Child Maltreatment in Enlisted Soldiers' Families During Combat-Related Deployments FREE

Child maltreatment includes neglect, physical abuse, emotional abuse, and sexual abuse. Children who are maltreated are at increased risk for negative health behaviors, depression, and chronic health conditions, with negative sequelae extending into adulthood.^1- 3 Few studies have examined child maltreatment within military families,^4- 10 of which there were more than 1.1 million with children younger than 18 years in 2004.¹¹ Limited knowledge regarding these families is of concern because of the possible impact of combat-related deployments on child maltreatment.

Military families have been found to demonstrate high levels of resilience^12- 13; nonetheless, deployments pose unique challenges. These deployments may affect the family's children, the soldier-parent preparing for (or returning from) deployment, and the parent remaining at home during deployment. Deployments have been associated with stress^14- 18 and behavioral problems^14,19 among children in military families, situations that can exacerbate parental stress.

Deployment also has been associated with increased stress^17,20- 21 among nondeployed parents, which may hamper their ability to appropriately care for their children. Parental stress (as mediated by their appraisals of the situation, available resources, and coping strategies) is believed to play a critical role in child maltreatment,^22- 23 particularly child neglect.²⁴ Hillson and Kuiper's stress and coping model of child maltreatment²² suggests that parents respond to stress either with positive adaptive behaviors or with dysfunctional behaviors that may result in child maltreatment.

The findings of 2 studies^25- 26 suggest a relationship between large-scale military deployments and increases in child maltreatment; however, the study designs did not permit the investigators to compare rates of child maltreatment events within abusive families during periods of deployment and nondeployment.

This study analyzed linked data from 2 confidential electronic data systems: the Army Central Registry and Army human resources data. The data were from a study period of approximately 40 months (September 11, 2001, through December 31, 2004). To protect the confidentiality of the families, all identifying information was recoded by Army personnel before the deidentified data were provided to the research team.

The Army Central Registry contains information on all substantiated incidents of child maltreatment and spousal abuse that have occurred within the total population of Army families worldwide. These data are reported by the Army Family Advocacy Program, the organization primarily responsible for family violence prevention, identification, evaluation, treatment, and follow-up.²⁷ At each Army installation, commanders and Army personnel in medical, social service, educational, and law enforcement roles are required (and other personnel are encouraged) to report known or suspected cases of child maltreatment. Army professionals coordinate with local civilian child protective service agencies to share information on child maltreatment reports involving military families, coordinate investigations, and provide services. Regardless of civilian agency involvement, the Family Advocacy Program assesses the reported maltreatment, makes decisions regarding substantiation based on its definitions and criteria, and monitors service provision and case resolution.²⁷

We examined several variables from the Army Central Registry. Of primary importance are variables that describe the characteristics of the substantiated child maltreatment incident. The Army Family Advocacy Program defines child maltreatment as mistreatment of a child by acts of omission or commission by a parent, guardian, foster parent, or caregiver in a way that the child's welfare is threatened or harmed. Maltreatment incidents are classified as neglect (deprivation of age-appropriate care, including failure to provide necessary care, inadequate supervision, medical neglect, educational neglect, and abandonment), physical abuse (physical harm, mistreatment, or injury), emotional abuse (acts or omissions with adverse effects on the child's psychological well-being), and sexual abuse (sexual activity with a child for the sexual gratification of an adult).²⁷ Maltreatment is categorized as mild or moderate or severe, with operational criteria defined for each type of maltreatment in terms of risk and harm to the child.²⁸ In addition, the date of the maltreatment report is recorded. Note that there is a median delay of 1 day between the actual occurrence of the maltreatment and the report of the maltreatment (C. D. Gable, PhD, written communication, May 2007).

Army Central Registry data also provide descriptions of the characteristics of the child maltreatment offenders, including their relationship to the military (soldier or civilian family member), age, sex, race and ethnicity, and whether offenders were using alcohol or other drugs at the time of the incident. Race and ethnicity classifications were based on information provided to the Army by the individual. Race was included as a possible covariate because it has been associated with variations in the rates of child maltreatment.²⁹ Characteristics of maltreated children include age and sex. Characteristics of the soldier family member include the soldier's sex and pay grade. We classified pay grades E1 through E4 as “lower” and E5 through E9 as “higher.” The lower group includes personnel up through the rank of corporal; the higher group includes personnel at the rank of sergeant or above.

