The effect of muscle warm-up on voluntary and evoked force-time parameters: A systematic review and meta-analysis with meta-regression

Highlights

• •Increasing muscle temperature, through active or passive means, moderately improves dynamic, fast-velocity force production and rate of force development but not maximal force capacity in both voluntary and electrically evoked contractions.
• •Factors such as temperature measurement method, warm-up specificity, sex, and training status do not seem to significantly impact the temperature increase-related performance enhancement, although current data are limited.
• •Active warm-up does not always lead to significantly greater performance increases than passive heating strategies, and the magnitude of enhancement may depend on the specifics of the active warm-up.

Abstract

Background

While muscle contractility increases with muscle temperature, there is no consensus on the best warm-up protocol to use before resistance training or sports exercise due to the range of possible warm-up and testing combinations available. Therefore, the objective of the current study was to determine the effects of different warm-up types (active, exercise-based vs. passive) on muscle function tested using different activation methods (voluntary vs. evoked) and performance test criteria (maximum force vs. rate-dependent contractile properties), with consideration of warm-up task specificity (specific vs. non-specific), temperature measurement method (muscle vs. skin), baseline temperatures, and subject-specific variables (training status and sex).

Methods

A systematic search was conducted in PubMed/MEDLINE, Scopus, Web of Science, Cochrane, Embase, and ProQuest. Random-effects meta-analyses and meta-regressions were used to compute the effect sizes (ES) and 95 % confidence intervals (95 %CIs) to examine the effects of warm-up type, activation method, performance criterion, subject characteristics, and study design on temperature-related performance enhancement.

Results

The search yielded 1272 articles, of which 33 met the inclusion criteria (n = 921). Increasing temperature positively affected both voluntary (3.7 %/°C ± 1.8 %/°C (mean ± SD), ES = 0.28 (95 %CI: 0.14 to 0.41)) and evoked (3.2 %/°C ± 1.5 %/°C (mean ± SD), ES = 0.65 (95 %CI: 0.29 to 1.00)) rate-dependent contractile properties (dynamic, fast-velocity force production, and rate of force development (RFD)) but not maximum force production (voluntary: –0.2 %/°C ± 0.9 %/°C (mean ± SD), ES = 0.08 (95 %CI: –0.05 to 0.22); evoked: –0.1 %/°C ± 0.8 %/°C (mean ± SD), ES = –0.20 (95 %CI: –0.50 to 0.10)). Active warm-up did not induce greater enhancements in rate-dependent contractile properties (p = 0.284), maximum force production (p = 0.723), or overall function (pooled, p = 0.093) than passive warm-up. Meta-regressions did not reveal a significant effect of study design, temperature measurement method, warm-up task specificity, training status, or sex on the effect of increasing temperature (p > 0.05).

Conclusion

Increasing muscle temperature significantly enhances rate-dependent contractile function (RFD and muscle power) but not maximum force in both evoked and voluntary contractions. In contrast to expectation, no effects of warm-up modality (active vs. passive) or temperature measurement method (muscle vs. skin) were detected, although insufficient data prevented robust sub-group analyses.

Graphical Abstract

1. Introduction

The temperature dependence of human skeletal muscle force production has been a subject of investigation since the early 20th century.^1,2 It has hence been well established that increased muscle contractile capacity occurs when muscle temperature increases within physiologically normal levels; resting muscle temperature is ∼35°C and can increase to ∼40°C. 3, 4, 5, 6, 7, 8, 9 In particular, increases in rate-dependent contractile properties such as the rate of force development (RFD), ^8,10 shortening velocity,⁴ and power output ^3,9 have been observed even with relatively little elevation in muscle temperature (∼1°C–2°C). However, the effect of increasing temperature on maximum force production capacity is less clear, with studies reporting increases 11, 12, 13 as well as decreases ^5,6 in maximum voluntary contraction (MVC) torque. Nonetheless, the increase in rapid force production is of primary interest in both sporting ^14,15 and clinical 16, 17, 18 populations under circumstances in which a limited time for force production exists.¹⁹

Much of the evidence for increased muscle contractility in response to muscle temperature elevation originates from studies using passive heating modalities,^1,3,5,9 which allow researchers to isolate the effect of temperature from non-temperature-related mechanisms such as neural drive,^20,21 blood/fluid influx,²² and motivation/arousal.²³ While passive warm-up has been shown to significantly enhance both voluntary^1,3,5,9 and evoked^5,24,25 contractile capacities (2 %/°C–5 %/°C), only a few studies have directly compared active and passive warm-up modalities,26, 27, 28, 29 with some reports favoring active over passive warm-up (∼6 %/°C vs. ∼3 %/°C).26, 27, 28, 29 It is therefore of interest to ascertain the magnitude of performance enhancement achieved following active vs. passive warm-ups, with the intent to better understand their role in sporting and athletic success.

While significant increases in both voluntary^1,3,5,9 and evoked^5,25 muscle contractility have been observed with temperature increases, few studies have directly compared the effect of muscle temperature on voluntarily vs. electrically evoked forces.^5,8,21,30,31 Some evidence suggests that evoked contractile properties may be more sensitive to temperature, with greater enhancements observed in evoked than in voluntary performance (e.g., 2 %/°C–17 %/°C vs. 2 %/°C–10 %/°C),^5,21,31 although the overall effect is currently unclear. This may have important implications for studies using voluntary and/or evoked contractions as markers of muscle performance (including fatigue) in response to acute or chronic interventions since it is unclear whether tests of evoked contractile function can be used as proxies for tests of voluntary function.32, 33, 34

Previous meta-analyses have highlighted a large between-study variability within the extant literature.^35,36 This might be explained by the variation in the types (e.g., active vs. passive) and configurations (e.g., intensity, duration, recovery, specificity) of the warm-up protocols used, along with the range of performance test criteria adopted (e.g., MVC vs. RFD; voluntary vs. evoked). However, it also may be explained by differences in study design. For example, few studies report pre- and post-intervention data for both experimental and control conditions,^8,11,37, 38, 39 with many making direct comparisons between post-intervention data (e.g., without presentation of pre- to post-intervention change scores,^5,6,9,40,41 potentially biasing the effect of muscle temperature on muscle function.³² Therefore, determining the consequences of study design is important for understanding the validity of study outcomes and, ultimately, for optimizing warm-up practices.

