Adequate vitamin D plays a central role in promoting calcium absorption and maintaining adequate serum calcium and phosphate concentrations, enabling normal bone mineralization and preventing hypocalcemia, and is essential for bone formation and resorption.1–4 Adequate serum concentration of 25-hydroxyvitamin D [25(OH)D] is associated with a lower prevalence of osteomalacia5–7 and a lower incidence of stress fractures.8
Serum 25(OH)D is the most accurate indicator of vitamin D nutritional status because it includes the effects of solar ultraviolet (UVB) exposure and oral intake from food and supplements, and it has a markedly longer half-life (approximately 3 weeks) than vitamin D (14 hours) or 1,25-dihydroxyvitamin D3 [1,25-(OH)2D3; 24 hours].2, 4
Skeletal stress fractures associated with cumulative trauma are common in military recruit populations and are an important cause of early attrition from service.8–12 Such studies have revealed a higher incidence of stress fractures in females than in males.13, 14 Another study found that within race/ethnic groups, age-adjusted mean BMD was higher in men than in women at all skeletal sites examined (ie, pelvis, arms, and lumbar spine; p < 0.05).15 Vitamin D deficiency is associated with an increased incidence of fractures in older adults of both sexes,16 and a relationship between vitamin D deficiency and risk of fractures in those 18 to 44 years of age has been identified only recently.8, 9
A randomized, placebo-controlled clinical trial by Lappe and colleagues reported that female recruits assigned to 800 IU/d of vitamin D3 and 2000 mg/d of calcium had a 20% lower incidence rate of stress fractures than the placebo group (p ≤ 0.01).9 Amenorrhea and age older than 25 years also were predictors of increased risk of stress fracture, whereas history of exercise was protective.9 In another study, serum 25(OH)D concentrations below the median (30 ng/mL) were associated with a higher odds ratio (OR) for stress fracture in young men (OR = 3.6, 95% CI 1.2–11.1).8
This study provides the first dose-response relationship between serum concentrations of 25(OH)D and the risk of stress fracture using sera collected before the fracture occurred.
Materials and Methods
This was a nested case-control study within a cohort of US Navy female recruits. Data and serum specimens for this study were collected for comprehensive military health surveillance and epidemiologic analysis in accordance with Department of Defense (DoD) Directive 6490.2, dated October 21, 2004, and DoD Instruction 6490.3, dated August 7, 1997. The data are stored in the Defense Medical Surveillance System (DMSS), the central repository of medical surveillance data for the US Armed Forces, which contains up-to-date and historical data on disease incidence and other medical events and longitudinal data on personnel and deployments. The DoD Serum Repository (SR) has been used for previous epidemiologic research.17
Sera from all active-duty service members, including recruits at the time of service entry, were collected by the DoD for surveillance and research purposes.18 The study population for this nested case-control study consisted of 1200 Navy female recruits, including 600 cases and 600 controls. Incident cases of stress fracture of the tibia or fibula that met the case definition and had stored sera were included in the study. The case definition was a stress fracture of the tibia or fibula (ICD 9-CM codes 733.16 or 733.93) identified in the first or second diagnosis positions that occurred during the first 180 days of active-duty service between January 1, 2002, and June 30, 2008, in female recruits. The most recent serum samples preceding the date of stress fracture were obtained.
Each case was individually matched to a female control by the following criteria: age (±2 years), race/ethnicity (non-Hispanic white, non-Hispanic black, or other), length of service equal to that of the case ± 30 days, and blood drawn within ± 2 days of the case. All controls were on active duty within ± 30 days of the date of diagnosis of stress fracture in the case. Hispanic persons were considered white. If more than one potential control met the criteria, the control whose date of serum collection was closest to that of the case was selected.
Independent variables included serum 25(OH)D concentration, BMI, and state of residence at entry into the military. BMI was included because heavier individuals have higher hip BMD values and more robust femur geometry.19 Furthermore, BMI and body weight are inversely associated with 25(OH)D concentrations.20, 21 PTH concentration was not available.
