Resistance training and inter-interindividual response differences on cardiorespiratory fitness in older adults: An ancillary meta-analysis of randomized controlled trials

Abstract

Examine true inter-individual response differences (IIRD) as a result of resistance training on cardiorespiratory fitness in older adults. Data from a recent meta-analysis of 22 randomized controlled trials representing 552 men and women (292 resistance training, 260 control) ≥ 60 years of age were included. The primary outcome was cardiorespiratory fitness (VO_2max) in ml^.kg^−1.min⁻¹. Using the inverse variance heterogeneity (IVhet) model, statistically significant treatment effect (resistance training minus control) increases in VO_2max in ml^.kg^−1.min⁻¹ were found (mean, 1.8, 95% CI, 0.4 to 3.3 ml^.kg^−1.min⁻¹, p = 0.01; Q = 82.8, p < 0.001; I²= 74.6%, 95% CI, 61.6 to 83.3%; $τ^{2}$ =1.1). The 95% prediction interval (PI) was −0.8 to 4.5 ml^.kg^−1.min⁻¹. However, no statistically significant IIRD was observed (mean, 0.6, 95% CI, −1.1 to 1.4 ml^.kg^−1.min⁻¹; $τ^{2}$ =1.5). The 95% PI was −1.8 to 2.0 ml^.kg^−1.min⁻¹. In conclusion, while progressive resistance training may increase VO_2max in ml^.kg^−1.min⁻¹, a lack of true resistance-training-associated IIRD exist.

Keywords

exercise resistance training meta-analysis cardiorespiratory fitness older adults

Introduction

Globally, the number of adults ≥ 60 years has been increasing at a significant rate and is expected to continue to increase in the future. For example, worldwide estimates project that the number of adults 60 years of age and older will increase from 962 million in 2017 to 2.1 billion in 2050.¹ Not surprisingly, the burden of health conditions in older adults has also increased. For example, it has been estimated that 23% of the worldwide burden of disease is the result of various disorders in adults 60 years of age and older, the leading causes being 1) cardiovascular diseases, 2) malignant neoplasms, 3) chronic respiratory diseases, 4) musculoskeletal diseases, and 5) neurological and mental disorders.² The healthcare costs for older adults are also substantial. In the United States, Levy et al. estimated that the one-year healthcare costs for eight different health conditions in adults 60 years of age and older were $63 billion.³

Given the increasing number of adults 60 years of age and older as well as the costs associated with such, it has been recommended that a need exists to reduce the burden of disease in older adults.² Higher cardiorespiratory fitness has been associated with a lower risk of cardiovascular as well as coronary heart disease, several cancers, chronic respiratory diseases, musculoskeletal diseases, neurological diseases, and common mental health disorders.⁴ In addition, higher cardiorespiratory fitness has been associated with a reduced risk of all-cause mortality.^5–9 While aerobic exercise training has traditionally been recommended for increasing cardiorespiratory fitness,⁴ the role of resistance training for increasing cardiorespiratory fitness has been more equivocal. To address such, Smart et al. recently conducted a systematic review with meta-analysis of randomized controlled trials to examine the effects of resistance training on cardiorespiratory fitness in adults ≥ 60 years of age.¹⁰ Of the 22 studies representing 552 men and women (292 exercise, 260 control), statistically significant improvements of 1.9 ml^.kg^.min⁻¹ (95% CI, 1.2 to 2.6 ml^.kg^.min⁻¹) were reported for maximum oxygen consumption (VO_2max in ml^.kg^−1.min⁻¹).¹⁰

Precision exercise prescription, i.e., the tailoring of exercise programs that take into account factors such as individual variability in one's genes, is considered to be one of the most important issues in exercise medicine.¹¹ However, the a priori assumption that true inter-individual response differences (IIRD) actually exist may not be tenable. For example, while heritability estimates ranging from 25% to 65% have been reported for cardiorespiratory fitness responses to aerobic exercise training,¹² a review that included six exercise intervention studies published prior to the NIH funded HEalth, RIsk factors, exercise Training And GEnetics (HERITAGE) Family Study as well as 180 published studies from the HERITAGE Family Study found that none had appropriately quantified IIRD, with only one non-HERITAGE study including a comparator arm, i.e. control group.¹³ Re-analysis of the one intervention study that included a control arm and appropriately quantified true IIRD found that the standard deviation of change in cardiorespiratory fitness, assessed as VO_2max in ml^.kg^−1.min⁻¹, was greater in the control arm (±5.6 ml^.kg^−1.min⁻¹) versus the aerobic exercise arm (±3.7 ml^.kg^−1.min⁻¹).¹³ Similarly, a recent individual participant data meta-analysis of eight randomized controlled trials representing 1879 adults found a lack of IIRD on cardiorespiratory fitness as a result of exercise training.¹⁴ Collectively, these findings suggest that factors other than IIRD to exercise training (random variation, physiological responses associated with behavioral changes that are not the result of an intervention such as exercise) are responsible for any observed variation.¹⁵ In addition, the lack of exercise-associated IIRD suggests that a search for potential moderators and mediators associated with exercise and cardiorespiratory fitness may not be necessary.¹⁶ From a practical perspective, these findings suggest that general versus specific exercise guidelines may be preferred with respect to the effects of exercise training on cardiorespiratory fitness.