Army human resources data for soldiers who had child maltreatment incidents in their families during the study period were provided by the Defense Manpower Data Center, which maintains military personnel files. Variables of interest were those that document dates of each soldier's entry to and exit from active duty Army service, and of soldier deployments related to Operation Enduring Freedom in Afghanistan or Operation Iraqi Freedom. During longer deployments, it is possible for soldiers to have short periods of time at home (ie, leave periods). Linking the Army human resources data to the Army Central Registry data allowed us to identify child maltreatment incidents that occurred during soldiers' deployments (or during leave periods) and those that occurred at times when soldiers were not deployed.

A total of 1985 families of active duty Army soldiers who experienced at least 1 combat-related deployment during the 40-month study period had records documenting substantiated child maltreatment by a parent (biological, step, or adoptive). A small number of families were excluded from analysis, including 156 families in which the soldier was not married to the civilian parent, 49 families in which the soldier was an officer, and 9 families in which both the husband and wife were service members. These exclusions were made because of the likely differences in these families' experiences of deployment and the relatively small sample sizes of these groups. Thus, the analysis included 1771 families of enlisted (ie, noncommissioned) soldiers.¹¹

Because some offenders committed maltreatment against more than 1 child on a single day during the study period while others committed maltreatment against 1 or more children on multiple days, the analysis examines incidents of child maltreatment. If an offender maltreated 1 of his/her children on only 1 day during the study period, we counted this as a single maltreatment incident. If the offender committed child maltreatment against 2 children on a single day, we counted this as 2 incidents. If an offender maltreated a single child on 2 different days, we counted this as 2 incidents. We also defined the period during which each child was at risk for maltreatment as being the number of days during the period when the child was 18 years or younger and the child's parent was an active duty soldier, classifying each day within that period as either during or not during a combat-related deployment of the child's soldier-parent. Finally, we characterized maltreatment incidents as occurring either during deployment or not during deployment.

The study protocol was reviewed by the US Army Human Subjects Research Review Board and the institutional review boards of both RTI International and the University of North Carolina at Chapel Hill. Each of these bodies approved the study protocol.

Conditional Poisson regression analysis methods³⁰ with random effects were used to estimate rate ratios (RRs) of interest (ie, RRs that compare the rate of child maltreatment during deployment with the rate of child maltreatment during nondeployment) and associated 95% confidence intervals (CIs). This maximum likelihood-based method of analysis, which uses family-specific comparisons of child maltreatment events during deployment and nondeployment, does not involve separately estimating a rate of child maltreatment during deployment and a rate of child maltreatment during nondeployment, but instead directly estimates the RR of interest. Moreover, this analysis takes into account differing family-specific risk periods of child maltreatment during deployment and nondeployment, accounts for family-specific correlated responses, and controls appropriately for family-specific covariates (including an underlying baseline tendency toward child maltreatment). We estimated an overall maltreatment RR (unadjusted for covariates), along with a 95% CI. We also estimated RRs and associated 95% CIs for RR parameters by stratifying on important covariates (including characteristics of the offenders, maltreated children, soldiers, and maltreatment incidents). Finally, we used similar analyses to examine child maltreatment committed by civilian female spouses during deployments and non-deployment of their soldier-husbands. Analyses were performed using SAS statistical software version 9.1 (SAS Institute Inc, Cary, North Carolina).

During the 40-month study period, 1858 parents in 1771 different families maltreated their children. Of these offenders, 1673 (90%) committed child maltreatment on only 1 day; 1028 (55%) maltreated 1 child on that day, 404 (22%) maltreated 2 children on that day, and 241 (13%) maltreated more than 2 children on that day (Table 1). An additional 163 offenders (9%) committed child maltreatment on 2 different days during the study period, and 22 parents (1%) maltreated children on 3 or more different days. The 1858 parent-offenders committed a total of 3334 child maltreatment incidents against 2968 individual children during the study period.

At the time of the initial incident, the parents' ages ranged from 18 years to 58 years (mean age, 29 years). The majority of offenders (54%) were non-Hispanic white. At the time of the initial experience of maltreatment, the 2968 children ranged in age from younger than 1 year to 18 years (mean age, 6 years). Male and female children were abused in approximately equal numbers.

There were a total of 3.3 million days at risk for maltreatment for the 2968 children (ie, days when the child was younger than 1 month through aged 18 years and the soldier-parent was an active duty member of the Army) during the study period. For a majority of these days at risk (78%), the soldier-parents were not deployed.

