Secondary Logo

Share this article on:

Aerobic Fitness Variables Do Not Predict the Professional Career of Young Cyclists


Medicine & Science in Sports & Exercise: April 2010 - Volume 42 - Issue 4 - p 805-812
doi: 10.1249/MSS.0b013e3181ba99bc

Purpose: The aim of this study was to examine the discriminant ability of aerobic fitness measures among junior cyclists of different competitive levels and to examine whether these variables were able to predict the cyclists who reached the professional level.

Methods: A total of 309 young cyclists (mean ± SD, age = 17.5 ± 0.5 yr, height = 178 ± 6 cm, weight = 66 ± 7 kg) performed an incremental maximal test to determine peak oxygen uptake (V˙O2peak) and respiratory compensation point. To examine the discriminant and predictive ability of these parameters, the cyclists were classified according to their competitive level and specialty: 1) national team (NAT) and nonnational team (non-NAT); 2) nonprofessionals (NP), and professional flat specialists and professional climbers; and 3) nonprofessionals (NP), professional continental, and ProTour. A logistic regression was used to test the accuracy of models generated using as predictors the laboratory measures of aerobic fitness and anthropometric data.

Results: The mean absolute and relative V˙O2peak were 4.7 ± 0.6 L·min−1 and 71 ± 7 mL·kg−1·min−1, respectively. NAT displayed higher V˙O2 values than non-NAT. Professional flat specialists showed higher absolute V˙O2 values than NP. Professional climbers showed higher relative V˙O2 values than NP. ProTour showed higher aerobic fitness measures than NP. Using the receiver operating characteristic curve, body mass, absolute V˙O2peak, and V˙O2 at respiratory compensation point were found to discriminate NAT from non-NAT. Although some of these variables influenced the odds of becoming professionals (odds ratios from 1.10 to 2.86), no models were able to correctly identify the cyclists who became professionals.

Conclusions: Traditional physiological measures of aerobic fitness are useful to identify junior cyclists who can excel in their category. However, these variables cannot be used for talent identification, if "talent" is interpreted as a young cyclist who will succeed in becoming a professional.

1Mapei Sport, Castellanza, ITALY; 2Department of Research and Development, Schulthess Klinik, Zurich, SWITZERLAND; and 3Research Centre for Bioengineering and Motor Sciences, Rovereto, ITALY

Address for correspondence: Franco M. Impellizzeri, Ph.D., Department of Research & Development, Schulthess Klinik, Lengghalde 2, 8008 Zürich, Switzerland; E-mail:

Submitted for publication May 2009.

Accepted for publication July 2009.

Talent identification has been defined as the "process of recognizing current participants with the potential to excel in a particular sport" (37). During the last decade, there has been increasing attention to this topic (1,34,37,39). A recent comprehensive review has underlined the complexity of talent identification and development (37). It is commonly believed that in individual endurance sports, the talent identification is easier than in other disciplines such as team sports (30,37). Although cycling is one of the most popular recreational and competitive sports, to our knowledge, few studies have attempted to identify key predictors of future professional cyclists (36). Nevertheless, many countries and professional teams are investing considerable amounts of money and resources in the identification of talents in cycling (36). These programs should allow recruiting athletes more likely to excel in adulthood. Early identification of future elite performers would give a competitive advantage to those organizations that are able to identify talented athletes (30). Indeed, it is thought that early identification of talent and well-structured development programs would increase the likelihood that these young athletes will become successful performers (30,37). A recent study by Schumacher et al. (36) suggested that those nations where cycling is not a popular sport (such as Australia and Germany) but initiatives for talent identification and development have been adopted had a higher percentage of junior cyclists selected for the national team that reached outstanding results in adulthood compared with those nations where cycling is most popular and practiced such as Spain, Italy, and France but where these programs are not used in a systematic way.

Maximal oxygen uptake (V˙O2max) and lactate or ventilatory threshold are considered important determinants of endurance performance (12,13,25,31). For this reason, the measurement of V˙O2max and the "anaerobic threshold" is quite common in cyclists. These physiological characteristics are considered prerequisites for competing at a high level in cycling (24,27). Furthermore, significant relationships between both off-road and on-road cycling performance and V˙O2max, peak power output, and ventilatory or lactate thresholds have been reported (3,5,7,8,11,21,22,25,32,38). For these reasons, these measures of aerobic fitness can potentially be candidate physiological variables for talent identification.

