Hypoalgesia after aerobic exercise in healthy subjects: A systematic review and meta-analysis

ABSTRACT

Exercise-Induced Hypoalgesia (EIH) refers to an acute reduced pain perception after exercise. This systematic review and meta-analysis investigated the effect of a single aerobic exercise session on local and remote EIH in healthy individuals, examining the role of exercise duration, intensity, and modality. Pressure pain thresholds (PPT) are used as the main measure, applying the Cochrane risk of bias tool and GRADE approach for certainty of evidence assessment. Mean differences (MD; Newton/cm²) for EIH effects were analysed. Thirteen studies with 23 exercises and 14 control interventions are included (498 participants). Most studies used bicycling, with only two including running/walking and one including rowing. EIH occurred both locally (MD = 3.1) and remotely (MD = 1.8), with high-intensity exercise having the largest effect (local: MD = 7.5; remote: MD = 3.0) followed by moderate intensity (local: MD = 3.1; remote: MD = 3.0). Low-intensity exercise had minimal impact. Neither long nor short exercise duration induced EIH. Bicycling was found to be effective in eliciting EIH, in contrast to the limited research observed in other modalities. The overall evidence quality was moderate with many studies showing unclear risk biases.

KEYWORDS:

• Exercise induced hypoalgesia
• exercise
• pressure pain threshold
• algometer
• pain
• intensity

1. Introduction

Exercise induced hypoalgesia (EIH) describes the phenomenon in which a single bout of physical exercise can lead to a reduction in pain perception afterwards (Rice et al., 2019). EIH has been scientifically studied for more than 40 years (Black et al., 1979). Meanwhile, the growing interest in the potential analgesic effect of exercise has led to numerous studies investigating this relationship (Vaegter & Jones, 2020).

Pain perception can be quantified semi-objectively using, among others, experimentally measured pain sensitivity or pain tolerance (Böing-Meßing et al., 2022; Koltyn, 2000). Pain sensitivity, whether measured at baseline or following physical exertion, can be assessed through the quantification of the individual’s pain threshold, defined as the minimal intensity of a noxious stimulus perceived as painful. Pain tolerance, on the other hand, measures the maximum intensity of a noxious stimulus that the subject is willing to tolerate (Vaegter & Jones, 2020). Various types of noxious stimuli, including thermal (e.g., heat, cold), chemical (e.g., capsaicin), and mechanical (e.g., pressure), can be employed for this purpose, while the utilization of mechanical pressure pain thresholds (PPT) is recommended for EIH (Naugle et al., 2012).

In one of the first meta-analyses, Naugle and colleagues examined the acute effect of three types of exercise (isometric, aerobic, or dynamic resistance) on pain perception in healthy participants. EIH was consistently observed across all types of exercise with effects ranging from small to large depending on the pain induction method and exercise protocol. Within aerobic exercise studies, a large mean effect was observed on PPT (d = 0.84), albeit only two studies were considered. In patients with chronic pain, effects were more variable and appeared to depend on the clinical pain condition (Naugle et al., 2012). These results were later supported in the review by Rice and colleagues, who state that a single bout of aerobic or resistance exercise typically leads to EIH in healthy participants, while it is still more variable in chronic pain populations and might be impaired in certain patients (Rice et al., 2019). The most current evidence on a meta-analytical level is provided by Wewege & Jones including only randomized controlled studies. They reported that, within healthy participants, aerobic exercise caused EIH with large effect sizes (Hedges’g = −0.85 [CI = −1.58, −0.13]; 7 studies), while resistance exercise caused small EIH (Hedges’g = −0.45 [CI = −0.69, −0.22]; 2 studies), and isometric exercise (Hedges’g = −0.16 [CI = −0.36, 0.05]; 3 studies) did not cause any EIH (Wewege & Jones, 2021).

Both aerobic and strength exercises are recommended by many guidelines for patients suffering from pain syndromes (e.g., Bannuru et al., 2019; Häuser et al., 2017; Hinman et al., 2023; Oliveira et al., 2018). Yet, none of these before mentioned analyses on EIH and, to the best of our knowledge, no other published meta-analysis aimed to provide a precise understanding of the optimal exercise regime (e.g., duration, intensity, and modality) to induce EIH. From the therapist’s point of view, this is of particular relevance to design and/or to “prescribe” the exercise as precise as possible to induce pain relieving effects. Even though aerobic exercise seems to be the exercise type yielding most robust EIH effects (Wewege & Jones, 2021), potential load-specific dose-response relationships are still discussed. To provide any recommendations, it is necessary to gain a basic applied physiological understanding of EIH in healthy individuals. Despite research conducted in single studies (e.g., Hoffman et al., 2004, Naugle et al., 2014, Tomschi et al., 2023, Tomschi et al., 2022, Vaegter et al., 2017), there is only little knowledge on a meta-analytical level about key variables regarding specific training load parameters for aerobic training, particularly exercise intensity and exercise duration, and their potential dose-response relationship as well as the exercise modality (e.g., running or cycling). It is further discussed whether EIH effects occur solely locally close to the working musculature and, or to what degree also remotely, i.e., away from the working musculature (Jones et al., 2017; Rice et al., 2019; Samuelly-Leichtag et al., 2018; Wu et al., 2022). A nuanced understanding of how these influencing factors affect EIH in healthy individuals is essential for the understanding of the “optimal” aerobic exercise regime to induce EIH and for providing specific recommendations for exercise as a potential pain-relieving therapy option.

