Abstract
Military families are exposed to a unique constellation of risk factors, which may impact maltreatment outcomes. The present study examined prospective relationships between demographic, health, birth-related, and military-specific risk factors identified prior to a child’s birth on their risk for maltreatment in the first two years of life. Data from the Millennium Cohort Study, Department of Defense (DoD) operational records and Family Advocacy Program data on met-criteria maltreatment, and Birth and Infant Health Research program data on suspected maltreatment were linked for 9076 service member parents. Discrete time survival analysis showed that preterm birth increased risk of maltreatment while parents’ older age, physical health, and service in the Navy or Air Force decreased risk. Building on DoD’s New Parent Support Program, findings suggest the need for universal and targeted prevention efforts, beginning during pregnancy, which limit or eliminate risk factors for maltreatment in military families.
Child maltreatment is a significant public health concern in both military and civilian populations. Referrals for alleged maltreatment occurred for 7.9 million U.S. children in 2019, with 16.7% of allegations substantiated following investigation (US Department of Health and Human Services [DHHS], 2021). Most children with substantiated maltreatment experience neglect (75%) and physical abuse (18%; DHHS, 2021). Maltreatment in childhood has been associated with risk for internalizing and externalizing behaviors (Crouch et al., 2018), physical health problems (Crouch et al., 2018), and substance misuse (Fava et al., 2019). The first two years of life are a particularly vulnerable period; children under age 2 represent the greatest rate of maltreatment at 37.2 per 1000 children in 2019. Very young children are also the most likely to die from maltreatment with 70.3% of all fatalities occurring before the age of 3 (DHHS, 2021). Abuse and neglect during infancy and early childhood have particularly pernicious and long-lasting effects on development (Nelson & Gabard-Durnam, 2020).
The healthy functioning of families is of great importance to the military, as strong families are a critical context for the mission readiness and retention of service members (National Academies of Sciences Engineering and Medicine [NASEM], 2019). To that end, the Department of Defense (DoD) tasked each service branch with establishing a Family Advocacy Program (FAP) in 1981 (Stander et al., 2018). FAP includes programs tasked with the prevention, detection, and response to violence in military families. Rather than reporting to civilian child protective services, military and civilian personnel report suspected maltreatment among military families to FAP, where reports are substantiated using uniform criteria across DoD (Stander et al., 2018). Though criteria are standardized, there are important differences in the provision of FAP services across branches, including in average caseloads, services offered, and staffing, which may impact detection across branches (Farris et al., 2019).
Drawing from a recent consensus report on military family wellbeing, the present study employs a risk and protective factors framework, in which risk factors are characteristics or experiences which threaten adjustment and increase the likelihood of negative outcomes, while protective factors decrease this likelihood (NASEM, 2019). Though differences in criteria make direct comparisons difficult, rates of maltreatment by subtype appear generally lower among military families compared to the civilian population (Milner, 2015). For example, in military samples, the proportion of neglect cases is lower, presumably because military families have access to healthcare and a housing allowance that provide a basic safety net, but the proportion of physical abuse cases appears to be higher, compared to civilian populations. However, definitions of what constitutes child maltreatment are different across these populations. For example, witnessing intimate partner violence is coded as emotional abuse within FAP, but does not necessarily constitute emotional abuse using civilian CPS definitions. Despite potentially lower overall rates, military families are exposed to a unique constellation of risk and protective factors, which may impact maltreatment outcomes. The present study explores the impact of demographic, health, birth-related, and military-specific risk factors identified at or prior to a child’s birth on risk for maltreatment before 24 months. This effort elucidates the relative contribution of these risk factors and points to possible targets for prevention and early intervention with families at increased risk.
