Femoral Neck Stress Fracture in Air Force Basic Trainees

ABSTRACT Stress fractures are a common overuse problem among military trainees resulting in preventable morbidity, prolonged training, and long-term disability following military service. Femoral neck stress fractures (FNSFs) account for 2% of all stress fractures but result in disproportionate burden in terms of cost and convalescence. The purpose of this study was to describe and investigate FNSF in U.S. Air Force basic trainees and to present new data on risks factors for developing FNSF. We examined 47 cases of FNSF occurring in Air Force basic trainees between 2008 and 2011 and 94 controls using a matched case-control model. Analysis with t tests and conditional logistic regression found the risk of FNSF was not associated with body mass index or abdominal circumference. Female gender (p< 0.001) and slower run time significantly increased risk of FNSF (1.49 OR, p< 0.001; 95% CI 1.19-1.86). A greater number of push-up and sit-up repetitions significantly reduced risk of FNSF (0.55 OR, p= 0.03; 95% CI 0.32-0.93; 0.62 OR, p= 0.04; 95% CI 0.4-0.98) for females. In this study body mass index was not correlated with FNSF risk; however, physical fitness level on arrival to training and female gender were significantly associated with risk of FNSF.

INTRODUCTION

Overuse injuries are a major problem for the U.S. military, incurring excessive cost and loss of operational capability.1-2 Overuse injuries related to fitness training, although preventable and frequently predictable, contribute to the largest and most expensive reason for disability following separation from the military.3,4 Within the context of basic training, the most frequent overuse injury that leads to discharge is stress fractures.1,5 Basic military trainees (BMTs) experience stress fractures at significantly higher rates than the general military population (43.75 per 1,000 person-years vs. 2.39 per 1,000 person-years).1 Furthermore, BMTs who experience a stress fracture are four times more likely to incur a medically related discharge during basic training.5

Stress fracture is an overuse injury that occurs among improperly conditioned individuals and is primarily caused by repetitive cortical microtrauma with insufficient time for recovery.2,6 - 8 The cumulative exposure, without sufficient time for recovery, leads to disruption of cortical bone integrity causing pain and susceptibility to a completed fracture. Although significant research2,7,9 investigating the incidence of stress fractures has been conducted, gaps remain in knowledge regarding preventive, as well as, predictive risk factors (e.g., intrinsic, extrinsic, hormonal, nutritional, and genetic).10 A review of the literature revealed that a significant predictor of stress fracture is level of physical fitness before basic military training.1,2,4-9 Female gender and low body mass index (BMI) have also been correlated to an increased risk of femoral neck stress fracture (FNSF); however, it remains unclear whether these findings are generalizable to the Air Force BMT population.7,10,11 Recent research12 suggests that low BMI (<19.2 kg/m3) is also correlated to the increased risk of stress fracture among military trainees.

Approximately 4 -5% of all military trainees will experience a stress fracture.1 One of the most significant stress fractures (in terms of cost, convalescence, and morbidity) is FNSFs.6,13,14 Although extremely rare in any population, it has been reported that FNSF accounts for approximately 2% of all stress fractures among military recruits, costing at least $100,000 per case before accounting for possible surgical expenses.1,6,13 Trainee Health Surveillance data from 2009 supports the cost burden for lower extremity stress fractures exceeded $4.8 million, with FNSF accounting for 10% of the lost training days cost burden and females accounting for 2/3 of FNSF although representing only 20% of the training population (personal communication). Recent research suggests evidence-based approaches to educating leadership, adherence to fitness training standards, and increased reporting and surveillance may significantly reduce FNSF and overuse injuries.6

The purpose of this study was to assess the risk of FNSF within Air Force BMTs. Furthermore, we sought to evaluate anthropometric predictors of FNSF and identify the contribution of physical fitness characteristics toward the risk of FNSF. The use of the Air Force basic trainee population for this type of study is ideal because all Air Force recruits must obtain a physical examination before service, live in environmentally controlled accommodations at one training site in San Antonio, Texas, have time-structured access to prepared meals, and have readily accessible medical care.

