Journal Logo


Iron Supplementation Improves Energetic Efficiency in Iron-Depleted Female Rowers


Author Information
Medicine & Science in Sports & Exercise: June 2014 - Volume 46 - Issue 6 - p 1204-1215
doi: 10.1249/MSS.0000000000000208


Iron deficiency (ID; serum ferritin [sFer] < 12.0 μg·L−1) is the most prevalent nutrient deficiency in the world, including in the United States, where ID with anemia (IDA, ID plus hemoglobin [Hgb] <12.0 g·dL−1) affects 3%–5% and ID without anemia (IDNA, sFer < 20.0 μg·L−1) affects ∼16% of young women (4). Active females, including military soldiers, are especially susceptible, as surveys have shown even higher rates of IDNA (25%–52%) (7,27,31). Although research has clearly shown the effects of IDA on physical performance (15), the high prevalence of IDNA in female athletes requires investigation into its effects on physical performance in this group. Beyond menstrual status, the increased prevalence of IDNA in active females may be due to one or a combination of the following factors: poor dietary Fe intake (16), hemolysis (foot strike and impact) (29), increased Fe losses (gastrointestinal tract, hematuria, and sweat) (38), or altered intestinal Fe absorption, including the effects of inflammation due to training (13).

Despite the inability to identify the exact mechanism in humans, endurance performance has been shown to be impaired in laboratory studies of IDNA women (7,17,30,40). Studies of how Fe status affects the performance of nonanemic endurance athletes are limited, and results vary due to several factors. These include poor study design, classification of baseline Fe status, supplementation dose used, and subject training status. Some of the research, however, does suggest that female endurance athletes and military soldiers with IDNA have decreased physical performance, and Fe supplementation benefits those with depleted Fe stores (11,18,26).

Energetic efficiency (EF) and blood lactate metabolism are two measures of endurance performance. Energetic (work) EF (%) is defined as work output per kilocalorie expended and is expressed as a percent relative to V˙O2peak (mL·kg−1·min−1) or to work (W) performed. It is hypothesized that IDNA affects Fe-containing enzymes of the TCA cycle, which are involved in the transformation of chemical to mechanical energy used to produce work output. Blood lactate concentration is positively correlated with the amount of lactate produced in muscle, which is an indication of the degree of anaerobic metabolism at the muscle tissue level. Blood lactate measured during exercise is the result of lactate production, release, and removal from the muscle. Although researchers have shown that ID and anemic humans and animals display an earlier increase in lactate production and decreased rate of lactate clearance (14,33), results in IDNA humans are not clearly understood.

Previous work from our laboratory, as well as from others, has shown that IDNA reduces the endurance capacity of women by improving energetic EF (17,18). In addition, our group and others have shown improvements in lactate metabolism in Fe-supplemented women, especially during the earlier phases of endurance time trials (TTs) (22,32,40). In a recent study of female rowers, sFer status was associated with decreased TT time and increased energetic EF (8). Despite the strong cross-sectional associations between IDNA and endurance performance, there is inadequate evidence demonstrating how Fe repletion affects performance in highly trained IDNA endurance athletes who have much less potential to benefit from improvements in Fe status.

Given the high prevalence of IDNA among female athletes, and the strong associations between Fe status and physical performance in both sedentary and active women, the current experimental study was designed to determine the efficacy of low-dose Fe supplementation to prevent decrements in Fe status as well as to test for a causal relationship between Fe status and performance in female rowers who are beginning their competitive season’s training program. The objectives of this study were 1) to examine the effects of low-dose Fe supplementation on female collegiate rowers’ Fe status, 2) to examine the effects of supplementation on metabolic adaptations to rowing (endurance) training during a 4-km TT (V˙O2peak, gross energetic EF, lactate concentration, and time to complete TT), and 3) to investigate the relation of change in Fe status indicators and changes in endurance capacity and energy metabolism during a TT after 6 wk of training. We hypothesized that Fe supplementation would prevent training-related declines in Fe status in female rowers, and that Fe-supplemented rowers would improve their Fe status (sFer) and performance outcomes after 6-wk of treatment compared with a placebo group.


Recruitment of subjects

Subjects were recruited at the beginning of the conditioning phases of their competitive rowing seasons. All varsity and second-semester novice female rowers were eligible to participate in the screening if at least 18 yr of age, nonsmoking, and were able to begin regular training for their sport. Training activities during this time included general aerobic conditioning (cycling, running, and rowing on ergometer), resistance training, and high intensity on-water rowing (as weather permitted).

A medical screening required by the National Collegiate Athletic Association before our study excluded all athletes not healthy enough to participate in their rowing team training (current, acute or chronic illness, severe asthma, musculoskeletal problems, etc.). All rowers provided written informed voluntary consent before participating in the study. This study was approved by the institutional review boards of the following colleges/universities: Binghamton University, Cornell University, Hobart and William Smith Colleges, Ithaca College, and Syracuse University. This study was registered in the public trial registry maintained by the U.S. National Library of Medicine ( as “How Does Iron Supplementation Affect Training and Performance in Female Collegiate Rowers?” (NCT01383798).

Study design

This study was part of a larger investigation examining the relationships between iron status, training, and performance in female rowers. This particular body of work was a randomized, double-blind, placebo-controlled Fe supplementation trial (RCT). A total of 165 female rowers from the five schools completed an Fe status screening (see Fig. 1) (7). Ten percent of rowers (n = 16) screened were identified as anemic (Hgb < 12.0 g·dL−1) and excluded from further participation in the study. Thirty percent of the rowers were identified as IDNA (sFer < 20.0 μg·L−1, Hgb > 12.0 g·dL−1). All 149 nonanemic subjects with and without IDNA were invited to participate in the baseline physical performance and body composition testing. Forty-eight rowers (n = 24 IDNA and 24 normal iron status) participated in the baseline data collection (5 from Hobart and William Smith Colleges, 14 from Cornell University, 9 from Ithaca College, 12 from Binghamton University, and from 8 Syracuse University) (8). Forty rowers were then randomized to receive supplemental Fe (n = 21, 12 IDNA) or placebo (n = 19, 11 IDNA). Subjects received Fe status, body composition, and fitness testing results as a benefit of participation, along with referral and recommendations to improve Fe status as necessary.

