Effects of resistance training performed to repetition failure or non-failure on muscular strength and hypertrophy:A systematic review and meta-analysis

Jozo Grgi ,Brd J.Shoenfeld *,John Orzem ,Filip Sbol

a Institute for Health and Sport(IHES),Victoria University,Melbourne,VIC 3011,Australia b Department of Health Sciences,Lehman College,Bronx,NY 10468,USA

c School of Health Sciences,Human Services and Nursing,Lehman College,Bronx,NY 10468,USA

d Fitness Academy,Zagreb 10000,Croatia

e Faculty of Kinesiology,University of Zagreb,Zagreb 10000,Croatia

Abstract

Keywords:1RM;Cross-sectional area;Data synthesis;Muscle size

1.Introduction

According to Henneman’s size principle,motor units are recruited in an orderly fashion.1This principle dictates that as force production requirements increase,motor units are recruited according to the magnitude of their force output,with small motor units being recruited first.2Theoretically,in a resistance exercise set using moderate loads,lower threshold motor units associated with type I muscle fibers are initially recruited to lift the load.2-4As the lower threshold motor units become fatigued,increased recruitment occurs of the higher threshold motor units associated with type II muscle fibers in order to maintain force production.2-4Therefore,performing resistance exercise sets to momentary muscular failure(i.e.,the maximum number of possible repetitions in a given set)is thought to be necessary to recruit all possible motor units.3,4Accordingly,some suggest this manner of training is optimal for achieving resistance training-induced increases in muscular strength and muscle size.3,4

Given the hypothesis that training to muscle failure is important for catalyzing resistance training-induced adaptations,several studies examined the effects that this type of training has on muscular strength and hypertrophy,as compared to the effects of training that does not include reaching muscle failure.5-21However,detailed scrutiny of these studies highlights inconsistent findings.For example,some report that training to muscle failure results in greater increases in muscular strength and/or hypertrophy.5,18However,others suggest that both training options(i.e.,training either to or not to muscle failure)can produce similar improvements with respect to these outcomes.9,16Some studies even indicate that training to failure has a detrimental effect.5,6The inconsistent evidence on this topic currently hinders the ability to draw practical recommendations for training program design.

In an attempt to provide greater clarity on the equivocal evidence on this topic,Davies and colleagues22,23performed a meta-analysis in which they pooled studies comparing the effects of training to muscle failure vs.non-failure on muscular strength gains.The analysis included 8 studies and indicated no significant difference between training to or not to muscle failure in terms of increases in muscular strength.Of the 8 studies included in this review,4 equated training volume between the groups and 4 did not equate training volume.Since publication of the meta-analysis by Davies et al.,22,238 additional studies have been published that examine the topic.8,11,13-17,21Thus,an updated meta-analysis would theoretically have approximately a 2-fold increase in the number of included studies.Furthermore,while the effects of training to or not to muscle failure on muscular strength have been explored via meta-analysis,the same is not true for hypertrophy.Therefore,in this review,we performed an updated meta-analysis exploring the effects of training to failure on muscular strength as well as conducted the first meta-analysis exploring the effects of training to muscle failure on hypertrophy outcomes.