In light of the aforementioned considerations, this meta-analysis aims to undertake a comprehensive assessment of the effects of acute aerobic exercise on EIH in healthy individuals. The primary focus of this study is to analyse and synthesize the influence of key exercise variables in order to gain a nuanced understanding of the relationship between aerobic training and pain perception. Thus, the following research questions are stated: (1) To what extent does an acute aerobic exercise result in local and remote EIH in healthy individuals? (2) How do exercise variables such as intensity, duration, and modality impact the outcomes of EIH following an acute aerobic exercise bout?

2. Materials and methods

2.1. Identification and selection of studies

A systematic review and meta-analysis of the existing scientific literature was conducted based on the guidelines recommended by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (Liberati et al., 2009). The study selection process, methodological quality assessment, and data analysis were performed by two independent investigators (AS, MS). This meta-analysis was prospectively registered on PROSPERO (ID: CRD42023413169).

From April to May 2023, the online databases PUBMED (Medline) and Web of Science were used for search of relevant articles. The search was updated in March 2024 (from 1 May 2023 up to 22 March 2024) to reflect the current body of literature. The search strategy was developed on the basis of the following PICOS selection criteria (Methley et al., 2014). Population (P): Healthy male and female subjects of any age or ethnicity without the presence of acute, sub-acute or chronic pain, or any other chronic disease (e.g., cardiovascular, respiratory, neurological, endocrine, mental, metabolic, or cancer); Intervention (I): Single aerobic exercise session including running, cycling, swimming, walking, rowing, or skating, lasting at least 5 min; Control (C): Inclusion of an adequate control condition (e.g., quiet rest or sham exercise); Outcome (O): Experimental pain outcomes in a semi-objective manner based on PPT; Study design (S): Crossover designs (n > 10) and controlled parallel group designs (n > 10 per group) with pain assessment immediate before and acutely following exercise. Studies with a multiple-week/-session design and studies using solely NRS/VAS scores or questionnaires were excluded. Additional measurement time points after the intervention period were not considered. Reviews and meta-analyses were excluded from analysis, yet checked for additional, possibly relevant literature, as were the reference sections of studies deemed suitable with regard to the inclusion criteria. Only human studies and articles published in English were considered. Initially, studies employing any type of experimental pain assessment (i.e., mechanical, thermal, chemical) were included. However, because most articles investigated mechanical stimuli to capture EIH, the assessment of PPT has been recommended in this context as most reliable (Naugle et al., 2012), and to ensure greater homogeneity across the different methodological approaches, only articles investigating PPT were considered.

The exact search strategy and the corresponding search strings per database are listed in Supplementary Table S1. Briefly, a combination of the primary terms “pain” and “exercise” and their associated sub-domains (e.g., pain thresholds, pain tolerance, cycling) was used for a comprehensive screening of the current literature.

2.2. Screening process and extraction of data

A first screening of the literature and eligibility assessment based on title and abstract was performed by two independent researchers (AS, MS). The remaining articles were subsequently extracted based on full-text screenings and imported in Microsoft Excel (Microsoft Corporation, 2018, Redmond, Washington) for further data management. An overview of the study selection process is presented in Figure 1. To resolve any disagreements, an additional researcher (FT) was consulted and consensus was reached.

Figure 1. Preferred reporting items for systematic reviews and meta-analysis (PRISMA) diagram of the study screening process for examining the effect of acute aerobic exercise on exercise-induced hypoalgesia in healthy participants.

Study-related variables, such as author, year of publication, study design, number and age of the participants, exercise modality and dose (intensity and duration), control condition, experimental pain measurement method, and measurement site were extracted for each of the included studies and imported into an electronic spreadsheet. An overview of these information is presented in Table 1.

Table 1. Overview included studies including characteristics of participants, exercise and control interventions, and experimental pain measurements. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, min = minutes, n.A. = not applicable, PPT = pressure pain threshold, RPE = rate of perceived exertion, SD = standard deviation, VO2max = maximum rate of oxygen consumption, W = Watt.