Timeline and flow of subjects through RCT.

Each subject consenting to participate in the 6-wk trial was randomly assigned to a treatment group by a research assistant who was not involved in data collection or contact with subjects. Randomization was done by assigning each subject a random number, with even and odd numbers being assigned to either treatment group. After initial randomization, any imbalance in the distribution of treatment or representation of school or baseline Fe status (sFer) was corrected by rerandomization. Rowers were randomly assigned to one of two groups: Fe supplementation with 50 mg FeSO4 twice per day (total dose 100 mg FeSO4 per day) or placebo in the form of identical red capsules. Both Fe and placebo capsules were prepared by a Registered Pharmacist (PharmD) at the Cornell University College of Veterinary Medicine Pharmacy (Ithaca, NY). The Fe supplement capsules contained FeSO4 per capsule with lactose filler, and the placebo capsules contained only lactose. The Fe content of both placebo and Fe capsules was analyzed via inductively coupled plasma emission spectroscopy by the U.S. Department of Agriculture’s Robert W. Holley Center for Agriculture and Health (Ithaca, NY). Twenty capsules were randomly selected for analysis from each of two batches. No differences in the average Fe content were found between the two batches of Fe-containing capsules (15.8 ± 0.5 mg elemental iron per capsule), and no Fe was detected in the placebo capsules. In previous studies, this level of Fe supplementation using FeSO4 over the course of 6–8 wk was sufficient to improve Fe stores (sFer) in compliant, nonathletic women (17,40).

Subjects were provided with 18 capsules each week and were instructed to consume two capsules per day, one capsule each at their morning and evening meals to minimize potential gastrointestinal side effects, and with a glass of citrus juice to enhance Fe absorption. Subjects were also instructed to avoid consumption of any other multivitamin/mineral supplements during the 6-wk study period. At the time of randomization, no rowers had been regularly consuming dietary supplements.

Compliance with the Fe treatment as well as current health, menstrual status, and physical activity was assessed by daily training logs. Subjects were instructed to record the number of capsules they consumed daily in their log, even if they had consumed less than the prescribed amount per day. In addition, weekly capsule counts were conducted by the researcher.

Information regarding current dietary supplement use, health and menstrual status, usual physical activity, and eating habits and attitudes was obtained using questionnaires and a 7-d food diary. In addition to supplement compliance, current menstrual status, rowing training regimen, and leisure-time physical activity outside of rowing training was quantified daily via detailed training and activity records. The session-RPE method was used to quantify daily training load, as previously described (8). We validated the session-RPE method using our VAS format with the summated HR zone method during 2 wk of training on a separate sample of 13 female rowers and found a significant positive correlation between the two methods (r = 0.85, P < 0.001) (9). In the current study, rowers were classified as either “high” trainers or “low” trainers based on the session-RPE cutoff of 3200, which was the median (50th percentile) session-RPE of the sample, and this measure of training status was used as a covariate in multiple regression analyses as appropriate.

Iron status variables measured from nonfasting venous blood samples (antecubital venipuncture, into two evacuated tubes, EDTA, and serum separator) included Hgb, hematocrit, red blood cell count (Beckman Coulter, Fullerton, CA); sFer (Immulite 2000; Siemens Healthcare Diagnostics, Deerfield, IL); soluble transferrin receptor (sTfR; Ramco Laboratories, Stafford, TX); and alpha-1-acid glycoprotein (AGP, radial immunodiffusion plate; Kent Labs, Bellingham, WA). Total body iron (TBI, mg·kg−1) was calculated using the ratio of sTfR to sFer as described by Cook et al. (5). Iron status was analyzed immediately after blood sampling. Every effort was made to obtain baseline, midpoint, and end point samples at the same time of day to control for diurnal variation in any measurement of Fe status. To control for potential variation in the nonautomated sTfR assay conditions, both baseline and end point serum samples for the same subject were analyzed at the same time after the supplementation trial was completed. Rowers were classified as either IDNA (sFer < 20.0 μg·L−1), normal (sFer ≥ 20.0 μg·L−1), or anemic (Hgb < 12.0 g·dL−1). Rowers with anemia were notified of their status immediately after blood test results (within one week of analysis), referred to their respective campus health services for further instruction and monitoring, and excluded from further participation in the study. All laboratory assays were performed in the Human Metabolic Research Unit by a qualified technician at Cornell University.

Anthropometric and body composition measurements were determined before and after the 6-wk study. Body weight and height were measured with standard procedures and equipment (23). For athletes for whom it was accessible, body fat and fat-free mass was assessed via air-displacement plethysmography (BodPod; Life Measurement, Inc., Concord. CA). For all subjects, percent body fat was calculated from triceps, suprailiac, and thigh skinfold thickness (Lange, Cambridge, MD) (20) and bioelectrical impedance analysis (BIA; RJL Systems, BIA-101) (34). The Siri equation was used to calculate percent fat from body density. For those athletes without access to the BodPod (n = 17), an average of their percent body fat values calculated from BIA and skinfold was used, as previously described (8).

Physical performance testing methods

Physical fitness and endurance performance was measured before and after the 6-wk study using a rowing ergometer (Concept2, Morrisville, VT) equipped with a digital readout monitor (PM2), displaying work (W), stroke rating (SPM), distance (m), and time (min:sec). A computerized metabolic cart (TrueMax 2400; ParvoMedics, Salt Lake City, UT) was used to measure V˙O2 and other physiological measures during all testing. Concentrations of O2 and CO2 in expired air were analyzed with each breath, and respiratory volume (V˙E) was measured with a respiratory pneumotachograph (Fitness Instrument Technologies, Farmingdale, NY) through a two-way breathing valve (Hans Rudolph, Kansas City, MO).