To explore the effects of muscle temperature triggered by both passive and active means on voluntary and evoked force production and RFD with sufficient sample size, a meta-analytic approach with meta-regression was adopted in the present study. We tested the hypotheses that: (a) increasing muscle temperature would have a greater effect on rate-dependent contractile properties than on maximum force production in both voluntary and evoked contractions, (b) increasing muscle temperature (irrespective of the warm-up modality) would have a greater effect on evoked than voluntary contractile function, and (c) active (i.e., exercise-based) warm-ups would induce greater increases in performance than passive warm-ups.

Additionally, some studies have reported that warm-up test exercise specificity, i.e., whether the warm-up and testing tasks were similar in motor pattern,²¹ temperature measurement method (muscle vs. skin),²³ as well as subject-specific variables such as training status^36,42 and biological sex^{42, 43} may influence the associations between increased muscle temperature and muscle function enhancements. We therefore ran analyses to explore the impact of these variables. It must be noted that heat stress studies (core temperature ≥ 38.5°C) were not included in the voluntary performance analyses because increasing body temperature (especially once core temperature exceeds 38.5°C) is associated with decreased voluntary muscle contractile function,^44,45 partly due to a reduced central (neural) drive to the muscle^44,45 and to the fact that whole body passive heating protocol treatments can easily reach such levels.⁴⁶

2. Methods

This systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA)⁴⁷ and PRISMA in Exercise, Rehabilitation, Sport Medicine, and Sports Science (PERSiST) recommendations.⁴⁸

2.1. Selection criteria

The eligibility criteria were informed by the PICOS framework (P: Population; I: Intervention; C: Comparator; O: Outcome(s); S: Study design)⁵⁴ and are presented in Table 1.

Table 1.

Criteria for inclusion and exclusion in the meta-analysis, according to the PICOS method.

	Definition	Inclusion criteria	Exclusion criteria
P	Population	•Healthy adults (male and female) •18–60 years	•Current or previous (<6 months) neuromuscular or musculoskeletal condition
I	Intervention	•Exercise-based (i.e., active) warm-up (e.g., general and specific) •Passive heating protocols (e.g., sauna, bath, hot packs, heated garments, irradiation, etc.)	•Interventions involving:CoolingRe-warm upFatiguing contractionsHeat stress (voluntary analyses only)Food supplements, nutritional or pharmacological aids
C	Comparator	•Pre-warm-up and/or non-warm-up control group data
O	Outcome(s)	•Voluntary contractile properties:Voluntary maximum muscle strength (e.g., peak force/torque)Voluntary rate-dependent contractile properties (e.g., isometric RFD, peak power output, velocity, etc.)Evoked contractile properties:Evoked maximum muscle strength (e.g., twitch/tetanic peak force/torque)Evoked rate-dependent contractile properties (e.g., twitch/tetanic RFD, TPT, etc.)Temperature measurement:Intramuscular (i.e., needle) or skin (i.e., thermistors, infrared thermometers, etc.) temperature	•No performance measure included (i.e., no primary outcome) •No temperature measurement included (i.e., no muscle or skin temperature recorded) •Long-duration performance (i.e., endurance task) •Testing in non-thermoneutral environment (i.e., warm-up before cold exposure)
S	Study design	•Experimental (EXP) designs:EXPFULL: Studies reporting pre- and post-intervention data for both experimental (i.e., warm-up) and control conditionsEXPWARM: Studies reporting pre- and post-intervention data for an experimental condition (i.e., warm-up) onlyEXPPOST: Studies reported post-intervention data for both experimental (i.e., warm-up) and control conditions (i.e., no pre-intervention data)	•Systematic reviews, conference abstracts, case reports, and letters to the editors •Articles not published in English or Portuguese

Abbreviations: PICOS = Population, Intervention, Comparator, Outcomes, and Study design; RFD = rate of force development; TPT = time-to-peak torque.

2.2. Data sources and searches

The following electronic databases were searched from the earliest date until September 20, 2022: PubMed/MEDLINE, Scopus, Web of Science, Cochrane, Embase, and ProQuest. Subsequently, reference lists and Google Scholar citations of the included studies were inspected for additional studies. All authors convened to develop the search strategy, which combined terms relating to strength, RFD, power, and evoked-contractile properties (i.e., twitch and tetanic force, RFD, etc.). The full syntaxes used for the literature search are presented in Supplementary Material 1.

2.3. Study selection

All results were imported to an EndNote database (EndNote Version X9.3.3; Clarivate Analytics, Philadelphia, PA, USA). The titles and abstracts of each study were screened for inclusion following the eligibility criteria presented in Table 1. Full-text articles of the included studies were then consulted to ensure eligibility. A PRISMA flow diagram showing the number of included/excluded studies and reasons for exclusion is presented in Fig. 1. A list of studies excluded with reasons is presented in Supplementary Material 2.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram illustrating the study screening process.

