The present review adhered to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (38). It was prospectively registered in the International Registry of Prospective Systematic Reviews (PROSPERO) database under registration number “CRD42023460252” on 5 September 2023. The following databases were electronically searched: PubMed, Web of Science, Cochrane Library, Scopus, and WanFang database. A comprehensive search was conducted for randomized controlled trials (RCTs) investigating the HIIT on body composition, cardiopulmonary function, and metabolic indicators in older adults. The search period ranged from the inception of each database to 1 July 2023. The search strategy is structured to include terms related to “aged,” “older adult,” “older,” “High-Intensity Interval Training,” “HIIT,” and “High-Intensity Intermittent Exercise.” Additionally, relevant literature and previously published systematic reviews were manually screened to identify any studies missed during the initial search process (search details in Additional File 1 in Supplementary Material ).

The retrieved literature was imported into NoteExpress, a literature management software. Two researchers independently conducted a comprehensive literature review, eliminating duplicate and irrelevant studies, extracting relevant data, and cross-verifying the selected literature. In case of any discrepancies, they consulted with each other or sought input from a third party for discussion. A screening flowchart is presented in Figure 1 . The research object of this paper is the older adult, and subjects were not restricted by country, race, or sex, or physical condition, except athletes. The experimental group was HIIT, and the control group was blank control or low-intensity exercise control. Articles in Chinese or English reporting at least one outcome measure were also included. The outcome measures included body mass index (BMI), body fat percentage (BF%), waist circumference (WC), maximal oxygen uptake (VO_2max), systolic blood pressure (SBP), diastolic blood pressure (DBP), resting heart rate (HR_rest), maximal heart rate (HR_max), respiratory exchange rate (RER), total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and fasting plasma glucose (FPG).

The data extraction encompassed the following components: basic information of the included literature (first author’s name, title, and publication year), characteristics of study subjects (age, gender, and number of subjects), training variables (intensity, form, frequency, and intervention period), main results, and key elements of risk of bias assessment. To calculate effect sizes for physical fitness measures in the intervention and control groups, baseline and follow-up mean as well as standard deviations (SDs) were extracted. In case any required data were missing, we contacted the corresponding author to obtain them. If relevant data could not be provided by the author, they were excluded from the analysis. The characteristics of the included studies are presented in Supplementary Table 1 (in Additional File 2).

RCTs were analyzed using the Cochrane Risk of Bias Tool 2.0 (39). There are three levels: low risk, high risk, and uncertain. Two researchers used ReviewManager 5.4.1 software to rigorously evaluate seven aspects of randomized allocation methods, allocation concealment of randomized methods, blinding of research subjects and interveners, blinding of outcome evaluators, integrity of outcome data, possibility of selective reporting, and other sources of bias. The risk of biased judgment in each domain was interpreted as low risk, moderate risk, severe risk, borderline risk, or no information (40). Two reviewers independently assessed the risk of bias, and any disagreements were resolved through a third party. In addition, the risk of publication bias was assessed using funnel plots when the meta-analysis included ≥10 studies.

Review Manager 5.3 (RevMan) software was used for statistical analysis. Because the literature had the same continuous outcome variable and the same unit of measurement, mean differences (MDs) and SDs along with 95% confidence intervals (95% CIs) were used to estimate the effect sizes observed in the studies. The random-effects model (overall effect p-value < 0.05) and fixed-effects model (overall effect p-value > 0.05 under the random-effects model) were the two main models used to synthesize all the results of forest plots. I ² was used to evaluate the degree of heterogeneity: above 25%, 50%, and 75% were low, medium, and high heterogeneity, respectively (41), and the level of meta-analysis was set as α = 0.05. When I ² ≤ 25%, the fixed-effects model was used for combined analysis. If I ² > 25%, the random-effects model was used for combined analysis. In order to increase the stability of the results, sensitivity analysis was performed by one-by-one elimination. These tests focused on significant heterogeneity associated with HIIT, in which individual studies were systematically removed from the meta-analysis and pooled effect estimates were recalculated to assess the impact of individual studies on the meta-analysis. As the objective of this study was to investigate the effects of HIIT on body shape, cardiopulmonary function, and metabolic parameters in the older adult, subgroup analyses were conducted according to intervention duration (≤12 weeks and >12 weeks) and disease type (CVD, diabetes, hypertension, cancer, and others) when I ² ≥ 50%. The possible publication bias was evaluated by drawing a funnel plot. Further analysis was performed when more than five studies were included in the subgroup.