Table 2 shows that 942 (28%) of the maltreatment incidents during the study period occurred while the soldier-parents were deployed and the remaining 2392 incidents (72%) occurred while the soldier-parents were not deployed. Conditional Poisson regression analyses showed that the overall unadjusted RR comparing the rates of child maltreatment was 1.42 (95% CI, 1.31-1.54) (942 incidents and 713 626 days at risk during deployments vs 2392 incidents and 2.6 million days at risk during nondeployment). Thus, in families with at least 1 substantiated report of child maltreatment during the study period, the rate of child maltreatment during soldier deployments was 42% higher than the rate of child maltreatment at times when soldiers were not deployed.

To take other potentially important variables into account, we conducted additional descriptive and conditional Poisson regression analyses, stratifying the analyses by each of the variables shown in Table 2. Of all incidents committed during deployment, 67.7% were moderate or severe in nature; the corresponding percentage during nondeployment was 59.4%. The rate of moderate or severe maltreatment was especially high during deployment compared with the rate on days when soldiers were not deployed (638 incidents and 447 647 days at risk during deployments vs 1421 incidents and 1.6 million days at risk during nondeployment; RR, 1.61 [95% CI, 1.45-1.77]). Although the rate of mild maltreatment was somewhat elevated during deployment compared with nondeployment, the effect size was modest and did not quite reach the .05 level of statistical significance (304 incidents and 286 005 days at risk during deployments vs 971 incidents and 1.1 million days at risk during nondeployment; RR, 1.15 [95% CI, 0.99-1.30]). The rate of child neglect when soldiers were deployed was almost twice the rate of child neglect during the times when the soldiers were not deployed (761 incidents and 470 657 days at risk during deployments vs 1407 incidents and 1.6 million days at risk during nondeployment; RR, 1.95 [95% CI, 1.77-2.14]). In contrast, the rates of physical abuse and emotional abuse were significantly lower during deployment than during nondeployment (for physical abuse, 97 incidents and 80 033 days at risk during deployments vs 451 incidents and 318 326 days at risk during nondeployment; RR, 0.76 [95% CI, 0.58-0.93]; for emotional abuse, 28 incidents and 125 565 days at risk during deployments vs 340 incidents and 475 581 days at risk during nondeployment; RR, 0.31 [95% CI, 0.19-0.43]). The rates of sexual abuse and more than 1 type of maltreatment did not differ significantly by deployment status.

Additional stratified analyses found that the child maltreatment incidents during deployment and nondeployment differed in terms of the characteristics of the parent-offenders. In families with at least 1 incident of child maltreatment during the study period, the rate of child maltreatment by female civilians was especially elevated during times of deployment (783 incidents and 382 480 days at risk during deployments vs 832 incidents and 1.2 million days at risk during nondeployment; RR, 3.33 [95% CI, 2.98-3.67]). The rate of maltreatment during deployment also was elevated among male civilians but not significantly so. In contrast, the rate of child maltreatment during deployment was lower among female soldiers (7 incidents and 19 411 days at risk during deployments vs 107 incidents and 97 696 days at risk during nondeployment; RR, 0.31 [95% CI, 0.06-0.55]) and male soldiers (98 incidents and 294 303 days at risk during deployments vs 1298 incidents and 1.2 million days at risk during nondeployment; RR, 0.27 [95% CI, 0.21-0.32]). Incidents of maltreatment by a soldier-parent during deployment were likely to have occurred while the soldier was home on leave. The rate of child maltreatment during deployment was elevated for offenders who were non-Hispanic white (629 incidents and 402 319 days at risk during deployments vs 1227 incidents and 1.4 million days at risk during nondeployment; RR, 1.80 [95% CI, 1.62-1.98]) but not for offenders who were black or Hispanic. Examination of other offender characteristics found that the rate of child maltreatment during deployment was higher than the rate of child maltreatment during nondeployment regardless of the offenders' age or whether the offender was using substances at the time of the child maltreatment incident.

With respect to characteristics of the maltreated children, the rate of maltreatment during deployments was elevated (relative to nondeployment) for children who were older than 2 years but were aged 5 years or younger (292 incidents and 208 493 days at risk during deployments vs 645 incidents and 745 847 days at risk during nondeployment; RR, 1.67 [95% CI, 1.43-1.92]) and for those who were older than 5 years but were aged 12 years or younger (377 incidents and 300 043 days at risk during deployments vs 986 incidents and 1.1 million days at risk during nondeployment; RR, 1.50 [95% CI, 1.31-1.68]). Elevations in the rate of child maltreatment during deployment did not reach statistical significance for children aged 2 years or younger, or for those who were older than 12 years. The rate of maltreatment during deployment was consistently elevated for both male and female children.

