Comprehensive Health Predictors of Injury, Illness, and Separation From Service After Enrollment in Special Forces Assessment and Selection: A Two‐Year Longitudinal Study

ABSTRACT

Individuals in physically and cognitively demanding professions face increased risk of injury, illness, and attrition due to repeated exposure to physical and psychological stressors. Identifying modifiable factors associated with these outcomes can guide risk mitigation strategies. This longitudinal study examined the 2‐year cumulative incidence of injuries, illnesses, and attrition following enrollment in the United States Army Special Forces Assessment and Selection (SFAS) course (n = 792) and assessed how physical, nutritional, cognitive, sleep, and social support at the time of enrollment were associated with illness, injury, and attrition outcomes. Upon SFAS enrollment, candidates completed psychological, sleep, and nutrition surveys. Anthropometrics and physical performance were assessed, and fasted blood was collected for physiological and nutritional biomarker analyses. Medical encounters and active‐duty status were monitored for all candidates, post 2 years enrollment, regardless of SFAS outcome. Logistic regression identified baseline factors associated with 2‐year incidence of injury, illness, or attrition. Slower road march times were associated with 25%–66% higher odds of injury, mental or behavioural illness, sleep disorder, and separation from service. Conversely, greater resistance training, better sleep quality, and higher protein intake were associated with 18%–45% lower odds of injury. Elevated morning cortisol at enrollment was associated with 23% lower odds of mental or behavioural illness, whereas negative mood, poorer sleep quality, and less social support were each associated with increased odds of separation by 28%–31%. This study identified modifiable factors that may be targeted to reduce injury and illness and improve resilience and retention in demanding occupations.

1. Introduction

Elite athletes, first responders, and military personnel are in physically demanding professions that routinely expose individuals to physical and psychological stressors which elevate risks for injuries or illnesses. For athletes, demanding training schedules, lengthy travel to competitions, and pre‐competition anxiety can contribute to suppressed immunity, mental health symptoms, and increased injury rates (Derman et al. 2022; Reardon et al. 2019; Simpson et al. 2017). The occupational demands of first responders and military personnel include extensive physical exertion while wearing heavy equipment, exposure to extreme environments, unpredictable schedules, and performing under intense stress in life‐or‐death situations (Stegerhoek et al. 2024). As a result, these occupations experience a high incidence of adverse health outcomes including, but not limited to, mental and behavioural health disorders, sleep disorders, and musculoskeletal injuries (MSKIs) (Alexander et al. 2016; Crum‐Cianflone et al. 2016; Lovalekar et al. 2021; Stegerhoek et al. 2024). Such occupational hazards underscore the need for comprehensive strategies to mitigate health risks following repeated exposure to physical and psychological stressors.

For military personnel, the health consequences of occupational demands extend beyond the individual, placing a substantial financial burden on the healthcare system and reducing the military's capacity to maintain a ready and capable force. Non‐combat MSKIs are six times more frequent than combat‐related MSKIs among United States (U.S.) service members, cost the military health system approximately $548 million annually, and were associated with 70% of U.S. Army medical disability discharges between 2011 and 2016 (Grimm et al. 2019; Molloy et al. 2020). Mental and behavioural health disorders have been on the rise among service members since the onset of Operations Enduring Freedom and Iraqi Freedom in 2001 (Crum‐Cianflone et al. 2016; Karney et al. 2008). Diagnoses of mental health disorders, including adjustment disorders, depression, and anxiety, increased by 65% among the active component of U.S. service members from 2000 to 2011 and accounted for the highest total number of occupied hospital bed‐days in 2023 (“Diagnoses of mental health disorders among active component U.S. Armed Forces, 2019–2023,” 2024). Poor sleep is not only associated with development of mental disorders and other medical conditions (Gehrman et al. 2013; J. J. Knapik et al. 2023), but also risk of injury (Ritland et al. 2021) and early discharge from military service (Seelig et al. 2016). Collectively, injuries and illnesses impact readiness by reducing availability for duty, decreasing deployability rates, and increasing medical separation rates (Lovalekar et al. 2021; Molloy et al. 2020). Although many of the physiological and psychological stressors associated with military service are unavoidable, prevention strategies aimed at minimising the risk of adverse health outcomes can not only improve performance, but also improve overall health and readiness (Bauer et al. 2018; Fallowfield et al. 2025; Rittenhouse and Khurana 2025; Wardle and Greeves 2017).

Over the past 10 years, the U.S. Army has embraced a cultural shift towards a multidisciplinary framework to optimize soldiers' individual performance and promote well‐being, and ensure soldiers are able to meet the extensive physical and cognitive demands of military service, often referred to as readiness (Deuster and O'Connor 2015; FM 7‐22 Holistic Health and Fitness 2020; Grier et al. 2018; S. Meyer and Cole 2024; V. M. Meyer 2018; Teyhen et al. 2018). The most current version of this framework, the Holistic Health and Fitness (H2F) program, was initiated in 2020 and incorporates subject matter experts, far‐forward medical care, and performance training approaches that originated in collegiate, professional, and elite sports organizations (U.S. Army Center for Initial Military Training 2023). The H2F program focuses on enhancing soldier readiness across five domains: physical, nutritional, mental, sleep, and spiritual (FM 7‐22 Holistic Health and Fitness 2020). Preliminary evidence suggests that H2F is effective in reducing injury rates, behavioral health profiles, and sleep disorders and improves readiness compared to the standard of care (H. Rhoades 2024; Szymanek et al. 2025). Identifying factors within each H2F domain that are associated with future risk of injury, behavioral illnesses, sleep disorders, or separation from service will help inform interventional targets for multidisciplinary training programs, such as H2F, to reduce risk of injury or illness and optimize performance in military, medical first responder, and athletic populations.

There were two objectives of this study. The first was to determine the 2‐year cumulative incidence of injuries and illnesses in a cohort of soldiers enrolled in the U.S. Army Special Forces and Assessment Selection (SFAS) course, a 19–20‐day intensive physical and cognitive evaluation referred to as “two weeks in hell” (Williams 2009), which is required to advance to approximately two years of Special Forces (SF) training (Farina et al. 2019). The second objective was to determine whether modifiable factors measured at SFAS enrollment—aligned with the five domains of H2F—are associated with 2‐year risk of injury, mental or behavioral illness, sleep disorder, or separation from service, irrespective of SFAS outcome. We hypothesized that more favorable status in these modifiable factors (e.g., higher physical fitness, better nutritional habits, improved sleep quality, and greater grit and resilience) would be associated with decreased 2‐year risk of injury, illness, and separation from service. Although these data were collected prior to the implementation of H2F, the modifiable factors measured remain directly applicable to current and future human performance optimization programs.

2. Methods

2.1. Participants

This study was part of a longitudinal study of active‐duty, male U.S. Army soldiers enroled in SFAS designed to identify predictors of success and resilience in SFAS, Special Forces Qualification Course (SFQC), and throughout military careers (Farina et al. 2025, 2020, 2019, 2022; Lieberman et al. 2025). There are two training courses that soldiers must pass to become a SF soldier, SFAS and SFQC. Most candidates (∼70%) do not successfully complete SFAS (Farina et al. 2019) and typically return to their previous unit or are reassigned within the Army, depending on their status and prior assignment. Candidates that successfully complete SFAS (∼30%) usually enrol in SFQC and complete approximately 2 years of specialised training, including military occupational specialty (MOS) training, Survival, Evasion, Resistance, and Escape (SERE) school, tactical skills training (e.g., land navigation, advanced weapons systems), and foreign language training (Farina et al. 2025). Most candidates selected in SFAS graduate SFQC (∼82%) and are assigned to an operational SF unit (Farina et al. 2025). For the present study, candidates who had passed SFAS were therefore either completing the final stages of SFQC or had recently graduated at the time of the 2‐year follow‐up.

