Dietary and training predictors of stress fractures in female runners.

PURPOSE To compare female runners with and without a history of stress fractures to determine possible predictors of such fractures. METHODS 27 female runners (age 18-40 yr) who had had at least 1 stress fracture were matched to a control sample of 32 female runners without a history of stress fractures. Bone mineral density (BMD) was measured by dual-energy X-ray absorptiometry (iDXA). Subjects answered questionnaires on stress-fracture history, training, menstrual status, and diet. RESULTS No significant differences were found in menstrual characteristics, diet and dairy intake, or bone measurements. Weekly servings of milk during middle school significantly predicted BMD at the femur (p = .010), femoral neck (p = .002), Ward's triangle (p = .014), and femoral shaft (p = .005). Number of menstrual cycles in the previous year predicted femoral-neck BMD (p = .004). Caffeine intake was negatively associated with BMD of the femur (p = .010), femoral neck (p = .003), trochanter (p = .038), and femoral shaft (p = .035). Weekly hours of training were negatively associated with total-body BMD (p = .021), total-body bone mineral content (p = .028), and lumbar-spine BMD (p = .011). Predictors for stress fractures included the number of years running, predominantly running on hard ground, irregular menstrual history, low total-body BMD, and low current dietary calcium intake when controlling for body-mass index (Nagelkerke R2 = .364). CONCLUSIONS Servings of milk during middle-school years were positively correlated with hip BMD, although current calcium intake, low BMD, irregular menstrual history, hard training surface, and long history of training duration were the most important predictors of stress fractures.