Models to predict injury, physical fitness failure and attrition in recruit training: a retrospective cohort study

Table 2 Retained predictors and associated areas under the ROC curves for the models for each course

Outcome	Course	Retained predictors	Nagelkerke R²	AUC
Any injury	28-day	Sex, Age, Init-SU, Init-shuttle^a	0.06	0.63
Any injury	80-day	Sex, Age, BMI, Init-SU, Init-shuttle^a	0.05	0.60
Overuse injury	28-day	Sex, Age, Init-SU, Init-shuttle^a	0.05	0.63
Overuse injury	80-day	Sex, Age, BMI, Init-shuttle^a	0.06	0.61
Stress fracture	28-day	(No model: only 1 stress fracture)	–	–
Stress fracture	80-day	Ht, Age, Init-shuttle	0.05	0.69
Neuromuscular injury	28-day	Sex, Age, Init-SU, Init-shuttle^a	0.07	0.64
Neuromuscular injury	80-day	Sex, Age, BMI, Init-shuttle^a	0.05	0.60
Traumatic injury	28-day	Sex, Age	0.02	0.59
Traumatic injury	80-day	Sex, Age, BMI, Init-SU^a, Init-shuttle^a	0.02	0.57
Attrition	28-day	Age	0.02	0.58
Attrition	80-day	Init-PU	0.01	0.51
Final BFA battery failure	28-day	Age, Wt, Init-PU^a, Init-SU^a, Init-shuttle^a	0.10	0.64
Final BFA battery failure	80-day	Ht, Wt, Init-PU^a, Init-SU^a, Init-shuttle^a	0.07	0.64

AUC area under the ROC curve, Init-PU push-up repetitions obtained upon arrival at the basic training site, Init-SU sit-up repetitions obtained upon arrival at the basic training site, Init-shuttle shuttle run level obtained upon arrival at the basic training sitem, Ht height in cm, Wt weight in kg, BMI Body mass index in kg/m², BFA Basic Fitness Assessment, ^afailure. dichotomized pass/fail version of the predictor variable. -. No data

ISSN: 2054-9369