Participants were recruited into this study on the day of arrival at SFAS occurring between May 2015 and March 2017. Details on recruitment, enrollment, and data collection procedures have been previously described (Farina et al. 2020, 2019, 2022). Briefly, all participants were active‐duty male U.S. Army soldiers recruited from 12 SFAS courses. Participants attended an informational study briefing prior to the start of SFAS and interested participants provided written, informed consent prior to participation. Demographics and psychological measures were collected following consent, and fasted morning blood samples were obtained the following morning prior to the start of the course. Measures of psychological and physical performance were administered and collected during SFAS by the US Army John F. Kennedy Special Warfare Centre and School (Camp Mackall, NC). In total, 1750 candidates were briefed, 821 consented, and 800 participants provided data at the start of SFAS. Data for the present analysis included 792 of the original 800 soldiers. Five soldiers requested to be withdrawn from the study prior to the 2‐year follow‐up and three were lost to follow‐up (Figure 1). The sample size of each variable varies due to participant availability or drop out during the SFAS course and are provided in the following sections.

FIGURE 1.

Study flowchart. Two‐year follow‐up data were available for 792 of the 800 candidates enroled in this longitudinal study. Injury and illness outcomes were compared between candidates that remained in service (Retained) and those who separated from service (Separated) within the 2‐year follow‐up period. Additional details are provided for the number of candidates in each group who graduated Special Forces Qualification Course (SFQC grad) and those who with not graduate SFQC (non‐SFQC grad). ¹Data from first enrollment excluded. ²No follow‐up data returned to study team. ³At the time of data retrieval participants were either still in training or SFQC outcome unknown.

2.2. Physical Readiness

The physical readiness domain of H2F refers to the ability to meet the physical demands of any duty or combat position (FM 7‐22 Holistic Health and Fitness 2020). The following measures collected prior to or during SFAS were included to assess physical readiness: weekly resistance training (n = 792), weekly aerobic training (n = 792), body mass index (BMI, n = 791), body composition (n = 790), Army Physical Fitness Test (APFT) total score (n = 782) and individual APFT event (n = 781) scores, and road march time (n = 679).

Soldiers completed an extensive, standardized background questionnaire at pre‐SFAS that included questions regarding the duration (minutes) and frequency (days per week) of resistance and aerobic training in a typical week preceding SFAS. Total duration of resistance and aerobic training exercises in a typical week were calculated by multiplying the midpoint of the self‐reported duration category (e.g., 1–15 min, 16–30 min, etc.) by the frequency (Farina et al. 2020). Weekly resistance training was categorised as: 1–199 min/week, 200–299 min/week, 300–399 min/week, or ≥ 400 min/week. Weekly aerobic training was categorised as: 1–199 min/week, 200–299 min/week, or ≥ 300 min/week (Farina et al. 2020).

Prior to SFAS, anthropometrics and body composition were assessed under fasted conditions between 0430 and 0630 while wearing lightweight clothing (Farina et al. 2025, 2019). Height was measured to the nearest 0.1 cm using a stadiometer (Hopkins Medical Products, Caledonia, MI) and body mass was measured to the nearest 0.1 kg with a calibrated electronic scale (Befour, Staukville, WI). The volunteers' body mass index (BMI) was calculated as weight/height² (kg/m²) and categorised into three groups using Centres for Disease Control (CDC) standards: healthy weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), or obese (≥ 30 kg/m²). None of the volunteers were underweight (< 18.5 kg/m²). Skinfold thickness was measured in duplicate at three sites (chest, tricep, and subscapular) with Lange calipers (Baty International, West Sussex, UK) to determine body fat percentage, fat mass, and fat‐free mass using a 3‐site skinfold equation (Jackson and Pollock 1985).

Measures of physical performance administered by the U.S. Army John F. Kennedy Special Warfare Centre and School (Camp Mackall, NC) during SFAS have been described elsewhere (Farina et al. 2020, 2019, 2022). Briefly, soldiers completed the APFT and two loaded road marches for time. Details on road march distance and times are not permitted to be released into the public domain due to agreements with the U.S. Army Special Operations Command. The APFT consisted of the maximal number of push‐ups and sit‐ups completed in 2 minutes and a maximal effort two‐mile run. The APFT score for each event ranged from 0 to 100, with a maximum APFT total score of 300. Only the first road march was included in this analysis as the second road march had a smaller sample size due to dropouts during SFAS (Farina et al. 2019).

2.3. Nutritional Readiness

Nutritional readiness refers to the ability to recognise, select, and consume adequate, healthy food and drink to meet the demands of any duty or combat position (FM 7‐22 Holistic Health and Fitness 2020). The following measures collected at pre‐SFAS were included to assess nutritional readiness: The Healthy Eating Index (HEI) total and component scores (n = 782), and blood/serum measures of haemoglobin A1C (HbA1c, n = 786), soluble transferrin receptor (sTfR, n = 742), vitamin D (n = 739), and glucose (n = 781).

HEI scores (version 2015) were calculated from the 127‐item 2014 Block Food Frequency Questionnaire (NutritionQuest, Berkley, CA) administered at pre‐SFAS (Block et al. 1990; Farina et al. 2020). HEI total score ranges from 0 to 100, calculated from the sum of 13 individual component scores, with higher scores indicative of greater compliance with federal guidelines (Krebs‐Smith et al. 2018). Six of the nine components that assess compliance with adequate intakes were included in this analysis (i.e., higher scores correspond to greater intake): total vegetables, total fruit, whole grains, dairy, total protein foods, and fatty acid ratio. Three of the four component scores that assess compliance for moderate intake were included in this analysis (i.e., higher scores correspond to lower intake): sodium, refined grains, and added sugar.

Fasted blood was drawn between 0430 and 0630 on the day prior to the start of SFAS. Whole blood samples were assessed for concentrations of glucose using Abbott i‐STAT analysers (Abbott Laboratories, Abbott Park, IL) then centrifuged, processed, shipped on dry ice to the Pennington Biomedical Research Centre (Louisiana State University, Baton Rouge, LA), and stored at −80°C until analyses. HbA1c were assessed using a UniCel DxC 600 clinical chemistry analyser (Beckman Coulter, Brea, CA). Enzyme‐linked immunosorbent assays (ELISA) were used to measure sTfR (R&D Systems Inc., Minneapolis, MN). Concentrations of 25‐hydroxy vitamin D were assessed using an iDS‐iSYS automated immunoassay analyser (Immunodiagnostic Systems, Boldon, UK).

2.4. Mental Readiness

Mental readiness refers to the ability to meet the mental demands of any combat or duty position, adapt successfully in the presence of adversity, and accomplish the mission (FM 7‐22 Holistic Health and Fitness, 2020). The following measures collected prior to and during SFAS were included to reflect mental readiness: Total Mood Disturbance (TMD) score (n = 782), Connor‐Davidson Resilience (CD‐RISC) score (n = 789), Grit score (n = 775), morning cortisol concentration (n = 789), and full‐scale intelligence quotient (FSIQ, n = 698).

