Altitude training (AT) has been a matter of extensive research for half a century, and despite some skeptical views (23), it still plays an important role in the preparation of athletes in many countries (37,47). The global theoretical concept behind this practice is the independent and combined effects of the physiological processes of acclimatization to chronic hypoxia and those derived from training under the additional stress imposed by exercising in a hypoxic environment.
The classical approach (“live high, train high,” Hi-Hi), used since the late 1960s, involves sea level (SL) resident athletes who travel to and subsequently live and train at moderate altitude, typically 1800–2500 m for 2–4 wk; a similar approach is also practiced by athletes who reside full time at altitude. Despite being used by very many elite swimmers and coaches, there is a remarkable lack of controlled studies on AT in swimming in the scientific literature, and there is no clear evidence that training at natural altitude enhances performance more than training at SL (31).
In a series of studies published in the 1990s, Levine and Stray-Gundersen (22) provided sound evidence that the “live high, train low” (Hi-Lo) strategy can improve 3000- to 5000-m running performance in US collegiate athletes. This approach was modified to limit the low-altitude training sessions to only high-intensity workouts and was subsequently termed “live high, train high and low” (41). Although there was substantial individual variability in the response (6), the improvement in running performance was associated with increase in red cell mass, the subsequent increase in maximal oxygen uptake (V˙O2max), (the high altitude effect), and the maintenance of high-intensity training velocities and oxygen flux to the muscles (the low altitude effect) (41). This paradigm has now been confirmed repeatedly by multiple investigators in elite endurance athletes performing different sports including running (41), orienteering (46), and cycling (15). However, these studies are difficult to compare with each other directly, given the many differences in experimental design including type of athletes being compared, performance measures, control of confounding variables, placebo/nocebo effects, and, most importantly, the nature of altitude exposure (i.e., true exposure to terrestrial altitude or nighttime exposure to normobaric hypoxic gas) (17). Even though in the last decade, the Hi-Lo approach has largely supplanted classical AT in the scientific literature and among many endurance athletes (37), no studies have been conducted using the Hi-Lo strategy in swimmers. Another key unanswered question, which is rarely addressed, concerns the proper timing of return to SL before competition (4).
In view of the disconnect between research evidence and practical use of AT, particularly in elite swimmers, an international group of investigators convened to conduct an international multidisciplinary and collaborative research project (The Altitude Project) to examine the effect of different current AT strategies on performance, technique, and health status of elite swimmers. This article is focused on performance, oxygen uptake kinetics, and hemoglobin mass and will be accompanied by other publications dealing with other aims of this project.
Consequently, this study aimed 1) to test the hypothesis that living at moderate altitude (2230 m) and training both at moderate and at lower altitude for 4 wk (Hi-HiLo) improves SL swimming performance more than living and training at altitude (classical terrestrial AT) for 3 (Hi-Hi3) or 4 wk (Hi-Hi) or than living and training at low altitude (conventional Lo-Lo training), 2) to elucidate whether the adaptive mechanisms conform with the “erythropoietic paradigm” (i.e., are mainly hematologic in nature, via the activation of erythropoiesis by induced hypoxia, with subsequent increase in V˙O2max), and 3) to quantify the eventual effect of the different interventions on performance on return to SL and to track changes during a lengthy period of 4 wk without concurrent tapering.
The study was designed as a controlled, nonrandomized, four parallel groups trial, comparing changes in swimming performance, V˙O2max, oxygen kinetics parameters, and total hemoglobin mass (tHbmass) after an experimental intervention consisting of training camps in four different conditions, as follows: 1) living and training at moderate altitude (2320 m above SL) for 4 wk (Hi-Hi), 2) identical intervention for 3 wk (Hi-Hi3), 3) living at altitude (2320 m) and training at both moderate and low altitude (690 m) for 4 wk (Hi-HiLo), and 4) living and training at near SL (190 or 655 m) for 4 wk (Lo-Lo).
All athletes were studied at the same point in their competitive season as noted in the following section. At the onset of the study, coaches were instructed to conduct a lead-in training program the week before initial testing, in which training load was reduced to avoid excessive fatigue before baseline swimming assessments. Then, all swimmers and their coaches traveled to Sabadell, Barcelona (190 m) or Madrid (655 m), Spain, where they stayed for 3–5 d for baseline testing. Next, all swimmers allocated to the AT groups traveled to the High Altitude Training Center at Sierra Nevada (2320 m), Spain, where they lived for 3 or 4 wk. The two Lo-Lo subsamples lived and trained in Sabadell (n = 4) and the High Performance Center (CAR) at Madrid (n = 7). One of the Hi-Hi3 subgroups (n = 6) was tested in Granada, Spain (690 m). In all cases, baseline and final testing were conducted at the same location and facility. To minimize instruction bias (placebo/nocebo effects), the group allocation was performed according to previous agreement with the team head coaches as to their preferred mode of intervention (i.e., Hi-Hi or Hi-HiLo altitude camps for 3 or 4 wk or SL camps) and coach and athlete acceptance of its characteristics and procedures. To minimize bias from differences in individual team training approaches, each experimental group was composed of swimmers of at least two different teams and nationalities.
To evaluate the effects on performance and physiological parameters, before the experimental intervention (PRE), all subjects took part in three testing sessions with the following content: 1) two swimming time trials over 50 m (TT50) and 100 or 200 m (TT100 or TT200), 2) one swimming time trial over 400 m (TT400), and 3) one 4 × 200 m incremental swimming test (T4×200). All measurements were repeated 1–2 d (POST) and 1 (PostW1), 2 (PostW2), 3 (PostW3), and 4 (PostW4) wk after completion of the training camp. This way, both the immediate effects and the off response of the intervention could be assessed. tHbmass was measured in duplicate (Pre1 and Pre2) on two separate days, and the average was adopted as a baseline value (PRE). Measurements were repeated once at the first (W1), second (W2), third (W3), and fourth (W4) week during the intervention period in all altitude groups and at W4 in the SL training group (Lo-Lo). An outline of the study design and testing schedule is shown in Figure 1.