Child Maltreatment in Military Families
In 2021, FAP reported a unique child victim rate of 4.3 per 1000 (U.S. Department of Defense [DoD], 2021a, 2021b). Very young military-connected children are overrepresented among incidents of substantiated maltreatment compared to the population of young children in these families overall, with over one-third of incidents occurring among children aged 2 or younger (DoD, 2019). Several studies have explored maltreatment within active duty service branches. A report spanning 1990 to 2004 employing Army FAP data suggested rates of maltreatment declined outside of two spikes during periods of large-scale deployments to the Middle East (McCarroll et al., 2008). Other studies have also found increased rates of maltreatment, particularly neglect, during periods of increased operational tempo (Gibbs et al., 2007; Rentz et al., 2007). More recently, increased risk for neglect among Army families has been associated with family factors, including size, access to childcare, and deployment experiences (Cozza, Ogle, et al., 2018; Cozza, Whaley, et al., 2018). Studies of maltreatment among Air Force families are often limited to families with suspected or substantiated cases of maltreatment and are consistent with Army studies regarding variability by demographic characteristics and exposure to military-specific stressors (Lorber et al., 2018; Travis et al., 2014). There may also be variability in maltreatment rates across branches as known correlates of maltreatment, like SES, age, and education differ across branches (DoD, 2020).
Demographic, Health, and Birth-Related Factors Associated with Maltreatment
In the general population, younger, lower SES, and single parents are more likely to perpetrate maltreatment (Centers for Disease Control and Prevention [CDC], 2018). Males are more likely to perpetrate physical or sexual abuse; neglect is more likely among female perpetrators (DHHS, 2005). Parental trauma exposure, substance use, and poorer mental health are associated with perpetration of maltreatment (Ayers et al., 2019; Chemtob et al., 2011; Hammond et al., 2017; Wolf & Freisthler, 2016). Parents’ physical health has received less attention (Stith et al., 2009), but studies conducted with military samples suggest that parental physical injuries may increase maltreatment risk (Hisle-Gorman et al., 2019). Though evidence is mixed (Stith et al., 2009), child and birth-related factors, including premature birth, male infant sex, and birth defects, may also increase risk (Gumbs et al., 2013; Van Horne et al., 2015).
Military-Specific Factors Associated with Maltreatment
Regarding military-specific risk factors, rank or paygrade has been associated with risk for maltreatment, as the majority of parent offenders appear to be junior enlisted service members (64% E4-E6 and 14% E1-E3; DHB, 2019). Further, increased rates of maltreatment and particularly neglect appear to be associated with the onset of post-9/11 conflicts and increased operational tempo (Gibbs et al., 2007; McCarroll et al., 2008). The deployment cycle is associated with a series of transitions and separations that create stress for families. In particular, deployment separation appears to increase military spouse distress (Mansfield et al., 2010), potentially increasing their difficulty meeting children’s developmental needs (Ross et al., 2020), which is hypothesized to increase incidence of child neglect observed during deployment separation (Gibbs et al., 2007). Though the reintegration phase of deployment has been linked to increases in maltreatment as well (Rentz et al., 2007; Taylor et al., 2016), the association of parental trauma exposure, whether combat-related or not, with child maltreatment outcomes has not been adequately explored in this population.
The Current Study
Previous research focusing on maltreatment in military families has explored demographic correlates of maltreatment and the risk for maltreatment in families during specific phases of the deployment cycle. With few exceptions (e.g., Gumbs et al., 2013), the vast majority of these studies have focused solely on substantiated cases of maltreatment and have explored risk factors that occur concurrently with maltreatment experiences. The present study extends previous literature to examine risk factors prior to birth and explore their associations with both suspected and met-criteria maltreatment outcomes during the first two years of life. The study employs big data methods (Hawkins et al., 2017) to link survey data from the Millennium Cohort Study (Gray et al., 2002; Ryan et al., 2007), DoD operational records, FAP data on met-criteria maltreatment, and TRICARE medical records of suspected maltreatment from the Birth and Infant Health Research (BIHR) program (Wood et al., 2017). These data provide a more complete picture of demographic, parent health, birth-related, and military-specific risk factors, which are hypothesized to increase risk for maltreatment in this population.