Soldiers completed the Profile of Mood States (POMS 2‐A), CD‐RISC, and Grit surveys at pre‐SFAS (Farina et al. 2019; Lieberman et al. 2025). The POMS 2‐A is a standardized inventory of mood states comprised of 65 mood‐related adjectives that assesses seven factor‐analytically derived mood subscales: tension/anxiety, depression/dejection, anger/hostility, vigour/activity, fatigue/inertia, confusion/bewilderment, and friendliness (Heuchert and McNair 2012). The TMD score is calculated by subtracting the vigour score from the sum of the five negative mood subscales. The CD‐RISC is a validated, 25‐item questionnaire that was used to measure stress coping ability on a 5‐point scale, with total scores ranging from 0 (lower resilience) to 100 (higher resilience) (Connor and Davidson 2003). The Grit short scale was used to measure passion and perseverance for long‐term goals on a five‐point scale, with total scores ranging from 1 (lowest level of grit) to 5 (highest level of grit) (Duckworth and Quinn 2009). Morning cortisol was included within the mental readiness domain as a physiological marker of stress. Cortisol concentrations were measured from pre‐SFAS blood samples using a chemiluminescent detection method with Immulite 2000 automated analyser (Siemens Healthcare Diagnostics, Tarrytown, NY).

During SFAS, soldiers completed an IQ assessment following the APFT. The specific assessment cannot be disclosed as it relates to ongoing selection procedures but is similar to the Wechsler Adult Intelligence Scale and comprised of 10 timed subsets (5 verbal and 5 performance) (Farina et al. 2019). The FSIQ includes all 10 subset and normative values and has a mean and standard deviation of 100 ± 15 in the general population.

2.5. Sleep Readiness

Sleep readiness refers to the ability to recognise and implement optimal sleep behaviours to support brain function, which ultimately contributes to the ability to meet the demands of any combat or duty position and accomplish the mission (FM 7‐22 Holistic Health and Fitness 2020). The Pittsburgh Sleep Quality Index (PSQI) was administered at pre‐SFAS (n = 764) to assess global sleep quality (Lieberman et al. 2025). The PSQI is a validated, self‐report 19‐item questionnaire used to assess sleep quality and disturbances over the previous month (Buysse et al. 1989). The PSQI yields a global total score (0–21), with higher scores indicating poorer sleep quality.

2.6. Spiritual Readiness

Spiritual readiness refers to the development of personal qualities needed to sustain an individual in times of stress, hardship, and tragedy to foster a sense of purpose or meaning (FM 7‐22 Holistic Health and Fitness 2020). Sense of purpose has previously been associated with positive social relationships and inversely associated with loneliness (Hill et al. 2023). Though this study did not include a direct measure of spiritual readiness, the Duke‐UNC Functional Social Support (DUFSS) questionnaire was included in this study as a relevant aspect of spiritual readiness (n = 784). Administered at pre‐SFAS, the DUFSS is a validated, eight‐item questionnaire that measures functional aspects of social support on a five‐point scale (Broadhead et al. 1988). Total scores for the DUFSS range from 8 to 40, with higher scores indicative of greater perceived social support.

2.7. Records of Injury, Illness, and Retention

Medical data on participants were obtained from the Defence Health Agency (DHA) Armed Forces Health Surveillance Branch's (AFHSB) Defence Medical Surveillance System (DMSS) (Defense Health Agency 2025). Investigators sent AFHSB a list of participant identifiers and time frame of interest (date of SFAS enrollment through 2 years later) via a secure, encrypted file transfer system. The AFHSB then provided investigators with a list of all medical encounters for each volunteer within that time frame. Medical encounters included outpatient and inpatient visits within military treatment facilities, as well as those occurring outside these facilities (civilian care), and paid for by the U.S. Department of Defence (reimbursable) recorded as International Classification of Diseases, Ninth or 10th Revision, Clinical Modification codes (ICD‐9‐CM and ICD‐10‐CM codes, respectively). The first volunteers were enroled in May 2015, prior to the implementation of ICD‐10‐CM codes in October 2015. Therefore, some records contained a combination of both ICD‐9‐CM and ICD‐10‐CM codes. A total of 76 ICD‐9‐CM codes were identified and converted into ICD‐10‐CM codes using General Equivalence Mapping (GEM) (Centers for Medicare & Medicaid Services 2025). Fifty‐nine (78%) codes were equivalent matches and 17 (22%) required judgment as ICD‐10‐CM codes were more specific than ICD‐9‐CM codes. In those cases, the corresponding ICD‐10‐CM code with the highest frequency in the Defence Medical Epidemiology Database (DMED) (Rubertone and Brundage 2002) was selected. For example, “Other malaise and fatigue” was a single diagnosis according to ICD‐9‐CM (780.79) but separated into “Other malaise” (R53.81) and “Other fatigue” (R53.83) in ICD‐10‐CM. In this case, “Other fatigue” was the more common diagnosis in DMED and selected as the corresponding ICD‐10‐CM code. For all medical encounters, only the primary ICD‐10‐CM code was used in the analysis.

All primary ICD‐10‐CM codes were categorised into 16 major diagnosis groups with subcategories, based on a taxonomy developed by Army Public Health Command (APHC) (Hauschild et al. 2017), now maintained by Defence Centres for Public Health—Aberdeen. The APHC taxonomy comprehensively categorises all injury types and was designed to differentiate between musculoskeletal and non‐musculoskeletal injuries. The APHC taxonomy defines injury as damage to normal function of body tissues resulting from energy transfer that exceeds the threshold of tissue tolerance, either suddenly (acute) or gradually (cumulative) (Hauschild et al. 2017). Injury type is further classified based on the tissue damaged as either musculoskeletal (e.g., bone, cartilage, muscle, tendon, fascia, joint, ligament, bursa, synovium) or non‐musculoskeletal (e.g., integumentary, nervous, circulatory, respiratory, or other organs) (Hauschild et al. 2017). Therefore, cumulative MSKIs refer to injuries resulting from cumulative overload of a repetitive low‐intensity force, whereas acute MSKIs refer to injuries resulting from a single, high intensity force that exceeds the tissue's capacity to withstand (Hauschild et al. 2017). Injury and illness outcomes were coded dichotomously based on whether the participant had at least one medical encounter for a given diagnosis within the 2‐year follow‐up period.

Active‐duty status was obtained from the Defence Manpower Data Centre (DMDC). Participants with a date of separation from service that was after the date of SFAS enrollment and prior to the same date 2 years later were classified as “separated.” Reasons for separation were not included. Therefore, “separated” included voluntary and involuntary (e.g., medical) exits from military service. All other participants were classified as “retained.”

2.8. Statistical Analysis

Analyses were completed using SAS (version 9.4; SAS Institute, Cary, North Carolina). The 2‐year cumulative incidence of injuries and illnesses was compared between retained and separated soldiers using Fisher's Exact Test, with differences considered statistically significant at p < 0.05. Logistic regression was used to identify risk factors associated with the 2‐year cumulative incidence of injury, mental or behavioural illness, organic sleep disorder, and separation from service [odds ratio (OR), 95% confidence interval (CI), and Wald χ²]. For categorical risk factors (e.g., training and BMI), each level was compared to a reference group assigned an OR of 1.00. The reference groups for resistance and aerobic training were 1–199 min/week and 18.5–24.9 kg/m² for BMI. Continuous risk factors were standardized (mean = 0, standard deviation = 1) to enable comparisons across variables measured in different units. Thus, the OR and 95% CI reflected the change in odds of the outcome per one standard deviation increase in the risk factor. Nagelkerke R ² (R ² _n ) was calculated to estimate the proportion of variation in the outcome explained by the logistic regression model.