Sixty-one swimmers, 34 female and 27 male participants, were recruited as subjects for the present study. Sample size was calculated on the basis of a potential increase of V˙O2max of 5% (mean ± SD, 3 ± 3.2 mL·kg−1·min−1) based on previous studies (22), requiring 11 athletes per intervention group (β = 0.80, α = 0.05), thus a total of 44 athletes. The subjects were swimmers from eight countries (Australia, Brazil, China, Great Britain, Netherlands, Slovenia, Spain, and Tunis). Selection criteria were to have competed internationally during the previous season and/or being preselected as a member of their national and/or Olympic teams. Exclusion criteria included residence at altitude greater than 1000 m in the previous 6 months, recent illness or injuries preventing normal training and racing, and having low ferritin levels at the beginning of the study (<20 ng·mL−1 for females and <30 ng·mL−1 for males). All subjects—and their legal guardians in swimmers under 18 yr of age—gave their informed written consent to the study that had received approval from the Ethics Committee for Clinical Sport Research of Catalonia. To quantify the competitive level of the subjects, the FINA Point Scoring system was used and a point score (range, 0–1100) was ascribed to each swimmer according to her or his best time in her or his main event, scaled up or down from 1000 points on the basis of the global 2012 fastest performance in each event. The swimmers were categorized as sprinters or nonsprinters—i.e., specialists in 50–100 m or in 200–1500 m, respectively—according to their best event. Before data analysis, one female participant was excluded because of low baseline ferritin levels, one male and one female subject were excluded from the training camp by their coach, one male and one female swimmer were excluded because they could not perform the POST tests, and one male and one female subject were excluded because they had to withdraw participation for personal reasons. Fifty-four subjects, 30 female and 24 male, successfully completed the intervention protocol (Table 1). After final group allocation, there were no significant differences among the four experimental groups in terms of performance level, body height or mass, V˙O2max, or tHbmass. The swimmers in the Lo-Lo group (19.6 ± 2.9 yr) were younger than those in the Hi-HiLo group (mean ± SD, 23.7 ± 3.5 yr) (P = 0.02), but there were no differences among the other groups.
Evaluation of Performance
Swimming time trials
The primary outcome measure of this study was swimming performance, as measured in time trials over three distances: 50-m front crawl (TT50), 400-m front crawl (TT400), and 100 or 200 m at personal best stroke for sprinters and nonsprinters, respectively (TT100 or TT200). All tests were conducted at the same 50-m indoor pool (temperature: water, 26°C–27°C; air, 27°C–28°C). After a standard competition warm-up, swimmers were instructed to achieve the best time possible in each trial, in which they swam alone. Start was given as in a competition, and time was manually recorded to the nearest 0.01 s by three experienced timers, one of whom was the swimmer’s personal coach. The median values were used for analysis. HR was continuously monitored (CardioSwim; Freelap, Switzerland), and V˙O2 was measured immediately after exercise for 3 min. The testing schedule at PRE and POST comprised two consecutive days. On day 1, two sessions were performed, as follows: T4×200 was administered in the morning session and TT100 or TT200, in the afternoon session. On day 2, T4×200 was repeated in the morning and TT50 and TT400 were performed in the afternoon, allowing a recovery period of at least 60 min in between. The testing schedule at POST to PostW4 comprised two afternoon sessions after a light morning training session. On day 1, TT50 and TT400 were administered. At day 2, TT100 or TT200 was performed.
The secondary outcome measure of the study was V˙O2max measured with a 4 × 200 m incremental swimming test (T4×200) at the same 50-m indoor pool. An incremental swimming exercise protocol based on a previous protocol in the swimming flume (34) was used. After approximately 30-min warm-up, subjects swam three times 200-m front crawl at paced speeds (0.9, 1.0, and 1.1 m·s−1 for women and 1.0, 1.1, and 1.2 m·s−1 for men). Swimming pace was controlled using a computer programmed to produce audio signals at set intervals. An assistant outside the pool entrained his walking pace to the audio signals and to marker cones placed at 5-m intervals at the poolside. The assistant carried a pole with an attached thin nylon line and a red ribbon at the end. Subjects were instructed to follow closely the red ribbon moving in front and below the water surface. Swimming laps were also timed each 50 m. After the submaximal paced swims (results reported separately), after 10 min of passive recovery, subjects completed an all-out 200-m front crawl swim to determine V˙O2max.
Oxygen uptake kinetics and V˙O2max
V˙O2 was measured using a telemetric portable gas analyzer (K4 b2; COSMED, Italy) that was held suspended over the water by an assistant following the swimmer along the pool with minimal intended interference with her or his swimming movements. This equipment was connected to the swimmer by a low-resistance respiratory snorkel and valve system (32). Pulmonary V˙O2 values during the T4×200 swimming test were measured breath by breath and then time-aligned to the start of exercise and plotted against time. No smoothing procedures were applied to avoid distortion of the underlying signal at the transient phase. V˙O2 data during the maximal 200-m swim were fitted using a nonlinear least-square regression technique implemented in Matlab R2010b (MathWorks, Inc.).
For the analysis of V˙O2max kinetics, the first two phases of the generally adopted three-phase model were identified because the exercise duration and intensity constrained the appearance of the slow component (phase III) (38). Phase I (cardiodynamic component) was determined as the time from the onset of exercise to a point of sharper increase in V˙O2, and its duration was computed as time delay for the primary component (TDp). Phase II (principal component) parameters were estimated using a monoexponential model according to the following equation:
where t (s) is the time from the onset of exercise, A0 is the baseline amplitude, Ap is the amplitude of the principal component, TDp (s) is the time delay of the first exponential term and equals the duration of phase I (cardiodynamic component), and τp is the time constant of the principal component. The total amplitude (Atot) was calculated as Atot = A0 + Ap. On a preliminary analysis, V˙O2max was calculated from the 200-m maximal swim as the asymptotic amplitude of the monoexponential equation (Atot) and as the last 20-s averaged values (3450 ± 711 vs 3364 ± 713 mL·min−1, respectively; P = 0.105). Because both values were not different, Atot was then chosen to best represent the highest V˙O2 attained during the maximal 200-m swim test and, thus, as the swimmer’s V˙O2max (30). The reliability of V˙O2max measurements was characterized by a typical error (TE) of 3.1% (95% confidence interval (CI), 1.1%–5.1%; n = 9).
Training and Monitoring
Individualized training plans were developed by the swimmers’ personal coaches, all of them very experienced in AT and well acquainted with the special characteristics of the environment. They were free to implement their own training program according to their previous experience and the swimmer’s fitness level and previous and expected individual response to altitude. All coaches emphasized endurance, short-interval speed, strength, and flexibility. Typically, training schedules included a morning and an afternoon pool session and a dry land workout generally oriented to strength and flexibility from Monday to Saturday morning. Throughout the training camp, careful resting and exercise HR and training time monitoring permitted calculation of the “training impulse” for each daily workout. The study was carried out during the first macrocycle (short-course season) of the Olympic year (October to December 2011) before the London 2012 Olympic Games. The intervention period is composed of a 3- or 4-wk mesocycle during the general preparation phase (early/mid-October up to early/mid-November). The performance follow-up period (POST) is composed of part of the specific preparation macrocycle (early/mid-November up to early/mid-December) and included some midseason races. Before participation, an agreement was attained with all coaches that training load would not substantially be reduced and there would be no tapering implemented during the 4-wk follow-up phase to allow the full assessment of the training intervention. Before all performance tests, coaches were asked to reduce training load the day before to minimize the influence of fatigue.
