Deployment and Preterm Birth Among United States Army Soldiers

American Journal of Epidemiology Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US. Vol. 187, No. 4 DOI: 10.1093/aje/kwy003 Advance Access publication: January 12, 2018 Original Contribution Deployment and Preterm Birth Among US Army Soldiers * Correspondence to Dr. Jonathan G. Shaw, Division of Primary Care and Population Health, School of Medicine, Stanford University, 1265 Welch Road, Stanford, CA 94305 (e-mail: jgshaw@stanford.edu). Initially submitted August 5, 2017; accepted for publication January 3, 2018. With increasing integration of women into combat roles in the US military, it is critical to determine whether deployment, which entails unique stressors and exposures, is associated with adverse reproductive outcomes. Few studies have examined whether deployment increases the risk of preterm birth; no studies (to our knowledge) have examined a recent cohort of servicewomen. We therefore used linked medical and administrative data from the Stanford Military Data Repository for all US Army soldiers with deliveries between 2011 and 2014 to estimate the associations of prior deployment, recency of deployment, and posttraumatic stress disorder with spontaneous preterm birth (SPB), adjusting for sociodemographic, military-service, and health-related factors. Of 12,877 deliveries, 6.1% were SPBs. The prevalence was doubled (11.7%) among soldiers who delivered within 6 months of their return from deployment. Multivariable discrete-time logistic regression models indicated that delivering within 6 months of return from deployment was strongly associated with SPB (adjusted odds ratio = 2.1, 95% confidence interval: 1.5, 2.9). Neither multiple past deployments nor posttraumatic stress disorder was significantly associated with SPB. Within this cohort, timing of pregnancy in relation to deployment was identified as a novel risk factor for SPB. Increased focus on servicewomen’s pregnancy timing and predeployment access to reproductive counseling and effective contraception is warranted. military deployment; posttraumatic stress disorder; pregnancy; preterm birth; servicewomen; United States Army Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; PTSD, posttraumatic stress disorder; SPB, spontaneous preterm birth; VA, Department of Veterans Affairs. premature child (7). The majority of preterm births are spontaneous (7, 8). While prediction and prevention remain elusive, the etiology of preterm birth is increasingly understood to be linked to psychological, physical, and environmental stressors, with maternal-fetal impacts varying by time of exposure (7, 9– 14). Various pathways have been hypothesized, including immunological, vascular, or neuroendocrine alterations triggered by such stressors or via stress-related health behaviors (7, 15). Given the unique exposures that female soldiers encounter during deployment and return from deployment, it is critical to determine whether there are associated increases in the risk of preterm delivery. Only a few studies of births taking place among servicewomen during or after active-duty deployment have been done, with generally reassuring results. However, one study was quite small (16), and the others were conducted a decade ago (17, 18); thus, the extant literature is unable to address the contemporary The US military has witnessed an important demographic shift over the past 2 decades as servicewomen have increasingly engaged in combat-related activity (1). As the current generation of young women takes on a growing role in wartime service, with recent policy changes expanding the role of women in combat (2, 3), not only are they exposed to unique physical and environmental health risks (4) but many return with trauma-related mental health issues (5, 6). While female soldiers start at a high level of fitness, little is known about the potential adverse reproductive impacts of military-specific exposures and stressors when deployed—for example, in support of recent conflicts in Afghanistan and Iraq. Preterm birth (defined as delivery before 37 weeks of gestation) is a critical health outcome, affecting approximately 12% of deliveries in the United States. It is a leading cause of infant morbidity and mortality, and is costly in terms of both medical expense and the burden to mother and family of caring for a 687 Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 Jonathan G. Shaw*, D. Alan Nelson, Kate A. Shaw, Kelly Woolaway-Bickel, Ciaran S. Phibbs, and Lianne M. Kurina 688 Shaw et al. METHODS We performed a retrospective cohort study using the Stanford Military Data Repository, which comprises deidentified medical and administrative data from official Department of Defense sources (see Web Table 1, available at https://academic.oup. com/aje, for details). The data set included month-level information for all soldiers who were on active duty in the US Army anytime between January 2011 and December 2014. We identified all live births occurring among active-duty women during this period and restricted the analysis to those with a minimum of 12 months of data prior to delivery, to ensure sufficient enlistment and health assessment exposure data. outside of the combat theater. We modeled antenatal deployment history with a binary indicator for any prior deployment as well as a categorical variable reflecting number of deployments (0, 1, 2 or ≥3). We also included a categorical variable for recency of return from deployment (“redeployment”), examining thresholds of both ≤6 months and ≤12 months from redeployment to delivery. Delivery within 6 months of redeployment would imply not only perideployment pregnancy but also, with rare exceptions, that conception occurred during deployment. Delivery 6–12 months after redeployment was generally expected to identify women who conceived shortly after returning from deployment. We identified PTSD with the ICD-9-CM code 309.81 in the health encounter data (inpatient and outpatient, within military facilities or purchased care in civilian facilities). The Army routinely screens for PTSD during primary care. The Primary Care PTSD Screen and follow-on PTSD Checklist used in Army clinics are confirmed to be valid screening measures in active-duty service members (26). We chose to include all clinician-assigned diagnoses of PTSD as clinically relevant for the detection of adverse reaction to trauma, given that primary care is the first line of screening and is universally and routinely accessed by soldiers, including for annual and predeployment assessments. We categorized PTSD status with a 3-level categorical variable: “no PTSD,” “current PTSD” (diagnosis recorded in the 12 months prior to delivery), or “history of PTSD” (diagnosis recorded in older clinical encounters but not present in encounters 12 months prior to delivery), based on prior evidence (6, 27) indicating that current PTSD is more likely to be related to preterm birth. Key covariates and potential confounders Primary outcome: spontaneous preterm delivery We focused on spontaneous (as opposed to medically indicated) preterm births, as they represent the majority of preterm births, are most pertinent to evaluating associations between potential unique deployment-related stressors and premature onset of labor and delivery (15), and are consistently identifiable in maternal data. We were unable to examine medically indicated preterm deliveries. Our Army data did not have infant records, and there is no specific maternal claims code for indicated preterm birth. Rather, study of indicated preterm birth typically requires direct chart review (22) or linkage to infant birth certificates (23–25) to determine prematurity, in conjunction with specific maternal procedures (e.g., cesarean delivery or induction of labor) to infer induction prior to term. We identified our primary outcome using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code 644.2 (“spontaneous onset of delivery before 37 weeks”) from digitally recorded inpatient data. Primary predictors: deployment and PTSD status We used longitudinal administrative data from the Defense Manpower Data Center to identify the number and timing of deployments prior to delivery. The Department of Defense’s administrative definition of deployment includes, for some soldiers, support of combat operations geographically located Data on all demographic variables came from official military personnel records. We included maternal age at the time of delivery (categorized as 19–21, 22–24, 25–29, 30–34, 35–39, or ≥40 years), marital status (married, never married, or formerly married), and race (white, black, Asian/Pacific islander, Native American, or other/unknown) in all models. Race was self-reported and was included because it is a well-established predictor of preterm birth (7). Information on race was reported as missing or “other” for less than 6% of our cohort. Socioeconomic covariates included attained educational level and military pay grade at the time of delivery. We identified obstetrical risk factors, including nonsingleton gestations, from the obstetrical inpatient data using validated ICD-9-CM codes (28), and we identified multiparity using data on soldiers’ claimed dependents as a proxy. We extracted information on preexisting hypertension and asthma from predelivery medical encounter data; these conditions are reasonably prevalent (>2%) and have been suggested as risk factors for preterm birth (7, 29–32). Diabetes was not included, because it frequently precludes military service and was extremely rare in our cohort. Body mass index (weight (kg)/height (m)2) was included and was based on prepregnancy height and weight. Antenatal alcohol use (defined as selfreported use more than 2–3 times per week) and tobacco use were identified from health encounter data, using the last known status reported by the patient prior to the delivery. Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 military climate. In a recent report from the Millennium Cohort Study, a study of military personnel, Ippolito et al. (19) found no evidence of increased infertility or miscarriage for deployed servicewomen, but they did not examine live births. In terms of mental health exposures, prior research carried out within the Department of Veterans Affairs (VA) health-care system showed that posttraumatic stress disorder (PTSD) is both highly prevalent among veterans and a predictor of spontaneous preterm birth (SPB) (6). However, the VA provides care only after the end of military service and only to a fraction of female veterans (20, 21). To our knowledge, no studies of PTSD in relation to preterm birth have been conducted among the activeduty component of servicewomen. Our study aims were therefore to 1) evaluate whether prior