This systematic review and meta-analysis followed the PRISMA with 2020 updates and Cochrane Collaboration guidelines (Supplementary File 1) (21) by a pre-registered protocol (INPLASY202530123).

Two authors (Y-GH and J-HS) performed a comprehensive search independently from inception until March 15, 2025, using PubMed, EMBASE, Web of Science, the China National Knowledge Infrastructure, Wanfang, and Cochrane Library. The search incorporated medical subject headings and keywords, specifically targeting terms like “β-hydroxy-beta-methylbutyrate” OR “beta-hydroxy-beta-methylbutyrate” OR “hydroxy methylbutyrate” AND “surgery” OR “operation” OR “operative,” without language or year restrictions. The search strategy is detailed in Supplementary File 2. Moreover, grey literature was explored through https://scholar.google.com and https://www.basesearch.net, and references of selected articles were examined for any eligible studies.

We included studies in the meta-analysis based on the following criteria. First, participants were adult patients over 18 years old undergoing surgery. Second, the intervention involved HMB in the experimental group, with no limitations on the dosage, administration route, duration of treatment, or use of additional supplements. Third, comparators were non-HMB interventions, placebo or conventional therapy. Fourth, only RCTs were included. Finally, the outcomes measured included clinical outcomes, muscle measurements, and nutritional status indicators. We excluded studies based on the following criteria: children or pregnant women, duplicate publications, or those designed as cohort studies, abstracts, reviews, or comments.

The two authors extracted relevant data from the tables, figures, texts or additional files from the included RCTs. These variables included trial characteristics (first author’s name, year of publication, country, and study design), patient characteristics (age, sex ratio, patient population, body mass index, and body weight), HMB and control regimens, and predefined outcomes. We preferred to use the intention-to-treat results of the included RCTs. For studies that provide results from assessments at different time points, we selected the longest assessment time points after treatment for inclusion in the meta-analysis. Any disagreements were resolved by consulting a third researcher (H-BH).

The primary outcomes were clinical indicators such as the length of stay (LOS) in the hospital and postoperative complications. Secondary outcomes included muscle measures [i.e., appendicular skeletal muscle mass (ASMM), mid-arm muscle-circumference (MAMC), lean body mass (LBM)], physical function [i.e., hand grip strength (HGS), gait speed, or 6-min walking distance (6-MWD)], and nutritional status [i.e., serum albumin, total albumin, body mass index (BMI), or body weight (BW)].

Y-GH and J-HS independently conducted quality assessments of each publication using the Cochrane Risk of Bias tool (version 2) (22). Publication bias was evaluated by visual inspection of funnel plots when 10 or more trials were available. We used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system to evaluate the quality of evidence (23). The disagreements between the two authors were resolved by consulting a third author (H-BH).

We used RevMan 5.4 software, as recommended by the Cochrane Library, for the meta-analysis. The mean differences (MD) or odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) were used to assess efficacy. For studies reporting median and interquartile range (IQR) but not SD, we estimated the mean and SD from the median and IQR, respectively (24). We conducted meta-analyses on predefined outcomes when at least two trials were available for pooling. We used the I² statistic to test for heterogeneity, with values of I² < 50% and I² > 50% indicating low and high heterogeneity, respectively. A fixed-effect model was used when I² < 50%, and a random-effect model was used when I² > 50% (25).

To test the outcomes’ robustness and explore the potential influence factors, we conducted sensitivity analyses to identify each study’s influence on the overall pooled estimate of the outcome of interest. We also performed subgroup analyses based on the following criteria: (1) exercise (with or without), (2) location (Asia or non-Asia), (3) patient age (≥65 years or <65 years), (4) HMB regimen (use of HMB alone or combined HMB with additional supplements), and (5) study design (double-blind or undouble-blind).