While the above findings are noteworthy, the investigative team is not aware of any previous research that has examined the IIRD of resistance training on cardiorespiratory fitness in older adults. From the authors’ perspective, an examination of such is important given the need to determine whether a search for potential moderators and mediators, including genetic interactions is worth pursuing as well as whether general versus specific exercise guidelines should be pursued. Given the former, the purpose of this study was to use the meta-analytic approach to examine whether resistance training-associated IIRD exists with respect to VO_2max in ml^.kg^−1.min⁻¹ in adults 60 years of age and older.

Methods

Registration

The protocol for this study was registered in Open Science Framework at https://osf.io/f2hzx.

Data source

Data for the current IIRD meta-analysis were derived from a previously published meta-analysis¹⁰ of progressive resistance training that did not examine for IIRD and which included the assessment of VO_2max in ml^.kg^−1.min⁻¹ from 22 randomized controlled trials representing 552 men and women (292 resistance training, 260 control) ≥ 60 years of age.^17–38 The reason for using this recent meta-analysis versus conducting an updated meta-analysis was based on 1) the recency of this prior work (2022),¹⁰ 2) decision tree analysis developed by others on when to update or conduct a new systematic review with meta-analysis,³⁹ and 3) a recent recommendation from others that previously published meta-analyses be used to conduct IIRD meta-analyses.⁴⁰ The original search took place from database inception up to January 21, 2022.

Details of the original meta-analysis and included studies are described in the original article.¹⁰ The characteristics of the included studies are described here in the Methods versus the Results section of the final manuscript because they were derived from the previous systematic review with meta-analysis.¹⁰ Briefly, nine studies were conducted in the United States,^{17,24–28,32,37,38} six in Brazil,^19–23,30 two in Denmark,^18,35 and one each in either Austria,³⁶ Australia,³¹ Finland,²⁹ Japan,³³ or Spain.³⁴ For those studies in which data were available,^{17–19,21–28,30–38} mean ± standard deviation (SD) ages were 68.1 ± 3.1 years in the resistance training groups (median = 67.7) and 67.5 ± 3.9 years in the control groups (median = 67.4). The percentage of men for those studies that provided such ranged from 0% to 100% in both the resistance training ( $\bar{X}$ ± SD = 23.9% ± 40.8%, median = 50%) and control $(\bar{X}$ ± SD = 23.9% ± 41.9%, median = 40%) groups.^{17–23,25–33,35–38} Baseline VO_2max, assessed using primarily maximal treadmill and cycle ergometer tests, ranged from 16.6 to 35.2 ml^.kg^.min⁻¹ in the resistance training groups ( $\bar{X}$ ± SD = 24.3 ± 5.6 ml^.kg^.min⁻¹, median = 21.8) and 16.0 to 41.7 ml^.kg^.min⁻¹ ( $\bar{X}$ ± SD = 24.8 ± 6.9 ml^.kg^.min⁻¹, median = 22.6) in the control groups.^{17–26,28–38}

Resistance training interventions ranged from 8 to 52 weeks ( $\bar{X}$ ± SD = 20.0 ± 9.0, median = 17), 2 to 5 days per week, ( $\bar{X}$ ± SD = 3.0 ± 1.0, median = 3).^17–38 The number of reported sets, repetitions, and exercises ranged from 1 to 8 ( $\bar{X}$ ± SD = 3.0 ± 1.0, median = 3.0),^17–36,38 6 to 20 ( $\bar{X}$ ± SD = 10.3 ± 1.8, median = 10),^{17–20,22–24,26–36,38} and 3 to 13 ( $\bar{X}$ ± SD = 8.0 ± 3.0, median = 8).^{17–25,27,30,32–38} For the 15 studies that reported such, intensity, based on 1 repetition maximum (RM), ranged from 45% to 90% ( $\bar{X}$ ± SD = 72.0% ± 9.0%, median = 70).^{17,19,20,23,25–27,29–33,36–38} All reported exercise sessions were supervised at least 75% of the time.^{18–22,24–38} The comparison groups consisted primarily of no exercise training,^{17–20,23–38} sham water-based training,²¹ or board games.²² Using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) instrument,^41,42 the previous meta-analysis considered the certainty of evidence to be “Moderate” to “High” for changes in VO_2max in ml^.kg^−1.min⁻¹.¹⁰

Data abstraction

Data were abstracted from the included meta-analysis by the first and second authors, independent of each other, using Microsoft® Excel® for Microsoft 365 MSO (Version 2203 Build 16.0.15028.20242). These included: 1) original study author names, 2) year of publication, and 3) sample sizes as well as change outcome means and standard deviations for VO_2max ml^.kg^−1.min⁻¹ for both resistance training and control groups. Upon completion of data abstraction, Gwet's AC1 statistic was used to assess inter-rater agreement.^43,44 The first two authors then met and reviewed their data abstraction for agreement. Any disagreements were resolved by consensus. If agreement could not be reached, the third author provided a recommendation.

Quality of original meta-analysis

The Assessment of Multiple Systematic Reviews (AMSTAR 2)⁴⁵ instrument was used to assess the quality of the included systematic review with meta-analysis.¹⁰ The overall confidence in the quality of the meta-analysis was considered as either “High”, “Moderate”, “Low” or “Critically low” based on previous recommendations.⁴⁵ AMSTAR2 assessments were conducted using the same procedures as for data extraction.

Data synthesis

Effect size metric

The original metric was used for the primary outcome, i.e. changes in VO_2max in ml^.kg^−1.min⁻¹.

