For just over a decade, β-hydroxy-β-methylbutyrate (HMB) has been studied for its potential as an anticatabolic lipolytic dietary supplement assisting gains in strength, muscle size, recovery, fat oxidation, and fat loss in humans when used in conjunction with a resistance training program. With some evidential support, HMB has become one of the more popular nutritional supplements taken by strength and body building athletes looking to improve strength and muscle size (12,16).
Several mechanisms of action of HMB have been proposed, largely on the basis of extrapolation from early research on animal carcasses and deduction from indirect measures in humans (17,27). Supplementation with HMB may attenuate training-induced proteolysis in the muscle via downregulation of proteolytic pathways (15,17,27,30) and may also stimulate protein synthesis similarly to leucine (30). In the sarcoplasm, HMB is thought to be metabolized to β-hydroxy-β-methylglutaryl-CoA, providing a readily available carbon source for cholesterol synthesis, which in turn provides material for muscle cell growth (17,27). Also, HMB may undergo polymerization and be used as a structural component of the cell membrane, leading to enhanced stability (27). Additionally, HMB has been proposed to increase muscle cell fatty-acid oxidation capacity via an unknown mechanism or mechanisms, leading to decreases in fat mass (27,34).
Despite largely indirect evidence to support a mechanism, several trials of the proposed ergogenic effects of HMB on strength, body composition, and markers of muscle damage have been conducted in humans in conjunction with resistance training. Researchers have found both gains (11,17,21,33) and no change in strength (13,14,19,28,35) with HMB supplementation. These equivocal strength results may be related to study-specific characteristics, such as resistance training experience of the subjects, the HMB daily dose, the intervention duration, or the intensity, form, and volume of the accompanying exercise intervention. Additionally, differential and inconsistent gains in upper- vs. lower-body lift performances have been reported, with the majority of improvements being in lower-body strength (6,17,33,35). This apparent selectivity indicates regional differences in the scope for adaptation that may have a physiological basis or that may be linked to previous training frequency or the relative intensity of training on various muscle groups.
In some studies, a pattern for reduction in markers of muscle damage with HMB supplementation has been observed (11,13,14,17,21), but the magnitudes of the outcomes are variable. Also equivocal is the effect of HMB supplementation on body composition, with 3 groups reporting increases in fat-free body mass (6,17,35) but the remainder showing no clear effect (6,11,13,14,17,18,21,24,28).
A previous meta-analysis was carried out by Nissen and Sharp (16) on a range of supplements popular with serious resistance training athletes that included HMB. The effects on strength and lean mass gains in conjunction with resistance training were quantified. For the effect of HMB, these authors have reported statistically significant effect sizes of 0.19 for strength (95% confidence limit [CL]: ±0.10) and 0.15 for fat-free mass gains (95% CL: ±0.09), but the trivial magnitude of these effects was not emphasized. Additionally, this meta-analysis did not account for the preintervention training status of the subjects, did not include measures of muscle damage, and has previously been criticized for collaboration of authors and institutes involved in several of the studies and the potential for publication bias because of the small number of studies included (4). Furthermore, since this review, an additional 4 studies have been published, warranting the present random effects meta-analysis. We have included training status, differentiated between upper- and lower-body strength outcomes, assessed the effect on muscle damage, and dealt with the study characteristics (as far as the quality of available information permits) and the nature and magnitude of the effects in more detail than the previous meta-analysis.
A literature search was conducted up to June 2007 on Medline (via Ovid), Science Citation Index (via Web of Science), Cambridge Scientific Abstracts, and TRIP+. Search terms included HMB, beta hydroxy beta methylbutyrate, hydroxy methylbutyrate, and sport supplements, and these search terms were then combined with supplement, resistance training, resistance exercise, weight training, and body composition. Literature inclusion criteria were as follows: full papers published in peer-reviewed journals in English, limited to supplementation intervention studies in humans in conjunction with resistance training. Eleven studies were found to fit the criteria (Table 1) (6,11,13,14,17,18,21,24,28,33,35). All studies were randomized, parallel-group, placebo-controlled trials, except for Ransone et al. (24), which was a randomized crossover with 1-week washout between conditions and post-only comparisons, which we normalized to the prestudy score. If insufficient data were provided in the publication for calculation of the mean effect and sampling variance (16,24), the corresponding authors were contacted and asked if they were willing to provide additional data. Nine studies had sufficient data to be included in effect size calculations for strength (6,11,14,17,18,21,24,28,33) and body composition (6,11,13,14,17,21,24,28,33), respectively. Of the studies found, sufficient comparisons (7) were available for analysis of the overall effect of HMB on the skeletal muscle membrane damage marker creatine kinase (CK) (11,13,14,17,21). Two datasets were excluded from the meta-analysis because they represented sampling outliers relative to the majority of studies published to date: the dataset of Vukovich et al. (35), because the average age of the sample was 70 years; and the group of women from the study of Panton et al. (21). Three studies investigated the effects of 2 doses of HMB (6,14,17), and 1 study compared (relative to control) the effects of standard-release vs. time-delay release HMB capsules (28). Together, a total of 14 comparisons, each with 3 strength outcomes (see below), were available for meta-analysis with samples from the population of previously resistance trained and untrained young adult men.
