Successful endurance training involves the manipulation of training intensity, duration, and frequency, with the implicit goal of maximizing performance and related physiological characteristics (10–13). Endurance athletes normally use a high-volume low-intensity training approach and incorporate moderate volumes of high-intensity training (5,6,25). This training model has been used by successful cross-country skiers throughout the decades (7,9,24,26), with their performance capability strongly related to maximal oxygen uptake (V[Combining Dot Above]O2max) and to oxygen uptake at the anaerobic threshold (15,17–19,24).
Several studies have shown that increased volume of aerobic high-intensity endurance training is effective for improving aerobic characteristics and endurance performance in highly trained endurance athletes (9,19,24). It has therefore been suggested that increased exercise intensity is the crucial factor for further improving performance in endurance trained athletes (19). However, the manipulation of both intensity and duration within the high-intensity sessions may play an important role for the endurance adaptations (7,25). By comparing the effects of 7 weeks of interval training, Seiler et al. (27) found that accumulating 32 minutes of work at 90% of maximal heart rate (HRmax) induced greater adaptive gains than accumulating 16 minutes at 94% in recreational cyclists. Whether similar findings can be found in highly trained athletes is yet to be examined.
In the study by Seiler et al. (27), interval sessions were matched for the overall effort, and the rate of perceived exertion (RPE) of interval training programs was more strongly related to the training intensity than the accumulated duration. Matching training for overall effort is in accordance with how athletes actually match their daily training (25,27), but in contrast to controlled intervention training studies on elite athletes that match training sessions for total work or energy expended (10,19). The latter type of matching can be argued to artificially constrain the training so that the overall effort associated with different interval workouts is not equivalent (27). When the exercise intensity exceeds the anaerobic threshold, small changes in the intensity are associated with large increases in fatigue and decreases in the tolerated exercise duration (25,27).
The purpose of this study was to test whether a long duration of aerobic high-intensity interval training is more effective than shorter intervals at a higher intensity in highly trained endurance athletes. Therefore, endurance performance and aerobic characteristics were tested before and after an 8-week intervention period where 2 weekly sessions of short- or long-duration intervals matched for effort were added to junior cross-country skiers' normal training program. We hypothesized that a long duration of aerobic high-intensity interval training improves endurance performance and aerobic characteristics more than shorter intervals at a higher intensity.
Experimental Approach to the Problem
To test the hypothesis that a long duration of aerobic high-intensity interval training improves endurance performance and aerobic characteristics more than shorter intervals at a higher intensity, a pre- to postintervention training study was executed. Initially, 21 junior national-level cross-country skiers followed the same high-volume low-intensity training program during an 8-week baseline training period. Thereafter, the skiers were randomly allocated into a short-interval group (SIG, n = 7), a long-interval group (LIG, n = 7), and a control group (CG, n = 7) stratified by gender, pretest scores, and baseline training volume. During an 8-week intervention, the SIG added 2 weekly sessions with short intervals (2–4 minutes) (15–20 minutes total work duration per session), whereas the LIG added 2 weekly sessions with long intervals (5–10 minutes) (40–45 minutes total work duration per session). The subjects were instructed to perform the interval sessions with their maximal sustainable intensity. The CG added 2 weekly sessions with low-intensity endurance training at 60–74% HRmax (approximately 100 minutes work duration per session). Training was performed on dry land during the summer preparation period with the distribution of exercise modes kept constant within and between the baseline and the intervention training periods in all groups. The participants were tested for 12-km roller-ski skating and a 7-km hill run, and for V[Combining Dot Above]O2max and oxygen uptake at the ventilatory threshold (V[Combining Dot Above]O2VT) during treadmill running before and subsequent to the training intervention.
Twelve male and 9 female, highly trained, national-level junior cross-country skiers volunteered to participate in the present study. All of the skiers were students at a Norwegian high school with a specialized program for cross-country skiing and had trained for cross-country skiing daily for more than 3 years. The skiers' training status, physiological characteristics, and anthropometrics are shown in Table 1. The study was preapproved by the Regional Ethics Committee, Trondheim, Norway. All subjects were fully acquainted with the nature of the study and informed of the experimental risks before signing written informed consent to participate. There were 12 participants who were younger than 18 years, and as a result, each of their parents was asked to provide parental consent for their child's participation. It was stated explicitly that subjects' could withdraw from the study at any point without given reason for doing so.