Separate, univariate logistic regression models were conducted for each risk factor. No covariates were included in these analyses to avoid underestimating the overall strength of associations or masking potential causal relationships among interrelated risk factors (Greenland et al. 1999). Risk factors were considered statistically significant where the 95% CI for the OR excluded 1.00 and p < 0.05 for Wald χ². Significant risk factors identified in the univariate analyses were subsequently entered into multivariate logistic regression models using forced entry. Multicollinearity was assessed using variance inflation factors (VIF), condition indices, and variance decomposition proportions. VIFs > 5 were considered indicative of multicollinearity, while condition indices > 30 with variance proportions > 0.5 were used to identify severe near‐dependency among risk factors in the model (Liao and Valliant 2012). For significant risk factors, differences between individuals with and without the outcome of interest were determined with chi‐square (categorical variables) or t‐test (continuous variables). Raw data, including means and standard deviations, for performance measures obtained during SFAS are not permitted to be released into the public domain due to agreements with the US Army Special Operations Command. A priori power calculations indicated that a sample size of 169–756 participants would be needed to detect risk differences of 20% or 30% (corresponding to risk ratios of 1.2 and 1.3) for injury incidences ranging from 20% to 50%, assuming β = 0.20 and α = 0.05. These calculations were based on injury incidence rates reported in prior military cohorts (Jones et al. 1993; J. Knapik et al. 1993; Reynolds et al. 2002).

3. Results

3.1. Cumulative Incidence of Injury and Illness

Among the 792 soldiers in the cohort, 69.8% sustained at least one injury of any kind within two years following SFAS enrollment, with 55.7% experiencing a cumulative MSKI and 22.4% an acute MSKI. Vector‐borne infections (e.g., upper respiratory infections, parasitic diseases) were the most common illnesses, affecting 35.1% of the cohort, followed by eye, ear, and oral conditions in 25.5% of the cohort. Additionally, 9.4% had at least one medical encounter for a mental or behavioural illness, and 4.7% for an organic sleep disorder within the 2‐year follow‐up period. Table 1 shows the 2‐year cumulative incidence for all injury and illness outcomes observed in the cohort and compares those retained (93.7%) and separated (6.3%). Those retained had lower likelihoods of acute MSKI, cumulative non‐MSKI, mental or behavioural disorders, and organic sleep disorders.

TABLE 1.

Prevalence of injuries and illnesses among special forces assessment and selection (SFAS) candidates retained and separated from service at 2‐years post SFAS course (N = 792).

	Retained (n = 742)	Separated (n = 50)	Total (N = 792)	Fisher's exact test
Injury or illness	n (%)	n (%)	N (%)	p‐value
Cancer	3 (0.4%)	0 (0.0%)	3 (0.4%)	1.000
Melanoma and other skin cancers	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Mouth and oropharynx cancers	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Testicular cancer	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Cardiovascular	41 (5.5%)	4 (8.0%)	45 (5.7%)	0.520
All other cardiovascular diseases	31 (4.2%)	2 (4.0%)	33 (4.2%)	1.000
Cerebrovascular disease	2 (0.3%)	1 (2.0%)	3 (0.4%)	0.178
Essential hypertension	6 (0.8%)	1 (2.0%)	7 (0.9%)	0.368
Inflammatory	3 (0.4%)	0 (0.0%)	3 (0.4%)	1.000
Degenerative, genetic MSK	49 (6.6%)	4 (8.0%)	53 (6.7%)	0.570
Degenerative/genetic disorder of the back or spine	5 (0.7%)	0 (0.0%)	5 (0.6%)	1.000
Degenerative/genetic disorder of the knee	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Degenerative/genetic disorder of the shoulder	3 (0.4%)	0 (0.0%)	3 (0.4%)	1.000
Osteoarthritis	8 (1.1%)	0 (0.0%)	8 (1.0%)	1.000
Other degenerative/genetic MSK disorder	35 (4.7%)	4 (8.0%)	39 (4.9%)	0.301
Digestive	56 (7.6%)	4 (8.0%)	60 (7.6%)	0.786
All other digestive diseases	38 (5.1%)	4 (8.0%)	42 (5.3%)	0.330
Appendicitis	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Cirrhosis of the liver	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Oesophagus disease	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Inguinal hernia	6 (0.8%)	0 (0.0%)	6 (0.8%)	1.000
Other gastroenteritis and colitis	15 (2.0%)	1 (2.0%)	16 (2.0%)	1.000
Peptic ulcer disease	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Eye, ear, oral	192 (25.9%)	10 (20.0%)	202 (25.5%)	0.406
All other oral conditions	11 (1.5%)	1 (2.0%)	12 (1.5%)	0.545
All other sense organ diseases	57 (7.7%)	5 (10.0%)	62 (7.8%)	0.582
Eye—glaucoma	3 (0.4%)	1 (2.0%)	4 (0.5%)	0.230
Eye—refraction, accommodation	137 (18.5%)	7 (14.0%)	144 (18.2%)	0.570
Hearing disorders	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Periodontal disease	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Genitourinary	62 (8.4%)	2 (4.0%)	64 (8.1%)	0.420
All other genito‐urinary diseases	51 (6.9%)	1 (2.0%)	52 (6.6%)	0.244
Kidney stones	2 (0.2%)	0 (0.0%)	2 (0.3%)	1.000
Nephritis and nephrosis	4 (0.5%)	0 (0.0%)	4 (0.5%)	1.000
Other breast disorders	8 (1.1%)	1 (2.0%)	9 (1.1%)	0.446
Ill‐defined conditions	298 (40.2%)	25 (50.0%)	323 (40.8%)	0.183
All other signs and symptoms*	244 (32.9%)	24 (48.0%)	268 (33.8%)	0.032
Respiratory and chest	49 (6.6%)	5 (10.0%)	54 (6.8%)	0.377
Abdomen and pelvis	77 (10.4%)	3 (6.0%)	80 (10.1%)	0.466
Infectious, vector‐borne	263 (35.4%)	15 (30.0%)	278 (35.1%)	0.541
Upper respiratory infections	154 (20.8%)	5 (10.0%)	159 (20.1%)	0.069
All other infectious and parasitic diseases*	86 (11.6%)	10 (20.0%)	96 (12.1%)	0.112
Lower respiratory infections	26 (3.5%)	1 (2.0%)	27 (3.4%)	1.000
Diarrhoeal diseases	29 (3.9%)	2 (4.0%)	31 (3.9%)	1.000
Chlamydia	18 (2.4%)	0 (0.0%)	18 (2.3%)	0.621
Sexually transmitted diseases	13 (1.8%)	1 (2.0%)	14 (1.8%)	0.602
Otitis media	6 (0.8%)	0 (0.0%)	6 (0.8%)	1.000
Unspecified viral infection	5 (0.7%)	0 (0.0%)	5 (0.6%)	1.000
Hepatitis B and C	0 (0.0%)	1 (2.0%)	1 (0.1%)	0.063
Intestinal nematode infection	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Malaria	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Tropical cluster	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Tuberculosis	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Injury	516 (69.5%)	37 (74.0%)	553 (69.8%)	0.633
Cumulative MSKI	409 (55.1%)	32 (64.0%)	441 (55.7%)	0.242
Acute MSKI*	174 (23.5%)	3 (6.0%)	177 (22.4%)	0.003
Subsequent and sequela injury encounters	164 (22.1%)	8 (16.0%)	172 (21.7%)	0.378
Acute non‐MSKI	144 (19.4%)	10 (20.0%)	154 (19.4%)	0.855
Cumulative non‐MSKI*	20 (2.7%)	7 (14.0%)	27 (3.4%)	0.001
Poisons—chemicals	66 (8.9%)	2 (4.0%)	68 (8.6%)	0.303
Environmental—thermal and light	25 (3.4%)	0 (0.0%)	25 (3.2%)	0.396
Other injury—operative/medical complication	13 (1.8%)	1 (2.0%)	14 (1.8%)	0.602
Non‐environmental—thermal	3 (0.4%)	0 (0.0%)	3 (0.4%)	1.000
Other injury—abuse/intentional*	4 (0.5%)	2 (4.0%)	6 (0.8%)	0.050
Other injury—unspecified/multiple injuries	9 (1.2%)	1 (2.0%)	10 (1.3%)	0.481
Poisons—toxins	5 (0.7%)	0 (0.0%)	5 (0.6%)	1.000
Environmental—altitude pressure	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Other injury—foreign body or food	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Other injury—lack of essential elements	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Poisons—drugs	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Other injury—reactions to external causes	0 (0.0%)	1 (2.0%)	1 (0.1%)	0.063
Environmental—cold	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Maternal, congenital	6 (0.8%)	1 (2.0%)	7 (0.9%)	0.368
All other congenital anomalies	5 (0.7%)	1 (2.0%)	6 (0.8%)	0.325
Congenital heart disease	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.0000
Mental, behavioral*	59 (8.0%)	15 (30.0%)	74 (9.4%)	< 0.001
Substance abuse disorder	23 (3.1%)	4 (8.0%)	27 (3.4%)	0.084
Adjustment disorder*	24 (3.2%)	9 (18.0%)	33 (4.2%)	< 0.001
Anxiety disorder*	11 (1.5%)	4 (8.0%)	15 (1.9%)	0.012
Mood disorder*	5 (0.7%)	6 (12.0%)	11 (1.4%)	< 0.001
All other mental disorders*	11 (1.5%)	4 (8.0%)	15 (1.9%)	0.012
Psychotic disorder	1 (0.1%)	1 (2.0%)	2 (0.3%)	0.122
Tobacco dependence	3 (0.4%)	0 (0.0%)	3 (0.4%)	1.000
Personality disorder	1 (0.1%)	1 (2.0%)	2 (0.3%)	0.122
Somatoform disorder	1 (0.1%)	1 (2.0%)	2 (0.3%)	0.122
Metabolic, endocrine	36 (4.9%)	2 (4.0%)	38 (4.8%)	1.000
All other endocrine disorders	1 (0.1%)	0 (0.0%)	1 (0.3%)	1.000
All other nutritional disorders	16 (2.2%)	2 (4.0%)	18 (2.3%)	0.316
Hypothyroidism	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Other metabolic disorders	18 (2.4%)	0 (0.0%)	18 (2.3%)	0.621
Overweight, obesity	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Neurologic*	45 (6.1%)	9 (18.0%)	54 (6.8%)	0.005
All other neurologic conditions	5 (0.7%)	2 (4.0%)	7 (0.9%)	0.067
Epilepsy	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Headache	13 (1.8%)	3 (6.0%)	16 (2.0%)	0.074
Organic sleep disorder*	29 (3.9%)	8 (16.0%)	37 (4.7%)	0.001
Other	33 (4.6%)	1 (2.0%)	34 (4.3%)	0.717
Blood—deficiency anaemia	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Blood—anaemia	0 (0.0%)	0 (0.0%)	0 (0.0%)	.
Blood—other disorder	2 (0.3%)	0 (0.0%)	2 (0.3%)	1.000
Non‐injury medical complications	5 (0.7%)	0 (0.0%)	5 (0.6%)	1.000
Other neoplasms (non‐cancer)	24 (3.2%)	1 (2.0%)	25 (3.2%)	1.000
Pulmonary	60 (8.1%)	1 (2.0%)	61 (7.7%)	0.168
All other respiratory diseases	28 (3.8%)	1 (2.0%)	29 (3.7%)	1.000
Allergic rhinitis	26 (3.5%)	0 (0.0%)	26 (3.3%)	0.401
Asthma	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Chronic obstructive pulmonary disease	4 (0.5%)	0 (0.0%)	4 (0.5%)	1.000
Chronic sinusitis	5 (0.7%)	0 (0.0%)	5 (0.6%)	1.000
Mechanical—acute non‐MSK	1 (0.1%)	0 (0.0%)	1 (0.1%)	1.000
Skin	149 (20.1%)	9 (18.0%)	158 (20.0%)	0.856
All other skin diseases	118 (15.9%)	9 (18.0%)	127 (16.0%)	0.691
Contact dermatitis	29 (3.9%)	1 (2.0%)	30 (3.8%)	1.000
Sebaceous gland diseases	16 (2.2%)	2 (4.0%)	18 (2.3%)	0.316