Traditional treatment effects meta-analysis

Preceding the IIRD meta-analysis, an aggregate data meta-analysis was conducted by pooling treatment effects for each outcome from each study. These were calculated as the change outcome difference between the resistance training and control groups along with the change outcome standard deviations for each group. Results were then pooled using the inverse variance heterogeneity (IVhet) model.^46,47 The IVhet model has been shown to be both more robust and plausible than random-effects models.^46,47 Ninety-five percent confidence intervals (CIs) that did not include zero (0) were considered statistically significant.

Statistical heterogeneity and inconsistency were assessed using the Cochran Q statistic and I-squared (I²) statistic.⁴⁸ For Q, alpha (p) values ≤0.10 were considered statistically significant while inconsistency (I²) was categorized as very low (<25%), low (25% to <50%), moderate (50% to <75%), or high (>75%).⁴⁸ The I² statistic is an indicator of the percentage of variance from true effects versus sampling error but not an indicator of how much an effect size like changes in VO_2max in ml^.kg^−1.min⁻¹ varies.⁴⁹ Thus, tau-squared ( $τ^{2}$ ), an absolute measure of between-study heterogeneity, was also calculated. Small-study effects (publication bias, etc.) were assessed qualitatively using the Doi plot and quantitively using the Luis Furuya-Kanamori (LFK) index.⁵⁰ The Doi plot has been reported to be more intuitive than the traditional funnel plot while the LFK index has been shown to be more robust than the usually used Egger's regression intercept test.⁵⁰ Previously suggested LFK cutpoints of ±1, between ±1 and ±2, and >±2, were used to denote no, minor, and major asymmetry.⁵⁰ Sensitivity analyses for changes in VO_2max in ml^.kg^−1.min⁻¹ were also conducted by deleting each study from the model once to see how it influenced the overall results, including heterogeneity and inconsistency. Outlier analysis was conducted by excluding study-level effect sizes in which their 95% confidence intervals (CIs) fell outside the pooled 95% CI.

Ninety-five percent prediction intervals (PIs) were calculated for changes in VO_2max in ml^.kg^−1.min⁻¹. Ninety five percent PIs are used to determine what result one might expect if a new randomized controlled trial was conducted in a population similar to those included in the meta-analysis.⁵¹ Thus, PIs indicate how much changes in VO_2max in ml^.kg^−1.min⁻¹ vary using the same metric.⁴⁹ The clinical importance for mean treatment effect changes (resistance training minus control) for VO_2max in ml^.kg^−1.min⁻¹ was based on a minimally clinically important difference (MCID) of 1.0 ml^.kg^−1.min⁻¹, derived from previous research showing a relative risk decrease of 9% in all-cause mortality.⁵² Based on previous suggestions, the following probabilistic anchors were used in interpreting the clinical importance of results: < 0.5% (most unlikely or almost certainly not), 0.5% to 5% (very unlikely), 5% to 25% (unlikely or probably not), 25% to 75% (possibly), 75% to 95% (likely or probably), 95% to 99.5% (very likely), > 99.5% (most likely or almost certainly).⁵³

IIRD meta-analysis

Based on previous recommendations,^54,55 the standard deviation of individual response approach (SD_ir), an absolute difference method, was chosen over others for the primary aim of this study, i.e. an IIRD meta-analysis for changes in VO_2max in ml^.kg^−1.min⁻¹. This was calculated as the difference between resistance training and control group standard deviations for changes in VO_2max in ml^.kg^−1.min⁻¹ from each study and treating them as point estimates as follows⁵⁴:

\sqrt{S D_{r t}^{2} - S D_{c}^{2}}

where

{SD}_{r t}^{2}

is the standard deviation for the resistance training group and

S D_{c}^{2}

is the standard deviation for the control group. The standard error of the variance for the point estimates was then calculated as follows⁵⁴:

S E = \sqrt{2 (\frac{S D_{r t}^{4}}{D F_{r t}} + \frac{S D_{c}^{4}}{D F_{c}})}

where DF_rt and DF_c represent the degrees of freedom minus 1 for the standard deviations in the resistance training and control groups. Individual response variances and their standard errors were then pooled into one overall point estimate and 95% CI using the IVhet model.⁴⁶ The SD for point estimates and 95% CIs were then calculated by taking the square root of each.⁵⁶ If the lower 95% CI was negative, the sign was first ignored, the square root computed, and the sign reapplied. Tau-squared (

τ^{2}

) was also calculated. Similar to the treatment effects meta-analysis, 95% PI were calculated for changes in VO_2max in ml^.kg^−1.min⁻¹ as well as the MCID and suggested probabilistic categories associated with clinical importance.

Software used for analysis

The following software was used for all statistical analyses: 1) Meta XL (version 5.3),⁵⁷ 2) Metan for Stata (version 16),⁵⁸ 3) KAPPAETC for Stata,⁴⁴ and 4) JMP (version 16.2).⁵⁹ All statistical tests were two-tailed.

Post hoc modifications

There were no post hoc modifications.

Results

Inter-coder agreement for data abstraction

Based on Gwet's AC1 statistic, the overall agreement rate for data abstraction prior to correcting differences was 0.96 (95% CI, 0.93, 0.98).