Pool training was monitored using waterproof HR monitors (CardioSwim; Freelap, Switzerland), which recorded beat-by-beat HR and lap times. Beacon transmitters (TX H2O; Freelap, Switzerland) were placed at the ends of the swimming pool so that the HR monitors’ microprocessor units could register the lap times, rest intervals, and 50-m average speed. HR was assessed from RR intervals, 1-s interpolated, and averaged for 5-s intervals. Dry land training was monitored using beat-by-beat Polar RS800CX monitors (Polar Electro Oy, Finland) with the same 5-s interval averaging. At the end of each training session, data were downloaded, processed, and stored using the Freelap Manager and Polar Pro-Trainer-5 software.
Estimation of training load
The training impulse method (TRIMP) (2) was used to quantify the internal training load. In addition, to improve estimation when monitoring interval training sessions, a modified calculation method (cumulative training impulse method (TRIMPc)) (13) was used, as follows:
where TRIMPc is cumulative TRIMP (a.u.) and n is the total (cumulative) number of exercise and rest intervals during the training session, each with its corresponding HR ratio. To allow the comparison of the TRIMPc values at SL and at 2320 m above SL, we used a correction factor based on the results of the experiments conducted by Wehrlin and Hallén (45) on the changes in V˙O2 and other cardiorespiratory and metabolic parameters with increasing altitude in treadmill running. On the basis of their data, for the same absolute load (55% of V˙O2max), HR was approximately 8% (90% CI, 6%–10%) higher at 2300 m of altitude compared with near SL (300 m), and maximal HR decreased by approximately 2.1% (90% CI, 1.6%–2.6%). To minimize the effect of the elevation in resting HR on TRIMPc calculations, we used the resting HR values during a variability test (6 min at supine position first time in the morning) measured daily (first 9 d) or twice a week (for the rest of the training camp).
Training log and clinical surveillance
During the training camp, each athlete kept a detailed training log, which included 1) a self-administered 10-item questionnaire to assess training intensity within 30 min after each training workout (session or s-RPE), 2) a 10-item fatigue questionnaire (TSF-10), modified from a previous seven-item questionnaire developed for swimmers (1), 3) the Lake Louise Score questionnaire for the assessment of acute mountain sickness symptoms, and 4) changes in health status (illness, injury, clinical symptoms, menstruation, etc.) and well-being, along with the outcome of the consultation with a health professional. Only the results of the two first assessments will be reported here.
Nutrition and Iron Supplementation
During the entire study, high-quality sports nutrition was provided to subjects under the supervision of their team nutritionist or physiologists with support from the nutrition staff of the training centers on demand. Subjects were provided ad libitum access to water at all times and were strongly advised to drink as often as possible. One month before starting the study, teams were asked to perform a blood test on all swimmers at their country of residence. Iron supplementation (100 mg of Fe++per os daily) was strongly recommended even for those with normal ferritin values but not imposed. For a few subjects with suboptimal ferritin level (<50 ng·mL−1) or history of iron deficiency or anemic episodes, treatment under prescription and supervision of their medical staff was a requirement for participation. During the lead-in period, serum ferritin concentration was analyzed using an electrochemiluminescence immunoassay (Roche Diagnostics GmbH, Mannheim, Germany). One female subject had low ferritin level (17 ng·mL−1) and was excluded from the study. The mean value (± SD) was 110 ± 73 ng·mL−1 (range, 28–380). Ferritin was also monitored weekly in all altitude groups during the intervention period.
Total Hemoglobin Mass
tHbmass was measured using the optimized CO rebreathing method, as described by Schmidt and Prommer (39) with some modifications (16,27). Briefly, the subjects inhaled a bolus of carbon monoxide (1.0 mL CO·kg−1 for males and 0.8 mL·kg−1 for females) followed by 3-L pure oxygen and rebreathed in a closed-system spirometer (SpiCo, Bayreuth, Germany) for 2 min. The volume of CO administered at altitude was adjusted according to barometric pressure. Arterialized blood was sampled from an earlobe before and after the rebreathing period (at 6 and 8 min) for analyzing carboxyhemoglobin (COHb) using a CO hemoximeter (OSM3; Radiometer, Denmark). COHb was measured in sextuplicate before and in triplicate at 6 and 8 min after starting the inhalation period. To calculate the amount of CO not taken up during the inhalation time and the amount exhaled after the test, the remaining CO in the spirometer and the end-tidal CO concentration were determined using a portable CO analyzer (Draeger, Luebeck, Germany). In our mobile laboratory, the reliability of this method, determined during this study by test–retest, was characterized by a TE of 1.3% in male athletes and 1.5% in females, averaging 1.35% (95% CI, 0.10%–2.65%).
Training Camp and Placebo/Nocebo Effects
All training camps were conducted in training centers of international standards, whether at SL or at altitude, where subjects lived and trained as a group for the whole intervention period. In the recruiting phase, coaches were offered to choose among the four different interventions and were asked to take to the training camp only those swimmers who had positive or neutral expectations regarding the effect of that intervention in their preparation and future performance. To evaluate eventual placebo or nocebo effects of the intervention, two ad hoc questionnaires were administered at the beginning and at the end of the intervention period, before PRE and POST testing, respectively, one for the head coaches and one for the swimmers. On their questionnaire, coaches were asked to state whether (yes, no, or not sure) they believed that the chosen intervention would help (PRE) or had helped (POST) the swimmers improve their swimming performance and whether they would choose again the same intervention as that at the time of entering the study (POST). On their questionnaire, swimmers were asked to state whether they believed that their training camp would (PRE) or did (POST) help them improve their swimming performance.
Descriptive data are presented as arithmetic means ± SD. Effects on performance, O2 kinetics parameters, and tHbmass are expressed as percent change values (Δ%) and 90% CI of the mean (±90% CI). Values of 95% CI are used for other variables as indicated. To assess the relation among V˙O2max, tHbmass, and performance indicators, the Pearson correlation coefficient (r) and coefficient of determination (r2) were used. To assess the effect of the intervention on swimming time trial performance over time, the primary outcome measure of this study, a comparison among all time points was made using percent change from PRE baseline values. We used the linear mixed modeling procedure for repeated measures (Proc Mixed) using the statistical package SAS (version 9.1.3; SAS Institute, Cary, NC) to estimate means for main effects and interaction of group (Hi-Hi, Hi-Hi3, Hi-HiLo, and Lo-Lo) and test (PRE, POST, PostW1, PostW2, PostW3, and PostW4). Where a significant effect was obtained, a post hoc analysis (Tukey test) was performed to identify the source of differences. The same analysis was performed for tHbmass measurements using the mean of Pre1 and Pre2 as baseline values (PRE) and interaction of group and test during the intervention (PRE, W1, W2, W3, and W4). Because preliminary analysis revealed differences in TRIMPc among groups, ANCOVA was carried out using TRIMPc as a covariate for performance in the three time trials. To evaluate the effects of the intervention on V˙O2max, a statistical comparison was conducted between the PRE score and the score obtained after the intervention period (POST) using a two-tailed paired t-test in the four experimental groups. Time trial swimming performance was assessed only once, considering the high reliability of these measurements on a previous study with trained swimmers for 100-m (TE, 1.4%; 95% CI, ±1.5%) and 400-m (TE, 1.5%; 95% CI, ±1.5%) time trials (34) and to avoid undue psychophysical stress to the subjects and risk of underperformance. Precise P values are reported, and significance level was set at P (probability of type I error) < α = 0.05.