First author (year)	Participants (males) age [years] mean ± SD	Study design	Exercise modality	Exercise intensity/duration #1 [n]	Exercise intensity/duration #2 [n]	Exercise intensity/duration #3 [n]	Control intervention [n]	Measurement instrument for PPT	Measurement sites
Hviid et al. (2019)	35 (17)	Cross-over	Walking	6 Min at mean RPE of 11 (6–20) [n = 35]	same as #1 [n = 35]	n.a.	6 Min quiet rest [n = 35]	Handheld algometer (Somedic)	local: Quadriceps femoris; remote: Trapezius
	26.0 ± 2.9
Jones et al. (2019)	16 (14)	Cross-over	Cycling	5 Min warm-up +15 Min at 60–70% HRR [n = 16]	n.a.	n.a.	20 Min light activity (RPE not above resting) [n = 16]	Handheld algometer (Wagner)	local: Rectus femoris; Remote: n.a.
	23.6 ± 3.9
Naugle et al. (2014)	27 (12)	Cross-over	Cycling	5 Min warm-up +20 Min at 70% HRR [n = 27]	5 Min warm-up +20 Min at 50% HRR [n = 27]	n.a.	25 Min quiet rest [n = 27]	Handheld algometer (Jtech)	local: n.a.; remote: Ventral forearm
	21.8 ± 4.1
Naugle et al. (2016)	^143 (20)	Cross-over	Cycling	5 Min warm-up +20 Min at 70% HRR [n = 43]	5 Min warm-up +20 Min at 50–55% HRR [n = 43]	n.a.	25 Min quiet rest [n = 43]	Handheld algometer (Jtech)	local: n.a.; remote: Ventral forearm
Niwa et al. (2022)	73 (38)	Cross-over	Cycling	30 Min at 70% HRR [n = 73]	30 Min at 50% HRR [n = 73]	30 Min at 30% HRR [n = 73]	30 Min quiet rest [n = 73]	Handheld algometer (Somedic)	local: Quadriceps femoris; remote: Biceps brachii
	20.6 ± 1.1
Pessoa et al. (2021)	^2Group #1 [n = 16 (5)]:	Parallel	Treadmill	2 Min warm-up +5 Min at 75–85% HRmax (220-age)+3 Min cool-down [n = 16]	n.a.	n.a.	Placebo mobilization [n = 16]	Handheld algometer (Wagner)	local: Tibialis anterior; remote: Interosseous dorsal
	24.7 ± 4.8
	Control group [n = 16 (4)]: 21.9 ± 2.9
Tomschi et al. (2023)	36 (18)	Cross-over	Cycling	30 Min at 75% VO2max [n = 36]	45 Min at 75% VO2max [n = 36]	60 Min at 75% VO2max [n = 36]	30 Min quiet rest [n = 36]	Handheld algometer (Wagner)	Local: Knee; remote: Elbow
	26.6 ± 3.2
Tomschi et al. (2024)	50 (50)	Cross-over	Rowing	30 Min at 70% (±5%) HRmax [n = 50]	n.a.	n.a.	30 Min quiet rest [n = 50]	Handheld algometer (Wagner)	Local: Knee; remote: Forehead
Tronarp et al. (2018)	36 (15)	Cross-over	Cycling	30 Min at 50% MAP (maximum aerobic power) [n = 36]	75 Min at 20% MAP [n = 36]	n.a.	30 Min standing [n = 36]	Handheld algometer (Somedic)	local: Quadriceps femoris; remote: Ventral forearm
	26.8 ± 7.6
Vaegter et al. (2014)	56 (28)	Cross-over	Cycling	2 Min warm-up +13 Min at 75% VO2max [n = 56]	n.a.	n.a.	15 Min quiet rest [n = 56]	Handheld algometer (Somedic)	local: Quadriceps femoris; remote: Biceps brachii
	^3median: 22
	range: 20–30
Vaegter et al. (2018)	34 (21)	Cross-over	Cycling	15 Min incremental cycling (start: 20 W, increment 20 W/Min) until RPE of 16 achieved. Thereafter intensity was maintained [n = 34]	same as #1 [n = 34]	n.a.	15 Min quiet rest [n = 34] before #1 and #2	Handheld algometer (Somedic)	local: Quadriceps femoris; remote: Trapezius
	25.8 ± 3.4
Van Weerdenburg (2016)	15 (9)	Cross-over	Cycling	10 Min warm-up +20 Min at 75–88% HRmax (Karvonen formula) [n = 15]	n.a.	n.a.	30 Min deep breathing [n = 15]	Handheld algometer (SBMEDIC)	local: Rectus femoris; Remote: Thenar muscle
	25.0 ± 6.5
Zheng et al. (2021)	Group #1 [n = 15 (8)]	Parallel	Cycling	5 Min warm-up +23 Min at 70% HRR [n = 15]	5 Min warm-up +25 Min at 50% HRR [n = 15]	n.a.	2 Min warm-up +23 Min quiet rest [n = 15]	Handheld algometer (Wagner)	local: Rectus femoris; remote: Thenar muscle
	24.7 ± 3.2
	Group #2 [n = 15 (7)]
	24.4 ± 2.2
	Group #3 [n = 15 (8)]
	24.3 ± 3.0

¹Of these: n = 25 (11) 21.7 ± 4.1 younger and n = 18 (9) 63.7 ± 6.6 older participants. ²Only the aerobic training group and the control group are presented. ³Indication of age provided only as median and range.

The primary outcome within the scope of this review was the mean difference (MD) in PPT after an exercise-based intervention compared with baseline values relative to the control condition. A positive difference between post and pre (post – pre) values indicates a hypoalgesic effect, whereas a negative difference implies a hyperalgesic effect. These data were extracted when available. Otherwise, the reviewers contacted the respective corresponding authors via email and followed this up two weeks later. In case of not getting a response, the outcomes of interest were extracted (see below). When studies formulated EIH opposingly (pre – post), results were redirected to match the definition used in this analysis. In the case of bilateral assessment of a particular landmark, the mean value was calculated and used for further analyses, considering that exercise applications included only cyclical and symmetrical modalities (i.e., cycling, running, rowing) that lead to equal bilateral loading. When either absolute (raw data of PPT) or relative (MD of PPT) data were illustrated graphically only and data were not provided by the respective authors, the online software WebPlotDigitizer (Version 4.6, San Francisco, CA) was used to translate illustration-based information into numerical values that could be used for further analysis as done before (Moeyaert et al., 2016). Using this software, specific points on the graph of any figure can be selected and displayed in numerical form. If data on the primary outcome were reported in a statistical unit other than mean and standard deviation (e.g., 95% confidence interval, standard error), standard formulae were used to convert them to ensure uniformity of data (Chi et al., 2023). For the conversion of median and interquartile range, the approximation method developed by Wan et al. (Wan et al., 2014) was used to estimate the mean and standard deviation of the respective data distribution as described elsewhere (Mok et al., 2021). If the standard deviation of difference was not explicitly reported, the following formula according to the Cochrane Handbook for Systematic Reviews of Interventions was applied (Higgins & Green, 2008):

$\mathrm{\Delta }\mathrm{SD}=\mathit{\hspace{0.17em}}\sqrt{\left(\mathrm{SDpr}{\mathrm{e}}^2+\mathrm{SDpos}{\mathrm{t}}^2-2\mathrm{x}\mathrm{corr}\mathrm{x}\mathrm{SDpre}\mathrm{x}\mathrm{SDpost}\right)}$

In concordance with Wewege & Jones (2021), a conservative paired-samples correlation of 0.85 was set for both study conditions.

