Health-Care Utilization Patterns of Maltreated Youth

Health-Care Utilization Patterns of Maltreated Youth Abstract To examine in detail the health-care utilization patterns of maltreated children, we studied electronic health records (EHRs) of children assigned maltreatment-related codes in a large medical system. We compared youth with maltreatment-related diagnoses (N = 406) with those of well-matched youth (N = 406). Data were based on EHRs during a 4-year period from the University of Minnesota’s Clinical Data Repository, which covers eight hospitals and over 40 clinics across Minnesota. A primary care provider (PCP) was assigned to over 80% of youth in both groups. As expected, however, the odds of not having a PCP were twice as high in the maltreated as in the comparison group. Also as expected, maltreated youth had higher rates of emergency department visits. We ruled out differences in age, gender, race, public insurance, duration in the medical system, type of specialty department, and clinic location as potential explanations for these differences. On the other hand, there were no significant differences between maltreated and comparison youth in hospitalizations, preventive visits, or office visits. Contrary to expectations, maltreated youth were not in the medical system for just a brief period of time and were not more likely to cancel or miss appointments. The current study adds to the research literature by providing more detailed information about the nature of health-care services used by children with maltreatment-related diagnoses. emergency department, health-care utilization, hospitalization, maltreatment, primary care provider Maltreatment affects a large proportion of children in the United States. Over a third (37%) of children have a maltreatment investigation by Child Protective Services (CPS) by the time they reach age 18 years (Kim, Wildeman, Jonson-Reid, & Drake, 2017), and about one in eight (12.5%) children have substantiated maltreatment (Wildeman et al., 2014). The latest National Survey of Children’s Exposure to Violence shows a lifetime prevalence rate of 38% for any maltreatment, based on self- and parent report (Finkelhor, Turner, Shattuck, & Hamby, 2015). The vast majority of what we know about the health-care needs and service utilization of maltreated children come from studies of children involved with CPS, rather than samples of children in the community. According to these studies, maltreated children have considerable mental and physical health needs (Christian & Schwarz, 2011). Most of the studies documenting these needs are based on the National Survey of Child and Adolescent Well-being (NSCAW), which include children who have been investigated by CPS (Burns et al., 2004; Casanueva, Cross, & Ringeisen, 2008; Farmer et al., 2010; McCue Horwitz et al., 2012; Stahmer et al., 2005), other cases of substantiated maltreatment (Schneiderman, Kools, Negriff, Smith, & Trickett, 2015), and children in foster care (Sullivan & van Zyl, 2008). Studies of children in NSCAW (Farmer et al., 2010; Florence, Brown, Fang, & Thompson, 2013; Hurlburt et al., 2004; Schneiderman et al., 2015), children in foster care or otherwise involved with CPS (Harman, Childs, & Kelleher, 2000; McMillen et al., 2004; Staudt, 2003; Thackeray, Leonhart, Yackey, Cooper, & Kelleher, 2016), and children exposed to domestic violence (Smith, Thompson, Johnson, Nitsche, & Kaslow, 2009) show that they have elevated rates of inpatient and outpatient mental health, emergency department (ED), general health services, and psychotropic medication use, and children assigned to maltreatment codes have an elevated rate of hospitalizations because of physical injuries (King, Farst, Jaeger, Onukwube, & Robbins, 2015). A review of studies of health-care costs of maltreated children and adults maltreated as children (Brown, Fang, & Florence, 2011) found maltreatment to be associated with higher mental health costs and utilization of multiple providers. Annual Medicaid costs of children in the NSCAW study who were maltreated or at risk for being maltreated were found to be $2,600 higher than costs of comparable children (Florence et al., 2013). Most of these studies have examined overall patterns of use, such as total Medicaid costs, or use of any mental health services. Despite these elevated rates of health-care utilization, maltreated children, at least those involved with CPS, do not receive adequate care given their level of need (Christian & Schwarz, 2011). For example, data based on the NSCAW study show that receipt and utilization of services for mental health and developmental problems are low compared with their level of problems (Burns et al., 2004; Casanueva et al., 2008; Farmer et al., 2010; Horwitz et al., 2012; Leslie, Hurlburt, Landsverk, Barth, & Slymen, 2004; Stahmer et al., 2005). Maltreated children are also less likely to have a medical home (Christian & Schwarz, 2011), defined as “accessible, continuous, comprehensive, family centered, coordinated, compassionate, and culturally effective” care “delivered or directed by well-trained physicians who provide primary care and help to manage and facilitate essentially all aspects of pediatric care … [and who] should be able to develop a partnership of mutual responsibility and trust with [the child and family]” (Medical Home Initiatives for Children With Special Needs Project Advisory Committee, 2002, p. 184). Similarly, a case-control study found that maltreated children were more likely to experience a change in their primary care providers (PCPs) in the year before their first maltreatment report that led to foster care placement (Friedlaender, 2005). On the other hand, this same case-control study found that maltreated children were not more likely to have ED or ambulatory visits during the year before their first maltreatment report (Friedlaender, 2005). Furthermore, a comparison of adolescents with substantiated maltreatment with a nonmaltreated group from the same low-income neighborhoods showed no differences in health insurance coverage and, in fact, greater receipt of preventive medical care in the maltreated than in the comparison youth (Schneiderman et al., 2015). Thus, there is some inconsistency regarding the extent to which maltreated children use different kinds of health-care services. In short, although most research indicates that maltreated children involved with CPS have higher health-care needs, and most, but not all, studies show higher levels of health-care use, many studies also show that these youth are not receiving adequate care. The discrepancy may be related to suboptimal patterns of health-care utilization. For example, they may rely too much on ED visits as opposed to preventive care; however, there are no studies directly comparing maltreated children’s use of ED services versus preventive visits. There is also no research on the extent to which maltreated children remain in one health-care system on a consistent basis, which would allow health-care providers more opportunities for effective interventions. Given the inconsistencies in their caregivers and instability in their living arrangements, it is also possible that maltreated children may have a high number of missed or canceled appointments, which might impact their health care. However, no studies have yet examined the frequency of missed appointments along with service use. Current Study To address the gaps in the research literature regarding the nature of health-care utilization and the limitation of research so far to children already involved with CPS, we compared the electronic health records (EHRs) of children assigned maltreatment-related codes in a large medical system with those of youth without maltreatment codes matched on age, gender, and clinic service specialty and location. Although these children were assigned maltreatment-related codes in their EHRs, study criteria did not include any involvement with CPS. Thus, we were able to assess a more community-based sample than previous studies. We examined the association of maltreatment codes with health-care utilization variables that could be extracted reliably from EHRs, namely, duration in the system, having a PCP, rates of hospitalizations, outpatient procedures/services, and missed appointments. Based on previous studies, we expected the maltreated group to be less likely to have a PCP, and to have higher rates of ED use. We also expected that maltreated youth might be more likely to be seen only once and to drop out of the system, and to miss more appointments. Method Overview Data were based on EHRs from the University of Minnesota’s Clinical Data Repository, which covers eight hospitals and over 40 clinics across Minnesota that are part of the Fairview Health Services (a nonprofit academic health-care system) or the University of Minnesota Physicians (a multispecialty group practice for medical faculty at the University). Close to 2.5 million, 0–21-year-olds had encounters at these sites between January 2011 and March 2015. We requested EHRs on all youth with an International Classification of Diseases-9 (ICD-9) (World Health Organization, 1978) code related to maltreatment. Next, we conducted a manual chart review to verify the maltreatment. The final sample consisted of 406 youth who had been assigned maltreatment-related codes and 406 matched comparison youth. Electronic Health Records Access to EHRs was provided by the Clinical and Translational Science Institute (CTSI) at the University of Minnesota. Participants were selected using the Epic EHR (EPIC—Electronic Privacy Information Center, n.d.), which includes longitudinally linked charts of all patients who had encounters at participating hospitals and clinics. Data entry practices across all clinics and hospitals became more consistent after 2011; thus, we limited participants to patients seen between January 1, 2011 and March 30, 2015, when we started the study. Data were generally based on encounters that occurred within the system; however, in cases where a patient was seen by providers outside the system covered by the Data Repository, EHRs contained reports sent to the patients’ clinicians by these outside providers. Maltreated Group Patients in the maltreated group were between 0 months to 21 years of age at the time the index maltreatment diagnosis was assigned. Eligible patients had at least one of the ICD-9 codes in Table I on the “encounter diagnoses” or “problem list” in their EHRs since 2011. We included all the codes designed to indicate that the provider suspects abuse or neglect that were included in ICD-9 (World Health Organization, 1978), the standard system used in the United States to assign diagnoses to patients. These codes were extracted automatically from the charts. Table I. Percentage of Participants With Maltreatment-Related ICD-9 Codes ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  Note. ICD-9 = International Classification of Diseases-9. Table I. Percentage of Participants With Maltreatment-Related ICD-9 Codes ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  Note. ICD-9 = International Classification of Diseases-9. When the index maltreatment diagnosis was “observation for abuse and neglect,” the type of maltreatment was further specified through manual chart review (see below). Because the automatically extracted information was restricted to data since 2011, manual chart review was restricted to the same time period. However, if a file from within this period revealed relevant information from before 2011 (e.g., mention of maltreatment or nature of the maltreatment), this information was included in coding. Comparison Group Eligible comparison youth had none of the maltreatment-related diagnoses above in their chart. Comparison patients were matched to eligible maltreated youth first on gender, then on age (to within 1 month). We also required comparison youth to have been seen at the same specialty department, at the same location, and at the same date (to within 1 week) as the encounter at which the index maltreatment diagnosis was assigned to the matched youth. If there were multiple patients matching these criteria, we selected the one that was closest on date of encounter. If they were equally close, we selected the one closest in age. If there were several such youth, the first one on the list was chosen. If there were no matches on these criteria, the temporal criterion for the encounter was expanded to 2 months. In cases where there were several matches with the expanded criterion, the one closest on date of encounter was selected. The vast majority (88.4%) of the patients fell within these criteria. If there were still no matches, comparison youth were matched on center, rather than department specialty, and the temporal