After the intervention period, all coaches (n = 8; 100%) responded to the ad hoc questionnaire that they would have chosen again the same intervention as that at the time of entering the study and that they expected that the chosen intervention would help the swimmers improve their performance. On their PRE questionnaire, all swimmers stated that (“yes”) they had chosen to participate in the training camp in the belief that it would help them improve their swimming performance after the intervention. On their POST questionnaire, their answers to the same question were “yes” (n = 49; 91%), or “not sure” (n = 5; 9%). These last subjects belonged to the Lo-Lo (n = 1), Hi-Hi (n = 2), and Hi-Hi3 (n = 2) groups. No subjects answered “no.”
Throughout the camp, daily average TRIMPc was greater in Hi-HiLo (258 ± 95) than those in Hi-Hi (205 ± 102; P = 0.01), Hi-Hi3 (177 ± 115; P < 0.001), and Lo-Lo (209 ± 100; P = 0.006). Mean daily TRIMPc values were greater in females than those in males (245 vs 185; P < 0.001). TRIMPc, an objective indicator of internal training load, showed strong correlation with perceived exertion after training (s-RPE score) (r = 0.724; P < 0.001). Mean daily s-RPE scores throughout the camp were greater in Hi-Hi3 (5.3 ± 1.8) than those in the other three groups (Hi-Hi: 4.4 ± 1.9, P < 0.001; Hi-HiLo: 4.8 ± 1.5, P = 0.01; and Lo-Lo: 4.6 ± 1.8, P < 0.001). Mean daily TSF-10 scores were also higher (P < 0.001) in Hi-Hi3 (25.0 ± 7.5) than those in the other three groups (Hi-Hi, 20.2 ± 6.1; Hi-HiLo, 18.6 ± 5.4; and Lo-Lo, 21.0 ± 10.0) (see Figure, Supplemental Digital Content 1, Comparing the three training load indicators throughout the camp, https://links.lww.com/MSS/A506).
Primary Analysis: Swimming Performance
The time course of time trial performance over time is presented in Table 2 and Figure 2.
Immediately after the training camp (POST), TT50 performance remained nearly stable in all groups but deteriorated in Hi-Hi3 and remained slower than at PRE until PostW3 (Fig. 2A). At PostW1, all groups improved performance (Hi-Hi3 could not be tested) compared with PRE, as follows: Lo-Lo (−2.0%; ±1.6%; P < 0.001), Hi-Hi (Δt = −4.0%; ±0.9%; P < 0.001), and Hi-HiLo (Δt = −4.8%; ±0.4%; P < 0.001). From that time point, the Hi-HiLo group tended to improve sprinting performance, reaching the highest change (Δt = −5.5%; ±1.0%; P < 0.001 from PRE) at the end of the 4-wk follow-up period and exceeding the change in the Lo-Lo control group at that point (Δt = −3.2%; ±1.1%; P < 0.001 from PRE) (group–test interaction, P = 0.01). The rest of the groups stabilized their performance, reaching equally significant changes from PRE values by the end of the study as compared with the Lo-Lo controls (Hi-Hi3: Δt = −3.4%, ±4.0%, P < 0.001; Hi-Hi: −3.7%, ±1.2%, P < 0.001).
Immediately after the training camp (POST), all groups tended to decrease 400-m TT performance, whereas the Hi-HiLo tended to improve and swam faster than the Hi-Hi group (group–test interaction, P = 0.03) (Fig. 2B). At PostW1, all groups experienced nearly identical improvement in performance (Δt approximately −2%), but the Hi-HiLo group continued to improve at PostW2 (Δt = −4.2%; ±0.9%; P < 0.001), the point at which their improvement from PRE was greater compared with that in all other groups (group–time interaction, P < 0.001). By the end of the follow-up period, both the Hi-HiLo (Δt = −4.7%; ±1.1%; P < 0.001) and the Hi-Hi swimmers (Δt = −3.3%; ±1.3%; P < 0.001) had improved more than the Lo-Lo controls (Δt = −1.6%; ±1.0%; P < 0.001) (group–test interaction, P = 0.001 and 0.03, respectively). Even if the linear mixed model was not able to detect significant differences between both 4-wk altitude groups (group–test interaction, P = 0.23), the size of the effect in the Hi-HiLo group was “most likely” greater than that in the Hi-Hi group.
TT100 or TT200
Since sprinters and nonsprinters swam 100 and 200 m, respectively, at their best personal stroke, this outcome was considered the most specific performance assessment in terms of distance and stroke and they are presented combined in Figure 2C. Immediately after the training camp, all groups performed similarly as compared with PRE, except for Hi-Hi3 (Δt = +1.9%; ±1.3%; P = 0.06) that was slower than PRE and got worse as compared with Lo-Lo, Hi-Hi, and Hi-HiLo (group–test interaction, P = 0.006, 0.03 and <0.001, respectively). At PostW1, all groups improved similarly to the Lo-Lo controls (mean Δt = −2% to 3.5%) and only the Hi-HiLo group improved more than Hi-Hi3 (group–test interaction, P = 0.03). From that point, the Hi-HiLo group progressed faster than the rest at PostW2 (Δt = −5.3%; ±1.4%; P < 0.001) and PostW4 (Δt = −6.3%; ±1.2%; P < 0.001), and by the end of the follow-up period, these improvements were substantially greater compared with Lo-Lo (Δt = −3.7%; ±1.0%; P < 0.001), Hi-Hi3 (Δt = −3.1%; ±0.9%; P < 0.001), and Hi-Hi (Δt = −3.4%; ±1.0%; P < 0.001) (group–test interaction, P = 0.02, 0.002, and <0.001, respectively).
As mentioned before, the daily average TRIMPc throughout the training camp was greater in Hi-HiLo than that in the other groups. To ascertain the eventual effect of training load on performance changes, an additional ANCOVA analysis was carried out using TRIMPc as a covariate for all time trials. A significant group–TRIMPc interaction was observed for TT400 only (P = 0.002), and, therefore, main effects were reassessed, adjusting for TRIMPc. The results confirmed the differences between groups at all time points, with the exception of differences between Hi-Hi and Lo-Lo at PostW4 in TT400 that became not significant (P = 0.08).