2.3. Risk of bias assessment

Two reviewers (AS & MS) evaluated the methodological quality of the included studies by using the Cochrane risk of bias, which is included in the software Review Manager (RevMan; Version 5.4, Copenhagen, Denmark) for windows (Higgins et al., 2011). For all included studies, seven domains for parallel designs were applied: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. The cross-over designs were assessed with three additional domains: appropriate crossover design (which includes (1) stable condition of the participants, (2) temporary effects of the intervention and (3) potential for carry-over effect), evaluation of the carry-over effect, and if unbiased data were presented (Ding et al., 2015). In general, this evaluation tool focuses on detection of bias arising from e.g., methodological errors or unreported data. Studies were considered as having a “low risk of bias”, a “high risk of bias”, or “unclear risk of bias”. For the assessment of the overall quality of evidence, the GRADE approach was applied, by evaluating the degree of imprecision (n < 300), inconsistency (I² ≥ 75%), risk of bias, and likelihood of publication bias (G. H. Guyatt et al., 2008). Discrepancies between the reviewers were settled through discussion with a third reviewer (FT).

2.4. Statistical analyses

The software RevMan was utilized to perform random-effects meta-analytical calculations and generate forest plots for the outcome variable of all included studies. The MD and 95% confidence intervals (95% CI) were computed after data for PPT were uniformly converted to units of Newton/cm².

To compensate for the heterogeneous number of landmarks assessed across studies, exactly two measurement points were selected for each study arm. The local landmark was defined as the one closest to the working musculature, being actively used in the intervention and located below the waist. The remote landmark was the one furthest peripheral from the working musculature, without active involvement in the exercise intervention and located above the waist. Hence, studies that exclusively reported data for either landmarks that were actively engaged (Jones et al., 2019) or passively used structures during exercise (Naugle et al., 2014, 2016) were excluded from sub-group analyses, yet included in the overall analysis. Taking into account the inclusion of multiple groups from one study (cross-over designs), the total number of participants in the corresponding control condition was evenly divided among the effects to avoid inappropriate weighting, as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins & Green, 2008) and recently conducted elsewhere (van Hooren et al., 2020).

To address the research questions formulated, multiple sub-group analyses were conducted. Studies were therefore categorized based on exercise intensity, duration, and modality (Liguori et al., 2022) for local and remote effects, respectively:

Exercise intensity: HRR (heart rate reserve) ≤ 50% (low intensity) vs. HRR > 50%/≤70% (moderate intensity) vs. HRR > 70% (high intensity). As no consistent unit for the exercise intensity was reported across all studies considered, the most commonly applied unit HRR was used for categorisation, which can be defined as the difference between the HRmax (maximum heart rate) and the resting heart rate. If the HRR was not reported, values for rate of perceived exertion (RPE) or percentages of either VO₂max (maximum oxygen uptake) or HRmax were transformed to HRR according to the conversion sheet published by the American College of Sports Medicine (ACSM) (Liguori et al., 2022).

Exercise duration: duration ≤20 min (low duration) vs. duration > 20/≤40 min (moderate duration) vs. duration >40 min (long duration).

Exercise modality: cycling-based interventions vs. treadmill-based interventions vs. rowing-based interventions.

The degree of heterogeneity was statistically tested using the I² statistic, which provides an estimate for a percentage-based variation across studies. The threshold for homogeneity among studies was defined as I² ≤ 40%, while I² ≥ 75% was used as a cut-off value for assessing the level of consistency among studies using the GRADE approach (Higgins & Green, 2008). Additionally, heterogeneity was assessed on the basis of the direction of the effect sizes among studies and overlapping CIs (G. Guyatt et al., 2023). Egger’s regression analysis and funnel plots were used to assess the presence of publication bias. An alpha-level of p ≤ 0.05 was applied.

3. Results

3.1. Study characteristics

In total, after removing the duplicates, 13 of the 2656 screened studies met the inclusion criteria and were therefore included in this meta-analysis. An overview of the selection process is presented in Figure 1.

