European Journal of Applied Physiology

, Volume 112, Issue 2, pp 513–523

Iron status in elite young athletes: gender-dependent influences of diet and exercise


    • Institute of Biochemistry, German Sport University
    • German Research Centre of Elite SportsGerman Sport University
  • Hans Braun
    • Institute of Biochemistry, German Sport University
    • German Research Centre of Elite SportsGerman Sport University
  • Silvia Achtzehn
    • German Research Centre of Elite SportsGerman Sport University
    • Institute of Training Science and Sport Informatics, German Sport University
  • Ursula Hildebrand
    • German Research Centre of Elite SportsGerman Sport University
    • Institute of Cardiology and Sport Medicine, German Sport University
  • Hans-Georg Predel
    • Institute of Cardiology and Sport Medicine, German Sport University
  • Joachim Mester
    • German Research Centre of Elite SportsGerman Sport University
    • Institute of Training Science and Sport Informatics, German Sport University
  • Wilhelm Schänzer
    • Institute of Biochemistry, German Sport University
Original Article

DOI: 10.1007/s00421-011-2002-4

Cite this article as:
Koehler, K., Braun, H., Achtzehn, S. et al. Eur J Appl Physiol (2012) 112: 513. doi:10.1007/s00421-011-2002-4


Iron depletion seems to occur more frequently among athletes than in the general population and may affect performance capacity. Only little information is available about the prevalence of iron status abnormalities in young elite athletes and whether iron depletion is associated with gender, sport, age or nutrition- and exercise-related factors in this group. Hence, diet, exercise and haematological data from 193 elite athletes (96 males, 97 females; 16.2 ± 2.7 years) from 24 different sports were analyzed retrospectively. Most female athletes failed to meet the recommended daily allowance for iron, even though dietary iron density was higher than in males (5.75 ± 0.78 vs. 6.17 ± 0.98 mg/1,000 kcal; P = 0.001). Iron depletion (serum ferritin < 35 μg/L) occurred in 31% of male and 57% of female athletes (P < 0.001). Low haemoglobin (males: <13 g/dL; females: <12 g/dL) and haematocrit (males: <40%; females: <36%) values were equally prevalent in both genders [haemoglobin: 7.3% (males), 6.2% (females); haematocrit: 13.5% (males); 15.5% (females)]. In females, reduced ferritin levels were associated with a lower dietary iron density (5.9 ± 0.8 vs. 6.6 ± 1.1 mg/1,000 kcal; P = 0.002). Males with iron depletion had a significantly higher estimated energy expenditure (48.7 ± 7.0 vs. 44.4 ± 7.6 kcal/kg/day; P = 0.009).


FerritinDietary iron densityExercise expenditureSupplementation


Iron is considered as one of the most critical micronutrients in the field of exercise nutrition (American Dietetic Association et al. 2009). Recent reviews of the literature conclude that athletes with suboptimal iron status may experience reduced exercise capacity and impaired sport performance (Rodenberg and Gustafson 2007; Zoller and Vogel 2004; Haas and Brownlie 2001).

In the diagnosis of iron status disorders, it is important to discriminate between reduced iron stores, depleted iron stores, and clinical conditions, such as iron deficiency anaemia (IDA). Apart from haemoglobin and haematocrit, which are typically used for the detection of IDA, serum ferritin is the most frequently used parameter for the screening of an athlete’s iron status and for the detection of iron depletion (Fallon 2004; Nielsen and Nachtigall 1998). Other diagnostic parameters such as serum transferrin receptor concentration or transferrin receptor index have also been suggested (Punnonen et al. 1997).

Among athletes, the prevalence of IDA lies in the range of about 3% and is comparable to the general population (Shaskey and Green 2000). IDA is defined by reduced haemoglobin concentrations and consequently exercise capacity is directly affected by the reduced oxygen-binding and transport capacity (Haas and Brownlie 2001).

There is a general consensus that depleted iron stores without clinical symptoms occur more frequently in athletes than in the general population (Rodenberg and Gustafson 2007). Prevalences of iron depletion as high as 58% have been reported for at-risk groups (Petersen et al. 2006). In larger cross-sectional studies, the pooled mean prevalence was 37% in athletes and 23% in sedentary controls (Fogelholm 1995). With respect to iron depletion, several recent studies have shown that even small changes in the available body iron can have a positive effect on exercise performance, even though the mechanisms are not fully understood (Hinton and Sinclair 2007; Brutsaert et al. 2003; Friedmann et al. 2001; Hinton et al. 2000).

