MIT appeared to be smaller than expected based on prior longitudinal studies in general populations with a variable amount of PA participation (31,32). The Baltimore Longitudinal Study of Aging (BLSA) (31) studied 810 healthy volunteers (aged 21–87 yr, 46% women) with a mean follow-up time of 7.9 yr and reported an accelerating decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ with increasing age_, exceeding 20% per decade in 70-yr-old men and women. Also, the HUNT Study of 4404 participants conducted in the same demographic area as the current reported a 20% decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ among 70-yr-olds over a decade (32). On this basis, a 10% decline would be expected over a 5-yr period. In the present study, the descriptive declines were generally lower. While this may reflect true differences in long-term exercise behavior, caution is warranted given the observational design, reliance on self-reported intensity, and the highly selected nature of this cohort, which contrasts with the broader and more heterogenous populations in BLSA and HUNT. Nevertheless, the observed decline among men in MIT was more than twice as large to what observed among men reporting HIT (−7.7% vs −3.1%), while no decline was observed in women reporting HIT, which is noteworthy but observational. Taken together, these patterns suggest that long-term exercise training at higher intensities may induce a more favorable $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectory in older adults.

The $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ levels observed in this study align with previous reports in Norwegian adults aged >70 yr (33), supporting generalizability of our findings to healthy older populations. Notably, comparing our sample of 70- to 77-yr-old men and women to the largest reference materials on $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ across different ages (FRIEND), the $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ levels in our sample appear comparable to those reported in 40- to 50-yr-old men and women in the United States (34). As previously emphasized by Peterman et al. (35), this underscores the need to further develop global reference standards, especially for older adults, where such data remain limited.

Contribution of Exercise Intensity and Duration on Long-term Changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$

Building on the observation of a smaller decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ in HIT, we explored how exercise intensity and exercise time/week were independently associated with change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. Our exploratory interaction analysis showed that each 1 MET increase in exercise intensity was associated with a 1.11 mL/kg/min smaller decline in men and 0.64 mL/kg/min smaller decline in women after over 5 yr. This association was stronger after 3 and 5 yr compared with 1 yr, which may suggest a cumulative contribution of higher exercise intensities. However, these analyses relied on the Borg-derived MET estimates, which may not fully capture true physiological workload in older adults and could introduce misclassification, thereby limiting the precision. Still, the consistency and increasing magnitude of the association across timepoints support exercise intensity as a potentially important determinant of long-term $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ preservation. In contrast, time/week showed weaker and less consistent association. For instance, among men, an additional 100 min/week was associated with just a 0.1 mL/kg/min better preservation of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ after 5 yr_, a nonsignificant and clinically negligible difference. The stronger association of exercise intensity with changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ align with the observations in the Norwegian HUNT study showing that both exercise volume and intensity was associated with smaller age-related decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$_, with the association being notably stronger for intensity (32). In contrast, the BLSA study found no attenuation of $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ decline with higher levels of PA (31). Importantly, the BLSA study did not account for exercise intensity. While current guidelines recommend a minimum of 150 min/wk of MIT or up to 150 min/wk of HIT as the upper bound (13), our findings suggest that exercising for this amount of minutes/week at higher intensities (≥90% of peak heart rate) may be more closely associated with smaller age-related decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. These observations may help inform future discussions on PA recommendations for older adults, although confirmatory studies are needed.

Clinical Implications

Given that high age-specific $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ appears to have major effects on public health and health related economics (11), our observations hold potential clinical relevance. Projections from the United Kingdom estimate a 36% increase in older adults (>65 yr) requiring high care needs by 2035 (36), underscoring the importance of strategies that may preserve functional capacity. For instance, the risk of dependency has been shown to increase by 14% for each 1 mL/kg/min lower $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, with approximately 18 mL/kg/min in men and 15 mL/kg/min in women representing critical thresholds (6,37). Moreover, using data from the Generation 100 Study, our group recently demonstrated that $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ levels below 26.5 mL/kg/min in men and 22.2 mL/kg/min in women were significantly associated with increased risk of all-cause mortality in older adults (38). In this context, the more favorable $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ trajectories observed among individuals who maintain higher exercise intensity may be meaningful from a population health perspective. Results from the present study may help to generate hypotheses for future work aimed at evaluating the role of exercise intensity in preserving $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ and functional capacity in older adults. Future research should also explore how such strategies can be implemented safely and effectively in broader aging populations. It is also important to clarify potential sex-specific barriers in long-term adherence, as we observed notably fewer women maintaining self-reported HIT in the present cohort.