deployment(s) and/or recent return from deployment independently predicts SPB and 2) test for an association between PTSD diagnosis and SPB among active-duty soldiers. Preterm Birth in Recently Returned Soldiers 689 Statistical analysis RESULTS Among the 12,877 deliveries occurring during active service, the prevalence of SPB was 6.1% (n = 785). Half of all mothers had prior deployment, and 21% had multiple prior deployments; these percentages were reasonably similar for women with and without SPB (Table 1). Of the SPBs, 16.3% occurred among servicewomen who had returned from deployment during the year prior to delivery, versus 13.3% of all other births; 6.4% of SPBs occurred within 6 months of the servicewoman’s return from deployment, versus 3.1% of all other births. Approximately 4% of soldiers delivering during the study period had prior PTSD diagnoses, three-quarters of which Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 Using discrete-time logistic regression to adjust for sociodemographic, obstetrical, and health covariates, including PTSD, we created 3 models to evaluate the association of deployment status with SPB. In model 1, we tested the association of number of prior deployments with SPB. In model 2, we tested recency of return from deployment at the time of delivery. In model 3, we tested the joint effects of prior deployment and recency of last deployment with a multilevel predictor variable combing 2 levels of recency of return (≤6 months or >6 months) and 4 levels of number of prior deployments (0, 1, 2, or ≥3). For our second aim, we examined PTSD’s associations with SPB in all of the above-described models. We also utilized a fourth model that did not adjust for deployment (model 4) and compared the findings for PTSD with those in models 1–3 to determine whether there was any evidence of an association with PTSD being attenuated by adjustment for deployment. Post hoc, we performed descriptive analysis of the subpopulation identified as being at highest risk, namely those delivering ≤6 months after return, comparing them with women who had longer postdeployment intervals before delivery. In a set of sensitivity analyses, we first tested how the odds ratio for deployment status varied when using a 12-month (instead of 6-month) threshold for “recent return” in model 3. We also tested whether excluding PTSD as a predictor in models 1–3 affected the odds ratio for deployment status, given that PTSD is hypothetically on the causal pathway through which deployment could affect preterm delivery. We also reran the analyses in our main model on the subpopulation of singleton pregnancies—that is, excluding twins/higher-order births—to see whether the associations of interest were sustained in a population at lower overall risk of preterm birth. Finally, we ran the analysis only on the subpopulation of women with any prior deployment to eliminate concern about possible systematic differences (16) between women who have deployed and those who have not. For this subcohort, we additionally included a binary variable for deployment location denoting whether a soldier’s most recent deployment was to Iraq/Afghanistan (as a proxy for risk of direct exposure to combat) or another locale (predominantly Kuwait, Qatar, United Arab Emirates, or Kyrgyzstan). We used Stata MP, version 14.2 (StataCorp LP, College Station, Texas) for all analyses. All reported P values are 2-sided. Regression results are expressed as adjusted odds ratios with 95% confidence intervals. were current PTSD diagnoses, and the prevalences were similar for women with and without spontaneous preterm deliveries (Table 1). Significant differences in race, hypertension, and multiple gestation were observed among women with and without SPB (all P’s < 0.001). The prevalence of SPB was slightly higher (7.5%) among women with 3 or more prior deployments and was particularly common (11.7%) among soldiers delivering within 6 months after return from deployment (Table 2). It was also higher among women with a history of PTSD (9.5%). Adjusting for all covariates and PTSD status (Table 3), model 1 suggested a positive stepwise association between number of deployments and SPB (for 1, 2, and ≥3 deployments, adjusted odds ratio (aOR) = 1.1 (95% confidence interval (CI): 0.9, 1.3), aOR = 1.3 (95% CI: 1.0, 1.6), and aOR = 1.4 (95% CI: 1.0, 2.0), respectively; test for trend: P = 0.05), but the elevation in risk was significant only for those with 3 or more deployments. When we modeled the 3 levels of recency of return from deployment (>12 months, 7–12 months, and ≤6 months), only those women returning most recently were at significantly increased risk (aOR = 2.1, 95% CI: 1.5, 2.9) compared with those with no prior deployments (model 2). Combining number of deployments and recency into a joint variable (model 3) demonstrated that repeated prior deployments were not associated with SPB among women with longer intervals between deployment and delivery (aORs ranged from 1.0 to 1.2; all P’s ≥ 0.3) but that recent return from deployment (≤6 months from return from deployment to delivery) was consistently associated with increased risk, regardless of number of prior deployments. The findings also indicated systematically higher odds of SPB with increasing numbers of deployments among those servicewomen who had recently returned (for 1, 2, and ≥3 deployments, aOR = 1.6 (95% CI: 1.1, 2.5), aOR = 2.7 (95% CI: 1.5, 4.9), and aOR = 3.8 (95% CI: 1.8, 7.8), respectively). (See Web Table 2 for full model details.) In all models, regardless of whether we did or did not adjust for deployment history, neither current PTSD nor historical PTSD was predictive of spontaneous preterm delivery (Table 3 and Web Table 3). Notably, among the covariates in our fully adjusted regression model, black race (versus white) remained a significant predictor of SPB (aOR = 1.4, 95% CI: 1.2, 1.7) (model 3; see Web Table 2 for full details). Women in the lowest pay grade (junior enlisted ranks) had a similar elevation in odds (aOR = 1.4, 95% CI: 1.0, 1.9). Among health status covariates, only the lowest body mass index category (<20 (underweight)) and history of preexisting hypertension were associated with SPB (aOR = 1.9 (95% CI: 1.0, 3.4) and aOR = 2.4 (95% CI: 1.7, 3.6), respectively). Post hoc analysis of the 428 soldiers with pregnancies proximate to deployment (delivery ≤6 months after return) showed that 317 (74%) were recorded as having been deployed during the 7th–10th calendar months preceding the month of their delivery, suggesting that conception occurred during deployment for the majority of those individuals (Web Table 4). These women were more likely to be in the youngest age categories (45% vs. 17.5% were aged 19–24 years; P < 0.001), to be unmarried, and to have a low pay grade and educational level compared with women whose postdeployment deliveries occurred more than 6 months after redeployment. Notably, 690 Shaw et al. Table 1. Sample Characteristics of US Army Women With Births During Active Duty, 2011–2014 Spontaneous Preterm Delivery Status Characteristic Total Population (n = 12,877) No. of Births SPB (n = 785) Not SPB (n = 12,092) % No. of Births % No. of Births P Valuea % Prior deployment and PTSD status No. of prior deployments 0.12 6,451 50.1 382 48.7 6,069 50.2 1 3,767 29.3 216 27.5 3,551 29.4 2 1,798 14.0 122 15.5 1,676 13.9 861 6.7 65 8.3 796 6.6 ≥3 Recency of return from deployment, months since return <0.001 >12 4,686 36.4 275 35.0 4,411 36.5 7–12 1,312 10.2 78 9.9 1,234 10.2 428 3.3 50 6.4 378 3.1 Current PTSD 361 2.8 20 2.5 341 2.8 History of PTSD 126 1.0 12 1.5 114 0.9 12,390 96.2 753 95.9 11,637 96.2 ≤6 PTSD statusb 0.25 No PTSD Demographic characteristics Maternal age, yearsc,d Marriede 27.3 (5.2) 27.4 (5.4) 27.2 (5.2) 0.38 9,148 71.0 545 69.4 8,603 71.1 6,523 50.7 341 43.4 6,182 51.1 Black 4,647 36.1 342 43.6 4,305 35.6 Asian/Pacific Islander or Native Americanf 1,000 7.8 60 7.6 940 7.8 707 5.5 42 5.4 665 5.5 Race 0.37 <0.001 White Other/unknown Army rank 0.15 Junior enlisted E1–E3 2,048 15.9 148 18.9 1,900 15.7 E4 5,036 39.1 284 36.2 4,752 39.3 E5–E6 3,031 23.5 185 23.6 2,846 23.5 E7–E9 529 4.1 33 4.2 496 4.1 140 1.1 13 1.7 127 1.1 1,572 12.2 94 12.0 1,478 12.2 521 4.0 28 3.6 493 4.1 High school 8,235 64.0 502 63.9 7,733 64.0 Some college 1,800 14.0 119 15.2 1,681 13.9 Bachelor’s degree 1,805 14.0 98 12.5 1,707 14.1 982 7.6 63 8.0 919 7.6 Noncommissioned officer Warrant officer (W1–W5) Commissioned officer O1–O3 O4–O10 g Educational level Graduate degree 0.78 Obstetrical profile Twins/higher-order gestation Prior childrenh 298 2.3 90 11.5 208 1.7 <0.001 4,321 33.6 232 29.6 4,089 33.8 <0.001 Table continues Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 0 Preterm Birth in Recently Returned Soldiers 691 Table 1. Continued Spontaneous Preterm Delivery Status Characteristic Total Population (n = 12,877) No. of Births % SPB (n = 785) No. of Births Not SPB (n = 12,092) % No. of Births P Valuea % Prior health status Prepregnancy body mass indexc,i Tobacco usej 25.1 (3.3) 24.9 (3.4) 25.1 (3.3) 0.10 30.7 237 30.2 3,713 30.7 0.76 Alcohol use more than 2–3 times per weekj 618 4.8 32 4.1 586 4.8 0.33 Hypertensionk 272 2.1 37 4.7 235 1.9 <0.001 Asthma 611 4.7 47 6.0 564 4.7 0.09 Abbreviations: PTSD, posttraumatic stress disorder; SPB, spontaneous preterm birth. a P values are 2-sided and based on χ2 tests for categorical variables and t tests for continuous variables. b Current PTSD = diagnosis code present in clinical encounters within 1 year prior to delivery; history of PTSD = diagnosis code present in prior clinical encounters, but none within 1 year prior to delivery. c Values are expressed as mean (standard deviation). d Age range was restricted to 19–52 years; for regression analysis, age was treated as categorical: 19–21, 22–24, 2529–, 30–34, 35–39, or ≥40 years). e For regression analysis, marital status was treated as categorical: married, never married, or formerly married. f Two racial categories were combined for display purposes only (not analysis) in order to suppress cell sizes with n < 10. g Data on educational level were missing for 55 participants. h Estimate was based on any children claimed as dependents. i Weight (kg)/height (m)2; based on the most recent data available more than 9 months prior to delivery. Data were missing for 453 participants. j Last known status reported by the patient; based on health encounter reports. k Preexisting hypertension; based on non-pregnancy-related International Classification of Diseases, Ninth Revision, Clinical Modification codes. they were not significantly different in terms of PTSD prevalence or number of prior deployments. In the multiple sensitivity analyses we performed (results not shown), the main findings remained robust. Utilizing a 12-month threshold for “recent return” in model 3, we found similar patterns but with attenuated joint effects of recent return and deployment counts, supporting the finding that the association with SPB was most strongly present in women who delivered within 6 months of return. Neither removing adjustment for PTSD nor restricting the analysis to singleton pregnancies materially changed the results. Finally, restricting the analysis to women with any prior deployment (half of our cohort) reconfirmed the association observed in the overall cohort. Additionally, among this subcohort, deployment to Afghanistan/Iraq (as a proxy for likely combat exposure) versus other locales was not itself associated with SPB (aOR = 0.8, 95% CI: 0.7, 1.1). DISCUSSION In this analysis of over 12,000 live births taking place among recently serving active-duty female US Army soldiers, we observed that proximity to deployment—but not the simple presence of prior deployment—was associated with an increased risk of SPB. Specifically, pregnancies that probably began during deployment (as indicated by birth within 6 months of return) Am J Epidemiol. 2018;187(4):687–695 were twice as likely to end in SPB. The overall rate of SPB we observed in our cohort (6.1%) was slightly lower than that reported in the general US population (8.1%) (calculated using the equivalent approach in the Nationwide Inpatient Sample, a cohort with the same mean age as ours) (33), which is consistent with the expectation that soldiers represent a healthy, low-risk population. Also, as expected, the SPB rate in our cohort was moderately lower than the total preterm birth rate (spontaneous and medically indicated) of about 7.5% reported in the 2 large cohort studies of servicewomen from the last decade (17, 18). Thus, it was all the more striking to find this subset of soldiers— those just recently returned from deployment—for whom 11.7% of deliveries were spontaneously premature. The concerns raised by these findings are heightened in the context of prior research documenting high rates of unintended pregnancy in the military (34) and emerging evidence that the most reliable forms of contraception (long-acting reversible contraceptives) are underutilized in the Army, especially around the time of deployment (35–37). Pregnancy presents unique consequences in the military, because pregnant soldiers cannot deploy and pregnancies occurring in-theater result in mandatory evacuation (34). Such pregnancies incur high costs both financially and in terms of military readiness. Our work now suggests that pregnancy among those just returned from deployment may inflict further hardship (7) in the form of preterm birth. Disparities in the use of effective contraceptives exist for young, lower-ranking Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 3,950 692 Shaw et al. Table 2. Unadjusted Prevalence of Spontaneous Preterm Delivery by Deployment and Posttraumatic Stress Disorder Status in US Army Women With Births During Active Duty, 2011–2014 Deployment and PTSD Status Total No. of Births SPB Deliveries % 785 6.1 0 6,451 382 5.9 1 3,767 216 5.7 2 1,798 122 6.8 861 65 7.5 >12 4,686 275 5.9 7–12 1,312 78 5.9 428 50 11.7 Current PTSD 361 20 5.5 History of PTSD 126 12 9.5 12,390 753 6.1 No. of prior deployments ≥3 Recency of return from deployment, months since return ≤6 PTSD statusa No PTSD Abbreviations: PTSD, posttraumatic stress disorder; SPB, spontaneous preterm birth. a Current PTSD = diagnosis code present in clinical encounters within 1 year prior to delivery; history of PTSD = diagnosis code present in prior clinical encounters, but none within 1 year prior to delivery. servicewomen (35, 36)—the same subpopulation disproportionately represented among the 428 servicewomen we observed to have pregnancies proximate to deployment. The 2-fold risk of SPB we observed in relation to potentially modifiable timing of pregnancy, in an otherwise young and fit Army population with universal access to health care, suggests the need for an increased focus on servicewomen’s access to reproductive counseling and the elimination of any barriers to accessing contraception, especially around the time of deployment. The observed SPBs proximate to deployment were likely due to conceptions that occurred during deployment, including during leave taken near the end of a deployment. However, some women may have deployed with early, unrecognized pregnancies or, alternatively, if the birth was very preterm, may have conceived immediately upon return. In any case, perideployment conceptions are inevitably subject to unique stressors, exposures, and risks, and our findings strongly implicate the period surrounding redeployment as a high-risk interval during which to conceive. The association of SPB with recent return from deployment seemed especially marked—a 3-fold risk—among women with a history of multiple deployments, suggesting that there may be an additive affect. One plausible interpretation is that with each subsequent redeployment, the stress incurred by transitioning (e.g., from combat back to noncombat life) is magnified and exerts greater stress on the maternal-fetal system. In addition to (or instead of) direct stress effects, our results could indicate that other effects of deployment—for example, sleep deprivation Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 No. 12,877 Overall (38, 39)—could have important short-term negative