Stretching and foam rolling are commonly used in sports practice as part of a warm-up routine. It is well known that both treatments can increase the range of motion (ROM) acutely (Konrad et al., 2019; Konrad and Tilp, 2020a; 2020b; Nakamura et al., 2021). According to a recent meta-analysis (Wilke et al., 2020), the magnitude of the effect on ROM following stretching and foam rolling is similar. Thus, when the goal is to increase ROM, both stretching and foam rolling can be considered as adequate warm-up routines. However, study findings about the acute effects of a single application of stretching or foam rolling on performance parameters are not as unambiguous. While a single static stretching exercise with a duration of ≥60 s likely causes a pronounced impairment in performance (-4.6%), shorter stretching durations (<60 s) show only minor changes (-1.1%) (Behm et al., 2016). However, if dynamic stretching is applied, Behm et al. (2016) reported mean increases in performance of 1.3%. Thus, performance changes following stretching depend on the stretch duration and stretching technique (Behm et al., 2016; Behm and Chaouachi, 2011; Behm et al., 2021a; Kay and Blazevich, 2012), and are also likely dependent on the muscles stretched (Konrad et al., 2021 . For foam rolling, there seems to be, at least, no detrimental effect on performance (Wiewelhove et al., 2019; Cheatham et al., 2015). Wiewelhove et al. (2019) even reported in their meta-analysis a tendency of improvement in sprint performance but no change in muscle strength. Thus, a single foam rolling exercise might be a practical approach for acutely increasing ROM while expecting an increased or at least stable athletic performance.

Less information is available on the combined effect of stretching and foam rolling. A recent review (Anderson et al., 2021) compared dynamic stretching to the combined effect of a foam rolling treatment and dynamic stretching of the hamstrings on ROM and performance parameters, and included four studies in their analysis. The authors concluded that the combined treatment had only a slight additional effect on ROM when compared to dynamic stretching alone. For performance, two out of the four included studies reported a greater increase in jump height with the combined treatment compared to dynamic stretching alone. Moreover, for agility, two studies (out of the three that assessed agility) reported that the combined effects improved agility to a greater extent than dynamic stretching alone (Anderson et al., 2021). There is therefore some evidence that the combination of foam rolling with dynamic stretching has an accumulative effect on the various parameters (e.g., jump height and agility) compared to dynamic stretching alone. However, these results do not include a quantification of the effects based on a meta-analysis or include muscle groups other than the hamstrings. Moreover, since static stretching and dynamic stretching can lead to different acute changes in ROM (Amiri-Khorasani et al., 2011) and performance (Behm and Chaouachi, 2011; Behm et al., 2021a), vibration foam rolling might have a more pronounced effect on ROM compared to foam rolling without vibration (Wilke et al., 2020), so these two modalities should also be considered.

Hence, the purpose of this meta-analysis was to compare the combined effect of a foam rolling (vibration and non-vibration foam rolling) and stretching exercise (including all techniques) to that of stretching or foam rolling alone on both ROM and physical performance. A further goal was to distinguish between the effects of the stretching technique (static stretching, dynamic stretching), the type of foam rolling (vibration foam rolling, non-vibration foam rolling), the tested muscles (hamstrings, quadriceps, triceps surae, hip, shoulder), and the order of the combined treatment (either foam rolling followed by stretching, or vice versa) by the use of a subgroup analysis.

This review was conducted according to the PRISMA guidelines and the suggestions from Moher et al. (2009) for systematic reviews with meta-analysis.

An electronic literature search was performed in PubMed, Scopus, and Web of Science. The search period ranged from 1990 until the 15th February 2021. The keywords for the online search were (“foam rolling” OR “self-myofascial release” OR “roller massage” OR “foam roller”) AND (stretch*), and they were the same for all the databases. The systematic search was done by three independent researchers (AK, MN, DB). In the first step, all the hits were screened by their abstract. If the content of a study remained unclear, the full text was screened to identify the relevant papers. Following this independent screening process, the researchers compared their findings. Disagreements were resolved by jointly reassessing the studies against the eligibility criteria. Overall, 169 papers were screened, from which nine papers were found to be eligible for this review. However, following the additional search of the references (search through the reference list) and citations (search through Google Scholar) of the nine already included papers, three more papers were identified as being relevant. Therefore, in total, 12 papers were included in this systematic review and were used for the meta-analysis. The whole search process is depicted in Figure 1.