Strength and Limitations

Several limitations of this study should be considered. By reclassifying participants according to self-reported exercise intensity rather than adhering to the original randomization, we deviated from the intention to treat principle. By this, selection bias should be considered, and residual confounding cannot be excluded. Therefore, the results should be interpreted cautiously. In addition, a stable nonexercising group was not available within our exposure definition, as very few participants reported no exercise across follow-ups. Accordingly, our findings should be interpreted as intensity-specific associations among participants reporting regular exercise training, rather than as exercise versus no exercise. However, the rationale for basing the analysis on self-reported exercise habits was grounded in the observed Hawthorne effect in the main Generation 100 Study, where a substantial proportion of the control group had high PA levels, and 50% of them were doing HIIT during the study period, making it difficult to draw robust conclusions about the true effect of exercise intensity on changes in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ (22). Thus, the exercise questionnaires were considered an appropriate proxy evaluating the association of consistent self-reported HIT or MIT on the longitudinal change in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$. We acknowledge that generalizability may be limited, as participants in this study were relatively healthy and may not reflect the general older population. This is further supported by the baseline differences in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, which suggest selection bias, as individuals with higher initial fitness may be more likely to choose or sustain higher exercise intensity, which potentially introduces confounding and reverse causality. Therefore, caution is warranted when extrapolating these findings to other demographic groups or clinical populations.

A major strength of this study is the longitudinal design, with repeated measures of the same individuals over time. Also, the use of objectively measured $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ from four timepoints in both sexes represents another major strength, as prior studies often have relied on estimated CRF or fewer assessments over longer intervals. Finally, presenting sex-specific results adds further value.

CONCLUSIONS

Consistent engagement in aerobic exercise training, particularly with HIT, was associated with a smaller age-related decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ over a 5-yr period in older adults. Our findings further suggest that higher exercise intensity may be more effective than time/week in mitigating the age-related decline in $\dot{\mathrm{V}}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$, an association that appeared to strengthen over time.

This work was funded by the Research Council of Norway; the K.G. Jebsen Foundation for Medical Research; Norwegian University of Science and Technology; Central Norway Regional Health Authority; St. Olavs University Hospital, Trondheim, Norway; and the National Association for Public Health, Norway. The funding organizations were not involved in the study design, data collection, analysis, and interpretation of the data, or in the preparation, and writing of the manuscript. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsements by the American College of Sports Medicine. The authors declare no competing interest, or any other relationships that could appear to have influenced the submitted work. S.M. designed the study, analyzed and interpreted data, and drafted the manuscript. S.L. provided statistical guidance, data interpretation, and critical review of the manuscript. L.K. contributed to supervision, literature search, data interpretation, and critical review of the manuscript. U.W. contributed to study design, interpreted data, supervision, literature search, and critical reviewed the manuscript. D.S. contributed to study design, supervision, literature search, interpreted data, and drafting of the manuscript. All authors have approved the final draft to be published. D.S. is the guarantor of the paper. We sincerely thank the participants of the Generation 100 Study for their contributions. All participants gave written consent before participation, and the study was approved by the Regional Committee for Medical Research Ethics (REK 2012/381 B). Individual data underlying the current trial cannot be shared, due to lack of participant consent for data sharing at the time of enrollment. However, we welcome scientific collaboration worldwide, that can have access to analyzed data from our university. A biobank containing blood and genetic material has been established as part of the project. While these biological samples can be shared with collaborators worldwide, any analysis must be conducted within our university.

Supplementary Material

msse-58-2020-s001.pdf ^{(289KB, pdf)}