AMSTAR 2 results

Evaluation of the included meta-analysis¹⁰ using AMSTAR 2 is shown in Supplementary File 1. The overall agreement rate prior to correcting disagreements was 0.34 (95% CI, −0.23, 0.91). More than half of the items (56.3%) were rated positively (yes/partial yes). The overall quality of the meta-analysis was considered to be “Moderate”.

Treatment effect results

Treatment effect results for changes in VO_2max in ml^.kg^−1.min⁻¹ are shown in Table 1 and Figure 1. As can be seen, statistically significant increases were observed as a result of resistance training (p = <0.001). These findings suggest that we are 95% confident that the mean across all included studies is between 0.4 and 3.3 ml^.kg^−1.min⁻¹. Compared to the overall baseline value, the overall increase was equivalent to 7.8% (95% CI, 1.7% to 13.9%). Statistically significant heterogeneity was observed while overall inconsistency based on I² was considered high. The 95% CI for I² (61.6% to 83.3%) was considered to reflect moderate to high inconsistency. Tau-squared ( $τ^{2}$ ) was 1.07. Major asymmetry suggestive of small study effects (publication bias, etc.) was observed (Supplementary File 2). One outlier which reported the highest increase in VO_2max in ml^.kg^−1.min⁻¹ ( $\bar{X} =$ 4.5, 95% CI, 3.4 to 5.8 ml^.kg^−1.min⁻¹) was observed.²⁰ As can be seen in Supplementary File 3, when that study was deleted from the model, results remained statistically significant ( $\bar{X} =$ 1.8, 95% CI, 0.5 to 3.0 ml^.kg^−1.min⁻¹) with statistically significant heterogeneity (Q = 59.0, p < 0.001) and overall moderate inconsistency (I² = 66.1%, 95% CI, 46.4% to 78.6%) . When the one study that comprised the highest weight (70.2%) was deleted from the model (See Supplementary File 3),²⁹ results also remained statistically significant ( $\bar{X} =$ 1.5, 95% CI, 0.4 to 2.5 ml^.kg^−1.min⁻¹) with statistically significant heterogeneity (Q = 76.5, p < 0.001) and overall moderate inconsistency (I² = 73.9%, 95% CI, 59.8% to 83.0%). When all studies were deleted from the model one time (see Supplementary File 3), increases remained statistically significant across all deletions, ranging from a low of 1.5 ml^.kg^−1.min⁻¹ (95% CI, 0.4 to 2.5 ml^.kg^−1.min⁻¹; Q = 76.5, p < 0.001; I²= 73.9%, 95% CI, 59.8% to 83.0%) to a high of 2.0 ml^.kg^−1.min⁻¹ (95% CI, 0.8 to 3.2 ml^.kg^−1.min⁻¹; Q = 76.5, p < 0.001; I²= 59.0%, 95% CI, 33.6% to 74.6%). The 95% PI for what result one might expect if a new study was conducted in a similar population was −0.8 to 4.5 ml^.kg^−1.min⁻¹. The percent chance, i.e. probability, of a clinically meaningful improvement of at least 1.0 in VO_2max in ml^.kg^−1.min⁻¹ was 74.4%, “possibly clinically important”, and only 0.6% from being considered “likely or probably clinically important”.

Figure 1.

Forest plot for treatment effect changes in VO_2max in ml^.kg^−1.min⁻¹, ordered from smallest to largest reductions.

Table 1.

Treatment effect and IIRD results.

Variable	Studies (#)	Participants (#)	$\bar{X}$ (95% CI)^c	95% PI^d
TE^a
- VO_2max ml^.kg^−1.min⁻¹	22	552	1.8 (0.4, 3.3)*	−0.8, 4.5*
SD_IR^b
- VO_2max ml^.kg^−1.min⁻¹	22	552	0.6 (−1.1, 1.4)	−1.8, 2.4

Notes: ^aTE, treatment effects; ^bSD_IR, standard deviation of individual response differences; ^c $\bar{X}$ (95% CI), mean and 95% confidence interval; ^d95% PI, 95% prediction interval; *, nonoverlapping 95% confidence intervals.

IIRD results

The results for the primary aim of the current study, i.e. IIRD, are shown in Table 1. As can be seen, the 95% CI included zero, suggesting a lack of statistically significant exercise-associated IIRD in VO_2max in ml^.kg^−1.min⁻¹. The 95% PI also included zero (0). Absolute between-study heterogeneity ( $τ^{2}$ ) was 1.5. The probability of clinically meaningful IIRD for changes in VO_2max in ml^.kg^−1.min⁻¹ as a result of resistance training was 51.2%, considered to be “possibly clinically important”. There were no outliers. With each study deleted once from the overall pooled model, results continued to lack statistical significance, ranging from a low of 1.1 ml^.kg^−1.min⁻¹ (95% CI, −1.3 to 1.3 ml^.kg^−1.min⁻¹) to a high of 1.1 ml^.kg^−1.min⁻¹ (95% CI, −0.6 to 1.6 ml^.kg^−1.min⁻¹)

Discussion

Overall findings

The primary aim of this study was to examine for IIRD in VO_2max in ml^.kg^−1.min⁻¹ as a result of resistance training in adults 60 years of age and older. The findings suggest a lack of resistance-training-associated IIRD on VO_2max in ml^.kg^−1.min⁻¹. These findings are supported by 1) the overlapping 95% confidence intervals, 2) overlapping 95% prediction intervals, 3) lack of outliers, and 4) the finding that results were only “possibly clinically important”. Given these findings, any IIRD is probably the result of random variation (measurement error, daily biological variation) and/or physiologic responses that are not the result of resistance training (diet, sleep, etc.).¹⁵ Thus, an examination for potential moderators and mediators, including genetic interactions, associated with resistance training on VO_2max in ml^.kg^−1.min⁻¹ in older adults may not be warranted.¹⁶