METHODS

The study objectives were to describe the prevalence of atraumatic FNSF among Air Force BMTs and determine the risk conferred by anthropometric and physical fitness characteristics using a matched case-control approach. The population of interest was Air Force BMTs who received their basic training between January 2008 and December 2011 at Lackland Air Force Base, San Antonio, Texas. The total number of individuals participating in Air Force Basic Training for the study period was 140,949 with 78% participant males (n = 110,204) and 22% females (n = 30,745 [Table I]). The data sources for this study were the Basic Training Management System (BTMS) database and the Armed Forces Health Longitudinal Technology Application medical record. This study was institutional review board approved and a waiver of informed consent and Health Information Portability and Accountability Act was granted.

Cases were defined as confirmed diagnosis of FNSF on X-ray, bone scan, or magnetic resonance imaging with at least 14-day exposure to basic training and were identified using a tracking roster maintained by the Trainee Health Surveillance Flight. This roster is updated weekly and populated using ICD9 coding obtained through the Armed Forces Health Longitudinal Application medical record system. The target sample size for cases was 100 based on an average of 17 to 20 cases of FNSF occurring annually in the training population. A one-to-one match to controls was planned to ensure adequate power to detect associations.

All cases of FNSF, as well as all diagnoses of femur stress fracture, pelvic stress fracture, and other (unclassified) stress fracture were identified (n = 547) to insure that we captured any secondary FNSFs. Records for all 547 pelvic region stress injuries were reviewed for radiographic evidence of FNSF. Forty-seven of the 547 identified cases satisfied the inclusion criteria whereas 500 were excluded based on injury occurrence after completing basic training or before 14-day exposure to training. Validity was verified via secondary screening of all identified cases and random sampling 25% of the excluded stress fractures. All collected data were screened for errors by both the investigator and coinvestigator and noted discrepancies were corrected only on mutual agreement. The smaller than anticipated number of cases required protocol modification to increase the ratio of cases to controls from 1:1 to 1:2 to maintain adequate power to detect a difference of half a standard deviation in BMI measure at the <0.05 a level for a two-tailed test.

Following identification of eligible cases, a reference population was requested from the BTMS custodian of all fellow classmates who graduated from basic training on schedule. Age on arrival at basic training was calculated and the two individuals closest in age were selected as controls. Controls were defined as members of the same gender who completed basic training on schedule from the same basic training flight as the case. Using this approach, cases and controls came from the same "community" and were similar in the following ways: same instructor, fitness schedule, time allowed for sleeping/eating, distance marching/traveling, flights of stairs for dormitory assignment, and season of the year with the only difference being diagnosis of fracture. Thus, we found no reason to stratify our study sample to control for these training environment factors.

Variables of interest for analysis were age, gender, ethnicity, training squadron assignment, height, weight, BMI on arrival to training (calculated from accession height and weight), days of exposure before injury, abdominal circumference, and fitness assessment scores (i.e., push-ups, sit-ups, run time). Fitness assessment scores were obtained from the baseline fitness test performed on all BMT in the first week of training. Clinical data included BMI at the time of injury (calculated from height and weight in medical chart on date of suspicion of injury); days spent convalescing, total number of days exposed to training environment before and following FNSF and whether the individual separated from the Air Force or returned to training. The following variables were sought but excluded from analysis because of insufficient reporting: self-reported calcium abstinence, self-reported chronic steroid use, and self-reported carbonated beverage consumption. Selfreported calcium abstinence was defined as strict vegan diet, lactose intolerance, or other avoidance of dairy products.