Cross-sectional comparison is the most commonly used method to identify the characteristics that may be used for talent identification. However, the characteristics that differentiate high- from low-level young cyclists do not necessarily imply that these characteristics are also able to predict their future competitive level (37). Longitudinal long-term observational studies are preferable. Although there is great interest in talent identification, no studies have examined if the physiological variables measured during the traditional and widespread incremental aerobic tests are different between junior cyclists of different competitive levels and if these variables are predictive of the future level achieved in adulthood. Such studies are difficult to conduct because large sample size and long-term longitudinal observations are required. Therefore, the first aim of this study was to examine the differences in aerobic fitness variables measured during incremental tests among junior cyclists (17-18 yr old) of dissimilar competitive levels. Subsequently, we examined whether these variables (and being selected for the national team) were able to predict the cyclists who reached the professional level some years later. A further aim was to provide reference aerobic fitness data for young cyclists.

Back to Top | Article Outline


Participants and study procedures.

For this study, we used retrospective data collected between 1996 and 2002 for cross-sectional comparisons and to prospectively examine the predictive ability of aerobic fitness variables measured in our laboratory. We analyzed the tests of 309 cyclists of the junior category according to the Union Cycliste Internationale (UCI; mean ± SD, age 17.5 ± 0.5 yr, height = 178 ± 6 cm, weight = 66 ± 7 kg). Inclusion criteria were as follows: competitive cyclists born between 1978 and 1984 and tested in our laboratory between the age of 17 and 18 yr at least twice or once but within the competitive season (between March and October). Data were analyzed at the end of December 2008, when the cyclists' age ranged from 24 to 30 yr old. This time frame (at least 5 yr) was selected to give them the time to become professionals, considering that it is common to be recruited in professional teams when cyclists are in the UCI under-23 category. Indeed, the mean ± SD age at which the junior cyclists of this study became professionals was 21.4 ± 1.8 yr. The cyclists involved in the study and tested in our laboratories came from Europe, Asia, and Oceania, but most of them (approximately 90%) belonged to Italian junior teams. The cyclists were not randomly selected from the population, but they were part of teams referring to our laboratories for physiological assessments. Among the 28 junior cyclists who became professionals, none of them was born in 1978, four were born in 1979, four in 1980, eight in 1981, seven in 1982, one in 1983, and four in 1984.

For the cross-sectional comparison, junior cyclists were assigned to the national team (NAT) or nonnational team (non-NAT) groups. Seventy-two of the 309 junior cyclists were selected from their respective national teams at the time of the tests. The cyclists were classified as professionals (P) if they competed at least 3 yr in professional teams (UCI rules) or nonprofessionals (NP). Because one of the weaknesses of ex post facto studies is the lack of control in the independent variables (2), we decided to use strict criteria. Therefore, we used the "3 yr of professional cycling" cutoff criterion, considering that the first neoprofessional cycling employment contract has to last at least 2 yr. The renewal of the contract for a third year has been considered the confirmation of the real attitude for professional cycling to exclude that they could have become professionals only temporarily (e.g., through the help of a sponsor). Using this criterion, seven cyclists were excluded because they remained in the professional category only 2 yr. All excluded cyclists belonged to professional continental (PC) teams and were retired at the time of the analysis. None of these cyclists retired for injuries or doping issues.

The junior cyclists classified as professionals were subsequently grouped on the basis of two criteria. Firstly, as in previous studies (18,19,26,33), they were classified into different specialty groups on the basis of the indications of coaches and/or cyclists and their role in competition: professional climber (PCL, n = 8), professional flat specialists (PFL, n = 11), and professional sprinters (n = 9). Secondly, they were classified as merely professionals (PC, n = 10) if they belonged to low-ranked professional teams (former GS2 or PC teams) and as ProTour cyclists (PT, n = 18) if they competed for top professional cycling teams (i.e., former GS1 or PT teams). Twelve of 18 PT riders won one or more races (6 of them collected overall 241 victories), whereas 4 of 10 PC riders collected one or more victories (15 overall). The PT riders won four medals at Olympic Games (3 gold and 1 silver), six medals at World Championships (3 gold, 1 silver, and 2 bronze), 1 Giro d'Italia, 12 stages at Grand Tours (Giro d'Italia and Tour de France), and seven classic 1-d races (Giro di Lombardia, Milano-SanRemo, Paris-Roubaix, and Amstel Gold Race). None of the professional riders has been tested positive in doping control.

