Women between the ages of 18 and 23 yr were recruited from the Oregon State University rowing team and the general student body. Exclusion criteria included: 1) the existence of conditions known to affect bone metabolism (e.g., uncontrolled diabetes); 2) injuries that would inhibit rowing performance; and 3) medications known to affect bone (e.g., steroid-derived asthma medication). Of the 43 athletes on the Oregon State University women’s rowing team, one subject was excluded due to a preexisting back injury. During the study, seven rowers discontinued the intervention when they left the team for personal reasons, and thus the team evaluated in this study was comprised of 19 first-year novice rowers (aged 19.5 ± 0.8 yr) with 3 months of rowing experience and 16 experienced rowers (aged 21.2 ± 1.2 yr) with 26 ± 10 months of rowing experience. For comparison, we used data from a control group recruited for a previous study in our laboratory (8). The 14 nonrowing controls (aged 19.2 ± 1.6 yr) were normally active college women, and their spine measurements were assessed with the same spine protocol as the rowers (DXA, Hologic QDR/1000-W, Waltham, MA); however, the time between scans for rowers and controls was 6 and 7 months, respectively. The Oregon State University Institutional Review Board approved the study, and all subjects gave written informed consent before participation.
During the observation period, all rowers participated in eight training sessions per week. Of the eight sessions, six were spent rowing on the water or on the rowing ergometer and two were spent cross-training that consisted of running, weight training, and stretching. The duration of each training session was approximately 90 min for rowing and 45 min for cross-training and, thus, the majority of total training time (83%) was spent rowing. On average, during each rowing session, the athletes took 1000–1200 repetitions (strokes) per session for a total of 6000 repetitions per week, regardless of experience level. During the observational period, there were 5158 potential rowing sessions, of which 120 were missed due to absence from practice and thus compliance was 97.6%.
All subjects completed the Oregon State University Bone Research Laboratory Health History Questionnaire. For the rowers, caloric consumption and calcium intake per day were assessed based on average food intake over the previous year by the Block Food Frequency Questionnaire, a previously validated frequency-amount questionnaire used by the National Cancer Institute (1). Controls completed 3-d diet records. Rowers and controls were eumenorrheic (10–12 menstrual cycles per year) and reported having regular cycles during the entire observational period. Five rowers (three experienced and two novices) but no controls reported taking birth control pills during the study. Mean calcium intake for rowers met the recommended intake by the National Institutes of Health Consensus Conference (7) of 1200 mg·d−1 for women of this age, but that of the control group was less (Table 1).
Bone mass measurements.
For rowers, BMD was assessed at the end of November and early June, whereas controls were assessed at the end of October and May. Spine BMD (g·cm−2) was measured by dual-energy x-ray absorptiometry (DXA) (QDR-1000/W, Hologic Inc.). The in-house coefficient of variation for the spine is ≤ 1.0%.
Rowing performance was assessed on a Concept 2 rowing ergometer (Concept 2, Model C, Morrisville, VT). All rowers performed timed 2000- and 6000-m tests once per month on separate days in January, February, and March.
Means and standard deviations were computed by standard statistical techniques. Before analysis, data were screened for normality, linearity, equal variances, and homogeneity of regression slopes for the covariates. In analysis of variance on all baseline variables, experienced rowers were significantly older than novices and controls, and both rowing groups had higher body mass index (BMI) than controls (Table 1). To evaluate the 6-month changes in spine BMD between groups, we calculated a spine BMD difference score [spine BMD difference score = post − pre in g·cm−2], we used ANCOVA and adjusted for BMI and age due to baseline differences (Table 1). In addition, we included baseline spine BMD as a covariate because of its potential to influence the response of bone to overload. To evaluate differences between novice and experienced rowers on the 2000- and 6000-m timed ergometer tests, we used factorial repeated measures ANOVA. All statistical analyses were performed with SPSS for Windows software, version 9.0 (SPSS, Inc., Chicago, IL).