BMI was calculated from the accession height and weight recorded in the BTMS. Height and weight were measured and recorded by trained instructor staff in the same room and on the same equipment within 5 days of arrival for basic training. Fitness scores were obtained from each trainee's initial fitness test and were recorded as continuous data. For the analysis, we selected to group run time in minutes, pushups and sit-ups in increments of 10. Days exposed to training were calculated by subtracting the date arrived to basic training from the date of clinical suspicion of FNSF. Date of clinical suspicion was defined as the initial presentation for medical care for pelvic region pain symptoms that led to provider ordering pelvic imaging. Days spent convalescing were calculated as the date of clinical suspicion subtracted from date the trainee separated from the military or returned to his/her respective training.

The BMI at the time of injury, ethnicity, and other variables were collected, but we chose not to include in the analysis because of the small sample size limiting the number of analyzable predictors. Three cases had nutritionally-related variables documented in their medical records, thus limiting any further analysis. Although abdominal circumference was collected for some within our study population, we were limited in assessing its effect on risk of FNSF because of nonrandom missing values and questionable validity. Abdominal circumference measurement was introduced midway through the study period, and is subject to significant rater bias.

IBM SPSS Statistics (version 20) (IBM Corporation, Armonk, New York) was used for statistical analysis and a two-sided p value of <0.05 was considered statistically significant. Data were evaluated for normality with the Kolmogorov- Smirnov test, confirming normal distributions. Paired t tests were used to compare means between groups. To account for the 1:2 matching design, the controls for each case were averaged and evaluated with the t test against the cases resulting in a 1:1 comparison. Mean, standard deviations, frequencies, and percentages were calculated for descriptive purposes as appropriate for distribution. Conditional logistic regression analysis was calculated using a modified cox regression. This accommodated for the matching design within SPSS to assess the relationship between FNSF and all variables and permitted multivariable analysis to assess the effect of combined fitness parameters on risk for FNSF. Analysis was performed on the entire data set, and then separately by gender where appropriate. Risk estimates were based on conditional logistic regression as a method to control for unmeasureable training parameters (i.e., training instructor, sleep cycles, distance marched throughout training, etc.).

RESULTS

Eighty-four FNSF cases were identified, of which 37 were excluded based on established exclusion criteria (<14 days exposure to training or occurrence after completion of basic training). A total of 47 eligible cases were identified followed by 94 matched controls yielding a total sample of 141 BMTs. Table I shows the distribution of the Air Force basic trainees population by year and gender from BTMS for the study period as well as FNSF prevalence per year and for the study period within this population. We found that the period prevalence for this 4-year study period was 0.04%. In addition, the period prevalence was significantly higher ( p < 0.001) among females (0.12%) when compared to males (0.01%).

We sought to compare anthropometric characteristics between cases and controls (Fig. 1). Overall, we observed males in the case and control group were heavier and taller than females. Male cases were of comparable height compared to controls, but were on average five pounds lighter and had a lower BMI compared to controls. Conversely, females in the case group were seven pounds heavier than female controls, and of comparable height. None of these findings achieved statistical significance. When we assessed the risk conferred by BMI (as a continuous measure), we did not detect statistical significance using conditional logistic regression (odds ratio [OR] = 0.99, p= 0.86; 95% CI 0.85-1.1 [Table II]). Furthermore, we found that abdominal circumference (an alternate measure of obesity) did not confer the risk of FNSF (OR =

We further sought to assess the contribution of collected physical fitness characteristics toward risk of FNSF. For both males and females, FNSF cases ran 1.5 miles significantly slower than controls (Fig. 2). Male and female controls completed more push-ups than cases; however, this difference was only statistically significant for females (Fig. 3). Males and female controls completed more sit-ups compared to cases but did not achieve significance (Table III).