The lack of IIRD observed in the current study is generally consistent with previous meta-analytic research on 1) aerobic exercise and body mass index (BMI in kg^.m²),⁶⁰ fat mass and percent body fat in children and adolescents with overweight and obesity,⁶¹ 2) tai chi,⁶² qigong,⁶³ walking,⁶⁴ and isometric exercise⁶⁵ on resting systolic and diastolic blood pressure in adults, 3) resistance training on muscular strength and hypertrophy,⁴⁰ 4) exercise and changes in body weight among adults,⁵⁶ and 5) exercise and cardiorespiratory fitness, waist circumference and body weight in humans.¹⁴

While not the primary purpose of the study, there was a statistically significant treatment effect increase in VO_2max in ml^.kg^−1.min⁻¹ as a result of resistance training in older adults. These results are reinforced by 1) the non-overlapping 95% CIs, 2) consistently non-overlapping 95% CIs when outliers as well each study was deleted from the model,²⁹ and 3) the finding that changes in VO_2max in ml^.kg^−1.min⁻¹ were close (0.6% away) to being considered “likely or probably clinically important” versus “possibly clinically important”. Alternatively, these findings could be questioned given the 1) moderate to high inconsistency based on the 95% CIs for I², 2) major asymmetry suggestive of small-study effects, and 3) overlapping 95% PIs. The results for the 95% PIs suggest that changes in VO_2max in ml^.kg^−1.min⁻¹ as a result of resistance training may increase in some older adults but decrease or have no effect in others. However, the majority of the interval fell on the side of benefit.

Implications for research

Based on the findings of this study, there are several implications for research with respect to resistance-training and VO_2max in ml^.kg^−1.min⁻¹ in older adults. First, the current findings, as previously mentioned, suggest that there is no need to examine for potential moderators, mediators and genetic interactions associated with changes in VO_2max in ml^.kg^−1.min⁻¹ as a result of resistance-training in older adults. Thus, such endeavors may be a waste of time and resources and potentially unethical.¹⁶ However, this does not negate the importance of future research on behavioral interventions to increase the number of older adults who participate in a regular program of progressive resistance training. This is especially true given the paucity of older adults who do so. For example, the prevalence of US adults 65 years of age and older who participate in strength training at least 2 times per week has been reported to be 9.6%,⁶⁶ while in Australia, the prevalence was reported to be 12.0%.⁶⁷ Second, for those original randomized controlled trials who nevertheless choose to include an examination of moderators, mediators and/or genetic interactions in their experimental design, it is suggested that this only be conducted after testing for true response variability associated with progressive resistance training. Applied and simple methods specific to exercise and nutrition have been described elsewhere.⁶⁸ Third, it is recommended that investigators of randomized controlled trials report sample sizes, as well as change outcome, means and standard deviations for progressive resistance training and control groups. This will allow for further, updated meta-analyses as well as minimizing the potential bias associated with missing data. The use of meta-analysis to examine IIRD is especially relevant given that the use of such has been recommended over such analyses in original trials in general, and progressive resistance training specifically.^40,69 Fourth, a major reason for participation in resistance training in older adults is to increase and/or maintain muscle strength and muscle quality. Along those lines, a recent meta-analysis of 13 randomized controlled trials found that resistance training increased both muscle strength and quality in older adults with sarcopenia.⁷⁰ However, the investigative team is not aware of any previous research that has examined resistance-training-associated IIRD on muscle strength and muscle quality in older adults. Given the former, it would appear reasonable to suggest that a need exists for such work.

Implications for practice

From the authors’ perspective, the results of this study suggest that general versus precision-based progressive resistance training guidelines may be adequate for improving VO_2max in ml^.kg^−1.min⁻¹ in many older adults. These findings may be particularly relevant given that a recent meta-analysis found that genotype-based advice did not affect either dietary or physical activity behavior more than general advice.⁷¹ With the former in mind, the American College of Sports Medicine (ACSM) recommends that older adults participate in progressive resistance training 2 days per week, executing 1 set of 8 to 10 exercises for 8 to 12 repetitions per set for the major muscle groups on each of the 2 days.⁷² In addition, the ACSM also recommends that older adults participate in aerobic and flexibility exercises.⁷² Similar international guidelines for older adults also exist,⁷³ as well as guidelines specific to progressive resistance training in older adults.⁷⁴

A focus on general guidelines may also provide greater reach given that precision-based guidelines may unknowingly disregard the very populations they are intended to reach (the poor, racial/ethnic minorities, etc.) given such factors as the high costs associated with such.⁷⁵ However, this does not negate the importance of adjusting progressive resistance programs based on the various conditions, including multiple co-morbidities, in older adults. For example, a systematic review of 41 studies by Marengoni et al., found that the prevalence of multimorbidity in older adults ranged from 55% to 98%.⁷⁶