Note: Subcategories represent key subclassifications within a main category and do not represent all possible subcategories.

Abbreviation: MSKI = musculoskeletal injury.

^* Significantly (p < 0.05) different between SFAS candidates retained and separated from service at 2‐years post‐SFAS.

3.2. Risk Factors for Injuries

Potential risk factors of injury (all causes) within 2 years following SFAS enrollment are shown in Supporting Information S1: Table 1, with significant risk factors shown in Table 2. Resistance training was associated with lower 2‐year all cause injury risk (χ ² = 8.58, p = 0.035, R ² _n  = 0.015), with 300–399 min/week reducing the odds by 45% compared to 1–199 min/week. Each one standard deviation (1‐SD) increase in the HEI added sugars score (i.e., lower added sugar intake) was associated with 16% lower odds of a 2‐year injury, while each 1‐SD increase in road march time (i.e., slower completion time), PSQI score (i.e., poorer sleep quality), and HEI refined grains score (i.e., less refined grain intake) was associated with 25%, 20%, and 18% higher odds, respectively. Resistance training (300–399 vs. 1–199 min/week), slower road march time, and higher refined grain scores remained significant in the multivariate model (Table 2) with low multicollinearity (VIF ≤ 1.3; maximum condition index < 28).

TABLE 2.

Significant risk factors associated with sustaining an injury within 2 years following special forces assessment and selection (SFAS) enrollment.

	Univariate OR (95% CI)	χ 2	p	R 2 n	Multivariate aOR (95% CI)	Dx	No Dx
Injury (all causes)
Resistance training
1–199 min/wk	REF	—	—	—	REF	142 (18%)	48 (6%)
200–299 min/wk	0.66 (0.42, 1.02)	3.46	0.063	—	0.62 (0.37, 1.03)	122 (15%)	63 (8%)
300–399 min/wk	0.55 (0.34, 0.89)	5.98	0.015	—	0.50 (0.29, 0.88)	85 (11%)	53 (7%)
≥ 400 min/wk	0.91 (0.60, 1.38)	0.21	0.650	—	0.85 (0.52, 1.39)	204 (26%)	76 (10%)
Road march time (minutes)	1.25 (1.05, 1.50)	6.20	0.013	0.014	1.22 (1.01, 1.47)	—	—
HEI—Refined grains score	1.18 (1.02, 1.36)	4.69	0.030	0.008	1.25 (1.05, 1.48)	9.2 ± 1.5	8.9 ± 1.7*
HEI—Added sugars score	0.84 (0.72, 0.99)	4.26	0.039	0.008	0.86 (0.71, 1.04)	7.6 ± 2.2	8.0 ± 1.9*
PSQI score	1.20 (1.02, 1.41)	4.91	0.027	0.009	1.16 (0.96, 1.41)	4.7 ± 2.7	4.2 ± 2.5
Injury (acute MSKI)
Road march time (minutes)	1.26 (1.06, 1.50)	6.78	0.009	0.015	1.21 (1.01, 1.46)	—	—
HEI—Total protein score	0.86 (0.74, 0.99)	3.90	0.048	0.007	0.91 (0.77, 1.08)	4.8 ± 0.5	4.9 ± 0.4*
PSQI score	1.29 (1.10, 1.52)	9.45	0.002	0.018	1.21 (1.01, 1.46)	5.1 ± 3.0	4.4 ± 2.6*
Injury (cumulative MSKI)
Aerobic training
1–199 min/wk	REF	—	—	—	REF	155 (20%)	100 (13%)
200–299 min/wk	0.80 (0.56, 1.13)	1.64	0.201	—	0.75 (0.51, 1.10)	143 (18%)	116 (15%)
≥ 300 min/wk	0.68 (0.48, 0.96)	4.70	0.030	—	0.63 (0.43, 0.92)	143 (18%)	135 (17%)
Road march time (minutes)	1.35 (1.15, 1.59)	12.96	< 0.001	0.027	1.33 (1.13, 1.57)	—	—
HEI—Total protein score	0.83 (0.70, 0.98)	4.88	0.027	0.001	0.77 (0.61, 0.96)	4.9 ± 0.5	4.9 ± 0.3*