This review considered studies that compared the combined effects of an acute bout of both stretching and foam rolling on ROM and/or performance parameters (e.g., strength, jump height) to the effects of foam rolling or stretching alone in healthy participants. We included studies in the English, German, and Japanese languages with crossover (pre- to post-comparison or post-comparison) or parallel group (pre- to post-comparison) designs. However, we excluded conference papers and theses.

From the included papers, the characteristics of the participants, the sample size, the study design, the characteristics of the intervention (e.g., stretching technique, vibration foam rolling vs. non-vibration foam rolling, duration), and the results of the main variables (ROM and/or performance parameters) were extracted. For the main variables, either the pre- and post-values (plus standard deviations) or the post-values (plus standard deviations) of the combined groups (stretching plus foam rolling) and the single intervention groups (foam rolling or stretching, but also from the control group,) were extracted. If the required data were missing, the authors of the studies were contacted via email.

The meta-analysis was performed using Comprehensive Meta-Analysis software, according to the recommendations of Borenstein et al. (2009). By the use of a random-effect meta-analysis, we assessed the effect size of the ROM in terms of the standardized mean difference between the combined effects (foam rolling and stretching together) and stretching alone, between the combined effects and foam rolling alone, and between the combined effects and a control condition. Due to the smaller number of available studies for the performance parameters and the restriction that a minimum of three studies was necessary to perform a meta-analysis, we could only assess the effect size of the combined effects, compared to stretching alone. Moreover, by using a mixed-effect model, we performed subgroup analyses with the stretching technique (static stretching, dynamic stretching), the type of foam rolling (vibration foam rolling, non-vibration foam rolling), the tested muscles (hamstrings, quadriceps, triceps surae, hip, shoulder), and the order of the combined treatment (either foam rolling followed by stretching or stretching followed by foam rolling) in both the analyses of the ROM and performance parameters. In addition, for the performance parameters, we also performed a subgroup analysis for the type of task (strength, jump height, sprinting). A minimum of two effect sizes per subgroup was necessary to perform a subgroup analysis. To determine if there were differences between the effect sizes of the subgroups, Q-statistics were applied (Borenstein et al., 2009). According to the recommendations of Hopkins et al. (2009), we defined the effects for a standardized mean difference of <0.2, 0.2–0.6, 0.6–1.2, 1.2–2.0, 2.0–4.0, and >4.0 as trivial, small, moderate, large, very large, and extremely large, respectively. I² statistics were calculated to assess the heterogeneity among the included studies, and thresholds of 25%, 50%, and 75% were defined as having a low, moderate, and high level of heterogeneity, respectively (Behm et al., 2021b; Higgins et al., 2003). An alpha level of 0.05 was defined for the statistical significance of all the tests

The methodological quality of the included studies was assessed using the PEDro scale. In total, 11 methodological issues were assessed by two independent researchers (AK, MN) and assigned with either one or no point. Hence, studies with a higher score represent higher methodological quality. If any conflict between the ratings of the two researchers was found, the methodological issues were reassessed and discussed. Moreover, the Egger’s regression intercept test was applied to detect possible publication bias.

In total, 12 studies compared the effects of a combined treatment (foam rolling plus stretching or stretching plus foam rolling) with the effects of stretching or foam rolling alone on ROM. Within these 12 studies, seven studies compared the effects of a combined treatment with stretching on performance parameters. Overall, 17 effect sizes could be extracted for the ROM parameters and 24 for the performance parameters. In summary, 267 participants (143 males and 124 females) with a mean age of 22.9 (±5.1 years) participated in the included studies. Out of the 267 participants 141 were athletes, 76 were physically active, and six were sedentary. The activity level of the remaining 44 was not defined. Table 1 presents the characteristics and outcomes of the 12 studies.