Methods
Datasets
Millennium cohort study (MilCo)
MilCo is a prospective cohort study launched in 2001, prior to the September 11 terrorist attacks, with the goal of providing a comprehensive picture of the impact of military service on the health and wellbeing of service members. Details about the study have been published elsewhere (Gray et al., 2002; Ryan et al., 2007). The MilCo sample was randomly selected from service rosters across both components (active duty and Reserves/National Guard) and all branches of service; certain groups were oversampled including previously deployed, Reserve/Guard, and female service members (Ryan et al., 2007). Four panels were enrolled between 2001 and 2013 for a total of 201,619 participants. Following consent and baseline survey completion, service members complete follow-up surveys at approximate 3-year intervals, though only baseline survey data were used here. The Institutional Review Board (IRB) of New York University approved this study.
DoD archival records
MilCo survey data has been linked to DoD personnel, administrative, operational, and health care records for service members and their families (Gray et al., 2002). For example, data regarding rank, branch of service, and deployment history are available through the Defense Manpower Data Center (DMDC).
Family advocacy program
Reports of alleged maltreatment among active duty personnel across service branches are submitted to a Central Registry managed by FAP. Referred cases are subjected to a standardized and rigorous substantiation process to determine whether they “meet criteria” for substantiated maltreatment. The Central Registry stores data with identifying information needed for linkage only on incidents meeting full criteria (Milner, 2015).
DoD birth and infant health research (BIHR) program
BIHR performs routine surveillance and research on infant health outcomes among DoD beneficiaries. Detailed methods have been previously described (Bukowinski et al., 2017). In brief, medical encounter data from the Military Health System Data Repository (MDR) are linked to military personnel data from the Defense Manpower Data Center to identify infants born to DoD beneficiaries. MDR data capture inpatient and outpatient encounters at military and civilian treatment facilities and are coded with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic and procedure codes as well as Current Procedural Terminology (CPT) codes. Diagnostic and procedure codes are used to define live births and outcomes of interest.
Data Linkage
Service member social security numbers were used to link records in the FAP database, and a DoD-specific identifier was used to link records in BIHR. Of the 201,619 service members who completed a baseline MilCo survey across all four panels, 98,195 had babies identified in the BIHR database, with 11,393 babies born within one year of survey completion. Records for service members who experienced a multiple birth (n=144), who had two children in the same 12-month period (n=11), who were in the Coast Guard (n=200), and those who did not have an identifier for the child in one or both linked databases (n=1702) were excluded. Multiple children in the same family were excluded to avoid biased estimates caused by nested data. Coast Guard families were excluded because the Coast Guard does not fall under DoD so experiences are likely distinct from other branches. Records for an additional 260 babies in the BIHR database were excluded because the service member sponsor was not the parent of the child, leaving 9076 Army, Navy, Marine Corps, or Air Force service members with a singleton birth within one year of their baseline MilCo survey date. See online supplemental material for more information on MilCo survey data collection and linked datasets.
Participants
Distribution of demographic characteristics and risk factors.
All service members indicating Hispanic ethnicity or non-white race included.
Constructed from MilCo and DMDC variables.
Constructed from MilCo and BIHR variables.