Experimental Approach to the Problem
The effect of HMB on strength was measured by performance in weight lifting. Within-study overall lift performance (overall average strength) was first compiled by averaging in absolute units (kg) all reported pretreatment strength measures and comparing these values with the total average posttreatment strength measures; SDs were generated from the root of the sum of the lift or exercise SD squared. Strength outcomes were further compiled and categorized into the pure average upper-body and pure average lower-body lift performances (Liftcode). Performance outcomes in composite lifts that included both upper- and lower-body components (e.g., power clean, hang clean) and measures of abdominal strength (e.g., total number of sit-ups to failure) were excluded from the pure upper-body and lower-body datasets but included in the overall average strength dataset. The effect of HMB on body composition was determined by meta-analysis of fat-free mass and body fat mass provided by each study, irrespective of measurement method.
We included 1 study with a dietary cointervention in which the HMB intervention group was also provided with a supplement containing whey protein-carbohydrate, vitamins, minerals, glutamine, and chromium picolinate (17). Fifty milligrams of potassium phosphate per capsule was also added to the HMB supplement in 2 studies (21,35), and glucose, taurine, disodium phosphate, and potassium phosphate were included in the intervention and control supplements in another (13); however, because of the difficulty in qualifying the physiological effects of these cointerventions, their influence was ignored from the meta-analysis.
The main outcome from this meta-analysis is a weighted mean of values of the outcome statistic from the various study comparisons, where the weighting factor is the comparison total sample size divided by the average sample size for all comparisons within a particular category of Liftcode and training experience. For a controlled trial, the comparison total sample size is 4n 1 n 2/n 1 + n 2, and for the crossover (24) it is 2n. Use of the inverse sampling standard error of the statistic derived from either the confidence interval (CI) or p value of the outcome statistic or from SDs of change scores as the weighting factor, as in standard random effects meta-analysis, was not possible, because too many studies presented a p value inequality (p > 0.05 or p ≤ 0.05) or insufficient inferential information to permit a comparison analysis. To exclude all these studies from the meta-analyses would have resulted in unacceptable bias, akin to the publication bias that arises from failure of authors to submit studies or outcomes with nonsignificant outcomes or failure of journal editors to accept them. The meta-analytic outcome from the current sample-size weighting method, nevertheless, is equivalent to that produced from the standard error method if it is assumed that the outcome has the same error of measurement in all studies (W. Hopkins, personal communication, 2007).
The meta-analyses were performed with a program for the mixed modeling procedure (PROC MIXED) in SAS (version 9.1; SAS Institute, Cary, NC) based on the method of Snowling and Hopkins (31). Exercise training experience (trained or untrained in resistance exercise before the study intervention) was the most important main effect in the fixed effects model. For strength outcome measures, the second fixed effect Liftcode and the interaction with training experience was added to determine the effect of HMB on upper- and lower-body strength by training experience. Four moderating covariates were added to the fixed effect model representing study intervention characteristics that were included in most studies and that might be expected based on physiological grounds to influence the effect of the supplement on strength and body composition outcomes (Table 1). The HMB daily dose (range 1.5-6.0 mg·d−1) and intervention duration (range 3-9 weeks) were included as numeric effects. Training load was included as a numeric effect and was quantified as the sum of session volume (h·wk−1), with training load having an integer value of 1 (average of 100-200 total repetitions per session) through 6 (average of 600 or more repetitions per session) and average session intensity values of 1 (low; average lift intensity <50-60% 1RM), 2 (moderate; average 50-80% 1RM), or 3 (high; >80% 1RM). Finally, dietary cointervention (16) was included as a binary variable and was interacted with training experience. In 6 studies, diet was recorded by dietary recall or diary methods (6,11,13,14,17,33). The postintervention minus preintervention change in average daily energy and the protein, carbohydrate, and fat macronutrient quantities were calculated and expressed as a factorial change relative to the preintervention value. Owing to insufficient sampling and between-study variation in change in reported average energy and macronutrient intake, the full model with these 2 variables would not run, leaving the final model to include all but these possible modifying effects.