Oxygen Uptake Kinetics and V˙O2max
Table 3 shows the V˙O2 kinetics parameters during T4×200.
There were no significant changes in V˙O2max (Δ% from PRE, P > 0.05) in either Lo-Lo (1.9%; ±1.5%), Hi-Hi3 (1.5%; ±2.5%), Hi-Hi (1.1%; ±2.6%), or Hi-HiLo (1.3%; ±1.4%), with larger variability in the individual changes in both groups living and training at altitude (see Figure, Supplemental Digital Content 2, Showing individual and group mean changes, https://links.lww.com/MSS/A507).
No significant relation between relative percent change in V˙O2max and percent change in TT400 performance was found for the entire group of subjects (r = −0.01, P = 0.95) or for the swimmers in each group (r = −0.22, 0.68, −0.12, and −0.39; P = 0.55, 0.09, 0.69, and 0.19, respectively, for Lo-Lo, Hi-Hi3, Hi-Hi, and Hi-HiLo). Likewise, there was no relation between change in V˙O2max and change in TT100 or TT200 performance neither for all subjects (r = 0.10, P = 0.50) nor for the swimmers in each group (r = 0.27, −0.002, 0.20, and 0.06; P = 0.46, 1.00, 0.52, and 0.84, respectively, for Lo-Lo, Hi-Hi3, Hi-Hi, and Hi-HiLo) (see Figure, Supplemental Digital Content 3, Showing the regression plots, https://links.lww.com/MSS/A508).
Total Hemoglobin Mass
tHbmass was not different among groups before the training camp (PRE, P = 0.49). The magnitude of tHbmass changes at altitude showed remarkable between- and within-groups variability (see Figure, Supplemental Digital Content 4, Showing individual and mean changes, https://links.lww.com/MSS/A509).
Figure 3 shows the time course of percent changes in tHbmass during the training camps. In the Hi-Hi group, tHbmass continuously increased from PRE (766 ± 187 g) for 3 wk (W1, 779 ± 192 g, P = 0.03; W2, 796 ± 196 g, P < 0.001; W3, 810 ± 204 g, P < 0.001) and remained nearly unchanged at W4 (815 ± 202 g, P < 0.001). In the Hi-Hi3, tHbmass did not change from PRE (816 ± 205 g) at W1 but increased at W2 (857 ± 223 g, P < 0.001) and W3 (851 ± 227 g, P = 0.02). In contrast, no significant changes were found in the Hi-HiLo group from PRE (896 ± 167 g) at any point during the training camp. Compared with that in PRE, increase in tHbmass was more pronounced in the Hi-Hi group (at W4, 6.2%; 90% CI, ±1.1%; P < 0.001) than that in the Hi-Hi3 group (at W3, 3.8%; ±2.3; P = 0.08; group–test interaction, P = 0.02), whereas no significant changes were found in the Hi-HiLo group (at W4, 1.3%; ±1.8; P = 0.71). Relative changes (Δ% compared with PRE) in tHbmass were not associated with changes in V˙O2max neither for all subjects (r = 0.01, P = 0.96) nor for each experimental group (r = −0.31, 0.16, −0.28, and 0.30, respectively, for Lo-Lo, Hi-Hi3, Hi-Hi, and Hi-HiLo; P = 0.3–0.6).
To the best of our knowledge, this is the first investigation to show performance improvements after a terrestrial AT intervention using a controlled design in swimmers and one of the few in truly elite athletes (12,37). The major findings of the study were that 1) there were no changes, or in some cases a worsening of performance, immediately after 3–4 wk of any training strategy in this group of elite swimmers, 2) swimming performance in stroke-specific 100 or 200 m improved significantly by approximately 3.1%–3.7% after 1–4 wk of recovery after completion of a coach-prescribed training camp conducted at SL or at moderate altitude (2320 m), 3) when two weekly sessions of high-intensity training at lower altitude were included (Hi-HiLo strategy), a greater improvement in performance occurred 2 and 4 wk after the training camp (5.3% and 6.3%, respectively), 4) similarly, the Hi-HiLo intervention also elicited further improvement in both 400- and 50-m stroke-nonspecific swimming performance, evidenced 2 wk (4.2% and 5.2%, respectively) and 4 wk (4.7% and 5.5%, respectively) after return to SL, and 5) this substantial delayed performance enhancement was not linked linearly to changes in V˙O2max, oxygen kinetics, or tHbmass and hence could not be attributed exclusively to enhanced oxygen transport capacity.
Effects on performance
The present results clearly show the potential benefit of conducting a well-implemented training camp whether at altitude or at SL after at least 1 wk of recovery from the training camp exposure. The potential improvement seemed to range from primarily anaerobic (50 m) to predominantly aerobic events (400 m), with no sex differences in the response. Previous analysis of competitive performance in Olympic swimmers estimated that the typical variation in an athlete’s maximum performance in competition was approximately 0.8%–1% (42) and approximately 2% improvement during the final 3 wk of preparation before the Sydney 2000 Olympic Games has also been reported (26). In a previous study of performance in training, typical variations in performance times were approximately 2%–4% in international Australian swimmers over a 4-month period (28). Taken together, these reports are in line with our finding of approximately 3.5% improvement in performance over a 7- to 8-wk period during the general preparation phase of the winter macrocycle in all groups, with an additional approximately 2.7% attributable to the Hi-HiLo group intervention.
Although the vast majority of studies in the AT literature are uncontrolled and underpowered—especially with elite athletes (3,23,37)—there seems to be growing consensus that when athletes are exposed to high enough altitude for a long enough amount of time and are able to preserve fitness by training hard under normoxic conditions, the majority may improve endurance performance (47). In a recent metaanalytic review, Bonetti and Hopkins (3) concluded that performance changes in studies using the conventional Hi-Hi approach were unclear, whereas changes using the terrestrial Hi-Lo strategy were considered “likely” both for elite and subelite athletes (approximately 4%) or a more realistic 1.5% when performance was predicted from uncontrolled studies. These estimations are in line with a recent review by Saunders et al. (37) in which, by using a regression analysis of average performance changes, it was estimated that a 3-wk terrestrial AT camp would elicit mean performance improvements of approximately 1.8% (Hi-Hi) and approximately 2.5% (Hi-Lo) (37).
The evidence in swimming is much less compelling. Six uncontrolled studies have tested the Hi-Hi strategy in swimmers. Three of them were entirely negative at either short or long distances from 100 m to 2000 yd (10,35). Two others showed modest and statistically unclear improvements in performance of approximately 1.6%–1.8% (11,25). In the only study with an SL control group, 10 male and female Korean elite swimmers lived and trained for 21 d at 1890 m (8). No statistical analysis was reported, but the small increase in performance in 100- and 200-m races (0.1%–0.7%) was below the smallest worthwhile enhancement effect of the intervention.