Both the baseline and the intervention training periods were performed as dry land training from June to September. Training plans, a training diary, and written instructions about how to record training were provided and explained to the subjects. Heart rate was registered for every training session using the Polar RS800 heart rate monitor (Polar Electro OY, Kempele, Finland) with 5-s registration intervals. The Borg RPE scale from 6 to 20 was used for monitoring the athletes' RPE of the interval sessions. The endurance training intensity was categorized into 4 zones:
- Low-intensity endurance training (INT1), performed as 1.5–3 hours continuous work, with an intensity of 60–74% of HRmax.
- Moderate-intensity endurance training (INT2), performed as 45 minutes to 1.5 hours continuous work, with an intensity of 75–87% of HRmax.
- High-intensity endurance training (INT3), performed as long duration intervals (5–10 minutes) with total work period durations of 40–45 minutes.
- Very high–intensity endurance training (INT4), performed as short intervals (2–4 minutes) with total work period durations of 15–20 minutes.
INT1 and INT2 were performed according to the heart rate zones presented, whereas the subjects were instructed to perform each interval session at INT3 and INT4 with their maximal sustainable intensity (isoeffort). Average HR for the last 25% of each interval and RPE for the entire sessions were recorded. All endurance training was quantified according to the session goal method described by Seiler and Kjerland (26). Strength and maximal speed training (<30 s) were recorded. Each training group was followed up daily by the investigators during the intervention training period.
The baseline training period included ∼60% INT1, ∼10% INT2, ∼10% INT3, and ∼10% INT4 (Table 2), which is a normal training distribution among Norwegian cross-country skiers. Approximately 10% of the total training was strength and maximal speed training. All the skiers carried out 80–85% of their endurance training as roller skiing, running, and running with poles, while the remainder of the training was performed as strength training or cycling. The dominant modes of endurance training were roller-ski skating (∼25%) and running (∼25%). There were no significant differences in the performed training with respect to volume, intensity, or exercise mode between the 3 groups, during the baseline training period. During the intervention training period, the LIG added 2 extra sessions per week of INT3, the SIG added 2 extra sessions per week of INT4 (both p < 0.01), whereas the CG added 2 extra sessions of INT1 (p < 0.05). No differences in the distribution of training modes from baseline to intervention training for any of the groups were revealed. There were no differences in the volume of strength and maximal speed training between groups in the 2 training periods, but all groups increased strength and maximal speed training to around 15% of total training in the intervention period (all p < 0.01).
Each skier was tested for 12-km time-trial performance on roller skis outdoors in the skating technique and a 7-km time-trial hill run. The tests had a 30-seconds start interval between the skiers in a random order. Before each of the performance test, a 30-minutes standardized warm-up procedure was performed. The 12-km roller-ski competition was performed on an asphalt trail standardized according to the International Ski Federation homologation manual (16). The subjects used identical skating roller skis (Swenor Roller-skis, Troesken, Norway), and each skier used the same pair of roller skis at pre- and posttesting. These roller skis were not used between pre- and posttesting to avoid changes in rolling resistance. The skiers used their own skating poles (pole length = 89 ± 2% of body height) with pole tips for roller skiing on asphalt. The 7-km uphill running test was performed on asphalt and had a mean inclination of 4.5%. Both competition trails were well known to all of the skiers and the skier's racing times were recorded by 2 synchronized stopwatches (Regnly RT3, Emit AS, Oslo, Norway).
There were stable weather conditions with no wind and dry asphalt both at the pre- and posttest situations. However, the temperature was 9° C colder at posttesting (18° vs. 9° C). To check whether these changes in temperature could have influenced the rolling resistance of the roller skis and consequently, the 12-km roller-ski skating performance, a rolling resistance test was performed before the 12-km skating roller-ski test at both test situations using the same pair of roller skis. The rolling resistance test was performed 8 times by one skier, 75 kg, rolling straight forward in a downhill tuck body position with 90° angles in elbow and knee joints at a constant decline of 10%. After a 30-minute run-in time was captured by 3 lasers (Speedtrap II Timing System, Brower Timing Systems, Draper, UT, USA), spaced at 3-minute intervals to guarantee a constant speed similar to the mean speed in the 12-km roller-ski test. From pre- to posttesting, mean speed during the rolling resistance test decreased by 1.5 ± 0.6%, from 5.8 ± 0.0 ms−1 to 5.7 ± 0.1 ms−1 (p = 0.03). Pilot tests before the study revealed that using one skier for testing rolling resistance gave representative values for the group changes in friction. However, we did not adjust performance for this in the present study, and posttest performances in the 12-km roller-ski test were slightly underestimated when compared with the pretest situation.