Energy expenditure (EE) was assessed via indirect calorimetry during exercise testing using a standard protocol that monitors expired gases for V˙E, V˙O2, VCO2, and RER continuously throughout testing (39). HR (Polar FS2; Polar Electro, Inc., Lake Success, NY) was also continuously monitored throughout testing. Cadence (strokes per minute) and work rate (WR; resistance, W) were monitored and recorded every 30 s.

Blood lactate concentrations were determined by the Lactate Pro analyzer (FaCT Canada; Quesnel, British Columbia, Canada), the validity and accuracy of which have been previously reported (8).

Subjects were instructed to not consume food (including supplements or medications) or beverages (other than water) or to perform strenuous physical activity 2 h before testing. Most women came in for testing at 7 a.m., which was before any regular activity, and before their first meal of the day. No rower came into the laboratory for any exercise testing within 4 h or less of strenuous practice/activity. To control for the effects of dietary intake and hydration status, subjects were instructed to record all food and fluid intake 7 d before testing as well as the day of exercise testing. Menstrual status was recorded in the daily training log, and we found no effect of menstrual cycle (indication of menstrual period) on any variable examined in this study). Subjects had the opportunity to warm-up for at least 10 min before all testing.

Rowers performed two tests in the laboratory. The first was a pretest done to acclimate subjects to testing protocol and laboratory procedures as well as to establish a V˙O2peak and target WR prescription of 85% of their maximal WR (WRmax) to be used in the subsequent 4-km TT. V˙O2peak was determined by a modified version of the maximum aerobic power (MAP) test, which is a ramped protocol used by rowing coaches to assess training progress (25). Rowers’ MAP in split time was converted into watts (W = 2.8/pace per 500 m3), and the test began 100 W below the predicted maximum. Each stage of the test lasted 90s, with a 10-s “gear-up” period between each stage. Every 90 s, the rower was asked to increase her WR by 20 W, until she was no longer able to maintain the WR. This test was designed to last between 8 and 10 min. V˙O2peak was identified as the highest V˙O2 value achieved and was confirmed by at least one of the following: 1) V˙O2 increased by <150 mL·min−1 with an increase in WR, 2) RER > 1.10, or 3) HRmax was within 10 beats of age-predicted maximum (220 − age) (24). A 15-min cooldown period followed testing at a self-selected WR, and HR was monitored for 10 min posttest. Blood sampling for lactate was collected pretest and posttest, as well as at 5- and 10-min posttest. Complete test time was approximately 45 min (10–15 min to acclimate to equipment, 10 min for actual testing, and 15 min cooldown). Either the participant or the investigator had the ability to stop the test at any time for any reason, such as equipment malfunction, subject symptoms of severe fatigue, and so on. The pretest was only performed at baseline.

Endurance capacity was assessed at both baseline and end point by time to complete a 4-km TT, and was administered within 3 d of the baseline pretest. This test consisted of a 4-km ergometer row at a submaximal WR prescription (WRRx) of 85% of rowers’ V˙O2peak reached in the pretest. This WR was maintained for 3600 m of the test, and the rowers were then asked to sprint the final 400 m of the test to simulate on-water racing. The 4-km TT was designed to last approximately 20–25 min. Subjects received standardized verbal encouragement during testing. Capillary blood samples were obtained by finger or ear punctures immediately before testing, and every 1000 m during testing, as well as 5- and 10-min posttesting. Complete testing time was ∼60 min total (15 min to acclimate to equipment, 5-min warm-up, 25 min actual testing, and 15 min cooldown). The 4-km TT was performed at baseline and end point at the same WRRx. V˙O2peak and other reported performance outcome variables were obtained from the 4-km TT.


Work output (kcal·min−1) was calculated as follows:

EE input was calculated as follows:

Gross EF was calculated as (12) follows:

Deviation from the target WRRx of 85% of their pretest WRmax (W) was calculated as follows:

Statistical analysis

All data were analyzed using SPSS Statistics version 21.0 (Chicago, IL). Using a significance level P < 0.05 at 80% power, a sample size of 20 rowers per treatment group was calculated based on the effect sizes expected in physical performance with Fe supplementation (treatment vs placebo) based on previous research in IDNA (sFer < 16 μg·L−1), nonathletic women (17,40).

Data were examined to verify normality of distribution, and skewed distributions were log-transformed. Results are reported as mean ± SD. All initial analyses were performed on an as-treated basis to examine the effect of Fe supplementation on both Fe status and performance outcomes. Additional subgroup analyses were then conducted on those subjects classified as IDNA at baseline. Independent Student’s t-test was used to examine treatment group differences at baseline; characteristics differing between treatment groups (P < 0.05) were considered potential confounders and were included as covariates in subsequent regression models. Three-factor repeated-measures ANOVA was used to assess the effect of baseline sFer on Fe status and physical performance outcomes.

When significant group–time–baseline sFer interactions were observed, data were stratified by baseline sFer (< or >20 μg·L−1) to further examine relationships. Repeated-measures ANOVA was used to test group and time effects as well as group–time interactions for both Fe status and physical performance outcomes. An exact F-test statistic was used for all repeated-measures analyses. Mixed effects linear regression analysis, including both random and fixed effects, was used to assess the effects of Fe supplementation on change in Fe status and change in performance variables. School was treated as a random effect to control for unmeasured potentially confounding factors related to school differences in Fe status, training and performance. The effects of change in Fe status on change in performance outcomes were analyzed by including baseline and change terms in regression models. Pearson’s correlation was used to examine the relation between Fe status performance variables. Main effects were considered statistically significant at P < 0.05, and because interactions are important in exploratory analyses, these effects were considered statistically significant at P < 0.20.