2.Methods

2.1.Search strategy

We performed the systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses(PRISMA)guidelines.24Electronic searches of PubMed/MEDLINE,Scopus,and SPORTDiscus databases were conducted using the following search syntax:(“resistance training”O(jiān)R“resistance exercise”O(jiān)R“strength training”O(jiān)R“strength exercise”O(jiān)R“weight training”O(jiān)R“weight exercise”)AND(“repetition failure”O(jiān)R“failure training”O(jiān)R“non-failure training”O(jiān)R“non failure training”O(jiān)R“muscular failure”O(jiān)R“muscle failure”O(jiān)R“to failure”O(jiān)R“not to failure”O(jiān)R“without resting”O(jiān)R“volitional interruption”O(jiān)R“high fatigue”O(jiān)R “l(fā)ow fatigue”)AND (“1 repetition maximum”O(jiān)R“1 RM”O(jiān)R“1RM”O(jiān)R“one repetition maximum”O(jiān)R“MVC”O(jiān)R“maximal voluntary contraction”O(jiān)R“muscle strength”O(jiān)R“muscular strength”O(jiān)R“muscle hypertrophy”O(jiān)R“muscular hypertrophy”O(jiān)R“muscle fibre”O(jiān)R“muscle fiber”O(jiān)R“muscle thickness”O(jiān)R“CSA”O(jiān)R“cross-sectional area”O(jiān)R“muscle size”).In addition to the primary search,we performed secondary searches by examining the reference lists of the included studies and by conducting forward citation tracking(i.e.,examining studies that have cited the included studies)in the Scopus database.Two authors of the review(JG and BJS)conducted these searches independently.Following the initial searches,the lists of included and excluded studies were compared between the authors.Any discrepancies between them were resolved through discussion and agreement.The search was finalized on January 2,2020.

2.2.Inclusion criteria

Based on the following criteria,we included studies that:(a)randomized participants(of any age)to the experimental groups;(b)compared the effects of resistance training to vs.not to muscle failure;(c)assessed changes in muscular strength and/or hypertrophy;(d)had a training protocol lasting for a minimum of 6 weeks;and(e)involved apparently healthy participants.For muscular strength outcomes,we considered studies that used either isometric or dynamic tests,or both.For muscular hypertrophy,we considered studies that assessed changes at the muscle fiber and/or whole muscle level.We considered studies with independent sample groups as well as those with dependent sample groups.We did not include studies that used blood flow restriction resistance training or concurrent training interventions(e.g.,combined resistance and aerobic training).

我們在大都會(huì)學(xué)習(xí)中心旁聽了一節(jié)歷史課。這是一所公立磁石學(xué)校，授課的是一名全美優(yōu)秀教師，學(xué)生則是九年級（相當(dāng)于中國初三的學(xué)生）。這位老師對課堂有很強(qiáng)的掌控力，全程很好地駕馭著課堂，引導(dǎo)學(xué)生積極參與，課堂氣氛十分活躍。我們跟著這位老師完整地上完了一節(jié)課，連中途校方請我們換教室繼續(xù)參觀，都沒舍得離開。

2.3.Data extraction

From each included study,we extracted the following data:(a)lead author and year of publication;(b)sample size and participant characteristics,including age and resistance training experience;(c)details of the resistance training programs;(d)muscular strength test(s)used and/or the site and tool used for the muscular hypertrophy assessment;and(e)pre-and post-intervention mean±SD of the strength and/or hypertrophy outcomes.Data extraction was performed independently by 2 authors(JG and BJS).Any discrepancies in the extracted data were resolved through discussion and consensus.

2.4.Methodological quality

We assessed the methodological quality of the included studies using the 27-item Downs and Black checklist.25This checklist addresses different aspects of the study design,including:reporting(Items 1-10),external validity(Items 11-13),internal validity(Items 14-26),and statistical power(Item 27).Given the specificity of the included studies(i.e.,exercise intervention),we modified the checklist by adding 2 items,1 pertaining to the training programs(Item 28)and 1 to training supervision(Item 29).22,26-28On this checklist,each item is scored with 1 if the criterion is satisfied and with 0 if the criterion is not satisfied.Based on the summary score,studies were classified as being of:good quality(21-29 points),moderate quality(11-20 points),or poor quality(less than 11 points).22,26,27Studies were independently rated by 2 reviewers(JG and FS)who settled any observed differences with discussion and agreement.

