The literature search was performed using the MEDLINE, CINAHL, and EMBASE databases from January 2010 to February 2021. Search terms included: acute coronary syndrome, myocardial ischemia, coronary artery disease, heart valve disease, cardiovascular disease, cardiac event, coronary artery bypass graft, percutaneous coronary intervention, angioplasty, stent, valve repair, valve prosthesis, valve replacement, cardiac surgery, cardiac catheterisation, cardiac rehabilitation, exercise training and secondary prevention. The full search strategy is included in the Supplementary Material.

Studies were included if they satisfied the following criteria: (i) original research published in English, (ii) contemporary, that is, published after January 2010, (iii) study population was adults over 18 years, (iv) studied Phase II (after discharge from hospital) or phase III CR programs (supervised or unsupervised outpatient CR), (v) described CR programs post myocardial infarction, PCI or cardiac surgery. Exclusion criteria were: (i) studies using only Phase I CR (in-patient, post event), (ii) inadequately described the exercise/rehabilitation protocol used, (iii) had fewer than 10 participants.

Two researchers (MS and JF) screened the searched articles by title and abstract and all studies meeting the inclusion criteria were reviewed by full text. Any issues with eligibility were resolved by discussion with a third researcher (JO).

Data were extracted from the studies onto an Excel (Microsoft Corporation, USA) spreadsheet for analysis. Data extracted included author, year of publication, country of study, study design, number of participants, age, sex ratios, indication for CR, level of fitness at entry measured by baseline cardiorespiratory function tests, supervision level, CR setting, phase of rehabilitation, exercise program type, intensity and duration, primary and secondary outcome measures and dropout rates. Assessment of study methodological quality (type of study, level of evidence, performance of a power analysis, treatment allocation, blinding, dropout rates and methodological flaws) was performed. Only studies with level of evidence IV or greater were included to minimize bias. This is shown in Table 1.

Statistical analysis was performed using Stata software (Version 14.0). Data from studies using the same outcome measures peak VO2max, 6-min walk test (6MWT) and Metabolic Equivalent of Task (METs) was pooled and analyzed with 95% confidence intervals (CIs) reported to assess for change in cardiorespiratory fitness from baseline to exit from rehabilitation. Where studies compared two different CR methods rather than a control group, both exercise protocols were included separately in the analysis. Where studies utilized two or more cardiorespiratory outcome measures, all were included. Heterogeneity assessed using an I² test and a fixed effects model was used. A meta-regression analysis was performed to determine the effect of other factors, specifically age and sex.

The initial search identified 2,726 studies (Figure 1). A total of 23 duplicates were removed, and 2,565 records were excluded based on title and/or abstract screening, leaving 138 for full-text review. A further 96 articles were excluded due to lack of exercise prescription, no cardiac event, having only Phase I CR, no outcome measures to assess change in cardiorespiratory status, or were protocols or review articles. Thus 42 articles were available for quantitative analysis, of which 13 were randomized controlled trials (33, 35, 36, 40, 41, 44, 45, 47, 48, 52, 54, 63, 65) and 29 were cohort studies (26–32, 34, 37–39, 42, 43, 46, 49–51, 53, 55–62, 64, 66, 67). Included papers are shown in Table 1, together with study results and the level of evidence of the study. Studies were conducted primarily in Europe (21 studies) and Asia (16 studies) but also included Canada (2) and Brazil (2) and the USA (1).

A total of 14,683 patients were included in the study data. The mean age of study participants was 59.5 years (range of means 45.3–80.3 years). Eighteen studies included comparison of differing exercise protocols (including one that separated into four study groups), creating 62 groups of patients that completed CR. Of the study groups 8/60 (13%) had a mean age <55 years [one study encompassing two groups did not report mean age of participants (36)]. The mean percentage of male participants within the study groups was 82.7%. In total, 46/61 (75%) had ≥75% of participants being male (one study did not specify the sex breakdown of study participants).