Additional analyses estimated RRs comparing child maltreatment during deployment and nondeployment, stratified by the characteristics of the family's soldier-parent. Rates during deployments were consistently elevated for both pay grade groups and for those with 1 or more deployments.

Because the rate of substantiated incidents of child maltreatment by the female civilian parent was more than 3 times greater during times of deployment, we conducted additional analyses concerning maltreatment by this group of parents during deployment and nondeployment. Table 3 presents stratified analyses of the characteristics of the maltreatment incidents committed by the female civilian spouse. Rates of child maltreatment were greater during periods of deployment, irrespective of maltreatment severity (for moderate or severe maltreatment, 522 incidents and 237 697 days at risk during deployments vs 485 incidents and 768 301 days at risk during nondeployment; RR, 3.85 [95% CI, 3.34-4.36]; and for mild maltreatment, 261 incidents and 153 881 days at risk during deployments vs 347 incidents and 498 744 days at risk during nondeployment; RR, 2.62 [95% CI, 2.17-3.07]). The rate of child neglect by civilian female spouses was especially elevated during the times that their soldier-husbands were deployed, being almost 4 times the rate of child neglect during other times (666 incidents and 303 555 days at risk during deployments vs 605 incidents and 967 362 days at risk during nondeployment; RR, 3.88 [95% CI, 3.43-4.34]). In addition, the rate of spouses' physical abuse of their children was elevated during deployments, nearly twice the nondeployment rate (73 incidents and 18 316 days at risk during deployments vs 141 incidents and 61 105 days at risk during nondeployment; RR, 1.91 [95% CI, 1.33-2.49]). The rate of incidents involving multiple types of abuse also was quite elevated during deployments (29 incidents and 15 964 days at risk during deployments vs 31 incidents and 52 674 days at risk during nondeployment; RR, 3.36 [95% CI, 1.58-5.15]).

The rate of maltreatment during soldier deployments was higher for civilian female spouses who were non-Hispanic white (571 incidents and 249 089 days at risk during deployments vs 478 incidents and 771 236 days at risk during nondeployment; RR, 4.12 [95% CI, 3.58-4.65]) rather than black or Hispanic (184 incidents and 117 191 days at risk during deployments vs 312 incidents and 414 482 days at risk during nondeployment; RR, 2.19 [95% CI, 1.77-2.62]), although both groups experienced elevated rates of maltreatment during soldier deployments. For all other characteristics of the civilian female spouses, their children, and their soldier-husbands, rates of child maltreatment were consistently elevated during times of deployment compared with nondeployment.

This study found that among Army families of enlisted soldiers with at least 1 substantiated report of child maltreatment who experienced deployments, the rate of child maltreatment was 42% greater during deployments compared with times when soldiers were not deployed. The rate of child neglect also was greater during times of soldier deployments, while rates of physical abuse were less. The rate of moderate or severe child maltreatment was more than 60% greater during soldier deployments than when soldiers were not deployed.

This elevation in the rate of maltreatment during soldiers' deployments is attributable to much higher rates of child maltreatment—more than 3 times higher—by civilian female spouses during those times. This pattern was particularly pronounced for rates of child neglect (almost 4 times greater), but also was significantly greater for child physical abuse compared with nondeployed periods. The elevation in civilian female spouses' rates of child maltreatment that was moderate or severe in nature was almost 4 times greater during deployment periods.

Our findings are consistent with those of Rentz et al,²⁵ in which population-level rates of child maltreatment within military families in a single state doubled during a period of large-scale deployments within the state, with the greatest increase occurring in maltreatment by civilian members of military families. These results also are consistent with trends identified by McCarroll et al,²⁶ who found that the rates of child neglect in US Army families increased sharply between 2001 and 2004, reversing a decade-long downward trend.

Increased rates of child neglect and physical abuse during times of deployment may be best understood in light of the fact that the deployment of a soldier often places much additional stress on the parent remaining behind.^{14- 18,20- 21} Parents' appraisal of these stresses and their ability to mobilize internal and external coping resources may result either in positive adaptation to the challenging situation, or maladaptive responses that result in child neglect or abuse.^22,24

Although the incidence of child maltreatment has been linked to characteristics such as socioeconomic status, parental substance use, and younger age of the children,^31- 32 the conditional Poisson regression models we used control for these possible family variations in the underlying tendency toward child maltreatment. Thus, the elevated RRs found for most subgroups of offenders, maltreated children, and soldiers indicate a consistent impact of deployment on child maltreatment among families studied in this investigation.

