Effect of a Universal Postpartum Nurse Home Visiting Program on Child Maltreatment and Emergency Medical Care at 5 Years of Age: A Randomized Clinical Trial | Child Abuse | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Visual Abstract. Effects of Universal Postpartum Nurse Home Visiting Program on Child Maltreatment and Emergency Medical Care
Effects of Universal Postpartum Nurse Home Visiting Program on Child Maltreatment and Emergency Medical Care
Figure 1.  CONSORT 2010 Flow Diagram for Family Connects Randomized Clinical Trial Implementation and Evaluation
CONSORT 2010 Flow Diagram for Family Connects Randomized Clinical Trial Implementation and Evaluation
Figure 2.  Cumulative Mean Number of Child Protective Services Investigations and Emergency Medical Care Episodes From Birth to 60 Months of Age, by Intervention Group (N = 531)
Cumulative Mean Number of Child Protective Services Investigations and Emergency Medical Care Episodes From Birth to 60 Months of Age, by Intervention Group (N = 531)
Table 1.  Preintervention Sample Characteristics
Preintervention Sample Characteristics
Table 2.  Descriptive Statistics for Child Maltreatment Investigations and Emergency Medical Care Use Through 5 Years of Age by Treatment Group and Subgroup
Descriptive Statistics for Child Maltreatment Investigations and Emergency Medical Care Use Through 5 Years of Age by Treatment Group and Subgroup
Table 3.  Main Effect Analyses Examining Family Connects Effect on Total Child Maltreatment Investigations and Emergency Medical Care Use Through 5 Years of Age (N = 531)
Main Effect Analyses Examining Family Connects Effect on Total Child Maltreatment Investigations and Emergency Medical Care Use Through 5 Years of Age (N = 531)
1.
U.S. Department of Health & Human Services, Administration for Children & Families, Administration on Children, Youth and Families, Children’s Bureau. Child Maltreatment 2019. Published February 2021. Accessed April 2, 2021. https://www.acf.hhs.gov/sites/default/files/documents/cb/cm2019.pdf
2.
Alpern  ER, Clark  AE, Alessandrini  EA,  et al; Pediatric Emergency Care Applied Research Network (PECARN).  Recurrent and high-frequency use of the emergency department by pediatric patients.   Acad Emerg Med. 2014;21(4):365-373. doi:10.1111/acem.12347 PubMedGoogle ScholarCrossref
3.
U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Early childhood home visiting models: reviewing evidence of effectiveness. OPRE report #2020-126. Updated December 2020. Accessed January 15, 2021. https://homvee.acf.hhs.gov/sites/default/files/2020-12/HomVEE_Summary_Brief.pdf
4.
Michalopoulos  C, Faucetta  K, Hill  CJ,  et al. Impacts on family outcomes of evidence-based early childhood home visiting: results from the Mother and Infant Home Visiting Program Evaluation. OPRE Report 2019-07. U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Published January 18, 2019. Accessed February 9, 2019. https://www.acf.hhs.gov/sites/default/files/documents/opre/mihope_impact_report_final20_508_0.pdf
5.
Lee  H, Crowne  SS, Estarziau  M,  et al. The effects of home visiting on prenatal health, birth outcomes, and health care use in the first year of life: final implementation and impact findings from the Mother and Infant Home Visiting Program Evaluation—Strong Start. OPRE Report 2019-08. U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Published January 18, 2019. Accessed February 9, 2019. https://www.acf.hhs.gov/sites/default/files/documents/opre/mihope_strong_start_final_report_final508_3.pdf
6.
Glasgow  RE, Vinson  C, Chambers  D, Khoury  MJ, Kaplan  RM, Hunter  C.  National Institutes of Health approaches to dissemination and implementation science: current and future directions.   Am J Public Health. 2012;102(7):1274-1281. doi:10.2105/AJPH.2012.300755 PubMedGoogle ScholarCrossref
7.
Health Resources & Services Administration. Maternal, Infant, and Early Childhood Home Visiting (MIECHV) Program. Accessed December 14, 2020. https://mchb.hrsa.gov/maternal-child-health-initiatives/home-visiting-overview
8.
Dodge  KA, Goodman  WB, Murphy  R, O’Donnell  K, Sato  J.  Toward population impact from home visiting.   Zero Three. 2013;33(3):17-23.PubMedGoogle Scholar