Implications for policy

The findings of this study suggest that policies aimed at promoting general versus precision-specific progressive resistance training programs in older adults may be preferable. However, since it is probably unlikely that policies limited to only progressive resistance training in older adults would be feasible, more general policies aimed at physical activity and exercise that include progressive resistance training may have greater success. With respect to adults, including older adults, these include policies aimed at increasing exercise and physical activity in the 1) health care sector, for example, the Exercise is Medicine® (EIM) program,⁷⁷ 2) transportation and planning sector, 3) parks and public spaces sector, and 4) worksite sector.⁷⁸ In addition, a recent systematic review of 57 reviews found convincing evidence of policy effectiveness in some areas, for example infrastructure, but not others, for example, economic.⁷⁹ It was suggested that a need exists to address issues such as the blending of physical activity policies and interventions as well as improvement in the methods for evaluating and collecting evidence on policies.⁷⁹

Strengths and potential limitations

To the best of the authors' knowledge, this is the first IIRD meta-analysis to examine the effects of progressive resistance training on changes in VO_2max in ml^.kg^−1.min⁻¹ in older adults, something that is considered to be the major strength of this study given its previously discussed implications for research, practice, and policy. However, several potential limitations exist. First, based on GRADE, the original study authors rated the strength of evidence with some trepidation (Moderate to High).¹⁰ Second, the quality of original meta-analysis¹⁰ was rated as “Moderate”. Third, based on previous research,⁵² a MCID increase of 1 ml^.kg^−1.min⁻¹ was considered clinically important in the current study. However, selection of a lower or higher MCID would have yielded different results for both IIRD and treatment effect analyses. Fourth, while the standard deviation of individual response approach has been recommended for examining IIRD,^54,55 a limitation of this method, despite random assignment, is that it is based on the assumption that random and within-subject variation is the same between an intervention and control group.⁸⁰ Fifth, while 95% PI's have been suggested to provide more accurate information regarding how much effect sizes vary across studies,⁴⁹ they are negatively affected when sample sizes are small and the data in a meta-analysis are not normally distributed.⁴⁹ Sixth, while sensitivity analyses were conducted, no subgroup, moderator, or meta-regression analyses were performed. However, in an aggregate data meta-analysis, analyses such as meta-regression are considered to be observational in nature, and thus, do not support causal interpretation.⁸¹ Rather, such analyses would need to be conducted in a large randomized controlled trial. Seventh, while plots such as the Doi plot are used for the assessment of small-study effects (publication bias, etc.), also referred to as reporting biases, it is also possible that asymmetrical plots can be caused by heterogeneity or chance.⁸² Eighth, any meta-analysis based on a previous meta-analysis, including the current one, suffers from the weaknesses in both the original trials as well as the meta-analysis itself. Finally, the potential for ecological fallacy, specifically Simpson's paradox, exists given that this was an aggregate data meta-analysis.⁸³

Conclusions

While progressive resistance training may increase VO_2max in ml^.kg^−1.min⁻¹ in many older adults, there is a lack of true resistance-training-associated IIRD. Therefore, other factors, especially random variation (measurement error, day-to-day biological variation), are most likely responsible for any observed variation in VO_2max in ml^.kg^−1.min⁻¹ among older adults. Given the former, general versus precision-based resistance training guidelines may suffice for older adults.

Supplemental Material

sj-docx-1-sci-10.1177_00368504241227088 - Supplemental material for Resistance training and inter-interindividual response differences on cardiorespiratory fitness in older adults: An ancillary meta-analysis of randomized controlled trials

Supplemental material, sj-docx-1-sci-10.1177_00368504241227088 for Resistance training and inter-interindividual response differences on cardiorespiratory fitness in older adults: An ancillary meta-analysis of randomized controlled trials by George A. Kelley, Kristi S. Kelley and Brian L. Stauffer in Science Progress

Supplemental Material

sj-docx-2-sci-10.1177_00368504241227088 - Supplemental material for Resistance training and inter-interindividual response differences on cardiorespiratory fitness in older adults: An ancillary meta-analysis of randomized controlled trials

Supplemental material, sj-docx-2-sci-10.1177_00368504241227088 for Resistance training and inter-interindividual response differences on cardiorespiratory fitness in older adults: An ancillary meta-analysis of randomized controlled trials by George A. Kelley, Kristi S. Kelley and Brian L. Stauffer in Science Progress

Footnotes

Authors’ contributions

GAK was responsible for the conception and design, acquisition of data, analysis and interpretation of data, drafting the initial manuscript and revising it critically for important intellectual content. KSK was responsible for the conception and design, acquisition of data, and reviewing all drafts of the manuscript. BLS was responsible for the conception and design, interpretation of data and reviewing all drafts of the manuscript. All authors read and approved the final manuscript.

Availability of data

All data for this study are available from the corresponding author upon reasonable request.

Declaration of conflicting interests

The author(s) declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Informed consent/ institutional review board approval

The proposed study is an aggregate data meta-analysis of previously reported summary data. Therefore, neither Informed Consent nor Institutional Review Board Approval is required.

ORCID iD

Brian L. Stauffer

George A. Kelley

Supplemental material

Supplemental material for this article is available online.

Author biographies

Dr. George A. Kelley is a Professor in the Department of Epidemiology and Biostatistics. His research focuses on using the meta-analytic approach to examine the effects of exercise and physical activity on health-related disease. He holds a doctorate in Exercise Science.