BMI was calculated by dividing body weight in kilograms by body height in meters squared. Weight and height measurements were available on 594 case/control pairs. Home of residence was included because 25(OH)D concentrations usually are inversely related to latitude.22 Quarter of first diagnosis of stress or pathologic fracture while in the military was categorized as winter (January to March), spring (April to June), summer (July to September), and autumn (October to December). Latitude of residence was determined by state or territory of residence at entry into the military. Northern latitudes were defined arbitrarily as states or territories at 41° latitude or higher, middle latitudes were states or territories at between 37 and 40° latitude, and southern latitudes consisted of states or territories lower than 37° latitude.
No data were available on history of oral intake of vitamin D before joining the military.
Univariate and multivariate analyses were performed using the SAS Version 9.1 conditional logistic regression procedure (PROC LOGISTIC; SAS, Inc., Cary, NC, USA). Quintiles of serum 25(OH)D were defined based on the control population. ORs of stress fracture were determined using the lowest quintile of serum 25(OH)D in the control population as the reference category. The criterion for statistical significance was p ≤ 0.05, two-tailed.
This study used aliquots of 0.5 mL each of frozen serum samples stored in the DoDSR to determine prefracture circulating serum 25(OH)D concentrations for cases and their matched controls. Samples were analyzed blindly at Heartland Assays, Inc. (Ames, IA, USA), using acetonitrile and a radioimmunoassay method.23
This study was conducted in accordance with the ethical standards of the relevant DoD Institutional Review Board (IRB) and the Helsinki Declaration of 1975, as revised in 1983, and IRB approval was obtained.
The majority of recruits with stress fracture were white (54.4%) and aged 20 years or younger (80.0%; mean age 19.5 years; Table 1). The mean serum 25(OH)D concentration was 28 ng/mL in cases and 31 ng/mL in controls. In the overall population, the mean and median serum 25(OH)D concentrations were 30 ng/mL. Blacks had the lowest mean 25(OH)D concentrations in both cases and controls (16 ng/mL in cases versus 20 ng/mL in controls; Table 1). The mean BMI was 23.9 kg/m2 in cases and 24.2 kg/m2 in controls (Table 1).
Table 1. Characteristics of Cases and Controls, US Navy Female Recruits, 2002–2008
Cases (n = 600)
Controls (n = 600)
*p < 0.05 according to unpaired t test.
Mean age in years (± SD)
Serum collected (%)
Mean BMI (± SD)
Mean serum 25(OH)D, ng/mL (± SD)a
Women in the highest quintile of serum 25(OH)D (range 39.9 to 112.5 ng/mL; mean 50 ng/mL ± standard deviation [SD] 10 ng/mL) had a lower risk of stress fractures than those in the lowest quintile (range 1.5 to 19.7 ng/mL; mean 14 ng/mL ± SD 4 ng/mL; OR = 0.51, 95% CI 0.34–0.76, p ≤ 0.01; Fig. 1). There was a dose-response relationship of higher serum 25(OH)D concentration with lower risk of stress fracture (ptrend = 0.02; Fig. 1). There was no association between BMI or latitude of home of residence and risk of stress fracture. Latitude of home of residence was included in the multivariate regression model because it was associated with risk at the α = 0.10 level of significance (p = 0.06). The association was tested using linear and nonlinear models.
The association of serum 25(OH)D concentration with risk of stress fracture also was analyzed by race. Quartiles were used for whites and other races, whereas division at the median was used for blacks, who were the demographic group with the fewest (n = 71) cases.
There was a clear inverse dose-response gradient in white women, similar to that for women of all races combined. In white women, ranges and means for quartiles of serum 25(OH)D and corresponding ORs were as follows:
Quartile 1: Range 6.5–26.9 ng/mL, mean 20 ng/mL, OR = 1.0 (reference group)
Quartile 2: Range 27–33.4 ng/mL, mean 30 ng/mL, OR = 0.68, 95% CI 0.44–1.1
Quartile 3: Range 33.5–40.1 ng/mL, mean 37 ng/mL, OR = 0.57, 95% CI 0.36–0.89
Quartile 4: Range 40.2–112.5 ng/mL, mean 50 ng/mL, OR = 0.49, 95% CI 0.31–0.79 (ptrend = 0.02)
Serum 25(OH)D concentrations among white women in the highest quartile (40.2 to 112.5 ng/mL, mean 50 ng/mL) were associated with half the risk of stress fracture as those in the lowest quartile (6.5 to 26.9 ng/mL, mean 20 ng/mL; p ≤ 0.05). Blacks with serum 25(OH)D concentrations above the median (20 ng/mL) had a substantially lower, although statistically borderline, risk of stress fracture than blacks below the median (OR = 0.54, 95% CI 0.28–1.06, p = 0.07). In women of other races, no association was observed in this study between serum 25(OH)D concentration and risk of stress fracture.