Measures
Met-criteria or suspected maltreatment
From the 166 reports of maltreatment associated with MilCo service members in FAP, 48 were excluded because they were subsequent reports on the same child. Reports were further restricted to those occurring before age 2, totaling 34 incidents of met-criteria maltreatment. Thirty-one cases of suspected maltreatment were identified by the presence of at least one ICD-9-CM code indicative of probable child maltreatment on an infant’s inpatient or outpatient medical record in the BIHR database. ICD-9-CM codes were identified from two previous child maltreatment studies (Schilling et al., 2020; Schnitzer et al., 2004) and included: 995.5x (child maltreatment syndrome), 994.2 (effects of hunger), 994.3 (effects of thirst), E967.x (perpetrator of child and adult abuse), E968.x (assault by other and unspecified means), E904.0 (abandonment or neglect of infants and helpless persons), E904.1 (lack of food), E904.2 (lack of water), E904.3 (exposure to weather conditions, not elsewhere classifiable), V15.41 (history of physical abuse), V15.42 (history of emotional abuse), V15.49 (history of abuse—other), V61.21 (counseling for victim of child abuse). Six cases appeared in both FAP and BIHR and were only counted once, leaving 59 unique reports of suspected or met-criteria maltreatment associated with the Millennium Cohort sample. Finally, a dichotomous indicator was created flagging all participants with any met-criteria or suspected maltreatment incident identified in either database.
Demographic factors
SM age (years) and race/ethnicity (non-Hispanic white/non-white) were derived from DMDC data. Marital status at the baby’s birth (married/unmarried) and child sex (male/female) were obtained from BIHR medical records.
Service member health factors
Self-reported mental and physical health status were measured on Millennium Cohort baseline surveys with the 36-item Veterans RAND (VR) scale, developed at RAND as part of the Medical Outcomes Study (Kazis, Lee, et al., 2004). The VR-36 includes 8 scales with two overarching summary scores: the physical component summary (PCS) and mental component summary (MCS). These summary scores were created using a validated approach through which scores are comparable to the original VR-36, with a mean of 50 and a standard deviation of 10 where higher scores reflect better health status in the US population (Kazis, Lee, et al., 2004; Kazis, Miller, et al., 2004). On the standardized instrument, one item asks participants to rate their overall health; this was modified on the MilCo survey into two items focused on physical and mental health, respectively, for a total of 37 items. Of these 37 items, 15 items are weighted more strongly to comprise the MCS, including: “During the past 4 weeks, how much of the time have you felt downhearted and blue.” Twenty-two items are weighted toward the PCS, including “I seem to get sick a little easier than other people.” Response options and component score calculation for the VR-36 are described elsewhere (Kazis, Lee, et al., 2004; Kazis, Miller, et al., 2004). Alcohol-related problems were measured with five dichotomous (yes/no) items from the Patient Health Questionnaire (Spitzer et al., 1999). Items reference the past 12 months and include “you had a problem getting along with people while you were drinking” and “you drove a car after having several drinks or after drinking too much.” A yes response on any item reflects problematic alcohol use.
Service member trauma exposure
Trauma exposure, which may or may not have been associated with combat, was measured with five items on the Millennium Cohort survey, drawn from the National Survey of Persian Gulf War Era Veterans (Smith et al., 2008). As all service members responded to these items regardless of whether they had been deployed, and as items are not specifically anchored to combat, these items were deemed to reflect trauma or violence exposure generally. Items began with the prompt, “Have you ever been personally exposed to the following (do not include TV, video, movies, computers or theater)” and included three response options: no, yes 1 time, and yes more than 1 time. Items included: “witnessing a person’s death due to war, disaster, or tragedy” and “witnessing instances of physical abuse (torture, beating, rape).” They were then asked to provide the date of most recent exposure. Those who endorsed any trauma item in the year prior to their survey received a score of 1 (exposure present).
Military-specific factors
Variables describing service member rank (enlisted, officer) and branch of service (Army, Navy, Marines, Air Force) were derived from Defense Manpower Data Center (DMDC) personnel files. To assess the proximal impact of deployment, data were drawn from DMDC’s Contingency Tracking System and recoded into a dichotomous variable reflecting deployment in the year prior to survey completion (Sullivan et al., 2020a; 2020b).