criterion was again expanded from within 1 week to 2 months. If several comparison patients met the center criteria, the one with the closest date of encounter and, when possible, a similar department in the same center was chosen. These criteria were met by 9.4% of the comparison patients. If no matches were found with these criteria, the age-matching requirement was relaxed 1 month at a time. For maltreated patients aged <12 months, the upper limit on the age difference was 3 months. If the target patient was between 12 and 36 months of age, the upper limit was 6 months. For older ages, the upper limit was 12 months. These criteria were met by 1.5% of the patients. If still no matches were found, the center requirement was relaxed to include other centers in the vicinity. Only a few (0.7%) of the comparison group were in this subset. Whenever possible, comparison youth seen at similar specialty departments in other centers were chosen. Each comparison patient was matched to only one maltreated patient. Other Variables Extracted Automatically From EHRs Vital Status We coded whether the patient was alive at the time of the chart review. Demographic Characteristics of the Patient These included age, gender, and race/ethnicity. Owing to small sample sizes, race/ethnicity was coded as “White” or “Not White” (Hispanic/Latino status was coded independently of White/non-White status). Insurance Status Patients were coded as having public insurance if they had insurance plans for low-income individuals (Medicaid, Prepaid Medical Assistance Program, or Minnesota Care) at the time of the chart review. Previous insurance status was not available. Date of the Index Maltreatment Diagnosis The date at which the maltreatment-related code was assigned was extracted. Duration in the System We coded the number of days between the first and last encounter involving direct contact in the patient’s chart between January 1, 2011 and March 30, 2015, and whether the only encounter in the chart happened at the time of the index maltreatment diagnosis. Thus, in the following analyses, duration refers only to the duration in the system between these two dates, regardless of the patient’s prior or subsequent time in the system. To determine whether maltreated patients came in just before the index maltreatment diagnosis was assigned or left shortly after it was assigned, we calculated the number of days between the index maltreatment diagnosis and the dates of the first and last encounters. Primary Care Provider To determine whether the patients had a consistent provider familiar with their history and managing their care, we coded whether the child had a PCP at the time of the most recent encounter. Previous PCP information was not available. Outpatient Procedures and Services Information on all procedures and services provided to the patients was extracted. Procedures with Current Procedural Terminology (CPT ®) codes were separated into preventive, ED, and office visits. ED (CPT codes 99281-5) and office visits (CPT codes 99211-5) were further divided into five severity levels, as defined by CPT, from least (1) to most severe (5). To simplify analyses, we collapsed severity levels for ED and office visits to create two dichotomous variables: any ED visits and any office visits. If an office visit included preventive medicine services, the office visit and the preventive visit would be coded and billed separately. Office visits that did not include preventive services could include, for, example, acute care for illness or injury. Definitions were accessed from the CTSI’s i2b2 tool (https://www.ctsi.umn.edu/researcher-resources/tools-and-software/i2b2). Definitions of these different types and levels of visits are provided in Appendix A. Hospitalizations As the distribution of the number of hospitalizations was too skewed for analysis as a continuous outcome, we coded hospitalizations as present or absent. Missed Appointments For each encounter or procedure with a service ID, data were extracted regarding whether the patient canceled, did not show up for the appointment, or left without being seen. If any of these three categories was recorded multiple times on the same date, we coded only one instance. The “missed appointment” variable was defined as presence versus absence of any of these three types of events. Manual Chart Review Eleven coders reviewed the charts manually to make sure there were references to child maltreatment in the maltreated group’s EHRs and no references to child maltreatment in the comparison group’s EHRs. Coders were trained on at least 23 charts, including at least 10 patients who were eligible. Next, they rated 10 additional charts of eligible participants. These charts had already been independently rated by at least two other raters who had discussed their ratings with each other and arrived at a consensus. Coders were required to achieve a minimum reliability of 0.6 for each code on these charts compared with the consensus ratings before proceeding further. After coders met this criterion, a random third of each coder’s charts (range: 26–39%) was reviewed by another coder. The two coders then met to resolve discrepancies. These random checks were conducted throughout the study. Interrater reliability was calculated with Cohen’s kappa for the charts used for calculating whether raters met the minimum reliability criterion. Fleiss’s kappa, averaged across all rater pairs, was used for calculating reliability for the remaining patients, as they were rated by different subsets of raters. For charts used for calculating whether raters met the minimum reliability criterion, we used a two-way random effects model. For the remaining charts, we used a one-way random effects model. Variables Extracted Through Manual Chart Review Exclusion Criteria The initial criteria yielded 631 patients with maltreatment-related codes. Eligibility was checked further through manual chart review for both groups. Patients initially assigned to the maltreated group were excluded from analyses (a) if their records were inaccessible because of EHR-assigned confidentiality restrictions (N = 15); (b) if there was no mention of maltreatment in their chart (N = 7); (c) if the only type of maltreatment mentioned in the chart was abuse by the patient’s own boyfriend or girlfriend (N = 21) or bullying by peers (N = 50); (d) if clinical examination showed no evidence of maltreatment despite an initial allegation, there was no mention in the chart of any other ongoing investigation, and there were no referrals by the clinicians aimed directly at addressing issues related to maltreatment (N = 50); (e) for a combination of these reasons (N = 3); or (f) for other reasons (e.g., sexual assault by a peer, the patient neglecting her own child, a maltreatment allegation concluded by the clinician to be a psychotic delusion, N = 22). Patients initially assigned to the comparison group were excluded (a) if their charts were inaccessible (N = 3), (b) if their charts indicated current or past history of any type of child maltreatment (N = 35), or (c) for other reasons (N = 4), such as having been adopted, despite no mention of maltreatment in the chart or history of sexual assault that did not qualify as child maltreatment. Finally, 44 patients in the maltreated group were excluded because they could not be matched to a comparison patient. Interrater reliability for the decision to include or exclude a patient was 0.71, with coders agreeing on 253 of 283 cases. In addition, the primary author (C.K.) reviewed all excluded cases and ensured that the criteria were being applied uniformly across patients. Foster Care To facilitate comparison with other studies, we coded whether the patient had been in foster care during the data collection period. Interrater reliability was 0.86. Timing of Maltreatment Manual chart review was used to document whether there was any maltreatment during the data collection period (i.e., January 1, 2011 to March 30, 2015). It should be noted that the assignment of the maltreatment-related diagnosis in the EHR could have occurred after the maltreatment, when the health-care provider became aware of the maltreatment. Eligibility for the study was determined based on the assignment of a maltreatment-related code during the data collection period, not necessarily on the actual occurrence of maltreatment during this period. Interrater reliability for this variable was 0.71. Data Analysis Data were checked for missing values, frequencies, minima, and maxima, and continuous data were also checked for outliers and deviations from normality. No outliers were found. As the maltreated and comparison patients were matched at an individual level, we used tests for dependent samples to analyze the data. For all tests, assignment of maltreatment-related codes was the key predictor variable. We identified the dichotomous variables of public insurance and race as potential covariates, and used McNemar’s tests of association to test associations between them and maltreatment status. Next, we used predictive mean matching to impute missing scores (the maximum proportion of missing data from any of the imputed variables was 2.5%). Linear mixed-effects models were used to compare groups on total duration in the system, with patients within matched pairs included as nested random effects, and race and public insurance status as covariates. For all dichotomous response variables, we used logistic mixed-effects models for binomial outcomes, with patients nested within matched pairs, and race, public insurance status, and duration in the system as covariates. Variance was adjusted by specifying random intercepts for the matched pairs. Significance level was set at p = .05. All analyses were conducted in R, Version 3.4.1 (R Core Team, 2017). Results Demographic Characteristics The maltreatment-related diagnoses of patients included in the maltreatment group are listed in Table I. Patients’ vital status and demographic characteristics are depicted in Table II. Patients in the two groups were matched on an individual basis on age and gender; thus, groups were not compared on these variables. Two children in the maltreated group and four in the comparison group were deceased by the end of the data collection period; a statistical test was not conducted on this difference because of the small sample sizes. McNemar’s tests of the association between maltreatment status versus race and insurance status were conducted to determine whether these variables should be used as covariates. Results indicated that patients in the maltreated group were more likely to be non-White, χ2 (1) = 13.6, p < .001, and to be on public insurance, χ2 (1) = 105.0, p < .001, than comparison patients. Both variables were used as covariates in subsequent analyses. For comparison purposes, it should be noted that in 2013, ∼23% of 0–17-year-olds in the State were on public insurance (https://pqc.health.state.mn.us/mnha/Welcome.action). Table II. Demographic Characteristics of the Maltreated and Comparison Groups Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  a Maltreated and comparison groups are significantly different, p < .001. b These are the ages of the matched youth at the time of the index diagnoses of the yoked maltreated patient. Table II. Demographic Characteristics of the Maltreated and Comparison Groups Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  a Maltreated and comparison groups are significantly different, p < .001. b These are the ages of the matched youth at the time of the index diagnoses of the yoked maltreated patient. Manual chart review indicated that 80% of the patients in the maltreated group had experienced maltreatment during the 4-year data collection period of the current study. Over a quarter (27%) of the patients in the maltreated group had evidence in their charts that indicated that they had been in foster care during the data collection period. In the comparison group, only 2% were coded as having been in foster care. Duration in the System During the 4-year data collection period, patients in the maltreated group (M = 853 days; SD = 525) were in the health-care system covered by the Clinical Data Repository for fewer days than comparison patients (M = 899; SD = 496). A linear mixed-effects model on number of days in the health-care system indicated that the difference was significant (Appendix B); thus, we used duration in the system both as an outcome and as a covariate in analyses of other outcomes, as it predicts likelihood of medical encounters. We also tested whether maltreated patients tended to be seen only once and to drop out once a maltreatment code was assigned. Only 6% of the patients in the maltreated group had a single encounter during the data collection period that