The primary search identified 135 records from the databases and additional searches. After removing duplicates, 91 records remained for title and abstract screening, of which 74 were excluded as they did not meet the inclusion criteria. A Subsequent full-text screening ruled out 6 RCTs, with the reasons for exclusion detailed in Supplementary File 3. Finally, 11 RCTs were included in the quantitative analyses (18–20, 26–33). The process of identification, screening and study inclusion is illustrated as a PRISMA flow diagram in Figure 1.

Table 1 summarizes the characteristics of the included RCTs. These trials, published from 2011 to 2025, were conducted in seven countries: Türkiye, China, Italy, Spain, Japan, Iran, and the United States. In total, 575 patients were analyzed, with 294 in the HMB group and 281 in the control group. Of these trials, nine were single-center (18–20, 26–29, 32, 33), while two were multi-center (30, 31), focusing on conditions like hip fracture (27, 28, 31), cardiac surgery (19, 20), cancer (30, 33), liver transplantation (18, 29), and endoscopic surgery (26). Five of the included RCTs administered HMB as a single supplement (20, 27, 29–31), while the other six combined HMB with arginine and glutamine (18, 19, 26, 28, 32, 33). All studies administered a daily dose of 3 g of HMB, taken as 1.5 g twice daily. Additionally, four trials incorporated exercises with HMB interventions (18, 20, 31, 32). Follow-up assessments were conducted in all RCTs, with the timing of outcome assessment ranging from 10 days to 12 months post-intervention. The details regarding the study strategies are summarized in Table 2. A total of six studies (18–20, 27, 30, 33) described the complications which were summarized in Supplementary File 6.

The results of the risk of bias assessment of the included RCTs are presented in Supplementary File 4. The risk of bias in RCTs ranged from low to high in all critical domains. Evaluation of publication bias by visually inspecting funnel plots showed potential publication bias in the included trials (Supplementary File 5). Through the GRADE method, we rated the evidence for pooled data for hospital LOS, complication, HG, BW, BMI, ASMM, albumin, and 6-MWD as moderate, moderate, low, very low, very low, very low, and low, respectively (Supplementary File 6).

Hospital LOS and postoperative complications were reported in nine (18–20, 28–33) and seven (18–20, 27, 30, 33) RCTs, respectively. Our analyses showed that HMB significantly reduced hospital LOS (MD −0.90 days; 95% CI, −1.79 to −0.01; I² = 0%, p = 0.05; Figure 2a) and decreased postoperative complications (RR 0.50; 95% CI, 0.32 to 0.79; I² = 0%, p = 0.003; Figure 2b) compared with the control group. Although we found no statistical heterogeneity, we conducted stratified analyses based on predefined main study characteristics and clinical conditions to explore any additional source of heterogeneity. The sensitivity analysis, which excluded any single study, yielded results closely aligned with the overall combined estimates. For hospital LOS, the p values ranged from 0.02 to 0.36, with all I² = 0%. For postoperative complications, the p values ranged from 0.003 to 0.05, with I² ranging from 0 to 26%. Subgroup analyses were also conducted, and most pooled subgroup results supported the reduction of hospital LOS and complications in the HMB group (Table 3).

As to muscle measures, MAMC, ASMM, SMM, and LBM were described by two, four, two, and two studies, respectively. The pooled estimates showed that compared with the control group, the HMB group had a more significant increase in changes regarding MAMC (MD 1.16 cm; 95% CI, 0 to 2.33; p = 0.05; Figure 3a) (28, 29) and ASMM (MD = 1.35 kg; 95% CI, 0.16–2.55; p = 0.03; Figure 3b) (20, 27, 29, 31), but had similar changes in SMM (MD = −0.45 kg; 95% CI, −1.42 to 0.53; p = 0.37; Figure 3c) (31, 33), and LBM (MD −0.22 kg; 95% CI, −3.04 to 2.60; p = 0.88; Figure 3d) (26, 33).

