Effects of Strength Training on the Physiological Determinants of Middle- and Long-Distance Running Performance: A Systematic Review

2.1 Literature Search Strategy

The PRISMA statement [43] was used as a basis for the procedures described herein. Electronic database searches were carried out in Pubmed, SPORTDiscus, and Web of Science using the following search terms and Boolean operators: (“strength training” OR “resistance training” OR “weight training” OR “weight lifting” OR “plyometric training” OR “concurrent training”) AND (“distance running” OR “endurance running” OR “distance runners” OR “endurance runners” OR “middle distance runners”) AND (“anaerobic” OR “sprint” OR “speed” OR “performance” OR “time” OR “economy” OR “energy cost” OR “lactate” OR “maximal oxygen uptake” OR “\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)” OR “aerobic” OR “time trial”). Searches were limited to papers published in English and from 1 January 1980 to 6 October 2017.

2.2 Inclusion and Exclusion Criteria

Using the mean \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) values provided within each study, participants training status was considered as moderately-trained (male \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) ≤ 55 ml kg⁻¹ min⁻¹), well-trained (male \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) 55–65 ml kg⁻¹ min⁻¹), or highly-trained (male \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) ≥ 65 ml kg⁻¹ min⁻¹) [10, 46]. For female participants, the \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) thresholds were set 10 ml kg⁻¹ min⁻¹ lower [46]). In the absence of \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) values, training status was based upon the training or competitive level of the participants: moderately-trained = recreational or local club, well-trained = Collegiate or provincial, highly-trained = national or international.

2.3 Study Selection

Figure 1 provides a visual overview of the study selection process. Search results were imported into a published software for systematic reviews [47], which allowed a blind screening process to be performed by two independent reviewers (RB and PH). Any disagreements were resolved by consensus. The initial search yielded 454 publications. Following the removal of duplicates (n = 190), publications were filtered by reading the title and abstract [inter-rater reliability (IRR): 95.3%, Cohens k = 0.86] leaving 19 review articles or commentaries, and 47 potentially relevant papers, which were given full consideration. Five additional records were identified as being potentially relevant via manual searches of previously published reviews on this topic and the individual study citations. These 52 studies were considered in detail for appropriateness, resulting in a further 26 papers [34, 37, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71] being excluded (IRR: 94.2%, Cohens k = 0.88) for the following reasons: not published in full in a peer-reviewed journal [50, 52, 60, 61], absence of a running only control group [48, 49, 54, 57, 59, 62, 63, 64, 65, 66, 67, 69], participants were non-runners [51, 53, 56, 68], no physiological parameters were measured [55], dissimilar running training was applied between groups [71], the ST intervention was poorly controlled [54], and ST did not involve one of the aforementioned types [34, 37, 58, 70].

Open image in new window

2.4 Analysis of Results

The Physiotherapy Evidence Database (PEDro) scale was subsequently used to assess the quality of the remaining 26 records [31, 32, 33, 36, 38, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92] by the two independent reviewers. Two studies reported their results across two papers [32, 38, 90, 92], therefore both are considered as single studies hereafter, thus a total of 24 studies were analyzed. The PEDro scale is a tool recommended for assessing the quality of evidence when systematically reviewing randomized-controlled trials [93]. Each paper is scrutinized against 11 items relating to the scientific rigor of the methodology, with items 2–11 being scored 0 or 1. Papers are therefore awarded a rating from 0 to 10 depending upon the number of items which the study methodology satisfies (10 = study possesses excellent internal validity, 0 = study has poor internal validity). No studies were not excluded based upon their PEDro scale score and IRR was excellent (93.2%, Cohens k = 0.86).

Results are summarized as a percentage change and the p value for variables relating to: strength outcomes, RE, \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), BL response, time trial, anaerobic performance, and body composition. Due to the heterogeneity of outcome measures in the included studies and the limitations associated with conditional probability, where possible, an effect size (ES) statistic (Cohens d) is also provided. Effect size values are based upon those reported in the studies or were calculated using the ratio between the change score (post-intervention value minus pre-intervention value) and a pooled standard deviation at baseline for intervention and control groups. Values are interpreted as trivial < 0.2; small 0.2–0.6; moderate 0.6–1.2; and large > 1.2.

3.1 Participant Characteristics

3.2 Study Design and PEDro Scores

Table 1 also provides an overview of several important features of study design, including PEDro scale scores. Studies lasted 6–14 weeks with the exception of two investigations, which lasted 24 [90, 92] and 40 weeks [33]. Fourteen studies provided detailed accounts of the running training undertaken by the participants. However, these were usually reported from monitoring records, thus only three studies were deemed to have appropriately controlled for the volume and intensity of running in both groups [32, 38, 73, 80, 90, 92]. Six studies provided little or no detail on the running training that participants performed [31, 33, 82, 84, 86, 91]. Strength training in all but three investigations [73, 78, 88] was supplementary to running training, and one paper provided the control group with alternative activities (stretching and core stability) matched for training time [77].

Studies were all scored a 4, 5, or 6 on the PEDro scale. All investigations had points deducted for items relating to blinding of participants, therapists, and assessors. Differences in the scores awarded were mainly the result of studies not randomly allocating participants to groups and failing to obtain data for more than 85% of participants initially allocated to groups; or this information not being explicitly stated.