After a 15-minute warm up at 60% of HRmax, V[Combining Dot Above]O2max and V[Combining Dot Above]O2VT were measured when running on a motorized treadmill, according to a traditional method of monitoring cross-country skiers in Norway (15). The test duration was 6–8 minutes, performed at a constant inclination of 10.5% with individual starting speeds and a stepwise increase by 1 km·h−1 every minute. Gas exchange was measured using the Metamax II portable analyzer (Cortex Biophysik GmbH, Leipzig, Germany). The Metamax II analyzer was calibrated according to standardized procedures and is reported to be precise in measuring O2-uptake within subjects (22). The test was considered to be a maximal effort if the following 3 criteria were met: (a) a plateau in V[Combining Dot Above]O2 with increasing exercise intensity, (b) respiratory exchange ratio more than 1.15, and (c) blood lactate concentration exceeding 8 mmol·L−1 (3). Oxygen uptake was measured continuously, and the average of the 3 highest 10-second consecutive measurements determined V[Combining Dot Above]O2max. Ventilatory threshold was defined as the intensity at which the ventilatory equivalent of oxygen (VE/V[Combining Dot Above]O2) began to rise without a concurrent rise in the ventilatory equivalent of carbon dioxide (VE/VCO2) (8,23). This method was performed according to earlier recommendations and is shown to be valid and reliable (1,2). To determine blood lactate concentration, blood samples were taken 1 and 3 minutes after finishing the test and the highest value represented the peak value. A 5-μl blood sample was taken from a fingertip and analyzed for blood lactate concentration by a Lactate Pro LT-1710t (ArkRay Inc., Kyoto, Japan) according to the manufacturer's instructions. This instrument has been found reliable for use in athletic testing (20,21). The highest HR value during the test was defined as HRmax.
The different tests were carried out with 24 hours in-between, and all tests were performed within one week. A 2-day standardized tapering period before the testing sessions were performed. Within subject variation was minimized by testing at the same time of day, and following a 1-hour fast before the tests. Participants abstained from exercise, alcohol, and caffeine in the 24 hours before tests. Subjects who had any health-related problems during the study period terminated their participation and were not included in the material.
All data were checked for normality and are presented as mean and standard deviation. 95% Confidence interval are presented for the changes in performance and physiological tests from pre- to posttest. Possible effects of gender and pretest scores were checked for before comparing V[Combining Dot Above]O2VT, V[Combining Dot Above]O2max, and performance between training groups using a one-way ANOVA with the gain score, that is, posttest-pretest, as the dependent variable. The measure of association (ω2) was calculated to estimate the percentage of the total variance that can be explained by the influence of the training. Pairwise comparisons following a significant main effect in the ANOVA involved the least squares distance post hoc procedure to evaluate significant differences in the adjusted means (M) between the groups. The effect size (ES) was calculated to measure the magnitude of a significant pairwise difference. Pre- to posttest changes within groups were tested by the paired samples t test procedure. Statistical significance was set at an alpha level of <0.05. Sample size is calculated on the basis of V[Combining Dot Above]O2max by sample size estimation for longitudinal studies of a 2-sided paired samples t test. With a level of significance at p < 0.05 and a power of 0.8, V[Combining Dot Above]O2max differences of 4 ml·kg−1·min−1 between groups can be revealed with 7 persons per group. To control for test-retest reliability, intraclass correlation coefficients were calculated for the dependent variables before the study started. Repeated measurements of V[Combining Dot Above]O2max by use of the Metamax II in our laboratory showed an intraclass correlation coefficient of 0.98. Repeated measurements of the roller-ski test showed an intraclass correlation coefficient of 0.95. All statistical tests were processed using SPSS 15.0 Software for Windows (SPSS Inc., Chicago, IL).
