Interindividual response variation to exercise-based cardiac rehabilitation on changes in cardiorespiratory fitness in heart transplant patients: A secondary meta-analysis of randomized controlled trials

Abstract

Background

Determine whether true exercise-associated interindividual response differences (IIRD) occur in cardiorespiratory fitness as a result of exercise-based cardiac rehabilitation in heart transplant patients.

Methods

Using data from a recent (2023) meta-analysis of 9 randomized controlled trials representing 296 patients (163 exercise, 133 control), an aggregate data meta-analysis of treatment effects (change outcome differences between exercise and control groups) was conducted as well as an IIRD meta-analysis using the inverse variance heterogeneity model. The primary outcome was cardiorespiratory fitness (VO_2max) in ml/kg/min.

Results

Statistically significant and clinically important increases equivalent to 14.5% were observed for VO_2max in ml/kg/min ($\stackrel{\circledR }{X}$, 3.0, 95% confidence interval (CI), 2.4-3.7 ml/kg/min, p < 0.001; Q = 11.8, p = 0.16; I² = 32.0%, 95% CI, 0%-68.8%; ${\tau }^2$ = 0.4). The 95% prediction interval (PI) was 1.2-4.7 ml/kg/min. However, no statistically significant IIRD were observed ($\stackrel{\circledR }{X}$, 0.6, 95% CI, −1.1 to 1.4 ml/kg/min; ${\tau }^2$ = 2.9). The 95% PI was −2.7 to 2.8 ml/kg/min.

Conclusions

While exercise-based cardiac rehabilitation increases VO_2max in ml/kg/min in heart transplant patients, a lack of true exercise-associated IIRD exists. A need exists for additional well-designed randomized controlled trials of longer duration to determine the long-term effects of exercise-based cardiac rehabilitation on VO_2max in ml/kg/min heart transplant patients.

Background

Worldwide, the number of heart transplants has been reported to be 8,409 in the year 2021,¹ while in the United States, 4,111 heart transplants were performed in the year 2022, an increase of 7.1% from 2021.² A major issue in heart transplant patients is post-transplant survival. For example, a recent study that included all adults (N = 30,606) from the Scientific Registry of Transplant Recipients who underwent isolated heart transplantation reported the survival rate to be 27% over a 12-year period.³ A significant prognostic factor in post-transplantation survival is cardiorespiratory fitness (CRF).⁴ In a recent retrospective analysis of 235 heart transplant patients, each 14% standard deviation increase in CRF, assessed as percent predicted peak oxygen consumption (%VO₂), was associated with a 32% decrease in mortality.⁴ Given that exercise is the best method for increasing CRF, the use of such may be especially important for increasing CRF in this population. Along those lines, a recent systematic review with aggregate data meta-analysis of 9 randomized controlled trials (RCTs)5, 6, 7, 8, 9, 10, 11, 12, 13 that examined the effects of exercise-based cardiac rehabilitation on changes in VO_2max in ml/kg/min in 299 heart transplant patients reported statistically significant improvements of 3.0 ml/kg/min (95% confidence interval (CI), 2.3, 3.8 ml/kg/min, p < 0.001; Q = 11.8, p = 0.16; I² = 32.0%; ${\tau }^2$ = 0.42) as a result of exercise.

The tailoring of exercise programs that attempt to account for factors, such as individual variability in one’s genes, that is, precision exercise prescription, is considered to be one of the most important issues in exercise medicine.¹⁴ Unfortunately, the a priori assumption that true exercise-associated interindividual response differences (IIRD) exist may not be reasonable. For example, while heritability estimates ranging from 25% to 65% have been reported with respect to CRF responses to aerobic exercise training,¹⁵ a review found that none had appropriately quantified exercise-associated IIRD.¹⁶ A reanalysis of the one trial that included a control arm and appropriately quantified true IIRD found that the standard deviation of change in CRF, assessed as VO_2max in ml/kg/min, was actually greater in the control (± 5.6 ml/kg/min) than aerobic exercise (± 3.7 ml/kg/min) arm, thereby negating any exercise-associated IIRD.¹⁶ In addition to the previous study, a recent individual participant data meta-analysis of 8 RCTs representing 1,879 adults found a lack of exercise-associated IIRD on cardiorespiratory fitness as a result of exercise training.¹⁷ These results suggest that factors other than IIRD to exercise training (random variation, physiological responses associated with behavioral changes that are not the result of the exercise intervention) are responsible for any observed variation in CRF.¹⁸ Consequently, a search for potential moderators and mediators associated with exercise and CRF may not be necessary.¹⁹ From an applied perspective, these findings suggest that general vs specific exercise guidelines may be preferred with respect to the effects of exercise training on CRF.