Complete Sample

Subject characteristics

Thirty-one rowers completed the study protocol (see Fig. 1). Six subjects from the Fe group and three from the placebo group dropped out of the study because of personal reasons (n = 4), injury (n = 3), or illness (n = 2), all unrelated to the study. Compared with the 31 rowers who completed the 6-wk trial, the nine rowers who did not complete the study had slower finishing times for the simulated 4-km TT at baseline (19.4 ± 2.6 vs 17.7 ± 1.2 min, P = 0.008), but no other differences in baseline Fe status, body composition, training, or performance measures were observed.

The placebo and supplemented groups were of similar age (19.8 ± 1.1 and 19.7 ± 0.9 yr, respectively) and height (170.5 ± 7.7 and 169.0 ± 6.5 cm, respectively) and had similar years of rowing experience (3.7 ± 2.6 and 2.6 ± 1.3 yr, respectively). Twenty-eight subjects had completed a 7-d food record before the start of the study. There were no significant differences in baseline food intake reported between the two treatment groups (n = 14 Fe, 18 placebo; mean ± SD = 1938 ± 610 kcal and 20 ± 14.3 mg Fe per day). Reported sources of dietary Fe included fortified breads and cereals, beans and legumes, dark leafy greens, and animal tissue. There were no differences between treatment groups in minutes per day spent training (58.6 ± 29.1 vs 63.6 ± 19.1 min·d−1 in placebo and Fe, respectively) or any other training characteristic at baseline. There was no significant difference in history of supplement usage, and there was no difference in the number of days since last menstrual period at baseline between treatment groups (17.2 ± 10.2 and 22.9 ± 16.2 d in placebo and Fe groups, respectively). Body weight (67.7 ± 9.5 kg in placebo and 67.0 ± 6.6 kg in Fe group) and composition (25.8% ± 4.3% fat in placebo and 25.2% ± 5.1% fat in Fe group) did not differ between the two groups before or after the study. Rowers in both groups significantly increased their FFM by 1.2 ± 1.2 kg and decreased their percent body fat by 1.5% ± 1.8% after 6 wk of training (P < 0.001).

Supplementation compliance

Supplementation compliance was assessed from a training log in which subjects recorded the number of capsules consumed daily. The amount of capsules consumed tended to be greater in the placebo group (80 ± 20 vs 64 ± 34 in Fe group, P = 0.11). There were no differences between the two groups in treatment-associated symptoms, as no adverse events or symptoms related to the supplementation were reported during the study.

Iron status responses

Results of the blood analyses measured in the complete sample at baseline and end point (0 and 6 wk) of the study are presented in Table 1. No significant group differences in any Fe status measure were observed at baseline. After the 6-wk trial, although significant time effects were observed for Hgb and moderate time effects were observed for log sFer and TBI, there were no treatment group or group-by-time differences in Fe status, although all major indicators of body/tissue Fe stores changed in the predicted direction in the Fe-supplemented group. There were no significant group, time, or group-by-time effects for AGP, indicating that inflammation did not change markedly over time.

Iron status of the complete sample.

Additional plausibility analyses revealed a moderate relationship between supplemental Fe consumed (mg elemental Fe per kilogram of body weight) and change in TBI (mg·kg−1) in the entire sample (n = 14 iron, 16 placebo). After excluding one outlier who was a noncompliant rower in the Fe treatment group with a large increase in TBI and controlling for TBI at baseline (β = −0.58, P < 0.001), total supplemental Fe consumed (0 mg in the placebo group) was a moderately significant predictor of change in TBI (β = 0.07, P = 0.13, R2 = 0.56). Examining those rowers in the Fe group separately (n = 14), total supplemental Fe consumed remained a moderately significant predictor of change in TBI (β = 0.12, P = 0.18, R2 = 0.51).

Fitness responses

Results of the 4-km TT before and after 6-wk of training and treatment are shown in Table 2. After 6 wk of training, rowers in both treatment groups improved V˙O2peak (+0.2 ± 0.2 L·min−1, P < 0.001) and maximal workload (+22.0 ± 36.3 W, P = 0.002); however, there were no significant group differences or group-by-time effects. Although change in average gross efficiency during the entire 4-km TT was not significantly different after 6 wk of training between groups (P = 0.18), total EE was significantly lower in the Fe-supplemented group (P = 0.03 for group-by-time interaction). Additional plausibility analyses revealed that after controlling for baseline efficiency and training, there was a significant interaction between treatment group and dose consumed (β = 0.02, P = 0.15, R2 = 0.49). The more supplement that was consumed in the Fe group, the greater positive change in gross efficiency after 6 wk of training and treatment.

Physical performance measures during 4-km TT before and after 6 wk of training and treatment.

There were no significant group differences in baseline or end point pretest lactate levels. At baseline, after 1000 m, lactate increased ∼2- to 2.5-fold above pretest levels during the 4-km TT test in the Fe and placebo groups, respectively, and remained elevated for the duration of the 4-km test. After 6 wk of training, blood lactate concentrations at all time points during the end point 4-km TT were significantly lower compared with baseline in both treatment groups (P < 0.05). When blood lactate concentration was expressed as a percent of maximal lactate concentration, there was a significant negative correlation between the consumption of supplemental Fe (total mg) and the percentage of maximal lactate at the 1000-m mark (r = −0.39, P = 0.03). Further examining lactate expressed as a percentage of maximal lactate achieved, there were significant differences between the two treatment groups. Rowers supplemented with Fe had a slower rise in blood lactate during the first half of the 4-km TT (1000 m: 40.3% ± 9.6% in Fe vs 50.6% ± 5.1% in placebo, P = 0.001; 2000 m: 45.4% ± 9.8% in Fe vs 55.1 ± 5.4 in placebo, P = 0.006). Rowers in the Fe group also showed a faster recovery 5 min after completing the TT compared with the placebo group (79.3% ± 6.4% in Fe vs 86.9 ± 5.4 in placebo, P = 0.001).