Reduced dietary iron and increased iron requirements have been identified as underlying causes of the increased prevalence of iron depletion in athletes. Dietary iron availability may be diminished due to inadequate intake or impaired intestinal absorption, whereas iron demands may be elevated because of increased losses via sweat and urine, gastro-intestinal bleeding, exercise-induced haemolysis or exercise-induced inflammatory processes (Peeling et al. 2008; Rodenberg and Gustafson 2007; Zoller and Vogel, 2004).

Due to increased growth requirements and potentially different food choices, junior athletes may be at risk for nutritional inadequacies (Anttila and Siimes 1996; Rowland et al. 1991). So far, only a small number of studies have addressed the prevalence and aetiology of iron depletion among elite young athletes. Data are available on male and female gymnasts and non-gymnasts (swimmers, tennis and table-tennis players) between 12 and 18 years (Constantini et al. 2000), female endurance athletes between 16 and 20 years (Malczweska et al. 2000) and 18-year-old male elite force soldiers (Merkel et al. 2009). In addition, young athletes have also been covered by a large cross-sectional study by Fallon (2008). However, most of these studies did not assess dietary iron intake or exercise behaviour.

Further comprehensive data on the haematological status, nutrition and exercise from a large group of young elite athletes is needed to understand the prevalence of iron depletion and other iron status abnormalities in this group. In addition, this data could also help to identify potential nutrition- and exercise-related risk factors for iron depletion in young athletes.

Therefore, the aim of our study was to determine the prevalence of iron status abnormalities in a large cohort of young athletes who train and perform on an elite level. As iron depletion was expected to be the most frequent iron status abnormality, we intended to further investigate whether iron depletion was associated with gender, sport, age, nutrition- and exercise-related factors such as iron intake, energy expenditure, supplement use or performance capacity in this group of athletes.


Study population

For the present study, we retrospectively analyzed data from athletes, who visited our centre between January 2007 and December 2008. All athletes participated in a routine examination, which included a comprehensive medical check-up, biomechanical and performance testing and a nutritional assessment amongst other tests. The aim of this program was to individually diagnose and improve the performance and performance-related aspects of young prospective athletes, who compete at national or international level. All athletes were selected for participation by their respective sports federation, so they can be considered to be among the (national) elite with respect to their age.

Written consent was obtained from each participant or in case of underage athletes from a legal representative. The study was approved by our University’s local ethics committee.

Data from athletes older than 25 years of age were excluded from analysis because this represents the maximal age at which athletes are eligible for international junior championships in certain sports.

Only those athletes for whom all necessary data (haematological profiles, nutrition and physical activity data) was available were included into the present analysis. Hence, from a total of 301 data sets, 88 were incomplete and rejected. For, 20 athletes, who had multiple examination dates within the time covered by the study, only the results of the first examination were included in this analysis. The anthropometric characteristics of the remaining 193 athletes can be found in Table 1.
Table 1

Anthropometric characteristics and reported training duration of the study participants


Overall (n = 193)

Males (n = 96)

Females (n = 97)

Mean ± SD (min; max)

Mean ± SD (min; max)

Mean ± SD (min; max)

Age (years)

16.2 ± 2.7 (11; 25)

16.1 ± 2.3 (13; 25)

16.3 ± 3.0 (11; 25)

Weight (kg)

65.6 ± 14.8 (33.4; 123.4)

72.4 ± 14.8 (42.0; 123.4)

58.8 ± 11.4 (33.4; 94.2)

Height (cm)

173.8 ± 11.7 (141; 202)

179.6 ± 10.0 (149; 202)

168.0 ± 10.3 (141; 188)

BMI (kg/m2)

21.4 ± 2.9 (16.0; 37.3)

22.2 ± 3.0 (17.2; 37.3)

20.6 ± 2.5 (16.0; 29.1)

Body Fat (%)

13.8 ± 6.0 (4.8; 34.4)

10.6 ± 4.8 (4.8; 30.5)

17.1 ± 5.1 (7.7; 34.4)

Training (min/day)

104 ± 58 (11; 347)

100 ± 47 (11; 266)

109 ± 67 (17; 347)

Twenty-four different sports were represented by the participants. Due to varying samples sizes, the athletes were clustered into the following sport groups: aesthetic sports [gymnastics (n = 10), synchronized swimming (n = 1)]; ball/team sports [basketball (n = 8), field hockey (n = 1), handball (n = 10), soccer (n = 14), volleyball (n = 1), water polo (n = 18)]; endurance sports [canoeing (n = 23), mountain biking (n = 1), Nordic combined (n = 5), rowing (n = 13), swimming (n = 9), triathlon (n = 3)]; combat sports [boxing (n = 4), judo (n = 13)]; racquet sports [badminton (n = 11), tennis (n = 12)]; winter/ice sports [bobsleigh (n = 1), luge (n = 11), skeleton (n = 4)]; and other sports, which did not fit into one of the above categories [archery (n = 4), fencing (n = 5)]. All track and field athletes were sprinters and jumpers (n = 8) or heptathletes/decathletes (n = 3) and thus were assigned to the group of other sports.