To the best of the authors’ knowledge, no previous study has examined exercise-associated IIRD on CRF, assessed as VO_2max in ml/kg/min, among heart transplant patients. Such a study is important for determining if a search for potential moderators and mediators, including genetic interactions, is justifiable in this population. In addition, an examination of such has important implications for exercise prescription. Thus, the purpose of this study was to fill this gap by conducting an aggregate data meta-analysis with a focus on exercise-associated IIRD on VO_2max in ml/kg/min as a result of exercise-based cardiac rehabilitation in heart transplant patients.

Methods

Data source

The a priori protocol for this study is registered in Open Science Framework (https://osf.io/axvqj/). For the current study, data from a recent (2023) systematic review with meta-analysis of RCTs examining the effects of exercise-based cardiac rehabilitation on VO_2max (ml/kg/min) in heart transplant patients was used.²⁰ This study included 9 RCTs5, 6, 7, 8, 9, 10, 11, 12, 13 representing 296 patients (163 exercise, 133 control).²⁰ While the previous meta-analysis²⁰ also included a separate meta-analysis based on 2 studies that compared high-intensity interval training to moderate-intensity continuous training,21, 22 they were not included in the current meta-analysis given the small sample size and lack of adequate control group for comparison. The rationale for using the existing meta-analysis vs conducting an updated systematic review with meta-analysis for the current study was based on the recency of this prior work (2023)²⁰ as well as a decision-tree algorithm from the Panel for Updating Systematic Reviews (PUGs) on when an updated or new systematic review with meta-analysis is warranted.²³

Details regarding the original meta-analysis are described in the original article.²⁰ The former notwithstanding, the characteristics of the studies included in the original meta-analysis are described here in the Methods section vs the forthcoming Results section of the manuscript because they were derived either directly or indirectly from the original meta-analysis.²⁰ Briefly, this previous systematic review with meta-analysis included 9 studies conducted in 8 different countries (Brazil,⁷ Canada,⁹ Denmark,⁸ Germany,¹² Italy,⁵ Norway,⁶ Taiwan,¹⁰ USA),11, 13 5 of which received funding from either Foundation8, 9, 11 or Government6, 7 sources.

Mean group ages between studies in heart transplant patients ranged from 45.0 to 60.6 years in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 52.2 ± 13.1, median = 54.4) and 45 to 59 years in the control groups ($\stackrel{\circledR }{X}$ ± SD = 52.4 ± 13.4, median = 59.0).5, 6, 7, 8, 9, 10, 11, 12, 13 Donor ages ranged from 30.0 to 34.9 years in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 32.9 ± 12.7, median = 34.0) and 25.0 to 38.0 years in controls ($\stackrel{\circledR }{X}$ ± SD = 32.5 ± 12.4, median = 34.0).5, 6, 9, 10, 13 The percentage of males ranged from 66.7% to 95% ($\stackrel{\circledR }{X}$ ± SD = 79.2 ± 8.3, median = 78.2) in the exercise groups and 55.6% to 91.7% ($\stackrel{\circledR }{X}$ ± SD = 75.3 ± 12.1, median = 77.6) in the control groups, while the percentage of females ranged from 5.0% to 33.3% in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 20.8 ± 8.3, median = 21.8) and 8.3% to 44.4% in the control groups ($\stackrel{\circledR }{X}$ ± SD = 24.7 ± 12.1, median = 22.4).6, 7, 8, 9, 10, 11, 12, 13 Dropouts ranged from 0% to 14.3% in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 5.6 ± 5.3, median = 6.4) and 0% to 30% in controls ($\stackrel{\circledR }{X}$ ± SD = 8.6 ± 11.0, median = 3.8).5, 6, 7, 8, 9, 10, 11, 13