9.
Dodge  KA, Goodman  WB.  Universal reach at birth: Family Connects.   Future Child. 2019;29(1):41-60. doi:10.1353/foc.2019.0003 Google ScholarCrossref
10.
Dodge  KA, Goodman  WB, Murphy  RA, O’Donnell  K, Sato  J, Guptill  S.  Implementation and randomized controlled trial evaluation of universal postnatal nurse home visiting.   Am J Public Health. 2014;104(1)(suppl 1):S136-S143. doi:10.2105/AJPH.2013.301361 PubMedGoogle Scholar
11.
Goodman  WB, Dodge  KA, Bai  Y, O’Donnell  KJ, Murphy  RA.  Randomized controlled trial of Family Connects: effects on child emergency medical care from birth to 24 months.   Dev Psychopathol. 2019;31(5):1863-1872. doi:10.1017/S0954579419000889 PubMedGoogle ScholarCrossref
12.
Dodge  KA, Goodman  WB, Bai  Y, O’Donnell  K, Murphy  RA.  Effect of a community agency–administered nurse home visitation program on program use and maternal and infant health outcomes: a randomized clinical trial.   JAMA Netw Open. 2019;2(11):e1914522. doi:10.1001/jamanetworkopen.2019.14522 PubMedGoogle Scholar
13.
Crane  J, Barg  M. Do early childhood programs really work? Washington, DC: Coalition for Evidence-Based Policy. Published 2003. Accessed October 13, 2018. https://pdfs.semanticscholar.org/90f5/fe541b5641037518041e66fe6b5af6d4a51a.pdf
14.
Ludwig  J, Sanbonmatsu  L, Gennetian  L,  et al.  Neighborhoods, obesity, and diabetes—a randomized social experiment.   N Engl J Med. 2011;365(16):1509-1519. doi:10.1056/NEJMsa1103216 PubMedGoogle ScholarCrossref
15.
Cohen  J.  Statistical Power for the Behavioral Sciences. 2nd ed. Erlbaum; 1988.
16.
Stroup  WW.  Generialized Linear Mixed Models: Modern Concepts, Methods, and Applications. CRC Press; 2012.
17.
Alonso-Marsden  S, Dodge  KA, O’Donnell  KJ, Murphy  RA, Sato  JM, Christopoulos  C.  Family risk as a predictor of initial engagement and follow-through in a universal nurse home visiting program to prevent child maltreatment.   Child Abuse Negl. 2013;37(8):555-565. doi:10.1016/j.chiabu.2013.03.012 PubMedGoogle ScholarCrossref
18.
Lawrence  CN, Rosanbalm  KD, Dodge  KA.  Multiple response system: evaluation of policy change in North Carolina’s child welfare system.   Child Youth Serv Rev. 2011;33(11):2355-2365. doi:10.1016/j.childyouth.2011.08.007PubMedGoogle ScholarCrossref
19.
Schafer  JL, Graham  JW.  Missing data: our view of the state of the art.   Psychol Methods. 2002;7(2):147-177. doi:10.1037/1082-989X.7.2.147PubMedGoogle ScholarCrossref
20.
Coxe  S, West  SG, Aiken  LS.  The analysis of count data: a gentle introduction to poisson regression and its alternatives.   J Pers Assess. 2009;91(2):121-136. doi:10.1080/00223890802634175PubMedGoogle ScholarCrossref
21.
Holm  S.  A simple sequential rejective method procedure.   Scand J Stat. 1979; 6: 65-70.Google Scholar
22.
Aiken  LS, West  SG.  Multiple Regression: Testing and Interpreting Interactions. Sage; 1991.
23.
Goodman  WB, Christopoulos  C, Quinn  J. Evaluation of the Family Connects Northeast Program in the North Carolina Race to the Top Early Learning Transformation Zone: final report. Duke University; 2016.
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    EXPAND ALL
    How Much Evidence Do We Need?
    Richard Krugman, MD | University of Colorado School of Medicine
    This study by the Goodman-Dodge group reported in this issue of the Journal, along with the decades of evidence from the Olds - Kitzman group (1) and others such as Healthy Families America (2) repeatedly demonstrate that home visitation gives parents and children highly positive outcomes. We now have more than three decades of solid data. So why are these programs not embedded in the US health care system, and their cost not included in all forms of public and private health insurance? Right now, there are hundreds of individual home visiting programs relying on short term foundation support or competing for their annual budget in communities where there are hundreds of chronically fiscally hungry not-for profits. Embedding them as a routine part of primary health care would also subject them to routine, ongoing quality and outcome assessments that cannot be assured in the community not-for-profit environment.