3.3 Training Programs

All studies utilized at least one multi-joint, closed kinetic chain exercise with the exception of two studies that used isometric contractions on the ankle plantarflexors [82, 84]. One study employed only resistance machine exercises for lower limb HRT [81], whereas all other studies used free weights, bodyweight resistance or a combination of machines and free weights. Strength training (using lower limb musculature) was scheduled once [33, 80, 81], twice [31, 32, 33, 38, 75, 78, 85, 86, 87, 89, 90, 92], three times [36, 72, 74, 75, 76, 77, 79, 82, 83, 88], or four times [84] per week. One study used 15 sessions over a 6-week period [91] and one study reported 2.7 h of ST activity per week [73].

Heavy RT was typically prescribed in 2–6 sets of 3–10 repetitions per exercise at relatively heavy loads (higher than 70% 1RM or to repetition failure). Plyometric training prescription consisted of 1–6 exercises performed over 1–6 sets of 4–10 repetitions, totaling 30–228 foot contacts per session. Most studies applied the principle of progressive overload and some authors reported periodized models for the intervention period [32, 33, 36, 38, 77, 88, 89]. Studies which included SpT tended to utilize short distances (20–150 m), over 4–12 sets at maximal intensity [73, 74, 88]. Strength training was supervised or part-supervised across all studies with the exception of three, one that was unsupervised [76] and two where it was unclear from the report [73, 74].

Running training varied considerably (16–170 km week⁻¹, 3–9 sessions week⁻¹) across the studies, with various levels of detail provided regarding weekly volume and intensity. Importantly, all studies that added ST reported that running training did not differ between groups.

3.4 Strength Outcomes

3.5 Running Economy

An assessment of RE was included in all but four [31, 85, 87, 90, 92] of the studies in this review (Table 3). Running economy was quantified as the oxygen cost of running at a given speed in every case, except in three studies where a calculation of energy cost was used [82, 84, 91]. Statistically significant improvements (2–8%, ES: 0.14–3.22) in RE were observed for at least one speed in 14 papers. A single measure of RE was reported in four of these papers [31, 79, 80, 88], and a further four studies assessed RE across multiple different speeds and found improvements across all measures taken [72, 74, 75, 84]. Six papers reported a mixture of significant and non-significant results from the intensities they used to evaluate RE [36, 73, 76, 77, 78, 86]. Six studies failed to show any significant improvements in RE compared to a control group [32, 81, 82, 83, 89, 91].

3.6 Maximal Oxygen Uptake

No statistically significant changes were reported in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) or \(\dot{V}{\text{O}}_{{2{\text{peak}}}}\) for any group in the majority of studies that assessed this parameter [31, 32, 36, 72, 74, 75, 77, 78, 79, 80, 85, 88, 89]. Three papers observed improvements for \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) in the intervention group, but the change in score did not differ significantly from that of the control group [33, 81, 91]. One study detected a significant improvement (4.9%) in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) for the control group compared to the intervention group [73].

3.7 Velocity Associated with \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)

Nine studies provided data on v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) or a similar metric [31, 32, 33, 36, 74, 78, 80, 85, 89]. Just two of these papers reported statistically significant improvements (3–4%, ES: 0.42–0.49) in the ST group compared to the control group [80, 89]. One study [74] reported a 2.6% improvement (ES: 0.57) and another [33] a 4.0% increase (ES: 0.9) after a 40-week intervention; however, these changes were not significantly different to the control group.

3.8 Blood Lactate Parameters

Blood lactate value was measured at fixed velocities in six studies [77, 78, 81, 82, 84, 92] and velocity assessed for fixed concentrations of BL (2–4 mmol L⁻¹) or lactate threshold (LT) in six studies [32, 33, 79, 81, 90, 91]. One study using young participants observed significantly greater improvements (11–12%) at two speeds compared to the control group [78]. Other studies found no significant changes following the intervention [32, 33, 77, 79, 82, 84, 91] or a change which was not superior to the control group [81, 90, 92].

3.9 Time-Trial Performance

To assess the impact of ST directly upon distance running performance, studies utilized time trials over 1000 m (preceded by 5 × 1 km) [90, 92], 1500 m [88], 2.4 km [87], 3 km [75, 80, 91], 5 km [31, 73], 10 km [88, 89], 5 min [32], and 40 min [38]. There were similarities to competitive scenarios in most studies, including performances taking place under race conditions [31, 75, 87, 90, 91, 92], on an outdoor athletics track [31, 87, 88, 89], on an indoor athletics track [73, 75, 80, 90, 91, 92], and following a prolonged (90-min) submaximal run [38]. Performance improvements were statistically significant compared to a control group for eight of the 12 trials. The exceptions were a 40-min time trial [38], a 1000-m repetition [90, 92], and two studies that used a 3 km time trial [75, 80]. Statistically significant 3 km improvements were observed for all groups in one case [80]; however, the ES was larger for the two intervention groups (0.37 and 0.46) compared to the control group (0.20). Improvements over middle-distances (1500–3000 m) were generally moderate (3–5%, ES: 0.4–1.0). Moderate to large effects (ES: > 1.0) were observed for two studies [31, 88] that evaluated performance over longer distances (5–10 km); however, the relative improvements were quite similar (2–4%) over long distances compared to shorter distances [31, 73, 88, 89].

3.10 Anaerobic Outcomes

Tests relating to anaerobic determinants of distance running performance were used in five investigations. Sprint speed over 20 m [73, 87] and 30 m [78] showed statistically significant improvements following ST (1.1–3.4%). Two studies provided evidence for enhancement of vMART [73, 78], and one further study showed no change in anaerobic running distance after 6 weeks of HRT [31]. A 30-s Wingate test was also used in one paper; however, no differences in performance were noted [89].