Two exceptions to this pattern are of interest. First, rates of child maltreatment were greater during soldier deployment for female civilian parents but not for male civilian parents, suggesting that these 2 groups may be different in terms of the stress that they experience during their spouses' deployment, how they cope with such stress, or how they mobilize resources such as assistance with child care. Second, the rate of maltreatment incidents during deployment was greater for offenders who were non-Hispanic white than for those who were black or Hispanic. This difference may reflect racial-ethnic patterns in factors that are potentially related to the stress associated with soldier deployments, such as civilian parent employment or use of formal and informal support services (R. Robichaux, PhD, LCSW, Behavioral Health Division, US Army Medical Command, written communication, June 2007).

The similar elevation in the rate of child maltreatment during deployment for families experiencing 1 deployment and those experiencing multiple deployments may appear counterintuitive if one imagines that repeated deployments would be particularly difficult for families. The stress and coping model of child maltreatment²² suggests some possible explanations, although we cannot test them with these data. For example, the civilian parent's evaluation of the situation may change after the experience of the initial deployment, moderating the experience of stress during subsequent deployments. In addition, the civilian parent may develop coping strategies during the soldier's initial deployment that help to alleviate the experience of stress during subsequent deployments. The Army may contribute to this process by selectively providing additional services to families who experience multiple deployments or to those who experienced difficulties during the first deployment. Any of these scenarios could moderate what might otherwise be elevated rates of maltreatment associated with multiple deployments.

Because of certain methodological limitations of this research, we urge caution in interpreting these findings. Discrepancies between the date of maltreatment and the reporting date used for this analysis may compromise the classification of some maltreatment incidents as occurring during deployments or nondeployment. Examination of later data shows that in the majority of cases, abuse was reported on the day that it occurred or on the following day; this finding suggests that reporting delays are unlikely to affect our overall findings related to the rates of maltreatment during deployments and other times. In addition, this study only examined child maltreatment committed by parents in families of enlisted Army soldiers who experienced at least 1 deployment during the study period. Our investigation was limited to combat-related deployments, and therefore we have no evidence that similar results would be obtained from the observation of soldier families experiencing non–combat-related deployments. In addition, we did not examine child maltreatment by other family members (eg, grandparents, aunts, or other relatives) who also may be affected by soldier deployments. Furthermore, due to the small number of officer families who were substantiated for child maltreatment, we were not able to examine whether combat-related deployments were associated with increased rates of maltreatment in these families. Using this data set, it is impossible to determine whether this relative paucity of substantiated child maltreatment incidents among the families of officers is due to a true lower prevalence of child maltreatment in officers' families, or is due to other factors, such as differences in the reporting of child maltreatment for families of officers vs enlisted soldiers. However, we conducted post hoc analyses to examine the source of maltreatment reports and found that military sources were responsible for similar percentages of maltreatment reports in the families of officers (60.0%) and enlisted soldiers (59.7%), suggesting that there is not a difference in the reporting of child maltreatment for families of officers vs enlisted soldiers.

All comparisons examined herein are based on substantiations using the Army's criteria, which are not exactly comparable with those used by civilian child protection agencies. In addition, because this study focused on Army families, the results may not be generalizable to the patterns of child maltreatment in other branches of the military, which differ in the characteristics of service members and their families.¹¹

Finally, we restricted our comparison of child maltreatment rates during periods of deployment and nondeployment to families who were identified as having at least 1 substantiated report of child maltreatment during the study period. Therefore, we could not estimate the risks of child maltreatment during combat-related deployments and during periods of nondeployment for all Army families. To address this question, families of soldiers would have to be followed up prospectively through periods of deployment and nondeployment to directly compare the risks of child maltreatment.

Despite these study limitations, our research findings provide information that may help to inform policy and practice regarding child maltreatment, including the Army's Family Advocacy Program and clinicians and child welfare professionals in communities with military populations. The findings confirm the need for supportive and preventive services for Army families during times of deployment. The Army supports a range of voluntary services for families during soldier deployments. These include Family Readiness Groups that provide mutual support, and Family Assistance Centers that offer 1-stop access to key support agencies.³³ Availability of other services may vary among installations but these typically include respite child care, home visiting for families of infants, and support groups for spouses of deployed soldiers. Nevertheless, the greater rate of child maltreatment associated with deployments suggests the need for enhanced support for civilian parents in terms of additional resources, more effective services, development of services that those parents at greatest risk will be likely to seek out and accept, and greater outreach to connect parents to services.