Most athletes participating in our study performed on the top national (n = 116) or international level (n = 41) with respect to their age group. Only a small number of athletes participated in senior world championships or Olympic competitions (n = 4) or were considered to reach this level within a given time (n = 14). This classification was performed by each athlete’s respective sports federation. For 22 participants, there was no classification available.

Study protocol

Following a period of 7 days in which the athletes recorded their nutrition and physical activity and an overnight fast, the athletes reported to our lab at 08:00 a.m. Body weight and body composition were measured by bioimpedance using a BA-418 MA balance (Tanita, The Netherlands). Body height was assessed to the nearest cm using a calibrated scale. A blood sample was also drawn from the anticubal vein. Athletes who failed to adhere to this protocol were classified as incomplete data sets.

Nutrition and exercise data

The parallel assessment of nutrition and exercise data was performed with a standardized food and activity record. The food record has recently been validated against doubly labelled water and 24-h urea excretion (Koehler et al. 2010). In brief, the standardized record contained a list of frequently used foods. Standard portion sizes were listed for all food items in order to minimize the respondents’ burden. All athletes were highly encouraged to add foods and to change portion sizes whenever necessary or applicable. The food record was analyzed using EBIspro software for Windows (version 7.0, 2005), which was used to calculate the average daily intake of nutrients and food groups based on the German Nutrient Data Base (Federal Research Centre for Nutrition and Food 2004). Nutrients originating from dietary supplements were not included into the present analysis.

For the evaluation of the intake of minerals, trace elements and vitamins, the recommended daily allowances (RDA) as published by the German Nutrition Society (Deutsche Gesellschaft für Ernährung 2008) were used.

Physical activity data was recorded simultaneously by the athletes using a structured record attached to the food record. The record listed 25 activities, which were divided into general activities (sleep, work, school, etc.), leisure activities (locomotion, sports) and exercise-related activities (training, competition, conditioning etc.). The athletes were instructed to write down the duration of all listed activities for each day as accurately as possible. The record has also been validated against doubly labelled water and indirect calorimetry (Koehler et al. 2010).

For the calculation of energy expenditure (EE), reference metabolic equivalent (MET) values were used (Ainsworth et al. 2000). Basal metabolic rate (BMR) was estimated based on body composition data (Cunningham 1991), which has been found to adequately predict BMR in athletes (Thompson and Manore 1996). For adolescent athletes, BMR was multiplied with age- and gender-specific factors (Harrell et al. 2005). Exercise EE (ExEE) represented the sum of the EE of all exercise-related activities. Daily total EE included BMR, leisure activity EE and ExEE (Koehler et al. 2010) and was calculated by adding the EE of all recorded activities.

Iron supplement use

The use of dietary supplements was documented by the athletes using a separate closed-end questionnaire, which has been described elsewhere (Braun et al. 2009). The athletes were instructed to record the use of specific supplements in the past (“never”/”past”) and within the last 4 weeks prior to the examination (“present”). For presently used supplements, the athletes also had to report the frequency of use, ranging from 1 to 2 times per month to daily. The supplement questionnaire was filled out by only 138 athletes (71% of the whole study population). The remaining 55 athletes were omitted only from analyses regarding supplement use.

Haematological data

The following routine measures were assessed from the blood samples: Haemoglobin, haematocrit, mean cellular volume (MCV), mean corpuscular haemoglobin (MCH), mean corpuscular haemoglobin concentration (MCHC), leukocyte count, and platelets were measured directly from the blood samples on a Sysmex KX-21N (Sysmex, Nordersted, Germany). Serum samples were analyzed for ferritin on an Elecsys 1020 (Roche, Mannheim, Germany) and for serum iron concentration and serum creatine kinase (CK) levels with a Cobas 400 (Roche, Mannheim, Germany).