All subjects pooled (n = 21) improved 12-km roller-ski performance by 2.9 ± 1.9%, 7-km hill run performance by 3.5 ± 2.6%, V[Combining Dot Above]O2max by 3.0 ± 1.7%, and V[Combining Dot Above]O2VT by 4.1 ± 3.0% from pre- to posttest (all p < 0.05). There were no gender effects in the group comparisons; therefore all further analyzes included both genders. Furthermore, there were no significant effects of pretest values on gain scores from pre- to posttesting. The average HR values over the last 25% of each work period during the interval sessions showed that INT3 was performed at 91 ± 1% of HRmax, whereas INT4 was performed at 95 ± 2% of HRmax. The average RPE for the entire sessions did not differ significantly between INT3 and INT4 (18.1 ± 1.2 vs. 18.5 ± 1.0).
The LIG improved performance in both 12-km roller skiing and 7-km hill run from pre- to posttesting by 6.8 ± 4.0% and 4.8 ± 2.6%, respectively (Figures 1A, B). There was a significant effect of type of training on changes in the 12-km roller-ski skating test (F2,18 = 4.25, p = 0.03, ω2 = 0.27) and the 7-km hill run (F2,18 = 4.89, p = 0.02, ω2 = 0.30; Figure 2). Follow-up tests showed that LIG (M = −139.6 seconds) improved roller-ski performance significantly more than the CG (M = −16.7 seconds, ES = 1.35) and the SIG (M = −17.1 seconds, ES = 1.35) (Figure 2). Follow-up testing also revealed that LIG (M = −94.0 seconds) improved in the hill run test as compared with the CG (M = −24.1 seconds, ES = 1.54) and the SIG (M = −33.9 seconds, ES = 1.33) (Figure 2).
The LIG and SIG improved V[Combining Dot Above]O2max by 3.7 ± 1.6% and 3.5 ± 3.2% from pre- to posttesting (both p < 0.01), whereas the performance in the CG did not change significantly (Figure 3A). There was a significant effect of type of training on changes in V[Combining Dot Above]O2max (F2,18 = 4.92, p = 0.02, ω2 = 0.31). Compared with the CG (M = −0.27 ml·kg−1·min−1), follow-up tests showed that both the LIG (M = 2.59 ml·kg−1·min−1, ES = 1.56) and the SIG (M = 2.16 ml·kg−1·min−1, ES = 1.32) improved their V[Combining Dot Above]O2max (Figure 2).
The LIG improved V[Combining Dot Above]O2VT by 5.8 ± 3.3% from pre- to posttesting (p = 0.004), whereas V[Combining Dot Above]O2VT in the SIG and CG did not change significantly (Figure 3B). There was a significant effect of type of training on changes in V[Combining Dot Above]O2VT (F2,18 = 9.44, p = 0.002, ω2 = 0.47; Figure 3B). Follow-up tests showed that the LIG (M = 4.1) significantly increased V[Combining Dot Above]O2VT as compared with the SIG (M = −0.48, ES = 2.31) and CG (M = 1.46, ES = 1.34) (Figure 2).
The purpose of the present study was to compare the effects of a long duration of aerobic high-intensity interval training with shorter intervals at a higher intensity on endurance performance and aerobic characteristics in highly trained junior cross-country skiers. The present study matched the interval sessions for overall effort and showed that both high-intensity intervals with a long duration and shorter intervals at a higher intensity were effective for improving V[Combining Dot Above]O2max. However, a long duration of intervals was most effective for improving endurance performance and oxygen uptake at the ventilatory threshold. These findings provide important information about the emphasis of high-intensity interval training for coaches and athletes.
Consistent with earlier findings (9,26), this study demonstrates that increasing the number of aerobic high-intensity sessions is more effective than increasing the number of low-intensity sessions for improving V[Combining Dot Above]O2max in highly trained cross-country skiers. There were no differences in the improvement of V[Combining Dot Above]O2max between the 2 interval training groups in this study, indicating that both high-intensity intervals with a long duration and shorter intervals at a higher intensity give an effective stimulus for enhancing V[Combining Dot Above]O2max. Overall, our findings corroborate earlier studies suggesting that aerobic high-intensity training seems to be a key factor in improving V[Combining Dot Above]O2max in well-trained populations (4,9,19,24–27). However, this is to our knowledge the first study to compare effects of duration and intensity of aerobic high-intensity intervals on V[Combining Dot Above]O2max in such highly trained endurance athletes.