The general linear model adjusted for BMI, age, and initial BMD revealed a significant difference in spine BMD difference score between groups (P = 0.03). In pair-wise comparisons, experienced rowers demonstrated a greater increase in spine BMD than novice rowers (P = 0.01, Fig. 1). Spine BMD difference scores were similar between the controls and the experienced (P = 0.37) and novice rowers (P = 0.11). In factorial repeated measures ANOVA, the 2000- and 6000-m ergometer times for the experienced rowers were significantly different than the novice rowers in each month (P = 0.0001, Figs. 2 and 3). The experienced athletes demonstrated better performance than the novice rowers for the 2000- and 6000-m ergometer tests at all time points (January, February, and March).
Our primary aim was to determine whether spine BMD differs in novice and experienced women rowers after 6 months of rowing training. We report that lumbar BMD increased significantly more in experienced rowers than the novice rowers and controls. Specifically, experienced rowers demonstrated a 2.14% increase in spine bone density, whereas the changes observed in the novice rowers and the controls were not greater than the in-house precision error for DXA spine measurements.
This study has several strengths. First, we compared the response of the spine to rowing in two groups of female athletes who were very similar. Experienced and novice rowers had comparable height and weight, and differed only by age. Furthermore, their normative values for lumbar spine bone density were 103% and 105% for experienced and novice, respectively, indicating that the group means were 2–5% higher than values for young normal women. All rowers participated in the same type and duration of training, took a similar number of strokes (repetitions) each session, and participated in the same day-to-day workouts. Due to the time required for team membership, participation in outside activities known to influence bone mass was minimal. In addition, due to the study design, compliance was high at 97.6%. Other prospective studies have reported low compliance rates and also participation in outside activity as confounding variables (12,16). Also, the conclusion of the observational period coincided with the end of the competitive racing season and thus included a progressive overload from training as team members prepared for the conference championships.
It is important to note limitations. Due to the study design, participation was limited to members of the Oregon State women’s rowing team; thus, it was not a randomized exercise intervention. However, our results provide a first step in developing a model to study the effects of rowing training as a strategy to build vertebral BMD in adults. Second, the control group had been recruited for an earlier study conducted in our laboratory (8) and thus were not measured over the same observational period as the rowers. Because there was a difference in BMI at baseline between the rowing groups and control group, we controlled for this difference by adjusting for initial BMI in the analysis. Third, we did not quantify the lumbar compressive or shear forces in the rowing groups, nor did we count the exact number of repetitions required to complete the ergometer tests. However, the rowers took an average of 28–30 strokes per minute for the 2000-m test and 26–28 for the 6000-m test. In addition, the novice and experienced rowers did not differ significantly in height and because of this presumably had a similar stroke length. Given the same number of strokes and the same length of stroke, the only way to cover the same distance faster is to apply more force. Because the experienced rowers were significantly faster on all tests, it follows that they also generated more force than the novice rowers. Although there is no question that efficiency and higher aerobic capacity (and higher anaerobic threshold) contribute to increased power output, in the case of our comparison, given similar strokes per session and strokes per minute in the timed ergometry tests, higher force production is what accounted for the difference in performance times. Lastly, we had low sample size and thus reduced statistical power (power = 0.65). This likely explains the lack of significant difference in spine changes between experienced rowers and controls.
Novice rowers had initial spine BMD that was 3.9% higher than experienced rowers. Although we cannot explain the higher BMD in novices, this difference was not significant between groups at baseline. Nevertheless, we covaried for initial BMD in our statistical analysis and the adjusted model was significant. Perhaps, if forces had been higher in the novices, they would have demonstrated increased spine BMD. In our previous work (13), we reported that gymnasts increase spine BMD over their 8-month training season despite high initial values. In more recent work (11), we observed this phenomenon over two separate training seasons. Despite initial spine BMD in gymnasts (year 1 = 1.143 and year 2 = 1.163 g·cm−2) that was similar to or higher than novices we observed in our current study, gymnasts responded to training over two separate 8-month training seasons by increasing spine BMD to 1.181 and 1.206 g·cm−2, respectively. These gains of 3.5% and 3.7% suggest that even with high initial values, bone responds to high force by increasing BMD.