If more than one test was available, the test in which the cyclist achieved the highest V˙O2max was used for the analysis (18,29). Cyclists were asked not to eat within 3 h of testing, not to drink beverages containing caffeine for at least 8 h before testing, and to avoid intense exercise for 24 h before testing. All the subjects, or their parents for 17-yr-old cyclists, signed an informed consent to the use of their data collected in routine evaluations for observational studies. The retrospective study was approved by an independent institutional review board according to the guidelines and recommendations for European ethics committees by the European Forum for Good Clinical Practice.

Back to Top | Article Outline

Incremental cycling test for V˙O2max determination

After the estimation of the body fat percentage using a skinfold technique (23), the participants performed an incremental maximal test for the determination of V˙O2max and ventilatory thresholds. Tests were performed on a standard racing bicycle with the rear wheel braked by a custom-made Monark-type pendulum system. The applied mechanical power was checked using an SRM Training System incorporating a 12-strain gauge (Schoberer Rad Messtechnik, Welldorf, Germany). The measured SRM power output and the nominal power were linearly related: SRM power output = power output × 1.03 + 32.3 (r 2 = 0.99). All nominal values were corrected accordingly (power output data are not presented and included in the models). The ergometer allowed the cyclists to use their own bicycles, so they could perform the test in their usual position. After a 15-min warm-up at 100 W, the test started at 100 W, and resistance increased by 25 W every 30 s until volitional exhaustion. Cyclists were instructed to maintain a cadence between 90 and 95 rpm throughout the test. Maximal and submaximal respiratory parameters were measured using a breath-by-breath automated gas analysis system (VMAX29; Sensormedics, Yorba Linda, CA). Before each test, flow and volume were calibrated using a 3-L capacity syringe (Sensormedics), while gas analyzers (Sensormedics) were calibrated using known oxygen and carbon dioxide concentrations. Achievement of V˙O2max was considered as the attainment of at least two of the following criteria: 1) a plateau in V˙O2 with increasing power output (<80 mL·min−1), 2) an RER > 1.10, or 3) an HR ±10 beats·min−1 of age-predicted maximum (220 − age) (16). Ventilatory threshold (VT) and respiratory compensation point (RCP) were identified by expert technicians according to Gaskill et al. (14). Approximately 15% did not reach the V˙O2max criteria, and therefore, for the rest of the article, the term V˙O2peak will be used when referring to maximal aerobic power. The maximal RER and HR values reached by the cyclists were 1.16 ± 0.07 and 197 ± 8, respectively.

In the present study, we used for the analyses the physiological measures scaled by body mass raised to 0 (i.e., absolute values) and 1. These two exponents have been selected to reflect the level ground and climbing ability of the cyclists (31). Although other exponents have been proposed (4,15), we decided to use these two values for an easier comparison with the data reported in previous studies. Because the physiological variables normalized using the exponents suggested in the literature are highly correlated to each other (e.g., V˙O2peak × body mass−1 vs V˙O2peak × body mass−0.79: r = 0.97), it is unlikely that this choice substantially influenced the results of the analyses.

Back to Top | Article Outline

Statistical analysis.

Data are presented as mean ± SD unless otherwise stated. Before all the statistical tests, we checked the uniformity of errors, and we applied the parametric tests assuming the central limit theorem. The comparison between NAT and non-NAT junior cyclists was verified using an unpaired t-test. The ability of anthropometric and aerobic fitness variables to discriminate NAT from non-NAT was examined using the area under the curve (AUC) of the receiver operating characteristics (ROC) curve (6). The ROC curve is a method of assessing the discriminating ability of a test used to classify individuals into two groups, and it is calculated plotting the sensitivity against 1 − specificity, where sensitivity is the proportion of individuals correctly identified by the test (e.g., NAT) and specificity is the proportion of individuals who are non-NAT and correctly identified by the test. In practice, it is an estimation of the probability that an individual randomly chosen from one group will exceed an individual randomly selected from the other group. A common summary measure of discrimination is the AUC of the ROC with 0.5 representing chance (e.g., tossing a coin) and 1 corresponding to perfect discrimination. An AUC > 0.70 and confidence intervals (CI) > 0.50 were used as generic benchmarks for interpreting acceptable the discriminant ability.