After inclusion of the 4 moderating effects of study characteristics, the remaining unexplained true variation within (strength and CK outcomes) and between (all outcomes) studies was estimated as 1 or more random effect. In the analysis of strength and CK, the random effect estimates were of the pure between-study variation in the effect of treatment (free of sampling variation attributable to error of measurement in the dependent variable giving rise to the meta-analyzed effect) and the within-study SD representing the residual random effect. The latter represents the standard error of a study estimate with the mean sample size (n) of the meta-analyzed estimates; this standard error was multiplied by √(n / 8) to provide an estimate of the mean standard error of measurement of the dependent variable (W. Hopkins, personal communication, 2007). For body composition, there were insufficient study estimates to include a pure between-study random effect; the residual random effect is, therefore, shown as the between-study SD.
The HMB effect in each study was converted to a factor relative to control derived from factor effect = (HMBpost/HMBpre)/(controlpost/controlpre). For presentation (Table 1), the factor was converted to a percentage by (factor effect − 1) × 100. We meta-analyzed the log-transformed factor for estimation of the mean effects. Estimated means were back-transformed to represent percent effects.
Back-transformed meta-analyzed percent effects were also expressed as standardized (Cohen) effects (2) by dividing by the average baseline between-subject SD (derived as the square root of the unweighted mean of variances) expressed as a percentage of the mean baseline score for the HMB and control groups. Magnitudes of the standardized effects were interpreted using thresholds of 0.2, 0.6, and 1.2 for small, moderate, and large effects, respectively, a modification of Cohen thresholds of 0.2, 0.5, and 0.8 (2); the modifications are based primarily on congruence with Cohen thresholds for correlation coefficients (8). In keeping with recent trends in inferential statistics (32), magnitude-based inferences about true (population) values of effects were made by expressing the uncertainty in the effects as the back-transformed 90% CLs. The less conservative 90% limits were selected in this report because the relatively low number of investigations will inherently increase the uncertainty of the true population estimate, making a higher CL unnecessarily conservative (3,32). To further qualify outcomes, an effect was deemed unclear if its CI overlapped the threshold for the smallest standardized effect (i.e., if the chances of the standardized effect being substantially greater than and less than 0.2 were both >5%); otherwise, the magnitude of the effect was reported as the magnitude of its observed value (1).
Three of the 10 publications included in the meta-analysis (Table 1) provided 2 outcomes, and 1 publication provided 3 outcomes, giving 8 estimates for previously trained subjects and 6 estimates for subjects previously untrained in gymnasium-based resistance exercise before intervention on the effects of HMB on strength, body composition, and muscle damage. The total subject pool was 394, comprising 135 untrained and 259 trained participants. The majority of studies used a daily HMB dose of 3 mg·d−1, with 1 study testing 1.5 mg·d−1 in untrained and 6 mg·d−1 in trained subjects, and another comparing 3 vs. 6 mg·d−1 in untrained subjects. Mean and between-study SDs for the meta-analysis were age 23 ± 2 years; duration 5 ± 2 weeks; training volume 5 ± 6 h·wk−1 overall, but 3 ± 0 h·wk−1 in untrained and 6 ± 7 h·wk−1 in trained subjects; training load rating of 9 ± 2 overall, 8 ± 1 for untrained subjects, and 8 ± 2 for trained subjects; and, finally, average study sample sizes of 26 ± 14 overall, and 22 ± 10 and 28 ± 16 for untrained and trained subjects, respectively.