Haematological parameters were classified as abnormal based on standard reference values. Serum ferritin was considered low at concentrations below 35 μg/L and critical at values under 12 μg/L (Peeling et al. 2008). Haemoglobin and haematocrit values were considered abnormal in females below 12 g/dL and 36% and in male athletes below 13 g/dL and 40% (Deakin 2006). For the MCV, a threshold of 80 fL was used (Deakin 2006). For the diagnosis of iron overload, threshold values were 31.3 μmol/L for serum iron and 150 (females) and 200 μg/L, respectively (males) for ferritin (Deakin 2006).

Physical performance capacity

On the examination day, all athletes performed an incremental running test in order to assess endurance performance capacity. The protocol and criteria for the assessment of peak oxygen uptake (VO2peak) and running velocities at blood lactate concentrations of 2 mmol/L (v2) and 4 mmol/L (v4) have been described in detail elsewhere (Sperlich et al. 2011). In short, following an individual warm-up and familiarization, the athletes started running at 2.4 m/s and speed increased every 5 min by 0.4 m/s until individual exhaustion. Respiratory gas exchange was measured with an open breath-by-breath spirograph (nSpire Health, Oberthulba, Germany). Reliable test results were available for 92 male and 89 female athletes.

Statistical analysis

Statistics were performed with R software (version 2.8.0, The R Foundation for Statistical Computing, 2008). For the detection of differences, the Mann–Whitney U test was used for continuous data and the two-sided Fisher test was used for discrete data. Differences were considered to be statistically significant at a probability of error below 5% (P < 0.05). If not stated otherwise, values were expressed as mean ± standard deviation (SD).


Dietary data

As shown in Table 2, dietary iron intake was significantly greater in male than in female athletes. 78 male athletes (81%) reached the RDA for iron. In the majority of the female athletes, dietary iron intake was below the recommendation (61 = 63%). After iron intake was normalized for body weight, values were almost identical for male and female athletes.
Table 2

Dietary intake data of nutrients associated with haematological values


Male athletes (n = 96)

Female athletes (n = 97)

P value

Iron intake (mg/day)

17.0 ± 5.3

13.8 ± 4.1


Iron intake (% RDA)

142 ± 49

93 ± 28


Athletes with intake <RDA

18 (19%)

61 (63%)


Iron intake (mg/kg/day)

0.24 ± 0.07

0.24 ± 0.07


Iron density (mg/1,000 kcal)

5.75 ± 0.78

6.17 ± 0.98


Energy intake (kcal/day)

2951 ± 800

2268 ± 677


Energy intake (kcal/kg/day)

41.7 ± 11.3

39.7 ± 13.1


Dietary iron intake and daily energy intake were significantly correlated in both male and female athletes (rmales = 0.91; rfemales = 0.83). Dietary iron density, which refers to the ratio between iron and energy intake, was significantly higher in female athletes (Table 2). Total energy intake was significantly greater in male athletes, but after normalization for body weight, this difference was not significant.

Haematological data

Ferritin, haemoglobin and haematocrit values are summarized in Table 3.
Table 3

Hematological data and prevalence of abnormal values by gender and sport


Aesthetic sports (n = 0)

Ball sports (n = 27)

Endurance sports (n = 32)

Combat sports (n = 8)

Racquet sports (n = 10)

Winter sports (n = 10)

Other sports (n = 9)

Overall (n = 96)

Male athletes




  Mean ± SD


60.3 ± 37.2

49.9 ± 33.2

33.9 ± 24.3

48.2 ± 30.3

73.2 ± 48.0

67.6 ± 41.7

55.4 ± 36.7

  <35 μg/L


7 (26%)

13 (41%)

4 (50%)

4 (40%)

1 (10%)

1 (11%)

30 (31%)

  <12 μg/L


2 (7%)

0 (0%)

2 (25%)

0 (0%)

0 (0%)

0 (0%)

4 (4%)

 Hemoglobin (g/dL)


  Mean ± SD


14.8 ± 1.4

14.6 ± 0.6

13.7 ± 1.8a

15.0 ± 1.0

14.7 ± 1.0

14.8 ± 0.8

14.7 ± 1.1

  Abnormal (<13 g/dL)


3 (11%)

1 (3%)

2 (25%)

0 (0%)

1 (10%)

0 (0%)

7 (7%)

 Hematocrit (%)


  Mean ± SD


43.0 ± 3.1

42.0 ± 1.7

40.5 ± 4.0a

43.0 ± 2.1

41.9 ± 3.0

42.7 ± 2.1

42.3 ± 2.6

  Abnormal (<40%)


4 (15%)

3 (9%)

3 (37%)

0 (0%)

2 (20%)

1 (11%)

13 (%)

Female athletes

Aesthetic sports (n = 11)

Ball sports (n = 25)

Endurance sports (n = 22)

Combat sports (n = 9)