Figure 1 describes the PRISMA process in detail. A total of 2,278 studies were retrieved from the database, and 7 studies were obtained from other sources. After the strict screening, 87 (42–128) studies were finally included, with a total of 4,213 older adult people. There were 2,099 subjects in the HIIT group, aged 66.8 ± 18.7 years, and 2,114 subjects in the control group, aged 63.1 ± 20 years. The exercise frequency ranged from 1 to 21 times per week, the duration ranged from 7 to 180 min per time, and the cycle ranged from 2 to 48 weeks. The measurement methods of outcome indicators were also reported. Among the included studies, 25 studies described medical supervision in detail, 29 did not describe it in detail, and 33 did not describe it. In addition, 24 studies reported a total of 196 people withdrawing due to family, subjective will, and other factors, accounting for 4.65% of the total number. There were 46 cases of adverse events, and the incidence of adverse events was 1.09%. Of note, adverse events were not attributed to HIIT. Figure 2 shows the geographical and sample distribution of the included studies, with 65.16% of the studies in Europe, 14.62% in North America and South America,14.24% in Asia, 5.03% in Australia, and 0.95% in Africa, involving 21 countries and regions. Of these, 658 were from Norway, 543 were from Switzerland, 537 were from Belgium, 191 were from Denmark, 182 were from Spain, 200 were from the United Kingdom, 148 were from Sweden, 144 were from Germany, 116 were from Italy, 26 were from France, 244 were from Canada, 222 were from the United States, 150 were from Brazil, 254 were from China, 133 were from Iran, 88 were from Japan, 72 were from South Korea, 29 were from Thailand, 24 were from Indonesia, 212 were from Australia, and 40 were from Egypt, for a total of 4,213 individuals.

A total of 87 RCTs were included, and the overall risk of bias was low. Of these RCTs, 84/87 evaluated using randomized sequence generation, 84/87 assessments used allocation concealment, 77/87 subjects and staff were blinded, 77/87 were blinded to the outcome data evaluation, 84/87 had complete data, 86/87 reported no selectivity, and 39/87 had no other bias among the included (the results are shown in Additional File 3 in Supplementary Material ).

The studies on the physical effects of the older adult can be divided into three aspects: body shape, cardiopulmonary function, and metabolic function. A total of 14 indicators were involved, and the detailed results of all indicators are recorded in Table 1 .

Owing to the intricate relationship between body shape and health as well as disease (129), it serves as a primary indicator for assessing the health status of older individuals. Two extensive cohort studies have demonstrated that both high BMI and BF% along with low BMI elevate the risk of mortality (130, 131). Therefore, it is crucial for older adult individuals to effectively manage their BMI and reasonably reduce their BF%. This meta-analysis aims to elucidate the impact of HIIT on three clinical parameters related to body shape in older adults. The results of the meta-analysis demonstrated that HIIT effectively improved BF% in older adults, which is consistent with the findings reported by Donghai et al. (132) and McLeod et al. (33), but contradicts the conclusions drawn from other studies (26, 29–32, 34–37). Unlike previous studies that primarily included individuals with normal weight as baseline data, the older adults we enrolled were mostly overweight or obese (BMI ≥25), although we did not restrict inclusion and exclusion criteria. HIIT stimulates lipolysis by increasing catecholamines and growth hormone (133). Meanwhile, HIIT improved BF%, including body composition and lipid profile, by increasing HDL in human and animal models (134–137), a mechanism consistent with our meta results for biochemical metrics. The meta-analysis results indicated insufficient evidence regarding the impact of HIIT on WC and BMI in older adults, aligning with most previous systematic reviews (26, 29–32, 34–37). To further assess the stability of these combined results, a sensitivity analysis was performed. The pooled results for BMI were stable, and for WC after excluding the study conducted by Tjønna et al. (117), I ² decreased from 80% to 48%, This change may be related to two aspects. First, the difference in the included population: Tjønna et al. (117) included the older adult with metabolic syndrome, while other studies included healthy people or people with one disease. Second, unlike BMI, WC directly reflects the characteristics of fat distribution. Given the relationship between the pathological mechanism of metabolic syndrome and visceral fat function, the effect size may be nonlinear with other populations (138). These factors may exacerbate heterogeneity. It is worth noting that, excluding the study by Madssen et al. (91), there was no significant change in I ²; however, WC exhibited a reversal (p = 0.04). Further analysis revealed that the exercise supervision implementation rate in the study was only 1/3, and this low compliance resulted in an increase in WC after HIIT intervention. This finding may further elucidate the insufficient effect of HIIT on WC combined results. Therefore, caution should be exercised when interpreting whether HIIT can improve WC in the older adult due to the lack of robustness of these results. HIIT did not significantly improve BMI in the older adult. The results of the study by Donghai et al. (132) suggested that although HIIT decreased BF% in the older adult, the increase in lean body mass did not change their total body weight significantly, resulting in no significant difference in BMI before and after exercise. This lack of significant difference in measurement results due to adaptive changes in the body is not related to the study hypothesis.