Three studies reported no dropouts (11,14,28). Two studies by the same group (17) had 2 and 4 dropouts, respectively, which were unrelated to the intervention, and incomplete data for 1 subject in 1 outcome measure (upper-body strength) because of injury. One study had 3 dropouts, which were unrelated to the intervention, and 1 subject was dropped because of abnormal results for an outcome not of interest to the present meta-analysis (21). Another study had 7 dropouts unrelated to the intervention, and 2 subjects were dropped for failure to comply with the study protocol (6). A further study reported 10 dropouts unrelated to the intervention; 2 subjects were dropped for failure to comply with the study protocol, and 3 subjects provided incomplete outcome measures (2 incomplete upper-body strength outcomes, 1 incomplete body composition outcome) because of injury (33). Training was supervised in most studies in the meta-analysis (6); for those studies in which exercise compliance was reported (11,17,21,24,28,33), it was high (range 87.5-100%).
Effect of β-Hydroxy-β-Methylbutyrate on Strength, Body Composition, and Muscle Damage
Funnel plots of the inverse of an estimate's weighting factor vs. the estimate's effect for the main outcomes grouped by training experience (not shown) identified (qualitatively) greater scedasticity in the untrained lifters and nonuniform spread in both groups of lifters, confirming the appropriateness of log-transformation of the factor effects. Between-study strength data were well represented, with the more powerful studies (greater sample size) showing less variation. There was no evidence for publication bias or outliers, as illustrated by a relatively uniform spread of outcomes between studies and no outcome greater than approximately 4 SDs from the center of the scatter.
The normalized effect of HMB supplementation on strength, body composition, and CK for each reporting study in the meta-analysis is shown in Table 2, along with the meta-analyzed mean effects and estimate precision. Figure 1 shows the meta-analyzed mean values and differences between the 2 levels of training experience for strength outcomes, all expressed in standardized units with an interpretation of effect size magnitudes, along with the random effects. Finally, Figure 2 shows the standardized effects of HMB on body composition measures and random effects.
In untrained lifters, there were clear-cut, small, beneficial increases in lower-body and overall average strength measures in response to HMB supplementation, but upper-body strength gains were negligible. In trained lifters, strength was not affected by HMB. Compared with the previous meta-analysis (20), the effects of HMB on overall average strength outcomes were trivial. There were clear, small gains in lower-body strength compared with the effect on upper-body strength in untrained lifters. There were small, clear gains in lower-body and overall average strength in untrained compared with trained lifters, but the effect of training experience on upper-body strength was inconclusive (Figure 1). For all studies combined, the standardized gain in lower-body strength was 0.44 more (±0.24; small) relative to upper-body strength in untrained lifters, but it was only 0.08 (±0.15; trivial) in trained lifters. The mean standard errors of measurement for strength outcomes are 4.3 and 6.2% in untrained and trained lifters, respectively.
Fat-free mass increases and changes in fat mass in trained and untrained lifters were negligible. The standardized effect of HMB on CK was −0.41 (±0.69, unclear). The mean standard error of measurement for CK was 6.9%.
Moderating Effects of Study Characteristics
Training load and HMB daily dose had neutral trivial influences on the effect of HMB on strength outcomes (Table 3). There was a reducing influence of increasing intervention duration on the magnitude of the HMB effect on strength, but the uncertainty allows this effect to be a trivial or small, enhancing effect. Dietary cointervention had a small, increasing persuasion on strength in trained lifters, but the uncertainty allows this effect to be a trivial to moderate, decreasing or increasing influence.
The influences of training load, HMB daily dose, and study intervention duration on fat-free and fat mass were negligible. Dietary cointervention was of increasing influence on both fat and fat-free mass outcomes, but the effects were either trivial in magnitude or assessed as unclear because of the uncertainty of the estimate caused by the limited number of studies with dietary cointerventions. Increasing the daily HMB dose reduced the effects of HMB on CK.
The unexplained variation between and within studies is illustrated by the standardized random effect. The between-study random effect in untrained lifters was small, but other between- and within-study effects were negligible (Figure 1). For fat and fat-free mass, the random effects were all negligible (Figure 2), showing that the meta-analytic model adequately accounted for the between-study variation in the effects of HMB on body composition. The between- and within-study standardized random effects for CK were unclear at 0.36 (90% CI: ±54) and 0.06 (±0.64), respectively.
The major findings of this meta-analysis are that HMB supplementation results in a small, beneficial increase to overall strength in untrained lifters but has a negligible effect on trained lifters. Furthermore, in untrained lifters, HMB results in a small to possibly moderate increase in lower-body strength, but it has only a negligible effect on upper-body strength. In contrast, all strength outcomes are insignificant in trained lifters. In both trained and untrained lifters, the effect of HMB supplementation on body composition is negligible.