2.4. Data extraction

The following data were extracted from the included studies and coded into a personalized Microsoft Excel spreadsheet (Microsoft Corp., Redmond, WA, USA): (a) study characteristics (first author, year, sample size, and study design (Experimental (EXP)_FULL, EXP_WARM, and EXP_POST)); (b) participant characteristics (age, sex, and training status (untrained or trained)); (c) details of the warm-up protocols (passive or active; see below); (d) details of the performance outcomes measured (see below); (e) temperature assessment method (muscle or skin); (f) means and SDs for performance variables; (g) means and SD for temperature measurements; and (h) relative performance change (percentage change in performance for a given change in temperature (%/°C)). All meta-analytic comparisons were made with standardized effect sizes (ES) computed from the raw performance and temperature data (detailed later). However, the relative performance change (%/°C) was computed to allow for practical comparisons between studies and conditions as well as to align with much of the literature reporting performance enhancement relative to temperature change.^3,9,49

The warm-up protocols were classified as passive (including hot bath, sauna, heating packs/garments, or irradiation/diathermy), active (general or specific exercise activity), or both (combination of passive and active strategies). Active exercise strategies were classified as general (e.g., cycling or jogging before a performance test) or specific (i.e., including trial repetitions of the performance test). Given that elevated muscle temperature has distinct effects on maximum force vs. rate-dependent contractile properties,^23,49 the selected outcomes were stratified into 4 categories: (a) voluntary maximum muscle strength; (b) voluntary rate-dependent variables (e.g., dynamic, fast-velocity force production, RFD, power output, velocity, jump height); (c) evoked maximum muscle strength (e.g., twitch/tetanic peak force/torque); and (d) evoked rate-dependent variables (e.g., twitch/tetanic RFD or rate of torque development (RTD), time-to-peak force/torque).

For studies in which a control condition served as a comparator for multiple experimental conditions, the sample size (n) of the control condition was adjusted by dividing n by the number of conditions to prevent sample size over-estimation (e.g., control condition of n = 20 serving as a comparator for 2 experimental conditions resulted in a sample size adjustment to n = 10). In cases where studies did not provide the required data for analyses, the authors were contacted to supply the data.^12,30,31,50, 51, 52, 53, 54, 55, 56, 57, 58 Consequently, data from 3 studies^31,50,58 were included in the analyses. In cases where mean and SD data could not be retrieved, a free web-based plot digitizing tool (Version.4.5; WebPlotDigitizer, Automeris) was used to extract data from figures. This resulted in the inclusion of data from 20 studies.6, 7, 8^,11,13,21,27, 28, 29, 30, 31^{,39,41,50,51,57,}59, 60, 61 Standard error and 95 % confidence intervals (95 %CIs) were converted to SD and included in the analyses.^{8,11,12,55,59,62}

2.5. Methodological quality

The modified Physiotherapy Evidence Database (PEDro) scale was employed to assess the methodological quality of the articles included in the review. Given that it is generally not feasible to blind the subjects and investigators in exercise-based or passive warm-up interventions, Items 5–7 (which are specific to blinding) were removed from the scale. This approach has been used in previous exercise systematic reviews.63, 64, 65 With the removal of these items, the maximum score on the modified “PEDro 8-point” scale was 7 because the first item, related to eligibility criteria, was not included in the total score. The studies were categorized as follows: poor, 0–3; moderate, 4; good, 5; and excellent, 6–7.^65,66

2.6. Data syntheses and analyses

2.6.1. Meta-analyses

Random-effects meta-analyses were performed using the Comprehensive Meta-analysis software (Version 2; Biostat, Tampa, FL, USA). Primary analyses were performed to determine the effect of increasing temperature on: (a) voluntary maximum muscle strength, (b) voluntary rate-dependent contractile properties, (c) evoked maximum muscle strength, and (d) evoked rate-dependent contractile properties. The study classified ES magnitudes as: trivial, 0.00–0.19; small, 0.20–0.49; moderate, 0.50–0.79; and  large, ≥0.80.⁶⁷ Significance was set at p < 0.05. Heterogeneity was explored using the I² statistic and the Cochrane Chi-squared test (χ²). I² < 50 % indicates low heterogeneity, 50 %–75 % indicates moderate heterogeneity, and >75 % indicates high heterogeneity; χ² p < 0.1 indicates the presence of heterogeneity. Data are presented as ES and 95 %CI. Positive effects indicate benefits for warm-up strategies. The ES was calculated differently for each study design (EXP_FULL, EXP_WARM, and EXP_POST), varying in dividend and divisor scores, but consistently defined as the mean difference between a warm-up and non-warm-up condition divided by the pooled SD of all conditions, following software-specific formulas (detailed below). For analytical purposes, primary meta-analyses were performed using the ES of each study design separately and then with the ES of all study designs together.

EXP_FULL

$$ES=\frac{\left(\text{mean}{W}_{\text{post}}-\text{mean}{W}_{\text{pre}}\right)-\left(\text{mean}{C}_{\text{post}}-\text{mean}{C}_{\text{pre}}\right)}{\sqrt{\left(W_{n-1}\times {\left(\left(SD{W}_{\text{post}}+SD{W}_{\text{pre}}\right)/2\right)}^2+C_{n-1}\times {\left((SD{C}_{\text{post}}+SD{C}_{\text{pre}})/2\right)}^2\right)/\left(W_{n-1}+C_{n-1}\right)}}$$ Eq1

$$SE=\sqrt{\left(\frac{1}{W_n}+\frac{1}{C_n}\right)+\left(\frac{E{S}^2}{2\left(W_n+C_n\right)}\right)}$$ Eq2