Why should swimming be different from land-based endurance sports regarding AT effects? First, swimming performance is more dependent on economy, i.e., the energy cost of swimming, than on maximal metabolic power (9); it follows that the benefit of enhanced V˙O2max could be outweighed by impaired technique and economy (24). Second, the benefit of AT might be more or less potent for swimmers of different events; for example, the energy percent share (phosphagenic–glycolytic–oxidative) at maximal competitive speed (front crawl) ranges from approximately 38%–58%–4% in 50 m to approximately 6%–21%–73% in 400 m (33), and more than three quarters of all competitive events are completed in less than 2.5 min by athletes of national standard or better. Third, in contrast to most endurance sports, a very large proportion of training is performed in intervals, with a growing emphasis on high-intensity and strength training (7).
Time course of performance changes
A distinguishing feature of the present study was that the follow-up period after the intervention covered a wide time span, up to 4 wk after intervention. This approach may provide useful evidence about the proper timing of return from AT for optimal SL performance, which remains largely unknown. If we focus on the stroke- or distance-specific assessment (TT100/200), it becomes clear that best performances were attained 4 wk after returning to SL, although the superior benefits of the Hi-HiLo intervention became evident already 2 wk after the training camp. A similar response was observed in TT400, but TT50 significantly improved on the first week after both 4-wk altitude camps. Therefore, on the basis of the present results, 50-m sprinters would likely perform at their best level 1 wk after completing an effective AT camp, whereas specialists in longer distances would attain their best performance 2–4 wk after the camp. The present findings are partially in line with the results reported by Levine and Stray-Gundersen (22) with college runners, who observed the effect on performance—in the Hi-Lo group only—immediately after and for 3 wk on return to SL.
Observational studies in swimmers have been less compelling. In an uncontrolled study, Friedmann et al. (11) found an unclear improvement in swimming performance (1.8%) 10 d after returning to SL. Gough et al. (20) analyzed the time course of performance on the basis of official competitive records after two types of 3-wk hypoxic exposure; swimming performances were substantially slower 1 d (approximately 1.5%) and 1 wk (0.9%–1.9%) after altitude/hypoxic training compared with controls and not different from prealtitude 2 and 3 wk later. Wachsmuth et al. (44) related tHbmass changes to competitive performance records over the course of 2 yr in various groups of elite swimmers undertaking classical AT for 3 or 4 wk; they found a nonsignificant drop in competition performance by approximately 1%–2% on days 1 and 7 after AT camps, and there was an unclear improvement of performance (0.8%) between 25 and 35 d after return in a small group of four athletes. It is worth emphasizing that in the present study, both Hi-Hi groups’ performance actually deteriorated immediately after the training camp (i.e., Hi-Hi3 decreased performance in all three tests and Hi-Hi got worse in TT50 and TT100/200) or 1 wk later (i.e., Hi-Hi was slower in TT50 and TT400). The mechanism(s) underlying the delayed effect of the different interventions is uncertain, although empirical evidence suggests that there is a necessary period for physiological and psychological recovery from the accumulated training stress and fatigue during the camp (29). In any case, the potential short-term detrimental effect of an AT camp should be acknowledged.
The effect of training
A key factor in the individual response to training, whether at altitude or at SL, is the training load. There is a methodological issue that needs to be addressed. For this study, we used a modification of the training impulse method (TRIMPc) deriving from cumulative exercise and recovery intervals (see Methods). However, to compare training loads at SL and altitude can be tricky. For example, when exercising at altitude at a given V˙O2 the submaximal HR response is increased (45), and, as a consequence of the reduced aerobic power (approximately15% at 2300 m) (45), endurance training with an identical absolute workload always translates to a more intense (greater relative workload) training at the higher altitude. Both components likely contribute importantly to the cardiovascular and metabolic response to exercise and thereby the training response. As one example in swimming, a group of competitive swimmers swam a 400-m time trial at 690 m above SL, and within 3–4 h on arrival to 2320 m at 92.5% of maximal speed at the previous test; peak HR (approximately 5%) and blood lactate (approximately 6%) were higher at altitude (24). Therefore, the weighting factors derived empirically by Bannister et al. (2) may not be as accurate or comparable when training at altitude. Interestingly, TRIMP data reported by Levine and Stray-Gundersen (22) calculated using the Banister method, although base training was achieved at slower speed and at a lower percentage of SL V˙O2max at moderate altitude, training HR was essentially the same; moreover, during interval training at moderate altitude, the athletes accomplished significantly lower training workloads and oxygen flux compared with that during SL (19% lower V˙O2) whereas training HR was only 5% lower. Taking all these factors into consideration, we calculated that for a given training intensity around the average training HR throughout the entire AT camp in our swimmers (approximately 125 bpm), an HR increase of 8% combined with reduction of 2.1% in maximal HR would increase TRIMPc values by 23% compared with that in normoxic conditions. Whether this difference translates into a true difference in training stimulus is unknown.
A few other points about training deserve attention. First, there were no differences in calculated training load between the two Hi-Hi groups compared with the Lo-Lo controls, empirically supporting the validity of the proposed correction for exercise in hypoxia. Second, the higher training workloads achieved during the two weekly low-level workouts in the Hi-HiLo group—i.e., the primary factor in comparing both 4-wk altitude interventions—are likely to account for at least part of the differences between the Hi-Hi and Hi-HiLo groups. Third, in contrast, the group perceiving the greatest training effort (s-RPE) and fatigue (TSF-10) throughout the training camp was the Hi-Hi3 group and not the Hi-HiLo (see Figure, Supplemental Digital Content 1, Comparing the three training load indicators throughout the camp, https://links.lww.com/MSS/A506); this suggests that training intensity may have been greater in Hi-Hi3, causing a certain degree of overreaching and a worse response in performance capacity immediately after the intervention, particularly in TT50. However, TT400 and TT100/200 performance was remarkably similar to the Hi-Hi group after 1 wk, suggesting that the effect of altitude acclimatization—i.e., the primary factor in comparing 3- and 4-wk altitude interventions—was commensurate. Fourth, regardless of these differences, when performance data were adjusted for TRIMPc as a covariate, there remained significant differences between groups at all time points, suggesting that the differences between groups are not likely to be attributed solely to training. Ultimately, the fact that the Hi-HiLo group achieved a greater training internal load—but not training effort—may be a core element of the Hi-HiLo training paradigm and hence a main factor likely contributing to the performance enhancement accomplished by this particular group.
There is now little doubt that altitude exposure, even in elite athletes of various sport types (14), including elite swimmers (20), causes an erythropoietic effect given adequate exposure in most athletes (19,22,36). Consistent with previous reports, we found that tHbmass clearly increased in swimmers living and training at altitude for 3 (3.8%) or 4 wk (6.2%) (Fig. 3). The magnitude, time course, and large variability of the erythropoietic response were in line with a recently published metaanalysis including data of 16 AT studies (19). For our study, therefore, it is unlikely that the magnitude and duration of altitude exposure were insufficient to induce a functional acclimatization response, at least for most athletes.