Ten of the 13 included studies were designed as controlled randomized cross-over trials (Hviid et al., 2019; Jones et al., 2019; Naugle et al., 2014, 2016; Niwa et al., 2022; Tomschi et al., 2023, 2024; Tronarp et al., 2018; Vaegter et al., 2014, 2018; van Weerdenburg et al., 2016), one was designed as a controlled cross-over trial (Vaegter et al., 2018), and the remaining two were randomized controlled trials (Pessoa et al., 2021; Zheng et al., 2021). Across all studies, a sum of 498 (274, 55.0% male) healthy participants were investigated. In total, 23 exercise interventions and 14 control interventions were analysed. In two studies (Hviid et al., 2019; Vaegter et al., 2018), participants underwent two sessions that included both an intervention and control condition, each considered separately in the meta-analytical calculations. In addition to cycling (10 studies) (Jones et al., 2019; Naugle et al., 2014, 2016; Niwa et al., 2022; Tomschi et al., 2023; Tronarp et al., 2018; Vaegter et al., 2014, 2018; van Weerdenburg et al., 2016; Zheng et al., 2021), which was predominantly employed as an aerobic exercise intervention, treadmill-based exercises (two studies) (Hviid et al., 2019; Pessoa et al., 2021) and rowing (one study) (Tomschi et al., 2024) were performed. Intervention protocols varied across duration and intensity in all included studies. Warm-ups were performed before the intervention in nine studies (Jones et al., 2019; Naugle et al., 2014, 2016; Niwa et al., 2022; Pessoa et al., 2021; Tomschi et al., 2024; Vaegter et al., 2014; van Weerdenburg et al., 2016; Zheng et al., 2021), only one also employed a cool-down phase (Pessoa et al., 2021). In one study (Vaegter et al., 2018), the load intensity was increased incrementally until the predetermined intensity level was reached, without explicitly declaring it as a warm-up. The shortest duration of the load was five minutes (Pessoa et al., 2021), whereas the longest duration was 75 minutes (Niwa et al., 2022; Tomschi et al., 2023; Tronarp et al., 2018). The intensities of the interventions were determined differently. In five studies, HRR was used, whereas in three studies HRmax was used (Pessoa et al., 2021; Tomschi et al., 2024; van Weerdenburg et al., 2016). Two other studies used the RPE (Hviid et al., 2019; Vaegter et al., 2018) and two others used VO2max (Tomschi et al., 2023; Vaegter et al., 2014). A single study reported intensity using maximal aerobic power (Tronarp et al., 2018). The most frequently selected control condition was quiet rest (eight studies) (Hviid et al., 2019; Naugle et al., 2014, 2016; Niwa et al., 2022; Tomschi et al., 2023, 2024; Vaegter et al., 2014, 2018), whereas in one study (Zheng et al., 2021) a brief warm-up period was followed by quiet rest. The remaining control sessions comprised light activity with RPE not exceeding resting level (Jones et al., 2019), a placebo mobilization program (Pessoa et al., 2021), standing (Tronarp et al., 2018), or a deep breathing protocol (van Weerdenburg et al., 2016). Measuring instruments in all included studies were manually operated pressure algometers. In terms of measurement points, 11 studies assessed PPT at a minimum of one local landmark (Hviid et al., 2019; Jones et al., 2019; Niwa et al., 2022; Pessoa et al., 2021; Tomschi et al., 2023, 2024; Tronarp et al., 2018; Vaegter et al., 2014, 2018; van Weerdenburg et al., 2016; Zheng et al., 2021) and 12 studies at a minimum of one remote landmark (Hviid et al., 2019; Naugle et al., 2014, 2016; Niwa et al., 2022; Pessoa et al., 2021; Tomschi et al., 2023, 2024; Tronarp et al., 2018; Vaegter et al., 2014, 2018; van Weerdenburg et al., 2016; Zheng et al., 2021). Eight studies selected the quadriceps femoris muscle as the local landmark (Hviid et al., 2019; Jones et al., 2019; Niwa et al., 2022; Tronarp et al., 2018; Vaegter et al., 2014, 2018; van Weerdenburg et al., 2016; Zheng et al., 2021), two studies the knee joint (Tomschi et al., 2023, 2024), and one study the tibialis anterior muscle (Pessoa et al., 2021). The measurement site on the quadriceps femoris muscle was uniformly located 10–20 cm proximal to the base of the patella, but was interchangeably referred to as either rectus femoris or quadriceps femoris muscle. For remote landmarks, a more heterogeneous selection was observed with two studies examining the biceps brachii (Niwa et al., 2022; Vaegter et al., 2014), two the thenar muscle (van Weerdenburg et al., 2016; Zheng et al., 2021), two the trapezius muscle (Hviid et al., 2019; Vaegter et al., 2018), three the ventral forearm (Naugle et al., 2014, 2016; Tronarp et al., 2018), one the elbow joint (Tomschi et al., 2023), one the forehead (Tomschi et al., 2024), and one the dorsal interosseus muscle (Pessoa et al., 2021). In nine studies, a unilateral assessment of the respective landmarks was performed, either by measuring exclusively the right side (Jones et al., 2019; Pessoa et al., 2021; Tronarp et al., 2018; Zheng et al., 2021), exclusively the dominant side (Vaegter et al., 2014; van Weerdenburg et al., 2016), exclusively the non-dominant side (Niwa et al., 2022), or the dominant quadriceps femoris muscle (local side) and the non-dominant trapezius muscle (remote side) (Hviid et al., 2019; Vaegter et al., 2018). In two studies, bilateral assessment of the landmarks of interest was conducted (Tomschi et al., 2023, 2024). Here, the mean value was used for further analyses. Two studies displayed pooled data of the corresponding landmarks which were used unchanged for analysis (Naugle et al., 2014, 2016). An overview of study characteristics and details on the specific intervention applied can be obtained from Table 1.

3.1.1. EIH after aerobic exercise

When considering the full spectrum of studies containing at least one landmark defined as local or remote, a significant effect of aerobic exercise on EIH could be observed for pooled data, yet with considerable heterogeneity across studies (MD = 2.33 [95% CI: 1.28, 3.37], p < 0.001, I² = 69%, k = 41). Likewise, when performing sub-group analyses solely for local (MD = 3.13 [95% CI: 1.22, 5.04], p = 0.001, I² = 64%, k = 19) and remote landmarks (MD = 1.82 [95% CI: 0.57, 3.06], p = 0.004, I² = 72%, k = 22), there were significant hypoalgesic effects found for both analyses, however with no significant difference between local and remote landmarks (Figure 2).

Figure 2. Forest plot of exercise-induced hypoalgesia for local and remote effects, respectively, and the risk of bias evaluation. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption.