Finally, we analyzed the risk conferred by physical fitness characteristics toward FNSF (Table II). Run time was significantly associated with increased risk of FNSF (OR =1.49, 95% CI 1.2-1.9, p= 0.00) for each additional minute required to run 1.5 miles. Controlling for BMI did not reduce this significance (OR = 1.5, 95% CI 1.2-1.9, p= 0.00). Run time attributed greater risk for males (3.24 OR; 95% CI 1.2 -9.0, p= 0.03) than for females (1.3 OR; 95% CI 1.1-1.7, p= 0.02). The number of push-ups accomplished in 1 minute was significant with each additional 10 push-ups decreasing the risk of FNSF for the sample (OR 0.58; CI 0.4 -0.9, p= 0.01). The risk conferred by the number of each additional 10 sit-ups accomplished in 60 seconds approached statistical significance (OR = 0.71, 95% CI 0.5 -1.02, p= 0.07). When examined by gender, the number of each additional 10 sit-ups or push-ups completed by females conferred reduction in risk (OR = 0.62, 95% CI 0.4-0.98, p = 0.04; and OR 0.55, 95% CI 0.3-0.9, p= 0.03) with no effect observed for males (p= 0.91) (Table II).

The mean number of days exposed to training before fracture was 31.26 (SD = 13.53) and was comparable between males (30.21, SD = 17.03) and females (31.70, SD = 12.05). The mean convalescing period was 126.37 (SD = 66.98) with males experiencing a slightly longer period of convalescence (133.19, SD = 54.26) when compared to females (123.28, SD = 72.43). None of these findings were statistically significant. We further found 15 of the 47 BMTs with FNSF (9 females, 6 males) were medically separated from the Air Force and did not complete basic training resulting in a discharge rate of 32% for subjects with FNSF (data not shown).

DISCUSSION

In this study population, for this period of time, we found that neither BMI nor abdominal circumference was predictor of FNSF risk when adequately powered to detect an effect. We confirmed that fitness and gender are significantly associated with risk of FNSF. Abdominal circumference was included in the BTMS data set in 2010 and inconsistently documented, limiting the ability to draw any definitive conclusions on the effect of abdominal circumference on risk because of nonrandom missing data and validity concerns.

The period prevalence of FNSF in the Air Force Basic Training population was small (0.04%); however, the disproportionate risk for females was highly significant and supports findings in previous research.7-9,11,12,15 This significance is further supported by Trainee Health and Surveillance data indicating that FNSF contributes to 10% of the overall cost burden and 3 times greater convalescence period compared to all other lower extremity stress fractures. Consistent throughout the literature, women experience 2 to 3 times higher stress fracture incidence than men.2,10,11 Wentz et al11 found bone health and normal body weight were more significant than gender in stress fracture risk in both military and athlete populations.

These findings suggest inherent differences in response to the gender-neutral training paradigm used for Air Force Basic Training, or may reflect underlying nutritional, hormonal, or fitness disparities associated with gender. Literature regarding baseline nutritional characteristics of military recruits is limited, the majority being subjective self-reporting with scant objective data available for analysis. Further prospective research evaluating the baseline nutritional, hormonal, and bone health status of female recruits would greatly enhance future stress fracture research. In addition, hormonal contraceptive practices in the months to years preceding service (e.g., oral contraceptive pill, medroxyprogesterone) and the use of over-the-counter proton pump inhibitors warrant clinical consideration with regard to overall bone health.

The significance of fitness level on risk of FNSF confirms what has previously been published regarding overall stress fracture risk.2 - 4,6,9 Individuals who run slower are significantly more likely to develop stress fractures; this may be a reflection of poor physical conditioning or undiagnosed preexisting injuries exacerbated by training. This is likely the result of decreased bone modeling or decreased physiological reserve to recover from cumulative exposure to impact activities. Similarly, push-up and sit-up performance may correlate to overall stamina and core muscle conditioning before military service. Undoubtedly, future studies will continue to include anthropometric measures available and relevant to the clinical assessment of suspected stress fracture for descriptive purposes; however, these measures have limited use as preventative or predictive tools. Exploring how gender differences in exercise patterns contribute to stress fracture disparities should be considered for any additional research on this topic as these may account for the area with greatest research affect.