EXP_WARM

$$ES=\frac{\text{mean}{W}_{\text{post}}-\text{mean}{W}_{\text{pre}}}{\sqrt{SD{W_{\text{post}}}^2+SD{W_{\text{pre}}}^2-\left(SD{W}_{\text{post}}+SD{W}_{\text{pre}}\right)}}$$ Eq3

$$SE=\sqrt{\left(\frac{1}{W_n}+\frac{1}{C_n}\right)+\left(\frac{E{S}^2}{2\left(W_n+C_n\right)}\right)}$$ Eq4

EXP_POST

$$ES=\frac{\text{mean}{W}_{\text{post}}-\text{mean}{W}_{\text{pre}}}{\sqrt{\left({(W_{n-1}\times SD{W}_{\text{post}})}^2+{(C_{n-1}\times SD{C}_{\text{post}})}^2\right)/\left(W_{n-1}+C_{n-1}\right)}}$$ Eq5

$$SE=\sqrt{\left(\frac{1}{W_n}+\frac{1}{C_n}\right)+\left(\frac{E{S}^2}{2\left(W_n+C_n\right)}\right)}$$ (6)

thereafter, ES and standard error (SE) were multiplied by Hedges' g J adjustment for small sample bias:

$$J=1-\frac{3}{4\left(W_{n-1}+C_{n-1}\right)-1}$$ Eq7

where W is warm-up (i.e., experimental condition), C is control, ES is effect size, SD is standard deviation, SE is standard error, and n is sample size.

2.6.2. Subgroup analyses

Subgroup analyses were performed to determine the effect of increasing temperature on primary outcomes (i.e., muscle contractile function). The χ² was used to compare effects between the following categorical variables: (a) study design (EXP_POST vs. EXP_WARM vs. EXP_FULL), (b) warm-up strategy (passive vs. active), (c) warm up-test exercise specificity (non-specific vs. specific), and (d) training status (untrained vs. trained).

2.6.3. Meta-regressions

Meta-regressions were performed to explore the influence of potential continuous moderators on the effect of warm-up on primary outcomes. The following moderator analyses were explored: (a) sex (% of men in the sample), (b) changes in performance with changes in temperature (skin vs. muscle measurement methods), and (c) baseline temperature. For the meta-regressions to explore the relationship of changes in temperature to performance, change in temperature (Δ_temp) was calculated using the approach used for calculating the change in primary outcomes (i.e., EXP_FULL, Δ_TEMP = (meanW_post– meanW_pre) – (meanC_post – meanC_pre); and EXP_WARM, Δ_TEMP = mean W_post – meanW_pre; EXP_POST, Δ_TEMP = meanW_post – meanC_post. For the meta-regressions to explore the influence of baseline temperature on performance, the following temperature measurements were considered: EXP_FULL: average meanW_pre and meanC_pre; EXP_WARM: meanW_pre; EXP_POST: meanC_post.

3. Results

3.1. Studies selection and search results

The initial literature search yielded 1524 results. After removing duplicates, titles, and abstracts, 603 studies were screened. We excluded 527 records, and retrieved one more study from other sources; thus, 77 full-text articles were screened for eligibility (Fig. 1). After screening the full-text articles for eligibility, 44 studies were excluded, resulting in 33 studies for the final analyses. From the 33 studies included (n = 921; age = 24.6 ± 4.0 years, mean ± SD), 7 studies used the EXP_FULL design (warm-up n = 161; control n = 125), 5 studies used the EXP_WARM design (warm-up n = 74), and 21 studies used the EXP_POST design (warm-up n = 293; control n = 268). Exercise warm-up strategies were performed as a heating intervention in 12 studies, passive heating strategies were applied in 18 studies, and a combination of active and passive strategies was used in 3 studies. Men-only samples were used in 18 studies, women-only samples in 5 studies, and 10 studies analyzed mixed samples. Nine studies included a trained sample. Seventeen studies measured skin temperature, 8 studies measured muscle temperature, and 8 studies measured both muscle and skin temperatures.

3.2. Characteristics of included studies

The methodological designs of the studies included are presented in Supplementary Material 3.

3.3. Methodological quality

The PEDro score for studies included in this review was 6.1 ± 0.7 (out of 7) (n = 33). Of the 33 studies, 26 studies (79 %) were rated excellent, 7 studies (25 %) were rated good, and no studies were rated moderate or poor (PEDro score ≤ 4). The PEDro scores for EXP_FULL (6.7 ± 0.8, n = 7) and EXP_WARM (7.0 ± 0.0, n = 5) were not statistically different (p = 0.102), but the PEDro score for EXP_POST (5.7 ± 0.5, n = 21) was significantly lower than EXP_WARM (p = 0.014). A comprehensive overview of the methodological quality assessment (PEDro criteria) for each study is reported in Supplementary Material 4.

3.4. Meta-analyses

3.4.1. Voluntary contractile properties

3.4.1.1. Voluntary maximum muscle strength

No significant effect was observed on voluntary maximum muscle strength when considering all study designs together (–0.2 %/°C ± 0.9 %/°C, ES = 0.08 (95 %CI: –0.05 to 0.22), p = 0.226; I² = 0 %, p = 0.915; k = 25) or when EXP_FULL (–0.1 %/°C ± 0.6 %/°C, ES = 0.21 (95 %CI: –0.18 to 0.59), p = 0.286; I² = 35 %, p = 0.188; k = 5), EXP_WARM (–1.7 %/°C ± 2.7 %/°C, ES = 0.02 (95 %CI: –0.19 to 0.23), p = 0.844; I² = 0 %, p = 0.436; k = 5), and EXP_POST (0.2 %/°C ± 1.4 %/°C, ES = 0.12 (95 %CI: –0.10 to 0.35), p = 0.288; I² = 0 %, p = 0.990; k = 15) were analyzed separately. No significant difference was revealed between study designs (χ² = 0.598, p = 0.742). Results are displayed in Fig. 2.