3.1.2. EIH across exercise intensities

Sub-group analysis regarding exercise intensity revealed that both moderate intensity (local: MD = 3.07 [95% CI: 0.75, 5.39], p = 0.010, I² = 50%, k = 10; remote: MD = 3.13 [95% CI: 0.53, 5.72], p = 0.020, I² = 82%, k = 10) and high intensity (local: MD = 7.46 [95% CI: 3.67, 11.24], p < 0.001, I² = 0%, k = 2; remote: MD = 2.99 [95% CI: 0.47, 5.51], p = 0.020, I² = 0%, k = 2) aerobic exercise interventions significantly induced EIH at local and remote landmarks. For low intensities, however, no significant effect on pain modulation could be observed (local: MD = 0.97 [95% CI: −1.69, 3.64], p = 0.470, I² = 58%, k = 6; remote: MD = 0.56 [95% CI: −1.11, 2.23], p = 0.510, I² = 73%, k = 6). Between sub-group analyses for local landmarks indicate a significant difference in terms of EIH between the different intensity levels but not for remote (Figures 3 and 4)

Figure 3. Forest plot of exercise-induced hypoalgesia of different exercise intensities for local effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption.

Figure 4. Forest plot of exercise-induced hypoalgesia of different exercise intensities for remote effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption.

3.1.3. EIH across exercise durations

For exercise duration, sub-group analysis indicated that only aerobic exercise bouts of moderate duration result in significant EIH both at local (MD = 3.38 [95% CI: 1.71, 5.05], p < 0.001, I² = 9%, k = 8) and remote landmarks (MD = 1.41 [95% CI: 0.39, 2.43], p = 0.007, I² = 35%, k = 8). A significant difference between duration sub-groups was observed forlocal landmarks only (Figures 5 and 6).

Figure 5. Forest plot of exercise-induced hypoalgesia of different exercise durations for local effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption.

Figure 6. Forest plot of exercise-induced hypoalgesia of different exercise durations for remote effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption..

3.1.4. EIH across exercise modalities

In terms of exercise modality, for both local (MD = 3.85 [95% CI: 2.06, 5.64], p < 0.001, I² = 50%, k = 14) and remote landmarks (MD = 1.98 [95% CI: 1.24, 2.72], p < 0.001, I² = 0%, k = 14) cycling-based interventions resulted in significant EIH. Treadmill/Walking-based (local: MD = − 1.64 [95% CI: −4.50, 1.22], p = 0.260, I² = 6%, k = 3; remote: MD = 6.25 [95% CI: −2.11, 14.62], p = 0.140, I² = 96%, k = 3) and rowing-based (local: MD = −1.40 [95% CI: −6.78, 3.98], p = 0.610, I² = not applicable, k = 1; remote: MD = −1.10 [95% CI: −3.45, 1.25], p = 0.360, I² = not applicable, k = 1) bouts of aerobic exercise yielded no significant effects on pain modulation, irrespective of the landmark assessed. Significant between-group differences between the different modalities were found at local and remote landmarks(Figures 7 and 8).

Figure 7. Forest plot of exercise-induced hypoalgesia of different exercise modalities for local effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption.

Figure 8. Forest plot of exercise-induced hypoalgesia of different exercise modalities for remote effects. HFmax = maximal heart frequency, HRR = heart rate reserve, MAP = maximum aerobic power, RPE = rate of perceived exertion, VO2max = maximum rate of oxygen consumption..

3.1.5. Risk of bias and quality assessment of studies

An overview of the risk of bias assessment of all included studies and their individual scores is presented in Figure 2. The cross-over domains that do not relate to the randomised controlled trial studies are not colour-coded for these two studies (Pessoa et al., 2021; Zheng et al., 2021). In the categories exclusively for cross-over designs, three studies were considered as high risk, two as unclear risk and five as low risk of bias for the “appropriate cross-over design”. For the three studies, the risk of bias was considered as high because the washout periods between interventions (<1 h) were considered as insufficient (Hughes & Patterson, 2020). There were three studies with unclear risk of bias due to the fact that stable conditions such as interventions on the same time of day or pain medication use were not reported. Seven cross-over studies were considered to be at low risk of bias for “carry over effect” as they evaluated the baseline outcomes and did not find statistically significant differences. In the other four studies no statistical assessment of the carry over effect was described. Regarding “unbiased data”, all studies were considered as low risk of bias as all provided data for all experimental periods within their respective investigations.

Considering the risk of bias categories that are applicable to both parallel and cross-over study designs, one study was judged as high risk of bias for “random sequence generation” due to the fact that no randomization was performed. Another seven studies were judged as unclear risk of bias because they did not provide information on how randomization was performed. The remaining studies were classified as low risk. In the “allocation concealment” category only one study adequately described this and thus had a low risk of bias. The other twelve were considered as unclear risk. The same result was obtained in the category “blinding of participants and personnel”. Accordingly, only one study was assessed as low risk and twelve as unclear risk of bias. None of the studies described attempts of implementing “blinding of outcome assessment”. Therefore, all were considered as unclear risk of bias. Ten studies were assessed as having low risk of bias concerning “incomplete outcome data” for comprehensively documenting the progression of patient participation throughout the trial; three were considered as unclear risk. All studies were considered as unclear risk of “selective reporting” bias, because no study provided information on prespecified outcomes. Regarding “other bias”, all studies were categorized to be at low risk.

Analysis of publication bias within the studies showed no statistically significant evidence of concern employing Egger’s regression (p = 0.374). Visual assessment via funnel plot can be obtained in Supplementary Figure S1.

Assessment of the overall quality of evidence for the effect of a single aerobic exercise intervention on pain perception revealed a moderate level of quality for local and remote landmarks (Supplementary Table S2). The quality of evidence for local landmarks was downgraded due to uncertainties in terms of risk of bias. For remote landmarks, the majority of studies demonstrated an uncertain risk of bias profile. The grading of subgroup analyses is presented in the Supplementary Table S2.

4. Discussion

This meta-analysis evaluated the impact of acute aerobic exercises on local and remote mechanical pain perception using PPT. These analyses for the first time elucidate the impact of specific influencing factors, i.e., measured body site (local/working, remote/non-working), intensity, duration, and modality (bicycle, walking/running, rowing) on EIH effects, providing an in-depth insight on potential dose-response relationships.