Birth-related factors
Major birth defects were defined according to current definitions from the National Birth Defect Prevention Network, which were based on ICD-9-CM codes in the range of 740.x-760.x. To qualify as a birth defect case, an infant must receive at least one inpatient diagnosis or two outpatient diagnoses on different dates. Preterm birth, or birth prior to 37 completed weeks’ gestation, was defined using ICD-9 codes from records of the infant or mother: weeks of gestation 756.2; preterm birth 765.x; postterm newborn 766.2x; preterm delivery 644.2x; or postterm pregnancy 645.[1,2]x. Finally, elapsed time between MilCo survey date and baby’s birth, measured in months, was included in models as a control variable.
Data Analysis
Data analysis began with univariate and bivariate analyses to estimate the prevalence of maltreatment and risk factors. Next, we employed discrete-time survival analysis using logistic regression to estimate the odds of maltreatment associated with identified risk factors (Singer & Willett, 2003). A person-period dataset was generated by including one observation per participant for every month their child was at risk of maltreatment up to 24 months of age, for a total of 185,485 person-months. Participants who did not experience maltreatment were censored either at the end of the 24-month observation period or when the service member separated from active military service. If the service member separated before their child turned 2 years old, they were censored early because linked datasets require TRICARE eligibility, thus we no longer had visibility to determine event occurrence. Approximately 24% of the sample was censored before the end of the observation period (median censor time 8 months); these families were only considered at risk for the number of months for which we had visibility to determine if they experienced maltreatment. All risk factors were modeled as time-invariant. No constraints were placed on the shape of the hazard function and thus the 24 monthly hazard intercept terms were completely general. Unadjusted and adjusted logistic regression were employed to examine the relationship of risk factors to maltreatment occurrence separately and in the presence of other risks. All analyses were conducted in SAS version 9.4. Missing data (less than 10% on all variables; see Table 1) were estimated using proc MI in SAS.
Results
Service member parents reported a mean mental component score of 52.16 (SD = 9.01) and a mean physical component score of 53.08 (SD = 7.65); almost 9% reported alcohol-related problems (Table 1). About 15% experienced a deployment in the year prior to survey completion and over 20% reported trauma exposure in the past year. Among babies, approximately 7% were born preterm and 3% were diagnosed with a birth defect. Less than 1% of families experienced a suspected or met-criteria maltreatment event in the 24 months following the child’s birth.
Life table describing maltreatment events.
*The number of participants who had not experienced a maltreatment event or been censored at the end of the month. Decreases in the risk set reflect both event occurrence and censoring due to service member separation from active military service.
Results of unadjusted and adjusted logistic regressions.
Note. SM= service member. Significant findings displayed in bold. 24 monthly intercept terms not displayed for simplicity.
In adjusted results, service member age remained associated with significantly decreased odds of maltreatment (OR = 0.914; 95% CI = 0.845–0.989), while female sex and being married no longer predicted maltreatment occurrence, likely reflecting the confounding effect of age, as female service members tend to be younger and married service members older. Service members in the Air Force (OR = 0.484; 95% CI = 0.250–0.937) and Navy (OR = 0.246; 95% CI = 0.086–0.704) had significantly decreased odds of a maltreatment event compared to those in the Army. Though mental health trended toward significance, those reporting better physical health prior to their child’s birth had significantly decreased odds of a maltreatment event during the first two years of their child’s life (OR = .968; 95% CI = 0.940–0.997), over a 3% reduction in risk for every unit increase in physical health. Finally, children born preterm were twice as likely to experience a maltreatment event (OR = 2.338; 95% CI = 1.136–4.810). Known risk factors including deployment, alcohol use, and trauma exposure did not reach significance.
Discussion
This study explored prospective relationships between risk factors experienced by active duty military service members and their families before or at the birth of their child and the child’s risk for maltreatment during the first two years of their life. To our knowledge, this study is the first to employ big data methods to link multiple existing datasets and to explore risk factors for maltreatment during the pre- and perinatal period. Overall, service members in this sample were relatively healthy, with self-reported mental and physical health scores both above the population mean (Kazis, Lee, et al., 2004). Births in this sample were also relatively healthy, with birth defects detected in 2.91% of births compared to 3% in the population overall (CDC, 2008) and 7.08% of babies born preterm compared to 9.76% in the general population (Purisch & Gyamfi-Bannerman, 2017). The rate of maltreatment before age 2 was also low, at less than 1%, compared to 3.7% in the general population (DHHS, 2019). Regarding comparisons to military data, the rate of maltreatment observed here is likely higher, though the unique child victim rate for children under 2 is not available (DoD, 2021). Findings suggest that exposure to risk factors, even before the birth of a child, may meaningfully increase the risk for maltreatment among very young military-connected children.