also occurred at the time of the index maltreatment diagnosis. Patients in the maltreated group were in the system for an average of 463 days (SD = 387) before a maltreatment code was assigned, and they remained in the system for an average of 536 days (SD = 418) between the assignment of the code and the end of the data collection period. The corresponding statistics for the comparison group were 459 days (SD = 381) and 548 days (SD = 400), based on the time of the maltreatment diagnosis of the matched patient. Group differences in these two intervals were not significant (Appendix B). Lack of PCP A PCP was listed for 89% of the comparison patients but only 80% of the maltreated patients. To determine if the maltreated patients were missing out on consistent care, we compared the maltreated and comparison groups on likelihood of not having a PCP. As can be seen in Appendix B, regression results indicated that the maltreated group had significantly greater odds of lacking a PCP than the comparison group. Outpatient Procedures and Services The proportions of patients in each group with different types of outpatient encounters are depicted in Table III. We conducted analyses to determine if the groups differed in the likelihood of preventive, office, or ED visits. There were no significant group differences in preventive visits or office visits (Appendix B). However, as shown in Appendix B, the odds of having had an ED visit was more than twice as high in the maltreated than in the comparison group. Table III. Percentages of Patients in the Maltreated and Comparison Groups Who Used Outpatient Procedures and Services Over the 4-Year Data Collection Period Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Note. Please see Appendix A for the definitions of the different types of visits. ED = emergency department. Table III. Percentages of Patients in the Maltreated and Comparison Groups Who Used Outpatient Procedures and Services Over the 4-Year Data Collection Period Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Note. Please see Appendix A for the definitions of the different types of visits. ED = emergency department. Hospitalizations We conducted analyses to determine if maltreated and comparison groups differed in the likelihood of hospitalizations. A third (33%) of the patients in the maltreated group had been hospitalized during the data collection period, compared with 26% of the patients in the comparison group; however, this difference did not reach statistical significance (Appendix B). Missed Appointments We conducted analyses to determine if the maltreated group would be more likely to have missed appointments. In the maltreated group, 22% of the youth canceled an appointment, 27% did not come in for a scheduled appointment, and 0.2% left without being seen during the 4-year data collection period. The corresponding percentages in the comparison youth were 19, 28, and 0.2%. The regression analysis confirmed that groups did not differ in the proportion of patients with missed appointments (Appendix B). Discussion Summary of Results We compared health-care utilization patterns of maltreated youth with those of age-, gender-, and clinic specialty/location-matched nonmaltreated youth. Results extend previous research by providing more detailed information about health-care use patterns, across a large medical care system, of maltreated children in the general population, not just in foster care. As expected, maltreated youth had elevated rates of ED visits, consistent with most previous research. We ruled out differences in age, gender, race, public insurance, duration in the medical system, type of specialty department, and clinic location as potential explanations for these differences in health-care use. On the other hand, there were no group differences in the proportion of patients who had any hospitalizations, or preventive visits or office visits. Previous results regarding health-care use are somewhat mixed, and results may depend on the sample (e.g., children at risk for maltreatment vs. children in foster care), data collection period (e.g., before or after reported maltreatment), data collection methods (e.g., self-report vs. EHRs), and utilization metric (e.g., services; costs; visits). Current results demonstrate that it is important to break down health-care utilization patterns by type of service. In future research, it would also be important to examine how these variables change longitudinally, as effects of maltreatment surface over time. It would also be important to document health-care utilization patterns in relation to need. In addition, we found that although a PCP was assigned to over 80% of the youth in both groups, the odds of not having a PCP was twice as high in the maltreated group. Results are consistent with those of previous studies showing that maltreated youth are more likely to change PCPs (Friedlaender, 2005) and less likely to have a medical home (Christian & Schwarz, 2011). Contrary to expectations, maltreated youth were not in the medical system covered by the Clinical Data Repository for a brief period of time. Over the 4-year data collection period, maltreated youth averaged 853 days in the system, only 46 fewer days than comparison youth. Few patients were seen only once when a maltreatment-related code was assigned and then dropped out. There were also no significant differences in the time between entry into the system from the beginning of the data collection period to the assignment of the index maltreatment diagnosis, and no differences in the time between assignment of this diagnosis and the last encounter in the system during the data collection period. Also contrary to what we expected, maltreated patients were not more likely to miss appointments. To our knowledge, this is the first study of maltreated children that has examined these variables. Limitations and Strengths First, we were not able to ascertain whether patients received medical care outside the system covered by the Clinical Data Repository. Second, EHRs provide no information on what patient needs motivated which types of health care use, or the outcomes of health-care encounters. Thus, it is not possible to know whether the care provided was appropriate to the need nor whether it was effective. Third, interrater reliabilities for the decision to include or exclude a patient and for determining whether maltreatment occurred during the data collection period were both 0.71; although this value is in the acceptable range (Hallgren, 2012), it was not as high as might be optimal. Fourth, health care in the United States is currently in flux, and results based on 2011–2015 may not be generalizable to other time periods. We previously reported (author names removed; in press) that only a fraction (0.02%) of youth in this medical system were assigned maltreatment codes over a 4-year period. In addition, the vast majority of studies of health-care utilization in maltreated children are based on youth in the child welfare system or on foster youth, and the most common type of maltreatment in these youth tends to be neglect. In contrast, the current sample was skewed toward physically and/or sexually abused youth, and only about a quarter had been in foster care during the data collection period. Further, “observation for maltreatment” was the most commonly assigned maltreatment-related code in our sample. Therefore, results are limited to this subgroup of youth. Yet, despite differences in sample characteristics, findings regarding elevated rates of ED visits replicate those of previous studies with maltreated youth (Florence et al., 2013; Harman et al., 2000; King et al., 2015; Smith et al., 2009; Thackeray et al., 2016). The study also had several important strengths, including access to a large medical system, a maltreated sample drawn from the general population rather than the foster care system, a well-matched comparison group, the use of EHRs (which may be more objective than self-report), and the availability of detailed information on frequency of health-care encounters. Clinical Implications and Suggestions for Future Research and Policy Research overwhelmingly indicates that although maltreated children have significantly elevated health needs, they do not receive the health care they need. Current results are consistent with this conclusion. Taken together with the fact that so few children in this system were assigned maltreatment-related codes (author names withdrawn, in press), we suggest that maltreatment is underidentified and that maltreated children may not be receiving services appropriate for their level of need, as seen by similar numbers of office visits between groups. Results also highlight opportunities for providers to improve the health care of maltreated children. These children are more likely to visit the ED; however, they are no different from comparison children in number of office visits, despite high health-care needs consistently reported in previous research. Two findings of the current study point to ways in which health-care utilization of maltreated children could be improved. First, maltreated children were less likely to have a PCP than comparison children. Future studies should examine more closely the role of a PCP in maltreated children’s health care and ways to make sure that the children have a PCP. For low-income children with chronic health conditions, having a consistent patient-centered medical home has been found to be associated with fewer ED visits and hospitalizations (Raphael et al., 2015). Similarly, for maltreated children or children at risk for maltreatment, a PCP can provide consistent care and identify red flags that could prevent (further) maltreatment and lead to fewer ED visits and hospitalizations. Second, data suggest that health-care providers have many opportunities to implement policies to identify children who have been maltreated or who are at risk of being maltreated, to provide appropriate referrals in time, and to advocate for and intervene with these families. That is, maltreated children remained in the same system for an average of over 2 years during a 4-year data collection period and were not more likely to miss appointments than comparison children. Over half came in for at least one preventive visit, 83% came in for at least one office visit, and they were not more likely to miss appointments. Office visits provide an important opportunity for medical professionals to identify signs of maltreatment or risk and to prevent (further) maltreatment. Pediatricians may be in a particularly important position to interface with families on this topic. A study of a random sample of parents contacted in a phone survey found that half (48%) of the parents were most likely to turn to a pediatrician regarding child discipline advice, including advice regarding corporal punishment, whereas only 18% said they would seek advice from mental health professionals (Taylor, Moeller, Hamvas, & Rice, 2013). A direction for future research aimed at improving health care for children who are maltreated or at risk for maltreatment is to examine the effects of screening all patients about adverse childhood experiences (ACEs) and recording this information in EHRs (Chapman, Dube, & Anda, 2007). A coordinated, ACE-informed approach across care providers could lead to more optimal utilization of health care, better prevention of maltreatment and health problems associated with ACEs, and reductions in unnecessary ED visits and hospitalizations. From a policy perspective, it should be noted that 87% of the children in the maltreated group were receiving public insurance, compared with 51% of the comparison group. Cuts to public insurance funding will disproportionately affect maltreated youth and result in even less optimal health care, which would likely lead to higher societal costs in the long run. Conflicts of interest: None declared. Appendix A CPT Definitions of Preventive Visits, Office Visits, and ED Visits Preventive visits (CPT codes: 99381-99429) are defined as initial or periodic “comprehensive preventive medicine evaluation [or reevaluation] and management of an individual, including an age and gender appropriate history, examination, counseling/anticipatory guidance/risk factor reduction interventions, and the ordering of laboratory/diagnostic procedures,” for new or established patients of different age ranges, brief (up to 60 min) “preventive medicine counseling and/or risk factor reduction intervention for individual,” brief screening, counseling and interventions related to smoking, tobacco use, alcohol and other substance use, “preventive medicine counseling and/or risk factor reduction intervention for an individual in group setting,” “administration and interpretation of health risk assessment instrument,” or other “preventive medicine service.” For office visits, Level 1 is “Office or other outpatient visit for the evaluation and management of an established patient that may not require the presence of a physician or other qualified healthcare professional. Usually, the presenting problem(s) are minimal. Typically, 5 minutes are spent performing or supervising these services.” Level 5 is defined as “Office or other outpatient visit for the evaluation and management of an established patient, which requires at least 2 of these 3 key components: A comprehensive history; A comprehensive examination; Medical decision making of high complexity. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s and/or family’s needs. Usually, the presenting problem(s) are of moderate to high severity. Physicians typically spend 40 minutes face-to-face with the patient and/or family.” A Level 1 ED visit is defined as an “emergency department visit for the evaluation and management of a patient, which requires these three components: A problem focused history; A problem focused examination; and Straightforward medical decision making. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s and or family’s needs. Usually, the presenting problem(s) are self limited or minor.” A Level 5 ED visit is defined as an “Emergency department visit for the evaluation and management of a patient, which requires these 3 key components within the constraints imposed by the urgency of the patient’s clinical condition and/or mental status: A comprehensive history; A comprehensive examination; and Medical decision making of high complexity. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s or family’s needs.” Appendix B Table AI. Results of the Regression Analyses   B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)    B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)  Notes. CI = confidence interval; OR = odds ratio; PCP = primary care provider; SE = standard error. Table AI. Results of the Regression Analyses   B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)    B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)  Notes. CI = confidence interval; OR = odds ratio; PCP = primary care provider; SE = standard error. References Brown D. S., Fang X., Florence C. S. ( 2011). Medical costs attributable to child maltreatment. American Journal of Preventive Medicine , 41, 627– 635. Retrieved from https://doi.org/10.1016/j.amepre.2011.08.013 http://dx.doi.org/10.1016/j.amepre.2011.08.013 Google Scholar CrossRef Search ADS PubMed  Burns B. J., Phillips S. D., Wagner H. R., Barth R. P., Kolko D. J., Campbell Y., Landsverk J. ( 2004). Mental health need and access to mental health services by youths involved with child welfare: A national survey. Journal of the American Academy of Child and Adolescent Psychiatry , 43, 960– 970. Retrieved from https://doi.org/10.1097/01.chi.0000127590.95585.65 http://dx.doi.org/10.1097/01.chi.0000127590.95585.65 Google Scholar CrossRef Search ADS PubMed  Casanueva C. E., Cross T. P., Ringeisen H. ( 2008). Developmental needs and individualized family service plans among infants and toddlers in the child welfare system. Child Maltreatment , 13, 245– 258. Retrieved from https://doi.org/10.1177/1077559508318397 http://dx.doi.org/10.1177/1077559508318397 Google Scholar CrossRef Search ADS PubMed  Chapman D. P., Dube S. R., Anda R. F. ( 2007). Adverse childhood events as risk factors for negative mental health outcomes. Psychiatric Annals , 37, 359– 364. Christian C. W., Schwarz D. F. ( 2011). Child maltreatment and the transition to adult-based medical and mental health care. Pediatrics , 127, 139– 145. Retrieved from https://doi.org/10.1542/peds.2010-2297 http://dx.doi.org/10.1542/peds.2010-2297 Google Scholar CrossRef Search ADS PubMed  EPIC—Electronic Privacy Information Center. ( n.d.). Retrieved May 28, 2017, Retrieved from https://epic.org/ Farmer E. M. Z., Mustillo S. A., Wagner H. R., Burns B. J., Kolko D. J., Barth R. P., Leslie L. K. ( 2010). Service use and multi-sector use for mental health problems by youth in contact with child welfare. Children and Youth Services Review , 32, 815– 821. Retrieved from https://doi.org/10.1016/j.childyouth.2010.01.019 http://dx.doi.org/10.1016/j.childyouth.2010.01.019 Google Scholar CrossRef Search ADS   Finkelhor D., Turner H. A., Shattuck A., Hamby S. L. ( 2015). Prevalence of childhood exposure to violence, crime, and abuse: Results from the National Survey of Children’s Exposure to Violence. JAMA Pediatrics , 169, 746– 54. Retrieved from https://doi.org/10.1001/jamapediatrics.2015.0676 Google Scholar CrossRef Search ADS PubMed  Florence C., Brown D. S., Fang X., Thompson H. F. ( 2013). Health care costs associated with child maltreatment: Impact on medicaid. Pediatrics , 132, 312– 318. Retrieved from https://doi.org/10.1542/peds.2012-2212 http://dx.doi.org/10.1542/peds.2012-2212 Google Scholar CrossRef Search ADS PubMed  Friedlaender E. Y. ( 2005). Patterns of health care use that may identify young children who are at risk for maltreatment. Pediatrics , 116, 1303– 1308. Retrieved from https://doi.org/10.1542/peds.2004-1988 http://dx.doi.org/10.1542/peds.2004-1988 Google Scholar CrossRef Search ADS PubMed  Hallgren K. A. ( 2012). Computing inter-rater reliability for observational data: An overview and tutorial. Tutorials in Quantitative Methods for Psychology , 8, 23– 34. http://dx.doi.org/10.20982/tqmp.08.1.p023 Google Scholar CrossRef Search ADS PubMed  Harman J. S., Childs G. E., Kelleher K. J. ( 2000). Mental health care utilization and expenditures by children in foster care. Archives of Pediatrics and Adolescent Medicine , 154, 1114– 1117. Retrieved from https://doi.org/10.1001/archpedi.154.11.1114 http://dx.doi.org/10.1001/archpedi.154.11.1114 Google Scholar CrossRef Search ADS PubMed  Horwitz S. M., Hurlburt M. S., Goldhaber-Fiebert J. D., Heneghan A. M., Zhang J., Rolls-Reutz J., Fisher E., Landsverk J., Stein R. E. ( 2012). Mental health services use by children investigated by child welfare agencies. Pediatrics , 130, 861– 869. Retrieved from https://doi.org/10.1542/peds.2012-1330 http://dx.doi.org/10.1542/peds.2012-1330 Google Scholar CrossRef Search ADS PubMed  McCue Horwitz S., Hurlburt M. S., Heneghan A., Zhang J., Rolls-Reutz J., Fisher E., Landsverk J., Stein R. E. ( 2012). Mental health problems in young children investigated by US child welfare agencies. Journal of the American Academy of Child and Adolescent Psychiatry , 51, 572– 581. Retrieved from https://doi.org/10.1016/j.jaac.2012.03.006 http://dx.doi.org/10.1016/j.jaac.2012.03.006 Google Scholar CrossRef Search ADS PubMed  Hurlburt M. S., Leslie L. K., Landsverk J., Barth R. P., Burns B. J., Gibbons R. D., Slymen D. J., Zhang J. ( 2004). Contextual predictors of mental health service use among children open to child welfare. Archives of General Psychiatry , 61, 1217– 1224. Retrieved from https://doi.org/10.1001/archpsyc.61.12.1217 http://dx.doi.org/10.1001/archpsyc.61.12.1217 Google Scholar CrossRef Search ADS PubMed  Kim H., Wildeman C., Jonson-Reid M., Drake B. ( 2017). Lifetime prevalence of investigating child maltreatment among US children. American Journal of Public Health , 107, 274– 280. http://dx.doi.org/10.2105/AJPH.2016.303545 Google Scholar CrossRef Search ADS PubMed  King A. J., Farst K. J., Jaeger M. W., Onukwube J. I., Robbins J. M. ( 2015). Maltreatment-related emergency department visits among children 0 to 3 years old in the United States. Child Maltreatment , 20, 151– 161. Retrieved from https://doi.org/10.1177/1077559514567176 http://dx.doi.org/10.1177/1077559514567176 Google Scholar CrossRef Search ADS PubMed  Leslie L. K., Hurlburt M. S., Landsverk J., Barth R., Slymen D. J. ( 2004). Outpatient mental health services for children in foster care: A national perspective. Child Abuse and Neglect , 28, 697– 712. Retrieved from https://doi.org/10.1016/j.chiabu.2004.01.004 http://dx.doi.org/10.1016/j.chiabu.2004.01.004 Google Scholar CrossRef Search ADS   McMillen J. C., Scott L. D., Zima B. T., Ollie M. T., Munson M. R., Spitznagel E. ( 2004). Use of mental health services among older youths in foster care. Psychiatric Services , 55, 811– 817. Google Scholar CrossRef Search ADS PubMed  Medical Home Initiatives for Children With Special Needs Project Advisory Committee. ( 2002). The medical home. Pediatrics , 110, 184– 186. Retrieved from https://doi.org/10.1542/peds.110.1.184 http://dx.doi.org/10.1542/peds.110.1.184 CrossRef Search ADS PubMed  R Core Team. ( 2017). R: A language and environment for statistical computing . Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/ Raphael J. L., Cooley W. C., Vega A., Kowalkowski M. A., Tran X., Treadwell J., Giardino A. P., Giordano T. P. ( 2015). Outcomes for children with chronic conditions associated with parent- and provider-reported measures of the medical home. Journal of Health Care for the Poor and Underserved , 26, 358– 376. Retrieved from https://doi.org/10.1353/hpu.2015.0051 http://dx.doi.org/10.1353/hpu.2015.0051 Google Scholar CrossRef Search ADS PubMed  Schneiderman J. U., Hurlburt M. S., Leslie L. K., Zhang J., Horwitz S. M. ( 2012). Child, caregiver, and family characteristics associated with emergency department use by children who remain at home after a child protective services investigation. Child Abuse and Neglect , 36, 4– 11. Retrieved from https://doi.org/10.1016/j.chiabu.2011.07.008 http://dx.doi.org/10.1016/j.chiabu.2011.07.008 Google Scholar CrossRef Search ADS PubMed  Schneiderman J. U., Kools S., Negriff S., Smith S., Trickett P. K. ( 2015). Differences in caregiver-reported health problems and health care use in maltreated adolescents and a comparison group from the same urban environment. Research in Nursing and Health , 38, 60– 70. Retrieved from https://doi.org/10.1002/nur.21634 http://dx.doi.org/10.1002/nur.21634 Google Scholar CrossRef Search ADS PubMed  Smith C. O., Thompson M. P., Johnson K., Nitsche A. M., Kaslow N. J. ( 2009). Service utilization patterns of maltreated and nonmaltreated children from low-income, African-American families. Psychiatric Services , 60, 1386– 1389. Retrieved from https://doi.org/10.1176/ps.2009.60.10.1386 http://dx.doi.org/10.1176/ps.2009.60.10.1386 Google Scholar CrossRef Search ADS PubMed  Stahmer A. C., Leslie L. K., Hurlburt M., Barth R. P., Webb M. B., Landsverk J., Zhang J. ( 2005). Developmental and behavioral needs and service use for young children in child welfare. Pediatrics , 116, 891– 900. Retrieved from https://doi.org/10.1542/peds.2004-2135 http://dx.doi.org/10.1542/peds.2004-2135 Google Scholar CrossRef Search ADS PubMed  Staudt M. M. ( 2003). Mental health services utilization by maltreated children: Research findings and recommendations. Child Maltreatment , 8, 195– 203. Retrieved from https://doi.org/10.1177/1077559503254138 http://dx.doi.org/10.1177/1077559503254138 Google Scholar CrossRef Search ADS PubMed  Sullivan D. J., van Zyl M. A. ( 2008). The well-being of children in foster care: Exploring physical and mental health needs. Children and Youth Services Review , 30, 774– 786. Retrieved from https://doi.org/10.1016/j.childyouth.2007.12.005 http://dx.doi.org/10.1016/j.childyouth.2007.12.005 Google Scholar CrossRef Search ADS   Taylor C. A., Moeller W., Hamvas L., Rice J. C. ( 2013). Parents’ professional sources of advice regarding child discipline and their use of corporal punishment. Clinical Pediatrics , 52, 147– 155. Retrieved from https://doi.org/10.1177/0009922812465944 Google Scholar CrossRef Search ADS PubMed  Thackeray J., Leonhart K., Yackey K., Cooper J., Kelleher K. ( 2016). Emergency department use for injuries by adolescents in foster care. Children and Youth Services Review , 62, 18– 21. Retrieved from https://doi.org/10.1016/j.childyouth.2016.01.011 http://dx.doi.org/10.1016/j.childyouth.2016.01.011 Google Scholar CrossRef Search ADS   Wildeman C., Emanuel N., Leventhal J. M., Putnam-Hornstein E., Waldfogel J., Lee H. ( 2014). The prevalence of confirmed maltreatment among US children, 2004 to 2011. JAMA Pediatrics , 168, 706– 13. Retrieved from https://doi.org/10.1001/jamapediatrics.2014.410 http://dx.doi.org/10.1001/jamapediatrics.2014.410 Google Scholar CrossRef Search ADS PubMed  World Health Organization. ( 1978). International classification of diseases : Ninth revision, basic tabulation list with alphabetic index . Geneva: World Health Organization. Retrieved from http://www.who.int/iris/handle/10665/39473 © The Author(s) 2018. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Pediatric Psychology Oxford University Press

Loading next page...
 
/lp/ou_press/health-care-utilization-patterns-of-maltreated-youth-Xf1tcg0H8Z
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
0146-8693
eISSN
1465-735X
D.O.I.