Baseline level of fitness was examined by assessing values for peak VO_2max, 6-minute walk test (6MWT) and METs on entry into CR, shown in Table 2. Mean baseline levels of cardiorespiratory function were: peakVO₂ 23 ml/kg/min (age-based normal value = 25.1 ml/kg/min), 6MWT 420.5m (normal = 555) and baseline METs 7 (normal = 7). 27% of study groups had a mean average cardiorespiratory function level below age-matched normative values (68–70).

Peak VO₂ was utilized by 28 studies (26–44, 56, 57, 59–62, 64, 65, 67), 12 studies examined 6MWT (34, 44–49, 60, 61, 63, 64, 66) and 9 studies utilized METs based on formal testing (50–55, 57–59). Multiple outcome measures were used in 7 studies (34, 44, 57, 59–61, 64). Dropout rate was specified in 16 of the studies (29, 33, 35, 36, 39–42, 44, 45, 51–54, 64, 65). The rate varied from 2 to 75%, with a mean dropout rate of 19.3%.

The duration of CR programs varied from 3 to 52 weeks, with a mean duration of 14.8 weeks. The prescribed programs are shown in Supplementary Table 1. The type of exercise prescribed varied, and included: walking (outdoor or treadmill), stationary bike, stationary stepping, stair climbing, swimming, arm ergometer, elliptical trainer, cross-country skiing, rowing and gymnastics. In some studies, strength training, calisthenics, or flexibility training were also included, but this was not examined in this study. Frequency of exercise was on average 3 sessions per week (range 2–10), with an average time of 38 min (range 10–90) min per session. Intensity levels varied between groups and were expressed as either: a percentage of VO₂peak (5 groups, average 59–75% of VO₂peak) a percentage of maximum or peak heart rate (11 groups, average 71–82% of maximum HR), a percentage of heart rate reserve (HRR) (13 groups, average 53.5–78% of HRR) or exercise at anaerobic or ventilatory threshold levels (6 groups). Nine study groups specified interval training, with the remainder focused on steady state training. Three studies were determined to be high-intensity interval training (HIIT) when defined as ≥85% of VO_2max (71).

Cardiorespiratory fitness increased in all measures assessed across all study groups examined in this study. Change in peakVO_2max was assessed in a total of 2,141 patients, with an observed change of 3.23 ml/kg/min (2.86–3.60, 95% confidence interval). Mean change in 6MWT was examined in 999 patients, with a mean change of 86.98 m (64.54–109.42, 95% confidence interval) on exiting CR. The utilization of METs to assess change in fitness was used in 12,248 patients in total, showing a mean average improvement in METs of 1.53 METs (1.28–1.78, 95% confidence interval) from entry to exit from CR (Figure 2).

The proportion of patients meeting population age and sex referenced normative values for cardiorespiratory fitness increased across each of the three domains. On entry to CR, 27% (11/41) of study groups had a mean peakVO_2max at the level or above age-matched normative values. This increased to 54% (22/41) on exit from CR. Improvements were also seen in 6MWT from 5% (1/22 study groups) to 32% (7/22 study groups) and METs from 69% (9/13 study groups) to 77% (10/13 study groups).

A further meta-regression analysis was performed to assess the independent effects of age and sex on change in cardiorespiratory fitness with CR. For each outcome measure, age and proportion of male participants had no significant impact upon improvement in cardiorespiratory function following CR. The results of this analysis are shown in Table 3.

Lower uptake of CR in women compared to men has been identified in multiple studies (4, 19–21). Similarly, our study showed that only 17% of patients completing CR were female. This is despite 36.3% of patients suffering AMI being female (18), which would be represented the AMI and PCI aspects of this patient cohort. In addition, three-quarters of study groups had a male to female ratio of >3:1. Reasons previously attributed for this discrepancy have included: inadequate CR education and referral, transport and health system reasons, and competing demands (such as caregiver responsibilities) (72). However, even after age and covariate adjustment, women were found to be 9% less likely to attend CR and 13% less likely to complete CR after commencement (4).