Our estimates for the magnitudes of the meta-analyzed effects are based on a generic statistical approach using mean effects standardized against the composite average between-subject SD of subjects at baseline to obtain the effect size (25,26). An assessment of magnitude directly related to performance outcomes for trained competitive weight lifters, lift performance, and aesthetic outcomes in trained body builders and related to functional strength and health benefits in previously untrained persons would require a meta-analysis of controlled trials of the effects of HMB on these outcomes directly; as far as we know, there are no such studies. However, models derived from analysis of the variation of competitive performance and the gap between finishers reveal that the smallest change in performance likely to increase the chances of winning is between 0.3 and 0.7 times the coefficient of variation for the performance (9). We were able to find 1 publication reporting a coefficient of variation of 2.4% for weight lifting performance, which is of similar magnitude to competitive performance in other sports (10,22). The value of 1.2% is, therefore, an appropriate midrange estimate for the smallest worthwhile enhancement of weight lifting performance, and 1RM is likely to be representative of competitive performance; the meta-analytic outcomes of 0.1-1.3% for trained lifters (Table 2) show that the effect of HMB on competitive performance is indeed likely to be of trivial magnitude, although for competitive lifting performance the uncertainty allows for substantial benefits or enhancements. Interpretation of the meta-analytic standardized effect provides a comparison of the magnitude of the effect against the recognized smallest effective shift in the mean relative to the population's normal variation (2,7,26). The meta-analytic standardized gain in overall average strength for all studies combined of 0.19 (90% CL: ±0.13) (Figure 2) is comparable with the composite estimate for the magnitude of the HMB effect in the meta-analysis by Nissen and Sharp (16) of 0.19 (±0.1), although it should be noted that the authors included the 1 study in women (21) and the 1 geriatric study (35), which were not included in the present meta-analysis. Nevertheless, Nissen and Sharp (16) conclude that the outcome justified the use of HMB as a supplement to support strength and lean mass gains during resistance training, presumably based on the statistical significance for the effect (p < 0.01). However, the outcomes from both meta-analyses show that, in fact, the overall effect of HMB is likely to be trivial, although the uncertainty allows for a small, beneficial effect. Nissen and Sharp (16) also report that in comparison with other supplements commonly used by resistance trainers, such as creatine, protein, chromium, dehydroepiandrosterone, and androstenedione, only creatine supplementation offered strength gains (0.36 ± 0.11) above the smallest effective threshold.
The current analysis suggests that HMB supplementation may produce a greater response in untrained lifters, relative to trained lifters, which could be related to less potential for adaptive strength responses to resistance training. The proposed mechanisms of HMB supplementation on strength gain are purported to be via inhibiting protein breakdown, and attenuation and/or enhanced repair of exercise induced muscle-cell membrane damage (15). With regular training, protein turnover (23) and resistance exercise-induced muscle damage are attenuated (5), which could be why highly trained lifters have not been found to respond to HMB supplementation to the same extent as previously untrained lifters. However, evidence in support of these hypotheses would be difficult to establish because the prestudy training status of trained participants was not qualified sufficiently in some studies to include the covariate in the meta-analysis. Within the scope of the definition of trained lifters, well-trained strength athletes, well-trained athletes using resistance exercise as a component of their sport-specific training, and moderately trained recreational lifters were included. On the other hand, untrained lifters included individuals who had not participated in resistance exercise during the previous 3 months or more (17). Classification of training status is a defined limitation of the present meta-analysis, and it is evident by the increased uncertainty of the untrained minus trained estimates for upper- and lower-body strength (Figure 1) that our polar definitions do not entirely capture the training-status-dependent effect of HMB on strength outcomes; only further studies with well-defined training status can resolve the uncertainty. The quality and quantity of the resistance training program and the addition of other or concurrent training during the HMB supplementation period are other potentially important confounding variables. In developing the meta-analytical model, we carefully evaluated the validity of quantifying the efficacy of the resistance training program employed in each study. Unfortunately, the nature of the reported training programs was variable, ranging from no detail (14), to vague detail (18,28), to clearer specifics (17), but, on the whole, the programs seemed relatively consistent with common and recommended practice, including repetitions to failure and progressive adaptation. The second problem we faced in coming up with a valid quantifier was that it is unknown what the optimal resistance training program is that would generate the greatest strength/hypertrophy gains with HMB supplementation; the question has not been investigated, and any attempt might have decreased rather than increased certainty in the outcome. Likewise, in 4 of the 9 comparisons included within the model for trained lifters (13,18,24,28), in addition to the prescribed resistance exercise, inadequately specified other training was concurrently performed during the supplementation period (e.g., agility and skills exercises, aerobic and anaerobic training). Consequently, we abandoned attempting to include training program quality and the influence of concurrent training in the meta-analytic model, favoring an evaluation of total resistance training load as the primary training affector (Tables 1 and 3). Nevertheless, training program quality and the presence of concurrent training may impact the efficacy of the supplement, but at this point in time there are insufficient data in the body of work available to objectively evaluate this contention, leaving us to recommend it as a factor to be considered in future work.