The present study demonstrates that a long duration of aerobic high-intensity intervals was more effective than shorter intervals at a higher intensity for improving endurance performances and V[Combining Dot Above]O2VT. This corroborates with Seiler et al. (27) who found that accumulating 32 minutes of work at 90% of HRmax induced greater adaptive gains than accumulating 16 minutes at 94% in recreational cyclists during 7 weeks of interval training matched for effort. Intensity and duration are integrated as signaling components in the adaptive response to endurance training, and the optimal design of interval sessions is subject for continuous discussions. Practically, small changes in the intensity are associated with large increases in fatigue and decreases in the tolerated exercise duration of high-intensity interval sessions (25,27). The present study indicates that it can be advantageous for endurance performance to slightly reduce intensity and use longer duration of intervals in endurance trained athletes. However, as this topic has revealed relatively little attention in the literature, future studies should further explore this on different groups of athletes.
The superior improvements in performance of LIG and V[Combining Dot Above]O2VT compared with the SIG indicate that the longer duration intervals at ∼90% of HRmax also improve endurance performance independently of alterations in V[Combining Dot Above]O2max. Basic exercise physiology explains improved endurance performance without changes in V[Combining Dot Above]O2max by enhanced anaerobic threshold or work economy (17), which is also supported by earlier studies on elite athletes (5,7). In the present study, an improved V[Combining Dot Above]O2VT with increased training at INT3 was revealed. This is likely to be due to the specificity principle, as this training intensity is slightly above the V[Combining Dot Above]O2VT. Such effects of endurance training are assumed to be caused by peripheral improvements at the muscle level, for example, mitochondria and aerobic enzyme function, and further to be linked to a greater utilization of oxygen at performance speed in endurance athletes (3,17).
Unique for the present approach was that interval training descriptions were based on isoeffort matching, which is different to other controlled studies that compare varying training programs using isoenergetic matching (19). We regard our method to have a high ecological validity because it is based on how athletes actually train (25,27). From highly trained athletes' point of view, it is easier to perform more work at lower exercise intensities, and isoenergetic matching underestimates the work that athletes are able to perform at lower intensities (25). This is confirmed by Seiler et al. (27) where the RPE of interval training programs was more strongly related to training intensity than accumulated duration. The present study showed similar values of RPE between INT3 and INT4 training, which demonstrates that the training effects were compared at similar efforts.
A limitation in this study may be that V[Combining Dot Above]O2max and V[Combining Dot Above]O2VT were measured during treadmill running and not when roller-ski skating. Earlier findings have revealed differences in physiological responses between techniques among cross-country skiers (14). However, in this study the distributions of training modes were kept constant between training periods and groups, and the changes in performance within groups were similar for roller-ski skating and running. Thus, we assume that the potential differences in V[Combining Dot Above]O2max and V[Combining Dot Above]O2VT between treadmill running and skating remained constant during the intervention period and represent valid measures in this study.
To optimize the adaptive responses and performance from endurance training, the design of aerobic high-intensity interval sessions is an important issue for coaches and athletes in endurance sports. This topic has not yet been fully resolved by the literature and is therefore subject for continuous discussions. The applications from this study apply both for coaches and scientists by showing that V[Combining Dot Above]O2max can be improved both by adding aerobic high-intensity interval sessions with a long duration and shorter intervals at a higher intensity. The fact that the longer duration intervals additionally enhanced endurance performance and the ventilatory threshold is an important finding for practice. However, this investigation was only performed over 8 weeks, and the effects of high-intensity interval training over a longer time-scale among highly trained athletes still requires further investigation.
Particular appreciation is directed to the skiers and coaches for their participation and cooperation during this study.
1. Amann M, Subudhi A, Foster C. Influence of testing protocol on ventilatory thresholds and cycling performance. Med Sci Sports Exerc 36: 613–622, 2004.
2. Amann M, Subudhi AW, Walker J, Eisenmann P, Shults B, Foster C. An evaluation of the predictive validity and reliability of ventilatory threshold
. Med Sci Sports Exerc 36: 1716–1722, 2004.
3. Bassett DR, Howley ET. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Med Sci Sports Exerc 32: 70–84, 2000.
4. Billat LV. Interval training for performance: A scientific and empirical practice. Med Sci Sports Exerc 31: 13–31, 2001.
5. Esteve-Lanao J, Foster C, Seiler S, Lucia A. Impact of training intensity distribution on performance in endurance athletes. J Strength Cond Res 21: 943–949, 2007.