10.1093/jpepsy/jsy004
Publisher site
See Article on Publisher Site

Abstract

Abstract To examine in detail the health-care utilization patterns of maltreated children, we studied electronic health records (EHRs) of children assigned maltreatment-related codes in a large medical system. We compared youth with maltreatment-related diagnoses (N = 406) with those of well-matched youth (N = 406). Data were based on EHRs during a 4-year period from the University of Minnesota’s Clinical Data Repository, which covers eight hospitals and over 40 clinics across Minnesota. A primary care provider (PCP) was assigned to over 80% of youth in both groups. As expected, however, the odds of not having a PCP were twice as high in the maltreated as in the comparison group. Also as expected, maltreated youth had higher rates of emergency department visits. We ruled out differences in age, gender, race, public insurance, duration in the medical system, type of specialty department, and clinic location as potential explanations for these differences. On the other hand, there were no significant differences between maltreated and comparison youth in hospitalizations, preventive visits, or office visits. Contrary to expectations, maltreated youth were not in the medical system for just a brief period of time and were not more likely to cancel or miss appointments. The current study adds to the research literature by providing more detailed information about the nature of health-care services used by children with maltreatment-related diagnoses. emergency department, health-care utilization, hospitalization, maltreatment, primary care provider Maltreatment affects a large proportion of children in the United States. Over a third (37%) of children have a maltreatment investigation by Child Protective Services (CPS) by the time they reach age 18 years (Kim, Wildeman, Jonson-Reid, & Drake, 2017), and about one in eight (12.5%) children have substantiated maltreatment (Wildeman et al., 2014). The latest National Survey of Children’s Exposure to Violence shows a lifetime prevalence rate of 38% for any maltreatment, based on self- and parent report (Finkelhor, Turner, Shattuck, & Hamby, 2015). The vast majority of what we know about the health-care needs and service utilization of maltreated children come from studies of children involved with CPS, rather than samples of children in the community. According to these studies, maltreated children have considerable mental and physical health needs (Christian & Schwarz, 2011). Most of the studies documenting these needs are based on the National Survey of Child and Adolescent Well-being (NSCAW), which include children who have been investigated by CPS (Burns et al., 2004; Casanueva, Cross, & Ringeisen, 2008; Farmer et al., 2010; McCue Horwitz et al., 2012; Stahmer et al., 2005), other cases of substantiated maltreatment (Schneiderman, Kools, Negriff, Smith, & Trickett, 2015), and children in foster care (Sullivan & van Zyl, 2008). Studies of children in NSCAW (Farmer et al., 2010; Florence, Brown, Fang, & Thompson, 2013; Hurlburt et al., 2004; Schneiderman et al., 2015), children in foster care or otherwise involved with CPS (Harman, Childs, & Kelleher, 2000; McMillen et al., 2004; Staudt, 2003; Thackeray, Leonhart, Yackey, Cooper, & Kelleher, 2016), and children exposed to domestic violence (Smith, Thompson, Johnson, Nitsche, & Kaslow, 2009) show that they have elevated rates of inpatient and outpatient mental health, emergency department (ED), general health services, and psychotropic medication use, and children assigned to maltreatment codes have an elevated rate of hospitalizations because of physical injuries (King, Farst, Jaeger, Onukwube, & Robbins, 2015). A review of studies of health-care costs of maltreated children and adults maltreated as children (Brown, Fang, & Florence, 2011) found maltreatment to be associated with higher mental health costs and utilization of multiple providers. Annual Medicaid costs of children in the NSCAW study who were maltreated or at risk for being maltreated were found to be $2,600 higher than costs of comparable children (Florence et al., 2013). Most of these studies have examined overall patterns of use, such as total Medicaid costs, or use of any mental health services. Despite these elevated rates of health-care utilization, maltreated children, at least those involved with CPS, do not receive adequate care given their level of need (Christian & Schwarz, 2011). For example, data based on the NSCAW study show that receipt and utilization of services for mental health and developmental problems are low compared with their level of problems (Burns et al., 2004; Casanueva et al., 2008; Farmer et al., 2010; Horwitz et al., 2012; Leslie, Hurlburt, Landsverk, Barth, & Slymen, 2004; Stahmer et al., 2005). Maltreated children are also less likely to have a medical home (Christian & Schwarz, 2011), defined as “accessible, continuous, comprehensive, family centered, coordinated, compassionate, and culturally effective” care “delivered or directed by well-trained physicians who provide primary care and help to manage and facilitate essentially all aspects of pediatric care … [and who] should be able to develop a partnership of mutual responsibility and trust with [the child and family]” (Medical Home Initiatives for Children With Special Needs Project Advisory Committee, 2002, p. 184). Similarly, a case-control study found that maltreated children were more likely to experience a change in their primary care providers (PCPs) in the year before their first maltreatment report that led to foster care placement (Friedlaender, 2005). On the other hand, this same case-control study found that maltreated children were not more likely to have ED or ambulatory visits during the year before their first maltreatment report (Friedlaender, 2005). Furthermore, a comparison of adolescents with substantiated maltreatment with a nonmaltreated group from the same low-income neighborhoods showed no differences in health insurance coverage and, in fact, greater receipt of preventive medical care in the maltreated than in the comparison youth (Schneiderman et al., 2015). Thus, there is some inconsistency regarding the extent to which maltreated children use different kinds of health-care services. In short, although most research indicates that maltreated children involved with CPS have higher health-care needs, and most, but not all, studies show higher levels of health-care use, many studies also show that these youth are not receiving adequate care. The discrepancy may be related to suboptimal patterns of health-care utilization. For example, they may rely too much on ED visits as opposed to preventive care; however, there are no studies directly comparing maltreated children’s use of ED services versus preventive visits. There is also no research on the extent to which maltreated children remain in one health-care system on a consistent basis, which would allow health-care providers more opportunities for effective interventions. Given the inconsistencies in their caregivers and instability in their living arrangements, it is also possible that maltreated children may have a high number of missed or canceled appointments, which might impact their health care. However, no studies have yet examined the frequency of missed appointments along with service use. Current Study To address the gaps in the research literature regarding the nature of health-care utilization and the limitation of research so far to children already involved with CPS, we compared the electronic health records (EHRs) of children assigned maltreatment-related codes in a large medical system with those of youth without maltreatment codes matched on age, gender, and clinic service specialty and location. Although these children were assigned maltreatment-related codes in their EHRs, study criteria did not include any involvement with CPS. Thus, we were able to assess a more community-based sample than previous studies. We examined the association of maltreatment codes with health-care utilization variables that could be extracted reliably from EHRs, namely, duration in the system, having a PCP, rates of hospitalizations, outpatient procedures/services, and missed appointments. Based on previous studies, we expected the maltreated group to be less likely to have a PCP, and to have higher rates of ED use. We also expected that maltreated youth might be more likely to be seen only once and to drop out of the system, and to miss more appointments. Method Overview Data were based on EHRs from the University of Minnesota’s Clinical Data Repository, which covers eight hospitals and over 40 clinics across Minnesota that are part of the Fairview Health Services (a nonprofit academic health-care system) or the University of Minnesota Physicians (a multispecialty group practice for medical faculty at the University). Close to 2.5 million, 0–21-year-olds had encounters at these sites between January 2011 and March 2015. We requested EHRs on all youth with an International Classification of Diseases-9 (ICD-9) (World Health Organization, 1978) code related to maltreatment. Next, we conducted a manual chart review to verify the maltreatment. The final sample consisted of 406 youth who had been assigned maltreatment-related codes and 406 matched comparison youth. Electronic Health Records Access to EHRs was provided by the Clinical and Translational Science Institute (CTSI) at the University of Minnesota. Participants were selected using the Epic EHR (EPIC—Electronic Privacy Information Center, n.d.), which includes longitudinally linked charts of all patients who had encounters at participating hospitals and clinics. Data entry practices across all clinics and hospitals became more consistent after 2011; thus, we limited participants to patients seen between January 1, 2011 and March 30, 2015, when we started the study. Data were generally based on encounters that occurred within the system; however, in cases where a patient was seen by providers outside the system covered by the Data Repository, EHRs contained reports sent to the patients’ clinicians by these outside providers. Maltreated Group Patients in the maltreated group were between 0 months to 21 years of age at the time the index maltreatment diagnosis was assigned. Eligible patients had at least one of the ICD-9 codes in Table I on the “encounter diagnoses” or “problem list” in their EHRs since 2011. We included all the codes designed to indicate that the provider suspects abuse or neglect that were included in ICD-9 (World Health Organization, 1978), the standard system used in the United States to assign diagnoses to patients. These codes were extracted automatically from the charts. Table I. Percentage of Participants With Maltreatment-Related ICD-9 Codes ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  Note. ICD-9 = International Classification of Diseases-9. Table I. Percentage of Participants With Maltreatment-Related ICD-9 Codes ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  ICD-9 code  Type of maltreatment  Number (%)  995.50  Child abuse, unspecified  40 (9.9)  995.51  Child emotional/psychological abuse  8 (2.2)  995.52  Child neglect, nutritional  38 (9.4)  995.53  Child sexual abuse  48 (11.8)  995.54  Child physical abuse  43 (10.6)  995.59  Other child abuse and neglect  13 (3.2)  E967.0  Child battering and other maltreatment by father, stepfather, or boyfriend  28 (6.9)  E967.1  Child battering and other maltreatment by other specified person  3 (0.7)  E967.2  Child battering and other maltreatment by mother, stepmother, or girlfriend  5 (1.2)  E967.3  Child and adult battering and other maltreatment by spouse or partner  3 (0.7)  E967.4  Child and adult battering and other maltreatment by child  1 (0.2)  E967.5  Child and adult battering and other maltreatment by sibling  3 (0.7)  E967.9  Child and adult battering and other maltreatment by unspecified person  1 (0.2)  E904.0  Abandonment or neglect of infants and helpless persons    V71.81  Observation for abuse and neglect  122 (30)    Multiple diagnoses  50 (12.3)  Note. ICD-9 = International Classification of Diseases-9. When the index maltreatment diagnosis was “observation for abuse and neglect,” the type of maltreatment was further specified through manual chart review (see below). Because the automatically extracted information was restricted to data since 2011, manual chart review was restricted to the same time period. However, if a file from within this period revealed relevant information from before 2011 (e.g., mention of maltreatment or nature of the maltreatment), this information was included in coding. Comparison Group Eligible comparison youth had none of the maltreatment-related diagnoses above in their chart. Comparison patients were matched to eligible maltreated youth first on gender, then on age (to within 1 month). We also required comparison youth to have been seen at the same specialty department, at the same location, and at the same date (to within 1 week) as the encounter at which the index maltreatment diagnosis was assigned to the matched youth. If there were multiple patients matching these criteria, we selected the one that was closest on date of encounter. If they were equally close, we selected the one closest in age. If there were several such youth, the first one on the list was chosen. If there were no matches on these criteria, the temporal criterion for the encounter was expanded to 2 months. In cases where there were several matches with the expanded criterion, the one closest on date of encounter was selected. The vast majority (88.4%) of the patients fell within these criteria. If there were still no matches, comparison youth were matched on center, rather than department specialty, and the temporal criterion was again expanded from within 1 week to 2 months. If several comparison patients met the center