Fig. 2.

Effect of increasing temperature on voluntary maximum muscle force production capacity. Statistical significance indicates, p < 0.05. 95 %CI = 95 % confidence interval; AM = ante meridiem (before midday);CA60 = conditioning activity, 60°/s; CA300 = conditioning activity, 300°/s; ES = effect size; EXP = Experimental; LL = lower limit of the confidence interval; PM = post meridiem (after midday); UL = upper limit of the confidence interval. Please refer to the respective studies for further information about the subgroups mentioned in parentheses.

3.4.1.2. Voluntary rate-dependent contractile properties

A significant effect was observed for voluntary rate-dependent contractile properties when considering all study designs together (3.7 %/°C ± 1.8 %/°C, ES = 0.28 (95 %CI: 0.14 to 0.41), p < 0.001; I² = 0 %, p = 0.999; k = 38) and for EXP_FULL (5.3 %/°C ± 2.1 %/°C, ES = 0.59 (95 %CI: 0.18 to 0.99), p = 0.005; I² = 0 %, p = 0.870; k = 8) and EXP_POST (2.8 %/°C ± 3.9 %/°C, ES = 0.28 (95 %CI: 0.09 to 0.47), p = 0.004; I² = 0 %, p = 1.000; k = 24) but not EXP_WARM (2.5 %/°C ± 5.4 %/°C, ES = 0.17 (95 %CI: –0.06 to 0.40), p = 0.152; I² = 0 %, p = 0.609; k = 6). No significant difference was revealed between study designs (χ² = 3.085, p = 0.214). Results are displayed in Fig. 3.

Fig. 3.

Effect of increasing temperature on voluntary rate-dependent contractile properties. Statistical significance indicates p < 0.05. 95 %CI = 95 % confidence interval; AM = ante meridiem (before midday); CYC = cycle; ES = effect size; EXP = Experimental; HI = high intensity; HPS = high-pull specific; LL = lower limit of the confidence interval; IR = light-emitting diode irradiation; MI = moderate intensity; PM = post meridiem (after midday). WBV = whole body vibration; UL = upper limit of the confidence interval. Please refer to the respective studies for further information about the subgroups mentioned in parentheses.

3.4.2. Evoked contractile properties

3.4.2.1. Evoked maximum muscle strength

No significant effect was observed for evoked maximum muscle strength when considering all study designs together (–0.1 %/°C ± 0.8 %/°C, ES = –0.20 (95 %CI: –0.50 to 0.10), p = 0.190; I² = 52 %, p = 0.034; k = 9) or when EXP_WARM (–0.4 %/°C ± 1.0 %/°C, ES = –0.26 (95 %CI: –0.62 to 0.11), p = 0.165; I² = 52 %, p = 0.100; k = 4) and EXP_POST (0.6 %/°C ± 3.5 %/°C, ES = –0.15 (95 %CI: –0.70 to 0.39), p = 0.576; I² = 59 %, p = 0.044; k = 5) were analyzed separately. No significant difference was revealed between study designs (χ² = 0.167, p = 0.683). Results are displayed in Fig. 4.

Fig. 4.

Effect of increasing temperature on evoked maximum muscle force production capacity. Statistical significance indicates p < 0.05. CA60 = conditioning activity, 60°/s; CA300 = conditioning activity, 300°/s. CI = 95 % confidence interval; ES = effect size; EXP = Experimental; LL = lower limit of the confidence interval; UL = upper limit of the confidence interval. Please refer to the respective studies for further information about the subgroups mentioned in parentheses.

3.4.2.2. Evoked rate-dependent contractile properties

A significant effect was observed for evoked rate-dependent contractile properties when considering all study designs together (3.2 %/°C ± 1.5 %/°C, ES = 0.65 (95 %CI: 0.29 to 1.00), p < 0.001; I² = 33 %, p = 0.198; k = 5). There was also a significant positive effect observed when EXP_WARM (3.7 %/°C, ES = 1.28 (95 %CI: 0.63 to 1.80), p < 0.001; I² = 0 %, p = 1.000; k = 1) and EXP_POST (3.2 %/°C ± 3.1 %/°C, ES = 0.56 (95 %CI: 0.18 to 0.95), p = 0.004; I² = 0 %, p = 0.699; k = 3), but not EXP_FULL (2.4 %/°C, ES = 0.20 (95 %CI: –0.50 to 0.90), p = 0.577; I² = 0 %, p = 1.000; k = 1), were analyzed separately. No significant difference was revealed between study designs (χ² = 5.312, p = 0.070). Results are displayed in Fig. 5.

Fig. 5.

Effect of increasing temperature on evoked rate-dependent contractile properties. Statistical significance indicates p < 0.05. 95 %CI = 95 % confidence interval; ES = effect size; LL = lower limit of the confidence interval; UL = upper limit of the confidence interval.

3.5. Subgroup analyses and meta-regressions

No significant differences were detected between warm-up modalities (active vs. passive), warm-up test exercise specificity (specific vs. non-specific active warm-ups), or training status (trained vs. untrained subjects) on maximum muscle strength or rate-dependent contractile properties outcomes (χ² test p > 0.05). Linear meta-regressions indicated no significant effect of sex (% of men) on strength or rate-dependent contractile properties outcomes (regression slope p > 0.05); they also revealed no association between change in performance and change in temperature or baseline temperature (skin or muscle). Results for subgroup comparisons and meta-regressions are presented in Supplementary Material 5.