Analyses stratified by baseline sFer

We further explored the relationship between baseline sFer status and responses to treatment using stratified analyses. We used the original sFer screening cutoff of 20 μg·L−1 to maintain adequate sample size within each stratum. The lower stratum (sFer < 20 μg·L−1) contained 16 subjects (n = 8 placebo, 8 Fe group). The upper stratum (sFer > 20 μg·L−1) contained 15 subjects (n = 8 placebo, 7 Fe group).

Baseline subject characteristics

Baseline subject characteristics were compared across sFer strata to ensure that only Fe status differed among those who began the study with low or high sFer concentrations. No significant differences were observed across the lower strata (sFer < 20 μg·L−1) for baseline anthropometry, physical performance, or training. In the upper strata (sFer > 20 μg·L−1), rowers in the Fe group had significantly faster 4-km TT times at baseline (P = 0.02). There were no other differences between the treatment groups.

Anthropometry across sFer strata

As shown in the original analysis, stratifying the data by sFer concentration showed significant effects of time for both FFM (P < 0.01) and percent body fat (P < 0.05) among both groups of subjects.

Iron responses across sFer strata

As expected, significant differences were observed in sFer across sFer strata (see Table, Supplemental Digital Content, Additional data by Fe status strata (sFer < 20 μg·L−1 as cutoff)]. There were no other significant differences among Fe status indicators. Among these subjects with sFer < 20 μg·L−1, significant time effects were observed for sFer (P = 0.005) and TBI (P = 0.04), but no treatment group or treatment group-by-time effects were observed. In the higher strata, there were no significant effects observed. Additional plausibility analyses of the subgroup of IDNA rowers (n = 7 iron, 8 placebo) were performed, less the same outlier who was a noncompliant rower in the Fe treatment group with a large increase in TBI. In the subgroup of rowers who were IDNA at baseline, after controlling for TBI at baseline (β = −0.68, P = 0.002), total supplemental Fe consumed was a moderately significant predictor of change in TBI (β = 0.11, P = 0.10, R2 = 0.59). Examining those IDNA rowers in the Fe group separately (n = 6), total supplemental iron consumed became a significant predictor of change in TBI (β = 0.23, P = 0.05, R2 = 0.79).

Additional interaction analyses were used to examine whether subjects with the poorest Fe status at baseline (greatest potential to benefit from supplementation) experienced the greatest improvements in Fe status. The interaction between treatment group and baseline Fe status (sFer) at baseline was significant, such that rowers with the lowest sFer at baseline showed the greatest improvement in Fe stores in the Fe group compared with the placebo group (β = −0.37, P = 0.07; for the treatment group-by-baseline sFer interaction term, see Fig. 2). There was also a modest interaction effect of treatment-by-baseline TBI (β = −0.30, P = 0.18, figure not shown).

Relationship between baseline sFer and change in sFer with Fe supplementation.

Physical performance across sFer strata

Physical performance results for subjects across sFer strata are presented in Table 3. Among all subjects, significant time effects were observed for V˙O2peak, but no group-by-time effects were observed. There were significant differences observed in EE and energetic EF during the 4-km TT across sFer strata. In the lower sFer strata, there were significant group-by-time effects observed for EE and energetic EF. Subjects with the poorest Fe status at baseline showed the greatest decrease in EE and thus greatest increase in energetic EF during the 4-km TT in the Fe group compared with the placebo group. There were no effects of supplementation observed in the higher strata. Additional plausibility analyses (adjusted for change in log sFer and supplemental Fe consumed) performed revealed that rowers who improved their sFer status and consumed more than 600 mg of the supplemental Fe during the 6-wk study (n = 5 in Fe treatment group) increased their gross EF more than those who consumed less than 600 mg of the supplement (n = 11, three of those in Fe treatment group; β = −0.006, P = 0.06 for the interaction).

Physical performance of rowers with baseline sFer < or >20 μg·L−1.

The change in EE during the 4-km TT divided by 1000-m bouts is shown in Figure 3. The significant overall EE effect (P = 0.01) between treatment groups observed in the lower sFer strata can be seen at the 1000 (P = 0.04), 2000 (P = 0.12), and 4000 m (P = 0.17) marks of the TT (group-by-time effects). No group-by-time effects were observed during any of the 1000-m bouts in the upper strata.

Mean ± SEM changes in EE (kcal) during each 1000-m bout of the 4-km time trial (TT) from baseline to posttreatment in (A) subjects with normal baseline sFer status [ placebo group (n = 8), iron group (n = 7)] and (B) IDNA subjects at baseline [ placebo group (n = 8), iron group (n= 8)]. *Significant group–time interaction, P < 0.05 (two-factor repeated-measures ANOVA). **Significant group–time interaction, P < 0.20 (two-factor repeated-measures ANOVA). A significant group–baseline sFer interaction (P < 0.20) for change in EE at 3000, 3600, and 4000 m prompted the stratified analysis in this figure.


The purpose of this RCT was to examine the effects of Fe supplementation on changes in Fe status and endurance performance in nonanemic female rowers who were training for their competitive season. sFer and Hgb were used to identify those to include in the supplementation trial, as done previously (17,40). In addition, we measured sTfR and calculated TBI to differentiate those with low total Fe stores from those with low tissue (functional) Fe (2,3,41). This study showed that 6 wk of Fe supplementation during training did lead to a greater improvement in Fe stores (sFer, TBI) compared with placebo, after controlling for baseline Fe stores, and that the Fe status of rowers with poorest Fe status at baseline benefitted the most from supplementation. These findings are important to female endurance athletes during training as well as all physically active women.

Other researchers have reported a deterioration of Fe status with moderate to high levels of physical activity. These decrements in Fe status have been shown to affect physical performance, especially in those women who begin training with poor Fe status (2,3,26), and Fe supplementation has been shown to prevent a deterioration in sFer during periods of moderate to heavy physical training (26). In the present study, we did observe a positive effect of Fe supplementation on sFer and TBI, especially in those rowers with low sFer at baseline. We did not, however, observe this positive effect on sTfR.