Note: Univariate odds ratios (OR) with 95% confidence intervals (CI) and Nagelkerke R ² (R ² _n) are presented alongside adjusted odds ratios (aOR) from the multivariate model. Significant risk factors in multivariate models are highlighted in bold. Continuous variables were standardized such that the OR reflects a change in the odds of the outcome per one standard deviation increase in the risk factor. Differences between individuals with (Dx) and without (No Dx) an injury diagnosis are presented as frequencies and proportions for each categorical variable or mean ± standard deviation for each continuous variable. Raw data SFAS performance measures (road march times) are not permitted to be displayed due to agreement with the Special Forces Command.

Abbreviations: HEI = Healthy Eating Index 2015; MSKI = musculoskeletal injury; PSQI = Pittsburgh Sleep Quality Index.

^* Significantly different (p < 0.05) between Dx and No Dx.

Potential risk factors of acute and cumulative MSKIs within 2 years following SFAS enrollment are shown in Supporting Information S1: Tables 2 and 3, with significant risk factors shown in Table 2. For acute MSKI, each 1‐SD increase in road march time and PSQI score (i.e., poorer sleep quality) was associated with 26% and 29% higher odds, respectively, while each 1‐SD increase in HEI protein score (i.e., greater protein intake) was associated with 14% lower odds. Slower road march time and poorer sleep quality remained independently associated with greater odds of acute MSKI in the multivariate model (Table 2). A significant linear dependency was noted between protein intake and sleep quality (condition index = 38; proportion = 0.85), with overall low multicollinearity (VIF ≤ 1.3).

For cumulative MSKI, each 1‐SD increase in road march time was associated with 35% higher odds, while each 1‐SD increase in HEI total protein score (i.e., greater total protein intake) was associated with 17% lower odds of 2‐year cumulative MSKI. Although the overall association between aerobic training and 2‐year cumulative MSKI risk was not significant (χ ² = 4.73, p = 0.094, R ² _n  = 0.008), post hoc comparison showed that ≥ 300 min/week versus 1–199 min/week of aerobic training was associated with 32% lower odds of a cumulative MSKI. All three variables remained independently associated with odds of cumulative MSKI in the multivariate model (Table 2). A significant linear dependency was noted between total protein intake and road march time (condition index = 33; proportion = 0.91), with overall low multicollinearity (VIF ≤ 1.3).

3.3. Risk Factors for Mental or Behavioural Illness and Sleep Disorders

Table 3 presents risk factors significantly associated with 2‐year mental or behavioural illness and organic sleep disorder, with all potential risk factors shown in Supporting Information S1: Tables 4‐5. Each 1‐SD increase in road march time (i.e., slower performance) was associated with 42% higher odds of 2‐year mental or behavioural illness, whereas each 1‐SD increase in APFT run score (i.e., faster 2‐mile run) and morning cortisol concentration was associated with 28% and 25% lower odds, respectively. In the multivariate model, only slower road march time remained independently associated with higher odds of mental or behavioural illness (Table 3). Severe multicollinearity was present between APFT run score and roach march time (condition index = 52; proportion = 88), although overall multicollinearity remained low (VIF ≤ 1.3).

TABLE 3.

Significant risk factors associated with a mental, behavioural or organic sleep disorder within 2 years following special forces assessment and selection (SFAS) enrollment.

	Univariate OR (95% CI)	χ 2	p	R 2 n	Multivariate aOR (95% CI)	Dx	No Dx
Mental, behavioural illness
APFT run score	0.72 (0.59, 0.88)	10.22	0.001	0.026	0.86 (0.59, 1.25)	—	—
Road march time (minutes)	1.42 (1.12, 1.80)	8.26	0.004	0.026	1.33 (1.02, 1.73)	—	—
Cortisol (μg/dL)	0.75 (0.60, 0.94)	6.21	0.013	0.017	0.79 (0.61, 1.03)	15.3 ± 5.0	16.7 ± 4.6
Organic sleep disorder
APFT run score	0.72 (0.56, 0.93)	6.12	0.013	0.021	0.81 (0.49, 1.32)	—	—
Road march time (minutes)	1.35 (1.00, 1.83)	3.86	0.049	0.017	1.23 (0.86, 1.76)	—	—
HEI—Sodium score	1.55 (1.12, 2.15)	6.91	0.009	0.028	1.51 (1.05, 2.18)	4.5 ± 2.3	3.5 ± 2.2

Note: Univariate odds ratios (OR) with 95% confidence intervals (CI) and Nagelkerke R ² (R² _n ) are presented alongside adjusted odds ratios (aOR) from the multivariate model. Significant risk factors in multivariate models are highlighted in bold. Continuous variables were standardized such that the OR reflects a change in the odds of the outcome per one standard deviation increase in the risk factor. Differences between individuals with (Dx) and without (No Dx) a mental, behavioural or sleep disorder diagnosis are presented as mean ± standard deviation. Raw data SFAS performance measures (APFT run score and road march times) are not permitted to be displayed due to agreement with the Special Forces Command.

Abbreviations: APFT = Army Physical Fitness Test; HEI = Healthy Eating Index 2015.

For sleep disorders, each 1‐SD increase in HEI sodium score (i.e., lower sodium intake) and road march time was associated with 55% and 35% higher odds, respectively, while each 1‐SD increase in APFT run score was associated with 28% lower odds of 2‐year organic sleep disorder. In the multivariate model, only lower sodium intake remained independently associated with higher odds of organic sleep disorder (Table 3). Severe multicollinearity was present between APFT run score and roach march time (condition index = 51; proportion = 90), although overall multicollinearity remained low (VIF ≤ 1.3).

3.4. Risk Factors for Separation From Service

Table 4 presents significant risk factors associated with separation from service within 2 years of SFAS enrollment. The odds of 2‐year separation from service increased with each 1‐SD increase in road march time (by 66%), total mood disturbance (by 31%), and PSQI score (by 34%). In contrast, the odds of 2‐year separation decreased for each 1‐SD increase in DUFSS score (by 28%), FSIQ score (by 29%), and morning cortisol concentration (by 26%). In the multivariate model, only slower road march time remained independently associated with increased odds of separation from service (Table 4). Moderate multicollinearity was present across road march time, FSIQ score, and DUFSS score (condition index = 56; proportion = 0.25–0.63), although overall multicollinearity remained low (VIF ≤ 1.3) All potential risk factors of separation from service are shown in Supporting Information S1: Table 6.

TABLE 4.