More research is also needed to determine why small strength benefits occur in untrained but not trained lifters, and why the effects are clearly apparent in the legs but not in the upper body. At this point in time, there is no clear mechanism to explain the differential training status effects and upper/lower-body outcomes. Muscle fiber-type differences between leg and arm muscles, the relative extent of recruitment and loading, and a differential relative responsiveness to HMB in type I compared with type II fibers may be factors (20). Another explanation is that upper-body muscle groups are at a higher level of conditioning relative to their biological maxima before the study and are, therefore, less susceptible to facilitated adaptation by HMB (33).
Between-study variation in diet is 1 possible confounding variable that we were unable to assess in the present meta-analysis because of insufficient information provided across studies with respect to dietary type and composition to validate an attempt. With the exception of the 3 comparisons that included dietary cointerventions (Table 1), diet was normal and standardized in all studies but was not strictly controlled. The outcomes of the present meta-analytic HMB effects, therefore, represent an estimate of the magnitude of gain in the population subsets in the presence of the normal dietary variation. Some uncertainty of the estimate and the unexplained effect (Figure 1) are likely associated with this variation, and further studies are required to determine whether HMB effects on strength and lean tissue accretion are coupled to nutrient availability or deficit under tissue stress (i.e., imposed through resistance training), as has been proposed (15,17). Alternatively, to account for an interaction with dietary protein-as the most important macronutrient in hypertrophy-a large study with clearly defined intakes of dietary protein coupled with HMB supplementation could be definitive.
With respect to body composition measures, the current meta-analytic estimates for the HMB effect on fat-free mass and fat mass were negligible in both trained and untrained lifters and were consistent in magnitude with the random effect (Table 2, Figure 2). A similar outcome was provided by the previous meta-analysis (16), with a standardized effect on fat-free mass of 0.15 (95% CL: ±0.09). It has been proposed that HMB reduces exercise-induced muscle protein catabolism and stimulates protein synthesis (30), promoting gains in fat-free mass, and it is linked to fat metabolism in skeletal muscles (29,34), possibly leading to reduced body fat. Although there were mean increases in fat-free mass (Figure 2), the magnitude suggests that the true effect of HMB is trivial; the meta-analyses also do not support a mechanism relating to body fat loss. Nevertheless, these findings are limited by the 3- to 9-week intervention duration of the body work analyzed; it is possible that worthwhile gains of lean mass (i.e., effect size >0.2) may be attainable with HMB over a longer time frame, and further well-controlled long-duration studies are warranted. Nevertheless, it is important to point out that it is difficult to control physical activity and diet and to retain subjects for durations much longer than 8-12 weeks in training, diet, or supplementation studies. The actions of most nutritional supplements are on the steepest part of the kinetic curve during the first few weeks of application and taper off toward their effective biological asymptote. There is little reason for HMB not to induce a similar kinetic pattern, making the present outcomes during the 4- to 9-week time frame indicative of the magnitude of the most likely gains.
On the basis of the biochemical pathway for HMB metabolism, it has been proposed that HMB may act either by decreasing cellular protein breakdown or by providing structural precursors for membrane cholesterol synthesis and, thereby, affecting cell membrane integrity (15). The HMB had a small, reducing effect on mean CK concentrations, supporting a mechanism, but the uncertainty suggests that moderate reductions or even small, detrimental effects on muscle damage are possible. With respect to moderating factors impacting on CK, there was some evidence for declining benefit with increasing HMB dose.