2.5.Statistical analyses

For each hypertrophy or strength outcome,the contrast between the training to failure vs.non-failure groups was calculated as the difference in effect sizes(ESs),where the ES was determined as the posttest-pretest mean change in each group,divided by the pooled pretest standard deviation,and multiplied by an adjustment for small sample bias.29ESs were interpreted as:small(≤0.20),moderate(0.21-0.50),large(0.51-0.80),and very large(＞0.80).30ESs are presented with their respective 95%confidence interval(95%CI).The variance of the difference in ESs depends on the within-subject posttest-pretest correlation,which was not available from the published data for many of the studies.Among studies for which this correlation could be estimated(back-solving from paired t test p values or SD of posttest-pretest change scores,when presented),the median value was 0.86;the moderately conservative value of 0.75 was used to calculate the variance for all studies.Sensitivity analyses(not presented)were performed using correlations ranging from 0.25 to 0.85;results were consistent with those using 0.75.Typically,when studies report multiple ESs,1 approach is to use study average ES,but this may result in a loss of information.31Therefore,we used a robust variance meta-analysis model,with adjustments for small samples,to account for correlated ESs within studies.32This meta-analysis model is specifically designed to be used when dealing with dependent ESs(e.g.,multiple strength tests in a single study).31Meta-analysis was conducted separately for the hypertrophy outcomes and strength outcomes.In addition,subgroup analyses were performed to explore the effects of training status(trained vs.untrained),training volume(volume equated vs.volume not equated),body region(upper vs.lower),exercise selection(multi-vs.single-joint exercises(only for strength)),and study design(independent vs.dependent groups).For hypertrophy outcomes,a sensitivity analysis was performed in which the muscle fibre data was excluded from the analysis.Publication bias was checked by examining funnel plot asymmetry and calculating trim-and-fill estimates.The trim-and-fill estimates(not presented)were similar to the main results.Calculations were performed using the robumeta package within R(Version 3.6.1;the R Foundation for Statistical Computing,Vienna,Austria).33All meta-analyses were performed using the robust variance random effects model.Effects were considered statistically significant at a p value of＜0.05.

3.Results

3.1.Search results

The primary search resulted in 1972 potentially relevant references.Of these results,15 studies7-21were identified that satisfied the inclusion criteria.A screening of the reference lists of the included studies and an examination of newer studies that cite them resulted in an additional 591 and 744 results,respectively.However,we did not find any additional relevant studies in the secondary searches.Therefore,the final number of included studies was 15,as presented in Fig.1.7-21

3.2.Study characteristics

Fifteen studies explored muscular strength outcomes(Table 1).The pooled number of participants in the studies was 394(265 males and 129 females).All participants in the studies were young adults.The sample sizes in the individual studies ranged from 9 to 89 participants,with a median of 25.Six studies7,10-12,17,21included resistance-trained participants,while the others were conducted on untrained individuals(Table 1).The duration of the training programs ranged from 6 to 14 weeks,with a median of 8 weeks.Training frequency ranged from 2 to 3 days per week.Muscular strength was most commonly assessed using the 1-repetition maximum(1RM)test.Other strength tests included the 6RMand 10RM,as well as different isometric or isokinetic strength(e.g.,knee extension,elbow flexion).

Seven studies11,13-17,19explored hypertrophy outcomes(Table 1).The pooled number of participants across studies was 219(130 males and 89 females).All participants in the studies were young adults.In the individual studies,sample sizes ranged from 10 to 89 participants,with a median of 25.Two studies11,17involved resistance-trained participants,while the others employed untrained individuals as study participants(Table 1).Resistance training programs in the studies lasted 6-14 weeks(10 weeks on average)with a training frequency of 2-3 days per week.Hypertrophy was most commonly assessed by the changes in muscle cross-sectional area or thickness of the quadriceps muscle.Some studies assessed alternative sites for muscle thickness,such as the elbow flexor and anterior deltoid.One study also assessed cross-sectional area changes in types I and II muscle fibers.19

Fig.1.Flow diagram of the search process.

