Protein intake and changes in whole-body lean/fat-free mass, measures of muscle size or mass, strength, and physical function in healthy adults: An updated systematic review and Bayesian model averaged dose-response meta-analysis

Background: Previous dose-response meta-analyses of protein intake have generally assumed linear relationships or imposed a fixed "breakpoint", rendering inferences conditional on a particular functional form. We therefore estimated the dose-response associations between protein intake and changes in whole-body lean/fat-free mass, measures of muscle size or mass, strength, and physical function in healthy adults, with and without resistance training (RT), while accounting for uncertainty about the shape of these associations. Methods: We performed an updated systematic review of parallel-group randomised controlled trials that assigned participants to dietary or supplemental protein interventions, placebo, or control conditions. We searched MEDLINE, Embase, CINAHL, and Web of Science Core Collection through 14 September 2026 and assessed risk of bias with Cochrane RoB 2. Standardised mean changes were synthesised using hierarchical arm-based Bayesian meta-regression. For each outcome, we fit 26 candidate models representing no association, linear relationships, linear splines with a knot at 1.6 g/kg/day, and nonlinear smooth functions. Models were then averaged using leave-one-out cross-validation stacking weights. Pairwise contrasts between protein intakes were evaluated against a preregistered smallest effect size of interest of ±0.1 standardised units, using a 95% posterior-probability criterion. We additionally fit a range of sensitivity analyses including Mundlak regression, strict within-study contrasts, common sample models of achieved- or assignment-derived protein intake, errors-in-variables models, residual-correlation models, and intervention strategy (addition or energy matched substitution). Finally, primary dose-response models examined body-mass relative total protein intake (g/kg/day) and thus we also examined models with protein as absolute total intake and absolute supplementation (both in g/day). Results: We identified 119 eligible studies, of which 95 included sufficient data to contribute 254 study arms, 4,931 participants, and 1,864 eligible baseline-to-follow-up contrasts to the primary analyses. Primary dose-response functions across outcomes generally showed modest positive associations, particularly with RT. However, no supported contrast between a current and proposed protein intake resulted in a conclusively meaningful gain in outcome from increasing intake or a meaningful loss from decreasing it. Increases or decreases of approximately 0.2 to 0.5 g/kg/day from a current intake were generally supported as practically equivalent. Contrasts between more widely separated intakes were uncertain rather than demonstrably meaningful or equivalent. We could not identify a clearly optimal intake. Inference was robust across the completed sensitivity families, although stricter within-study analyses produced dose-response functions of attenuated magnitude. The absolute total intake and absolute supplementation dose analyses did not materially alter the overall interpretation based on relative total intake. Most trials raised some concerns or were at high risk of bias. Publication-bias adjustment attenuated outcome-level estimates towards zero, and certainty was low or very low for all GRADE-assessed contrasts. Conclusions: Protein intake showed a modest positive association with outcomes and so those wishing to maximise outcomes may see it as reasonable to aim for higher protein intakes. However, the estimates were not sufficiently precise to establish that the gain from increasing intake, or the loss from decreasing it, exceeds a practically meaningful threshold. Comparisons between more widely separated intakes, and the extent to which the observed associations represent causal dose effects, remain uncertain particularly as data are sparse at low and high intakes (2.0% of primary-analysis arms were below 0.8 g/kg/day and 8.3% were above 2.2 g/kg/day). As such, whilst our results do not establish adequacy of the RDA, we suggest that practical recommendations for changes to intakes from typical habitual values should be based upon considerations other than just their effectiveness for the outcomes examined here (e.g., cost, appetite and food preferences, energy restriction). Lastly, we suggest that future research should make concerted attempts to address the shortcomings of the current body of literature. Funding: Kieser Australia provided in-kind support through consultancy and employment. Registration: Open Science Framework, w2qpx.