Regarding the first research question of the present meta-analysis, results indicate that EIH effects generally occur following aerobic exercises compared to control conditions (Figure 2). These results are in line with prior observations of EIH at a meta-analytical level, which demonstrated large effects following aerobic exercise (Naugle et al., 2012; Wewege & Jones, 2021). Furthermore, the present meta-analysis reveals that hypoalgesia is only slightly (not significantly) lower at remote body sites compared to local ones. It is still discussed whether EIH effects are more pronounced and more consistently observed depending on the distance to the primary exercising body parts during cycling, as central and peripheral pain inhibitory processes are involved in EIH (Gomolka et al., 2019; Jones et al., 2017; Tomschi et al., 2022; Vaegter et al., 2014; Wu et al., 2022) and no conclusive effects can be drawn from the present analysis.

In the light of the second research question, results further reveal that EIH is highest after high-intensity exercises close to the working musculature compared to both lower intensities. The two interventions using a high intensity were conducted by Vaegter et al. 2018 and began with an initial intensity of 20 Watts. The resistance was subsequently raised by 20 Watts every minute until the participants reached a RPE of 16 equalling a hard to very hard subjective intensity. Upon achieving this level of exertion, the subjects maintained that intensity for the remaining duration of the 15-min exercise. Here, in both interventions the PPT were higher post exercise at the quadriceps and trapezius while no difference was observed in the rest control condition (Vaegter et al., 2018). However, EIH was also achieved locally using moderate intensities, such as 70% of HHR (30 min (Niwa et al., 2022) and 25 min (Zheng et al., 2021)), 75% VO2max (Vaegter et al., 2014), 75–88% of HRmax (van Weerdenburg et al., 2016). Remotely, EIH was only observed following 70% of HRR (Niwa et al., 2022; Zheng et al., 2021). Though, EIH effects were noted remotely after moderate as well as high-intensity exercise on the subgroup level, while no difference was observed between the intensities (Figure 4). Interestingly, EIH following low-intensity exercise was not superior compared to the control condition locally and remotely, which was the case in moderate and high intensity (Figure 3). On a physiological level, the degree of EIH, among others, depends on the intensity of the exercise, as high intensity exercise activates the hypothalamic-pituitary-adrenal (HPA) axis and triggers the release of catecholamines more potently (Hilberg et al., 2003; Tanaka et al., 2023). Consequently, neuroendocrine substances, such as endogenous opioid-related substances (e.g., beta-endorphin) and endocannabinoids (e.g., anandamide (AEA) and 2-arachidonoylglycerol (2-AG)), are synthesized and released from multiple sites within the central and peripheral nervous system and activate pain-inhibiting pathways centrally and peripherally (Crombie et al., 2018; McMurray et al., 1987; Rice et al., 2019; Schwarz & Kindermann, 1989; Vaegter & Jones, 2020).

It must be considered that the low intensities were performed in two interventions via a walking exercise of 6 min using an intensity of RPE = 11 (equalling a light subjective exertion). These two interventions revealed no EIH effects compared to the respective control conditions and even hyperalgesic effects remotely following one exercise bout (Hviid et al., 2019). Low intensities were also determined by using 30% and 50% of the HRR, respectively (Niwa et al., 2022), 50% of the HRR (Zheng et al., 2021), and 20% of MAP (Tronarp et al., 2018). Here, results on an individual level reveal that EIH was observed following 50% HRR locally and 30% and 50% remotely (Niwa et al., 2022). When considering the estimates of exercise intensity, it becomes evident that specific intensity levels are defined differently in single studies. Therefore, it was necessary to convert the different intensity estimates (based on the conversions recommended by the ACSM) into a common unit (HRR) allowing a meaningful and standardized comparison between individual studies. Yet, it must be noted that this conversion might result in a loss of specificity.

Only a limited number of interventions examined the EIH effects following long-duration exercises (>40 min), showing that the focus of research revolves around shorter duration exercises (Figures 5 and 6). Low duration (≤20 min) interventions did not result in EIH compared to the respective control interventions locally and remotely, which can most likely be explained by the low intensities of RPE = 11 alongside the very low duration of 6 min used in two interventions (Hviid et al., 2019). Yet, the four other interventions of this subgroup using higher intensities (i.e., RPE of 16 (Vaegter et al., 2018), 75–85% of HRmax (van Weerdenburg et al., 2016), 75% of VO2max (Vaegter et al., 2014)) resulted in local EIH but only two also revealed remote EIH (Pessoa et al., 2021; Vaegter et al., 2018). Important to note is also that these three latter studies used a longer duration of 15 to 20 min. Contrastingly, moderate duration exercise effects were observed locally and remotely, which might be explained by the rather consistently sufficient intensity employed in these interventions. However, based on the study by Tomschi et al. 2023, no differences were observed between exercise durations (30, 45, and 60 minutes) , revealing that the exercise duration has probably only a minor influence on EIH (Tomschi et al., 2023). Only three studies examined the effect of long exercise durations (>40 min) and no EIH is observed locally and remotely. Here, it must be considered that one intervention used a very low intensity of 20% MAP for 75 min (Tronarp et al., 2018), which is most likely an insufficient intensity to induce any hypoalgesia. Besides, two of the long duration interventions used an intensity of 75% VO2max for 45 and 60 min (Tomschi et al., 2023), which can be seen as a sufficient intensity to induce EIH as observed in other shorter duration interventions in the present meta-analysis (e.g., Vaegter et al. (2014)). Yet, that study used bony (articular) structures for PPT measurements (Tomschi et al., 2023) compared to muscular structures used by all other studies except for one other (Tomschi et al., 2024), in which also no EIH effects were observed. To the best of the authors knowledge, it has not been scientifically studied whether bony and muscular measurement sites yield comparable results in the context of EIH even though differences in baseline sensitivity seem to be evident (Maeda et al., 2011).