In the multivariate regression model, serum 25(OH)D concentration persisted as being inversely associated with risk of stress fracture (ptrend = 0.02; Table 2). The association of latitude of home with ORs for stress fracture remained borderline (p = 0.07). Similar to the univariate analysis, risk of stress fractures was lowest in individuals whose serum 25(OH)D concentration was in the highest quintile (39.9 to 112.5 ng/mL) compared with those in the lowest quintile (1.5 to 19.7 ng/mL; OR = 0.51, 95% CI 0.34–0.78, p = 0.01). For latitude of home of residence, persons living in southern latitudes at the time of entry into the Navy had a higher risk of stress fracture (OR = 1.43, 95% CI 1.05–1.95, p = 0.05).
Table 2. Multivariate Analysis of Stress Fracture and Serum 25(OH)D Concentration and Home of Residence Using Conditional Logistic Regression in 600 Cases and 600 Controls, 2002–2008
p Value for stratum
Overall p value
Serum 25(OH)D range (mean), ng/mL
Home of residence (latitude)
An inverse association between serum 25(OH)D and risk of stress fracture was not detected among blacks or individuals of other races, although the finding in blacks approached statistical significance. Previous data suggest that blacks have higher mean total BMD than whites.15 Our inability to detect an association of serum 25(OH)D with risk of stress fracture may be due to overall higher BMD in blacks.15 The positive association with serum 25(OH)D, although of borderline statistical significance, suggests that vitamin D still could be of value for the prevention of stress fractures in blacks. A larger observational study or randomized, controlled trial similar to that of Lappe and colleagues9 could help in further evaluation of this association.
Another possible explanation for the lack of a statistically significant association between serum 25(OH)D level and lower risk of fractures in blacks is the low concentration of serum 25(OH)D found in this population. Owing to increased levels of skin pigmentation, blacks synthesize 25(OH)D at one-third the rate of whites.24 In this study, blacks had mean serum 25(OH)D concentrations that were approximately half those of whites (Table 1). Since the greatest statistically significant reduction in risk was found in subjects in the highest quintile of serum 25(OH)D (39.9 to 112.5 ng/mL), serum 25(OH)D in black subjects may not have been high enough for a protective effect on fractures to occur.
Previous research has shown that amenorrhea and history of exercise were independently associated with risk of fractures.9 In this study, it was not possible to include amenorrhea and exercise history (prior to basic training) in the regression model as confounders because data on these factors were not available for the population under study. Therefore, results of the analysis should be interpreted with caution. However, all recruits undergo the same training regimen in basic training. Therefore, all recruits have the same exercise history with respect to basic training.
No association was detected between latitude of home of residence and risk of stress fracture. This was unexpected because there are greater amounts of solar ultraviolet B exposure at lower latitudes. Home of residence is the location of residence indicated at the time of entry into the military. The home of residence variable does not take into account the length of time spent in a particular state or territory and is not necessarily indicative of usual location or the location of birth. Alternatively, persons living in southern latitudes may use sunscreen more frequently, preventing solar ultraviolet B from penetrating the skin to synthesize vitamin D.
BMI over the range observed in the study population was not associated with risk of stress fracture. BMI was analyzed as a continuous rather than categorical variable because the standard categories associated with BMI ranges are specific to adults 20 years of age and older, and the majority (61%) of our study population was aged 19 years or younger.