3.11 Body Composition

Body mass did not change from baseline in 18 of the studies [32, 33, 36, 38, 72, 73, 74, 75, 77, 79, 80, 81, 83, 84, 86, 87, 88, 89]; however, one investigation reported a significant increase (2%, ES: 0.32) following ST [78]. This study also documented changes in the thickness of quadriceps femoris muscle in both the intervention (3.9%, ES: 0.35) and control group (1.9%, ES: 0.10) [78]. Similarly, an increase in total lean mass (3%) and leg lean mass (3%) was found following 12 weeks of ST despite little alteration in cross-sectional area of the vastus lateralis and body mass being noted [90, 92]. Another study observed a significant decrease (− 1.2%) in body mass in the control group, with no change in the intervention group [32]. A significant increase in leg mass (3.1%, ES: 1.69) was also noted in this study [32, 38]. Other indices of body composition that exhibited no significant changes were: fat mass [33, 36, 72, 73, 78, 86], fat-free mass [36, 72, 86], lean muscle mass [33, 78], skinfolds [83, 89], and limb girth measurements [72, 73, 83].

4.1 Running Economy

Running economy, defined as the oxygen or energy cost to run at a given sub-maximal velocity, is influenced by a variety of factors, including force-related and stretch–shortening cycle qualities, which can be improved with ST activities. In general, an ST intervention, lasting 6–20 weeks, added to the training program of a distance runner appears to enhance RE by 2–8%. This finding is in agreement with previous meta-analytical reviews in this area that show concurrent training has a beneficial effect (~ 4%) on RE [10, 26]. In real terms, an improvement in RE of this magnitude should theoretically allow a runner to operate at a lower relative intensity and thus improve training and/or race performance. No studies attempted to demonstrate this link directly, although inferences were made in studies, which noted improvements in RE and performance separately [73, 80, 88]. Other works provide evidence that small alterations in RE (~ 1.1%) directly translate to changes (~ 0.8%) in sub-maximal [94] and maximal running performance [95]. The typical error of measurement of RE has been reported to be 1–2% [96, 97, 98, 99] and the smallest worthwhile change ~ 2% [94, 98, 100], which is thought to represent a “real” improvement and not simply a change due to variability of the measure. Taken together, it is therefore likely that the improvements seen in RE following a period of concurrent training would represent a meaningful change in performance.

Improvements were observed in moderately-trained [72, 76, 84, 86], well-trained [33, 36, 73, 75, 79, 80, 88] and highly-trained participants [74, 77], suggesting runners of any training status can benefit from ST. Different modes of ST were utilized in the studies, with RT or HRT [72, 78, 79, 84, 86], ERT [80], PT [75, 76, 80], and a combination of these activities [33, 36, 77], all augmenting RE to a similar extent. Single-joint isometric RT may also provide a benefit if performed at a high frequency (4 day week⁻¹) [84]. Several studies adopted a periodized approach to the types of ST prioritized during each 3- to 6-week cycle [33, 36, 77, 88], which is likely to provide the best strategy to optimize gains long-term [101].

Six studies [32, 81, 82, 83, 89, 91] failed to show any improvement in RE and a further six [36, 73, 76, 77, 78, 86] observed both improvements and an absence of change at various velocities. This implies benefits are more likely to occur under specific conditions relating to the choice of exercises, participant characteristics, and velocity used to measure RE. In most studies that observed a benefit, exercises with free weights were utilized [33, 36, 72, 74, 86, 88]. Multi-joint exercises using free weights are likely to provide a superior neuromuscular stimulus compared to machine-based or single-joint exercises as they demand greater levels of co-ordination, multi-planar control, activation of synergistic muscle groups [102, 103] and usually require force to be produced from closed-kinetic chain positions. These types of exercise also have a greater biomechanical similarity to the running action so are therefore likely to provide a greater level of specificity and hence transfer of training effect [104]. An insufficient overload or a lack of movement pattern specificity may therefore be the reason for the absence of an effect in studies that used only resistance machines [32, 81] or a single-joint exercise [82]. These studies were also characterized by a lower frequency of sessions compared to studies that used similar RT exercises but did observe an improvement in RE [78, 84].

Moderately-trained runners were used in three of the six studies showing an absence of effect [81, 83, 91] and one used triathletes who performed a relatively low volume of running (34.8 km week⁻¹) as part of their training [83]. However, a similar number of studies who used recreational athletes did show a positive effect [72, 76, 84, 86], suggesting that training level is unlikely to be the reason for the lack of response in these studies. This is also confirmed by recent observations that showed improvement in RE following a period of concurrent training was similar across individuals irrespective of training status and the number of sessions per week ST was performed [10].