Assessedoutcomes 6RMbenchpress Isometrickneeextensionand flexion 1RMandisometrickneeextension 1RMbenchpressandsquat 1RMbenchpressandsquat,vastus medialis,elbowflexor,anterior deltoidmusclethickness 1RMsquat 1RMandisometrickneeextension,rectusfemorisandvastus lateralisCSA 1RMkneeextension,quadriceps CSA Trainingdurationandweekly frequency 6weeks,3daysperweek 6weeks,2daysperweek 9weeks,3daysperweek 11weeks,2daysperweek 6weeks,2daysperweek 14weeks,3daysperweek 14weeks,2-3daysperweek 8weeks,2daysperweek Volume equated Yes Yes Yes Yes Yes No Yes Yes Setandrepetitionscheme Failure:4sets×6repetitions Non-failure:8sets×3repetitions Failure:25repetitionsinasfewsetsaspossible Non-failure:5sets×5repetitions Failure:4sets×10repetitions Non-failure:40repetitionswith30ofrestbetweeneachrepetition Failure:3sets×(6-10repetitions)Non-failure:6sets×(3-5repetitions)Failure:4sets×10repetitions)Non-failure:8sets×5repetitions Failure:1set×(8-12repetitions)Non-failure:3sets×10repetitions Failure:3-4setsperformedtofailure Non-failure:totalnumberofrepetitionsinthegrouptrainingtofailurewas dividedintomultiplesets Failure(highload):3setstomusclefailure Non-failure(highload):60%ofthetotalrepetitionsinthegrouptrainingto failurewasusedperset;additionalsetswereaddedtomatchthetotalnumberof 6RM 6RM Failure:6-10RM,or80%6-10RM Non-failure:80%-105%Trainingload Failure:80%-105%Failure:80%ofmaximaltorque Non-failure:80%ofmaximaltorque 6-10RM Non-failure:6-10RM,or80%10RM Failure:75%1RM Non-failure:75%1RM Failure:75%1RM Non-failure:75%1RM Failure:8-12RM Non-failure:90%-100%Failure:50%-60%1RM Non-failure:50%-60%1RM Failure(highload):80%1RM Non-failure(highload):80%1RM 26elitejuniormaleteam Participant gameplayerswithprevious experienceinresistance training 9younguntrainedmen 23younguntrainedmenand women playerswithpreviousexperienceinresistancetraining men men Table1 Summaryofstudiesincludedinthereview.Study Drinkwateretal.(2005)7 Fisheretal.(2016)8 Follandetal.(2002)9 Izquierdoetal.(2006)1029youngmaleBasqueball Karstenetal.(2021)1118youngresistance-trained Krameretal.(1997)1230youngresistance-trained Lacerdaetal.(2020)1310younguntrainedmen Laseviciusetal.(2019)1425younguntrainedmen 1RMandisokineticelbowflexion,elbowflexormusclethickness 1RMkneeextension,vastus lateralisCSA 1RMsquat,quadricepsCSA,musclefiberCSA 1RMandisometricelbowflexion 1RMandisometricelbowflexion,elbowflexorCSA 1RMsquat 1RMbenchandlegpress,10RM benchpress,legpress,seatedrow,andsquatmachine 10weeks,2daysperweek 12weeks,3daysperweek 8weeks,2daysperweek 6week,3daysperweek 12weeks,3daysperweek 8weeks,3daysperweek 8weeks,3daysperweek No Yes/No Yes Yes No No Yes repetitionsbetweenthegroups Failure(lowload):3setstomusclefailure Non-failure(highload):60%ofthetotalrepetitionsinthegrouptrainingto failurewasusedperset;additionalsetswereaddedtomatchthetotalnumberof repetitionsbetweenthegroups Failure:3setstomusclefailure Non-failure(volumeequated):4sets×7repetitions Non-failure(volumenon-equated):3sets×7repetitions Failure(highload):3setstomusclefailure Non-failure(highload):3setsnottomusclefailure(1-3repetitions in“reserve”)Failure(lowload):3setstomusclefailure Non-failure(lowload):3setsnottomusclefailure(1-3repetitions in“reserve”)Failure:velocitylossof40%Non-failure:velocitylossof20%Failure:1set×(6-10repetitions)Non-failure:6-10sets×1repetition Failure:4sets×6repetitions Non-failure(rapidshortening):4sets×4repetitions Non-failure(stretch-shortening):4sets×4repetitions Failure:1set×(8-12repetitions)Non-failure:(3-5sets)×(2-10repetitions)Failure:3sets×10repetitions Non-failure:3sets×10repetitions 1RM 1RM 1RM 1RM 1RM 1RM of2-10RM Failure(lowload):30%Non-failure(highload):30%1RM Failure:70%1RM Non-failure(volumeequated):70%Non-failure(volumenon-equated):70%Failure(highload):80%1RM Non-failure(highload):80%1RM Failure(lowload):30%Non-failure(lowload):30%Failure:70%-85%1RM Non-failure:70%-85%1RM Failure:6RM Non-failure:6RM Failure:85%1RM Non-failure(rapidshortening):85%Non-failure(stretch-shortening):85%1RM Failure:8-12RM Non-failure:80%-100%Failure:10RM Non-failure:90%oftheloadusedinthegroup trainingtofailure=repetitionmaximum.Martorellietal.(2017)1589younguntrainedwomen N′obregaetal.(2018)1627younguntrainedmen Pareja-Blancoetal.(2017)1722resistance-trainedmen Rooneyetal.(1994)1827younguntrainedmenand women Sampsonetal.(2016)1928younguntrainedmen Sanbornetal.(2000)2017younguntrainedwomen Vieiraetal.(2019)2114youngresistance-trained men Abbreviations:CSA=cross-sectionalarea;RM