Cross-sectional data report that rowers have higher lumbar BMD than nonrowing controls. Morris et al. (5) compared BMD values of 14 female rowers (aged 19.7 ± 1.6 yr) with 14 female matched controls. All rowers had been training for a minimum of 3 yr and were rowing at least 5× wk−1. They found that the rowers had greater lumbar spine BMD but were not different than the controls at the other sites measured. Smith and Rutherford (10) compared total body and spine BMD in male athletes with controls. The cohort was comprised of 12 rowers (aged 20.8 ± 2.4 yr) who trained on average 25 h·wk−1, 8 triathletes (aged 29.1 ± 5.4 yr) who trained 20 h·wk−1, and 13 nonexercising controls (aged 21.7 ± 3.6 yr). Results revealed that the rowers had higher BMD at the spine and total body than both the triathletes and the controls. Wolman et al. (17) compared bone density in women athletes and found that, despite a similar prevalence of menstrual irregularities, national team lightweight rowers (aged 25.1 ± 3.5 yr) had significantly higher lumbar BMD than both elite runners (aged 25.9 ± 2.7 yr) and professional dancers (aged 22.7 ± 3.8 yr). Although these studies support that rowing targets the spine, they do not provide information relative to load magnitude and the effect of rowing on spine BMD over time.
Two longitudinal studies have reported the benefits of rowing on the lumbar spine BMD and our study corroborates these findings (2,5). Morris et al. (5) showed that, in adolescent girls aged 14–15 yr, 18 months of rowing training resulted in a significant 6.2% increase in lumbar spine BMD compared with a 1.1% increase in the control group. In that study, the girls participated in three to five on-water rowing sessions and three land-based training sessions per week. Cohen et al. (2) showed a 2.9% increase in lumbar BMD in 17 male novice college oarsmen after 7 months of rowing compared with eight aged-match controls. Training included 8 h of rowing, 1 h of weight training, and 1 h of running per week. It is likely that the collegiate men were stronger at baseline than the novice women in our study and, thus, able to generate greater forces at the spine earlier in their intervention.
In rowing, to maximize the propulsive effect of the oar, the back extensors must transfer the forces generated by the legs to the oar handle and, thus, hands. Morris et al. (6) demonstrated that rowing produced spinal compression forces that vary as a function of the force applied to the handle during a rowing cycle. Based on our direct measures in pilot experiments and reflecting the findings of Morris et al. (6) and Saul and Hosea (9), the forces in the spine are in dynamic equilibrium with both the forces at the hand (oar) and with those produced in the legs. Spinal forces can thus be determined from those produced by either the arms or the legs. In our case, because we had data on power output at the hands and were interested in differences that might explain the differential increase in spinal BMD, it was natural to examine power differences measured at the hands to support the changes in BMD at the spine in experienced rowers. Toward this end, we assessed the power differences measured at the hands between novice and experienced rowers by analyzing the results from standard race simulation rowing ergometer tests (Concept 2, Model C). As part of normal training, all rowers were tested in January, February, and March over two different distances performed on different days (Figs. 2 and 3). Given that power = force × time/distance, the measured differences in power output and the known differences in ergometer times for both experienced and novice rowers indicate that the experienced rowers generated on average 15% more hand force at the beginning of the test period for both the 2000- and 6000-m trials and about 10% more in March. Because compressive forces at the spine are directly related to hand forces, we conclude that the observed increase in spine BMD in experienced rowers was a consequence of the increased forces on their spine.