The number of years since heart transplantation ranged from 0.2 to 6.8 in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 4.8 ± 4.8, median = 5.0) and 0.2 to 7 in control groups ($\stackrel{\circledR }{X}$ ± SD = 4.1 ± 3.5, median = 4.0). Body mass index in kg/m² ranged from 24.4 to 28.3 in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 26.7 ± 4.1, median = 26.5) and 25.1 to 28.7 in controls ($\stackrel{\circledR }{X}$ ± SD = 26.4 ± 4.5, median = 26.2).6, 7, 8, 9, 10, 11, 12 For baseline VO_2max (ml/kg/min), values ranged from 12.1 to 27.7 in the exercise groups ($\stackrel{\circledR }{X}$ ± SD = 20.5 ± 7.1, median = 20.0) and 13.2 to 28.5 in the control groups ($\stackrel{\circledR }{X}$ ± SD = 20.0 ± 7.7, median = 19.0).5, 6, 7, 8, 9, 10, 11, 12 One study that did not include standard deviation data for calculating pooled baseline results across studies reported a baseline VO_2max of 9.2 ml/kg/min in the exercise group and 10.4 ml/kg/min in the control group.¹³ Assessment of VO_2max in ml/kg/min was conducted using a variety of treadmill and cycle ergometer protocols.5, 6, 7, 8, 9, 10, 11, 12, 13

All exercise groups participated in home and/or center-based moderate-intensity continuous training and/or high-intensity interval training consisting of one or more of the following: walking, jogging, cycling, staircase running.5, 6, 7, 8, 9, 10, 11, 12, 13 Four studies also included strengthening exercises.9, 10, 11, 13 Length of training ranged from 8.0 to 26.1 weeks ($\stackrel{\circledR }{X}$ ± SD = 15.6 ± 7.5, median = 12.0),5, 6, 7, 8, 9, 10, 11, 12, 13 frequency from once every 2 weeks to 5 times per week ($\stackrel{\circledR }{X}$ ± SD = 3.2 ± 1.2, median = 3.0),5, 6, 7, 8, 9, 10, 11, 12, 13 and duration from 16 to 60 minutes per session ($\stackrel{\circledR }{X}$ ± SD = 32.3 ± 14.0, median = 30.0).5, 6, 7, 8, 9, 10, 11, 12, 13 Compliance, defined as the percentage of exercise sessions attended and limited to the 5 studies (56%) in which data were calculable,6, 8, 9, 12, 13 ranged from 47% to 100% ($\stackrel{\circledR }{X}$ ± SD = 79% ± 23%, median = 89%). Dose of exercise, calculated as the product of frequency and minutes per week, ranged from 90 to 120 minutes per week for the 3 studies (33%) in which this data could be calculated.5, 7, 11 Using American College of Sports Medicine categories for exercise intensity given the different methods used to assess such,²⁴ intensity ranged from “Light” to “Maximal.”5, 6, 7, 8, 9, 10, 11, 12, 13 No serious adverse events associated with the exercise interventions were reported.5, 6, 7, 8, 9, 10, 11, 12, 13

Based on the Cochrane Risk of Bias Assessment Instrument for randomized controlled trials (RoB2),²⁵ the authors of the previous meta-analysis²⁰ considered the overall risk of bias across all studies to be “high,” with 7 of the 9 studies5, 6, 9, 10, 11, 12, 13 classified as “high risk” and 2 as “some concerns.”7, 8 The overall strength/certainty of the evidence based on the Grading of Recommendations, Assessment, Development, and Evaluations' Instrument²⁶ was considered to be “moderate.”²⁰

Data extraction

Independent data extraction from the previous meta-analysis was conducted by the first and second authors of the current study using Microsoft Excel for Microsoft 365 MSO (Version 2301).²⁷ Information extracted included the following: (1) study authors, (2) year of publication, (3) sample sizes as well as change outcome means and standard deviations for exercise and control groups, as well as treatment effects and their 95% CI. The first 2 authors extracted all data independent of each other and then met to review their selections. Any disagreements were resolved by consensus. If consensus could not be achieved, the third author provided a recommendation. Prior to meeting, Gwet’s AC1 statistic was used to assess inter-rater agreement.²⁸

