Study population
Launched in 2001, the Millennium Cohort Study is an ongoing prospective cohort study designed to investigate health effects associated with military service and has been previously described [25]. Briefly, four panels of US military personnel were enrolled in 2001, 2004, 2007, and 2011, leading to a total current enrollment of 201,619 from all service branches (Army, Navy, Marine Corps, Air Force, Coast Guard) and components (active duty, Reserve, National Guard). Enrolled participants are followed via self-administered surveys approximately every 3 years, regardless of military status at the time of follow-up. Baseline and follow-up surveys collected information on service-related experiences and behavioral, physical, and mental health. Data on demographic and military service-related characteristics were also obtained from electronic personnel files maintained by the Defense Manpower Data Center (DMDC). The study protocol was approved by the Naval Health Research Center Institutional Review Board (NHRC.2000.0007).
We initially restricted the analysis to 123,761 participants who were serving on active duty when they completed the baseline survey, due to the fact that active duty personnel have comprehensive access to the Military Health System during their time in service, while access to this system is limited for Reserve and National Guard personnel. We further excluded participants with evidence of autoimmune disease prior to baseline, either by self-report or from medical encounter data (n = 2221). After further excluding participants who were missing data on demographics, history of PTSD, or history of other mental health conditions (n = 968), the present analysis included 120,572 participants.
Ascertainment of selected autoimmune diseases
The primary endpoints were RA, SLE, MS, IBD, and a composite outcome of any of the selected autoimmune diseases. Incident cases of selected autoimmune diseases were identified from medical encounter records in the Military Health System Data Repository (MDR) using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes. Cases were defined as service members with one inpatient or at least two outpatient encounters with the corresponding ICD-9-CM code: 714 for RA; 710.0 for SLE; 340 MS; and 555 or 556 for IBD. The date of diagnosis was assigned as the earliest medical encounter noting the ICD-9-CM code of interest.
Assessment of PTSD and covariates
PTSD and other mental health conditions were assessed at all available time points. Participants were classified as having a history of PTSD if they 1) self-reported receiving a health care provider’s diagnosis of PTSD, or 2) screened positive using the PTSD Checklist−Civilian Version (PCL-C). The PCL-C is a validated instrument used to rate the severity of 17 PTSD symptoms [26]. Using criteria established in the Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV-TR), participants were classified as having a positive screen for PTSD if they reported a moderate or greater level of at least one intrusion symptom, three avoidance symptoms, and two hyperarousal symptoms [27].
Participants were classified as having a history of another mental health condition, with the exception of PTSD, if they 1) self-reported receiving a health care provider’s diagnosis of depression, schizophrenia, or bipolar disorder, or 2) screened positive for major depression, panic disorder, or other anxiety disorders using standardized Patient Health Questionnaire (PHQ) algorithms [28,29,30].
Age, sex, race/ethnicity, service branch, and pay grade were obtained from DMDC at the time of enrollment. Deployment in support of operations in Iraq and Afghanistan was determined based on dates in and out of theater from the Contingency Tracking System maintained by DMDC. Combat experience was defined as reporting at least one of the following: personally witnessing or being exposed to a person’s death due to war or disaster; witnessing physical abuse (torture, beating, rape); dead or decomposing bodies; maimed soldiers or civilians; or prisoners of war or refugees. Deployment dates in combination with the combat variable were used to assess if participants experienced no deployments, deployments without combat, or deployments with combat.
Data from all available study time points were used to ascertain height, weight, smoking status, alcohol intake, and prior physical/sexual trauma. Body mass index (BMI) was calculated as self-reported weight divided by the square of self-reported height (kilograms ÷ meters squared). Participants were classified as “never” smokers if they did not report that they had smoked at least 100 cigarettes (5 packs) in their lifetime; smokers were classified as former smokers if they reported having quit successfully or not having smoked in the past year. Heavy drinking was defined as exceeding recommended weekly limits of more than 14 and 7 alcoholic drinks per week for men and women, respectively. Alcohol misuse was measured as an affirmative response to any problem drinking item on the PHQ (e.g., “You drove a car after having several drinks or after drinking too much”) [28, 29]. Prior physical or sexual trauma was assessed based on any positive endorsement at baseline or follow-up to three items (e.g., “suffered forced sexual relations or sexual assault”, “experienced sexual harassment”, “suffered a violent assault”).
Statistical analysis
Participants accrued follow-up time from the date of baseline survey completion to the date of diagnosis of the selected autoimmune disease (depending on the outcome of interest), separation from active duty status, or the end of the follow-up period (September 30, 2015), whichever occurred first. Multivariable hazard ratios (HR) and 95% confidence intervals (CI) were estimated with Cox proportional hazards models and adjusted for age, sex, enrollment panel, race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, other), pay grade (enlisted or officer), service branch (Army, Navy, Marine Corps, Air Force), and history of another mental health condition. In sensitivity analyses, models were further adjusted for the following health behaviors: smoking status (never, former, or current), heavy or problem drinking (yes, no), and BMI (< 25, 25–29, ≥30 kg/m2). History of PTSD and other mental disorders, educational attainment, marital status, active duty status, military service branch, military rank, combat and deployment status, smoking status, BMI, alcohol intake, and prior physical/sexual trauma were updated as time-dependent variables in the analysis.
All analyses were stratified by sex due to known differences in autoimmune disease incidence between men and women [31]. Additional analyses were stratified by prior combat deployment experience and by prior physical or sexual trauma. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).