Our results support the theory that force is a key variable in osteogenesis. These results may provide preliminary data from which to develop an exercise prescription for building spine bone density and reducing vertebral osteoporosis.
The authors wish to thank the Oregon State University Women’s Rowing Team for participating in this study and gratefully acknowledge the support provided by the Oregon State University Bone Research Laboratory Clinical Program. We also thank Dr. Daniel Williams, University of Utah, for his statistical assistance, and Dr. Wilson C. Hayes, Oregon State University, for his biomechanics expertise.
1. Block, G. Human dietary assessment: methods and issues. Prev. Med. 18: 653–660, 1989.
2. Cohen, B., P. J. Millet, B. Mist, M. A. Laskey, and N. Rushton. Effect of exercise training programme on bone mineral density in novice college rowers. Br. J. Sports Med. 29: 85–88, 1995.
3. Cummings, S. R., D. M. Black, and M. C. Nevitt. Bone density at various sites for prediction of hip fractures. Lancet 341: 72–75, 1993.
4. Marcus, R. Editorial. Exercise: moving in the right direction. J. Bone Miner. Res. 13: 1793–1796, 1998.
5. Morris, F. L., W. R. Payne, and J. D. Wark. The impact of intense training on endogenous estrogen and progesterone concentrations and bone mineral acquisition in adolescent rowers. Osteoporos. Int. 10: 361–368, 1999.
6. Morris, F. L., R. M. Smith, W. R. Payne, M. A. Galloway, and J. D. Wark. Compressive and shear forces generated in the lumbar spine of female rowers. Int. J. Sports Med. 21: 518–523, 2000.
7. NIH Consensus Development Panel. Osteoporosis prevention, diagnosis, and therapy. JAMA 285: 785–795, 2001.
8. Robinson, T. L., C. Snow-Harter, D. R. Taafe, D. Gillis, J. Shaw, and R. Marcus. Gymnasts exhibit higher bone mass than runners despite similar prevalence of amenorrhea and oligomenorrhea. J. Bone Miner. Res. 10: 26–35, 1995.
9. Saul, J. S., and T. M. Hosea. Rowing. In: The Spine in Sport, R. G. Watkins (Ed.). St. Louis, MO: Mosby, 1996, pp. 578–591.
10. Smith, R., and O. M. Rutherford. Spine and total body bone mineral density and serum testosterone levels in male athletes. Eur. J. Appl. Physiol. 67: 330–334, 1993.
11. Snow, C. M., D. P. Williams, J. Lariviere, R. K. Fuchs, and T. L. Robinson. Bone gains and losses following seasonal training and detraining in gymnasts. Calcif. Tissue Int. 69: 7–12, 2001.
12. Snow-Harter, C., M. L. Bouxsein, B. T. Lewis, D. R. Carter, and R. Marcus. Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial. J. Bone Miner. Res. 7: 761–769, 1992.
13. Taaffe, D. R., T. L. Robinson, C. M. Snow, and R. Marcus. High-impact exercise promotes bone gain in well-trained female athletes. J. Bone Miner. Res. 12: 255–260, 1997.
14. Taaffe, D. R., C. M. Snow-Harter, D. A. Connolly, T. L. Robinson, M. D. Brown, and R. Marcus. Differential effects of swimming versus weight-bearing activity on bone mineral status of eumenorrheic athletes. J. Bone Miner. Res. 10: 586–592, 1995.
15. Watts, N. B. Osteoporotic vertebral fractures (©2001 American Association of Neurological Surgeons). Neurosurg. Focus
16. Witzke, K. A., and C. M. Snow. Effects of plyometric jump training on bone mass in adolescent girls. Med. Sci. Sports Exerc. 32: 1051–1057, 2000.
17. Wolman, R. L., P. Clark, E. Mcnally, M. Harries, and J. Reeve. Menstrual state and exercise as determinants of spinal trabecular bone density in female athletes. Br. Med. J. 301: 516–518, 1990.