3.3.Methodological quality

The median score on the modified Downs and Black checklist was 21 points(range:19-24 points).Five studies7,14,18-20were classified as being of moderate methodological quality,whereas all other studies were considered to be of good methodological quality(Table 2).None of the studies were classified as being of low quality.

3.4.Meta-analysis results

When considering all available studies,the meta-analysis for muscular strength gains indicated no significant difference between the training conditions(p=0.198;ES=-0.09,95%CI:-0.22 to 0.05;Fig.2).Subgroup analysis for studies that did not equate training volume showed a moderate significant effect favoring non-failure training on strength gains(p=0.025;ES=-0.32,y 95%CI:-0.57 to-0.07).In the subgroup analyses for studies that did equate training volume,however,there was no significant difference between training conditions with respect to strength gains(p=0.860;ES=0.01,95%CI:-0.12 to 0.15).Subgroup analyses that stratified the studies according to training status,body region,exercise selection,or study design showed no significant differences between training conditions(Table 3).

When considering all available studies,the meta-analysis for hypertrophy indicated no significant difference between the training conditions (p=0.152;ES=0.22,95%CI:-0.11 to 0.55;Fig.3).The sensitivity analysis did not have a meaningful impact on the results.Notably,in the subgroup analysis for resistance-trained individuals,the analysis showed that training to failure had a significant effect on muscle hypertrophy(p=0.039;ES=0.15,95%CI:0.03-0.26).Subgroup analyses that stratified the studies according to training volume,body region,or study design,however, did not demonstrate significant differences between training conditions(Table 3).

4.Discussion

The results of this systematic review and meta-analysis suggest that training to muscle failure may produce similar increases in muscular strength and muscle size as non-failure training.This finding remained consistent among subgroup analyses,which suggests that the impact of training to failure is not likely to be moderated by variables such as body region,exercise selection,or study design.The subgroup analyses of studies that did not equate training volume between the groups stood out because it found that muscular strength gains favored training that did not include muscle failure.On the other hand,another subgroup analysis found that training to failure might be a benefit in terms of muscle hypertrophy for resistancetrained individuals.