Some efforts have been made to improve CR uptake and completion among female cardiac patients. One commonly used approach is to have sex-specific or women-only CR programs (22). This can lead to greater adherence (73), improved mental health outcomes (74) and comparable functional outcomes when compared to mixed sex CR (75), as well as improved satisfaction from participants (76). These classes more commonly offer alternative forms of exercise, such as dance and yoga. However, as an example in Australia, only 1% of programs offer women-only classes (22). Home-based or community-based programs have also been suggested as a potential improvement, as these have been shown to be equally effective as traditional CR (72) and may circumvent some of the potential barriers to attendance for women.

Previous studies have shown that younger age is associated with a higher drop-out rate (77). There is also evidence that older patients are more likely to complete CR (78, 79). This study appears to support these findings. The mean age of subjects completing CR in the groups analyzed was 59.5 years, but the average ages of study groups were closely centered around this point, showing little variability. Only 13% of study groups had a mean age <55 years, when this age group represents 23% of patients suffering AMI (80). As a result, when assessing the efficacy of CR programs, our data indicate that the research is skewed toward the needs of those aged older than 55.

In Australia and the UK, exercise recommendations for cardiac patients are largely toward light- to moderate-intensity aerobic exercise (81–83). Lower intensity exercise is easier to implement, particularly within restricted resource situations and is more acceptable to older adults and those with physical limitations (11), but this could be a significant deterrent to younger, more athletic patients, for whom the program is “too easy” (effectively detraining). Exercise interventions in coronary artery disease patients that recommend moderate to vigorous intensity exercise have shown greater improvements in cardiorespiratory fitness and cardiac risk factors than lower intensity interventions (11), as well as being safe and well accepted. Applying these measures to younger patients may lead to better uptake and completion, and better health outcomes.

Our study shows that the entry-level baseline level of fitness for patients completing CR was lower than that of normative data for sedentary 60–69 year-olds. We found that 73% of study groups were below average compared to their sedentary matched counterparts. That is, the majority of CR patients are below average fitness for their age.

It can be argued that acute events such as MI, PCI and cardiac surgery would be expected to acutely lower these cardiorespiratory functional measures. However, these low values may also represent drop-out or failure of inclusion of patients coming from a high level of baseline fitness capacity, and that these patients are disproportionately represented in the 19.3% of dropout cases from CR. Although this group may have lower risk for future cardiac events, the importance of transitioning them back to an acceptable level of function is high. A lack of guidance on how to progress back to normal function may increase the risk of further cardiac issues in this patient cohort.

The average CR program lasted 15 weeks and included cardiovascular training with 3 sessions per week. The average duration of each session was 38 min. Intensity levels were on average 59–75% of VO₂peak, 71–82% of maximum HR, 53.5–78% of HRR or at anaerobic or ventilatory thresholds. Only 16% of groups were specified as interval training, with the remainder focused on steady state training. HIIT programs represented only 5% of prescribed programs.

Despite this study only including phase II or III CR programs, there was great variability in the programs prescribed to patients following cardiac events. This variability encompassed duration of programs, type of exercise program, frequency and duration of sessions as well as the intensity level and individualization of rehabilitation. This variation in prescription of exercise programs makes drawing conclusions for the type of rehabilitation that works optimally for different subsets of cardiac rehabilitation very difficult.

Cardiorespiratory fitness outcomes increased across all study groups examined in this study. Mean improvement in cardiorespiratory function was 3.23 ml/kg/min for VO_2max, 86.98 m for 6MWT and 1.53 METs. The results of this study mirror those of a multitude of others showing that over the time period of CR, cardiorespiratory outcome measures improve on average in all participants.

Meta-regression analysis found that neither age nor the proportion of females to males in study groups had any significant impact upon improvement in cardiorespiratory function following CR. This highlights the importance of CR and its effectiveness for all patients, as well as the benefits that can be attained regardless of age and sex. However, it raises the question of whether this improvement could be increased further and whether rehabilitation is challenging enough or tailored enough to the patients involved.