The majority of interventions were conducted using cycling exercises, which resulted in local and remote EIH. Only a limited number of studies used walking or running (Hviid et al., 2019; Pessoa et al., 2021) and one study was conducted on a rowing ergometer (Tomschi et al., 2024) with no effects observed in any of these interventions (Figures 7 and 8). Of the included three interventions using the exercise modality walking or running, two interventions were of low intensity and duration (RPE = 11 for 6 min) (Hviid et al., 2019), while one was of moderate intensity (75–85% HRmax for 5 min). This might explain the lack of EIH as exercise intensity seems to be a key determinant to induce EIH. The rowing ergometer intervention was conducted at a moderate intensity of 70% HRmax for 30 min and did not induce any EIH locally or remotely at articular landmarks.

Based on the GRADE system, the overall quality of the evidence was at a moderate level of quality for the main analysis for local and remote landmarks. The quality of evidence for local landmarks was downgraded due to uncertainties in terms of risk of bias. For remote landmarks, the majority of studies demonstrated an unclear risk of bias profile. Regarding the individual analyses, due to concerns related to inconsistency, lack of precision, or potential bias risks the quality of evidence had to be downgraded on an individual level from very low to moderate (Supplementary Table S2). Regarding the risk of bias, some categories were generally considered to be unclear i.e., wash-out periods, description of how randomization was performed (allocation concealment), prospective study protocol, and statistical reporting of potential carry-over effects. Besides, risk of bias regarding blinding was reported regularly. However, from a practical point of view, it is almost impossible to blind – either the participants or the testers – with respect to whether the exercise or control session is performed.

This meta-analysis is subject to several limitations: The study’s strict inclusion and exclusion criteria led to a limited number of studies and interventions, since the analysis is based on high-quality research. Non-controlled trials or grey literature were not included, introducing the potential for publication bias. Therefore, only studies utilizing controlled trials as part of parallel-group or crossover designs were included aiming to provide a more accurate estimate of the causal nature of observed EIH effects. This decision was grounded on the notion that single-group trials are particularly prone to bias especially when the primary outcomes are semi-objective experimental pain measurements, such as PPT. Due to familiarization or adaptation effects, there may be certain distortions in the outcomes that can only be ruled out when incorporating a control condition (Song et al., 2021; Wewege & Jones, 2021). Additionally, the inclusion of four studies (five interventions) with small sample sizes (n ≤ 16) per intervention may reduce the statistical power and robustness of our results. Given the unavailability of some raw data, certain data points were obtained through recalculations and visual extraction methods. Dose-response relationships between exercise intensity as well as duration and EIH could not be calculated as the independent variables had to be scaled categorically due to the low diversity in exercise variable ranges. Future studies should therefore be encouraged to employ more divergent exercise regimes in the context of experimentally testing EIH. Of note is also the exclusive focus on healthy individuals, limiting the applicability to clinical populations. High heterogeneity among studies poses a challenge, despite the use of random-effects modelling. Furthermore, our analysis exclusively includes mechanical pain threshold measurements, potentially limiting the generalizability of our findings to other types of experimental pain modalities, such as thermal or electrical stimuli, albeit research has proposed mechanical PPT as the most valid measurement method to assess EIH effects (Naugle et al., 2012).

In forthcoming research, it is advisable that studies utilize a stringent methodology (e.g., randomized controlled design, sufficient wash-out periods e.g., more than 24 h (Hughes & Patterson, 2020), familiarization with the measurement procedure) to ascertain the causal impact of a single exercise session in healthy participants and to reduce the risk of bias. More concretely, specific different exercise intensities and durations should be compared in one study to gain more precise insights into dose-response relationships (as done e.g., (Tomschi et al., 2022)), by also including a control session. Besides, to determine the influence of intensity and duration on EIH, the poles of this continuum should be used in upcoming studies, i.e., very short duration but maximal intensity and vice versa. Most exercises were conducted on a bicycle ergometer, while only three interventions were performed with running/walking and one was performed with rowing as the exercise modality, which calls for more studies using above all treadmills but also other aerobic exercise modalities. Moreover, as EIH effects might depend on the body structure measured (e.g., muscular, bony, articular), it is warranted to compare different structures with respect to EIH effects. The gained knowledge of this meta-analysis can be incorporated (in adapted form) into studies dealing with clinical populations suffering from pain syndromes. In this context, there is a clear need for more research investigating the EIH effect of specific aerobic exercise in these populations (as done e.g., patients with musculoskeletal pain (Wewege & Jones, 2021)), where, in addition to experimental pain measurements, also subjective pain (e.g., via NRS/VAS) can be assessed to provide a more holistic view of exercise effects on pain perception.

5. Conclusion

The findings of the present analyses advance the understanding of the intricate relationship between aerobic exercise and acute pain perception by providing crucial insights into possible dose-response relationships. EIH effects typically manifest following aerobic exercises locally and remotely. Higher-intensity exercises near the working musculature seem to produce highest hypoalgesia. In contrast, lower-intensity exercises, especially of long duration, yielded negligible effects. The overall evidence quality was moderate for local and remote effects, with many studies showing unclear risk biases.

Acknowledgments

The authors would like to thank Mr. Joschua Wieseor his support in conducting this meta-analysis. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online https://doi.org/10.1080/02640414.2024.2352682

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.