4.1.Muscular strength

In 2009,the American College of Sports Medicine published a position stand on resistance training prescription for healthy adults.34Even though training to muscle failure is briefly mentioned,the position stand stops short of making any recommendations in regard to this training variable for the development of strength.Critics of this position stand3,4suggested that individuals seeking to improve strength should perform repetitions to muscle failure based on the premise that this method of training is optimal for maximizing strength gains.As such,there is an apparent disagreement in the literature relative to recommendations for this training variable.Based on the current evidence and our pooled analysis comprising approximately 400 participants,it seems that training to muscle failure is not necessary for increases in muscular strength.Nonetheless,training in this manner does not appear to have detrimental effects on these adaptations,suggesting that the choice of training to failure vs.non-failure can be based more or less on personal preference alone.Finally,the upper and lower limits of the 95%CIs were within the zones of small to moderate ES suggesting that even if there were a benefit to either of these methods of training,the benefit is likely to be negligible for most individuals.

As previously noted,the subgroup analysis for training volume showed significant favoring for the effects of non-failure training on muscular strength gains.However,in the majority of studies that did not equate training volume between the groups,participants that did not train to muscle failure performed more sets(i.e.,more volume)than did the individuals training to muscle failure.12,17,19,20For example,in a study done by Kramer et al.,12the group that trained to muscle failure performed a single set per exercise for 8-12 repetitions,whereas the group that did not train to muscle failure performed 3 sets of 10 repetitions(while not reaching muscle failure).This is relevant to emphasize because it has been previously shown that training volume increases strength in a linear dose-response manner.35Therefore,the significant effect of training that does not include muscle failure seems to be primarily related to the differences in training volume between the groups.Indeed,when considering only studies that equated for training volume between the groups,the pooled ES amounted to 0.01 nested within a 95%CI of-0.12 to 0.15,suggesting highly similar increases in strength regardless of whether an individual does or does not reach failure during training.

4.2.Muscle hypertrophy

Totalscore 20 21 21 23 23 21 24 19 22 24 23 20 20 20 23 supervision(Items27-29)29 1 1 1 1 1 1 1 1 1 1 1 1 0a 1 1 Power,compliance,28 0 0 1 1 1 1 1 0 1 1 1 0 1 0 1 27 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 26 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 25 1 1 1 1 1 0 1 0 1 1 0 1 0 0 1 24 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a Internalvalidity(Items14-26)23 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 22 0a 0a 0a 1 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 21 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 20 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 19 0a 0a 1 1 1 1 1 0a 1 1 1 0a 1 0a 1 18 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 16 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 15 0a 0a 0a 0a 0a 0a 1 0a 0a 0a 1 0a 0a 0a 0a 14 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a ResultsofthemethodologicalqualityassessmentusingthemodifiedDownsandBlackchecklist.Externalvalidity(Items11-13)13 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 12 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 10 0 1 0 0 1 0 1 0 0 1 1 1 1 1 0 9 1 1 0 1 1 1 1 1 1 1 1 0 1 1 1 Reporting(Items1-10)8 0 0 0 0 0 1 1 0 0 1 1 0 0 1 1 7 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 6 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 1 1 1 1 1 0 0 1 1 1 0 1 0 0 1 4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Table2 Study Drinkwateretal.(2005)7 Fisheretal.(2016)8 Follandetal.(2002)9 Izquierdoetal.(2006)101 Karstenetal.(2021)111 Krameretal.(1997)121 Lacerdaetal.(2020)131 Laseviciusetal.(2019)141 Martorellietal.(2017)151 N′obregaetal.(2018)161 Pareja-Blancoetal.(2017)171 Rooneyetal.(1994)181 Sampsonetal.(2016)191 Sanbornetal.(2000)201 Vieiraetal.(2019)211 Note:1=criteriamet;0=criterianotmet.aItemwasunabletobedetermined,scored0.