Racquet sports (n = 13)

Winter sports (n = 6)

Other sports (n = 11)

Overall (n = 97)



  Mean ± SD

29.2 ± 13.4

35.4 ± 19.4

41.2 ± 31.3

34.9 ± 21.2

34.1 ± 17.4

47.6 ± 26.2

25.3 ± 12.4

35.4 ± 22.1

  <35 μg/L

8 (73%)

14 (56%)b

12 (54%)

5 (56%)

7 (54%)

3 (50%)

8 (72%)c

57 (59%)

  <12 μg/L

1 (9%)

1 (4%)

1 (4%)

1 (11%)

1 (8%)

0 (0%)

2 (18%)

7 (7%)

 Hemoglobin (g/dL)


  Mean ± SD

13.5 ± 0.7

13.1 ± 0.9

13.2 ± 0.9

12.9 ± 0.5

12.8 ± 0.7

14.1 ± 0.8

13.2 ± 1.1

13.2 ± 0.9

  Abnormal (<12 g/dL)

0 (0%)

4 (16%)

1 (5%)

0 (0%)

0 (0%)

0 (0%)

1 (9%)

6 (6%)

 Hematocrit (%)


  Mean ± SD

39.6 ± 2.0

38.6 ± 2.4

38.4 ± 2.5

38.4 ± 2.1

37.5 ± 2.1a

40.8 ± 1.7

38.4 ± 2.9

38.6 ± 2.4

  Abnormal (<36%)

1 (9%)

3 (12%)

5 (22%)

1 (11%)

4 (31%)

0 (0%)

1 (9%)

15 (15%)

aSignificantly different from other sports, same gender (P < 0.05)

bSignificantly different from male athletes, same sport (P < 0.05)

cSignificantly different from male athletes, same sport (P < 0.01)

Serum ferritin concentrations were higher in male (55.4 ± 36.7 μg/L) than in female athletes (35.4 ± 22.0 μg/L, P < 0.001) but the relatively large standard deviation indicates a strong interindividual variation. Low ferritin values indicative of iron depletion (<35 μg/L) occurred more frequently in female athletes (57 = 59%) than in male athletes (30 = 31%, P < 0.001). The prevalence of critical values (<12 μg/L) was low (males: 4, females: 7).

In male athletes under 18 years of age iron depletion occurred in 35% (28 of 80), whereas only 13% (2 of 16) aged 18 years or older were iron depleted. This difference did not reach statistical significance (P = 0.14). In females, the prevalence of iron depletion was comparable between younger and older athletes [45/76 = 59% (<18 years); 12/21 = 57% (≥18 years); P = 0.99]. With respect to the small group sizes, especially for older male athletes with iron depletion (n = 2), we abstained from further differentiations between age groups with regard to other parameters.

In male athletes, lowest ferritin levels were seen in combat sports (33.9 ± 24.3 μg/L). Among females, those athletes engaged in aesthetic sports (29.2 ± 13.4 μg/L) and in other sports (25.3 ± 12.4 μg/L) exhibited the lowest ferritin concentrations. The differences to the other groups were not statistically significant.

Haemoglobin levels were higher in male (14.7 ± 1.1 g/dL) than in female athletes (13.2 ± 0.9 g/dL, P < 0.001). Under consideration of the different normal ranges for male and females, the prevalence of abnormally low values was comparable (7.3 vs. 6.2%). In male combat sport athletes, haemoglobin values were significantly lower than in all other sports. In female athletes, haemoglobin concentrations were lowest in racquet sports and in combat sports but this effect did not reach significance.

For haematocrit data, there were very similar trends. Average values were lower in female athletes (38.6 ± 2.4%) than in male athletes (42.3 ± 2.6%, P < 0.001) and again, in male athletes haematocrit was significantly lower in those involved in combat sports than in those involved in all other sports. In female athletes, haematocrit values were significantly lower only in racquet sports. The prevalence of low haematocrit values was also comparable between male and female athletes (13.5 and 15.5%, respectively).

The prevalence of other abnormal values was rather low. Eight athletes (4 males, 4 females) had an MCV below 80 fL and 4 of those (3 males, 1 females) also had low haemoglobin and haematocrit values and ferritin below 12 μg/L, which suggested the presence of IDA. High serum iron concentrations indicative of iron overload were observed in 10 male and 13 female athletes. One female athlete had a ferritin concentration of 156 μg/L, but normal serum iron levels.