In the present study, birth outcomes were the strongest predictor of risk for maltreatment. Though birth defects did not increase risk in our sample, babies born prematurely were more than twice as likely to experience maltreatment before their second birthday. While preterm birth is a known risk factor for maltreatment (Spencer et al., 2006), due to the increased stress of challenging infant health issues (Enlow et al., 2017), findings from a recent study of pre-term civilian babies yielded a 60% increase in maltreatment risk (Mason et al., 2018), compared to over 130% here. Stressors associated with pre-term birth could be exacerbated by military-specific stressors, such as disruptions to support systems and care continuity precipitated by frequent moves among active duty families, in addition to more widely-studied transitions within the context of the deployment cycle (Meadows et al., 2017).
Regarding demographics, in our adjusted results, only the age of the service member was a significant predictor of risk for maltreatment. Babies born to older service members were at significantly decreased risk of maltreatment, which is consistent with a sizeable body of literature in the general population that highlights risk for maltreatment among younger parents (CDC, 2018). Though not novel, these findings are nevertheless concerning in this population, as service members aged 25 years or younger represent almost half of the active component (DoD, 2020), and there is evidence that service members tend to marry and start families at younger ages (Lundquist & Xu, 2014), suggesting that targeted maltreatment prevention programs tailored to younger families may be particularly important.
Service members who reported better mental and physical health may be at lower risk of maltreatment, though only the relationship between physical health and maltreatment remained significant in adjusted results. While the relationship between parental mental health and child maltreatment has been established in prior literature (Stith et al., 2009), our results suggest that poor physical health among parents may also increase risk for maltreatment, a finding which has received significantly less attention in prior work. Further, the mental and physical health of service members was measured up to a year before their child was born and potentially up to three years before a maltreatment event occurred, suggesting that it may be possible to identify military families at higher risk for maltreatment during the prenatal period. Most critically, service members are at particular risk for physical health problems (Williamson et al., 2019), and this may be a particular risk for future maltreatment.
Service members who reported recent trauma exposure, whether in the context of combat or not, were not at elevated risk of maltreatment. While these findings run counter to results from civilian studies (Cohen et al., 2008), a number of factors may account for this discrepancy. First, service members reported trauma exposure in the year prior to survey completion, which could be up to two years before their child was born and four years before the end of the observation period for maltreatment events. It is possible that the impact of trauma exposure, particularly in the absence of lasting mental health symptoms, may be more proximal. Second, the measure used here does not explicitly capture prior exposure to maltreatment in one’s family of origin, which has been the most consistent predictor of future maltreatment events (Stith et al., 2009).
Finally, while rank was not a significant predictor, branch of service was significantly related to maltreatment. Children born to service members in the Navy and the Air Force were less likely to experience maltreatment relative to children born into Army families. This may be an indication of real maltreatment differences by branch that are worthy of replication in future research. Furthermore, correlates of maltreatment may vary across branches; for example, service members in the Army tend to be younger, a risk factor for maltreatment (DoD, 2020). We controlled for many of these demographic differences, including age, in our analyses, but there are potentially other differences, like SES which we were unable to include in models. These may account for some of the observed difference between branches. Additionally, prior research has consistently found a connection between deployment separations and increased rates of maltreatment (Gibbs et al., 2007; Rentz et al., 2007). Service members in the Army deploy significantly more than in other branches (Baiocchi, 2013). Though our analyses accounted for recent deployments, the effects of cumulative deployment experiences, which are likely to be elevated among Army families relative to other branches, may partially explain the increased rates of maltreatment observed among these families.