    These programs immunize children and families from devastating health, mental health and social effects and should be applied to our population the same way we immunize against measles, mumps, rubella and polio.

    References

    (1) Prenatal and Infancy Nurse Home Visiting Effects on Mothers: 18-Year Follow-up of a Randomized Trial
    David L Olds, Harriet Kitzman, et al. Pediatrics (2019) Dec;144(6):e20183889.doi: 10.1542/peds.2018-3889.

    (2) https://www.healthyfamiliesamerica.org/our-impact/state-evaluations/

    CONFLICT OF INTEREST: I am the Chair of the Board of the National Foundation to End Child Abuse and Neglect.
    READ MORE
    Original Investigation
    Public Health
    July 7, 2021

    Effect of a Universal Postpartum Nurse Home Visiting Program on Child Maltreatment and Emergency Medical Care at 5 Years of Age: A Randomized Clinical Trial

    Author Affiliations
    • 1Sanford School of Public Policy, Duke University, Durham, North Carolina
    • 2Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, Durham, North Carolina
    JAMA Netw Open. 2021;4(7):e2116024. doi:10.1001/jamanetworkopen.2021.16024
    Key Points

    Question  What is the association of the Family Connects (FC) program, a universal newborn nurse home visiting program, with child maltreatment investigations and child emergency medical care use between birth and 5 years of age?

    Findings  In this randomized clinical trial, analyses of administrative records indicated that families assigned to FC had 39% fewer Child Protective Services investigations for suspected child abuse and neglect. Families assigned to FC also had a 33% decrease in total child emergency medical care use.

    Meaning  These findings indicate that, when implemented with high quality and broad reach, the FC program can have positive long-term benefits for population rates of child well-being.

    Abstract

    Importance  The Family Connects (FC) program, a community-wide nurse home visiting program for newborns, has been shown to provide benefits for children and families through the first 2 years of life. Potential longer-term outcomes for child well-being remain unknown.

    Objective  To determine the effect of randomization to FC on child maltreatment investigations and emergency medical care through 5 years of age.

    Design, Setting, and Participants  In this randomized clinical trial, families of all 4777 resident births in Durham County, North Carolina, from July 1, 2009, to December 31, 2010, were randomly assigned to receive the FC program or treatment as usual. Impact evaluation was on an intent-to-treat basis and focused on a subsample of 549 families randomly selected from the full population and included review of hospital and Child Protective Services (CPS) administrative records. Statistical analysis was conducted from November 6, 2020, to April 25, 2021.

    Interventions  The FC programs includes 1 to 3 nurse home visits beginning at the infant age of 3 weeks designed to identify family-specific needs, deliver education and intervention, and connect families with community resources matched to their needs. Ongoing program engagement with service professionals and an electronic resource directory facilitate effective family connections to the community.

    Main Outcomes and Measures  Two primary trial outcomes were CPS-recorded child maltreatment investigations and emergency medical care use based on hospital records.