Quality of included meta-analysis

To assess the quality of the included systematic review with meta-analysis,²⁰ the Assessment of Multiple Systematic Reviews (AMSTAR 2) instrument²⁹ was used.³⁰ AMSTAR 2 includes 16 questions for RCTs of health care interventions.²⁹ Response choices for RCTs, depending on the question, are “Yes,” “No,” “Partial Yes,” or “No meta-analysis conducted,” with “Yes” and “Partial Yes” signifying that the item was addressed satisfactorily.²⁹ While an overall score is not a primary goal of AMSTAR 2, the following categories have been suggested with respect to confidence in the quality of a systematic review with meta-analysis: “High,” “Moderate,” “Low,” or “Critically low.”²⁹ Independent assessment using AMSTAR 2 was conducted by the first 2 authors. They then met to assess agreement, with any disagreements resolved by consensus. If consensus could not be attained, a recommendation was provided by the third author. Prior to meeting, Gwet’s AC1 statistic was used to assess inter-rater agreement.²⁸

Data synthesis

Effect size metric

The effect size metric for changes in CRF was VO_2max in ml/kg/min. The rationale for using the original metric (VO_2max in ml/kg/min) is based on the belief that it is more clinically interpretable than a metric such, as the standardized mean difference effect size.

Meta-analysis of treatment effects

While not the primary aim of the current study, a meta-analysis of treatment effects for changes in VO_2max in ml/kg/min was first conducted by pooling results from each study into one overall pooled result. These were calculated from the change outcome differences in VO_2max in ml/kg/min between the exercise and control groups along with their standard deviations. Treatment effects (exercise minus control) from each study were pooled into one overall treatment effect using the inverse variance heterogeneity model of Doi et al, a model that incorporates heterogeneity into the analysis and has been shown to produce more accurate 95% CIs than the traditional random-effects model of Dersimonian and Laird used in the original meta-analysis.²⁰ Improvements in VO_2max in ml/kg/min was indicated in the direction of benefit, that is, positive values favoring exercise-based cardiac rehabilitation. Two-tailed z-alpha values ≤ 0.05% and 95% CIs that did not include zero (0) were considered statistically significant.

Heterogeneity was examined using the Cochran Q statistic with alpha values ≤0.10 considered to be statistically significant heterogeneity.³¹ Inconsistency was examined using the I-squared (I²) statistic along with its 95% CI.³² Increasing values of I², a relative statistic that is an extension of Q, are indicative of greater inconsistency. In addition to Q and I², tau-squared (${\tau }^2$), an absolute measure of between-study heterogeneity, was also calculated.³³ While not examined in the original meta-analysis,²⁰ the Doi plot and Luis Furuya-Kanamori (LFK) index were used to examine for small-study effects (publication bias, etc).³⁴ The Doi plot is reported to be more intuitive than the commonly used funnel plot and the LFK index more accurate than Egger’s regression intercept test.³⁴ Based on previous recommendations, small-study effects for the LFK index were classified as no (<±1), minor (between ±1 and ±2), or large (>±2) asymmetry.³⁴ Influence analysis, a form of sensitivity analysis, was also conducted by removing each study from the model once. Furthermore, outlier analysis was performed by deleting results for those effect sizes from studies in which their 95% CI fell completely outside the pooled 95% CI. Neither influence analysis nor outlier analysis was apparently conducted in the original meta-analysis.²⁰ Finally, given the small number of studies included, an a priori decision was made to not conduct any subgroup, moderator, or meta-regression analyses.

Meta-analysis of IIRD

To examine, for the first time exercise response variation for changes in VO_2max in ml/kg/min in heart transplant patients, true IIRD between exercise and control group standard deviations for VO_2max in ml/kg/min were considered as point estimates, calculated from each study as follows³⁵:

$$\sqrt{{\mathit{SD}}_e^2-{\mathit{SD}}_c^2}$$

where ${\text{SD}}_e^2$ is the standard deviation for the exercise and ${\mathit{SD}}_c^2$ is the standard deviation for the control group. The standard error of the variance for point estimates was then computed as follows³⁵:

$$\mathit{SE}=\sqrt{2\left(\frac{{\mathit{SD}}_e^4}{{\mathit{DF}}_e}+\frac{{\mathit{SD}}_c^4}{{\mathit{DF}}_c}\right)}$$

where DF are the degrees of freedom minus 1 for the exercise and control groups’ standard deviations. Positive values for IIRD, that is, individual response variation, were suggestive of greater variability for changes in VO_2max in ml/kg/min in the exercise vs control groups.³³ Findings were then pooled by combining individual response variances and their standard errors into one overall point estimate and 95% CI using the inverse variance heterogeneity model.³⁶ The SD for point estimates and their 95% CIs were then calculated by computing the square root of each.³³ If the lower 95% CI was negative, the sign was initially disregarded, the square root calculated, and the sign reapplied. Absolute between-study heterogeneity was calculated using tau ($\tau$).³³