criteria, the one with the closest date of encounter and, when possible, a similar department in the same center was chosen. These criteria were met by 9.4% of the comparison patients. If no matches were found with these criteria, the age-matching requirement was relaxed 1 month at a time. For maltreated patients aged <12 months, the upper limit on the age difference was 3 months. If the target patient was between 12 and 36 months of age, the upper limit was 6 months. For older ages, the upper limit was 12 months. These criteria were met by 1.5% of the patients. If still no matches were found, the center requirement was relaxed to include other centers in the vicinity. Only a few (0.7%) of the comparison group were in this subset. Whenever possible, comparison youth seen at similar specialty departments in other centers were chosen. Each comparison patient was matched to only one maltreated patient. Other Variables Extracted Automatically From EHRs Vital Status We coded whether the patient was alive at the time of the chart review. Demographic Characteristics of the Patient These included age, gender, and race/ethnicity. Owing to small sample sizes, race/ethnicity was coded as “White” or “Not White” (Hispanic/Latino status was coded independently of White/non-White status). Insurance Status Patients were coded as having public insurance if they had insurance plans for low-income individuals (Medicaid, Prepaid Medical Assistance Program, or Minnesota Care) at the time of the chart review. Previous insurance status was not available. Date of the Index Maltreatment Diagnosis The date at which the maltreatment-related code was assigned was extracted. Duration in the System We coded the number of days between the first and last encounter involving direct contact in the patient’s chart between January 1, 2011 and March 30, 2015, and whether the only encounter in the chart happened at the time of the index maltreatment diagnosis. Thus, in the following analyses, duration refers only to the duration in the system between these two dates, regardless of the patient’s prior or subsequent time in the system. To determine whether maltreated patients came in just before the index maltreatment diagnosis was assigned or left shortly after it was assigned, we calculated the number of days between the index maltreatment diagnosis and the dates of the first and last encounters. Primary Care Provider To determine whether the patients had a consistent provider familiar with their history and managing their care, we coded whether the child had a PCP at the time of the most recent encounter. Previous PCP information was not available. Outpatient Procedures and Services Information on all procedures and services provided to the patients was extracted. Procedures with Current Procedural Terminology (CPT ®) codes were separated into preventive, ED, and office visits. ED (CPT codes 99281-5) and office visits (CPT codes 99211-5) were further divided into five severity levels, as defined by CPT, from least (1) to most severe (5). To simplify analyses, we collapsed severity levels for ED and office visits to create two dichotomous variables: any ED visits and any office visits. If an office visit included preventive medicine services, the office visit and the preventive visit would be coded and billed separately. Office visits that did not include preventive services could include, for, example, acute care for illness or injury. Definitions were accessed from the CTSI’s i2b2 tool (https://www.ctsi.umn.edu/researcher-resources/tools-and-software/i2b2). Definitions of these different types and levels of visits are provided in Appendix A. Hospitalizations As the distribution of the number of hospitalizations was too skewed for analysis as a continuous outcome, we coded hospitalizations as present or absent. Missed Appointments For each encounter or procedure with a service ID, data were extracted regarding whether the patient canceled, did not show up for the appointment, or left without being seen. If any of these three categories was recorded multiple times on the same date, we coded only one instance. The “missed appointment” variable was defined as presence versus absence of any of these three types of events. Manual Chart Review Eleven coders reviewed the charts manually to make sure there were references to child maltreatment in the maltreated group’s EHRs and no references to child maltreatment in the comparison group’s EHRs. Coders were trained on at least 23 charts, including at least 10 patients who were eligible. Next, they rated 10 additional charts of eligible participants. These charts had already been independently rated by at least two other raters who had discussed their ratings with each other and arrived at a consensus. Coders were required to achieve a minimum reliability of 0.6 for each code on these charts compared with the consensus ratings before proceeding further. After coders met this criterion, a random third of each coder’s charts (range: 26–39%) was reviewed by another coder. The two coders then met to resolve discrepancies. These random checks were conducted throughout the study. Interrater reliability was calculated with Cohen’s kappa for the charts used for calculating whether raters met the minimum reliability criterion. Fleiss’s kappa, averaged across all rater pairs, was used for calculating reliability for the remaining patients, as they were rated by different subsets of raters. For charts used for calculating whether raters met the minimum reliability criterion, we used a two-way random effects model. For the remaining charts, we used a one-way random effects model. Variables Extracted Through Manual Chart Review Exclusion Criteria The initial criteria yielded 631 patients with maltreatment-related codes. Eligibility was checked further through manual chart review for both groups. Patients initially assigned to the maltreated group were excluded from analyses (a) if their records were inaccessible because of EHR-assigned confidentiality restrictions (N = 15); (b) if there was no mention of maltreatment in their chart (N = 7); (c) if the only type of maltreatment mentioned in the chart was abuse by the patient’s own boyfriend or girlfriend (N = 21) or bullying by peers (N = 50); (d) if clinical examination showed no evidence of maltreatment despite an initial allegation, there was no mention in the chart of any other ongoing investigation, and there were no referrals by the clinicians aimed directly at addressing issues related to maltreatment (N = 50); (e) for a combination of these reasons (N = 3); or (f) for other reasons (e.g., sexual assault by a peer, the patient neglecting her own child, a maltreatment allegation concluded by the clinician to be a psychotic delusion, N = 22). Patients initially assigned to the comparison group were excluded (a) if their charts were inaccessible (N = 3), (b) if their charts indicated current or past history of any type of child maltreatment (N = 35), or (c) for other reasons (N = 4), such as having been adopted, despite no mention of maltreatment in the chart or history of sexual assault that did not qualify as child maltreatment. Finally, 44 patients in the maltreated group were excluded because they could not be matched to a comparison patient. Interrater reliability for the decision to include or exclude a patient was 0.71, with coders agreeing on 253 of 283 cases. In addition, the primary author (C.K.) reviewed all excluded cases and ensured that the criteria were being applied uniformly across patients. Foster Care To facilitate comparison with other studies, we coded whether the patient had been in foster care during the data collection period. Interrater reliability was 0.86. Timing of Maltreatment Manual chart review was used to document whether there was any maltreatment during the data collection period (i.e., January 1, 2011 to March 30, 2015). It should be noted that the assignment of the maltreatment-related diagnosis in the EHR could have occurred after the maltreatment, when the health-care provider became aware of the maltreatment. Eligibility for the study was determined based on the assignment of a maltreatment-related code during the data collection period, not necessarily on the actual occurrence of maltreatment during this period. Interrater reliability for this variable was 0.71. Data Analysis Data were checked for missing values, frequencies, minima, and maxima, and continuous data were also checked for outliers and deviations from normality. No outliers were found. As the maltreated and comparison patients were matched at an individual level, we used tests for dependent samples to analyze the data. For all tests, assignment of maltreatment-related codes was the key predictor variable. We identified the dichotomous variables of public insurance and race as potential covariates, and used McNemar’s tests of association to test associations between them and maltreatment status. Next, we used predictive mean matching to impute missing scores (the maximum proportion of missing data from any of the imputed variables was 2.5%). Linear mixed-effects models were used to compare groups on total duration in the system, with patients within matched pairs included as nested random effects, and race and public insurance status as covariates. For all dichotomous response variables, we used logistic mixed-effects models for binomial outcomes, with patients nested within matched pairs, and race, public insurance status, and duration in the system as covariates. Variance was adjusted by specifying random intercepts for the matched pairs. Significance level was set at p = .05. All analyses were conducted in R, Version 3.4.1 (R Core Team, 2017). Results Demographic Characteristics The maltreatment-related diagnoses of patients included in the maltreatment group are listed in Table I. Patients’ vital status and demographic characteristics are depicted in Table II. Patients in the two groups were matched on an individual basis on age and gender; thus, groups were not compared on these variables. Two children in the maltreated group and four in the comparison group were deceased by the end of the data collection period; a statistical test was not conducted on this difference because of the small sample sizes. McNemar’s tests of the association between maltreatment status versus race and insurance status were conducted to determine whether these variables should be used as covariates. Results indicated that patients in the maltreated group were more likely to be non-White, χ2 (1) = 13.6, p < .001, and to be on public insurance, χ2 (1) = 105.0, p < .001, than comparison patients. Both variables were used as covariates in subsequent analyses. For comparison purposes, it should be noted that in 2013, ∼23% of 0–17-year-olds in the State were on public insurance (https://pqc.health.state.mn.us/mnha/Welcome.action). Table II. Demographic Characteristics of the Maltreated and Comparison Groups Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  a Maltreated and comparison groups are significantly different, p < .001. b These are the ages of the matched youth at the time of the index diagnoses of the yoked maltreated patient. Table II. Demographic Characteristics of the Maltreated and Comparison Groups Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  Variable  Maltreated (N = 406)  Comparison (N = 406)  Deceased (N)  2  4  Mean age (years) at index diagnosis (SD)  7.7 (5.8)  7.7 (5.8)b  Range: 0–20.8 years  Range: 0–20.8 years  Mean age (years) at final encounter during the study period  9.0 (6.0)  8.7 (6.1)  Range: 0–24.6 years  Range: 0–21.6 years  Gender (male:female) (%)  45:55  45:55  Public insurance (%)  87a  51a  Race (White:non-White) (%)  60:40a  71:29a  a Maltreated and comparison groups are significantly different, p < .001. b These are the ages of the matched youth at the time of the index diagnoses of the yoked maltreated patient. Manual chart review indicated that 80% of the patients in the maltreated group had experienced maltreatment during the 4-year data collection period of the current study. Over a quarter (27%) of the patients in the maltreated group had evidence in their charts that indicated that they had been in foster care during the data collection period. In the comparison group, only 2% were coded as having been in foster care. Duration in the System During the 4-year data collection period, patients in the maltreated group (M = 853 days; SD = 525) were in the health-care system covered by the Clinical Data Repository for fewer days than comparison patients (M = 899; SD = 496). A linear mixed-effects model on number of days in the health-care system indicated that the difference was significant (Appendix B); thus, we used duration in the system both as an outcome and as a covariate in analyses of other outcomes, as it predicts likelihood of medical encounters. We also tested whether maltreated patients tended to be seen only once and to drop out once a maltreatment code was assigned. Only 6% of the patients in the maltreated group had a single encounter during the data collection period that also occurred at the time of the index maltreatment diagnosis. Patients in the maltreated group were in the system for an average of 463 days (SD = 387) before a maltreatment code was assigned, and they remained in the system for an average of 536 days (SD = 418) between the assignment of the code and the end of the data collection period. The corresponding statistics for the comparison group were 459 days (SD = 381) and 548 days (SD = 400), based on the time of the maltreatment diagnosis of the matched patient. Group differences in these two intervals were not significant (Appendix B). Lack of PCP A PCP was listed for 89% of the comparison patients but only 80% of the maltreated patients. To determine if the maltreated patients were missing out on consistent care, we compared the maltreated and comparison groups on likelihood