Significant risk factors associated with separation from service within 2 years following special forces assessment and selection (SFAS) enrollment.

	Univariate OR (95% CI)	χ 2	p	R 2 n	Multivariate aOR (95% CI)	Separated	Retained
Separation from Service
Road march time (minutes)	1.66 (1.27, 2.16)	13.81	< 0.001	0.053	1.51 (1.13, 2.03)	—	—
POMS—TMD score	1.31 (1.03, 1.67)	4.73	0.030	0.015	1.28 (0.93, 1.74)	7.0 ± 26.6	−0.2 ± 21.4*
FSIQ score	0.71 (0.52, 0.96)	4.96	0.026	0.019	0.79 (0.57, 1.11)	—	—
Cortisol (μg/dL)	0.74 (0.56, 0.97)	4.78	0.029	0.016	0.74 (0.55. 1.00)	15.2 ± 4.6	16.7 ± 4.7
PSQI score	1.34 (1.03, 1.73)	4.84	0.028	0.016	1.07 (0.75, 1.51)	5.3 ± 2.5	4.5 ± 2.7
DUFSS score	0.72 (0.58, 0.88)	10.51	0.001	0.029	0.85 (0.63, 1.15)	35.5 ± 7.0	37.7 ± 4.1*

Note: Univariate odds ratios (OR) with 95% confidence intervals (CI) and Nagelkerke R ² (R² _n ) are presented alongside adjusted odds ratios (aOR) from the multivariate model. Significant risk factors in multivariate models are highlighted in bold. Continuous variables were standardized such that the OR reflects a change in the odds of the outcome per one standard deviation increase in the risk factor. Differences between separated and retained Soldiers are presented as mean ± standard deviation. Raw data SFAS performance measures (road march times and FSIQ scores) are not permitted to be displayed due to agreement with the Special Forces Command.

Abbreviations: DUFSS = Duke/UNC Functional Social Support; FSIQ = Full Scale Intelligence Quotient; POMS = Profile of Mood States; PSQI = Pittsburgh Sleep Quality Index; TMD = Total Mood Disturbance.

^* Significantly different (p < 0.05) between separated and retained.

4. Discussion

This longitudinal study of 792 SFAS candidates determined that 70% had at least one medical encounter for an injury within two years of SFAS enrollment, predominantly cumulative MSKIs. In addition, 9% had at least one medical encounter for a mental or behavioural illness, 5% for an organic sleep disorder, and 6% were separated from service. More weekly aerobic training or better aerobic fitness was associated with lower odds of injury, mental or behavioural illnesses, sleep disorders, and separation from service. Greater protein intake and better sleep quality were important factors associated with decreased MSKI risk, while mood disturbance and lower social support were associated with increased odds of separation. These findings demonstrate modifiable factors across physical, nutritional, mental, sleep, and spiritual domains are potential intervention targets to reduce risk of injury or illness and enhance retention in military populations. Given that elite athletes and medical first responders can be exposed to similar physical and psychological demands, these findings may also have broader application for enhancing health and performance in these populations.

4.1. Injuries

This study cohort exhibited a higher occurrence of injuries (70%) compared to the entire active‐duty population (53%–56%) (U.S. Army Public Health Center 2018, 2019, 2020) and may suggest that soldiers in this cohort were exposed to greater physical demands during the 2‐year surveillance period. However, other established injury risk factors such as age and injury history were not accounted for in this analysis and may have also contributed to the higher injury occurrence observed in this cohort (Sammito et al. 2021). Cumulative MSKIs, resulting from cumulative overload of a repetitive low‐intensity force, were the most common injury type followed by acute MSKIs, resulting from a single, high intensity force that exceeds the tissue's capacity to withstand (Hauschild et al. 2017). Notably, acute MSKIs occurred more frequently in retained than separated soldiers, likely due in part to greater exposure to physical training among those who continued in service and less follow‐up time among those who separated.

Slower road march times were consistently associated with greater odds of 2‐year injury (all causes), acute MSKI, and cumulative MSKI, consistent with prior evidence that low physical fitness is a modifiable risk factor for injuries (Lovalekar et al. 2021; Sammito et al. 2021; Tingelstad et al. 2024; Wardle and Greeves 2017). Similarly, low muscular strength has been associated with elevated injury risk (Sammito et al. 2021). Results from the present study support the protective role of resistance training as 300–399 min/wk (∼5–6 hours/week) of resistance training was associated with 45% lower odds of an injury (all causes) compared to 1–199 min/week. However, high training volumes have been linked to increased injury risk (Tingelstad et al. 2024; Wardle and Greeves 2017), emphasising the importance of well‐designed strength and conditioning programs that incorporate progressive overload and tailored training to enhance fitness and reduce injury risk (Tingelstad et al. 2024).

Beyond physical fitness, sleep quality and nutrition also influenced the odds of 2‐year injury risk. Poorer sleep quality (higher PSQI scores) was associated with greater odds of injury (all causes) and acute MSKI, likely attributable to impaired decision‐making, slowed reaction time, and increased risk propensity associated with poor sleep quality (Haskell et al. 2025). Poor sleep quality often stems from modifiable behaviours (e.g., irregular sleep/wake times, light exposure, stimulating activities prior to bed) that can be improved through education and behaviour change (Haskell et al. 2025). Similarly, higher dietary protein was associated with lower odds of MSKI and higher refined grain and lower added sugar intake were associated with lower odds of injury (all causes). Dietary protein is an essential nutrient that provides amino acids to support bone matrix formation, muscle protein synthesis, and tissue repair (Wallace and Frankenfeld 2017). DoD nutritional standards recommend 0.8–1.6 g/kg/d of protein from a variety of food sources and 1.6 g/kg/d during periods of energy deficit to support muscle mass retention (Army Regulation 40‐25, 2017). During prolonged, intense physical activities, carbohydrates are the primary fuel source and important for maintaining glycogen stores (Amawi et al. 2023). Although refined carbohydrate intake can delay the onset of fatigue and improve endurance capacity during prolonged exercise (Ramos‐Campo et al. 2024), limited data are available to support the observed associations between greater refined grain intake, lower added sugar intake, and lower injury risk. Collectively, these results suggest that the nutritional needs of military personnel are unique due to the physical demands of their occupations (Army Regulation 40‐25, 2017).

4.2. Mental or Behavioural Illness and Sleep Disorders

During the 2‐year study follow‐up period from 2017 to 2019, approximately 9% of the study cohort had at least one medical encounter for a mental or behavioural illness and 5% for a sleep disorder. In contrast, the entire active‐duty Army population exhibited a higher occurrence of behavioural health (15%–16%) and sleep (12%–14%) disorders during that same period (U.S. Army Public Health Center 2018, 2019, 2020). This finding may be attributed to hesitancy among service members in elite units to report symptoms during medical evaluations due to negative career consequences (Garcia et al. 2022) or that SFAS‐eligible candidates engage in more healthy behaviours than the general Army population (Cooper et al. 2020). Mental or behavioural and sleep disorders occurred more frequently in separated than retained soldiers, suggesting these disorders may be interrelated or have contributed to separation from service. Veterans with mental health disorders experience a higher frequency of sleep disorder diagnoses compared to those without mental disorders (Alexander et al. 2016). Taken together, an integrated approach to human performance optimization for mental health, such the mental readiness domain of H2F, may help reduce stigma, improve retainment, and enhance health outcomes for soldiers during service and their transition to civilian life (Garcia et al. 2022).