The rowers in our study were not anemic; thus,there were few with sTfR > 8.0 mg·L−1 (n = 6 at baseline). At the end of the study, sTfR was not significantly different between treatment groups (P = 0.20), although rowers in the placebo group tended to have higher sTfR compared with the Fe group. Researchers have shown that sTfR is affected by muscle growth (2,28), and in the current study, after 6 wk of training, all rowers increased their FFM by ∼1.2 kg. Although we observed no correlation between change in FFM and change in sTfR, it is possible that this factor may have played a role in diminishing the effects of supplementation on this measure of Fe status.

It is possible that rowers’ training may have decreased the response to Fe supplementation. Endurance exercise does increase body Fe turnover and may increase basal Fe loss (sweat, urine, etc.), and the inflammation due to heavy training may affect hepcidin and, thus, downregulate Fe absorption (13). It has been suggested that physical training may increase the estimated average requirement (EAR) for Fe in female athletes by 30%–70%, from 8 mg to 10–14 mg·d−1 (19). The level of Fe supplementation prescribed in this study was more than the recommended daily allowance for women (RDA, 18 mg·d−1) and should have been adequate to improve Fe stores if taken as directed, especially given that our subjects were consuming an adequate amount of dietary Fe at the beginning of the study. In previous studies, 100 mg FeSO4 (∼20 mg·d−1 elemental Fe) over the course of 6–8 wk was sufficient to improve Fe stores (sFer) in compliant, nonathletic women (17,40). In our sample, although rowers in the Fe group consumed on average ∼60% of the prescribed Fe dose (actual consumption ∼15 mg elemental iron per day), we still saw an improvement in Fe status with supplementation, and no side effects were reported with this dose. FeSO4 is highly absorbable (15%–20%) (6) and is three to four times more absorbable than ferric preparations, and its bioavailability is comparable with supplements containing heme Fe (35).

Although no side effects of the Fe supplement were reported, rowers in the Fe group were less compliant throughout the trial. Diminished compliance in the Fe group (more so than the placebo group) may have been due to unreported GI or other unreported side effects. During the trial, at weekly log collection and pill counts and again upon debriefing, forgetfulness was the most common reason cited for noncompliance by the rowers in both treatment groups. These rowers were not habitual supplement users before the study; thus, consuming a supplement twice daily may have been difficult to make part of their daily routines. In the current study, compliance did not peak until around the midpoint of the trial, before which rowers were only consuming ∼30% of the prescribed Fe dose. If we had given a larger dose of Fe to compensate for the less frequent consumption, we may have seen a greater improvement in Fe stores. Alternatively, at the same consumption rate, we may have seen greater improvement in Fe stores after 9 wk of supplementation compared with 6 wk. More infrequent (weekly) supplementation with 60 mg of Fe for 7 months has been shown to be as effective as daily supplementation in increasing sFer in IDNA premenopausal women (37), and several researchers report improved Fe status and better compliance with this mode of supplementation in pregnant women and children (1). Future studies should focus on ways to improve supplement compliance in training female athletes to confer maximum benefit to athletes’ Fe status.

A limitation of our subgroup analysis is the use of the sFer cutoff of 20.0 μg·L−1 to identify rowers as IDNA. Any misclassification of baseline Fe status in these subjects could have diluted the differences in change in Fe status between the two treatment groups. Although sFer is the most common index of Fe stores reflecting Fe stored in the liver, it is an acute-phase protein and can be elevated in an inflammatory state (e.g., infection and postexercise), potentially masking ID (13). However, an inflammatory marker such as C-reactive protein (CRP) or alpha-1-acid glycoprotein (AGP) can partially rule out falsely elevated sFer (36). In our study, although there were no differences in AGP between the two treatment groups at the beginning of the study, inflammation did not change over the course of the trial, and only one subject had AGP levels above the threshold indicative of inflammation (140 mg·dL−1). Some IDNA subjects may still have been misclassified as normal.

In previous cross-sectional analyses, we reported that at the beginning of a training season, IDNA rowers had slower 4-km TT times and lower V˙O2peak compared with rowers with normal Fe status, and that these effects of poor Fe status on performance were more pronounced in rowers who trained less hard compared with those who trained harder (8). In the current RCT analyses, time to complete the 4-km TT and V˙O2peak were unaffected by Fe supplementation. We did, however, observe significant treatment effects for EE and energetic EF in rowers who were IDNA at baseline. This means that after being supplemented with Fe for 6 wk, IDNA rowers were able to perform the same workload at a lower energy cost (lower level of physical exertion = more energetically efficient). The total EE difference between the Fe and the placebo groups in the IDNA strata was 15 kcal over the 4-km TT (∼20 min). This is equivalent to approximately 45 kcal·h−1 and would be at least a 135-kcal difference during a 3-h training session. After the 6-wk trial, the IDNA rowers supplemented with Fe showed greater improvements in EF (+1.3% compared with the placebo group) during the 4-km TT. From the analysis of the 1000-m bouts of the 4-km TT, the improvements in EE and energetic EF are most prominently seen during the first half as well as final sprint of the TT in the IDNA subgroup.

These findings are consistent with similar Fe supplementation studies of nonanemic, nonathletic women. Researchers have found that O2 consumption (as %V˙O2max) during an endurance test was significantly less (−3%) after Fe supplementation and significantly greater (+3%) in a placebo group (22). Zhu and Haas (40) showed that after 8 wk of Fe supplementation, nonathletic women increased their EF by decreasing their EE by 5.1% (P = 0.016) compared with women in the placebo group, and that this treatment effect on %V˙O2peak was mediated by a change in Hgb.

Hinton et al. (17) found that after a 4-wk training program imbedded in a 6-wk Fe supplementation trial, although both placebo and Fe treatment groups increased their work EF (through training), there were no significant differences in EF between the treatment groups. The Fe group, however, decreased their O2 consumption by 5% (as a %V˙O2peak) during the last 5 km of 15-km TT. Most recently, Hinton et al. found that after 6 wk of Fe supplementation, posttrial work EF in recreational athletes was significantly increased (+1.1%) compared with placebo (+0.7%) (18).