4. Discussion

The main finding of the present systematic review with meta-analysis was that increasing muscle temperature (within normal physiological ranges: 33.9°C–39.7°C), irrespective of the warm-up modality used (active, exercise-based vs. passive), had a significant positive effect on rate-dependent contractile function (3.5 %/°C ± 2.1 %/°C, ES: 0.28 to 0.65, voluntary and evoked pooled) but no effect on maximum muscle force production (–0.2 %/°C ± 0.6 %/°C, ES: –0.20 to 0.08, voluntary and evoked contractions pooled). Results demonstrate that increasing muscle temperature improves the RFD and power produced during maximal effort contractions.

4.1. Effect of increasing temperature on evoked contractile properties

Across all studies, irrespective of the type of warm-up, the present analyses revealed a significant increase in evoked rate-dependent muscle contractile function (3.2 %/°C ± 1.5 %/°C) but not maximum force production (–0.1 %/°C ± 0.8 %/°C) with increasing temperature. Such results are consistent with the results of within-study comparisons showing that elevated muscle temperature has a greater effect on rate-dependent contractile properties (i.e., muscle power, RFD, etc.^40,68) than maximum muscle strength²³ during voluntary contractions. The magnitude of performance enhancement following passive warm-up computed in the present analysis (∼3 %/°C) is consistent with the 2 %/°C–5 %/°C performance enhancement often reported with passive heating interventions.⁴⁹ Therefore, it is possible that the greater potential for performance enhancement following active warm-up in the present analysis (∼7 %/°C) could be explained by other phenomena such as post-activation potentiation ((PAP) the increase in muscle twitch force following prior activity⁶⁹), which may particularly enhance rate-dependent contractile function. Nonetheless, it is also plausible that the limited data sets available for comparisons (k = 9) and between-study differences in variables assessed (e.g., in stimulation intensity, frequency, and duration as well as the muscle tested) contributed to the lack of effect of increasing temperature on evoked maximum muscle strength following active warm-up in the present study. Consistent with this possibility is the moderate between-study variability computed across all contractile properties tested in the present analysis (I^2: 52 %–59 %). There were insufficient studies to perform sub-group analyses between different stimulation types or parameters, thus future research should examine the specific effect of temperature on parameters obtained during electrical stimulation to better understand the possible temperature-dependent effects. This is of particular importance for researchers who use evoked contractile properties as markers of muscle function or fatigue, but it is also relevant to exercise professionals who use electrical muscle stimulation techniques as training and testing tools in clinical populations.

4.2. Effect of increasing temperature on voluntary contractile properties

The present analyses revealed a significant improvement in voluntary rate-dependent muscle contractile function (3.7 %/°C ± 1.8 %/°C) but not maximum force production (–0.2 %/°C ± 0.9 %/°C) with increasing temperature, confirming previous within-study findings.^23,40,68 Contrary to our hypothesis, active, exercise-based warm-ups did not induce a greater performance enhancement than passive heating alone (active: 3.6 %/°C ± 1.1 %/°C vs. passive: 0.5 %/°C ± 1.4 %/°C for rate-dependent contractile properties, p = 0.284). There is evidence from individual studies that certain active warm-up protocols (i.e., task-specific with correct intensity and volume) are more effective than passive26, 27, 28, 29^,70 and other (sub-optimal) active warm-up protocols.^21,50 For example, within-study comparisons show a trend towards greater effects for active than passive warm-up protocols in studies included in the present analysis (i.e., ES = 0.27 vs. 0.0826, 27, 28, 29^,70). Yet, subgroup analyses did not reach statistical significance (i.e., active vs. passive, p = 0.093). Therefore, the lack of statistical significance may be explained by methodological limitations of the present meta-analytic approach (i.e., the inability to examine factors affecting performance across active warm-up programs, detailed later). Likewise, rate-dependent contractile properties included both isometric (RFD) and dynamic (power) variables, which were pooled in the present study. Thus, it remains a possibility that the effect of increased temperature might differentially influence these outcomes. Therefore, further work is required to better understand the effect of different warm-up modalities (active vs. passive) on voluntary contractile function. This includes exploring the mechanisms underpinning these effects to improve current warm-up practices in sporting, research, and clinical environments where comprehensive warm-ups, as described by Blazevich et al.,⁶⁹ are routinely utilized to optimize voluntary contractile function. Additionally, incorporating the recommendations by MacIntosh et al.³² and others on study designs (e.g., familiarization, randomization, blinding) will be crucial to enhance the robustness of future research. It is important to recognize that individual responses result from optimal and suboptimal combinations of all the factors involved (e.g., warm-up task, test, individual characteristics (sex, age, training status)⁷¹). This might explain the large inter-individual and between-study differences often reported. These aspects will be further discussed in the next section to provide a comprehensive understanding and to improve the design and application of warm-up protocols.