However, in contrast to the Hi-Hi groups, mean tHbmass did not change in our Hi-HiLo swimmers (at W4, 1.3; ±1.8%) who were also exposed to the same degree of sustained hypobaric hypoxia for 4 wk. The simplest explanation for these contrasting results may be the individual variability in tHbmass changes because half of the subjects actually showed increase of tHbmass over the TE of the measurement (see Figure, Supplemental Digital Content 4, Showing individual and mean changes, https://links.lww.com/MSS/A509). Comparable results were reported by Gore et al. (18) in a small group of eight elite track cyclists living and training at 2690 m for 31 d. We must consider also that wide variability in the erythropoietic response to moderate hypoxia has been consistently shown (6,11,19) also in swimmers (3.0% in males and 2.7% in females) followed over 2-yr period (44). Other mechanisms such as superimposed inflammation or perhaps medications like nonsteroidal anti-inflammatory drugs, which may suppress the erythropoietic response in some athletes, should also be considered (43).
Relation between changes in tHbmass and V˙O2max
In the present study, changes in tHbmass were not associated with changes in V˙O2max, neither in the whole group of swimmers nor in any of the individual experimental groups. Under controlled conditions involving athletes training at SL, cross-sectional and longitudinal data show that V˙O2max of elite athletes is closely related to tHbmass and a random variation of 1 g is associated with a change in V˙O2max by approximately 4 mL·min−1 (36,40). However, after real or simulated altitude, the relation between the gain in tHbmass and V˙O2max has been less robust (36,40). For example, in runners performing the original Hi-Lo protocol, the correlation for the change in V˙O2max and red cell volume yielded r2 = 0.14 (22). These results are in line with those reported in a recent review (36) of 10 recent studies involving four different sports, which estimated a mean approximately 3% increase in tHbmass and V˙O2max and a similarly significant, albeit weak, correlation between both parameters (r2 = 0.15). It should be emphasized that this relation between changes in tHbmass and V˙O2max after altitude exposure is the same order of magnitude as that observed after recombinant human erythropoietin administration (e.g., r2 = 0.28) (36,40), highlighting that V˙O2max is a complex parameter that is not exclusively determined by the red cell mass. It is, therefore, not surprising that the present study could not identify a significant linear relation between tHbmass and V˙O2max after any of the training interventions.
Role of systemic oxygen delivery in performance
No significant relations were found between changes in swimming performance and changes in V˙O2max in the entire group of swimmers or in any of the experimental groups (see Figure, Supplemental Digital Content 3, Showing the regression plots, https://links.lww.com/MSS/A508). This observation highlights the apparent independence of relatively short swimming performances on V˙O2max (and tHbmass), although this concept is not limited to swimming. For example, in an elegant mechanistic study by Garvican et al. (15), blood was actually removed to eliminate the effect of the erythropoietic response in female cyclists exposed to simulated (normobaric) hypoxia (Hi-Lo, 26 d, 16 h·d−1, 3000 m) while undergoing training specifically designed to improve short-duration power. Both the “clamped” (phlebotomy to restrain increased tHbmass) and unclamped groups improved 4-min supramaximal performance (power output above that required to elicit V˙O2max) to a similar degree (approximately 4%), despite the fact that by design, tHbmass (and consequently peak V˙O2) increased only in the latter. Like our study, this outcome emphasized the success of the training camp on short-term performance. However, performance on a following ride to exhaustion at peak power output was substantially worse in the “clamped” group (38%), which reinforced the role of tHbmass and V˙O2max for repeated high-intensity, endurance-type efforts. The present study adds to the body of knowledge in this field, highlighting the complex interaction among altitude acclimatization effects (such as tHbmass among others), altitude and SL training effects, V˙O2max, and performance in events of different sports and different durations/intensities requiring widely divergent metabolic demands (11,18,44).
In our study, despite failing to demonstrate an increase in tHbmass or V˙O2max, the swimmers in the Hi-HiLo group improved performance more than the altitude controls 2 and 4 wk after the intervention. Several possible explanations should be considered. First, swimming, especially in the shorter distances, may not be as dependent on systemic O2 delivery as endurance running or cycling. Second, there are other factors (e.g., differences in training intensity) that may have played greater role in improving swimming performances through as yet undetermined mechanisms, e.g., improved muscle efficiency probably at a mitochondrial level, greater muscle buffering, and the ability to tolerate lactic acid production (17), or improved O2 flux to the exercising muscles (41). This would be particularly true for the differences between Hi-Hi and Hi-HiLo, where the altitude exposure was matched, and the only difference was the twice weekly low-altitude training sessions. For the comparison between Lo-Lo and Hi-HiLo, data interpretation is more challenging, and it seems that some other aspects of altitude exposure independent of erythropoiesis must be playing a role. Finally, because the most relevant performance differences occurred after some period of training at SL, we cannot completely exclude the possibility that uncontrolled factors related to the post-camp training experience could have influenced the swimming performances.
Collectively, our findings strongly support the involvement of factors not related to systemic O2 delivery in the substantial improvement of performance observed in our Hi-HiLo swimmers, especially compared with the low-altitude controls. Therefore, our view is that an “integrative model” of adaptation to hypoxia—not opposed but complementary to the prevailing “erythropoietic/central” or to alternative “nonhematological/peripheral” models—should be considered. This paradigm would recognize 1) the multifactorial nature of physiological adaptations to exposure to midterm moderate hypoxia and to intensive training, 2) the multilevel control mechanisms of these adaptations (i.e., acting at the systemic, local, and cellular levels in a synergistic or antagonistic manner), and 3) the nonlinear behavior of complex, dynamic systems.
Design and limitations of the study
From the early steps of this study, we were committed to recruit truly elite athletes. Working with such unique individuals in a real-world setting, particularly during an Olympic season, certainly added substantial complexity to our study. Therefore, we were confronted with the virtual impossibility to conduct a fully controlled experimental design without seriously compromising the ecological validity of the study (i.e., conducted in conditions closely matched to those in which in this unique type of athletic population usually live, train, and compete) or limiting its external validity (i.e., the ability to generalize the outcome to the elite swimming population). Regarding subject assignment, we acknowledge that, as subjects were not allocated randomly, selection bias may have occurred. Similar to other sports, there exists a team–coach binomial that cannot be separated from different intervention groups as would be required by strict randomization. In an effort to reduce the likelihood of assignment bias, first, we chose to allocate swimmers from at least two different squads and nations in each group and tested that groups were properly matched for performance level and key physiological characteristics (V˙O2max and tHbmass). Finally, we chose to use mixed linear modeling for the primary data analysis because of its greater flexibility to model the variance/covariance structure of repeated measures data sets. This modeling approach proved here to be able to detect and quantify changes despite relatively small subsample sizes likely owing to the following: 1) the high reliability of the time trial performance swimming tests (34), in which test–retest reliability matched measures from competition performance in top swimmers, 2) the multiple observations during the follow-up period in the four experimental groups, 3) the large magnitude of the fixed effects in performance observed, and 4) the robustness of the model against loss of statistical power due to missing values in the different assessments (e.g., HR monitoring, time trial tests, V˙O2max measures, etc.).