Additional analyses

While not calculated in the original meta-analysis,²⁰ 95% prediction intervals (PI) based on the pooled mean effect, standard error, and $\tau$ were calculated for the results from both the treatment effects and IIRD meta-analyses.³⁷ Ninety-five percent PIs provide an estimate of the range of expected effects if a new study was conducted. In addition to 95% PIs, the magnitude for exercise minus control group changes in VO_2max in ml/kg/min for both treatment effects and IIRD meta-analyses were compared to a minimally clinically important difference (MCID) in VO_2max of 1.0 ml/kg/min based on previous research demonstrating a relative risk decrease of 9% in all-cause mortality.³⁸ While the MCID was not reported in the original meta-analysis,²⁰ the current authors felt that this analysis was important from the perspective of clinical relevance. As a further assessment of clinical importance, the following probabilistic categories were used: (1) <0.5% (most unlikely or almost certainly not), (2) 0.5% to 5% (very unlikely), (3) 5% to 25% (unlikely or probably not), (4) 25% to 75% (possibly), (5) 75% to 95% (likely or probably), (6) 95% to 99.5% (very likely), (7) >99.5% (most likely or almost certainly).³⁹

Software used for analysis

Data extraction and AMSTAR2 assessment were conducted using Microsoft Excel for Microsoft 365 MSO (Version 2301),²⁷ while statistical analyses were conducted using Meta XL (version 5.3),⁴⁰ SSC-Stat (version 3.0),⁴¹ as well the most recent user-written versions of metan⁴² and KAPPAETC43, 44 within Stata (version 16.1).⁴⁵ Statistical tests were 2-tailed.

Amendments to protocol

There were no amendments to the a priori protocol.

Results

Intercoder agreement for data abstraction

The overall agreement rate for data abstraction prior to correcting differences was 0.98 (95% CI, 0.95, 1.0). All disagreements were resolved by the first 2 authors.

AMSTAR 2 results

Study and item-level results for AMSTAR 2 are shown in Supplementary file 1. The overall agreement rate prior to correcting disagreements was 0.58 (95% CI, 0.13, 1.0). All disagreements were resolved by the first two authors. Seventy-five percent of the items were rated positively (yes/partial yes). The overall quality of the included meta-analysis was considered “Moderate.”

Treatment effect results

The overall results for changes in VO_2max in ml/kg/min are shown in Table 1 while individual and pooled study results are shown in Figure 1. At the individual study level, 8 of 9 studies (88.9%) yielded nonoverlapping 95% CI. Across all studies, statistically significant increases in VO_2max (p < 0.001) were observed as a result of exercise-based cardiac rehabilitation, ranging from a low of 1.5 (95% CI, 0-3.0) to 5.6 (95% CI, 3.4, 7.8) ml/kg/min. Overall changes were equivalent to a relative increase of 14.5% (95% CI, 10.8%-18.1%). No statistically significant heterogeneity was observed while the 95% CI for inconsistency based on I² was wide (0.0%-68.8%). Tau-squared as an absolute measure of between-study heterogeneity was 0.42. Minor asymmetry suggestive of small-study effects (publication bias, etc) was observed (Supplementary file 2). With each study deleted from the model once, results remained statistically significant across all deletions, ranging from a low of 2.8 ml/kg/min (95% CI, 2.1, 3.4) to 3.3 ml/kg/min (95% CI, 2.6, 3.9). No outliers were found. The 95% PI for what result one might expect to achieve if they conducted their own randomized controlled trial in the same population did not include zero (Table 1). The percent chance, that is, probability, of a clinically meaningful improvement of at least 1.0 ml/kg/min in VO_2max was 98.3% (very likely to be clinically important).

Table 1.