Kristi S. Kelley is a Research Instructor in the School of Public and Population Health at Boise State University. Her research focuses on using the meta-analytic approach to examine the effects of exercise and physical activity on health-related disease. She possesses a Master’s Degree in Health Promotion.

Dr. Brian L. Stauffer is a Professor of Medicine, Division of Cardiology. His research focuses on translational cardiovascular experimentation in clinical populations and in developing and using animal models of human disease. He holds a doctorate in Medicine.

References

United Nations Department of Economic and Social Affairs Population Division. World population ageing 2017 - Highlights (ST/ESA/SER.A/397), 2017.

Prince

Guo

, et al. The burden of disease in older people and implications for health policy and practice. Lancet 2015; 385: 549–562.

Levy

Slade

Chang

, et al. Ageism amplifies cost and prevalence of health conditions. Gerontologist 2020; 60: 174–181.

2018 Physical Activity Guidelines Advisory Committee. 2018 Physical activity guidelines advisory committee scientific report. 2nd ed. Washington, DC U.S. Department of Health and Human Services, 2018.

Kodama

Saito

Tanaka

, et al. Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis. JAMA 2009; 301: 2024–2035.

Ezzatvar

Ramirez-Velez

de Asteasu

MSL

, et al. Cardiorespiratory fitness and all-cause mortality in adults diagnosed with cancer systematic review and meta-analysis. Scand J Med Sci Sports 2021; 31: 1745–1752.

Nocon

Hiemann

Müller-Riemenschneider

, et al. Association of physical activity with all-cause and cardiovascular mortality: a systematic review and meta-analysis. Eur J Cardiovasc Prev Rehabil 2008; 15: 239–246.

Qiu

Cai

Sun

, et al. Is estimated cardiorespiratory fitness an effective predictor for cardiovascular and all-cause mortality? A meta-analysis. Atherosclerosis 2021; 330: 22–28.

Laukkanen

Isiozor

Kunutsor

. Objectively assessed cardiorespiratory fitness and all-cause mortality risk: an updated meta-analysis of 37 cohort studies involving 2,258,029 participants. Mayo Clin Proc 2022; 97: 1054–1073.

10.

Smart

TFF

Doleman

Hatt

, et al. The role of resistance exercise training for improving cardiorespiratory fitness in healthy older adults: a systematic review and meta-analysis. Age Ageing 2022; 51: 9.

11.

Ross

Goodpaster

Koch

, et al. Precision exercise medicine: understanding exercise response variability. Br J Sports Med 2019; 53: 1141–1153.

12.

Teran-Garcia

Rankinen

Bouchard

. Genes, exercise, growth, and the sedentary, obese child. J Appl Physiol 2008; 105: 988–1001.

13.

Williamson

Atkinson

Batterham

. Inter-individual responses of maximal oxygen uptake to exercise training: a critical review. Sports Med 2017; 47: 1501–1513.

14.

Bonafiglia

Swinton

Ross

, et al. Interindividual differences in trainability and moderators of cardiorespiratory fitness, waist circumference, and body mass responses: a large-scale individual participant data meta-analysis. Sports Med 2022; 52: 2837–2851.

15.

Hecksteden

Kraushaar

Scharhag-Rosenberger

, et al. Individual response to exercise training - a statistical perspective. J Appl Physiol 2015; 118: 1450–1459.

16.

Atkinson

Batterham

. True and false interindividual differences in the physiological response to an intervention. Exp Physiol 2015; 100: 577–588.

17.

Ades

Ballor

Ashikaga

, et al. Weight training improves walking endurance in healthy elderly persons. Ann Intern Med 1996; 124: 568–572.

18.

Andersen

Schmidt

Nielsen

, et al. Effect of football or strength training on functional ability and physical performance in untrained old men. Scand J Med Sci Sports 2014; 24: 76–85.

19.

Bocalini

Serra

Dos

. Moderate resistive training maintains bone mineral density and improves functional fitness in postmenopausal women. J Aging Res 2010; 2010: 760818.

20.

Brentano

Cadore

da Silva

, et al. Physiological adaptations to strength and circuit training in postmenopausal women with bone loss. J Strength Cond Res 2008; 22: 1816–1825.

21.

Costa

Kanitz

Reichert

, et al. Water-based aerobic training improves strength parameters and cardiorespiratory outcomes in elderly women. Exp Gerontol 2018; 108: 231–239.

22.

Elsangedy

Oliveira

GTA

Machado

, et al. Effects of self-selected resistance training on physical fitness and psychophysiological responses in physically inactive older women: a randomized controlled study. Percept Mot Skills 2021; 128: 467–491.

23.

Guido

Lima

Benford

, et al. Effects of 24 weeks of resistance training on aerobic fitness indexes of older women. Rev Bras Med Esporte 2010; 16: 259–263.

24.

Hagberg

Graves

Limacher

, et al. Cardiovascular responses of 70- to 79-yr-old men and women to exercise training. J Appl Physiol 1989; 66: 2589–2594.

25.

Hagerman

Walsh

Staron

, et al. Effects of high-intensity resistance training on untrained older men. I. Strength, cardiovascular, and metabolic responses. J Gerontol 2000; 55: B336–B346.

26.

Haykowsky

McGavock

Vonder Muhll

, et al. Effect of exercise training on peak aerobic power, left ventricular morphology, and muscle strength in healthy older women. J Gerontol Biol Sci 2005; 60: 307–311.

27.