Documentation of relevant nutritional history was absent in nearly all treatment encounters. It is well established that nutritional status is correlated with risk of stress fracture; 16,17 however, this information has limited use in clinical management of stress fractures. There are currently no systematic efforts to prospectively evaluate nutritional status of military recruits before arriving for training, and limited efforts to collect baseline status on arrival to training. Knowledge of baseline nutritional status could enhance treatment decisions and preventative approaches for stress fracture mitigation. In particular, clinicians should include evaluation of known nutritional behaviors that negatively affect bone density such as consumption of phosphoric acid-containing carbonated beverages, chronic steroid use, and average calcium intake as these have been correlated to poor bone quality and may affect clinical diagnosis and treatment decisions.16,17

Potential sources of bias within this study may include misclassification bias based in the control population, or diagnostic bias in which potential cases were not identified because of failure to present for care or failure to detect underlying pathology. In addition, measurement bias was possible as the height and weight scores were collected by nonmedical personnel for nonresearch purposes. These potential biases were managed by screening all cases of stress fracture in the pelvic region with intent of minimizing diagnostic bias, and by selecting flightmates as controls to mitigate systematic differences in accession measures. We believe this technique was effective in minimizing bias and error in this study in combination with a conservative statistical approach. The sample size of this study population limited the ability to perform more complex multivariate analysis, and may have limited the ability to discriminate other predictive factors. Although we believe our method of selecting controls from flightmates that graduated basic training on time by age was appropriate for this study, it is unknown if selected controls were treated for medical issues other than stress fractures, or if they developed stress fractures following successful completion of basic training. Our study was not designed to review the medical records of selected controls, limiting the ability to ascertain medical history of controls.

Future research should focus on evaluating nutritional status of female trainees on arrival to training. Additional research should consider the influence of exogenous hormonal contraception use during years before military service, not just while in training. Researchers should seek to understand baseline fitness histories for females entering military service and participation in impact sports before service in military. Finally, to minimize overuse injuries, training policy makers should consider individualizing fitness training programs to entry levels of fitness with a goal of improving overall health, decreasing frequency of high impact activities, and optimizing recovery time and nutrition. This last recommendation has proven highly successful in Army Basic Training at one installation for reducing annual incidences of FNSF by approximately 50% for men and women.6

CONCLUSION

Atraumatic FNSF remains an uncommon but debilitating stress fracture among the Air Force trainee population. FNSF disproportionately affects females by a significant margin and does not appear to be correlated with BMI. Entry fitness testing performance may be a powerful predictor for FNSF risk. Commanders may consider enhanced emphasis on entry fitness testing scores as criteria for training, or adopting individualized conditioning programs for those with marginal fitness scores on arrival. Further studies with larger populations are warranted to correlate these results with overall stress fracture rates. These findings are, nonetheless, significant and justify further research into factors underlying the gender disparity noted here.

ACKNOWLEDGMENT

My sincere appreciation to Dr. Juste Tchandja, Dr. Leo Cropper, Mr. Bob Wilson as well as the support of the Trainee Health Surveillance Department and 737th Training Group.

REFERENCES

1. Lee D: Stress fractures, active component, U.S. armed forces, 2004 - 2010. MSMR 2011; 18(5): 8-11.

2. Technical bulletin medical 592. Prevention and control of musculoskeletal injuries associated with physical training. Washington, DC, Headquarters Department of the Army, 2011. Available at http://armypubs .army.mil/med/dr_pubs/dr_a/pdf/tbmed592.pdf; accessed April 9, 2013.

3. Bullock SH, Jones BH, Gilchrist J, Marshall SW: Prevention of physical training-related injuries recommendations for the military and other active populations based on expedited systematic reviews. Am J Prev Med 2010; 38(1): S156- S181.

4. Jones BH, Thacker SB, Gilchrist J, Kimsey CD, Sosin DM: Prevention of lower extremity stress fractures in athletes and soldiers: a systematic review. Epidemiol Rev 2002; 24(2): 228 - 47.