Treatment Effect and IIRD Results

Variable	Studies (#)	Participants (#)	$\stackrel{\circledR }{X}$ (95% CI)a	95% PIb
TEc
VO2max ml/kg/min	9	296	3.0 (2.4, 3.7)d	1.2, 4.7d
SDIRe
VO2max ml/kg/min	9	296	0.6 (−1.1, 1.4)	−2.7, 2.8

^a$\stackrel{\circledR }{\mathit{X}}$ (95% CI), mean and 95% confidence interval.

^b95% PI, 95% prediction interval.

^cTE, treatment effects.

^dNonoverlapping 95% confidence intervals.

^eSD_IR, standard deviation of individual response differences.

Figure 1.

Forest plot for treatment effect changes in VO_2max in ml/kg/min, ordered from smallest to largest increases.

IIRD results

Results for exercise-associated IIRD, the primary aim of the current study, are shown in Table 1. As can be seen, the pooled 95% CI included zero (0), suggesting a lack of statistically significant exercise-associated IIRD in VO_2max in ml/kg/min. The 95% PI was wide and also included zero (0) (Table 1). Absolute between-study heterogeneity based on ${\tau }^2$ was 2.9. With each study deleted from the model once, no statistically significant IIRD were found across all deletions, ranging from a low of 0.3 (95% CI, −1.8, 1.9) to a high of 1.3 (95% CI, −0.7, 2.0). No outliers were observed. The percent chance, that is, probability, of clinically meaningful exercise-associated IIRD for changes in VO_2max in ml/kg/min as a result of exercise-based cardiac rehabilitation was 50.6% (only possibly clinically important).

Discussion

Overall findings

For the primary purpose of this study, the overall findings suggest a lack of exercise-associated IIRD in VO_2max in ml/kg/min as a result of exercise-based cardiac rehabilitation in heart transplant patients. These results are reinforced by (1) overlapping 95% CI, (2) overlapping 95% PI, (3) overlapping 95% PI when each study was deleted from the model once, (4) absence of outliers, and (5) the finding that results were only possibly clinically important. Thus, based on the existing evidence, it appears that any response variability is due to factors other than exercise, for example, random variation (measurement error, biological day-to-day variation), and/or behavioral factors (sleep, diet, etc). Consequently, a search for potential moderators and mediators, including genetic interactions, associated with exercise IIRD may not be warranted. These findings are consistent with previous reviews on exercise and CRF.16, 17

While not the primary purpose of the current study, statistically significant and clinically important increases in VO_2max in ml/kg/min were observed as a result of exercise-based cardiac rehabilitation in heart transplant patients. These findings are supported by (1) nonoverlapping 95% CI, (2) nonoverlapping 95% PI, (3) absence of outliers, (4) statistically significant findings when each study was deleted from the model once, (5) lack of statistically significant heterogeneity and inconsistency, although the 95% CI for I² was wide, and (6) the finding that results were considered “very likely to be clinically important.” These results are similar to the overall results reported in the original meta-analysis by Costa et al.²⁰ From the current investigative team's perspective, these findings provide compelling evidence for the improvement of CRF, as assessed by VO_2max in ml/kg/min, for exercise-based cardiac rehabilitation in heart transplant recipients.

Implications for research

There are several implications for research on exercise-associated IIRD on CRF in heart transplant patients. First, it is suggested that future RCTs examine for IIRD before attempting any moderator or mediator analysis, including testing for genetic interactions. An applied and recommended method for doing so has been described by Swinton et al.⁴⁶ A detailed discussion of exercise-associated IIRD and its assessment has also been provided by Ross et al.¹⁴ Second, it is recommended that future studies report complete data with respect to the dose of exercise that was not only prescribed, but what was performed. For example, in the current study, only 5 (56%) studies reported calculable data for compliance,6, 8, 9, 12, 13 defined as the percentage of exercise sessions attended, while only 3 (33%) reported sufficient data to calculate the dose of exercise (minutes per week calculated from the product of frequency per week and minutes per session).5, 7, 11 None of the studies provided sufficient information so that the dose of exercise could be adjusted for compliance. Thus, it is strongly suggested that future studies provide data for the following: (1) length of the study, (2) number of days per week of exercise, (3) minutes of exercise per session, (4) intensity of the exercise, and (5) compliance to exercise. The provision of such data would allow for the calculation of exercise dose with respect to (1) minutes per week of exercise, (2) minutes per week of exercise, adjusted for compliance, (3) intensity minutes of exercise per week as defined by the 2018 Physical Activity Guidelines for Americans,⁴⁷ and (4) intensity minutes per week of exercise, adjusted for compliance. Finally, additional studies that examine the association between exercise-based cardiac rehabilitation, CRF, and survival in heart transplant recipients are needed.