An important issue with all training studies is the likelihood of information bias, eliciting the occurrence of placebo, nocebo, and training camp effects (3,17,21,23). Because all camps were conducted in training centers of high international standard, a site-specific training camp effect is less likely but of course still possible. An effort was made to counterbalance the information bias by ensuring that all coaches and swimmers were fully informed before their involvement about the characteristics of the main intervention and by stressing the unpredictability of the final outcome of each intervention compared with the others. Only then were coaches offered to choose the type of intervention they and their swimmers would finally be involved in. Finally, the responses to the ad hoc questionnaires to swimmers and coaches strongly suggested that nocebo and placebo effects are unlikely to have substantially affected the results of the study.
Practical implications for training and performance
On the basis of the present results with elite athletes, 1) after at least a week of recovery, stroke distance-specific swimming performance might be expected to substantially improve as a result of a well-implemented training camp, regardless of whether it is held at altitude or not and 2) a greater benefit can be expected by “living high-training high and low” (Hi-HiLo) for 4 wk after 2–4 wk of reacclimatization to SL; this benefit can also expected to be superior in 50- and 400-m performance. However, care should be taken not to generalize these improvements to all swimmers because substantial individual variability was noted in this as well as in other studies including swimmers performing AT (6). Another important practical implication of our results is that performance is likely to be unchanged or worsened immediately after a 3- or 4-wk intensive training camp regardless of the training environment and that benefits can only be expected to occur after 1–4 wk after completing the intervention. This delayed response could eventually provide a time window for tapering before competition, as has been suggested (29). However, after a training camp at altitude, this outcome should be interpreted with caution because evidence suggests that the combination of the deacclimatization response of hematological, ventilatory, and biomechanical factors with return to SL likely interact to determine the best timing for competitive performance (5). Moreover, the potential short-term detrimental effect of an AT camp should not be neglected. Monitoring individual training load and adaptation (e.g., resting, exercise, and recovery HR response, HR variability, exercise perception, and state of fatigue) during and after the altitude camp to avoid excessive overload or detraining as well as assessing individual peaking performance profile are strongly recommended before directly applying these rules to individual cases.
SL swimming performance of elite swimmers in 100- (sprinters) or 200-m (nonsprinters) time trials was not altered, or in some cases impaired immediately, but improved significantly by approximately 3.1%–3.7% after 1–4 wk of recovery after completion of a coach-prescribed training camp whether it was conducted at SL or at moderate altitude (2320 m). By including two weekly sessions of high-intensity training at lower altitude (Hi-HiLo strategy), greater improvement in performance occurred 2 and 4 wk after the training camp (5.3% and 6.3%, respectively). Similarly, further improvement in 400- and 50-m freestyle time trial performance was noted 2 (4.2% and 5.2%, respectively) and 4 wk (4.7% and 5.5%, respectively) after return to SL after the Hi-HiLo intervention. This delayed performance enhancement was not linked to changes in V˙O2max, O2 kinetics, or tHbmass and hence could not be attributed exclusively to enhanced systemic oxygen delivery capacity. We conclude that 1) a well-implemented 3- or 4-wk training camp, whether conducted at SL or at moderate altitude, enhances performance in most elite swimmers after a 1- to 4-wk delay period, with substantial individual variability in the response and 2) “living high-training high and low” for 4 wk has the potential to improve swimming performance above and beyond altitude and SL controls through complex mechanisms involving altitude acclimatization and training effects.
This work was partially supported by grants awarded by the Ministry of Science and Innovation of Spain (DEP2009-09181), Higher Sports Council of Spain (CSD 35/UPB/10, 005/UPB10/11, 112/UPB10/12, CAR-UGr 2009, and CAR-UGr 2011), Dutch Olympic Committee (NOC*NSF WOT/44090101), and INEFC (research support grants 2011 and 2012).
We gratefully acknowledge the valuable contribution of Christopher Gore, James Stray-Gundersen, Martin Truijens, and Uwe Hoffmann in the first design of the project and, together with three anonymous referees, for their useful comments on earlier versions of the article. We also acknowledge the technical and scientific support given by Marlen Klein, Christian Völzke, and Torben Hoffmeister (Bayreuth University), Thorsten Schuller (German Sports University Cologne), Esa Hynynen (KIHU—Research Institute for Olympic Sports of Finland), Marek Anestik (Scottish Institute of Sport), Boro Štrumbelj (University of Ljubljana), Jun Qiu (Shanghai Institute of Sports Science), Jordi J. Mercadé and Amador García (University of Granada), Vicent Nebot, Matías Pérez-Sánchez (Virgen de las Nieves University Hospital), Josefa León-López (San Cecilio University Hospital), and Javier Argüelles (CAR Sierra Nevada). We also acknowledge the contribution of Anna Barrero (INEFC) in data acquisition, Iñaki Pérez-Medina† in data analysis, Roser Aliberas and Marta Rodríguez-Aliberas in blood sampling, and Rafael Tarragó and Ricard Fàbrega in logistics and administration. A special note of gratitude goes to president Joaquim Torres, Eloi Gómez and the staff of C. N. Sabadell, to president Fernando Carpena and technical director Luis Villanueva (Royal Spanish Swimming Federation, RFEN), to deputy director Alfonso Sánchez Bernard (CAR Sierra Nevada), and to director Andreu Camps (INEFC) for their valuable support.
We are profoundly indebted to the coaches and staff of the participating teams: Fred Vergnoux (C. N. Sabadell and RFEN); Jordi Murio, Juan J. Castillo, and Víctor Mancha (RFEN); David Lyles, Jenny Lyles, and Xu Feng Jie (Shanghai Province Swimming and Chinese Swimming Federation); Miha Potočnik, Gorazd Podržavnik, and Roni Pikec (Slovenian Swimming Federation); Rohan Taylor, Jeremy Oliver, and Danielle Stefano (Victorian Institute of Sport); and Patrick Pearson (EIFFEL Swimmers PSV-Eindhoven). A very special note of appreciation goes to each and all the swimmers who served as subjects and rendered their valuable time and effort.
A nonpublished Spanish version of this study has been awarded the “Tercer Premio Nacional de Investigación en Medicina del Deporte 2013, Escuela de Medicina del Deporte, Universidad de Oviedo.”
The authors declare no conflicts of interest, financial or otherwise.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.