6. Fiskerstrand A, Seiler KS. Training and performance characteristics among Norwegian international rowers 1970-2001. Scand J Med Sci Sports 14: 303–310, 2004.
7. Gaskill SE. Fitness Cross-Country Skiing. Champaign, IL: Human Kinetics, 1988.
8. Gaskill SE, Ruby BC, Walker AJ, Sanchez OA, Serfass RC, Leon AS. Validity and reliability of combining three methods to determine ventilator threshold. Med Sci Sports Exerc 33: 1841–1848, 2001.
9. Gaskill SE, Serfass RC, Bacharach DW, Kelly JM. Responses to training in cross-country skiers. Med Sci Sports Exerc 31: 1211–1217, 1999.
10. Gorostiaga EM, Walter CB, Foster C, Hickson RC. Uniqueness of interval and continuous training at the same maintained exercise intensity. Eur J Appl Physiol Occupat Physiol 63: 101–117, 1991.
11. Hickson RC, Hagberg JM, Ehsani AA, Holloszy JO. Time course of the adaptive responses of aerobic power and heart rate to training. Med Sci Sports Exerc 13: 17–20, 1981.
12. Hickson RC, Kanakis C Jr, Davis JR, Moore AM, Rich S. Reduced training duration effects on aerobic power, endurance, and cardiac growth. J App Physiol 53: 225–229, 1982.
13. Hickson RC, Rosenkoetter MA. Reduced training frequencies and maintenance of increased aerobic power. Sci Sports Exerc 13: 13–16, 1981.
14. Holmberg H-C, Rosdahl H, Svedenhag J. Lung function, arterial saturation and oxygen uptake in elite cross country skiers: influence of exercise mode. Scand J Med Sci Sports 17: 437–444, 2007.
15. Ingjer F. Maximal oxygen uptake as a predictor of performance ability in woman and men elite cross-country skiers. Scand J Med Sci Sports 1: 25–30, 1991.
16. International Ski Federation. FIS Cross-Country Homologation Manual (5th ed.). 2009. Available at: http://www.fis-ski.com/data/document/homologation-manual-2009.pdf
17. Joyner MJ, Coyle EF. Endurance exercise performance: the physiology of champions. J Physiol 586: 35–44, 2008.
18. Larsson P, Olofsson P, Jakobsson E, Burlin L, Henriksson-Larsèn K. Physiological predictors of performance in cross-country skiing from treadmill tests in male and female subjects. Scand J Med Sci Sports 12: 347–353, 2002.
19. Laursen PB, Jenkins DG. The scientific basis for high-intensity interval training. Sports Med 32: 53–73, 2002.
20. McLean SR, Norris SR, Smith DJ. Comparison of the Lactate Pro and the YSI 1500 sport blood lactate analyzers. Int J Appl Sports Sci 16: 22–30, 2004.
21. Medbø JI, Mamen A, Holt Olsen O, Evertsen F. Examination of four different instruments for measuring blood lactate concentration. Sca J Clin Lab Invest 60: 367–380, 2000.
22. Medbø JI, Mamen A, Welde B, von Heimburg E, Stokke R. Examination of the Metamax I and II oxygen analyzers during exercise studies in the laboratory. Scand J Clin Lab Investig 62: 585–598, 2002.
23. Reinhard U, Müller PH, Schmülling RM. Determinations of anaerobic threshold by the ventilation equivalent in normal individuals. Respiration 38: 36–42, 1979.
24. Sandbakk Ø, Welde B, Holmberg H-C. Endurance training and sprint performance in elite junior cross-country skiers. J Strength Cond Res 25: 1299–1305, 2011.
25. Seiler KS. What is best practice for training intensity and duration distribution in endurance athletes? Int J Sports Phys Perf 5: 276–291, 2010.
26. Seiler KS, Kjerland GØ. Qantifying training intensity distribution in elite endurance athletes: Is there evidence for an “optimal” distribution? Scand J Med Sci Sports 16: 49–56, 2006.
27. Seiler S, Jøranson K, Olesen BV, Hetlelid KJ. Adaptations to aerobic interval training: interactive effects of exercise intensity and total work duration. Scand J Med Sci Sports, 2011. doi: 10.1111/j.1600-0838.2011.01351.x.