Implications for practice

Given the observed improvements in CRF as assessed by VO_2max in ml/kg/min as well as the absence of exercise-associated IIRD, it would appear plausible to suggest that adherence to general guidelines for exercise-based cardiac rehabilitation in heart transplant patients would be appropriate. Along those lines, we suggest adherence to the guidelines set forth by the American Association of Cardiovascular and Pulmonary Rehabilitation.⁴⁸ Importantly, these programs should be adapted to the individual needs of each patient, for example, baseline levels of VO_2max in ml/kg/min, choice of exercise, etc.

Implications for policy

The current findings suggest that policies aimed at promoting exercise-based cardiac rehabilitation in heart transplant patients are important. This may be especially relevant since exercise-based cardiac rehabilitation, heart transplant patients, or otherwise, has been reported to be underutilized. For example, despite recommendations regarding the benefits and importance of participation in cardiac rehabilitation, including heart transplant patients,49, 50, 51, 52 participation in these programs has been reported to range from only 19% to 34%.⁵³ As described by Chindhy et al, policies aimed at physician factors (low referral rates, physician endorsement), patient factors (gender, race/ethnicity, medical comorbidities, socioeconomic, psychological), as well as systemic factors (travel and transportation, costs of attendance, fragmented care), would all contribute to increased utilization of cardiac rehabilitation services.⁵¹ In addition, given the costs associated with policies based on genotype-based advice, a focus on general guidelines may be more cost-effective given the findings of a recent meta-analysis showing no difference between the two.⁵⁴

Strengths and potential limitations

The major strength of the current study is that to the best of the investigative team’s knowledge, this is, the first study to examine true exercise-associated IIRD on VO_2max in ml/kg/min as a result of exercise-based cardiac rehabilitation in heart transplant patients. This is important given that CRF is considered to be a clinical vital sign.⁵⁵ In addition, a retrospective study in 178 heart transplant patients found that higher CRF was a strong predictor for long-term survival.⁵⁶ The authors also suggested that VO_2max is a crucial measurement and should be more frequently used after heart transplantation.⁵⁶ Furthermore, a more recent retrospective study by Hanff et al in 235 heart transplant patients found that post-transplant VO_2max was a highly significant prognostic indicator of long-term post-transplant survival.⁴

While the results of this study are important, they should be considered with respect to several potential limitations. First, the fact that the longest intervention only lasted approximately 26 weeks limits one’s ability to determine the long-term effects of exercise-based cardiac rehabilitation on VO_2max in heart transplant patients. Second, an MCID of 1.0 ml/kg/min was used for VO_2max in the current study. However, use of a larger MCID may have led to different findings for those analyses that included the MCID. Third, the standard deviation of individual response approach used in the current meta-analysis has been recommended by others as the preferred method when examining exercise-associated IIRD.35, 57 However, a limitation of this method is that it is grounded on the assumption that both random and within-subject variation are similar in both the exercise and control group.⁵⁸ Fourth, the small number of studies included may have led to poor coverage for the 95% PIs.⁵⁹ Fifth, while sensitivity analyses were performed, no covariate analyses were conducted because of the small number of studies included as well as missing data for different variables from different studies so that covariates, such as the dose of exercise, could not be calculated and examined. Sixth, like any meta-analysis, results are limited by the quality of available data. For example, in the original study by Costa et al,²⁰ the overall risk of bias based on the RoB2²⁵ was considered to be “high,” while the overall strength/certainty of evidence based on the Grading of Recommendations, Assessment, Development, and Evaluations' Instrument²⁶ was considered “moderate.”²⁰

Conclusions

The results of this study suggest that while exercise-based cardiac rehabilitation increases VO_2max in ml/kg/min in heart transplant patients, there is a lack of exercise-associated IIRD. However, additional, well-designed RCTs of longer duration are needed to determine lasting effects.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding and Acknowledgments: None.

Appendix A. Supplementary material

Supplementary material

.