The Egger’s regression intercept test indicated that no reporting bias was likely for the meta-analysis dealing with ROM and for the comparison of the combined treatment with stretching (intercept -0.59; P = 0.31), foam rolling (intercept 0.78; P = 0.81), or the control condition (intercept 0.32; P = 0.91). When comparing the effect on performance parameters between the combined treatment and stretching alone, the Egger’s regression intercept test indicated a potential for reporting bias (intercept -1.51; P = 0.00). The average PEDro score value is 6.92 (±1.31), indicating a low risk of bias (Maher et al., 2003; Moran et al., 2021). The two assessors agreed for 125 out of the 132 criteria (12 studies × 11 scores). The mismatched outcomes were discussed and the assessors finally agreed on the scores presented in Table 2.

The meta-analysis with 17 effect sizes from 12 studies revealed no significant difference in ROM changes between the combined condition and the stretching condition alone (ES = 0.032; Z = 0.517; CI (95%) -0.090 to 0.155; P = 0.61; I² = 0.00) (see Figure 2). Fourteen out of the 17 effect sizes allowed us to calculate pre- to post-changes. The remaining three effect sizes were based on post-values only. The combined condition and the stretching condition alone (pre- to post-comparison) showed an average increase in ROM of 6.83% (CI (95%) -0.34% to 14.00%) and 5.26% (CI (95%) -1.59% to 12.10%), respectively. None of the subgroup analyses, including the stretching technique (static stretching, dynamic stretching) (Q = 0.01; P = 0.92), the type of foam rolling (vibration foam rolling, non-vibration foam rolling) (Q = 0.05; P = 0.83), and the order of the combined treatment (Q = 0.66; P = 0.42), revealed significant differences in ROM. However, we did not perform a subgroup analysis with the tested muscles, since there were subgroups with only one effect size.

The meta-analysis with six effect sizes from six studies revealed no significant difference in ROM changes when the combined condition was compared to the foam rolling condition alone (ES = -0.225; Z = -1.182; CI (95%) -0.597 to 0.148; p = 0.24; I² = 62.89) (see Figure 3). The combined condition and foam rolling condition alone (pre- to post-comparison) showed an average increase in ROM of 13.77% (CI (95%) -0.61% to 28.16%) and 7.19% (CI (95%) 2.48% to 11.89%), respectively.

The subgroup analyses differentiating between different stretching techniques revealed no significant difference in ROM (Q = 0.58; p = 0.45). A subgroup analysis including the type of foam rolling or the order of the combined treatment was not possible since only non-vibration foam rolling studies and studies where stretching was followed by foam rolling were part of this meta-analysis. However, we did not perform a subgroup analysis with the tested muscles since there were subgroups with only one effect size.

The meta-analysis with six effect sizes from three studies revealed a significantly higher increase of a small magnitude in ROM in the combined condition compared to the control condition without any intervention (ES = -0.332; Z = -2.499; CI (95%) -0.593 to -0.072; p = 0.012; I² = 32.38) (see Figure 4). The six effect sizes of the combined condition and the control condition (pre- to post-comparison) showed an average change of 1.51% (CI (95%) -4.74% to 7.76%) and 0.21% (CI (95%) -1.16% to 1.59%), respecttively. The subgroup analyses of the stretching technique (Q = 0.78; p = 0.38), the tested muscles (Q = 1.44; p = 0.49), and the order of the combined treatment (Q = 0.78; p = 0.38) revealed no significant difference in ROM changes. A subgroup analysis with the type of foam rolling was not possible since only non-vibration foam rolling studies were part of this meta-analysis.

A meta-analysis of the effects on performance was only possible with the combined condition and the stretching condition alone since there were insufficient studies and effect sizes for the foam rolling condition available.

The meta-analysis with 24 effect sizes from seven studies revealed no significant difference in performance parameters between the combined condition and the stretching condition alone (ES = -0.029; Z = -0.503; CI (95%) -0.142 to 0.084; p = 0.62; I² = 0.00) (see Figure 5). Ten out of the 24 effect sizes allowed us to calculate pre- to post-changes. The remaining 14 effect sizes were based on post-values only. The combined condition and stretching condition alone (pre- to post-comparison) showed changes in performance of -0.41% (CI (95%) -3.02% to 2.19%) and -0.07% (CI (95%) -3.34% to 3.22%), respectively.