Data on other sites of lower extremity stress fracture were not available. Stress fractures of the pelvis, femoral neck and shaft, other parts of the femur, metatarsals, and other lower extremity sites were formerly classified under general ICD-9-CM codes, and the site of stress fracture had to be specified separately. Because the specific site affected was not available in the database, only fractures of the tibia and fibula could be identified. However, in military recruits, the most common sites of stress fractures are the tibia and fibula.9
This study may be interpreted in the context of the findings of a study by Priemel and colleagues,6 who examined iliac crest bone biopsies in Germany and found that pathologic accumulation of osteoid was completely prevented only when the circulating 25(OH)D concentration was greater than 30 mg/mL. The authors concluded that 30 ng/mL of 25(OH)D is the minimum circulating concentration to maintain bone health. This study confirmed this assertion, but our data suggest that 40 ng/mL would be a more appropriate target for 25(OH)D with regard to prevention of stress fractures because the risk of stress fractures continued to be lower through at least 40 ng/mL.
Both the proposed 30 and 40 ng/mL targets for bone health are higher than the 20 ng/mL target for 25(OH)D concentration proposed by the Committee to Review Dietary Reference Intakes for Vitamin D and Calcium of the National Academy of Sciences–Institute of Medicine, Food and Nutrition Board.25 It is possible that the committee may have included only the data points for observed excessive osteoid volume above 2% solely from the subjects whose serum 25(OH)D concentrations also were less than 20 ng/mL. Such an analysis would have excluded at least 6 data points where osteoid volume was excessive and the subjects' serum 25(OH)D concentrations were higher than 20 ng/mL. It is not clear why or whether these data points were excluded. If they had been included, they would have indicated that 30 ng/mL rather than 20 ng/mL was the minimal 25(OH)D concentration threshold for preventing excessive osteoid volume and the minimum target for maintaining bone health. Based on this study, it appears that a higher serum 25(OH)D concentration, specifically 40 ng/mL, would be a more appropriate target for maintaining bone health with respect to stress fractures of the tibia and fibula.
There was either a lack of a statistically significant benefit or, alternatively, a beneficial trend of borderline significance between serum 25(OH)D concentration and risk of stress fracture in African Americans in this study. This finding is reminiscent of the lack of association between prediagnostic serum 25(OH)D concentration and risk of multiple sclerosis in African Americans in an earlier study.17
In that study, African Americans had lower risk of multiple sclerosis than whites, and there was no significant association of serum 25(OH)D concentration with risk of the disease.17 To account for this racial difference, the authors proposed that a gene associated with racial admixture26 might have reduced the incidence of multiple sclerosis and its association with low 25(OH)D concentration in African Americans. On the other hand, though, the dose-response gradients for the association of lower serum 25(OH)D concentration with higher risk of stress fracture were similar in African Americans and whites in this study. All considered, low serum 25(OH)D concentration appears to be associated with a risk of stress fractures of the tibia and fibia, probably regardless of race. However, more research would help to confirm this because the findings in African Americans were borderline.
The latitude of training is that of Great Lakes, IL, 43°N. The military does not, at present, routinely provide 25(OH)D testing or vitamin D supplements to recruits.
Risk of stress fractures was paradoxically higher in recruits who identified the South as their home of record. It is possible that regional differences in per-capita dietary intake of calcium could account at least in part for the higher rate of stress fractures in women from the South because lower intake of calcium in the South theoretically might contribute to a higher risk of osteomalacia, potentially a precursor of fractures. Per-capita mean intake of dietary calcium is somewhat lower in the South (685 mg/d) than in the North Central (755 mg/d) or Northeastern (728 mg/d) sectors of the United States.27 Race also may have contributed this finding because the incidence of stress fracture was substantially lower in black than in white women, and recruits from the South were more likely than those from other regions to be black.
The findings in this study emphasize the importance of vitamin D status in risk of stress fractures in young women who are taking part in physical activity. Although this study was conducted in a military population exposed to rigorous physical training over an 8-week period, vitamin D is also likely to benefit young women in the general population who engage in strenuous physical activity. This could include females who take part in high school or college athletic programs, marathons and/or triathlons, and moderate- to high-impact weight-bearing activities. Women with a history of stress fracture also may benefit from the findings of this study. Recent studies have shown that vitamin D deficiency is highly prevalent in otherwise healthy young people.28–30
Since the risk of fractures increases in older adults and fractures can occur at any age, vitamin D supplementation is recommended for all ages. According to the Dietary Reference Intakes of the Food and Nutrition Board, Institute of Medicine (IOM)–National Academy of Sciences, the current Recommended Daily Intake of vitamin D for persons from birth to 50 years is 200 IU/d of vitamin D, 51 to 70 years is 400 IU/d, and 71 years and older is 600 IU/d.2 Historically, serum 25(OH)D concentrations of 20 ng/mL or more (or 50 nmol/L or more) have been considered adequate for bone and overall health in healthy individuals.2 However, based on the findings of this study, only serum 25(OH)D concentrations in the highest quintile (39.9 to 112.5 ng/mL) were associated with a lower risk of stress fracture. A similar association was present when the comparison was limited to whites (40.2 to 112.5 ng/mL).