To examine whether aerobic fitness measures derived from the laboratory incremental tests and being selected for the National team were able to predict the future career of junior cyclists, we tested two models using multimodal logistic regression. The predictor variables for both models were absolute and relative V˙O2peak and belonging to the National team. Absolute and relative V˙O2 at RCP and at VT were excluded from this analysis because they were highly correlated with V˙O2peak (r > 0.80). For the same reason, we did not include the power output measures. In the first model, the predicted variable was categorized as NP, PFL, and PCL. Sprinters were collapsed in the PFL group. In the second model, the predicted variable was categorized as NP, PC, and PT. As a measure of effect size for the logistic regression models, we reported two measures of pseudo R 2 (McFadden and Negelkerke), and we calculated the classification table. Univariate analyses (ANOVA) were also performed to examine differences among NP, PFL, and PCL and among NP, PC, and PT. When a significant F value was found, the Bonferroni post hoc test was applied. As a measure of effect size, the partial eta squared (η 2) was calculated for the ANOVA, and values of 0.01, 0.06, and >0.15 were considered small, medium, and large, respectively (10). ANOVA's assumptions were verified before running the analysis. Reference data were provided, reporting the values corresponding to each decile and presenting the frequency histogram. The probability of type I error (α) was set a priori at 0.05 in all statistical analyses.

Back to Top | Article Outline


Comparisons between groups.

The comparison between junior cyclists of NAT and non-NAT showed that NAT displayed higher body dimensions, absolute V˙O2peak, absolute and relative V˙O2 at RCP, and absolute V˙O2 at VT (Table 1). The relative V˙O2peak was greater in NAT [difference = 1.4 (95% CI = 0.414-3.211) mL·kg−1·min−1], but the effect size was trivial.



The results of the univariate analyses on the predictor variables are presented in Tables 2 and 3. A significant main effect for group (NP, PFL, and PCL) was found in all the measures of aerobic fitness, with the exception of height and body fat. The partial η 2 were, however, small for all the variables (<0.057). The post hoc tests after the ANOVA on NP, PFL, and PCL showed that PFL showed higher body weight, absolute V˙O2peak, V˙O2 at RCP, and V˙O2 at VT than NP. Conversely, PCL showed lower body weight, higher relative V˙O2peak, V˙O2 at RCP, and V˙O2 at VT than NP. A significant main effect was also found for the factor group with cyclists classified as NP, PC, and PT. Post hoc tests showed that the differences between aerobic fitness measures were found between NP and PT, with the latter showing greater values.





Back to Top | Article Outline

Multimodal logistic regression.

The results of the two models are presented in Table 4. The first model was significant (χ2 = 34.285, df = 6, P < 0.001). The likelihood ratio tests indicated that all the variables significantly influenced the model: relative V˙O2peak2 = 9.193, df = 2, P = 0.010), absolute V˙O2peak2 = 5.797, df = 2, P = 0.055), and belonging to NAT (χ2 = 9.171, df = 2, P = 0.010). The parameter estimates of the multimodal logistic regression showed that the odds of becoming PFL increased [odds ratio (OR) = 2.860] for every unit of increase in absolute V˙O2peak, whereas the odds decreased (OR = 0.210) if the cyclist was non-NAT. The odds of becoming PCL increased (OR = 1.206) for every unit of increase in relative V˙O2peak. However, the pseudo R 2 was low (McFadden = 0.155 and Negelkerke = 0.205), and this model, although significant, was not able to correctly classify the cyclists in the professional groups (all cyclists classified as NP). This suggests a poor predictive ability of these measures.



The second model was also significant (χ 2 = 28.985, df = 6, P < 0.001). The likelihood ratio tests showed that the national team factor and the relative V˙O2peak significantly influenced the model (P = 0.005 and P = 0.031, respectively). None of the variables was able to predict junior cyclists becoming P. On the other hand, the odds of becoming a PT slightly increased for every unit of increase in relative V˙O2peak (OR = 1.099). The cyclists not competing for the national team were less likely to become PT cyclists (OR = 0.174) than the riders selected for the national team. Similarly to the first model, the pseudo R 2 of the second model was low (McFadden = 0.129 and Negelkerke = 0.173), and the model was not able to correctly classify the cyclists in the groups, with all cyclists classified as NP.