Several factors related to physical, nutritional, and mental readiness were associated with the odds of 2‐year mental or behavioural illness or sleep disorder risk. Better aerobic fitness (i.e., higher APFT run scores and faster road march times) reduced the odds of both mental or behavioural illness and sleep disorders, likely due to the well‐documented benefits of exercise on mood regulation and sleep quality through cardiovascular health and central regulatory factors such as changes in brain function (Basso and Suzuki 2017; Chennaoui et al. 2015). Cortisol concentrations typically rise within the first hour after awakening but are blunted under chronic stress, a pattern that has been linked to depression, anxiety (Knezevic et al. 2023), and overtraining syndromes (Armstrong et al. 2021). Although higher morning cortisol concentrations at SFAS enrollment were associated with lower odds of 2‐year mental or behavioural illness, serial sampling is recommended to assess chronic stress (Knezevic et al. 2023). Similarly, there is limited biological rationale to support the observed association between lower sodium intake and increased odds of 2‐year sleep disorder. Therefore, these observations should be interpreted with caution and warrant further investigation to assess the potential relationships. Lastly, some established risk factors, such as BMI (J. J. Knapik et al. 2023), were not significantly associated with injury or illness in this cohort. This may be due, in part, to the low prevalence of obesity in this sample (8%) compared to the broader active‐duty population (17%) (J. J. Knapik et al. 2023), suggesting that BMI may be a less sensitive risk factor of injury and illness in highly active or more physically fit populations.

4.3. Separation From Service

Fifty of the 792 SFAS candidates (6%) separated from service within two years of SFAS enrollment. Risk factors encompassing physical, mental, sleep, and spiritual readiness influenced separation from service, highlighting the importance of integrative health programs for sustaining military careers. Previous reports have demonstrated that lower aptitude, mental health symptoms, low physical activity or fitness, and frequent sleep difficulties are associated with increased risk of attrition in U.S. Armed Forces (J. J. Knapik et al. 2004). Likewise, lower intelligence scores, higher TMD, slower road march times, poorer sleep quality, and lower morning cortisol concentrations at the time of SFAS enrollment were associated with greater odds of 2‐year separation. Lower morning cortisol may be associated with depression and anxiety (Knezevic et al. 2023) or overtraining syndromes (Armstrong et al. 2021), as previously discussed, whereas higher cortisol concentrations have been associated with improved performance in operational settings, suggesting allostatic load management or an adaptive response to stress (Szivak et al. 2018; Taylor et al. 2007).

Contrary to our hypothesis, neither grit nor resilience was associated with odds of 2‐year separation, which may be attributable to low variability or a ceiling effect in these traits among highly motivated cohorts (Arias Gonzalez et al. 2015; Benedict et al. 2023). However, SFAS candidates reporting higher perceived social support, an aspect of spiritual readiness, had lower odds of separation within 2 years. Military unit support from leaders or peers has been shown to have a positive impact on military retention (Vasterling et al. 2015). In the civilian workforce, organizational support has been associated with job satisfaction and occupational commitment (L. Rhoades and Eisenberger 2002). These findings emphasise the importance of community and support for retaining soldiers. While only road march time remained statistically significant in the multivariate model, severe dependency was observed among several variables and underscores the strong interrelatedness of factors across H2F domains. Taken together, these findings suggest that factors related to physical, mental, and spiritual readiness likely influence separation from service through both independent and shared pathways.

4.4. Strengths and Limitations

Key strengths of this longitudinal study included the large sample size (N = 792), comprehensive measures collected at SFAS enrollment, high study retention rate over the 2‐year follow‐up period (99%), and use of a robust medical database that provided detailed documentation of diagnosed medical conditions. However, several limitations should be considered when interpreting these findings. First, the observational design of this study precludes causal inference and findings should be interpreted as associations rather than direct causes. Second, all risk factors were measured only once before SFAS, whereas physical performance, nutrition, sleep quality, psychological state, social support and biomarkers likely fluctuated over the 2‐year follow‐up period. Third, many comparisons of 2‐year cumulative injury and illness incidence between retained and separated soldiers were underpowered due to the low separation rate (6%) and the combination of voluntary and involuntary exits. Fisher's exact test was used to account for small cell sizes, but these results should still be interpreted with caution. Fourth, additional aspects of sleep and spiritual readiness, such as actigraphy to assess sleep or spiritual surveys, were not captured in this study and may also contribute to the risk of injury, illness, or separation. Fifth, this analysis involved multiple univariate regression models to identify potential associations between risk factors aligned with the five domains of H2F and outcomes related to injury, illness, and separation from service. While this approach avoids underestimating overall effect sizes or masking interrelated causal relationships (Greenland et al. 1999), it did not adjust for multiple comparisons, which increases the risk of type I errors. Results based on univariate analyses should be viewed as preliminary and interpreted with caution. Multivariate regressions revealed significant linear‐dependencies among risk factors, highlighting their strong interrelatedness. Future confirmatory studies should incorporate multiple testing corrections to validate true effects. Lastly, injury and illness outcomes were coded dichotomously based on whether participants had at least one encounter. It is possible that risk factors may differ, or be more strongly associated, among participants with recurrent visits for the same condition.

5. Conclusion

The present study highlights modifiable factors related to physical, nutritional, mental, sleep, and spiritual readiness that significantly contribute to risk of injury, illness, and separation from service within 2 years. These findings demonstrate there are unique nutritional needs of military personnel as well as the importance of baseline aerobic fitness, sleep quality, psychological well‐being, and social support to reduce the odds of injury, mental illness, sleep disorders and attrition. Interventions aimed at reducing risk of injury and illness in physically demanding professions should consider a multidisciplinary approach, such as H2F, to enhance performance, health, and well‐being.

Author Contributions

E.F.K, J.J.K, S.M.P, J.P.M, and H.R.L. conceived and designed the research, E.F.K, L.A.T., J.J.K, S.M.P, J.P.M, and H.R.L. conducted research and investigation processes, E.F.K. and M.E.B. analyzed data, M.E.B., E.F.K., J.J.K and H.R.L. interpreted results, M.E.B. prepared figures, M.E.B. drafted manuscript, M.E.B., E.F.K, L.A.T., J.J.K, S.M.P, J.P.M, and H.R.L. edited and revised manuscript, M.E.B., E.F.K, L.A.T., J.J.K, S.M.P, J.P.M, and H.R.L. approved final version of manuscript.

Funding

This work was funded by the Defense Medical Research and Development Program (DMRDP) and U.S. Army Medical Research and Development Command (USAMRDC).

Ethics Statement

This study was conducted under a memorandum of agreement between the U.S. Army Special Operations Command and the U.S. Army Research Institute of Environmental Medicine (USARIEM). The USARIEM Institutional Review Board (IRB) initially approved the study and the U.S. Army Medical Research and Development Command IRB has conducted continuing reviews. Investigators adhered to the policies for protection of human subjects in Department of Defence Instruction 3216.02 and the research was conducted in adherence with the provision of 32 Code of Federal Regulations Part 219 and the ethical principles of the Declaration of Helsinki.

Consent

Participants attended an informational study briefing and interested participants provided written, informed consent prior to participation.

Conflicts of Interest

The authors declare no conflicts of interest.

Disclaimers

The results of this study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the Army or the Department of War. Any citations of commercial organizations and trade names in this report do not constitute an official Department of the Army endorsement or approval of the products or services of these organizations.

Supporting information

Supporting Information S1

Data Availability Statement

Data cannot be made publicly available due to agreements with United States Special Operations Command (USSOCOM). Data are not permitted to be displayed by the U.S. Army John F. Kennedy Special Warfare Center and School (USAJFKSWCS) as releasing this information may interfere with ongoing training and selection procedures.