Differential rates of maltreatment may also be a product of differences in surveillance and service provision across branches. A recent RAND report (Farris et al., 2019) describes key differences in how FAP services are organized and their relationship to other military support resources across branches. Though the within branch sample size was too small to evaluate whether differences were statistically significant, findings suggest there may be important variability. For example, 100% of Air Force and 96% of Navy FAP centers engaged in outreach and prevention services compared to only 67% of Army offices. Caseloads at each installation also appear to vary by branch, with caseloads in the Air Force and Navy at 58 and 32, respectively, while Army and Marine installations averaged 165 and 135, respectively (Farris et al., 2019). This RAND report, which triangulates information from interviews with stakeholders within and outside DoD and a survey of FAP offices, offers a blueprint for further investigation. A similar approach with a larger and more representative sample of FAP offices in each branch would be suited to identifying critical differences in surveillance and service provision with the potential to inform specific policy recommendations at the DoD and branch levels.
Strengths and Limitations
To our knowledge, this study represents the largest effort to link existing data on military families to examine risk for maltreatment among very young children. Linkage across four different datasets offers two advantages over other approaches. First, combining administrative, operational, health, and self-report survey data provided a more comprehensive picture of potential risk exposure and avoided many of the pitfalls of using only self-report data. Second, by including both FAP and medical records from BIHR, this study attempts to address problems with underreporting in both systems separately (Wood et al., 2017), potentially providing a more complete picture of maltreatment in military-connected families. Additionally, this study detected a low base rate of maltreatment incidents, which is expected when studying rare events; however, the large sample size nevertheless allowed us to detect significant relationships between a number of risk factors and our outcome. Finally, prospective relationships between pre- and perinatal risk factors and maltreatment outcomes highlight avenues for prevention.
Despite these strengths, the results of this study should be considered in light of a number of limitations. First, though linking multiple datasets is preferable to other methods, it is still likely that the present data did not fully capture maltreatment events in this population. Further, given the low base rate of maltreatment, we were not able to examine the impact of risk factors on subtypes of maltreatment, including neglect, physical, emotional, or sexual abuse. As there is evidence that military stressors like deployment may have differential effects on different maltreatment types (Gibbs et al., 2007; Rentz et al., 2007), examination of these more nuanced relationships is warranted. Similarly, other potential risk factors for maltreatment, like birth order, family size, or health of the non-military parent, were not available for inclusion (Stith et al., 2009). Additionally, there is the potentially for confounding and detection bias specific to our findings regarding pre-term birth. Infants born prematurely are likely to have more contact with medical providers (Korvenranta et al., 2010), which may increase the potential for maltreatment detection. Further, confounding variables, like unwanted pregnancy, may be associated with increases in pre-term birth (Seng et al., 2016) and increases in maltreatment (Guterman, 2015).
Further, this study focuses solely on risk factors and does not take into account the strengths of military families, which may be critical to understanding maltreatment outcomes consistent with the risk and protective factors framework employed here. Social support, for example, is consistently associated with lower maltreatment risk (Stith et al., 2009), but was not modeled here. Additionally, some variables included in models did not perfectly capture relevant constructs. For example, we were unable to include SES background, and the trauma exposure variable measured only trauma in the last year and did not represent military parents’ exposure to maltreatment during their own childhoods, a strong predictor of future maltreatment risk in other studies (Madigan et al., 2019; Stith et al., 2009). Finally, though the prospective design of this study is critical to identify risk factors which could be targeted for prevention efforts, the time elapsed between the measurement of some risk factors and the outcome may have been too long to observe more proximal effects on maltreatment. For example, deployment appears to have an effect on maltreatment risk that may last up to 6 months beyond the family’s reunification (Taylor et al., 2016), but our observation period included a wider margin of time that could begin a year beyond the end of the most recent deployment for families in this sample.