By comparing the order of the combined treatment (either foam rolling followed by stretching or vice versa), the subgroup analysis revealed a significant difference between the subgroups of “foam rolling followed by stretching vs. stretching alone” and “stretching followed by foam rolling vs. stretching alone” (Q = 5.53; p = 0.02). While stretching followed by foam rolling showed a similar magnitude of change as stretching alone (ES = 0.10; p = 0.21), the foam rolling followed by stretching exercise revealed a significant but trivial better effect than stretching alone (ES= -0.17; p = 0.04) (see Figure 5). Moreover, further subgroup analyses of the stretching technique (Q = 0.11; p = 0.74), the type of foam rolling (Q = 3.21; P = 0.07), and the type of task (strength, jump height, sprinting) (Q = 0.82; p = 0.66) revealed no significant difference in performance. However, we did not perform a subgroup analysis of the tested muscles since there were subgroups with only one effect size.

The purpose of this review was to compare the effects of an acute bout of a combined treatment of stretching and foam rolling to the effects of stretching or foam rolling alone on both ROM and performance parameters in healthy subjects. For ROM, the meta-analysis revealed the superior effect of a combined treatment compared to the control condition (no stretching or foam rolling); however, no superior effect was found for the combined treatment compared to either stretching or foam rolling alone. In addition, further subgroup analyses of ROM (e.g., stretching technique, muscle groups tested) showed no differences between the modalities. With regard to performance, the meta-analysis revealed no difference between the combined treatment compared to stretching or foam rolling alone; however, the subgroup analysis of performance revealed the trivial but superior effect of the combined treatment compared to stretching alone, but only if the foam rolling was followed by stretching.

Three meta-analyses were performed for ROM. Thereby, the combined effects (stretching and foam rolling) were compared to a control condition, stretching alone, and foam rolling alone. The combined treatment had a superior effect on ROM compared to the control condition. Increases in ROM following a single stretching (Behm et al., 2016; Behm et al., 2021a; Behm and Chaouachi, 2011) or foam rolling treatment (Wilke et al., 2020) were reported by several studies. However, it is not clear if the increases in ROM have an accumulated effect when foam rolling and stretching are performed within one training session, compared to foam rolling or stretching alone. Anderson et al. (2021) reported in their review only a small additional effect when comparing a combined treatment with foam rolling and dynamic stretching to dynamic stretching alone. However, the authors did not perform a meta-analysis or include the static stretching technique, or consider other muscles than the hamstrings. Our meta-analysis included static stretching and different muscles, but neither showed any superior effect for a combined treatment compared to stretching alone or foam rolling alone.

We would have expected that a change in tissue compliance (e.g., muscle or tendon stiffness), which has been reported following an acute bout of stretching (Kato et al., 2010; Kay et al., 2015; Konrad et al., 2017), and the additional changes in stretch or pain tolerance following foam rolling (Nakamura et al., 2021), might have led to greater gains in ROM compared to stretching or foam rolling alone. However, this was not observed in this meta-analysis. A possible reason for the lack of an additional effect in the combined treatment might be a saturation effect for ROM, which has also been observed following a certain duration of stretching (Mizuno, 2019) and foam rolling (Nakamura et al., 2021). Mizuno (2019) reported a similar amount of increase in ROM when comparing 10 s of static stretching with 100 s. Moreover, Nakamura et al. (2021) reported no further changes in ROM following a single foam rolling treatment applied for 300 s compared to 90 s. Hence, if the single treatment of stretching or foam rolling exceeds a certain duration, no additional changes in ROM can be expected. The duration range of the combined treatment of the included studies when foam rolling was combined with static stretching was between 40 s and 210 s (mean: 134.0 ± 77.9 s). Hence, the assumed saturation following a combined treatment for ROM is not unlikely. According to the results of our meta-analysis, it can be recommended that foam rolling (with a duration range between 30 s and 120 s; mean 66.0 ± 36.9 s) or stretching (with a duration range between 10 s and 90 s; mean 65.7 ± 28.8 s) could be performed to increase the ROM of a joint acutely, rather than a more time-consuming combined treatment This goes in line with previous recommendations on foam rolling (Behm et al., 2020) or stretching (Behm, 2018). Wilke et al. (2020) reported in a recent meta-analysis that foam rolling and stretching have a similar magnitude of change on ROM, so that athletes could apply either stretching or foam rolling to increase ROM, according to their preference. However, several studies have reported enhanced recovery and reduced delayed-onset muscle soreness (DOMS) following foam rolling (MacDonald et al., 2014; Nakamura et al., 2020; Pearcey et al., 2015), but not following a stretching exercise (Afonso et al., 2021; Henschke 2011). Therefore, if a secondary goal is to reduce DOMS, besides the increase in ROM, foam rolling is likely the better treatment.