Reference data for V˙O2peak are provided in Table 5, and the frequency histogram is presented in Figure 1. The mean relative V˙O2peak was 71 ± 7 mL·kg−1·min−1 (range = 50-89 mL·kg−1·min−1). The mean absolute V˙O2peak was 4.7 ± 0.6 L·min−1 (range = 2.9-6.7 L·min−1).





Back to Top | Article Outline


The main finding of this study is that the traditional measures of aerobic fitness determined during incremental tests in the laboratory are not able to predict the competitive level that junior cyclists can reach in adulthood. However, these measures can differentiate the competitive level of the young cyclists, suggesting that they can be useful for team selections in the junior category.

Traditionally, the candidate variables that can be used for talent identification are identified examining the parameters that are different among young cyclists of dissimilar competitive level (37). In the present study, the competitive level was based on being selected or not for the national team. The results showed that the NAT were taller and heavier (small and medium effect size, respectively). In addition, they had higher absolute maximal and submaximal oxygen uptake than non-NAT. These results were confirmed by the analysis of the AUC derived from the ROC curves of each measure. This test is more appropriate to examine the discriminant ability because a difference, although significant, does not necessarily imply that the variable is able to discriminate (20). This is clear examining the AUC of the Table 1 showing that among the variables that were significantly different between groups, only body mass, absolute V˙O2peak, V˙O2 at RCP, and VT showed acceptable discriminant ability (i.e., AUC close to 0.70). Because these characteristics are common among flat specialists, it seems that the NAT group mainly included riders with level-ground cycling ability. This was not surprising and can be explained by the altimetric profile of both international and national junior races. In fact, junior competitions are limited in their length by international and national rules, and the stage races' duration is limited up to 4 d. Although there are no restrictions on the total altitude climbed, analyzing the five recent junior World and European Championships, we found that the mean altitude climbed was of 1557 ± 290 m, whereas in the five professional World Championships, it was 4179 ± 1080 m (data supplied by the Italian Cycling Federation). These considerations may explain the overall advantage of possessing level-ground abilities for the junior cycling performance. Given the altimetric profile of the junior races and the fact that the junior national teams are tailored to be competitive in these races, it seems that flat specialists have an overall advantage at being selected.

A first analysis to examine whether the selected independent variables were different between P and NP and able to predict the competitive level reached in adulthood was conducted classifying the junior cyclists who became P on the basis of their specialty (flat specialists and climbers). The oxygen uptake values were superior in P versus NP, but, although significant, the effect sizes of the main effects were small. It is interesting to note that the characteristics of these junior cyclists who became professionals are very similar to those reported for professional adult cyclists (26). Indeed, the anthropometric characteristics of the PCL and PFL were similar to the PCL and time trialists reported by Lucia et al. (26). Likewise, the measures of aerobic fitness found in the present study for P were similar to the absolute V˙O2peak reported for flat specialists (approximately 5.2 L·min−1) and the relative V˙O2peak shown in climbers (approximately 78 mL·kg−1·min−1) (26). This shows that the strongest junior cyclists already displayed physiological and anthropometric characteristics similar to professional adult cyclists. This also suggests that other characteristics may be important to excel in cycling and compete at the professional level because it is unlikely that junior cyclists even displaying similar profiles can be competitive in professional races.

A logistic regression was used to examine the ability of anthropometric characteristics and the selected measures of aerobic fitness to predict the competitive level reached by these junior cyclists. An increase of one unit in absolute V˙O2peak significantly increased the OR of becoming a PFL (OR = 2.860), and not being selected for the national team decreased this probability (OR = 0.210). On the contrary, a higher V˙O2peak increased the chances of becoming PCL (OR = 1.206). This confirmed that measures of aerobic fitness scaled by body mass reflect the climbing ability of the cyclists, whereas the absolute values are important for level-ground cyclists. Moreover, to become PCL, being NAT had no influence. This supports the assumption that climbers have greater difficulty in demonstrating their potential in the junior category. However, although significant, the model showed no predictive ability because it was not able to correctly classify the cyclists in the P group.