Two female athletes had elevated leukocyte counts (>10×109 /L) and one female athlete had a mildly elevated platelet count (413×109 /L), but in all cases ferritin was in the normal range. The majority of the male athletes (65 = 68%) had serum CK levels above 200 U/L. Only 14 males had CK values between 500 and 1,500 U/L, but one of these athletes also had a ferritin concentration of 171 μg/L. In females, CK was higher than 200 U/L in 35 athletes (36%) and higher than 500 U/L in 5 athletes (5%). None of these athletes had abnormally high ferritin values.

Neither in male nor in female athletes was there a significant correlation between ferritin concentrations and inflammatory markers.

Nutrition- and exercise-associated factors in relation to iron depletion

There were no significant differences between male athletes with iron depletion and those with normal ferritin values with regard to total dietary iron intake (P = 0.99) or to dietary iron density (P = 0.81; Fig. 1). Female athletes with normal ferritin levels consumed more iron than athletes with low ferritin values. When express as percentage of the RDA, the difference of 8.5% was not statistically significant (P = 0.27). After normalization for energy intake, female athletes with normal ferritin levels had an iron density which was 0.66 mg/1,000 kcal higher than female athletes with low ferritin concentrations (P = 0.002).
Fig. 1

Dietary iron intake expressed as % of the RDA (upper panel) and as iron density (lower panel) in athletes with ferritin levels indicative of iron depletion (“low”) and athletes with normal ferritin levels, separated by gender

In male athletes with reduced ferritin levels, total daily energy expenditure (Fig. 2) was higher than in those with normal ferritin concentrations (48.7 ± 7.0 vs. 44.4 ± 7.5 kcal/kg/day, P = 0.009). The difference in exercise-related EE between male athletes with low and normal ferritin levels did not reach statistical significance (11.7 ± 5.0 vs. 10.4 ± 5.2 kcal/kg/day, P = 0.27) and there were no differences in the self-reported training duration [99 ± 48 min/day (low ferritin levels) vs. 100 ± 47 min/day (normal ferritin levels)]. For leisure time EE there was also a trend towards higher levels in athletes with low ferritin levels, but again the difference was not statistically significant [7.5 ± 1.7 (low) vs. 5.7 ± 2.4 kcal/kg/day (normal), P = 0.17].
Fig. 2

Total daily energy expenditure (upper panel) and the ratio between energy intake and energy expenditure (lower panel) in athletes with ferritin levels indicative of iron depletion (“low”) and athletes with normal ferritin levels, separated by gender

The higher energy expenditure was accompanied by a higher dietary energy intake [46.2 ± 12.2 kcal/kg/day (males with low ferritin) vs. 39.6 ± 10.3 kcal/kg/day (males with normal ferritin), P = 0.011] so that the ratio of energy intake and energy expenditure was not significantly different (0.95 ± 0.22 (reduced ferritin) vs. 0.90 ± 0.20 (normal ferritin), P = 0.30).

In female athletes, there were no statistically significant differences between athletes with low and normal ferritin levels with respect to energy expenditure [45.8 ± 9.8 (low) vs. 43.2 ± 6.5 kcal/kg/day (normal); P = 0.85], exercise-related EE [11.1 ± 5.6 (low) vs. 9.2 ± 4.7 kcal/kg/day (normal); P = 0.21], practice duration [117 ± 74 (low) vs. 98 ± 74 min/day (normal); P = 0.41], leisure EE [6.3 ± 1.5 (low) vs. 6.5 ± 1.3 kcal/kg/day (normal), P = 0.41], energy intake [40.6 ± 13.8 (low) vs. 38.5 ± 12.0 kcal/kg/day (normal); P = 0.79] and the ratio of energy intake and expenditure [0.89 ± 0.24 (low) vs. 0.89 ± 0.23 (normal); P = 0.85].

Male athletes with normal ferritin values consumed significantly more meat and fish (80 ± 47 g/day) than male athletes with low ferritin values (57 ± 57 g/day, P = 0.008). This trend did not occur in female athletes [32 ± 34 (normal) vs. 35 ± 33 g/day (low); P = 0.53].

Use of iron supplements

Information on the use of dietary supplement was available for only 138 athletes (73 males, 65 females). The majority of the athletes (51 males, 44 females) reported that they had never used iron supplements. Among the remaining 43 supplement users (22 males, 21 females), a total of 15 (6 males, 9 females) reported to have taken iron supplements within four weeks prior to the examination. In average, self-reported frequency of use was 4.8 ± 2.8 times per week (range: twice per month–daily).