The pleiotropic effects of maintaining a high level of cardiopulmonary function on the health of the older adult have been widely recognized (139), and it has been identified as a priority for promoting the health of the older adult by the World Health Organization (WHO) (12). A total of six cardiopulmonary function indicators were included in this meta-analysis to evaluate the effects of HIIT on cardiopulmonary function in the older adult. The existing evidence shows that HIIT can effectively improve VO_2max and HR_max in the older adult. The analysis of 55 studies showed that the increase of VO_2max by 2.46 mL/kg/min is very significant, and VO_2max is recognized as the gold standard of cardiopulmonary function. A 44-year follow-up study showed that VO_2max was inversely associated with the risk of cancer, cardiovascular events, and all-cause mortality (140), and each 1-MET increase reduced cardiovascular events by 15% and all-cause mortality by 13% (141). Subgroup analysis found that the effect of HIIT on VO_2max in the older adult was generalized and was independent of the intervention period and the health of the subjects. In addition, HR_max is also one of the important indexes to evaluate the cardiopulmonary function of the older adult, and it is positively correlated with VO_2max. The improvement of HR_max is consistent with VO_2max, but the improvement of HR_max is less extensive than VO_2max. Although the aim of this study was not to investigate the underlying physiological mechanisms of HIIT, based on the current evidence, we highlighted some mechanisms related to the obtained results, such as high physiological stress caused by HIIT, high recruitment pattern of type I and type II fibers, and intense muscle contraction during exercise, leading to imbalance of the ATP/ADP relationship, increased PGC1-α activation and, thus, increased PGC1-α activation, and strong body VO_2max (134, 142, 143). The results of this study show that there is insufficient evidence for HIIT to improve blood pressure in the older adult, which is consistent with the results of previous studies (30–37, 132), but in sharp contrast to the study by Du et al. (29), the difference in conclusions may be related to the training method [isometric exercise has a better effect on blood pressure change than other exercise methods (144)] and the sensitivity of the subjects to HIIT. Therefore, the effects of aging on HIIT-induced fitness may be interesting to investigate in the future. Blood pressure is the most important modifiable risk factor for all-cause morbidity and mortality; however, our evidence found that HIIT had no significant effect on SBP and DBP, and neither changed significantly after individual removal, possibly because the majority of non-hypertensive people we included undermined this effect. After the study by Villelabeitia-Jaureguizar et al. (124) was excluded by RER, the results were reversed (p = 0.04). Since this study compared the improvement difference between moderate-intensity exercise and HIIT, the effect of HIIT was reduced. The results for RER need to be interpreted with caution, given the lack of robustness of the sensitivity analyses. Unfortunately, there is insufficient evidence on the effect of HIIT on HR_rest in the older adult; thus, it was omitted from this meta-analysis.

Metabolic abnormalities can cause a variety of chronic diseases, including obesity, CVD, diabetes, and cancer, which bring huge public health problems and medical burdens (145). Metabolic disorders are often more serious in the older adult. Priority should be given to reducing the related risks in the older adult at this stage while preventing them. A total of five metabolic indicators were included in this meta-analysis, and the existing evidence showed that HIIT effectively improved HDL, which was in sharp contrast with previous studies (26, 29, 31) probably because previous studies were based on comparing the differences between HIIT and moderate-intensity exercise, while most of our work was based on the comparison between HIIT and blank control. Because of the strong association between dyslipidemia and CVD, HIIT is of great significance for the improvement of HDL levels. However, HIIT had no significant effect on TG and LDL, which is consistent with previous studies (33, 37). The results of TC and FPG were reversed after excluding the study of Gjellesvik et al. (119), which may be due to the differences in the baseline levels of our included studies, two types of exercise in some studies, and the use of drugs to treat metabolic abnormalities or affect metabolic indicators. These factors may reduce the improvement effect of HIIT on metabolic indicators. It also indicates the lack of robustness of the TC and FPG results, and this result needs to be interpreted with caution.

Among the included studies, 25 studies provided comprehensive descriptions of medical supervision, while 29 studies lacked detailed information in this regard, and 33 studies did not mention it at all. Additionally, withdrawal from the intervention was reported in 24 studies due to reasons such as familial obligations, personal preferences, and other factors, involving a total of 196 participants, accounting for approximately 4.65% of the overall sample size. Adverse events were documented in 46 cases with an incidence rate of approximately 1.09%. Importantly, none of these adverse events were attributed to HIIT.

The large sample size and high heterogeneity observed in our study were expected due to differences in methodology and study subjects, as the range of studies included all older adults except those with contraindications to exercise. Therefore, subgroup analyses and sensitivity points were conducted to assess the role and stability of the pooled results. Sensitivity analysis revealed that WC, RER, TC, and FPG lacked robustness. Despite implementing a rigorous search strategy, language bias was inevitable as only Chinese and English literature was retrieved within our constraints. Furthermore, variations in exercise equipment, interval time, and intervention duration among the HIIT studies included prevented us from conducting subgroup analysis; thus, specific exercise doses cannot be recommended at this stage. Lastly, our included studies did not assess medication use or baseline/daily physical activity, which are critical factors for evaluating the hypothesis that HIIT improves health in older adults. Additionally, we lacked data on other influencing factors such as gender and exercise capacity. Finally, it should be emphasized that most of the included samples were from the Northern Hemisphere, and the unbalanced sample distribution may limit the promotion of HIIT in the Southern Hemisphere.