The velocity used to assess RE may also explain the discrepancies in results across studies. It has been suggested that runners are most economical at the speeds they practice at most [98], and for investigations that utilized PT, stretch–shortening cycle improvements are likely to manifest at high running speeds where elastic mechanisms have greatest contribution [83, 105]. Therefore a velocity-specific measurement of RE may be the most valid strategy to establish whether an improvement has occurred. For example, Saunders and associates [77] observed an improvement (p = 0.02, ES: 0.35) at 18 km h⁻¹ in elite runners, but an absence of change at slower speeds. Similarly, Millet and colleagues [74] noted large (ES: > 1.1) improvements at speeds faster than 75% v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (~ 15 km h⁻¹) in highly-trained triathletes, and Paavolainen et al. [73] detected changes at 15 km h⁻¹ but not slower speeds in well-trained runners. Furthermore, Piacentini and co-workers [86] found improvement at race-pace in recreational marathon runners but not at a slower and a faster velocity. Improvements observed at faster compared to slower speeds may also reflect improvements in motor unit recruitment as a consequence of ST. As running speed increases there is a requirement for greater peak vertical forces due to shorter ground contact times, which elevates metabolic cost [25]. To produce higher forces, yet overcome a reduction in force per motor unit as a consequence of a faster shortening velocity, more motor unit recruitment is required [106]. Thus, an increase in absolute motor unit recruitment following a period of ST would result in a lower relative intensity reducing the necessity to recruit higher threshold motor units during running [25]. Several studies that failed to show any response used a single velocity to assess RE [32, 83, 89], perhaps indicating that the velocity selected was unsuitable to capture an improvement. Furthermore, only a small number of studies used relative speeds [33, 74, 79, 81, 82], with most choosing to assess participants at the same absolute intensity. A given speed for one runner may represent a high relative intensity, whereas for another runner it may be a relatively low intensity. Therefore selecting the same absolute speed in a group heterogeneous with respect to \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), may not provide a true reflection of any changes which take place following an intervention. Moreover, this may also confound any potential improvements observed in fractional utilization of \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\).

Several common procedural issues exist in the studies reviewed, which may influence the interpretation of results and therefore conclusions drawn. The majority of studies quantified RE and \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) as a ratio to body mass; however, oxygen uptake does not show a linear relationship with increasing body size [107]. It is also known that the relationship between body size and metabolic response varies across intensities, with a trend for an increasing size exponent as individuals move from low-intensity towards maximal exercise [108, 109]. Moreover, allometric scaling is likely to decrease interindividual variability [110], potentially improving the reliability of observations [99]. Ratio-scaling RE for all velocities to body mass is therefore theoretically and statistically inappropriate [111]. Just two studies [79, 80] used an appropriate allometric scaling exponent (0.75) to account for the non-linearity associated with oxygen uptake response to differences in body mass, both establishing a large ES in their results. The unsuitability of ratio-scaling as a normalization technique when processing physiological data is likely to have influenced the statistical outcomes of some studies and thus inaccurate conclusions may have been generated.

Running economy was expressed as oxygen cost in all but three studies [82, 84, 91], which quantified RE using the energy cost method. As the energy yield from the oxidation of carbohydrates and lipids differs, subtle alterations in substrate utilization during exercise can confound measurement of RE when expressed simply as an oxygen uptake value. Energy cost is therefore the more valid [112, 113] and reliable [99] metric for expressing economy, compared to traditional oxygen cost, as metabolic energy expenditure can be calculated using the respiratory exchange ratio, thus accounting for differences in substrate utilization. Despite attempts to control for confounding variables such as diet and lifestyle in most studies, equivalence in inter-trial substrate utilization cannot be guaranteed, which may have impacted upon the measurement of RE.

4.2 Maximal Oxygen Uptake

Maximal oxygen uptake is widely regarded as one of the most important factors in distance running success [114], therefore the objective for any distance runner is to maximize their aerobic power [9]. An individual’s \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) is limited by their ability to uptake, transport and utilize oxygen in the mitochondria of working muscles. Endurance training involving prolonged continuous bouts of exercise or high intensity interval training induces adaptations primarily within the cardiovascular and metabolic systems that results in improvements in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) [9, 115]. Conversely, ST is associated with a hypertrophy response that increases body mass and has been reported to decrease capillary density, oxidative enzymes and mitochondrial density [116, 117, 118], which would adversely impact aerobic performance. Theoretically there is therefore little basis for ST as a strategy to enhance aerobic power. However it is important to address whether in fact \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) is negatively affected when distance running is performed concurrently with ST.

Thirteen works in this review found no change in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) following the intervention period, demonstrating that although ST does not appear to positively influence \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), it also does not hinder aerobic power. Although ST in most studies was supplementary to running training, it appears that the additional physiological stimulus provided by ST was insufficient to elicit changes in cardiovascular-related parameters [119]. Three studies did observe significant increases in aerobic power that did not differ to the change observed in the control group [33, 81, 91], and one further study found an improvement in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) in the control group only [78]. It is perhaps surprising that more studies did not find an increase in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (in any group) given that participants continued their normal running training through the study period. Improvements in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) of 5–10% have been shown following relatively short periods (< 6 weeks) of endurance training [9]; however, the magnitude of changes is dependent upon a variety of factors including the initial fitness level of individuals and the duration and nature of the training program [120]. Maximal oxygen uptake is known to have an innate upper limit for each individual, therefore in highly-trained and elite runners, long-term performance improvement is likely to result from enhancement of other physiological determinants, such as RE, fractional utilization and v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) [4, 121, 122]. A number of studies used moderately-trained participants [23, 72, 76, 81, 91], who would be the most likely to show an improvement in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) following a 6- to 14-week period of running, with two investigations demonstrating improvements for both groups [81, 91]. The absence of \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) improvement in other papers suggests that the duration of the study and/or the training stimulus, was insufficient to generate an improvement [120]. Indeed, one study of 40 weeks’ duration in Collegiate level runners observed similar improvements (ES: 0.5–0.6) in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) in both groups [33], suggesting a longer time period may be required to detect changes in runners with a higher training status. High-intensity aerobic training (> 80% \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)) is a potent stimulus for driving changes in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)[123]; however, some studies reported runners predominantly utilized low-intensity (< 70% \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)) continuous running [74, 78, 89], which may also explain the lack of changes observed.

4.3 Velocity Associated with \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\)

An individual’s v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) is influenced by their \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), RE and anaerobic factors including neuromuscular capacity [4, 124]. The amalgamation of several physiological qualities into this single determinant appears to more accurately differentiate performance, particularly in well-trained runners [3, 98, 125, 126], therefore v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) has been labelled as an important endurance-specific measure of muscular power [127].

Improvements for v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (3–4%, ES: 0.42–0.49) were found in two investigations [80, 89], with a further two studies observing improvements (2.6–4.0%, ES: 0.57–0.9) that could not be ascribed to the training differences between the groups [33, 74]. A number of studies also found little change in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) following an intervention [31, 32, 36, 78, 85]. As v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) is the product of the interaction between aerobic and anaerobic variables, a small improvement in one area of physiology may not necessarily result in an increase in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\). Damasceno et al. [89] found an improvement in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (2.9%, p < 0.05, ES: 0.42) despite detecting no change in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), RE or Wingate performance, therefore attributed the change to the large improvements (23%, ES: 1.41) in the force-producing ability they observed in participants. Conversely, Berryman and associates [80] found changes in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (4.2%, ES: 0.43–0.49) alongside improvements in RE (4–7%, ES: 1.01), moderate increases in power output, and no change in \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) scores. Beattie and co-workers [33] credited the change in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) they observed (20-weeks: 3.5%, ES: 0.7) to the accumulation of improvements in RE, \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) and anaerobic factors; however, these were not sufficiently large enough to provide a significant group × time interaction. Millet and colleagues [74] found notable improvements in RE (7.4%, ES: 1.14); however, changes in RE could not explain the changes observed in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) (r = − 0.46, p = 0.09). It may also be the case that longer periods of ST are required before an improvement in v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) is detected, as studies showing an improvement (2.6–4.0%, ES: 0.57–0.9) from baseline lasted 14 weeks or more [33, 74], and studies showing little change tended to be 6–8 weeks in duration [31, 78, 85].

The conflicting results could also be explained by the inconsistency in methods used to define v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\). A number of different protocols and predictive methods have been suggested to assess v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) [4], including determination from the \(\dot{V}{\text{O}}_{2}\)-velocity relationship [128] and the peak running speed attained during a maximal test using speed increments to achieve exhaustion [21, 127]. All studies that measured v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) in this review did so via an incremental run to exhaustion progressed using velocity. Velocity at \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) was taken as the highest speed that could be maintained for a full 60-s stage [78, 80, 85], an average of the final 30-s [31, 36], the mean velocity in the final 120-s [32], or the minimum velocity that elicited \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) [33, 74]. Although a direct approach to the measurement of v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) has been recommended [4], due to the velocity increments (0.5–1.0 km h⁻¹) used in these investigations, this may not provide sufficient sensitivity to detect a change following a short- to medium-term intervention. Damasceno and associates [89] calculated v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) using a more precise method based upon the fractional time participants reached through the final stage of the test multiplied by the increment rate. This perhaps provided a greater level of accuracy which allowed the authors to identify the differences in changes which existed between the groups. Taken together, there is weak evidence that v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) can be improved following an ST intervention, despite constituent physiological qualities often exhibiting change. Differences in the protocols used to determine v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) makes comparison problematic; however, a more precise measurement of v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) that accounts for partial completion of a final stage is likely to provide the sensitivity to identify subtle changes that may occur.

The critical velocity model, which represents exercise tolerance in the severe intensity domain, potentially offers an alternative to measurement of v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) that is currently uninvestigated in runners [35, 129]. Two main parameters can be assessed using the critical velocity model; critical velocity itself, which is defined as the lower boundary of the severe intensity domain which when maintained to exhaustion leads to attainment of \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), and the curvature constant of the velocity–time hyperbola above critical velocity, which is represented by the total distance that can be covered prior to exhaustion at a constant velocity [130]. Middle-distance running performance (800 m) is strongly related to critical velocity models (r = 0.83–0.94) in trained runners [131], and may be more important than RE in well-trained runners [35]. Evidence from studies using untrained participants has demonstrated that the total amount of work that can be performed above critical power during high-intensity cycling exercise is improved (35–60%) following 6–8 weeks of RT [132, 133]. Future investigations should therefore address the dearth in literature around how ST might positively influence parameters related to the critical velocity model [35].

4.4 Blood Lactate Markers

A runner’s velocity at a reference point on the lactate-velocity curve (e.g., LT) or BL for a given running speed are important predictors of distance running performance [134, 135, 136]. A runners LT also corresponds to the fractional utilization of \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) that can be sustained for a given distance [114], therefore an increase in LT also allows a greater proportion of aerobic capacity to be accessed.

In contrast to RE, ST appears to have little impact upon BL markers. This is quite surprising as an improvement in RE should theoretically result in an enhancement in speed for a fixed BL concentration. This suggests that adaptations to RE can occur independently to changes in metabolic markers of performance. An absence of change in BL also implies that ST does not alter anaerobic energy contribution during running, thus assuming aerobic energy cost of running is reduced following ST, it can be inferred that total energy cost (aerobic plus anaerobic energy) is also likely to be reduced. Previous studies have shown as little as 6 weeks of endurance training can improve BL levels or the velocity corresponding to an arbitrary BL value in runners [137, 138, 139]. The intensity of training is important to elicit improvement in BL parameters [140], therefore it appears that the running training prescription may have been insufficient to stimulate improvements, or the training status of participants meant a longer period was required to realize a meaningful change. In addition, the inter-session reliability of BL measurement between 2–4 mmol L⁻¹ is ~ 0.2 mmol  L⁻¹ [99], therefore over a short study duration this metric may not provide sufficient sensitivity to detect change.

Training at an intensity above the LT is likely to result in a reduction in the rate of BL production (and therefore accumulation), or an improved lactate clearance ability from the blood [9]. Short duration high-intensity bouts of activity generate high levels of BL so drive metabolic adaptations which can result in an improvement in performance [141, 142, 143]. Studies that have utilized high-repetition, low-load RT in endurance athletes therefore have the potential to produce high BL concentrations so may provide an additional stimulus to improve performance via BL parameters. This theory is supported by works that have demonstrated improvements in BL-related variables in endurance athletes following an intervention that uses a strength-endurance style of conditioning with limited rest between sets [54, 62, 144]. The ST prescription in the studies reviewed was predominantly low-repetition, high-intensity RT or PT, which is unlikely to have provided a metabolic environment sufficient to directly enhance adaptations related to BL markers.

4.5 Time-Trial Performance

Physiological parameters such as \(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\), RE and LT are clearly important determinants that can be quantified in a laboratory; however, for a runner, TT performance possesses a far higher degree of external validity. Similar improvements in TT performance were observed for middle-distance events (3–5%, ES: 0.4–1.0) and long-distance events up to 10 km (2–4%, ES: 1.06–1.5). In the majority of these studies, time trials took place in a similar environment and under comparable conditions to a race, therefore these findings have genuine applicability to “real-life” scenarios. These improvements are likely to be a consequence of significant enhancements in one or more determinants of performance. Interestingly, Damasceno and co-authors [89] found an improvement in 10 km TT performance due to the attainment of higher speeds in the final 3 km, despite observing no change in RE during a separate assessment. This suggests that greater levels of muscular strength may result in lower levels of relative force production per stride, thereby delaying recruitment of higher threshold muscle fibers and thus providing a fatigue resistant effect [145]. This subsequently manifests in a superior performance during the latter stages of long-distance events [89].

Four studies observed no difference in performance change compared to a control group [38, 75, 80, 90, 92]. Vikmoen and colleagues [38] attributed a lack of effect in their 40 min TT to the slow running velocity caused by the 5.3% treadmill inclination used in the test. This was also the only study to use a treadmill set to a pre-determined velocity which participants could control once the test had commenced. The absence of natural self-pacing may therefore have prevented participants achieving their true potential on the test. Spurrs et al. [75] and Berryman et al. [80] both found improvements in 3 km performance compared to a pre-training measure of a comparable magnitude to other studies (2.7–4.8%, ES: 0.13–0.46); however, changes were not significantly different to a control group, suggesting ST provided no additional benefit or there was a practice effect associated with the test.

It could be possible that enhancement of physiological qualities in some studies could be attributed to RT being positioned immediately after low-intensity, non-depleting running sessions [146]. This arrangement of activities in concurrent training programs has been shown to provide a superior stimulus for endurance adaptation compared to performing separate sessions, and without compromising the signaling response regulating strength gains [147, 148]. This, however, appears not to be the case, as most studies reported ST activities took place on different days to running sessions [85, 88, 89] or were at least performed as separate sessions within the same day [33, 36, 38, 72, 75, 78]. Only three studies performed ST and running immediately after one another, with one positioning PT before running [87] and one lacking clarity on sequencing [76]. Schumann and colleagues [90, 92] observed no additional benefit to both strength and endurance outcomes compared to a running only group, when ST was performed immediately following an incremental running session (65–85% maximal heart rate), citing residual fatigue which compromised quality of ST sessions as the reason.

4.6 Anaerobic Running Performance

The contribution of anaerobic factors to distance running performance is well established [127, 149]. In particular, anaerobic capacity and neuromuscular capabilities are thought to play a large role in discriminating performance in runners who are closely matched from an aerobic perspective [124, 150]. An individual’s v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) perhaps provides the most functional representation of neuromuscular power in distance runners; however, measures of maximal running velocity and anaerobic capacity are also potentially important [127].

Tests for pure maximal sprinting velocity (20–30 m) were used in three studies [73, 78, 87] and showed improvements (1.1–3.4%) following ST in every case. This confirms results from previous studies that have shown sprinting performance can be positively affected by an ST intervention in shorter-distance specialists [151, 152, 153]. This finding has important implications for distance runners, as competitive events often involve mid-race surges and outcomes are frequently determined in sprint-finishes, particularly at an elite level [154, 155, 156, 157]. Middle-distance runners also benefit from an ability to produce fast running speeds at the start of races [158], therefore improving maximum speed allows for a greater “anaerobic speed reserve” [159], resulting in a lower relative work-rate, and thus decreasing anaerobic energy contribution [41]. Interestingly, endurance training in cyclists has been shown to improve critical power [160] but reduce work capacity for short duration exercise [161, 162]. It is unknown whether long-term aerobic training has a similar effect on anaerobic running qualities; however, ST offers a strategy to avoid this potential negative consequence.

The velocity attained during a maximal anaerobic running test provides an indirect measure of anaerobic and neuromuscular performance, and has a strong relationship (r = 0.85) to v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) [19]. The vMART is particularly relevant to middle-distance runners because it requires athletes to produce fast running speeds under high-levels of fatigue caused by the acidosis and metabolites derived from glycolysis [163]. Both studies that included this test observed significant improvements in vMART (1.1–3.4%), which can be attributed to changes observed in neuromuscular power as a result of the ST intervention [73, 78]. One study showed no alteration in the predicted distance achieved on an anaerobic running test following 6 weeks of HRT; however, the validity and reliability of the test was questioned by the authors [31]. Performance on a 30 s Wingate test was also unchanged following 8 weeks of running training combined with HRT in recreational participants [89]. This finding perhaps underlines the importance of selecting tests which are specific to the training which has been performed in the investigation.

4.7 Strength Outcomes

Changes in strength outcomes were evident in most studies despite all but one [78] observing no change in body mass. Since strength changes can be ascribed to both neurological and morphological adaptations [164], it is therefore likely that improvements are primarily underpinned by alterations in intra- and inter-muscular co-ordination. It is also known that initial gains in strength in non-strength trained individuals are the consequence of neural adaptations rather than structural changes [118]. An improvement in force producing capability is perhaps expected in individuals who have little or no strength-training experience [165]; however, concurrent regimens of training have consistently been shown to attenuate strength-related adaptation [30].

The seminal paper published by Hickson et al. [48] was the first to identify the potential for endurance exercise to mitigate strength gains, when both training modalities were performed concurrently within the same program. Follow-up investigations have since shown mixed results [166, 167, 168, 169, 170, 171], but evidence from this review clearly demonstrates that, for the distance runner at least, strength-related improvements are certainly possible following a concurrent period of training. Nevertheless, the study designs adopted by the works under review did not include a strength-only training group, thus it is not possible to determine whether strength adaptation was in fact negated under a concurrent regimen. One study using well-trained endurance cyclists with no ST experience, observed a blunted strength response in a group who added ST to their endurance training compared to a group who only performed ST [170]. Based upon this finding and other similar observations [167, 172, 173] it seems likely that although distance runners can significantly improve their strength using a concurrent approach to training, strength outcomes are unlikely to be maximized. Moreover, the degree of interference with strength-adaptation also appears to be exacerbated when volumes of endurance training are increased and the duration of concurrent training programs is longer [30, 146].

4.8 Body Composition

Resistance training performed 2–3 times per week is associated with increases in muscle cross-sectional area as a principal adaptation [174]. Although gains in gross body mass may appear to be an unfavorable outcome for distance runners, the addition of muscle mass to proximal regions of the lower limb (i.e., gluteal muscles) should theoretically provide an advantage, via increases in hip extension forces, minimizing moment of inertia of the swinging limb, and reducing absolute energy usage [25]. It is somewhat surprising that virtually all studies demonstrated an absence of change in body mass, fat-free mass, lean muscle mass, and limb girths. Other than one investigation [33], the duration of the studies that observed no effect on measures of body composition was < 14 weeks, suggesting this may not have been sufficiently long to demonstrate a clear hypertrophic response. There is also a possibility that small increases in muscle mass within specific muscle groups (e.g., gluteals) were present, and contributed to the improvements observed in RE, but these may not have been detectable using a gross measure of mass. Evidence for this may have occurred in the Schumann et al. study [90, 92], who observed increases in total lean mass (3%) despite noting no significant change in body mass or cross-sectional area of the vastus lateralis compared to baseline measures.

The interference effect observed during concomitant integration of endurance and ST as part of the same program may also provide an explanation for the lack of change in measures of mass. Following a bout of exercise, a number of primary and secondary signaling messengers are up regulated for 3–12 h [175], which initiate a series of molecular events that serve to activate or suppress specific genes. The signaling messengers which are activated, relate to the specific stress which is imposed on the physiological systems involved in an exercise bout. Strength training causes mechanical perturbation to the muscle cell, which elicits a multitude of signaling pathways that lead to a hypertrophic response [176]. In particular, the secretion of insulin-like growth factor-1 as a result of intense muscular contraction is likely to cause a cascade of signaling events which increase activity of phosphoinositide-3-dependent kinase (Pl-3 k) and the mammalian target of Rapamycin (mTOR) [177, 178, 179]. There is strong evidence that mTOR is responsible for mediating skeletal muscle hypertrophy via activation of ribosome proteins which up regulate protein synthesis [180]. Prolonged exercise bouts, such as those associated with endurance training, activate metabolic signals related to energy depletion, uptake and release of calcium ions from the sarcoplasmic reticulum and oxidative stress in cells [181]. Adenosine monophosphate activated kinase (AMPK) is a potent secondary messenger which functions to monitor energy homeostasis [182] and when activated, modulates the release of peroxisome proliferator co-activator-1α, which along with calcium-calmodulin-dependent kinases increase mitochondrial function to enhance aerobic function [181, 183, 184]. Crucially though, AMPK also acts to inhibit the Pl-3 k/mTOR stage of the pathway via activation of the tuberous sclerosis complex thereby suppressing the ST induced up regulation of protein synthesis [185, 186]. This conflict arising at a molecular signaling level therefore appears to impair the muscle fiber hypertrophy response to ST and attenuate increases in body mass [186].

4.9 Muscle–Tendon Interaction Mechanisms

The potential mechanisms for the positive changes observed in physiological parameters underpinning running performance were directly investigated in three studies [82, 84, 91], and were inferred from gait measures [36, 73, 74, 75, 77] and strength outcomes in others. It is well documented that muscle–tendon unit stiffness correlates well with RE [187, 188, 189]. Tendons are also highly adaptable to mechanical loading and have been shown to increase in stiffness in response to HRT and PT [84, 190, 191]. Despite observing no statistical effect for HRT on RE, Fletcher and colleagues [82] also found a relationship between the change in RE and the changes observed in Achilles tendon stiffness. Despite these associations, it is likely that improvements in RE are a consequence of the interaction between adaptations to tendon properties and improvements in motor unit activation which influence behavior of force–length-velocity properties of muscles [25]. It tends to be assumed that improved tendon stiffness allows the body to store and return elastic energy more effectively, which results in a reduction in muscle energy cost due to a greater contribution from the elastic recoil properties of tendons [192]. Indeed, authors of studies in the present review have argued that the improvements observed in RE following a period of ST are due to an enhanced utilization of elastic energy during running [36, 73, 74, 75]. An alternative proposal, based upon more recent evidence, suggests the Achilles tendon provides a very small contribution to the total energy cost of running therefore improvements in stiffness provide a negligible reduction in energy cost [193, 194]. Instead, a tendon with an optimal stiffness contributes to reducing RE by minimizing the magnitude and velocity of muscle shortening, thus allowing muscle fascicles to optimize their length and remain closer to an isometric state [25]. A reduction in the amount and velocity of fiber shortening therefore reduces the level of muscle activation required and hence the energy cost of running [193].

The improvements observed in maximal and explosive strength, which can be attributed to increases in motor unit recruitment and firing frequency, enable the lower limb to resist eccentric forces during the early part of ground contact [165] and thus contribute to the attainment of a near isometric state during stance. As the force required to sustain speed during distance running performance is submaximal, the level of motor unit activation needed can be minimized when fascicles contract isometrically [25]. This enables the Achilles tendon in particular to accommodate a greater proportion of the muscle–tendon unit length change during running thereby reducing metabolic cost [194]. Variables which provide an indirect measure of the neuromuscular systems ability to produce force rapidly and utilize tendon stiffness were found to improve in other studies that showed improvements in running performance and/or key determinants [73, 74, 78, 79, 80, 87]. However, some studies found improvements in running-related parameters despite observing no alterations in jump performance [33, 76, 77, 78, 91], rate of force development [36, 75, 77], or stiffness [33, 74, 89] illustrating that measures were insufficiently sensitive to detect change, or a combination of mechanisms is likely to be contributing towards the enhancements observed.

Heavy RT causes a shift in muscle fiber phenotype, from the less efficient myosin heavy chain (MHC) IIx to more oxidative MHC IIa, [195, 196]. A higher proportion of MHC IIa has been shown to relate to better running economy [91, 197, 198]; however, whether changes to MHC properties as a result of ST contribute to an improvement in RE and performance remains to be determined. One previous study provided evidence that 4 weeks of sprint running (30-s bouts) improve RE and also the percentage of MHC IIx [199]; however, the absence of endurance training may partly explain the shift in phenotype. Over a longer period (6 weeks), Pellegrino and co-workers [91] found no measurable changes in MHC isoforms following a PT intervention despite a significant improvement in 3 km TT performance, suggesting that a contribution from this mechanism is unlikely for distance running.

It could also be speculated that improvements in RE due to improved strength might have resulted in subtle changes to running kinematics, thus enabling participants to perform less work for a given submaximal speed [72]. There is currently little direct support for this conjecture; however, previous work has shown that running technique is an important component of RE [200, 201], and improving hip strength can reduce undesirable frontal and transverse plane motion in the lower limb during running [202]. One study in this review did observe a reduction in EMG amplitude in the superficial musculature of the lower limb following ST; however, this wasn’t accompanied by an improvement in RE [83]. This suggests that favorable adaptations in neuromuscular control do not necessarily translate to reducing the metabolic cost of running. Additionally, two studies showed significant increases (3.0–4.4%) in ground contact time during submaximal running after an ST intervention [36, 81]; however, only Giovanelli and colleagues [36] found a corresponding improvement in RE. Several papers have demonstrated an inverse relationship between RE and ground contact times [201, 203, 204], since a lower peak vertical force is required to generate the same amount of impulse during longer compared to short ground contacts [25]. Although there is currently minimal evidence to suggest an ST intervention increases ground contact time during sub-maximal running, this mechanism may in part explain the improvements in RE.

4.10 Strength-Training Prescription

4.11 Limitations

In addition to the limitations already highlighted in this review, there are other weaknesses that should be acknowledged. For many of the studies reviewed, calculation of an ES was possible for the variables measured, which provides insight into the meaningfulness and substantiveness of results. However, despite the qualitative nature of this review, interpretation of findings was predominantly based upon reported probability values, which can be misleading due to low sample sizes and the heterogeneity in the pool of participants studied. A relatively large number of studies have been included in this review; however, several parameters (e.g., v\(\dot{V}{\text{O}}_{{2{ \hbox{max} }}}\) and BL) were measured in only a small number of studies, which increases the possibility that false conclusions may be drawn.

There was also a lack of detail concerning several important confounding variables in studies, such as the nature of running training prescription and participant’s previous experience in ST. All but seven studies [31, 73, 74, 76, 84, 86, 90, 92] identified that participants had not been engaged in a program of ST for at least 3 months prior to the study commencing. Although it is perhaps unlikely that participants in these seven studies were strength-trained, this cannot be discounted and may therefore have influenced findings in these investigations.