A second analysis was performed, grouping the junior cyclists who became P according to the professional cycling level (PT and PC). All the oxygen uptake values were significantly higher in the PT than in the NP group, but the effect sizes were small, suggesting a nonsubstantial difference. Furthermore, logistic regression analysis showed that none of these variables influenced the chances of becoming P. On the other hand, a unit increase in relative V˙O2peak increased the odds of becoming PT but not substantially (OR = 1.099). Instead, not being selected by the national team decreased importantly the odds of becoming PT (OR = 0.174). This might be because being selected by the national team gave the cyclists higher chances to be recruited by the best U23 teams, hence to have good results and afterward to become PT. The odds of becoming a PT tend to increase also according to relative V˙O2peak. This may be explained by the different kind of races performed by the professional cyclists of different team levels. PT riders are more likely to participate in "Grand Tours" than PC riders. Therefore, PT teams attempt to select the strongest climbers (i.e., with the highest relative physiological values). However, it must be taken into account that the PC group included only two climbers, and this may have influenced the results. Nevertheless, this model was not able to predict the junior cyclists who became professionals either.

The second aim of this study was to provide normative data for the junior cycling category (Table 5). In this study, we presented the data of a large cohort. Few studies have reported data similar to the present study for junior cyclists. Bunc et al. (9) tested 11 male juniors (17.7 ± 1.8 yr, 176.5 ± 3.4 cm, 65.7 ± 3.6 kg) among the best Czech cyclists. The lower V˙O2max values (4.27 ± 0.32 L·min−1, 65.4 ± 5.1 mL·kg−1·min−1) are possibly because, in the Czech Republic, cycling is not as popular as in Italy, so their average performance level may be not as high. A study by Perez-Landaluce et al. (35) analyzed 26 male Spanish cyclists (18.3 ± 0.9 yr, 176.5 ± 6.4 cm, 66.4 ± 6.4 kg). Also, in this study, V˙O2max values were lower (4.4 ± 0.4 L·min−1, 65.5 ± 3.9 mL·kg−1·min−1) than our results. The authors reported that cyclists involved in the study during training covered approximately 11,000-13,000 km·yr−1. To our knowledge, Italian juniors usually train approximately 20,000 km·yr−1. The lower values of the physiological parameters, when compared with our data, may be due to the different amount of training, as previously suggested for professional cyclists (35). The only study analyzing a sample of high-level junior cyclists showed results that are consistent with our findings. Woolford et al. (40) tested 10 high-level juniors (17.4 ± 0.4 yr, 183.8 ± 3.5 cm, 71.5 ± 3.8 kg) at different cadences. At cadence similar to our study, they reported absolute and relative V˙O2max (5.3 ± 0.2 L·min−1, 74.2 ± 2.3 mL·kg−1·min−1, respectively) similar to those measured in our investigation. Comparing the relative V˙O2max of the Australian riders with our normative data, they are close to the 70th percentile, whereas the absolute values are close to the 90th percentile. Again, this seems to confirm that, in junior cyclists, high absolute values characterize the strongest riders of this category.

Cross-sectional comparisons are the most typical method to identify which characteristics may be relevant for talent identification (37). Even if these characteristics are not necessarily able to differentiate young athletes performing at different competitive levels, they are able to predict the level these cyclists can reach in adulthood (37). One of the reasons suggested by Vaeyens et al. (37) is that the athletes may not necessarily retain those relevant characteristics in adulthood. However, the anthropometric and the aerobic characteristics of these cyclists are similar to those reported for professional adult riders. The lack of predictive validity of these measures of aerobic fitness is probably mainly due to the multifactorial nature of cycling performance, and hence, it is unlikely that a single incremental test may provide all the physiological determinants of the performance. Even if the V˙O2max is considered a prerequisite, other factors such as economy and anaerobic characteristics may be important for excelling in cycling (12,24). In addition, in predicting sporting success, the contribution of aspects other than physiology is often neglected. These aspects include technical ability, tactical skills, and socioeconomic conditions. Psychobiologic factors such as mental fatigue and perception of effort are also important determinants of endurance performance, as recently shown by Marcora et al. (28). All these aspects should be taken into account by professionals, coaches, researchers, and anybody who would like to have a better understanding of junior cyclists and cycling performance in general.

The present study has two main potential limitations: the statistical power and the possible "hidden" influence of doping practices. Although the sample size may seem relatively high, in logistic regression, the sample size needed usually ranges from hundreds to thousands of subjects, depending on various factors including the prevalence of the event to predict (17). The sample size of the predicted groups of the present study is relatively small compared with the total sample size, raising the concern of possible type II errors. However, the models showed no predictive ability at all. Indeed, none of the junior cyclists who became professional was correctly classified. This suggests that the poor predictive ability was unlikely to be caused by a low statistical power. The second concern is related to doping issues. Because more effective doping controls have been only recently introduced, we cannot completely exclude that some of the professional cyclists included in this study were not tested positive because of the lack of appropriate detection methods. However, irrespectively from the reason (doping or actual poor predictive ability of the selected parameters), the study showed that the variables measured during incremental laboratory tests do not predict the career of cyclists. Moreover, the potential influence of doping may be a concern for all the studies involving professional cyclists published in the last 30 yr.