Hunter

Wetzstein

McLafferty

Jr , et al. High-resistance versus variable-resistance training in older adults. Med Sci Sports Exerc 2001; 33: 1759–1764.

28.

Irving

Lanza

Henderson

, et al. Combined training enhances skeletal muscle mitochondrial oxidative capacity independent of age. J Clin Endocrinol Metab 2015; 100: 1654–1663.

29.

Kallinen

Sipilä

Alen

, et al. Improving cardiovascular fitness by strength or endurance training in women aged 76–78 years. A population-based, randomized controlled trial. Age Ageing 2002; 31: 247–254.

30.

Kanegusuku

Queiroz

Chehuen

, et al. Strength and power training did not modify cardiovascular responses to aerobic exercise in elderly subjects. Braz J Med Biol Res 2011; 44: 864–870.

31.

Lovell

Cuneo

Gass

. Strength training improves submaximum cardiovascular performance in older men. J Geriatr Phys Ther 2009; 32: 117–124.

32.

Madden

Levy

Stratton

. Exercise training and heart rate variability in older adult female subjects. Clin Invest Med 2006; 29: 20–28. https://pubmed.ncbi.nlm.nih.gov/16553360/

33.

Okazaki

Kamijo

Takeno

, et al. Effects of exercise training on thermoregulatory responses and blood volume in older men. J Appl Physiol 2002; 93: 1630–1637.

34.

Romero-Arenas

Blazevich

Martínez-Pascual

, et al. Effects of high-resistance circuit training in an elderly population. Exp Gerontol 2013; 48: 334–340.

35.

Schmidt

Hansen

Andersen

, et al. Cardiovascular adaptations to 4 and 12 months of football or strength training in 65- to 75-year-old untrained men. Scand J Med Sci Sports 2014; 24: 86–97.

36.

Strasser

Keinrad

Haber

, et al. Efficacy of systematic endurance and resistance training on muscle strength and endurance performance in elderly adults–a randomized controlled trial. Wien Klin Wochenschr 2009; 121: 757–764.

37.

Thomas

Pierzga

Kenney

. Aerobic training and cutaneous vasodilation in young and older men. J Appl Physiol 1999; 86: 1676–1686.

38.

Vincent

Braith

Feldman

, et al. Improved cardiorespiratory endurance following 6 months of resistance exercise in elderly men and women. Arch Intern Med 2002; 162: 673–678.

39.

Garner

Hopewell

Chandler

, et al. When and how to update systematic reviews: consensus and checklist. Br Med J 2016; 354: i3507.

40.

Steele

Fisher

Smith

, et al. Meta-analysis of variation in sport and exercise science: examples of application within resistance training research. SportRxiv (Preprint) 2022: 1–23. https://doi.org/10.51224/SRXIV.214.

41.

Schunemann

Brozek

Guyatt

, et al. GRADE handbook for grading quality of evidence and strength of recommendation [updated October 2013]. The GRADE Working Group 2013, 2013.

42.

GRADEpro

GDT.

GRADEpro guideline development tool [Software]. McMaster University and Evidence Prime, 2022.

43.

Gwet

. Computing inter-rater reliability and its variance in the presence of high agreement. Br J Math Stat Psychol 2008; 61: 29–48.

44.

Klein

. Implementing a general framework for assessing interrater agreement in Stata. Stata J 2018; 18: 871–901.

45.

Shea

Reeves

Wells

, et al. AMSTAR 2: A critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. Br Med J 2017; 358: j4008.

46.

Doi

Barendregt

Khan

, et al. Advances in the meta-analysis of heterogeneous clinical trials I: the inverse variance heterogeneity model. Contemp Clin Trials 2015; 45: 130–138.

47.

Doi

SAR

Furuya-Kanamori

Thalib

, et al. Meta-analysis in evidence-based healthcare: a paradigm shift away from random effects is overdue. Int J Evid Based Healthc 2017; 15: 152–160.

48.

Higgins

JPT

Thompson

Deeks

, et al. Measuring inconsistency in meta-analyses. Br Med J 2003; 327: 557–560.

49.

Borenstein

. In a meta-analysis, the I-squared statistic does not tell US how much the effect size varies. J Clin Epidemiol 2022; 152: 281–284.

50.

Furuya-Kanamori

Barendregt

Doi

SAR

. A new improved graphical and quantitative method for detecting bias in meta-analysis. Int J Evid Based Healthc 2018; 16: 195–203.

51.

IntHout

Ioannidis

Rovers

, et al. Plea for routinely presenting prediction intervals in meta-analysis. BMJ Open 2016; 6: e010247.

52.

Laukkanen

Zaccardi

Khan

, et al. Long-term change in cardiorespiratory fitness and all-cause mortality: a population-based follow-up study. Mayo Clin Proc 2016; 91: 1183–1188.

53.

Hopkins

Marshall

Batterham

, et al. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 2009; 41: 3–13.

54.

Hopkins

. Individual responses made easy. J Appl Physiol 2015; 118: 1444–1446.

55.

Hrubeniuk

Bonafiglia

Bouchard

, et al. Directions for exercise treatment response heterogeneity and individual response research. Int J Sports Med 2022; 43: 11–22.

56.

Williamson

Atkinson

Batterham

. Inter-individual differences in weight change following exercise interventions: a systematic review and meta-analysis of randomized controlled trials. Obes Rev 2018; 19: 960–975.

57.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB

0.06 MB