As to outcomes of physical function, changes in HGS, 6-MWD, and GS was reported by seven, two, and one trials, respectively. When pooled, no significant differences were found in changes in HGS (MD = 1.82; 95% CI, −0.69 to 4.34; p = 0.16; Figure 4a) (18, 20, 27–29, 31, 33) between the two groups. However, patients in HMB had significantly more 6-MWD (MD = 52.36 m; 95% CI, 13.99 to 90.72; p = 0.007; Figure 4b) than control group. In addition, one study reported a significant improvement in GS than the controls (p = 0.0002) (20).

Regarding other variables of nutritional status, compared to the control group, the HMB group did not show significant improvement from baseline after treatment, including BW (MD = 0.21 kg; 95% CI, −0.15 to 0.58; p = 0.25; Figure 5a) (26–28, 31–33), BMI (MD = 0.07 kg/m²; 95% CI, −0.22 to 0.36; p = 0.63; Figure 5b) (26, 28, 31), serum albumin (MD = 3.40 g/L; 95% CI, −0.66 to 7.46; p = 0.10, Figure 5c) (27, 30, 31), and total albumin (MD = 7.65 days; 95% CI, −2.40 to 17.69; p = 0.14; Figure 5d) (27, 31).

For many years, HMB has been used in athletes for muscle building, strength, endurance enhancement (34, 35), and recovery after exercise-induced muscle injury (12). In recent years, its interest has rapidly expanded to include the elderly ill populations (15), sarcopenia (14), cancer (16), and critically ill patients (17). A meta-analysis suggested that 12 weeks of HMB supplementation improved muscle mass, strength, and physical function in the elderly population (15). Another meta-analysis that included nine RCTs suggested that HMB improved muscle mass and strength, but there was no evidence of benefits for physical function in patients with sarcopenia (14). Similarly, Prado et al. (16) pooled the results from 15 studies of cancer patients treated with HMB and showed that HMB had a beneficial effect on muscle mass and function in this patient population. These studies consistently demonstrated the beneficial effects of HMB in elderly or frail populations and were consistent with some of our findings. Conversely, in studies of critically ill patients, HMB did not improve mortality or other clinical outcomes in ICU patients (17). This ineffectiveness may be related to the HMB strategy used in the included trials. Our study included surgical patients who were also from a population with advanced age, tumors, and heart diseases and were subjected to varying degrees of stress from surgery. We analyzed the effects of HMB comprehensively on clinical outcomes, muscle strength, mass, body function, and nutrition in this population, initially showing the benefits of HMB and/or its additives. Our study adds a new population for clinical HMB application in terms of meta-analysis evidence.

Although HMB can benefit surgical patients, several issues are worth exploring. First, our primary outcomes were hospital LOS and postoperative complications, as they were the most selections among the included studies (18–20, 28–33). This selection reflected that clinicians pay more attention to patient-centered clinical outcomes. However, the effect of HMB on the outcome of hospital LOS is indirect. Meanwhile, hospital LOS is a relatively subjective outcome since it is often influenced by clinical practices such as bed availability, turnover, and patient wishes. Fortunately, the clinical aspect of the benefits is supported by the positive outcome findings of complications, muscle mass, and functional activity. Moreover, most subgroup analyses based on predefined influencing factors showed a tendency to benefit HMB supplements in surgical patients.

In contrast, the outcome of complications is relatively objective. Most of the complications reported in the included articles focused on nosocomial infections, including surgical site, lung, and urinary tract infections (19, 20, 27, 33). HMB has shown effects in promoting wound healing (36), lowering CPK (37), and increasing serum growth hormone levels (33). The latter may promote wound healing (38). Reducing these complications contributes to the success of the surgery and subsequent recovery.