4.3. Confounding factors

4.3.1. Study design

A fundamental limitation of meta-analyses is the pooling of studies with different study designs and methods to quantify the overall effect of an intervention. For example, only 7 studies in the present review included pre- and post-intervention data for both experimental and control conditions (i.e., the ideal approach;⁷² EXP_FULL); thus, to increase sample size, studies that did not include a control condition (EXP_WARM: n = 5) or pre-intervention data for both the experimental and control conditions (i.e., post-intervention data only; EXP_POST: n = 21) were also included in the meta-analysis. Of note, in the 3 models, warm-up interventions were compared to a non-warm-up condition. The 3 study designs were also evaluated individually; however, this can be problematic as computing within-group (i.e., in EXP_WARM) differences can overstate the effect of an intervention, and significant results are easier to find than for between-group differences.^73,74 For example, in the analyses of evoked rate-dependent contractile function, a larger ES was observed for EXP_WARM (ES = 1.21, k = 1) and EXP_POST (ES = 0.56, k = 3) compared to EXP_FULL (ES = 0.20, k = 1). A comparison of ES between study designs in this case, however, may be misleading due to the limited number of studies.⁷⁵ Nonetheless, for this comparison and all others presented here, meta-regressions revealed that study design (EXP_FULL vs. EXP_WARM vs. EXP_POST) had no significant impact. Therefore, the data obtained by studies using all 3 study designs thus provided an overall effect that can be considered more representative of the extant literature (all designs pooled: n = 33; EXP_FULL only: n = 7). Discrepancies between study designs might reflect the lack of within- (EXP_POST) or between-session (EXP_WARM) control conditions, as forces or torques produced during voluntary contractions are inherently more variable76, 77, 78 and affected significantly more by non-temperature-related factors than evoked contractions.⁶⁹ Notably, larger SD were reported by studies that employed EXP_POST and EXP_WARM conditions than those that employed EXP_FULL.

4.3.2. Type of warm-up and warm-up specificity

The degree of voluntary performance change observed in the present analysis ranged from –10.3 %/°C⁷⁹ to +18.5 %/°C⁵⁷ across studies. Speculatively, this variation may be attributed to factors such as the type of warm-up used (active vs. passive⁸⁰) and warm-up test exercise specificity.²¹ However, there were no significant differences in performance enhancement between active, exercise-based warm-ups and passive heating protocols in the present analysis (ES = 0.26 vs. 0.06, p = 0.093, maximum muscle strength and rate-dependent contractile properties pooled). This finding was unexpected as active warm-ups often include the extensive practice of the performance test,^21,81 which should optimize neural activation patterns^20,21 and thus enhance muscle function via non-temperature-related factors, such as increased neural drive^20,21, water/blood influx⁸², and/or increased motivation/arousal.²³ Thus, the non-temperature-related factors associated with active warm-up should theoretically provide an additional performance enhancement separate from the temperature effect (i.e., larger effect for active compared to passive warm-up protocols, ES = 0.27 vs. 0.0826, 27, 28, 29^,70). For example, in the study of Barnes et al.,⁵⁰ a high pull-specific warm-up elicited a considerably larger effect (ES = 0.93) than whole-body vibration or cycling warm-ups (ES = 0.10 and 0.14, respectively) despite invoking a smaller increase in temperature. Speculatively, the greater effect of any active warm-up condition over others might be explained by a greater motor pattern similarity leading to higher or more effective muscle recruitment,^19,21,69 or at least minimizing motor pattern interference effects (i.e., reduced and/or altered muscle activation pattern^21,69). Alternatively, the intent to produce large or rapid forces^19,20 or the use of optimum (especially non-fatiguing⁸³) exercise volumes might affect warm-up efficacy. Such possibilities highlight the potential influence of non-temperature-related processes on muscle contractile function and may partly explain the lack of statistical significance in the present subgroup analyses (active vs. passive). Given these possibilities, pooling studies using various active warm-up strategies does not allow for a detailed examination of the factors affecting performance across active warm-up programs, although it did allow for confirmation that its effects did not meaningfully affect the outcomes of the present analyses.

Intriguingly, non-specific warm-up protocols tended to induce greater performance enhancements than specific warm-up protocols in the present analysis (ES = 0.40 vs. 0.21, p = 0.092). This is contrary to our hypothesis as well as the findings of previous work on warm-up test exercise specificity.^20,21,69 This result may speculatively be explained by the way in which “specific” was defined in the present study (i.e., specific = includes trial repetitions of the performance test; e.g., knee extension warm-up before knee extension test). This binary approach (specific or non-specific) does not consider the warm-up design (i.e., intensity, volume, recovery) in the analysis. Thus, it is plausible that various “specific” warm-up protocols included in the present analysis were sub-optimal, induced fatigue and/or a motor pattern interference, and had a limited (or negative) effect on performance.

Given the lack of well-controlled, quality research on this topic, the vast range of types (active vs. passive) and configurations (intensity, duration, recovery, specificity, etc.) of warm-up protocols must be explored under ideal experimental conditions (i.e., EXP_FULL) to better understand the effects of different warm-up modalities as this will have a significant impact on warm-up practices in sporting, clinical, or research environments where optimizing muscle contractile function is critical.

4.3.3. Temperature measurement method and baseline temperature

Intramuscular temperature recording is considered the most valid and accurate measure of muscle temperature;⁸⁴ however, only 8 studies in the present analysis used intramuscular temperature measurements. Therefore, 17 studies using skin temperature as a proxy for muscle temperature were also included to increase sample size while reflecting current research and clinical practices.⁸⁵ Of particular interest, meta-regressions did not reveal an effect of temperature measurement method in the present study (skin vs. muscle, p: 0.175–0.521), indicating that the relative changes in skin temperature may be associated with changes in muscle temperature, even though the absolute temperature values will differ. This is likely explained by the muscle temperature range in the included studies (33.9°C–39.7°C). However, care must be taken when extrapolating these findings as (a) skin and muscle temperatures seem to have different time kinetics;⁸⁶ (b) beyond this temperature range, the relationship between changes in skin temperature and muscle temperature dissociates with greater temperature changes (i.e., greater changes in skin temperature with extreme environmental temperatures; e.g., cooling < 30°C and whole-body heating > 38°C^87,88); and (c) after passive heating strategies, muscle temperature varies depending on the assessment depth and also depends on participant's fat and muscle thicknesses,⁸⁹ factors which do not affect the skin temperature. Moreover, it is important to note that the timing of evaluations was not specifically considered in this analysis, meaning comparisons between studies could have included post-warm-up measurements performed between 0 and 15 min following the cessation of exercise, which could potentially influence our findings. However, we did not specifically consider the timing of evaluations in our analysis for several reasons. Firstly, incorporating the timing of evaluations as an additional variable would have further limited our already constrained sample size. Secondly, we faced challenges in determining an appropriate cutoff point for the timing of evaluations given the variability in protocols and time points across the studies included in our analysis. Thirdly, not all studies clearly reported the timing of assessment. Establishing a standardized cutoff without clear consensus could introduce additional biases or inconsistencies. The timing of post-exercise temperature measurements needs further study to clarify the impact of increasing muscle temperature on performance outcomes.