This study displays the outcomes of CR in those who complete the rehabilitation process. This study compares the demographics of these patients to those eligible to complete CR and shows that younger, female and moderate-to-high fitness individuals are under-represented. CR appears to largely meet the needs of male, older and low fitness individuals, with either a failure to be referred, a failure to attend or a failure to complete rehabilitation in those outside of these demographics. It may be that due to rehabilitation being tailored to the more unfit members of the cohort, those of higher fitness feel that their needs are not met and drop out.

This study has a number of limitations. First, it examines CR research literature which includes both CR programs and research studies rather than a snapshot of CR programs themselves solely. Examining demographic trends within these data are beneficial and has been performed previously, but to our knowledge, no assessment of this demographic data has been performed on research study participants to this point. It is also important to reflect on the need for research looking at these specified groups and whether they benefit from rehabilitation as much as the more studied cohorts of males, sedentary and older patients.

Although this systematic review included all available literature, there were a small number of studies using each individual cardiorespiratory outcome measure resulting in high heterogeneity. Therefore, we have included only the three most widely reported outcome measures to minimize this heterogeneity. Additionally, regarding the delivery of cardiac rehabilitation, there is considerable variation particularly in the duration of cardiac rehabilitation. This varies greatly from study-to-study and country-to-country, leading to different effect sizes. The variation in the way that cardiac rehabilitation is prescribed and delivered in terms of frequency, intensity, type, and time, also means that there is unavoidable heterogeneity in results.

Most studies included did not specify baseline level of physical function and the level of physical activity prior to commencement of CR. Only 8/60 (13%) of studies specified a previous physical activity level. Baseline cardiorespiratory function testing was used for this purpose but may not adequately describe how physically active a rehabilitation participant was prior to commencement in CR. When grouped among other participants of low cardiac fitness, it will be difficult to ascertain whether the benefit gained from CR is greater or smaller than the mean for those who were previously physically active.

Dropout rate amongst studies was difficult to accurately determine as some studies only examined patients who completed CR whereas others included dropout rates from the commencement of CR. Whether these dropout rates preferentially affect younger, female and fitter patients is impossible to conclusively ascertain from this data. This is a potential area of future research, alongside further investigation into the reasons behind the poorer completion rate of each of these subgroups.

Cardiorespiratory assessment measures also appeared to be variable. In assessing baseline level of cardiorespiratory fitness, the percentage of study groups meeting age-matched normative values varied depending on which of the cardiorespiratory measures were used [peak VO_2max (27%), 6MWT (5%) and METs (69%)]. This may reflect inaccuracies in the normative values but may also show the inaccuracy of cardiorespiratory function testing in general, particularly with 6MWT and METs.

This study highlights a number of concerns regarding the equity of CR. There appears to be a lack of research and exercise programs focused on the needs of female, young and fit patients when recovering from AMI, PCI or cardiac surgery. Taken together, this indicates large gaps in CR, with the possibility of patients “falling through the cracks” in recovering from a major life event. It also represents a potential risk of increased future cardiac events if the needs of these patients are not met.

In order to improve CR completion, it is important to address the reasons behind a lack of uptake and a lack of completion of CR in these patients. We need to ensure that CR is made more accessible and suitable to a broader range of people, as well as more appealing to those who are referred. By making CR more personalized, we can change CR from the current model of modest efficacy to an effective therapy.

Our opinion on how this is achieved is through a more patient-centric approach. Commencing any form of exercise is a two-stage process: motivation, as well as of performance. By addressing the goals, interests and motivations of the patient, greater long-term self-efficacy can be fostered. Moving away from the current paternalistic model of care does not have detrimental effects on patient outcomes and can be cost-effective, decrease fear and enhance motivation (45, 52, 84).

The future will involve a model of care that is more personalized and may be that a triage model is utilized involving clinical determinants of health as well as goals, risk factors and social determinants. Smart technology will allow a more specific, patient-centered approach, which can work better for those with time-constraints, whilst reducing stress and giving equivalent benefits to supervised training (45, 52). This can be done remotely and reach patients outside of suburban areas. Risk factors for coronary events appear to be changing, with a greater burden from sedentary behavior and lower physical fitness. By using more innovative methods to improve activity, risk factor modification can be sustainably improved.