Study design and participants
This study was approved by the U.S. Army Medical Research and Development Command Human Institutional Review Board (Fort Detrick, MD). Investigators adhered to the policies regarding the protection of human subjects as prescribed in Army Regulation 70-25, and the research was conducted in adherence with the provisions of 32 CFR Part 219. Data collection took place October 2021 to July 2022 at three Army installations: Fort Liberty, NC; Fort Gregg-Adams, VA; and at the U.S. Military Academy, West Point, NY (cadets were excluded). All participants provided written informed consent. Exclusion criteria included current pregnancy and large amounts of metal in the body that would impact body scanning procedures.
The study population was a representative sample of 1904 (643 Females) active-duty Soldiers (Suppl Table 1). The population was selected using validated methodology similar to the National Health and Nutrition Examination Survey (NHANES) in order to produce a study population that was representative of the U.S. Army demographic distributions of sex, race/ethnicity (R/E, American Indian or Alaskan Native, Asian/Pacific Islander, Black non-Hispanic, Hispanic, White non-Hispanic, and other) and age category (17-20, 21-27, 28-39, >40-years old) that existed in the Army in December 2020 [17]. Women and groups comprising less than 10% of the populations (American Indian or Alaskan Natives, Asians and those >40 years old) were oversampled to better evaluate between groups differences in these smaller populations.
Data collection
Demographic and physical performance data
Participants provided baseline demographic and military information by consenting to provide their Department of Defense (DoD) identification number to allow access to current Army Combat Fitness Test (ACFT) scores. At the time of data collection, the ACFT included six events: 3 repetition maximum deadlift, standing power throw, hand-release push-up, sprint-drag-carry, 2-mile run and the leg tuck. The ACFT events were scored according to age and sex adjusted standards (maximum of 100 points/event and maximum cumulative total score of 600). In March 2022, the leg tuck was dropped from the ACFT and replaced with a timed plank [18]. To account for this change within analyses containing the ACFT, scores for the leg tuck were dropped from the total ACFT scores and it was assumed that all passed the timed plank at the minimum score needed for each threshold score analysis. MSKI data was accessed through a data repository maintained by the US Army Research Institute of Environmental Medicine, the Soldier Performance Health and Readiness database. For this study, MSKI was defined as any injury that resulted in lost or limited duty time in the one year following the participant’s body composition assessment.
Body composition assessment
Participants were asked to come to data collection site fully hydrated, with at least 12 hours since the last exercise bout and three hours since their last meal. Anthropometric measurements were made in lightweight shirt, shorts, and sports bra (for women) with stocking feet. Standing height was measured using a stadiometer (model 217, SECA, Chino, CA) and body mass was measured using a calibrated electronic scale (model DS6150, Doran, Batavia, IL).
Body composition was determined using four techniques: dual-energy x-ray absorptiometry (%BFDXA, GE Lunar Prodigy Advanced, GE Healthcare, Madison, WI), bioelectrical impedance analysis (%BFBIA, InBody 770, InBody, Cerritos, CA) and CBE. Female participants produced a negative pregnancy test prior to scanning procedures. For CBE, manual circumference for both men and women were measured at the neck, waist, abdomen, and hips in triplicate. The measurements were conducted by trained research staff using a calibrated fiberglass tape measure and recorded to the nearest 0.5inch, body weight (BW) to the nearest pound and height to the nearest inch. The circumference measured were used to estimate %BF using two separate, sex-specific equations (more details on these equations are provided below).
Hodgdon equations from the 2019 Army Regulation (%BFHE) [19-20]:
Female: [163.205 x Log10 (waist + hip – neck)] – [97.684 x Log10 (height)] – 78.387
Male: [86.010 x Log10 (abdomen – neck)] – [70.041 x Log10 (height)] + 36.76
Taylor-McClung equations updated for use in 2023 Army Regulation (%BFTM) [21-22]:
Female: -9.15 – (0.015 x BW) + (1.27 x abdomen)
Male: = -26.97 – (0.12 x BW) + (1.99 x abdomen)
The Taylor-McClung equation [21] was developed as a simpler CBE for estimation of %BF with error equitable across as compared to %BFDXA, a research standard of measure [23]. The in vivo coefficient of variation in an external population for soft tissue and %BFDXA was 0.4 to 1.0% [24].
The TM equation was further examined herein to determine whether a built-in offset would more closely meet the needs of the Army while improving the %BF estimation to accurately categorize individuals as meeting the Army %BF standards. To do this, a 1% and 2.5% offset was built into the base TM Equation [21]. This allowed the %BF of all individuals to be slightly underestimated with the goal of reducing error where an individual would be falsely failed by the CBE method. In other words when there is error in categorizing an individual for meeting the Army %BF standards, it would be in favor of the Soldier.
Statistical analysis
The population was weighted to provide results that would represent a population with the same distribution of age, R/E, and sex as the current U.S. Army. This was done by classifying each individual into a combined age, R/E, and sex category. This weight was calculated using the following equation:
where each individual was assigned, a weight based on the proportion (Pr) of their specific age, R/E and sex category in both the study sample and in the entire Army. This weighting redistributed the distribution of age, R/E, and sex within the study population to be equal to that of the Army.
Weighted proportions were estimated for the pass and fail rates and the false pass and false fail rates for each of the circumferences-based equations using the AR 600-9 body composition standards. A false pass was indicated when an individual passed the AR 600-9 based on the CBE but should have failed if the %BFDXA was used. A false failure was indicated when an individual failed the AR 600-9 using the CBE but would have passed if %BFDXA was used. To estimate accuracy in the %BF as estimated by the CBEs and BIA, weighted root mean squared error were calculated using %BFDXA as the gold standard. Results were presented by R/E and sex, to evaluate potential differences in accuracy of the estimated %BF.
To evaluate the associations of body composition and PF on MSKI risk, logistic regression models were used. Separate models were run to evaluate how passing the Army %BF standard and how %BF as continuous variables effected MSKI risk. Additionally, models were built to evaluate how achieving three different thresholds of scores on the ACFT effected injury risk. The three ACFT thresholds evaluated were: 1) scoring above 80 points on all six events (ACFT80), 2) scoring above 90 points on all six events (ACFT90), and 3) scoring at least 80 points on all six events while simultaneously scoring a total score above 540 points, i.e., > 80 for enough events to accumulate 40 additional points, (ACFT540). ACFT models were adjusted for age and sex to account for the age and sex standardized scoring system of the ACFT. The goal of these models was to identify a threshold score on the ACFT that significantly protected against injury and had similar pass rates for both men and women. Once an ideal ACFT threshold was identified, individuals were classified as meeting that standard or not. A logistic regression model with an interaction term between meeting the identified ACFT threshold score and continuous %BFDXA was performed to evaluate if meeting the ACFT threshold modifies the association of %BF on MSKI risk. Approximately 20% of the participant pool had missing ACFT data. Sensitivity analysis was conducted, and with no bias was associated with missing ACFT data. To include all observations in the analysis, multiple imputation with 5 imputations were used for analyses conducted with ACFT data. Statistical analyses were conducted using R (v4.2.1; R Core Team 2022) and SAS statistical software (Version 9.4, SAS Institute Inc Cary, NC, USA 2023).