Prevalence and Factors Associated with Eating Disorders in Military First Line of Defense against COVID-19: A Cross-Sectional Study during the Second Epidemic Wave in Peru

Eating disorders (ED) encompass conditions such as anorexia nervosa, bulimia nervosa, and binge eating disorder that alter eating habits and adversely affect physical and mental health and relationships. Pre-pandemic global ED prevalence was estimated at 0.91%, with specific rates of AN (0.16%), BN (0.63%), and BED (1.93%). Military populations face unique risk factors for ED, including exposure to military-related trauma, PTSD, depression, substance use, and strict weight and fitness standards. The COVID-19 pandemic exacerbated mental health burdens and ED risk due to social restrictions, disrupted routines, sleep problems, and limited access to care. In Peru, military personnel were heavily involved in enforcing lockdown measures and supporting immunization, potentially increasing their vulnerability to mental health problems affecting eating behaviors. The study aimed to estimate the prevalence of ED symptoms and identify associated factors among military personnel in Lambayeque, Peru, during the second COVID-19 wave in 2021.

Prior research indicates elevated ED risk among military service members and veterans, linked to trauma exposure, PTSD, depression, and stringent weight standards. International reports during COVID-19 show increased ED symptoms due to routine disruptions, food access concerns, and fear of infection. Before the pandemic, binge eating was a notable issue in military groups, with risk factors including insomnia, anxiety, depression, and food insecurity. In Peru, evidence on military eating behaviors during the pandemic is scarce, with reports mainly noting high prevalence of stressors such as nervousness and sadness that could precede mental health problems.

Design and setting: Secondary analysis of an analytic cross-sectional observational study assessing mental health in frontline military personnel during the second COVID-19 wave. Data were collected 2–9 November 2021 in Lambayeque, Peru. Population and sample: The target population comprised 820 military personnel engaged in COVID-19 defense activities in Lambayeque. The primary study estimated a required sample of 582 (expected prevalence 12.8%, 99% confidence, 2.5% precision, +20% for losses). A convenience sample of 710 (86.6% of the population) was enrolled. For this analysis, 550 participants who completed the EAT-26 were included; those with missing variables of interest were excluded. Outcome: Eating disorder symptomatology measured with the Spanish EAT-26 (26 items; 6-point response options scored 0–3; three subscales: Diet, Bulimia/Food preoccupation, Oral control). Cutoff ≥20 indicated positive ED symptomatology requiring further clinical evaluation. Psychometrics: sensitivity 88.9%, specificity 97.7%; Cronbach’s alpha in this study 0.93. Covariates and instruments: - Insomnia: Insomnia Severity Index (ISI; 7 items, 0–28; cutoff ≥8); alpha 0.88. - Food insecurity: Household Food Insecurity Access Scale (HFIAS; 9 items; three domains); alpha 0.87. - Physical activity: IPAQ-Short (9 items; categories: intense, moderate, mild, inactive); alpha 0.64. - Resilience: CD-RISC-10 (0–40; classification around score 30); alpha 0.97. - Fear of COVID-19: FCV-19S (7 items, 1–5; cutoff 16.5); alpha 0.94. - Burnout: Maslach Burnout Inventory (22 items; dimensions: emotional exhaustion, depersonalization, personal accomplishment); alpha 0.91. - Anxiety: GAD-7 (0–21; categories none/mild/moderate/severe); alpha 0.93. - Depression: PHQ-9 (0–27); alpha 0.92. - PTSD: PTSD Checklist–Civilian Version (PCL-C; 17 items, cutoff 43); alpha 0.95. - Sociodemographic/occupational/psychosocial variables: age, gender, marital status, religion, children, frequent alcohol and tobacco use, comorbidities (hypertension, diabetes), BMI category, work time in frontline (1–6, 7–12, 13–18, ≥19 months), personal/family mental health history, help-seeking for mental health during COVID-19, trust in government handling of COVID-19. Procedures: After ethics and military authorization, supervised, in-person group sessions in ventilated settings with distancing, masks, and hand hygiene were conducted. Participants completed a REDCap-based questionnaire after providing electronic informed consent. Participation was voluntary without coercion. Statistical analysis: Data were exported from REDCap and analyzed in Stata 17. Descriptive statistics summarized variables (frequencies for categorical, appropriate central tendency/dispersion for continuous). Bivariate associations between ED symptoms and covariates were examined using chi-square tests. Generalized linear models (Poisson family with log link) estimated crude and adjusted prevalence ratios (PR) with 95% CIs. Variables significant in simple models (p<0.05) were included in multiple regression. Collinearity among covariates was assessed. Ethics: Approved by the Ethics Committee of Universidad San Martin de Porres. Data were anonymized; informed consent obtained; principles of the Declaration of Helsinki observed.