    Results  Of the 4777 randomized families, 2327 were allocated to the intervention, and 2440 were allocated to services as usual. Among the children in the full study population, 2380 (49.8%) were female, 2397 (50.2%) were male, and 3359 (70.3%) were from racial/ethnic minority groups; of the 531 children included in the impact evaluation follow-up, 284 (53.5%) were female, 247 (46.5%) were male, and 390 (73.4%) were from racial/ethnic minority groups. Negative binomial models indicated that families assigned to FC had 39% fewer CPS investigations for suspected child maltreatment through 5 years of age (95% CI, −0.80 to 0.06; 90% CI, −0.73 to −0.01; control = 44 total investigations per 100 children and intervention = 27 total investigations per 100 children); intervention effects did not differ across subgroups. Families assigned to FC also had 33% less total child emergency medical care use (95% CI, −0.59 to −0.14; 90% CI, −0.55 to −0.18; control = 338 visits and overnight hospital stays per 100 children and intervention = 227 visits and overnight hospital stays per 100 children). Positive effects held across birth risk, child health insurance, child sex, single-parent status, and racial/ethnic groups. Effects were larger for nonminority families compared with minority families.

    Conclusions and Relevance  The findings of this randomized clinical trial suggest that, when implemented with high quality and broad reach, a brief postpartum nurse home visiting program can reduce population rates of child maltreatment and emergency medical care use in early childhood.

    Trial Registration  ClinicalTrials.gov Identifier: NCT01406184

    Introduction

    Efforts to promote population health in early childhood remain a significant public health challenge in the United States. In 2019, almost 3.5 million children were subject to Child Protective Services (CPS) investigations for suspected maltreatment,1 and previous research suggests that children account for more than 28 million emergency department (ED) encounters annually.2 The risk is greatest for children from birth to 3 years, minority families, and families with low income. The Maternal, Infant, and Early Childhood Home Visiting (MIECHV) Program allocates $400 million annually to support the implementation of evidence-based home visiting models, providing more than 1 million home visits for children and families living in high-need communities. Although funding supports the dissemination of programs with demonstrated positive effects through randomized clinical trials (RCTs),3 the results do not always replicate when programs are scaled to serve larger and more diverse populations,4,5 highlighting the need for a framework for scaling and population impact.6

    Family Connects (FC) is a MIECHV-approved,7 postpartum nurse home visiting program designed to reduce child maltreatment rates and improve health outcomes at a population level. Family Connects is a systems approach to supporting families, combining top-down engagement and alignment of community resources with bottom-up identification of family-specific needs through short-term nurse home visits for every birth in a community.8 Because home visits are offered to all families, no subsequent scaling is required, and no stigma is attached to participation, maximizing community acceptance at modest cost ($500-$700 per birth9).

    Impact evaluation findings previously reported from the present RCT indicate that randomization to FC is associated with a variety of positive child and family outcomes, including increased connections to community resources, higher-quality home environments and parenting during infancy, reduced postpartum mental health symptoms, and a 37% reduction in child emergency medical care use through 24 months of age.10,11 Findings from a second, independent FC RCT identified multiple positive effects: increased community connections, reduced postpartum mental health symptoms, reduced emergency medical care use for infants with 1 or more medical risks at birth, and a 44% decrease in child maltreatment investigations through 24 months of age.12 Although these results indicate a consistent, positive pattern of benefits through the first 2 years of life, long-term child and family outcomes have yet to be investigated. This is an important lacuna, because the short-term benefits of many early childhood interventions are not sustained over time.13 In addition, rates of child maltreatment, the primary outcome for both FC RCTs, have not been examined in the present trial.

    The primary goal of the present RCT is to examine the effect of FC on child maltreatment investigations and emergency medical care use through 5 years of age. A secondary goal is to examine whether program benefits for maltreatment and emergency medical care use differ across multiple child and family characteristics. It was hypothesized that (1) randomization to FC would be associated with lower rates of child maltreatment investigations and emergency medical care use and (2) moderation analyses would identify positive effects for all subgroups, but with larger effects for high-risk families that typically have worse outcomes but in FC received more intensive nurse interventions and connections to community services for long-term support.