Implications
Our findings lend support to the conceptualization of maltreatment as a public health issue. Herrenkohl et al. (2016) suggest this effort requires comprehensive, coordinated, multisector initiatives—inclusive of universal and targeted prevention efforts to limit or eliminate risk. Considering the overall low base rate and significant elevation in risk associated with several factors examined here, these findings support targeted prevention, focusing on critical risk factors for maltreatment. As preterm birth was the most salient predictor here, interventions that include universal preterm birth prevention strategies set forth by Newnham et al. (2014) are also worthy of consideration. Ensuring that preterm birth prevention resources are available to all military families may be a viable child maltreatment prevention strategy.
Investing in targeted supports for families who experience preterm birth may also be critical to preventing maltreatment. The New Parent Support Program (NPSP) is an installation-based home visiting program available to expecting military families, as well as those with young children, that is primarily designed to assist families with the transition to parenthood within the military context. Specifically, NPSP is designed to support parents in managing additional parenting demands due to deployment-related separation and stress. Tailoring a version of NPSP to mitigate the impact of parenting stress associated with preterm birth could function as a selective prevention strategy to reduce maltreatment risk (Kaaresen et al., 2006). Integrating content that has been shown to reduce risk of preterm birth itself also may hold promise. Findings from a recent systematic review of primary and secondary interventions to reduce preterm birth revealed positive effects of nutritional education, health promoting behavioral and lifestyle changes, and nutritional supplements (Matei et al., 2019); integrating such content into existing NPSP programming may maximize impact by also addressing parental health status prior to birth, another risk factor for maltreatment revealed by the present analyses.
Conclusion
In the U.S., the problem of child maltreatment has primarily been conceptualized as a social issue, leading to solutions that rely largely on tertiary social service interventions. Shifting to a public health approach that embraces both universal and targeted prevention initiatives such as those described above has the potential to yield substantial population-level impacts. In contrast to family-based supports linked to deployment transitions (DeVoe et al., 2017; Julian et al., 2017), our findings suggest that a selective prevention approach focused on families who experience a broader array of risk factors, begins during pregnancy, and prioritizes cross-sector collaboration holds great promise for expectant military families. Finally, these findings suggest that there may be important differences in rates of maltreatment across service branches which could result from a combination of differential rates of risk factors across branches as well as differences in services offered to families at risk. More research is needed to replicate and understand potential causes for these differences.
Footnotes
Acknowledgments
The authors also gratefully acknowledge Nida Corry, PhD; Sharmini Radakrishnan, PhD; Alicia Sparks, PhD; and Christianna Williams, PhD from Abt Associates; Ernestine Briggs-King, PhD and John Fairbank, PhD from the Center for Child and Family Health; and contributions of the Millennium Cohort Study Team. In addition, the authors want to express their gratitude to the Family Study participants without whom this study would not be possible. The authors express gratitude to the other contributing members of the Millennium Cohort Family Study Team from the Naval Health Research Center, including Lauren Bauer, MPH; Alejandro Esquivel, MPH; Hope McMaster, PhD; Alexis Takata; and Kelly Woodall, MPH.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Disclaimer
Authors on this paper are military service members or employees of the U.S. Government. This work was prepared as part of their official duties. Title 17, U.S.C. §105 provides that copyright protection under this title is not available for any work of the U.S. Government. Title 17, U.S.C. §101 defines a U.S. Government work as work prepared by a military service member or employee of the U.S. Government as part of that person’s official duties. Report No. 18-XX was supported by the U.S. Navy Bureau of Medicine and Surgery under work unit no. N1240. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, nor the U.S. Government. The study protocol was approved by the Naval Health Research Center Institutional Review Board in compliance with all applicable Federal regulations governing the protection of human subjects. Research data were derived from an approved Naval Health Research Center Institutional Review Board protocol number NHRC.2015.0019.