The lowest serum 25(OH)D concentration shown to adequately reduce the risk of stress fracture was at least double the formerly recommended minimum serum concentration (>20 ng/ml).2 In normal adults, 4000 IU/d of vitamin D3 would be needed to maintain serum 25(OH)D concentrations at no less than 100 nmol/L or 40 ng/mL.31 Although 25(OH)D concentrations of up to 400 ng/mL (or 1000 nmol/L) were not detected as being associated with harm in an animal model,32 no randomized, controlled trials have been conducted to establish the safety of 2000 to 4000 IU/d by young adults.
While no randomized trials of vitamin D3 intakes of 2000 to 4000 IU/d have been reported, the 2010 IOM report endorsed 4000 IU/d as being the tolerable upper level of intake (TULI).2
According to this study, 75% of female recruits were at higher than necessary risk of stress fractures of the tibia and fibula owing to low serum 25(OH)D concentrations (<40 ng/mL). The higher risk of stress fractures owing to low serum 25(OH)D could be addressed in young women by ensuring that their serum 25(OH)D concentration is maintained at 40 ng/mL or higher.
This study did not measure serum 25(OH)D concentrations in male recruits. Therefore, it is not known from this study whether a similar association of lower serum 25(OH)D concentrations with higher risk of stress fracture applies to males. However, a previous study by Ruohola and colleagues in Finland identified such an association in males that is similar to that of this study of females.8 It is reasonable to expect that the association found in this study in women also would be present in men, considering the findings of Ruohola and colleagues.8 Future research on male recruits would be desirable to confirm this, though, in the United States or other countries.
Vitamin D intake has been reported to be safe for adults in doses up to 9000 IU/d.33 Full-body solar exposure provides the equivalent of 10,000 IU/d, suggesting that this may be a physiologic limit.31, 34, 35 These results suggest that supplementary vitamin D3 intakes of 2000 to 4000 IU/d would promote bone health and reduce the risk of stress fractures and that the benefits of this intake would exceed its risks.
Recruit training begins shortly after entry into the US military. It may be beneficial to test serum 25(OH)D concentrations at military processing centers to identify individuals at higher risk of fracture during recruit training. Vitamin D supplementation during boot camp has been shown to decrease the risk of stress fracture in female Navy recruits.9 Vitamin D3 supplementation prior to boot camp would be even more beneficial because it may take 2 to 3 months to increase serum 25(OH)D concentrations to optimal in those who are insufficient.
All the authors state that they have no conflicts of interest.
We express our deepest thanks and appreciation to Dr Angie Eick, special studies lead at the Armed Forces Health Surveillance Center, for her invaluable assistance, expert advice, and thoughtful consultation and review. Funding was provided by the US Army Medical Research and Materiel Command's Military Operational Medicine Research Program for Bone Health and Military Medical Readiness (Fort Detrick, MD) under Work Unit No. 63706N-60836.
Authors' roles: AAB performed the statistical analyses, arranged for access to the DoD blood samples, wrote the first and several subsequent drafts of the manuscript, and incorporated reviews by coauthors in the period preceding submission. EDG coordinated the overall concept and design of the study, proposed statistical analyses, and reviewed and approved multiple drafts of the manuscript. CFG helped with the overall concept and design of the study and reviewed and approved multiple drafts of the manuscript. SBM reviewed and approved drafts of the manuscript and made numerous key edits to the manuscript. FCG proposed the concept and design of the study and reviewed and approved multiple drafts of the manuscript. KZ reviewed the literature on vitamin D and stress fractures, reviewed the manuscript, provided literature citations, and shared in the drafting of revisions. KT and JML offered important suggestions regarding design of the study, proposed the diagnostic inclusion criteria, and reviewed and approved drafts of the manuscript.