After 6 wk of training, all rowers improved their lactate response during the 4-km TT. The Fe-supplemented group, though, had much slower rise from pretest lactate (10% lower, as a percentage of maximal lactate) during the first two phases of the TT, as well as a faster recovery 5 min after completing the test compared with the placebo group. Although many researchers have found no effect of oral Fe supplementation on lactate concentration during exercise in IDNA women (10,11,17,21), results from the current study are similar to those of Zhu and Haas (40) who found that IDNA women supplemented with Fe showed a slower rise in lactate concentration during the first leg of a 15-km TT (5-km mark) on a cycle ergometer as well as an inverse association between lactate concentration at 5 km and Hgb, even in marginal ID. Even in the absence of frank anemia (Hgb < 12.0 g·dL−1), impaired O2 transport capacity due to IDNA appears to affect lactate metabolism, resulting in impaired oxidative metabolism and, ultimately, increased reliance on anaerobic metabolism to produce energy (greater lactate production at an earlier stage of exercise). In a state of IDNA, lactate metabolism may be directly affected, resulting in the prevention or slowing of lactate clearance (14,33).

The 4-km TT test protocol was designed to give rowers an incentive to finish “the race” as they would on the water, similar to the manner in which they train. This familiar format should have minimized the influence of motivational status on test time and level of exertion that is a common problem in time-to-exhaustion protocols. However, there was still a subjective component of the TT protocol, as it was up to each rower to control her WR on the ergometer at all times, so despite the given WRRx (85% of max), it was up to each rower to maintain that WR using the monitor on the ergometer. However, deviation from WRRx was not significantly different between the two treatment groups at baseline or end point. In addition, the difference in energetic EF should not have been due to differences in psychological factors between the two treatment groups, as motivation scores measured throughout the trial were not significantly different between the two groups at baseline and end point. Furthermore, in these studies, we measured gross EF, which does not account for energy expended during an unloaded (0 W) condition (termed net energetic EF). Measuring EE during an unloaded bout would have enabled us to calculate net EF, which many have been more sensitive to changes in Fe status, and strengthened the relationships between Fe status and performance.

Despite our limitations, this RCT showed that after controlling for baseline sFer, Fe supplementation improved Fe stores during training as well as decreased EE and increased EF during a 4-km TT. Also, we observed a slowed lactate response during the early phases of the 4-km TT and a faster recovery post-TT with Fe supplementation, indicating that IDNA affects lactate metabolism in the absence of anemia. Taken together, these results indicate that that IDNA increases rowers’ exertion and energy cost to do the same load of work, and that Fe supplementation enhanced rowers’ adaptation to training, adding to the growing body of evidence that Fe supplementation improves the Fe status and physical performance of active women. These results are important for female endurance athletes whose dietary patterns and physical training levels increase their risk of IDNA and suggest that Fe supplementation may maximize the benefits of endurance training. Future studies should focus on implementation of Fe status screening programs for female athletes in training, as well as ways to improve supplement compliance in training female athletes to confer maximum benefit to athletes’ Fe status.

The authors are grateful to the coaches and rowers who participated in this study, as well as to our undergraduate research assistants. Special thanks to Victoria Simon (Human Metabolic Research Unit) for her technical laboratory assistance and analysis of blood samples.

This study was funded by grants from the American Dietetic Association Foundation and Cornell University’s Division of Nutritional Sciences.

Authors declare no conflicts of interest.

The results do not constitute endorsement by the American College of Sports Medicine.