effects on pregnancy. Alternatively, environmental or man-made exposures that soldiers experience in preparation for and/or during deployment—for example, environmental pollutants (13), pesticides (40), or biological immunizations and prophylactic medications—may have transient adverse effects on pregnancy. However, studies to date of servicewomen exposed to burn pits or smallpox and anthrax vaccination have not demonstrated increased preterm birth risk (18, 41, 42). Other deployment-related health factors may be alternative pathways to increased risk. For example, deployed service members may be at increased risk of genitourinary infections due to adverse hygiene and toileting conditions, and they may also be more vulnerable to sexually transmitted infection due to lack of availability or adherence to condom use during deployment (4). Regardless, our reassuring results that multiple deployments do not increase SPB risk in the absence of a recent return from deployment suggest that the increased risk observed among those with recent deployments is not the result of acquired chronic effects of exposures. Interestingly, we found no indication that PTSD lies on the causal pathway from recent deployment to SPB. Contrary to what was previously demonstrated in a VA cohort, we found no association between active PTSD and preterm birth (6). The simplest explanation may be that the VA population consists of soldiers seeking health care for problems such as PTSD after discharge from service, and therefore represents a very different study denominator. The selected subset of military personnel cared for by the VA might have more severe or refractory cases of PTSD than do women in the total active-duty population. It is also possible that any true association of PTSD with SPB, should it exist in servicewomen, could have been attenuated by our reliance on ICD-9-CM codes. Our study had several limitations. First, because we only had access to maternal records, we lacked infant-specific gestational age estimates. This meant that we could not quantify prematurity in terms of severity, nor could we explore associations with the minority of preterm births that are medically indicated, since identifying induced deliveries as preterm requires linking maternal records to detailed gestational age data. However, SPBs (the result of preterm labor or premature preterm rupture of fetal membranes) account for the majority of preterm births in developed countries (7) and were thus an appropriate primary focus. Moreover, we externally validated our use of ICD-9-CM code 644.2 to identify SPB, examining its accuracy in obstetrical discharge claims linked to birth certificate data in a combined state database (2000–2012; n = 7 million) maintained by the California Office of Statewide Health Planning and Development. We found that the maternal code was a reliable indicator of preterm birth (79.7% specific to <37 weeks of gestation and 88.7% specific to <38 weeks of gestation, as reported on the infant birth certificate) associated with a median gestation of 35 weeks. Likewise, comparing our ICD-9-CM code alone with a complex algorithm (43) developed for the California database to subtype preterm births into the categories 1) “spontaneous,” 2) “medically indicated,” or 3) “unknown,” we found that the maternal code identified subtype 1 with a sensitivity of 89.4% and a specificity of 99.3%. Still, our reliance on this single code to identify SPB inevitably entailed some misclassification. However, we expect that it would have been nondifferential Preterm Birth in Recently Returned Soldiers 693 Table 3. Adjusteda Odds of Spontaneous Preterm Birth According to Deployment and Posttraumatic Stress Disorder Status in US Army Women With Births During Active Duty, 2011–2014 Predictor No. of Births Model 1b aOR 95% CI Model 2c aOR 95% CI Model 3d aOR 95% CI 1.00 Referent No. of prior deployments 0 1.00 Referent 1 1.06 0.88, 1.29 2 1.26 0.97, 1.62 ≥3 1.41 1.01, 1.96 No prior deployments 1.00 Referent >12 1.00 0.82, 1.23 7–12 1.07 0.82, 1.40 ≤6 2.11 1.52, 2.93 Deployment history interaction (deployment/months since return) Never deployed 6,451 First/>6 3,504 1.00 0.82, 1.22 First/≤6 263 1.64 1.06, 2.54 Second/>6 1,691 1.13 0.86, 1.47 Second/≤6 107 2.74 1.53, 4.94 Third or more/>6 803 1.20 0.85, 1.71 Third or more/≤6 58 3.76 1.81, 7.83 PTSD statuse No PTSD 1.00 Referent 1.00 Referent 1.00 Referent Current PTSD 0.86 0.54, 1.38 0.87 0.54, 1.40 0.86 0.54, 1.39 History of PTSD 1.39 0.74, 2.61 1.45 0.77, 2.71 1.44 0.76, 2.70 Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; PTSD, posttraumatic stress disorder. a ORs were adjusted for demographic (age, race, marital status, Army rank, educational level), obstetrical (multiple gestation, multiparity), and health-status (body mass index, tobacco use, alcohol use, hypertension, asthma) covariates as detailed in Web Table 2. b Modeling deployment count as a predictor. c Modeling deployment recency as a predictor. d Modeling the interaction of deployment count and recency as predictors. e Current PTSD = diagnosis code present in clinical encounters within 1 year prior to delivery; history of PTSD = diagnosis code present in prior clinical encounters, but none within 1 year prior to delivery. with regards to deployment status; thus, it is plausible that the effect sizes we arrived at were underestimates. Finally, we were unable to capture detailed information on the exact timing of conception in relation to deployment. While pregnancies occurring/detected in-theater, by policy, result in mandatory evacuation (34), we lacked field medical records with which to clarify whether a pregnancy was identified during deployment or to elucidate deployment-related health issues that could contribute to an increased risk of SPB. In conclusion, conception occurring during or just after return from deployment appears to increase the risk of SPB. Especially given women’s growing role in combat, priority should be given to enabling servicewomen to defer conception until after deployment and completion of the postdeployment reintegration transition. Am J Epidemiol. 