With regard to the effects on performance parameters, due to the lack of studies and effect sizes, a comparison was only possible between the combined treatment and a stretching treatment alone (but not foam rolling or a control condition). The meta-analysis revealed no significant changes between the combined treatment and stretching alone. This is in contrast to the review by Anderson et al. (2021), who concluded that foam rolling and dynamic stretching might have a superior effect compared to dynamic stretching alone. The authors reported that two out of four studies found a greater increase in vertical jump height compared to dynamic stretching alone. Moreover, two out of three studies included in their review reported a greater effect on agility following a combined treatment (foam rolling and dynamic stretching) compared to dynamic stretching alone. However, Anderson et al. (2021) only included studies with dynamic stretching and studies assessing ROM with a sit and reach test. Consequently, several studies that investigated the combined effects on performance parameters with other stretching techniques and other muscles were excluded. In the present meta-analysis, we analyzed 12 studies in total, including seven studies of dynamic stretching and five of static stretching. Several reviews reported major differences in the effects between static and dynamic stretching if applied prior to a sports event. Static stretching with a long duration (≥60 s) likely causes a decrease in performance, while dynamic stretching can lead to an increase in performance (Behm et al., 2016; Behm et al., 2021a; Behm and Chaouachi, 2011). A recent meta-analysis of the acute effects of foam rolling on performance (Wiewelhove et al., 2019) reported a tendency of improvement (P = 0.06) in sprint performance (+0.7%), but negligible effects on jump or strength performance. This is in accordance with another review (Cheatham et al., 2015) that reported that a single bout of a foam rolling exercise likely does not induce changes in performance parameters. Therefore, similar to the results of Anderson et al. (2021), we expect that a combination of a foam rolling and dynamic stretching treatment will likely lead to a greater increase in performance than dynamic stretching alone. However, our subgroup analysis of the different stretching techniques (static stretching, dynamic stretching) did not reveal a significant effect of the combined treatment including dynamic stretching compared to dynamic stretching alone. Hence, according to our meta-analysis, dynamic stretching alone can lead to a similar performance changes as the combined treatment. Therefore, foam rolling does not necessarily have to be included if the goal is to increase performance. Similar to the results for the dynamic stretching group, no difference between the combined treatment and the static stretching treatment alone was shown by the meta-analysis.

Further subgroup analyses showed no significant differences between the different muscles tested (e.g., quadriceps, hamstrings), the type of foam rolling (vibration foam rolling, non-vibration foam rolling), or the type of task (e.g., strength, power). However, a significant difference could be detected between the orders of the combined treatment. Stretching followed by foam rolling showed a similar magnitude of change on performance as stretching alone; however, the foam rolling followed by stretching exercise revealed a significantly but trivial better effect than stretching alone (see Figure 5). A possible mechanism for why foam rolling before stretching can lead to a favorable effect compared to stretching alone is unclear and should be investigated in future studies.

It can be concluded that athletes under time constraints do not have to combine stretching with foam rolling in order to increase their ROM because the combination does not lead to any additional effects. However, if the goal is to also increase performance (e.g., strength, speed), the combination of foam rolling followed by stretching (but not vice versa) should be favored compared to stretching alone.