Neither in males nor in female athletes, there was a significant difference in serum ferritin levels between iron supplement users and non-users. Only in female athletes, serum iron concentrations were significantly higher in those who reported iron supplement use within the last four weeks prior to the investigation compared to athletes who did not use iron supplements at the time of investigation (P < 0.05) or who did not use them at all (P < 0.05).

Among those participants who took iron supplement at the time of investigation, four athletes (all females) had abnormally high serum iron concentrations indicative of an iron overload.

Physical performance capacity

There were no differences in VO2peak between athletes with low and normal iron status, neither in male [53.0 ± 8.3 (low) vs. 51.8 ± 10.4 ml/kg/min (normal), P = 0.56] nor in female athletes [45.5 ± 6.3 (low) vs. 45.6 ± 6.9 ml/kg/min (normal), P = 0.98]. In addition, there was no association between iron status and running velocities at blood lactate concentrations of 2 and 4 mmol/L (data not shown).


This retrospective study was aimed at assessing the prevalence of abnormal haematological data and at identifying associations between iron depletion and nutrition- and exercise-related factors in elite young athletes.

Haematological data

More than one third of our athletes had low ferritin values indicative of iron depletion, which is in the range of other studies. Among gymnasts, who are generally considered as an at-risk group for nutritional deficiencies (Calabrese et al. 1985), the prevalence of low ferritin levels was 30–36% (Constatini et al. 2000). According to a review by Fogelholm (1995), the pooled prevalence of abnormal ferritin levels was 37% in female athletes.

Thirteen athletes from our cohort had haemoglobin concentrations indicative of anaemia. At 6.7%, the prevalence of anaemia was about twofold higher than reported for adult athletes (Fogelholm 1995).

Based on our data, female athletes were about twice as likely to have reduced ferritin levels (58.8 vs. 31.2%). Low ferritin levels were also found more frequently among female athletes by Fallon (2004) and Fogelholm (1995). Especially female adolescent athletes have been identified as a group at-risk of iron depletion. In female runners between 15 and 18 years, 34% had serum ferritin levels below 12 μg/L (Nickerson et al. 1989). In female swimmers, critically low ferritin levels (<12 μg/L) occurred in 46–58% (Petersen et al. 2006). In comparison to these values, the prevalence of critically low ferritin levels was rather low in our group of athletes (7% females, 4% males).

Reduced serum ferritin levels seemed to occur more frequently in younger male athletes, even though this trend did not reach significance. In female athletes, there was no difference between younger and older athletes. This is in agreement with other population studies, which showed that in males but not in females, there is a slow increase in serum ferritin through adolescence and a more pronounced increase in late adolescence (Cook et al. 2003). At least in males, these pubertal changes have been connected to increased growth requirements (Anttila and Siimes 1996).

Dietary data

Iron intake was below the recommendations in the majority of the female athletes, whereas most male athletes consumed sufficient iron. Female athletes had a significantly higher iron density in their diet but this did not fully compensate for the higher RDA (15 vs. 10–12 mg/day for males). This is in agreement with a study in endurance athletes, in which female athletes failed to meet the RDA for iron (Weight et al. 1992). In pubescent athletes, dietary iron intake was also found to be lower in females than in males, even though in this group recommendations were met by most subjects (Fogelholm et al. 2000). In contrast to our study, Fogelhom et al. (2000) did not find a difference in dietary iron density between male and female athletes.

Nutrition- and exercise-associated factors in relation to iron depletion

In female athletes in our study, low ferritin levels were significantly associated with a reduced dietary iron density. This implies that female athletes with a high dietary iron density were more likely to have a normal iron status. When iron density was higher than 6 mg/1,000 kcal, which represents a typical Western diet (Beard and Tobin 2000), the relative risk for low ferritin values was 2.3-fold lower. This is in agreement with review by Beard and Tobin (2000) who identified dietary iron density as one of the main causes of iron depletion in female athletes. Female athletes with normal ferritin values also had a higher dietary iron intake (97 vs. 83% of the RDA) but this effect did not reach statistical significance. We can only speculate about the reasons why iron density was linked to the iron status more directly than total iron intake but dietary assessment studies have shown that relative nutrient densities are more robust against underreporting than absolute intakes (Kant 2002).

Male athletes with low ferritin values had a higher level of energy expenditure than those with normal ferritin concentrations. We found a significant difference between athletes with low and normal iron status only for estimated total EE, whereas the differences in exercise-associated EE and leisure EE did not reach statistical significance. It seems unlikely that the iron status is directly associated with energy expenditure but EE can rather be seen as an indicator of the level of physical activity.