To conclude, within the context, scope, and limitations of the present body of published work, there were sufficient studies to allow us confirm that 3-9 weeks of HMB supplementation during resistance training in previously untrained subjects incurs small but clear lower-body (leg) and overall strength gains, but the effects on upper-body lifting performance were less and were trivial in magnitude. In contrast, the HMB effect on strength gains in trained lifters was of trivial magnitude. The small lower-body and overall average strength gains in untrained lifters seem unrelated to effects of HMB on body composition, which were negligible, or to muscle damage estimates (via CK), which remain inconclusive. Further research is required to quantify the effectiveness of HMB over a longer time frame (e.g., 24-52 weeks) and to elucidate possible interactions with training experience and nature, and with diet.
For previously untrained individuals starting a resistance training program, small gains in lower-body and overall strength with HMB supplementation may be expected during the first 1-2 months. However, the absence of even small benefits to lean mass gain and body fat reduction in both untrained and trained lifters, and the trivial effect on strength in well-trained lifters, suggest that, over the short term, this supplement is of negligible worth to the majority of experienced athletes performing resistance exercise as part of their training programs. The effectiveness of HMB supplementation over the longer term is unknown.
Examples of SAS code for the meta-analytic random effects model and advice on random effects modeling were kindly provided by Prof. W.G. Hopkins, AUT University, Auckland, New Zealand.
1. Batterham, AM and Hopkins, WG. Making meaningful inferences about magnitudes. Int J Sport Physiol Perf
1: 50-57, 2006.
2. Cohen, J. Statistical Power Analysis for the Behavioral Sciences
(2nd ed). Hillsdale, NJ: Lawrence Erlbaum, 1988.
3. Curran-Everett, D and Benos, DJ. Guidelines for reporting statistics in journals published by the American Physiological Society. Physiol Genomics
18: 249-251, 2004.
4. Decombaz, J, Bury, A, Hager, C, Nissen, S, and Sharp, R. HMB meta-analysis and the clustering of data sources. J Appl Physiol
95: 2180-2182, 2003.
5. Faulkner, JA, Brooks, SV, and Opiteck, JA. Injury to skeletal muscle fibers during contractions: conditions of occurrence and prevention. Phys Ther
73: 911-921, 1993.
6. Gallagher, PM, Carrithers, JA, Godard, MP, Schulze, KE, and Trappe, SW. β-Hydroxy-β-methylbutyrate ingestion, part I: effects on strength and fat free mass. Med Sci Sports Exerc
32: 2109-2115, 2000.
7. Glass, G. Integrating findings: the meta-analysis of research. Rev Res Educ
5: 351-379, 1977.
9. Hopkins, WG, Hawley, JA, and Burke, LM. Design and analysis of research on sport performance enhancement. Med Sci Sports Exerc
31: 472-485, 1999.
10. Hopkins, WG, Schabort, EJ, and Hawley, JA. Reliability of power in physical performance tests. Sports Med
31: 211-234, 2001.
11. Jówko, E, Ostaszewski, P, Jank, M, Sacharuk, J, Zieniewicz, A, Wilczak, J, and Nissen, S. Creatine and β-hydroxy-β-methylbutyrate (HMB) additively increase lean body mass and muscle strength during a weight-training program. Nutrition
17: 558-566, 2001.
12. Kreider, RB. Dietary supplements and the promotion of muscle growth with resistance exercise. Sports Med
27: 97-110, 1999.
13. Kreider, RB, Ferreira, M, Greenwood, M, Willson, M, Grindstaff, P, Plisk, J, Reinardy, J, Cantler, E, and Almada, AL. Effects of calcium β-HMB supplementation during training on markers of catabolism, body composition, strength and sprint performance. J Exerc Physiol
3: 48-59, 2000.
14. Kreider, RB, Ferreira, M, Wilson, M, and Almada, AL. Effects of calcium β-hydroxy-β-methylbutyrate (HMB) supplementation during resistance-training on markers of catabolism, body composition and strength. Int J Sports Med
20: 503-509, 1999.
15. Nissen, S, and Abumrad, N. Nutritional role of the leucine metabolite β-hydroxy β-methylbutyrate (HMB). J Nutr Biochem
8: 300-311, 1997.
16. Nissen, SL and Sharp, RL. Effect of dietary supplements on lean mass and strength gains with resistance exercise: a meta-analysis. J Appl Physiol
94: 651-659, 2003.
17. Nissen, S, Sharp, R, Ray, M, Rathmacher, JA, Rice, D, Fuller, JC, Connelly, AS, and Abumrad, N. Effect of leucine metabolite β-hydroxy-β-methylbutyrate on muscle metabolism during resistance-exercise training. J Appl Physiol
81: 2095-2104, 1996.