The meta-analysis for hypertrophy outcomes suggests that similar increases in muscle size can be attained regardless of whether or not training is carried out to muscle failure.This means that,based on the current body of literature,training to momentary muscle failure does not seem to be required for increases in muscle size.However,we should again highlight that training to muscle failure does not appear to produce any detrimental effects on muscle hypertrophy.Still,it should be considered that the upper limit of the 95%CI in this analysis was 0.55,which is in the range of a large effect.Therefore,while we did not show significant differences between training to failure vs.non-failure,the wide 95%CI also underlines the need for future research on the topic.

The subgroup analysis performed for resistance-trained participants indicated that,for them,training to failure had a significant effect on muscle hypertrophy.Indeed,it is conceivable that,as an individual approaches his or her genetic ceiling for muscular adaptations,a greater intensity of effort may be required to elicit further gains.However,this analysis was limited by the small number of included studies.Specifically,only 2 studies11,17were included:one that equated training volume between the groups,and one that did not.While the results presented in our review offer preliminary support for training to failure in resistance-trained participants,future studies are needed to provide greater clarity on the influence of training status when exercise variables are strictly controlled,particularly in highly trained individuals.

The finding observed in the main meta-analysis for hypertrophy could be explained by the loads used in the majority of included studies.In general,studies used moderate to high loads(e.g.,60%-90%1RM)in their resistance training programs(Table 1).This aspect is relevant because the upper limit of motor unit recruitment is thought to be around 60%-85%of maximum force(depending on the muscle group).36-38In other words,when exercising with such training loads,high-threshold motor units tend to be recruited from the onset of the exercise,and the additional increase in force beyond the upper limit of motor unit recruitment is accomplished by rate coding.36-38Therefore,training to muscle failure may not be needed for motor unit recruitment when using moderate or high loads.However,it should be noted that simply because a fiber has been recruited does not mean that it has been sufficiently stimulated to hypertrophy.Thus,while the level of recruitment may provide a partial mechanistic explanation of these findings,it would appear that other factors are involved as well.39

Recently,it has been hypothesized that training to muscle failure becomes increasingly more important when exercising with lower loads(e.g.,30%of maximum force),due to the delayed recruitment of larger motor units.40In support of this idea,Lasevicius et al.14compared training to muscle failure vs.non-failure with loads of 30%and 80%1RM.The study used a within-subject unilateral design whereby 1 limb trained to failure at the given load and the other did not.Results indicated that training to failure promoted greater increases in muscle size in groups training with low loads.Alternatively,in the groups performing high-load training,similar increases in muscle size were noted with and without training to muscle failure.N′obrega et al.16performed a similar experiment and reported comparable hypertrophy effects in both high-and low-load training groups,regardless of whether or not they trained to failure.However,in this study,the groups not training to failure performed only 1-2 repetitions less per set than the group training to failure.In the Lasevicius et al.14study,the limb that did not exercise to failure,trained with 60%of the total repetitions(per set)performed by the limb that trained to failure;additional sets were added to match the total number of repetitions between the conditions.These methodological differences are likely to account for the conflicting evidence.As such,this is an area requiring further scientific attention.

Fig.2.The forest plot from the meta-analysis of the effects of training to failure vs.non-failure on muscular strength.The X axis denotes Cohen’s d(ES)while the whiskers denote the 95%CI.a The sum of the percentages is not 100%due to the rounding.95%CI=95%confidence interval;ES=effect size;MVC=maximal voluntary contraction;RM=repetition maximum;RS=rapid speed;SSC=stretch-shortening cycle.

Table 3 Results of the subgroup meta-analyses.