2018;187(4):687–695 ACKNOWLEDGMENTS Author affiliations: Division of Primary Care and Population Health, Department of Medicine, School of Medicine, Stanford University, Stanford, California (Jonathan G. Shaw, D. Alan Nelson, Lianne M. Kurina); Department of Obstetrics and Gynecology, School of Medicine, Stanford University, Stanford, California (Kate A. Shaw); Office of the Surgeon General of the United States Army, Falls Church, Virginia (Kelly WoolawayBickel); VA Palo Alto Health Care System, US Department of Veterans Affairs, Palo Alto, California (Ciaran S. Phibbs); and Department of Pediatrics, School of Medicine, Stanford University, Stanford, California (Ciaran S. Phibbs). This study was supported by a Clinical and Translational Science Award (CTSA) to the Stanford Center for Clinical and Translational Research and Education (grant UL1 TR001085). Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 Recency of return from deployment, months since return 694 Shaw et al. REFERENCES 1. Street AE, Vogt D, Dutra L. A new generation of women veterans: stressors faced by women deployed to Iraq and Afghanistan. Clin Psychol Rev. 2009;29(8):685–694. 2. McGraw K, Koehlmoos TP, Ritchie EC. Women in combat: framing the issues of health and health research for America’s servicewomen. Mil Med. 2016;181(1 suppl):7–11. 3. Carter A. Memorandum for Secretaries of the Military Departments, Under Secretary of Defense for Personnel and Readiness, Chiefs of the Military Services, Commander, US Special Operations Command—Subject: Approval of Final Implementation Plans for the Full Integration of Women in the Armed Forces. Washington, DC: US Department of Defense; 2016. 4. Braun LA, Kennedy HP, Womack JA, et al. Integrative literature review: US military women’s genitourinary and reproductive health. Mil Med. 2016;181(1):35–49. 5. Seal KH, Bertenthal D, Miner CR, et al. Bringing the war back home: mental health disorders among 103,788 US veterans returning from Iraq and Afghanistan seen at Department of Veterans Affairs facilities. Arch Intern Med. 2007;167(5):476–482. 6. Shaw JG, Asch SM, Kimerling R, et al. Posttraumatic stress disorder and risk of spontaneous preterm birth. Obstet Gynecol. 2014;124(6):1111–1119. 7. Institute of Medicine, National Academy of Sciences. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: National Academies Press; 2007. 8. Ananth CV, Vintzileos AM. Epidemiology of preterm birth and its clinical subtypes. J Matern Fetal Neonatal Med. 2006; 19(12):773–782. 9. Copper RL, Goldenberg RL, Das A, et al. The Preterm Prediction Study: maternal stress is associated with spontaneous preterm birth at less than thirty-five weeks’ gestation. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol. 1996;175(5):1286–1292. 10. Green NS, Damus K, Simpson JL, et al. Research agenda for preterm birth: recommendations from the March of Dimes. Am J Obstet Gynecol. 2005;193(3):626–635. 11. Holzman C, Senagore P, Tian Y, et al. Maternal catecholamine levels in midpregnancy and risk of preterm delivery. Am J Epidemiol. 2009;170(8):1014–1024. 12. Class QA, Lichtenstein P, Långström N, et al. Timing of prenatal maternal exposure to severe life events and adverse pregnancy outcomes: a population study of 2.6 million pregnancies. Psychosom Med. 2011;73(3):234–241. 13. Basu R, Pearson D, Ebisu K, et al. Association between PM2.5 and PM2.5 constituents and preterm delivery in California, 2000–2006. Paediatr Perinat Epidemiol. 2017;31(5):424–434. 14. Vadillo-Ortega F, Osornio-Vargas A, Buxton MA, et al. Air pollution, inflammation and preterm birth: a potential mechanistic link. Med Hypotheses. 2014;82(2):219–224. 15. Wadhwa PD, Entringer S, Buss C, et al. The contribution of maternal stress to preterm birth: issues and considerations. Clin Perinatol. 2011;38(3):351–384. 16. Katon J, Cypel Y, Raza M, et al. Deployment and adverse pregnancy outcomes: primary findings and methodological considerations. Matern Child Health J. 2017;21(2):376–386. 17. Ryan MA, Jacobson IG, Sevick CJ, et al. Health outcomes among infants born to women deployed to United States military operations during pregnancy. Birth Defects Res A Clin Mol Teratol. 2011;91(2):117–124. 18. Conlin AM, DeScisciolo C, Sevick CJ, et al. Birth outcomes among military personnel after exposure to documented openair burn pits before and during pregnancy. J Occup Environ Med. 2012;54(6):689–697. 19. Ippolito AC, Seelig AD, Powell TM, et al. Risk factors associated with miscarriage and impaired fecundity among United States servicewomen during the recent conflicts in Iraq and Afghanistan. Womens Health Issues. 2017;27(3): 356–365. 