18. O'Connor, DM, and Crowe, MJ. Effects of six weeks of β-hydroxy-β-methylbutyrate (HMB) and HMB/creatine supplementation on strength, power, and anthropometry of highly trained athletes. J Strength Cond Res
21: 419-423, 2007.
19. O'Connor, PM, Kimball, SR, Suryawan, A, Bush, JA, Nguyen, HV, Jefferson, LS, and Davis, TA. Regulation of translation initiation by insulin and amino acids in skeletal muscle of neonatal pigs. Am J Physiol
285: E40-E53, 2003.
20. Ostaszewski, P, Kostuik, S, Balasinska, B, Jank, M, Papet, I, and Glomot, F. The leucine metabolite β-hydroxy-β-methylbutyrate (HMB) modifies protein turnover in muscles of laboratory rats and domestic chickens in vitro. J Anim Physiol Anim Nutr (Berl)
84: 1-8, 2000.
21. Panton, LB, Rathmacher, JA, Baier, S, and Nissen, S. Nutritional supplementation of the leucine metabolite β-hydroxy-β-methylbutyrate (HMB) during resistance training. Nutrition
16: 734-739, 2000.
22. Paton, CD and Hopkins, WG. Seasonal changes in power of competitive cyclists: implications for monitoring performance. J Sci Med Sport
8: 375-381, 2005.
23. Phillips, SM, Tipton, KD, Ferrando, AA, and Wolfe, RR. Resistance training reduces the acute exercise-induced increase in muscle protein turnover. Am J Physiol
276: E118-E124, 1999.
24. Ransone, J, Neighbors, K, Lefavi, R, and Chromiak, J. The effect of β-hydroxy β-methylbutyrate on muscular strength and body composition in collegiate football players. J Strength Cond Res
17: 34-39, 2003.
25. Rhea, MR. Synthesizing strength and conditioning research: the meta-analysis. J Strength Cond Res
18: 921-923, 2004.
26. Rosenthal, R and Dimatteo, MR. Meta-analysis: recent developments in quantitative methods for literature reviews. Ann Rev Psychol
52: 59, 2001.
27. Slater, GJ and Jenkins, D. β-Hydroxy-β-methylbutyrate (HMB) supplementation and the promotion of muscle growth and strength. Sports Med
30: 105-116, 2000.
28. Slater, G, Jenkins, D, Logan, P, Lee, H, Vukovich, M, Rathmacher, JA, and Hahn, AG. β-Hydroxy-β-methylbutyrate (HMB) supplementation does not affect changes in strength or body composition during resistance training in trained men. Int J Sport Nutr Exerc Metab
11: 384-396, 2001.
29. Slater, GJ, Logan, PA, Boston, T, Gore, CJ, Stenhouse, A, and Hahn, AG. β-Hydroxy β-methylbutyrate (HMB) supplementation does not influence the urinary testosterone: epitestosterone ratio in healthy males. J Sci Med Sport
3: 79-83, 2000.
30. Smith, HJ, Mukerji, P, and Tisdale, MJ. Attenuation of proteasome-induced proteolysis in skeletal muscle by β-hydroxy-β-methylbutyrate in cancer-induced muscle loss. Cancer Res
65: 277-283, 2005.
31. Snowling, NJ and Hopkins, WG. Effects of different modes of exercise training on glucose control and risk factors for complications in type 2 diabetic patients. Diabetes Care
29: 2518-2527, 2006.
32. Sterne, JAC and Smith, GD. Sifting the evidence-what's wrong with significance tests? BMJ
322: 226-261, 2001.
33. Thomson, JS, Watson, PE, and Rowlands, DS. Effects of nine-weeks B-hydroxy-B-methylbutyrate supplementation on strength and body composition in resistance trained men. J Strength Cond Res
34. Van Koevering, MT, Dolezal, HG, Gill, DR, Owens, DR, Strasia, FN, Buchanan, CA, Lake, RDS, and Nissen, S. Effects of β-hydroxy-β-methyl butyrate on performance and carcass quality of feedlot steers. J Anim Sci
72: 1927-1935, 1994.
35. Vukovich, MD, Stubbs, NB, and Bohlken, RM. Body composition in 70-year-old adults responds to dietary β-hydroxy-β-methylbutyrate similarly to that of young adults. J Nutr
131: 2049-2052, 2001.