Authors

Institutions

Publication Details

Published
2026-11-17
DOI
https://doi.org/10.51224/sportrxiv.1166
Primary Topic
Muscle metabolism and nutrition
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Protein intake and changes in whole-body lean/fat-free mass, measures of muscle size or mass, strength, and physical function in healthy adults: An updated systematic review and Bayesian model averaged dose-response meta-analysis

Matthew J. Lees, James Steele, Andrea Viloria Medina, Stuart Phillips et al.
1 links
Muscle metabolism and nutrition
preprint

Protein intake and changes in whole-body lean/fat-free mass, measures of muscle size or mass, strength, and physical function in healthy adults: An updated systematic review and Bayesian model averaged dose-response meta-analysis

Matthew J. Lees, James Steele, Andrea Viloria Medina, Stuart Phillips, Lauren Roxburgh, Myles Moore, Everson Nunes
preprint en

Abstract

Background: Previous dose-response meta-analyses of protein intake have generally assumed linear relationships or imposed a fixed "breakpoint", rendering inferences conditional on a particular functional form. We therefore estimated the dose-response associations between protein intake and changes in whole-body lean/fat-free mass, measures of muscle size or mass, strength, and physical function in healthy adults, with and without resistance training (RT), while accounting for uncertainty about the shape of these associations. Methods: We performed an updated systematic review of parallel-group randomised controlled trials that assigned participants to dietary or supplemental protein interventions, placebo, or control conditions. We searched MEDLINE, Embase, CINAHL, and Web of Science Core Collection through 14 September 2026 and assessed risk of bias with Cochrane RoB 2. Standardised mean changes were synthesised using hierarchical arm-based Bayesian meta-regression. For each outcome, we fit 26 candidate models representing no association, linear relationships, linear splines with a knot at 1.6 g/kg/day, and nonlinear smooth functions. Models were then averaged using leave-one-out cross-validation stacking weights. Pairwise contrasts between protein intakes were evaluated against a preregistered smallest effect size of interest of ±0.1 standardised units, using a 95% posterior-probability criterion. We additionally fit a range of sensitivity analyses including Mundlak regression, strict within-study contrasts, common sample models of achieved- or assignment-derived protein intake, errors-in-variables models, residual-correlation models, and intervention strategy (addition or energy matched substitution). Finally, primary dose-response models examined body-mass relative total protein intake (g/kg/day) and thus we also examined models with protein as absolute total intake and absolute supplementation (both in g/day). Results: We identified 119 eligible studies, of which 95 included sufficient data to contribute 254 study arms, 4,931 participants, and 1,864 eligible baseline-to-follow-up contrasts to the primary analyses. Primary dose-response functions across outcomes generally showed modest positive associations, particularly with RT. However, no supported contrast between a current and proposed protein intake resulted in a conclusively meaningful gain in outcome from increasing intake or a meaningful loss from decreasing it. Increases or decreases of approximately 0.2 to 0.5 g/kg/day from a current intake were generally supported as practically equivalent. Contrasts between more widely separated intakes were uncertain rather than demonstrably meaningful or equivalent. We could not identify a clearly optimal intake. Inference was robust across the completed sensitivity families, although stricter within-study analyses produced dose-response functions of attenuated magnitude. The absolute total intake and absolute supplementation dose analyses did not materially alter the overall interpretation based on relative total intake. Most trials raised some concerns or were at high risk of bias. Publication-bias adjustment attenuated outcome-level estimates towards zero, and certainty was low or very low for all GRADE-assessed contrasts. Conclusions: Protein intake showed a modest positive association with outcomes and so those wishing to maximise outcomes may see it as reasonable to aim for higher protein intakes. However, the estimates were not sufficiently precise to establish that the gain from increasing intake, or the loss from decreasing it, exceeds a practically meaningful threshold. Comparisons between more widely separated intakes, and the extent to which the observed associations represent causal dose effects, remain uncertain particularly as data are sparse at low and high intakes (2.0% of primary-analysis arms were below 0.8 g/kg/day and 8.3% were above 2.2 g/kg/day). As such, whilst our results do not establish adequacy of the RDA, we suggest that practical recommendations for changes to intakes from typical habitual values should be based upon considerations other than just their effectiveness for the outcomes examined here (e.g., cost, appetite and food preferences, energy restriction). Lastly, we suggest that future research should make concerted attempts to address the shortcomings of the current body of literature. Funding: Kieser Australia provided in-kind support through consultancy and employment. Registration: Open Science Framework, w2qpx.

University of Tasmania (AU), University of Guelph (CA), McMaster University (CA)
Good health and well-being
Muscle metabolism and nutrition
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon