1 Introduction
Maximal aerobic capacity (
\(\dot{V}{\text{O}}_{2\max }\)) is a well-established marker of cardio-respiratory fitness and related to long-term health outcomes [
1‐
3], predicting longevity in a dose–response-dependent manner, alongside numerous other phenotypic changes [
4]. In addition to the ‘protective’ health effects ascribed to cardiorespiratory fitness [
5],
\(\dot{V}{\text{O}}_{2\max }\) is also a primary determinant of endurance performance, explaining ~ 20–60% of the variation in performances of different mode and distance, which can be realised by athletes when combined with other sub-maximal endurance performance determinants [
6‐
8]. Despite the suggested limited trainability of
\(\dot{V}{\text{O}}_{2\max }\) [
9], a number of different training approaches have been adopted to support its adaptation, such as repeated high-intensity or continuous endurance exercise, which typically confers
moderate effects [
10]. To augment these responses, or prepare individuals for maximal endurance exercise in hot conditions, there has been substantial interest in manipulation of environmental stressors, on the basis that exposure to a combination of exercise and hot environmental temperatures will exaggerate the effector response and subsequent stimulus for endurance adaptation [
11].
Heat acclimation or acclimatisation (HA) describes processes of serial exposure to artificial or outdoor heated environments, respectively, often conducted in combination with exercise [
12]. Engaging in HA enhances the capacity to thermoregulate in the heat, thus improving heat tolerance via the enhancement of various thermoregulatory mechanisms [
13,
14]. Whilst it is largely agreed that HA improves exercise capacity and
\(\dot{V}{\text{O}}_{2\max }\) in the heat, its transfer to thermoneutral environments has been debated [
15]. Some studies demonstrate 4–13% changes [
16‐
21] and others report no change or a reduction following a range of HA protocols [
22‐
28]. It was reasoned recently that insufficient post-HA adaptation periods could explain these discrepancies [
21], alongside other factors, such as inter-individual differences in adaptation capacity [
29]. It is possible that the HA dose (i.e. thermal load or training load) also explains the magnitude of the observed
\(\dot{V}{\text{O}}_{2\max }\) adaptation [
30]. Indeed, heat acclimation is most commonly conducted in combination with exercise in an ‘isothermic’ (fixed period of time at a fixed pre-determined core temperature) or ‘iso-intensity’ mode (fixed exercise intensity), which can vary in duration but typically ranges between 4 and 14 days [
12]. Iso-intensity modes are preferred for acclimatisation, owing to the less controllable environment, and their time scale is often longer to account for natural variation in the environment; however, this is thought to result in adaptations that are more specific to competition, if planned correctly [
12,
14]. The selected type of heat exposure will drastically alter the subsequent stimuli for adaptation [
12,
14]; however, it is not known how the selected HA modality and loading characteristics affect the adaptation of
\(\dot{V}{\text{O}}_{2\max }\) in hot or thermoneutral environments. Lastly, many HA studies have been appropriately questioned [
31] for the absence of control groups (i.e. participants receiving no heat exposure) in their research design. This increases the risk of biased outcome estimates ascribed to HA and
\(\dot{V}{\text{O}}_{2\max }\) adaptation and requires further investigation.
To date, there have been meta-analyses conducted to evaluate the efficacy of HA on acclimation status and a number of physiological and performance outcomes [
30,
32,
33]; however, while these articles provide detailed insights into broader questions regarding HA, their analytical focus has not been
\(\dot{V}{\text{O}}_{2\max }\) adaptation. This means that a substantial number of papers have been overlooked, and the potential for disparate conclusions on this important measure of cardiorespiratory fitness is possible. Thus, there has been no comprehensive meta-analysis of all HA studies to have measured
\(\dot{V}{\text{O}}_{2\max }\) as an outcome variable. Furthermore, no study has investigated the collective moderating effect of the abovementioned factors, such as HA mode, thermal or training load and post-testing periods on
\(\dot{V}{\text{O}}_{2\max }\) adaptation. Owing to the historical debate of this topic and ongoing consistency of evidence, we sought to meta-analyse all studies (with and without control groups) that have investigated the effect of HA on
\(\dot{V}{\text{O}}_{2\max }\) adaptation in thermoneutral or hot environments. We also performed a number of meta-regressions to establish the moderating effect of selected variables on the variability in
\(\dot{V}{\text{O}}_{2\max }\) adaptation.
4 Discussion
We conducted four meta-analyses to evaluate the efficacy of HA on
\(\dot{V}{\text{O}}_{2\max }\) adaptation in thermoneutral or hot environments, as well as establishing the moderating effect of selected variables on the magnitude of adaptation reported across studies. This has particular relevance for those intending to utilise HA to increase aerobic capacity in cool or hot environments, such as athletes or military personnel. Across all meta-analyses, there was an improvement in
\(\dot{V}{\text{O}}_{2\max }\) following HA, ranging from 0.30 (
small) to 0.76 (
moderate-large) standardised mean changes. These significant improvements were found, irrespective of the hot or thermoneutral environment used during testing, although there were stronger effects in the hot
\(\dot{V}{\text{O}}_{2\max }\) test results. This is consistent with the general recommendation that hot training confers adaptations to performance tests in environments that mimic that of training [
63] and were anticipated based on the principle of environmental training specificity [
64,
65]. Indeed, cardiovascular adaptations to HA (see [
66]) should more thoroughly prepare participants for the severe blood flow conflicts experienced during hot exercise, as compared to thermoneutral [
67], and could explain the descriptively larger trend in hot
\(\dot{V}{\text{O}}_{2\max }\) changes in the heat across the meta-analyses conducted here. This is consistent with the understanding that
\(\dot{V}{\text{O}}_{2\max }\) is primarily limited by central factors (i.e. cardiac output and skeletal muscle blood flow [
68])—reductions of which also impair performance in the heat [
67]. However, the finding of greatest importance revealed herein is that cross-adaptation does occur, with training in the heat augmenting the effect of exercise when targeting thermoneutral
\(\dot{V}{\text{O}}_{2\max }\) changes. This finding has implications for cardiorespiratory health in the general population [
4] and contributes methods with which to enhance the classically described model of endurance performance determinants [
69].
The consistent, positive effect of HA on
\(\dot{V}{\text{O}}_{2\max }\) across the within- and between-participant meta-analyses implies that the absence of a control group (those not receiving HA) does not appear to substantially affect the outcome. Comparison of between- (controlled) and within-participant meta-analyses analyses demonstrated that both the thermoneutral (Hedges’
g = 0.42 and 0.30, respectively) and hot (Hedges’
g = 0.63 and 0.75, respectively) resulted in similar changes in
\(\dot{V}{\text{O}}_{2\max }\). These results should not encourage the use of within-participant time-series designs but should arrest the debate [
15] that HA is equivalent to thermoneutral training in eliciting
\(\dot{V}{\text{O}}_{2\max }\) adaptations, when all studies are considered.
On the basis that the mode of HA could elicit different physiological responses, we conducted sub-analyses to determine the potential moderating effect of isothermal or iso-intensity programmes, which are most commonly adopted in the available literature. However, there was no effect of the adopted HA mode on
\(\dot{V}{\text{O}}_{2\max }\) adaptation (Table
3). This was not anticipated, but was also supported by the finding that neither mechanical intensity nor indices of ambient thermal load during the HA programmes had a moderating effect on
\(\dot{V}{\text{O}}_{2\max }\) changes (Table
3). Unfortunately, the internal core temperature responses could not be statistically assessed owing to the inconsistent or incomplete reporting of these data but descriptive observation demonstrated no trend in the relationship between
\(\dot{V}{\text{O}}_{2\max }\) adaptation and core temperature during HA, in any environment (Table
3). Together, these results indicate that the associated increase in mechanical work rates elicited by iso-intensity models (i.e. more intense exercise) or characteristics of the isothermal models, did not influence the level of adaptation observed. These findings are at odds with the theorised necessity of higher-intensity exercise for
\(\dot{V}{\text{O}}_{2\max }\) adaptation during HA [
21]. Indeed, the lack of difference in
\(\dot{V}{\text{O}}_{2\max }\) adaptation between iso-intensity and iso-thermal designs was not anticipated, as iso-intensity models have been repeatedly shown to augment endurance performance [
21,
43,
70], which is not always the case in isothermal studies [
27,
41]. It was equally surprising that the lack of moderating effect was consistent for hot
\(\dot{V}{\text{O}}_{2\max }\) adaptations, since isothermal heat induction is thought to elicit the greatest thermoregulatory effects [
14,
71], for the reason that core temperature can be controlled by the investigator during HA [
14]. Thus, it is thought that isothermal HA is more likely to enhance the magnitude of daily thermo-effector responses and, in turn, enhances these thermoregulatory defences to a subsequent heat stimulus—as is necessary for the heat acclimated phenotype [
12]. Irrespective of this, the current collection of results contradict the seemingly logical inference that isothermal approaches would confer some additional benefit for maximal testing in hot conditions. This could be related to the more recently reported neutral relationship between internal thermal load (time spent > 38.5 °C) and changes in hallmark acclimation responses, such as core temperature or heart rate [
29]. Collectively, it is likely that gross, multi-organ systemic outcome measures, such as
\(\dot{V}{\text{O}}_{2\max }\), require a mixture of thermal and exercise stimuli, which varies between individuals. The recommendation, based on the current evidence, is that the choice of isothermal or iso-intensity will not affect the
\(\dot{V}{\text{O}}_{2\max }\) outcome in hot or thermoneutral environments.
HA varies in the number of days over which it can be conducted, with short-term heat acclimation (< 7 days) facilitating partial adaptation [
72‐
74], and long-term heat acclimation (often ≥ 7 days) completing this process [
75,
76]. This notion was partially supported in the control group meta-analysis of hot
\(\dot{V}{\text{O}}_{2\max }\) adaption, where the number of HA days significantly moderated the overall effect, such that for every additional day of HA, a 0.29 (
small) standardised mean increase in
\(\dot{V}{\text{O}}_{2\max }\) was observed. However, this was not the case in any of the other meta-analyses conducted herein and this particular meta-analysis was based on a total of four articles, which somewhat limits our confidence in the result. Similarly, the ambient temperature of the HA programme also explained significant variance (Table
1) in the outcome of hot
\(\dot{V}{\text{O}}_{2\max }\) increases in the control group analysis, with hotter temperatures eliciting greater
\(\dot{V}{\text{O}}_{2\max }\) changes, up to a ceiling value of 40 °C. Thus, notwithstanding the smaller sample of studies, longer and hotter HA programmes appear to confer the greatest effects on hot
\(\dot{V}{\text{O}}_{2\max }\) when compared to control groups. This could also relate to the specificity of the stimulus and the total heat exposure [
64,
65] and agree with the notion that longer adaptation periods might be necessary for full hot adaptation [
75,
76]. Interestingly, a recent meta-analysis reported no moderating effect of induction length (HA days) or any other parameter of HA programmes on
\(\dot{V}{\text{O}}_{2\max }\) adaptation [
32] but the number of studies considered for analysis was markedly less and not sub-analysed in the same manner as the current analysis. For example,
\(\dot{V}{\text{O}}_{2\max }\) assessments in hot and thermoneutral environments were amalgamated for analysis, which will produce mixed results in comparison to the current analysis, since these tasks offer distinctly different challenges.
In the meta-analysis of within-group hot
\(\dot{V}{\text{O}}_{2\max }\) adaptation, we found that the post-acclimation testing period was significantly related to the outcome. Across the six studies in this analysis, we report that for every additional post-testing day immediately after the final HA intervention, a 0.17 (
small) standardised change in
\(\dot{V}{\text{O}}_{2\max }\) can be expected, up to a seven-day limit. In other words, testing or planned performance too close to the final day of HA is not advisable if complete hot
\(\dot{V}{\text{O}}_{2\max }\) adaptation is to be realised. Extending this up to 7 days appears to be optimal. It should be acknowledged, however, that reporting of this was inconsistent, with many identifying the exact number of days and others stating a time period (i.e. within 7 days). Therefore, there is likely to be some variability in this moderator. Nevertheless, this is an important finding for those utilising heat acclimation to prepare for endurance exercise in hot environments and means that the suggested urgency (see [
65]) to acclimate/acclimatise near to competition might be unwarranted if
\(\dot{V}{\text{O}}_{2\max }\) is considered to be of greatest importance. This is inconsistent with the immediate decays reported in other physiological measures, such as core temperature and heart rate, following HA [
12,
73,
77] and is likely to be related to the delayed adaptive responses (absence of immediate decay) to heat acclimation that have been demonstrated in some studies [
21,
78,
79]. This phenomenon can be likened to classical dose–response theories of Seyle [
80] and explained directly by the severity of the imposed internal thermal load and necessary recovery that subsequently ensues to permit phenotypic adaptation [
81]. Given that this was observed in the within-participant hot
\(\dot{V}{\text{O}}_{2\max }\) analysis, rather than the thermoneutral equivalents, it is possible that the specificity of the hot environment during post-testing underpins this theory. However, experimental work has recently demonstrated a similar pattern of adaptation following heat acclimation when testing in thermoneutral environments [
21]. Therefore, further work is needed to understand this phenomenon and its physiological determinants.
As we anticipated, there were numerous inconsistencies between studies, which limit some of the conclusions of the current meta-analyses and should be considered when using these results to inform HA programme design. For example, core body temperature in response to HA sessions was often reported as either a mean or a final temperature reached. Whilst there is typically a relationship between these variables, it would be helpful for readers if more complete reporting of the mean, standard deviation and final core body temperatures is provided. In addition, the measurement of core body temperature (rectal, oesophageal, tympanic, ingestible pill) often varies between studies and could alter the magnitude of the response. Whilst we used SMD to control for differences in measurement type in the current meta-analyses, readers should be aware of this when evaluating the raw data we presented herein from previous studies. Finally, the number of days between HA completion and post-testing of \(\dot{V}{\text{O}}_{2\max }\) should be reported more consistently among studies, perhaps through submission of raw data to support summary findings, since this appears to affect the \(\dot{V}{\text{O}}_{2\max }\) measurement. More exact reporting of this would help to understand the consistency of this conclusion and improve the design of HA programmes, if \(\dot{V}{\text{O}}_{2\max }\) improvement is assumed to be a desirable outcome.