1. Beard J. Weekly iron intervention: the case for intermittent iron supplementation. Am J Clin Nutr. 1998; 68 (2): 209–12.
2. Brownlie T IV, Utermohlen V, Hinton PS, Giordano C, Haas JD. Marginal iron deficiency without anemia impairs aerobic adaptation among previously untrained women. Am J Clin Nutr. 2002; 75 (4): 734–42.
3. Brownlie T IV, Utermohlen V, Hinton P, Haas J. Tissue iron deficiency without anemia impairs adaptation in endurance capacity after aerobic training in previously untrained women. Am J Clin Nutr. 2004; 79 (3): 437–43.
4. Cogswell ME, Looker AC, Pfeiffer CM, et al. Assessment of iron deficiency in US preschool children and nonpregnant females of childbearing age: National Health and Nutrition Examination Survey 2003–2006. Am J Clin Nutr. 2009; 89 (5): 1334–42.
5. Cook JD, Flowers CH, Skikne BS. The quantitative assessment of body iron. Blood. 2003; 101 (9): 3359–63.
6. Davila-Hicks P, Theil EC, Lönnerdal B. Iron in ferritin or in salts (ferrous sulfate) is equally bioavailable in nonanemic women. Am J Clin Nutr. 2004; 80 (4): 936–40.
7. DellaValle DM, Haas JD. Impact of iron depletion without anemia on performance in trained endurance athletes at the beginning of a training season: a study of female collegiate rowers in central NY state. Int J Sp Nutr Ex Metab. 2011; 21 (6): 501–6.
8. DellaValle DM, Haas JD. Iron status is associated with endurance performance and training in female rowers. Med Sci Sports Exerc. 2012; 44 (8): 1552–9.
9. DellaValle DM, Haas JD. Quantification of training load and intensity in female collegiate rowers: validation of a daily assessment tool. J Strength & Cond Res. 2013; 27 (2): 540–8.
10. Fogelholm M, Jaakkola L, Lampisjarvi T. Effects of iron supplementation in female athletes with low serum ferritin concentration. Int J Sports Med. 1992; 13 (2): 158–62.
11. Friedmann B, Weller E, Mairbaurl H, Bartsch P. Effects of iron repletion on blood volume and performance capacity in young athletes. Med Sci Sports Exerc. 2001; 33 (5): 741–6.
12. Gaesser GA, Brooks GA. Muscular efficiency during steady-rate exercise: effects of speed and work rate. J Appl Physiol. 1975; 38 (6): 1132–9.
13. Gaffney-Stomberg E, McClung JP. Inflammation and diminished iron status: mechanisms and functional outcomes. Curr Opin Clin Nutr Metab Care. 2012; 15 (6): 605–13.
14. Gregg SG, Willis WT, Brooks GA. Interactive effects of anemia and muscle oxidative capacity on exercise endurance. J Appl Physiol. 1989; 67 (2): 765–70.
15. Haas J, Brownlie T IV. Iron deficiency and reduced work capacity: a critical review of the research to determine a causal relationship. J Nutr. 2001; 131 (2): 676–90S.
16. Hassapidou M, Manstrantoni A. Dietary intakes of elite female athletes in Greece. J Hum Nutr Diet. 2001; 14 (5): 391–6.
17. Hinton PS, Giordano C, Brownlie T, Haas JD. Iron supplementation improves endurance after training in iron-depleted, nonanemic women. J Appl Physiol. 2000; 88 (3): 1103–11.
18. Hinton P, Sinclair L. Iron supplementation maintains ventilatory threshold and improves energetic efficiency in iron-deficient nonanemic athletes. Eur J Clin Nutr. 2007; 61 (1): 30–9.
19. Institute of Medicine (IOM). Dietary Reference Intakes (DRIs) for (Micronutrients). Washington (DC): National Academy Press; 2001. p. 773.
20. Jackson A, Pollock M, Ward A. Generalized equations for predicting body density of women. Med Sci Sports Exerc. 1980; 12 (3): 175–81.
21. Klingshirn L, Pate R, Bourque S, Davis J, Sargent R. Effect of iron supplementation on endurance capacity in iron-depleted female runners. Med Sci Sports Exerc. 1992; 24 (7): 819–24.
22. LaManca J, Haymes E. Effects of iron repletion on V˙O2max, endurance, and blood lactate in women. Med Sci Sports Exerc. 1993; 25 (12): 1386–92.
23. Lohman T, Roche A, Martorell R. Anthropometric Standardization Reference Manual. Champaign, IL: Human Kinetics; 1988. p. 177.
24. MacDougall J, Wenger H, Green H. Physiological Testing of the High-Performance Athlete. Champaign, IL: Human Kinetics; 1991. p. 432.
25. Maestu J, Jurimae J, Jurimae T. Monitoring of performance and training in rowing. Sports Med. 2005; 35 (7): 597–617.
26. McClung JP, Karl JP, Cable SJ, et al. Randomized, double-blind, placebo-controlled trial of iron supplementation in female soldiers during military training: effects on iron status, physical performance, and mood. Am J Clin Nutr. 2009; 90 (1): 124–31.
27. McClung JP, Marchitelli LJ, Friedl KE, Young AJ. Prevalence of iron deficiency and iron deficiency anemia among three populations of female military personnel in the US Army. J Am Coll Nutr. 2006; 25 (1): 64–9.
28. Murray-Kolb L, Beard J, Joseph L, Davey S, Evans W, Campbell W. Resistance training affects iron status in older men and women. Int J Sport Nutr Exerc Metab. 2001; 11 (3): 287–98.
29. Peeling P, Dawson B, Goodman C, et al. Training surface and intensity: inflammation, hemolysis, and hepcidin expression. Med Sci Sports Exerc. 2009; 41 (5): 1138–45.
30. Rowland TW, Deisroth MB, Green GM, Kelleher JF. The effect of iron therapy on the exercise capacity of nonanemic iron-deficient adolescent runners. Am J Dis Child. 1988; 142 (2): 165–9.
31. Sandström G, Börjesson M, Rödjer S. Iron deficiency in adolescent female athletes—is iron status affected by regular sporting activity? Clin J of Sp Med. 2012; 22 (6): 495–500.
32. Schoene R, Escourrou P, Robertson H, Nilson K, Parsons J, Smith N. Iron repletion decreases maximal exercise lactate concentrations in female athletes with minimal iron-deficiency anemia. J Lab Clin Med. 1983; 102 (2): 306–12.
33. Stanley W, Mazer C, Neese R, et al. Increased lactate appearance and reduced clearance during hypoxia in dogs. Horm Metab Res. 1990; 22 (9): 478–84.
34. Sun SS, Chumlea WC, Heymsfield SB, et al. Development of bioelectrical impedance analysis prediction equations for body composition with the use of a multicomponent model for use in epidemiologic surveys. Am J Clin Nutr. 2003; 77 (2): 331–40.
35. Swain JH, Johnson LK, Hunt JR. Electrolytic iron or ferrous sulfate increase body iron in women with moderate to low iron stores. J Nutr. 2007; 137 (3): 620–7.
36. Thurnham DI, McCabe LD, Haldar S, Wieringa FT, Northrop-Clewes CA, McCabe GP. Adjusting plasma ferritin concentrations to remove the effects of subclinical inflammation in the assessment of iron deficiency: a meta-analysis. Am J Clin Nutr. 2010; 92 (3): 546–55.
37. Viteri FE, Ali F, Tujague J. Long-term weekly iron supplementation improves and sustains nonpregnant women’s iron status as well or better than currently recommended short-term daily supplementation. J Nutr. 1999; 129 (11): 2013–20.
38. Weaver CM, Rajaram S. Exercise and iron status. J Nutr. 1992; 122 (3 Suppl): 782–7.
39. Weir JB. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol. 1949; 109 (1–2): 1–9.
40. Zhu YI, Haas JD. Altered metabolic response of iron-depleted nonanemic women during a 15-km time trial. J Appl Physiol. 1998; 84 (5): 1768–75.
41. Zhu Y, Haas J. Response of serum transferrin receptor to iron supplementation in iron-depleted, nonanemic women. Am J Clin Nutr. 1998; 67 (2): 271–5.


© 2014 American College of Sports Medicine