4.3.Generalizability of the results

While this meta-analysis showed no significant differences between the effects of training to muscle failure or non-failure on muscle strength and hypertrophy,these results are specific to the population analyzed in all included studies—young adults.Therefore,future work is needed to explore the effects of training to failure vs.non-failure among middle-aged and older adults.Additionally,our results are specific to studies that used isolated traditional resistance training programs.There is evidence that avoiding muscle failure may be important when using blood flow restriction training and in concurrent exercise programs.5,6,41For example,in a study by Carroll et al.,5the participants coupled resistance training with a low-volume sprint interval training.While this study did not satisfy our inclusion criteria due to its utilization of concurrent exercise programs,its results did indicate that failure training had a detrimental effect on muscle hypertrophy in resistancetrained men.Therefore,the findings presented herein cannot necessarily be generalized to adaptations that occur with concurrent training.

4.4.Areas for future research

Although our findings provide evidence that consistently training to failure is not obligatory for enhancing muscular strength and hypertrophy,the current literature is not sufficient to determine the level of effort necessary to maximize these adaptations.It is currently unclear whether the same effects would be achieved if an individual stops the set,for example,5 repetitions before failure vs.2 repetitions before failure.Future research should seek to quantify the lower threshold as to how many repetitions short of failure would be sufficient to elicit an optimal adaptive response.This should be quantified across various repetition ranges,as the relative magnitude of load will necessarily influence results.

4.5.Methodological quality and limitations

All included studies were classified as being of moderate or good methodological quality.Therefore,the results presented in this review are not confounded by the inclusion of studies that were of low methodological quality.However,there is 1 significant limitation noted in some of the included studies.Specifically,5 studies4,7,8,18,20did not report participants’adherence to the training programs(Table 2).In the studies that did report adherence to the training interventions,it was very similar between the groups(Table 1).Thus,while there is no reason to believe that adherence was not similar between the groups in papers that did not report these data,future studies should ensure that this information is clearly presented.

An important methodological consideration of this review is that we included studies with independent groups as well as those with dependent groups.In a design with dependent groups,limbs are assigned to perform 1 of 2 training routines(e.g.,either training to or not to failure).This design has certain advantages,such as minimizing the variability in responses between individuals.Still,this model’s limitation is the possible cross-education effect,which dictates that training 1 limb increases strength in both limbs.42However,we also conducted subgroup analyses where the studies were stratified according to their study design.There was no significant difference between training to failure vs.non-failure in subgroup analyses for studies with independent vs.dependent groups,therefore reinforcing the primary analysis results.As mentioned previously,training to muscle failure may be more important with lower as opposed to higher loads.In the present review,we included studies that utilized both high and low loads in their respective training routines,which might be a limitation of the review,even though it should be considered that only 2 studies used very low loads(i.e.,30%1RM).14,16

5.Conclusion

The findings of this review suggest that training to or not to muscle failure may produce similar increases in muscular strength and muscle size.This finding generally remained consistent in subgroup analyses that stratified the studies according to body region,exercise selection,or study design.Still,when volume was not controlled for,there was favoring of non-failure training on strength gains,as well as favoring of training to failure for hypertrophy in resistance-trained individuals.More studies should be conducted among older adults and highly trained individuals in order to improve the generalizability of these findings.

Fig.3.The Forest plot from the meta-analysis on the effects of training to failure vs.non-failure on muscle hypertrophy.The X axis denotes Cohen’s d(ES)while the whiskers denote the 95%CI.a The sum of the percentages is not 100%due to the rounding.95%CI=95%confidence interval;CSA=cross-sectional area;ES=effect size;RS=rapid speed;SSC=stretch-shortening cycle.

Authors’contributions

JG conceived the idea for the review,performed the searches,data extraction,and methodological quality assessment,and drafted the manuscript;BJS conceived the idea for the review,performed the searches and data extraction,and critically revised the manuscript content;JO analyzed the data and critically revised the manuscript content;FS performed the methodological quality assessment and critically revised the manuscript content.All authors have read and approved the final version of the manuscript,and agree with the order of presentation of the authors.

Competing interests

The authors declare that they have no competing interest.