of not having a PCP. As can be seen in Appendix B, regression results indicated that the maltreated group had significantly greater odds of lacking a PCP than the comparison group. Outpatient Procedures and Services The proportions of patients in each group with different types of outpatient encounters are depicted in Table III. We conducted analyses to determine if the groups differed in the likelihood of preventive, office, or ED visits. There were no significant group differences in preventive visits or office visits (Appendix B). However, as shown in Appendix B, the odds of having had an ED visit was more than twice as high in the maltreated than in the comparison group. Table III. Percentages of Patients in the Maltreated and Comparison Groups Who Used Outpatient Procedures and Services Over the 4-Year Data Collection Period Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Note. Please see Appendix A for the definitions of the different types of visits. ED = emergency department. Table III. Percentages of Patients in the Maltreated and Comparison Groups Who Used Outpatient Procedures and Services Over the 4-Year Data Collection Period Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Type of visit  Maltreated (%)  Comparison (%)  Preventive visits  52  55  Office visits (Level 1 = least severe)       Level 1  16  19   Level 2  39  49   Level 3  75  80   Level 4  53  50   Level 5  13  13   Any level  83  85  ED visits (Level 1 = least severe)       Level 1  2  1   Level 2  29  22   Level 3  52  44   Level 4  45  39   Level 5  37  21   Any level  70  57  Note. Please see Appendix A for the definitions of the different types of visits. ED = emergency department. Hospitalizations We conducted analyses to determine if maltreated and comparison groups differed in the likelihood of hospitalizations. A third (33%) of the patients in the maltreated group had been hospitalized during the data collection period, compared with 26% of the patients in the comparison group; however, this difference did not reach statistical significance (Appendix B). Missed Appointments We conducted analyses to determine if the maltreated group would be more likely to have missed appointments. In the maltreated group, 22% of the youth canceled an appointment, 27% did not come in for a scheduled appointment, and 0.2% left without being seen during the 4-year data collection period. The corresponding percentages in the comparison youth were 19, 28, and 0.2%. The regression analysis confirmed that groups did not differ in the proportion of patients with missed appointments (Appendix B). Discussion Summary of Results We compared health-care utilization patterns of maltreated youth with those of age-, gender-, and clinic specialty/location-matched nonmaltreated youth. Results extend previous research by providing more detailed information about health-care use patterns, across a large medical care system, of maltreated children in the general population, not just in foster care. As expected, maltreated youth had elevated rates of ED visits, consistent with most previous research. We ruled out differences in age, gender, race, public insurance, duration in the medical system, type of specialty department, and clinic location as potential explanations for these differences in health-care use. On the other hand, there were no group differences in the proportion of patients who had any hospitalizations, or preventive visits or office visits. Previous results regarding health-care use are somewhat mixed, and results may depend on the sample (e.g., children at risk for maltreatment vs. children in foster care), data collection period (e.g., before or after reported maltreatment), data collection methods (e.g., self-report vs. EHRs), and utilization metric (e.g., services; costs; visits). Current results demonstrate that it is important to break down health-care utilization patterns by type of service. In future research, it would also be important to examine how these variables change longitudinally, as effects of maltreatment surface over time. It would also be important to document health-care utilization patterns in relation to need. In addition, we found that although a PCP was assigned to over 80% of the youth in both groups, the odds of not having a PCP was twice as high in the maltreated group. Results are consistent with those of previous studies showing that maltreated youth are more likely to change PCPs (Friedlaender, 2005) and less likely to have a medical home (Christian & Schwarz, 2011). Contrary to expectations, maltreated youth were not in the medical system covered by the Clinical Data Repository for a brief period of time. Over the 4-year data collection period, maltreated youth averaged 853 days in the system, only 46 fewer days than comparison youth. Few patients were seen only once when a maltreatment-related code was assigned and then dropped out. There were also no significant differences in the time between entry into the system from the beginning of the data collection period to the assignment of the index maltreatment diagnosis, and no differences in the time between assignment of this diagnosis and the last encounter in the system during the data collection period. Also contrary to what we expected, maltreated patients were not more likely to miss appointments. To our knowledge, this is the first study of maltreated children that has examined these variables. Limitations and Strengths First, we were not able to ascertain whether patients received medical care outside the system covered by the Clinical Data Repository. Second, EHRs provide no information on what patient needs motivated which types of health care use, or the outcomes of health-care encounters. Thus, it is not possible to know whether the care provided was appropriate to the need nor whether it was effective. Third, interrater reliabilities for the decision to include or exclude a patient and for determining whether maltreatment occurred during the data collection period were both 0.71; although this value is in the acceptable range (Hallgren, 2012), it was not as high as might be optimal. Fourth, health care in the United States is currently in flux, and results based on 2011–2015 may not be generalizable to other time periods. We previously reported (author names removed; in press) that only a fraction (0.02%) of youth in this medical system were assigned maltreatment codes over a 4-year period. In addition, the vast majority of studies of health-care utilization in maltreated children are based on youth in the child welfare system or on foster youth, and the most common type of maltreatment in these youth tends to be neglect. In contrast, the current sample was skewed toward physically and/or sexually abused youth, and only about a quarter had been in foster care during the data collection period. Further, “observation for maltreatment” was the most commonly assigned maltreatment-related code in our sample. Therefore, results are limited to this subgroup of youth. Yet, despite differences in sample characteristics, findings regarding elevated rates of ED visits replicate those of previous studies with maltreated youth (Florence et al., 2013; Harman et al., 2000; King et al., 2015; Smith et al., 2009; Thackeray et al., 2016). The study also had several important strengths, including access to a large medical system, a maltreated sample drawn from the general population rather than the foster care system, a well-matched comparison group, the use of EHRs (which may be more objective than self-report), and the availability of detailed information on frequency of health-care encounters. Clinical Implications and Suggestions for Future Research and Policy Research overwhelmingly indicates that although maltreated children have significantly elevated health needs, they do not receive the health care they need. Current results are consistent with this conclusion. Taken together with the fact that so few children in this system were assigned maltreatment-related codes (author names withdrawn, in press), we suggest that maltreatment is underidentified and that maltreated children may not be receiving services appropriate for their level of need, as seen by similar numbers of office visits between groups. Results also highlight opportunities for providers to improve the health care of maltreated children. These children are more likely to visit the ED; however, they are no different from comparison children in number of office visits, despite high health-care needs consistently reported in previous research. Two findings of the current study point to ways in which health-care utilization of maltreated children could be improved. First, maltreated children were less likely to have a PCP than comparison children. Future studies should examine more closely the role of a PCP in maltreated children’s health care and ways to make sure that the children have a PCP. For low-income children with chronic health conditions, having a consistent patient-centered medical home has been found to be associated with fewer ED visits and hospitalizations (Raphael et al., 2015). Similarly, for maltreated children or children at risk for maltreatment, a PCP can provide consistent care and identify red flags that could prevent (further) maltreatment and lead to fewer ED visits and hospitalizations. Second, data suggest that health-care providers have many opportunities to implement policies to identify children who have been maltreated or who are at risk of being maltreated, to provide appropriate referrals in time, and to advocate for and intervene with these families. That is, maltreated children remained in the same system for an average of over 2 years during a 4-year data collection period and were not more likely to miss appointments than comparison children. Over half came in for at least one preventive visit, 83% came in for at least one office visit, and they were not more likely to miss appointments. Office visits provide an important opportunity for medical professionals to identify signs of maltreatment or risk and to prevent (further) maltreatment. Pediatricians may be in a particularly important position to interface with families on this topic. A study of a random sample of parents contacted in a phone survey found that half (48%) of the parents were most likely to turn to a pediatrician regarding child discipline advice, including advice regarding corporal punishment, whereas only 18% said they would seek advice from mental health professionals (Taylor, Moeller, Hamvas, & Rice, 2013). A direction for future research aimed at improving health care for children who are maltreated or at risk for maltreatment is to examine the effects of screening all patients about adverse childhood experiences (ACEs) and recording this information in EHRs (Chapman, Dube, & Anda, 2007). A coordinated, ACE-informed approach across care providers could lead to more optimal utilization of health care, better prevention of maltreatment and health problems associated with ACEs, and reductions in unnecessary ED visits and hospitalizations. From a policy perspective, it should be noted that 87% of the children in the maltreated group were receiving public insurance, compared with 51% of the comparison group. Cuts to public insurance funding will disproportionately affect maltreated youth and result in even less optimal health care, which would likely lead to higher societal costs in the long run. Conflicts of interest: None declared. Appendix A CPT Definitions of Preventive Visits, Office Visits, and ED Visits Preventive visits (CPT codes: 99381-99429) are defined as initial or periodic “comprehensive preventive medicine evaluation [or reevaluation] and management of an individual, including an age and gender appropriate history, examination, counseling/anticipatory guidance/risk factor reduction interventions, and the ordering of laboratory/diagnostic procedures,” for new or established patients of different age ranges, brief (up to 60 min) “preventive medicine counseling and/or risk factor reduction intervention for individual,” brief screening, counseling and interventions related to smoking, tobacco use, alcohol and other substance use, “preventive medicine counseling and/or risk factor reduction intervention for an individual in group setting,” “administration and interpretation of health risk assessment instrument,” or other “preventive medicine service.” For office visits, Level 1 is “Office or other outpatient visit for the evaluation and management of an established patient that may not require the presence of a physician or other qualified healthcare professional. Usually, the presenting problem(s) are minimal. Typically, 5 minutes are spent performing or supervising these services.” Level 5 is defined as “Office or other outpatient visit for the evaluation and management of an established patient, which requires at least 2 of these 3 key components: A comprehensive history; A comprehensive examination; Medical decision making of high complexity. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s and/or family’s needs. Usually, the presenting problem(s) are of moderate to high severity. Physicians typically spend 40 minutes face-to-face with the patient and/or family.” A Level 1 ED visit is defined as an “emergency department visit for the evaluation and management of a patient, which requires these three components: A problem focused history; A problem focused examination; and Straightforward medical decision making. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s and or family’s needs. Usually, the presenting problem(s) are self limited or minor.” A Level 5 ED visit is defined as an “Emergency department visit for the evaluation and management of a patient, which requires these 3 key components within the constraints imposed by the urgency of the patient’s clinical condition and/or mental status: A comprehensive history; A comprehensive examination; and Medical decision making of high complexity. Counseling and/or coordination of care with other physicians, other qualified health care professionals, or agencies are provided consistent with the nature of the problem(s) and the patient’s or family’s needs.” Appendix B Table AI. Results of the Regression Analyses   B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)    B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)  Notes. CI = confidence interval; OR = odds ratio; PCP = primary care provider; SE = standard error. Table AI. Results of the Regression Analyses   B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)    B (95% CI or SE)  t/z  p  OR (95% CI)  Total duration in system            Intercept  1,000.70 (931.25 0–1,070.15)          Maltreatment status  −72.65 (−141.07 to −4.23)  −2.09  .038      Race  −143.41 (−218.46 to −68.35)  −3.76  <.001      Public insurance  −118.91 (−200.13 to −37.70)  −2.88  .004    Duration between beginning of data collection period and assignment of maltreatment diagnosis            Intercept  490.11 (437.31–542.91)          Maltreatment status  −19.31 (−64.48 to 25.85)  −0.84  .41      Race  −14.48 (−69.62 to 40.65)  −0.52  .61      Public insurance  −63.01 (−122.43 to −3.59)  −2.09  .04    Duration between assignment of maltreatment diagnosis and end of the data collection period            Intercept  573.34 (518.42–628.26)          Maltreatment status  −1.95 (−51.52 to 47.63)  −0.08  .94      Race  −93.31 (−152.86 to −33.76)  −3.08  .002      Public insurance  −19.14 (−81.86 to 43.58)  −0.60  .55    No PCP            Intercept  −2.28 (0.37)          Maltreatment status  0.81 (0.25)  3.29  <.001  2.26 (1.39–3.67)    Race  1.05 (0.25)  4.16  <.001  2.85 (1.74–4.66)    Public insurance  0.45 (0.28)  1.60  .011  1.56 (0.91–2.67)    Duration in system  −1.64 (0.39)  −4.17  <.001  0.19 (0.09–0.42)  Preventive visits            Intercept  −1.93 (0.33)          Maltreatment status  −0.06 (0.21)  −0.29  .77  0.94 (0.63–1.41)    Race  −0.40 (0.23)  −1.75  .08  0.67 (0.43–1.05)    Public insurance  −0.20 (0.25)  −0.80  .42  0.82 (0.51–1.33)    Duration in system  4.47 (0.49)  9.04  <.001  87.22 (33.10–229.87)  Office visits            Intercept  4.95 (1.16)          Maltreatment status  −.0.46 (0.47)  −0.99  .32  0.63 (0.25–1.58)    Race  −1.48 (0.61)  −2.42  .02  0.23 (0.07–0.75)    Public insurance  −0.49 (0.67)  −0.73  .47  0.62 (0.17–2.27)    Duration in system  8.83 (1.52)  5.81  <.001  6,811.88 (347.58–133,500.6)  ED visits            Intercept  0.43 (0.34)          Maltreatment status  0.90 (0.23)  3.86  <.001  2.45 (1.56–3.86)    Race  −0.07 (0.25)  −0.26  .79  0.94 (0.57–1.54)    Public insurance  −0.42 (0.27)  −1.60  .11  0.65 (0.39–1.10)    Duration in system  2.09 (0.48)  4.40  <.001  8.07 (3.18–20.47)  Hospitalizations            Intercept  −0.31 (.86)          Maltreatment status  0.47 (0.27  1.77  .08  1.60 (0.95–2.68)    Race  0.53 (0.33)  1.60  .11  1.70 (0.89–3.24)    Public insurance  −0.45 (0.34)  −1.32  .19  0.64 (0.33–1.24)    Duration in system  1.23 (0.66)  1.86  .06  3.44 (0.93–12.65)  Missed appointments            Intercept  −1.58 (0.28)          Maltreatment status  −0.01 (0.18)  −0.06  .95  0.99 (0.69–1.41)    Race  −0.07 (0.20)  −0.38  .71  0.93 (0.63–1.36)    Public insurance  −0.29 (0.21)  −1.35  .18  0.75 (0.49–1.14)    Duration in system  1.87 (0.34)  5.51  <.001  6.45 (3.33–12.56)  Notes. CI = confidence interval; OR = odds ratio; PCP = primary care provider; SE = standard error. References Brown D. S., Fang X., Florence C. S. ( 2011). Medical costs attributable to child maltreatment. American Journal of Preventive Medicine , 41, 627– 635. Retrieved from https://doi.org/10.1016/j.amepre.2011.08.013 http://dx.doi.org/10.1016/j.amepre.2011.08.013 Google Scholar CrossRef Search ADS PubMed  Burns B. J., Phillips S. D., Wagner H. R., Barth R. P., Kolko D. J., Campbell Y., Landsverk J. ( 2004). Mental health need and access to mental health services by youths involved with child welfare: A national survey. Journal of the American Academy of Child and Adolescent Psychiatry , 43, 960– 970. Retrieved from https://doi.org/10.1097/01.chi.0000127590.95585.65 http://dx.doi.org/10.1097/01.chi.0000127590.95585.65 Google Scholar CrossRef Search ADS PubMed  Casanueva C. E., Cross T. P., Ringeisen H. ( 2008). Developmental needs and individualized family service plans among infants and toddlers in the child welfare system. Child Maltreatment , 13, 245– 258. Retrieved from https://doi.org/10.1177/1077559508318397 http://dx.doi.org/10.1177/1077559508318397 Google Scholar CrossRef Search ADS PubMed  Chapman D. P., Dube S. R., Anda R. F. ( 2007). Adverse childhood events as risk factors for negative mental health outcomes. Psychiatric Annals , 37, 359– 364. Christian C. W., Schwarz D. F. ( 2011). Child maltreatment and the transition to adult-based medical and mental health care. Pediatrics , 127, 139– 145. Retrieved from https://doi.org/10.1542/peds.2010-2297 http://dx.doi.org/10.1542/peds.2010-2297 Google Scholar CrossRef Search ADS PubMed  EPIC—Electronic Privacy Information Center. ( n.d.). Retrieved May 28, 2017, Retrieved from https://epic.org/ Farmer E. M. Z., Mustillo S. A., Wagner H. R., Burns B. J., Kolko D. J., Barth R. P., Leslie L. K. ( 2010). Service use and multi-sector use for mental health problems by youth in contact with child welfare. Children and Youth Services Review , 32, 815– 821. Retrieved from https://doi.org/10.1016/j.childyouth.2010.01.019 http://dx.doi.org/10.1016/j.childyouth.2010.01.019 Google Scholar CrossRef Search ADS   Finkelhor D., Turner H. A., Shattuck A., Hamby S. L. ( 2015). Prevalence of childhood exposure to violence, crime, and abuse: Results from the National Survey of Children’s Exposure to Violence. JAMA Pediatrics , 169, 746– 54. Retrieved from https://doi.org/10.1001/jamapediatrics.2015.0676 Google Scholar CrossRef Search ADS PubMed  Florence C., Brown D. S., Fang X., Thompson H. F. ( 2013). Health care costs associated with child maltreatment: Impact on medicaid. Pediatrics , 132, 312– 318. Retrieved from https://doi.org/10.1542/peds.2012-2212 http://dx.doi.org/10.1542/peds.2012-2212 Google Scholar CrossRef Search ADS PubMed  Friedlaender E. Y. ( 2005). Patterns of health care use that may identify young children who are at risk for maltreatment. Pediatrics , 116, 1303– 1308. Retrieved from https://doi.org/10.1542/peds.2004-1988 http://dx.doi.org/10.1542/peds.2004-1988 Google Scholar CrossRef Search ADS PubMed  Hallgren K. A. ( 2012). Computing inter-rater reliability for observational data: An overview and tutorial. Tutorials in Quantitative Methods for Psychology , 8, 23– 34. http://dx.doi.org/10.20982/tqmp.08.1.p023 Google Scholar CrossRef Search ADS PubMed  Harman J. S., Childs G. E., Kelleher K. J. ( 2000). Mental health care utilization and expenditures by children in foster care. Archives of Pediatrics and Adolescent Medicine , 154, 1114– 1117. Retrieved from https://doi.org/10.1001/archpedi.154.11.1114 http://dx.doi.org/10.1001/archpedi.154.11.1114 Google Scholar CrossRef Search ADS PubMed  Horwitz S. M., Hurlburt M. S., Goldhaber-Fiebert J. D., Heneghan A. M., Zhang J., Rolls-Reutz J., Fisher E., Landsverk J., Stein R. E. ( 2012). Mental health services use by children investigated by child welfare agencies. Pediatrics , 130, 861– 869. Retrieved from https://doi.org/10.1542/peds.2012-1330 http://dx.doi.org/10.1542/peds.2012-1330 Google Scholar CrossRef Search ADS PubMed  McCue Horwitz S., Hurlburt M. S., Heneghan A., Zhang J., Rolls-Reutz J., Fisher E., Landsverk J., Stein R. E. ( 2012). Mental health problems in young children investigated by US child welfare agencies. Journal of the American Academy of Child and Adolescent Psychiatry , 51, 572– 581. Retrieved from https://doi.org/10.1016/j.jaac.2012.03.006 http://dx.doi.org/10.1016/j.jaac.2012.03.006 Google Scholar CrossRef Search ADS PubMed  Hurlburt M. S., Leslie L. K., Landsverk J., Barth R. P., Burns B. J., Gibbons R. D., Slymen D. J., Zhang J. ( 2004). Contextual predictors of mental health service use among children open to child welfare. Archives of General Psychiatry , 61, 1217– 1224. Retrieved from https://doi.org/10.1001/archpsyc.61.12.1217 http://dx.doi.org/10.1001/archpsyc.61.12.1217 Google Scholar CrossRef Search ADS PubMed  Kim H., Wildeman C., Jonson-Reid M., Drake B. ( 2017). Lifetime prevalence of investigating child maltreatment among US children. American Journal of Public Health , 107, 274– 280. http://dx.doi.org/10.2105/AJPH.2016.303545 Google Scholar CrossRef Search ADS PubMed  King A. J., Farst K. J., Jaeger M. W., Onukwube J. I., Robbins J. M. ( 2015). Maltreatment-related emergency department visits among children 0 to 3 years old in the United States. Child Maltreatment , 20, 151– 161. Retrieved from https://doi.org/10.1177/1077559514567176 http://dx.doi.org/10.1177/1077559514567176 Google Scholar CrossRef Search ADS PubMed  Leslie L. K., Hurlburt M. S., Landsverk J., Barth R., Slymen D. J. ( 2004). Outpatient mental health services for children in foster care: A national perspective. Child Abuse and Neglect , 28, 697– 712. Retrieved from https://doi.org/10.1016/j.chiabu.2004.01.004 http://dx.doi.org/10.1016/j.chiabu.2004.01.004 Google Scholar CrossRef Search ADS   McMillen J. C., Scott L. D., Zima B. T., Ollie M. T., Munson M. R., Spitznagel E. ( 2004). Use of mental health services among older youths in foster care. Psychiatric Services , 55, 811– 817. Google Scholar CrossRef Search ADS PubMed  Medical Home Initiatives for Children With Special Needs Project Advisory Committee. ( 2002). The medical home. Pediatrics , 110, 184– 186. Retrieved from https://doi.org/10.1542/peds.110.1.184 http://dx.doi.org/10.1542/peds.110.1.184 CrossRef Search ADS PubMed  R Core Team. ( 2017). R: A language and environment for statistical computing . Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/ Raphael J. L., Cooley W. C., Vega A., Kowalkowski M. A., Tran X., Treadwell J., Giardino A. P., Giordano T. P. ( 2015). Outcomes for children with chronic conditions associated with parent- and provider-reported measures of the medical home. Journal of Health Care for the Poor and Underserved , 26, 358– 376. Retrieved from https://doi.org/10.1353/hpu.2015.0051 http://dx.doi.org/10.1353/hpu.2015.0051 Google Scholar CrossRef Search ADS PubMed  Schneiderman J. U., Hurlburt M. S., Leslie L. K., Zhang J., Horwitz S. M. ( 2012). Child, caregiver, and family characteristics associated with emergency department use by children who remain at home after a child protective services investigation. Child Abuse and Neglect , 36, 4– 11. Retrieved from https://doi.org/10.1016/j.chiabu.2011.07.008 http://dx.doi.org/10.1016/j.chiabu.2011.07.008 Google Scholar CrossRef Search ADS PubMed  Schneiderman J. U., Kools S., Negriff S., Smith S., Trickett P. K. ( 2015). Differences in caregiver-reported health problems and health care use in maltreated adolescents and a comparison group from the same urban environment. Research in Nursing and Health , 38, 60– 70. Retrieved from https://doi.org/10.1002/nur.21634 http://dx.doi.org/10.1002/nur.21634 Google Scholar CrossRef Search ADS PubMed  Smith C. O., Thompson M. P., Johnson K., Nitsche A. M., Kaslow N. J. ( 2009). Service utilization patterns of maltreated and nonmaltreated children from low-income, African-American families. Psychiatric Services , 60, 1386– 1389. Retrieved from https://doi.org/10.1176/ps.2009.60.10.1386 http://dx.doi.org/10.1176/ps.2009.60.10.1386 Google Scholar CrossRef Search ADS PubMed  Stahmer A. C., Leslie L. K., Hurlburt M., Barth R. P., Webb M. B., Landsverk J., Zhang J. ( 2005). Developmental and behavioral needs and service use for young children in child welfare. Pediatrics , 116, 891– 900. Retrieved from https://doi.org/10.1542/peds.2004-2135 http://dx.doi.org/10.1542/peds.2004-2135 Google Scholar CrossRef Search ADS PubMed  Staudt M. M. ( 2003). Mental health services utilization by maltreated children: Research findings and recommendations. Child Maltreatment , 8, 195– 203. Retrieved from https://doi.org/10.1177/1077559503254138 http://dx.doi.org/10.1177/1077559503254138 Google Scholar CrossRef Search ADS PubMed  Sullivan D. J., van Zyl M. A. ( 2008). The well-being of children in foster care: Exploring physical and mental health needs. Children and Youth Services Review , 30, 774– 786. Retrieved from https://doi.org/10.1016/j.childyouth.2007.12.005 http://dx.doi.org/10.1016/j.childyouth.2007.12.005 Google Scholar CrossRef Search ADS   Taylor C. A., Moeller W., Hamvas L., Rice J. C. ( 2013). Parents’ professional sources of advice regarding child discipline and their use of corporal punishment. Clinical Pediatrics , 52, 147– 155. Retrieved from https://doi.org/10.1177/0009922812465944 Google Scholar CrossRef Search ADS PubMed  Thackeray J., Leonhart K., Yackey K., Cooper J., Kelleher K. ( 2016). Emergency department use for injuries by adolescents in foster care. Children and Youth Services Review , 62, 18– 21. Retrieved from https://doi.org/10.1016/j.childyouth.2016.01.011 http://dx.doi.org/10.1016/j.childyouth.2016.01.011 Google Scholar CrossRef Search ADS   Wildeman C., Emanuel N., Leventhal J. M., Putnam-Hornstein E., Waldfogel J., Lee H. ( 2014). The prevalence of confirmed maltreatment among US children, 2004 to 2011. JAMA Pediatrics , 168, 706– 13. Retrieved from https://doi.org/10.1001/jamapediatrics.2014.410 http://dx.doi.org/10.1001/jamapediatrics.2014.410 Google Scholar CrossRef Search ADS PubMed  World Health Organization. ( 1978). International classification of diseases : Ninth revision, basic tabulation list with alphabetic index . Geneva: World Health Organization. Retrieved from http://www.who.int/iris/handle/10665/39473 © The Author(s) 2018. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Journal

Journal of Pediatric PsychologyOxford University Press

Published: Feb 2, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off