In conclusion, the traditional physiological measures of aerobic fitness measured in the laboratory such as V˙O2peak, V˙O2 at RCP, V˙O2 at VT, and the anthropometric characteristics are useful in identifying the young cyclists who can excel in their age category. Some of these variables increase or decrease the odds of becoming professional cyclists. However, none of them can be used for talent identification and to predict the competitive level that can be reached in adulthood.

The authors thank Andrea Morelli, Massimiliano Coppini, Ermanno Rampinini, and Domenico Carlomagno for their help in the data collection. The authors also thank Caterina Cazzola for the English revision of the manuscript and Samuele M. Marcora for his helpful suggestions.

The authors dedicate this study to the coach of the Italian Professional Cycling Team, Franco Ballerini, who passed away on February 7, 2010.

No external financial support was required for this project. The authors have no professional relationships with companies or manufacturers who may benefit from the results of the present study. The results of the present study do not constitute endorsement by American College of Sports Medicine.

Back to Top | Article Outline


1. Abbott A, Collins D. Eliminating the dichotomy between theory and practice in talent identification and development: considering the role of psychology. J Sports Sci. 2004;22(5):395-408.
2. Ary D, Cheser Jacobs L, Razavieh A, Sorensen C. Introduction to Research in Education. 7th ed. Belmont (CA): Wadsworth; 2006.p. 359-66.
3. Balmer J, Davison RC, Bird SR. Peak power predicts performance power during an outdoor 16.1-km cycling time trial. Med Sci Sports Exerc. 2000;32(8):1485-90.
4. Batterham AM, Tolfrey K, George KP. Nevill's explanation of Kleiber's 0.75 mass exponent: an artifact of collinearity problems in least squares models? J Appl Physiol. 1997;82(2):693-7.
5. Bentley DJ, McNaughton LR, Thompson D, Vleck VE, Batterham AM. Peak power output, the lactate threshold, and time trial performance in cyclists. Med Sci Sports Exerc. 2001;33(12):2077-81.
6. Bewick V, Cheek L, Ball J. Statistics review 13: receiver operating characteristic curves. Crit Care. 2004;8(6):508-12.
7. Bishop D, Jenkins DG, Mackinnon LT. The relationship between plasma lactate parameters, W peak and 1-h cycling performance in women. Med Sci Sports Exerc. 1998;30(8):1270-5.
8. Bishop D, Jenkins DG, McEniery M, Carey MF. Relationship between plasma lactate parameters and muscle characteristics in female cyclists. Med Sci Sports Exerc. 2000;32(6):1088-93.
9. Bunc V, Heller J, Horcic J, Novotny J. Physiological profile of best Czech male and female young triathletes. J Sports Med Phys Fitness. 1996;36(4):265-70.
10. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Mahwah (NJ): Lawrence Erlbaum; 1988. p. 16-8.
11. Coyle EF. Integration of the physiological factors determining endurance performance ability. Exerc Sport Sci Rev. 1995;23(1):25-63.
12. Coyle EF. Physiological determinants of endurance exercise performance. J Sci Med Sport. 1999;2(3):181-9.
13. Faria EW, Parker DL, Faria IE. The science of cycling: physiology and training-part 1. Sports Med. 2005;35(4):285-312.
14. Gaskill SE, Ruby BC, Walker AJ, Sanchez OA, Serfass RC, Leon AS. Validity and reliability of combining three methods to determine ventilatory threshold. Med Sci Sports Exerc. 2001;33(11):1841-8.
15. Heil DP. Scaling of submaximal oxygen uptake with body mass and combined mass during uphill treadmill bicycling. J Appl Physiol. 1998;85(4):1376-83.
16. Howley ET, Bassett DR Jr, Welch HG. Criteria for maximal oxygen uptake: review and commentary. Med Sci Sports Exerc. 1995;27(9):1292-301.
17. Hsieh FY. Sample size tables for logistic regression. Stat Med. 1989;8(7):795-802.