There are several studies reporting that serum ferritin levels are reduced with increasing physical activity such as at the onset of vigorous training (Ashenden et al. 1998) or with an increase in training load (Dallongueville et al. 1992), but other studies did not find a direct association between low ferritin values and energy expenditure (Fogelholm et al. 1992a; Telford et al. 1993). The potential mechanisms for exercise-related increases in iron requirements have been discussed in detail elsewhere (Peeling et al. 2008).

Only about 8% of our athletes used iron supplements at the time of investigation. Four female athletes had abnormally high serum iron concentrations indicative of iron overload, a condition that has been associated with iron supplementation or iron injections (Zotter et al. 2004; Deugnier et al. 2002). The consequences of iron overload are not well described but excessive unabsorbed iron may cause mucosal damage in the colon and free iron is also known a potent pro-oxidant (Reddy and Clark 2004).

Apart from increased serum iron concentrations, there were no significant differences in biochemical indices between users and non-users. This has also been reported in another cross-sectional study (Fogelholm et al. 1992a). Even though all athletes were advised to comply with the overnight fast and not to use any medication in order to avoid acute effects on blood levels, it is possible that high serum iron levels resulted from supplements, which were consumed closely before the blood collection.

Limitations of the study

The use of ferritin as indicator of an athlete’s iron status has been criticized by several authors. There are no standardized thresholds for the detection of iron depletion (Rodenberg and Gustafson 2007) and in recent supplementation studies, cut-offs between 16 and 25 μg/L have been used for the inclusion of iron-deficient participants (Hinton et al. 2000; Friedmann et al. 2001; Brownlie et al. 2002; Brutsaert et al. 2003; Fogelholm et al. 1992b). In our cross-sectional study, a threshold of 35 μg/L was applied, which is commonly accepted (Peeling et al. 2008) and often used in the treatment of German athletes (Nielsen and Nachtigal 1998).

Questions to whether ferritin is susceptible to false positive or false negative results have also been raised (Mast et al. 1998; Fallon et al. 2001). As an acute-phase protein, serum ferritin concentrations may be elevated following acute strenuous exercise or inflammation (Deakin 2006). Leukocytes and platelets were mostly within the normal ranges in our group but serum CK levels were moderately elevated in about two-thirds of the male and one-third of the females. Systematically elevated CK levels have also been reported in judo athletes (Malczewksa et al. 2004) but in contrast to this study, we did not find an association between CK and ferritin levels. Furthermore, inflammation-associated increases in serum ferritin would have caused false negative results (normal ferritin levels despite iron depletion) and it is unlikely that this effect caused significant bias with regard to our aims.

A recent study in trained and untrained subjects found a negative correlation of ferritin levels with total haemoglobin mass and endurance performance capacity (Robinson et al. 2010). The authors argued that an increase in total haemoglobin mass may lead to an iron storage shift and consequently cause ferritin levels to drop. Since all of our athletes represented the national elite in their respective sport, they were considered to be in a relative steady state with regard to performance capacity, training intensity and total haemoglobin mass. Furthermore, we did neither find a positive nor a negative association between ferritin levels and performance capacity. However, small performance differences would have been blunted by the variety of sports included in the study.

Despite these limitations, serum ferritin is still one of the most commonly used indices of an athlete’s iron status and allows frequent and cost-effective measurements (Fallon 2004; Nielsen and Nachtigal 1998). Nevertheless, the inclusion of other parameters of the iron status such as serum transferrin receptor concentration (Schumacher et al. 2002) would have been useful and should be considered for future studies.

Unfortunately, we did not systematically assess menstruational losses in female athletes. It is well known that the onset of menstruation has an effect on the iron status (Milman et al. 1998). Malczewska et al. (2000) reported that the iron status was significantly associated with duration and intensity of the menses in endurance athletes. It can be expected that at least in some of our younger athletes menarche has not yet occurred, as 76% of elite athletes at an age of 13.5 ± 1.6 years were premenarche (Constantini et al. 2000).


In spite of the limitations, our investigation represents a cross-sectional study with a large and rather heterogeneous sample of elite young athletes. The high prevalence of iron depletion especially among females indicates that regular assessment of the iron status is also necessary in elite young athletes. Our results also suggest that iron depletion in young athletes may be associated with diet and exercise. When working with athletes, practitioners should consider these aspects and help to improve the athletes’ diet in order to prevent exercise- or diet-induced iron depletion. Longitudinal studies are necessary to investigate the efficacy of nutritional interventions and supplementation in this group of athletes.

The experiments were carried out in compliance with current German federal and state laws


The authors would like to acknowledge institutional funding by the German Research Centre of Elite Sports.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer-Verlag 2011