20. Katon JG, Washington DL, Cordasco KM, et al. Prenatal care for women veterans who use Department of Veterans Affairs health care. Womens Health Issues. 2015;25(4):377–381. 21. Veterans Health Administration, US Department of Veterans Affairs. Women veterans health care—facts and statistics about women veterans. https://www.womenshealth.va.gov/ WOMENSHEALTH/latestinformation/facts.asp. Accessed November 27, 2017. 22. Stout MJ, Macones GA, Tuuli MG. Accuracy of birth certificate data for classifying preterm birth. Paediatr Perinat Epidemiol. 2017;31(3):245–249. 23. Klebanoff MA, Yossef-Salameh L, Latimer C, et al. Development and validation of an algorithm to determine spontaneous versus provider-initiated preterm birth in US vital records. Paediatr Perinat Epidemiol. 2016;30(2):134–140. 24. Baer RJ, Chambers CD, Bandoli G, et al. Risk of preterm birth by subtype among Medi-Cal participants with mental illness. Am J Obstet Gynecol. 2016;215(4):519.e1–519.e9. 25. Jelliffe-Pawlowski L, Baer R, Blumenfeld Y, et al. Maternal characteristics and mid-pregnancy serum biomarkers as risk factors for subtypes of preterm birth. BJOG. 2015;122(11): 1484–1493. 26. Bliese PD, Wright KM, Adler AB, et al. Validating the Primary Care Posttraumatic Stress Disorder Screen and the Posttraumatic Stress Disorder Checklist with soldiers returning from combat. J Consult Clin Psychol. 2008;76(2):272–281. 27. Seng JS, Low LK, Sperlich M, et al. Post-traumatic stress disorder, child abuse history, birthweight and gestational age: a prospective cohort study. BJOG. 2011;118(11):1329–1339. 28. Lydon-Rochelle MT, Holt VL, Nelson JC, et al. Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records. Paediatr Perinat Epidemiol. 2005;19(6):460–471. 29. Ofori BD, Le Tiec M, Bérard A. Risk factors associated with preterm birth according to gestational age at birth. Pharmacoepidemiol Drug Saf. 2008;17(6):556–564. Am J Epidemiol. 2018;187(4):687–695 Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 The CTSA program is led by the National Center for Advancing Translational Sciences at the National Institutes of Health. All data used in the study were provided under a cooperative agreement with the US Army Medical Command. We thank Vilija Joyce and Dr. Susan Schmitt of the VA Palo Alto Health Care System and Jonathan Mayo of Stanford University for technical support of our external validation work. This work was presented in abstract form at Stanford Medicine’s Spectrum Innovation Research Symposium, Stanford, California, January 31, 2017. This research was approved by the Stanford University Institutional Review Board and by the Defense Health Agency’s Human Research Protection Office. The content is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health, the Department of Veterans Affairs, or the US Army. Conflict of interest: none declared. Preterm Birth in Recently Returned Soldiers 695 Am J Epidemiol. 2018;187(4):687–695 38. Okun ML, Schetter CD, Glynn LM. Poor sleep quality is associated with preterm birth. Sleep. 2011;34(11): 1493–1498. 39. Kajeepeta S, Sanchez SE, Gelaye B, et al. Sleep duration, vital exhaustion, and odds of spontaneous preterm birth: a casecontrol study. BMC Pregnancy Childbirth. 2014;14:337. 40. Hanke W, Jurewicz J. The risk of adverse reproductive and developmental disorders due to occupational pesticide exposure: an overview of current epidemiological evidence. Int J Occup Med Environ Health. 2004;17(2):223–243. 41. Ryan MA, Gumbs GR, Conlin AM, et al. Evaluation of preterm births and birth defects in liveborn infants of US military women who received smallpox vaccine. Birth Defects Res A Clin Mol Teratol. 2008;82(7):533–539. 42. Conlin AM, Bukowinski AT, Gumbs GR. Analysis of pregnancy and infant health outcomes among women in the National Smallpox Vaccine in Pregnancy Registry who received anthrax vaccine adsorbed. Vaccine. 2015;33(36): 4387–4390. 43. Shachar BZ, Mayo JA, Lyell DJ, et al. Interpregnancy interval after live birth or pregnancy termination and estimated risk of preterm birth: a retrospective cohort study. BJOG. 2016; 123(12):2009–2017. Downloaded from https://academic.oup.com/aje/article-abstract/187/4/687/4802710 by guest on 07 June 2020 30. Haas JS, Fuentes-Afflick E, Stewart AL, et al. Prepregnancy health status and the risk of preterm delivery. Arch Pediatr Adolesc Med. 2005;159(1):58–63. 31. Wingate MS, Bronstein J, Hall RW, et al. Quantifying risks of preterm birth in the Arkansas Medicaid population, 2001–2005. J Perinatol. 2012;32(3):176–193. 32. Sorensen TK, Dempsey JC, Xiao R, et al. Maternal asthma and risk of preterm delivery. Ann Epidemiol. 2003;13(4):267–272. 33. Elixhauser A, Wier LM. Complicating Conditions of Pregnancy and Childbirth, 2008. (Healthcare Cost and Utilization Project (H·CUP) statistical brief #113). Rockville, MD: Agency for Healthcare Research and Quality; 2011. 34. Tepe V, Yarnell A, Nindl BC, et al. Women in combat: summary of findings and a way ahead. Mil Med. 2016;181(1 suppl):109–118. 35. Erickson AK, Nelson DA, Shaw JG, et al. Long-acting reversible contraceptive placement among active-duty US army servicewomen. Obstet Gynecol. 2017;129(5):800–809. 36. Witkop CT, Webber BJ, Chu KM, et al. Contraceptive prescriptions for US servicewomen, 2008–2013. Contraception. 2017;96(1):47–53. 37. Harrington LA, Shaw KA, Shaw JG. Contraception in US servicewomen: emerging knowledge, considerations, and needs. Curr Opin Obstet Gynecol. 2017;29(6):431–436.