Sample characteristics (n=550): 95.5% male; median age 22 years. Frequent alcohol and tobacco use reported by 17.6% and 6.7%, respectively. Hypertension 9.6%. BMI: overweight 33.6%, obesity 6.8%. Work time ≥19 months reported by 36.9%. Food insecurity affected 48.7%. Low physical activity 11.6%. Burnout syndrome 9.3%. PTSD 7.5%. Fear of COVID-19 19.2%. Suicidal risk 14%. ED symptoms prevalence was 10.2% (56/550). EAT-26 item signals: 19.1% always exercise a lot to burn calories; 14.2% always consider calories in foods; 10% cut food into small pieces. Bivariate associations with ED symptoms: Significant associations for labor time (p=0.022), insomnia (p<0.001), fear of COVID-19 (p<0.001), burnout syndrome (p=0.005), depression (p=0.009), anxiety (p<0.001), and PTSD (p<0.001). Simple regression PRs (selected): labor time 7–12 months PR 3.79 (95% CI: 1.51–9.49); 19+ months PR 2.79 (1.17–6.65); insomnia PR 2.71 (1.67–4.42); fear of COVID-19 PR 3.36 (2.05–5.49); burnout PR 2.39 (1.32–4.33); anxiety PR 2.50 (1.53–4.09); PTSD PR 4.14 (2.47–6.92). (Depression also showed a positive crude association.) Multiple regression (adjusted): ED symptoms were more prevalent among those with work time 7–12 months PR 2.97 (1.24–7.11) and ≥19 months PR 2.62 (1.11–6.17), fear of COVID-19 PR 2.20 (1.26–3.85), burnout syndrome PR 3.73 (1.90–7.33), and PTSD PR 2.97 (1.13–7.83). Insomnia, anxiety, and depression were not significant after adjustment.

The study addressed the prevalence of eating disorder symptoms and their correlates among frontline military personnel during the second COVID-19 wave in Peru. Approximately one in ten service members screened positive for ED symptoms. Longer duration working on the COVID-19 frontline, heightened fear of COVID-19, burnout syndrome, and PTSD were independently associated with higher ED symptom prevalence. These findings align with prior literature linking occupational stressors, trauma exposure, and pandemic-related fears to disordered eating. While insomnia, anxiety, and depression showed crude associations, they did not remain significant in the adjusted model, suggesting overlapping effects with PTSD, burnout, and pandemic-related fear. The results underscore the need for integrated mental health and occupational interventions targeting high-risk subgroups within the military to mitigate disordered eating behaviors exacerbated by prolonged pandemic duties and psychosocial stressors.

About 10% of frontline military personnel in Lambayeque, Peru, exhibited eating disorder symptoms during the second COVID-19 wave. Higher prevalence was associated with longer frontline service time, fear of COVID-19, burnout syndrome, and PTSD. Military institutions should prioritize screening and preventive interventions for EDs within comprehensive mental health strategies that address trauma, burnout, and pandemic-related fears. Future research should include longitudinal designs, broader geographic samples, and additional psychosocial variables (e.g., body image, social media use, self-esteem) to clarify causal pathways and inform targeted prevention and treatment.

- Cross-sectional design precludes causal inference. - Potential nonresponse bias due to voluntary participation and variable motivation, possibly leading to over- or under-reporting. - Membership (cluster) bias related to hierarchical subgroups within the military that may share attributes influencing outcomes. - Information bias from self-reported measures. - EAT-26 screens for ED risk but does not establish clinical diagnosis; positive screens require professional evaluation. - Unmeasured confounders (e.g., race, sexual trauma, social media use, self-esteem, body satisfaction) due to secondary data constraints. - Selection bias: participants were from a single region (Lambayeque), limiting generalizability to other Peruvian military populations. - Some discrepancies in sample counts reported across sections of the study documentation.