Meta-regression also did not reveal an effect of baseline (starting) temperature. That is, the effect of increasing muscle temperature on performance (when an effect was observed) was similar regardless of the starting temperature. However, the temperature-dependence of skeletal muscle is non-linear,90, 91, 92 with the most significant functional changes occurring outside normal physiological ranges (i.e., <28°C).⁹³ For example, De Ruiter et al.²⁴ reported that Q₁₀ (a measure of temperature-dependence⁹⁴) for maximal muscle power production was 2.0 in the range 31.4°C–37.1°C but 6.9 in the range 22.2°C–25.6°C, indicating an increase in temperature-dependence at lower temperatures. Therefore, the lack of effect of baseline temperature observed in the present analysis is likely explained by our decision to exclude cooling studies from the analysis and thus examine only normal physiological temperatures (muscle: 36.1°C ± 0.9°C; skin: 32.0°C ± 2.5°C).

4.3.4. Subject-specific characteristics

Previous meta-analyses^36,42 have reported that the magnitude of performance enhancement observed following warm-up is strongly influenced by subject-specific moderators such as training status. However, our subgroup analyses detected no significant difference in the effect of temperature between trained (ES = 0.30) and untrained (ES = 0.11) subjects (p = 0.128). Moreover, meta-regressions did not reveal an effect of sex (p = 0.274) on the effect of increasing temperature. These findings differ from those of previous meta-analyses,^36,42 which might speculatively be explained by the way in which “trained” status is defined (e.g., a minimum of 6 consecutive months of dedicated training vs. >1 year⁴² or >2 years³⁶). Further, our definition did not differentiate between resistance or aerobic training due to poor statistical power associated with the small sample size, so our analysis did not factor in the effect of training modality (i.e., fiber type changes, fatigability, etc.) on performance outcomes. Moreover, the present study was limited to healthy adults (18–60 years), thus the potential role of age on the warm-up effect remains unknown and should be the target of future investigation. The lack of sex effect in our analysis might be explained by the small sample size of studies involving female subjects (i.e., men-only = 18 studies vs. women-only = 5 studies). Therefore, given the rise in global professionalism and participation of female athletes⁹⁵ and the impact of warm-up on sporting performance,^96,97,98,99 future research should explicitly explore the effects of moderating variables such as sex and training status in well-designed and well-controlled studies.

4.3.5. Confounding factors – conclusion

Contrary to our hypotheses, the present analyses demonstrated no significant effects of moderating variables, supporting our decision to pool the data from studies with different study designs and methods to provide an overall effect that is more representative of the extant literature. However, considerations still need to be given to the factors that may potentially moderate the effect of increasing temperature on performance to gain a better understanding of the topic. Therefore, future research that explicitly investigates the effect of moderating variables using ideal study designs (i.e., randomized controlled trials) is needed. This will ultimately improve warm-up practices in both practical and research settings.

5. Conclusion

Increasing muscle temperature, irrespective of the warm-up modality (active vs. passive), significantly and positively affects rate-dependent contractile function (e.g., force production in high-speed tasks and RFD) but has no effect on maximum muscle force production in both evoked and voluntary contractions. Subgroup analyses revealed that active, exercise-based warm-ups had no greater effect on contractile function enhancements than passive heating, potentially indicating that much of the warm-up effect could be attributed to increasing temperature rather than non-temperature-related factors. However, several previous studies have reported additional benefits of active warm-ups, and factors such as movement pattern or movement speed similarity between warm-up and test exercises, or warming up with the necessary intent (effort) while minimizing fatigue may be important to extract the greatest benefit from active warm-ups; the pooling of data within meta-analysis prevented exploration of such possibilities. Important to the present analyses, meta-regressions indicated no effect of study design (EXP_FULL vs. EXP_WARM vs. EXP_POST), temperature measurement method (muscle vs. skin), baseline temperature (within normal physiological ranges), or subject-specific factors such as sex or training status on analysis outcomes. However, data sets from studies using less robust designs were included in the categorical and meta-regression analyses due to limited data sets available using the ideal study design (EXP_FULL). Importantly, pooling studies with a vast range of warm-up test exercise configurations and methodological differences may have contributed to the lack of significant effect for many of the moderating factors. Thus, it is recommended that more research explicitly tests the effects of these moderating factors using well-controlled study designs to provide a greater wealth of interrogatable data to ultimately improve current warm-up practices.

Author's contributions

CJW was responsible for the conception and design of the review, literature search, screening of studies, data extraction, methodology quality assessment, and drafting the article; JPN assisted with literature screening and data extraction, and conducted the statistical analyses; AJB contributed to the conception and design of the review, and provided supervision and guidance. All authors contributed to the interpretation of the data, provided critical revisions, contributed to the intellectual content, and approved the final version. 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 interests.