18. Impellizzeri FM, Ebert T, Sassi A, Menaspa P, Rampinini E, Martin DT. Level ground and uphill cycling ability in elite female mountain bikers and road cyclists. Eur J Appl Physiol. 2008;102(3):335-41.
19. Impellizzeri FM, Marcora SM. The physiology of mountain biking. Sports Med. 2007;37(1):59-71.
20. Impellizzeri FM, Marcora SM. Test validation in sport physiology: lessons learned from clinimetrics. Int J Sports Physiol Perform. 2009;4(2):269-77.
21. Impellizzeri FM, Marcora SM, Rampinini E, Mognoni P, Sassi A. Correlations between physiological variables and performance in high level cross country off road cyclists. Br J Sports Med. 2005;39(10):747-51.
22. Impellizzeri FM, Rampinini E, Sassi A, Mognoni P, Marcora SM. Physiological correlates to off-road cycling performance. J Sports Sci. 2005;23(1):41-7.
23. Jackson AS, Pollock ML. Generalized equations for predicting body density of men. Br J Nutr. 1978;40(3):497-504.
24. Lucia A, Hoyos J, Chicharro JL. Physiology of professional road cycling. Sports Med. 2001;31(5):325-37.
25. Lucia A, Hoyos J, Perez M, Santalla A, Earnest CP, Chicharro JL. Which laboratory variable is related with time trial performance time in the Tour de France? Br J Sports Med. 2004;38(5):636-40.
26. Lucia A, Joyos H, Chicharro JL. Physiological response to professional road cycling: climbers vs. time trialists. Int J Sports Med. 2000;21(7):505-12.
27. Lucia A, Pardo J, Durantez A, Hoyos J, Chicharro JL. Physiological differences between professional and elite road cyclists. Int J Sports Med. 1998;19(5):342-8.
28. Marcora SM, Staiano W, Manning V. Mental fatigue impairs physical performance in humans. J Appl Physiol. 2009;106(3):857-64.
29. Martin DT, McLean B, Trewin C, Lee H, Victor J, Hahn AG. Physiological characteristics of nationally competitive female road cyclists and demands of competition. Sports Med. 2001;31(7):469-77.
30. Morris T. Psychological characteristics and talent identification in soccer. J Sports Sci. 2000;18(9):715-26.
31. Mujika I, Padilla S. Physiological and performance characteristics of male professional road cyclists. Sports Med. 2001;31(7):479-87.
32. Nichols JF, Phares SL, Buono MJ. Relationship between blood lactate response to exercise and endurance performance in competitive female master cyclists. Int J Sports Med. 1997;18(6):458-63.
33. Padilla S, Mujika I, Cuesta G, Goiriena JJ. Level ground and uphill cycling ability in professional road cycling. Med Sci Sports Exerc. 1999;31(6):878-85.
34. Pearson DT, Naughton GA, Torode M. Predictability of physiological testing and the role of maturation in talent identification for adolescent team sports. J Sci Med Sport. 2006;9(4):277-87.
35. Perez-Landaluce J, Fernandez-Garcia B, Rodriguez-Alonso M, et al. Physiological differences and rating of perceived exertion (RPE) in professional, amateur and young cyclists. J Sports Med Phys Fitness. 2002;42(4):389-95.
36. Schumacher YO, Mroz R, Mueller P, Schmid A, Ruecker G. Success in elite cycling: a prospective and retrospective analysis of race results. J Sports Sci. 2006;24(11):1149-56.
37. Vaeyens R, Lenoir M, Williams AM, Philippaerts RM. Talent identification and development programmes in sport: current models and future directions. Sports Med. 2008;38(9):703-14.
38. Westgarth-Taylor C, Hawley JA, Rickard S, Myburgh KH, Noakes TD, Dennis SC. Metabolic and performance adaptations to interval training in endurance-trained cyclists. Eur J Appl Physiol Occup Physiol. 1997;75(4):298-304.
39. Williams AM, Reilly T. Talent identification and development in soccer. J Sports Sci. 2000;18(9):657-67.
40. Woolford SM, Withers RT, Craig NP, Bourdon PC, Stanef T, McKenzie I. Effect of pedal cadence on the accumulated oxygen deficit, maximal aerobic power and blood lactate transition thresholds of high-performance junior endurance cyclists. Eur J Appl Physiol Occup Physiol. 1999;80(4):285-91.


©2010The American College of Sports Medicine