Second, the effectiveness of HMB may be influenced by the strategy of its application. The previous meta-analysis of the critically ill population not benefiting from HMB may be related to ICU patients receiving HMB later (17). In the study by Supinski et al. (39), patients had received mechanical ventilation for an average of 6 days before HMB supplementation. Theoretically, HMB administered after muscle weakness did not improve muscle function. Therefore, this delay may have hindered the beneficial effects of HMB therapy. Meanwhile, ICU patients often suffer from gastrointestinal dysfunction, as well as fasting and gastric decompression (40), which may impair drug absorption and limit the effectiveness of HMB in improving muscle function. On the contrary, the timing of HMB administration in surgical patients can be initiated very early in the preoperative period, i.e., some studies have started patients on HMB as early as 10–30 days before surgery (19, 20, 30). Most patients can take HMB orally before surgery and get good absorption. Importantly, surgical patients are very receptive and compliant with HMB therapy. For example, compliance was as high as 95% in the study by Ogawa et al. (20). Moreover, some investigators have controlled and promoted compliance by asking patients treated with HMB to return the HMB/placebo bag the day before surgery (19). These differences may be why surgical patients benefit more from HMB than critically ill patients. In addition, rehabilitation performed preoperatively and postoperatively in surgical patients is easier to implement and works well with HMB (20, 32). Conversely, ICU patients who receive HMB rarely undergo rehabilitation. Even if they receive rehabilitation, the effect is not accurate. One ICU study reported that their patients were poorly trained (<10 min/day) (39). Therefore, using perioperative HMB, especially preoperatively, may be a promising area.

In addition, the beneficial effects of HMB may be influenced by the synergistic effect of common additives, including ARG and Glu (18–20, 26, 28, 32, 33). However, these included trials did not compare individuals and combinations, making it difficult to assess the effects of HMB alone. Some previous studies have suggested that ARG and Glu, as immune nutrients, can significantly improve the immunity and infection of postoperative patients (41, 42). However, Kuhls et al. (43) compared HMB alone or in combination with arginine and glutamine in trauma patients and found no difference between use alone and in combination, and HMB significantly improved nitrogen balance. In addition, a meta-analysis suggests that the presence of HMB may exacerbate the prognosis of ICU patients (44). Therefore, future studies need to elucidate the specific role of these amino acids and their optimal combination in perioperative nutritional support.

HMB improves muscle measurements and physical function and is thought to promote protein synthesis and inhibit protein catabolism in vivo through complex mechanisms (45, 46). Recently, clinical emphasis has been placed on the importance of adequate nutritional support combined with rehabilitation in muscle protein maintenance and synthesis (47). The included studies support this view. For example, the results of HMB improvement in ASMM were pooled from four RCTs (20, 27, 29, 31), all of which reflected an emphasis on adequate nutritional support (20, 29, 31) and the implementation of early rehabilitation (20, 31). Similarly, HMB improved 6-MWD by pooling results from two studies describing detailed nutrition programs and early rehabilitation (20, 29).

Our study has some limitations. First, most included RCTs were small-sample, open-label studies with potentially high selection bias. Second, only a few studies provided data for pooling in the secondary outcomes, which limited our implementation of subgroup analyses of these outcomes. Therefore, interpretation of these outcomes requires caution. Third, although there was no statistically significant heterogeneity in the primary outcomes, some potential clinical heterogeneity remained unresolved. For example, there were variations in the gender distribution and types of surgery among study participants. These differ substantially in physiological stress, postoperative recovery patterns, nutritional risk, and potential for muscle catabolism. Meanwhile, the included study’s definition of complications varied substantially, which depended on the surgical procedures. Fourth, since nearly half of the included RCTs administered HMB as a single supplement, more future studies should clarify the independent influencing role of HMB. Fifth, exploring the ideal timing (or regimen) of HMB preoperative supplementation remains unclear due to the limited availability of data from the included studies. Sixth, most studies provided results at different time points for assessments. Since many of the outcomes assessed (e.g., physical function, nutritional markers) are time-sensitive, these inconsistencies in measurement timing may influence the pooled effect estimates. Finally, some postoperative patients had ICU admissions. However, we could not evaluate the efficacy of these ICU patients separately due to insufficient data.