5. Reis JP, Trone DW, Macera CA, Rauh MJ: Factors associated with discharge during marine corps basic training. Mil Med 2007; 172(9): 936 -41.

6. Scott S, Feltwell DN, Knapik JJ, et al: A multiple intervention strategy for reducing femoral neck stress injuries and other serious overuse injuries in U.S. army basic combat training. Mil Med 2012; 177(9): 1081-9.

7. Knapik J, Montain SJ, McGraw S, Grier T, Ely M, Jones BH: Stress fracture risk factors in basic combat training. Int J Sports Med 2012; 33(11): 940 - 46.

8. Kuhn KM, Riccio AI, Saldua NS, Cassidy J: Acetabular retroversion in military recruits with femoral neck stress fractures. Clin Orthop Relat Res 2010; 468(3): 846 -51.

9. Shaffer RA, Rauh MJ, Brodine SK, Trone DW, Macera CA: Predictors of stress fracture susceptibility in young female recruits. Am J Sports Med 2006; 34(1):108 -15.

10. Rome K, Handoll HH, Ashford R: Interventions for preventing and treating stress fractures and stress reactions of bone of the lower limbs in young adults. Cochrane Database Syst Rev 2005; CD000450.

11. Wentz L, Liu PY, Haymes E, Ilich JZ: Females have a greater incidence of stress fractures than males in both military and athletic populations: a systemic review. Mil Med 2011; 176(4): 420 -30.

12. Finestone A, Milgrom C, Evans R, Yanovich R, Constantin N, Moran DS: Overuse injuries in female infantry recruits during low-intensity basic training. Med Sci Sports Exerc 2008; 40(11): S630- S635.

13. Lee CH, Huang GS, Chao KH, Jean JL, Wu SS: Surgical treatment of displaced stress fractures of the femoral neck in military recruits: a report of 42 cases. Arch Orthop Trauma Surg 2003; 123(10): 527-33.

14. Pihlajamaki HK, Ruohola JP, Kiuru MJ, Visuri TI: Displaced femoral neck fatigue fractures in military recruits. J Bone Joint Surg Am 2006; 88A(9): 1989 -97.

15. Armstrong DW, Rue JP, Wilckens JH, Frassica FJ: Stress fracture injury in young military men and women. Bone 2004; 35(3): 806 -16.

16. Kelly HW, Van Natta ML, Covar RA, Tonascia J, Green RP, Strunk RC: Effect of long-term corticosteroid use on bone mineral density in children: a prospective longitudinal assessment in the childhood Asthma Management Program (CAMP) study. Pediatrics 2008; 122(1): e53- 61.

17. Takeda E, Yamamoto H, Yamanaka-Okumura H, Taketani Y: Dietary phosphorus in bone health and quality of life. Nutr Rev 2012; 70(6): 311- 21.

Maj Kevin R. Kupferer, USAF BSC*; Col David M. Bush, USAFR MC[dagger]; John E. Cornell, PhD[dagger]; Valerie A. Lawrence, MD[dagger]; Jeffrey L. Alexander, PhD[double dagger]; Rosemarie G. Ramos, PhD, MPH§; Denice Curtis, DDS, MPH, DHSc[double dagger]

*Air Force Medical Support Agency/SG5I, 7700 Arlington Boulevard, Falls Church, VA 22042-5164.

[dagger]University of Texas Health Science Center, San Antonio, IIMS/Research C-TREO - MC 7757, 7703 Floyd Curl Drive, San Antonio, TX 78229-3900.

[double dagger]A.T. Still University, 5850 E. Still Circle, Mesa, AZ 85206.

§59th MDW Office of the Chief Scientist, 2200 Berquist Drive, Room 9B11, Wilford Hall Ambulatory Surgical Center, Lackland AFB, TX 78236.

The opinions expressed on this document are solely those of the author(s) and do not represent an endorsement by or the views of the U.S. Air Force, the Department of Defense, or the U.S. Government.

doi: 10.7205/MILMED-D-13-00154