    Methods
    Participants, Intervention Assignment, and Evaluation Design

    Participants included families of all 4777 resident Durham County, North Carolina, births from 2 county hospitals between July 1, 2009, and December 31, 2010. As shown in Table 1, 2380 of all children born (49.8%) were female, 2397 (50.2%) were male, and 3359 (70.3%) were from racial/ethnic minority groups. All families randomly assigned to the FC program provided written informed consent for the intervention. A randomly selected subsample of 549 intervention and control group families (11.5% of the RCT population) participated in an independent outcome evaluation study, providing written informed consent for an in-home interview when the infant was 6 months of age and allowing study access to child hospital and maltreatment administrative records through 5 years of age. All study protocols were approved by the Duke Medicine institutional review board (trial protocol and statistical analysis plan in Supplement 1). This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

    Discharge records for all county births at the 2 Durham County birthing hospitals were reviewed for possible inclusion. As shown in Figure 1, families of all 4777 resident births were randomly assigned a priori to intervention or treatment as usual based on infant birth date. Families of all 2327 even-date births were assigned to receive the FC program. Program staff attempted to engage and enroll all of these families in the intervention. Families of all 2450 odd-date births were assigned to receive treatment as usual. Control families were not offered the FC program but received all other services in the community, as usual. This approach allowed for the inclusion and evaluation of the implementation for all eligible families (not only families willing to participate in an RCT) with experimental rigor, without exception, and with ethical care for privacy.

    Independent of the RCT implementation, investigators selected a random subsample of 549 families from the full population of 4777 families to participate in an impact evaluation study beginning when the infant was 6 months of age. Use of the random subsample allowed for an impact evaluation of a population-level intervention while maintaining feasible evaluation costs (eg, Moving to Opportunity for Fair Housing intervention14). Consistent with this evaluation strategy, a computer algorithm used electronic short-form birth records to randomly select 1 family per birth date for each day of the RCT implementation enrollment period (ie, from July 1, 2009, to December 31, 2010). Families that declined research participation were replaced with a randomly selected family with the same infant birth date, matched on birthing parent’s race/ethnicity to minimize possible sampling bias. Statistical power was estimated following the method of Cohen15 using 2-tailed tests with 0.80 power and a 0.05 significance level. Using techniques described by Stroup16 for power analyses with Poisson distributions, the study was sufficiently powered to detect a 27% decrease in the primary outcome.

    Interviewers attempted to contact all families selected for impact evaluation follow-up, inviting them to participate in a research study examining associations between family use of community resources and child and family well-being. Families were asked to participate in an in-home interview when the infant was 6 months of age and to consent to administrative record review through 5 years of age. Families were blinded to the study goal of evaluating the FC program, and interviewers were blinded to FC program participation status. In total, 682 families were randomly selected, and 549 agreeed to participate (80.5%; 269 from the intervention group and 280 from the control group). After all families were enrolled and interviewed, a post hoc reconciliation of birth rosters from hospital discharge and public birth records identified 18 participating families (9 from the intervention group and 9 from the control group) lacking an accurate discharge record (13 without a discharge record, 3 with an incorrect birth date, and 2 with incorrect addresses affecting residency determination). These families were removed from the analyses (the final number familes was 531).

    Study participation rates did not differ between intervention-assigned families (260 of 320 [81.3%]) and control-assigned families (271 of 344 [78.8%]). Intervention participation rates for the 260 FC program–assigned families were greater than those for the full population of intervention families (net complete, 204 of 260 [78.5%] vs 1596 of 2327 [68.8%]; P = .02). Among families participating in the FC program, dosage did not differ between the full population of intervention families (mean [SD] number of sessions, 1.52 [0.69]) and families participating in the impact evaluation study (mean [SD] number of sessions, 1.59 [0.71]) (P = .18). One participating child in the control group died at 12 months of age from a genetic condition. This child was retained in the analyses owing to the availability of partial outcome data; the inclusion or exclusion of this child does not meaningfully alter the results.

    The baseline characteristics of the consenting families were collected from hospital discharge records and the in-home interview when the infant was 6 months of age. Table 1 provides these characteristics for the full birth population, the randomly selected group for evaluation, the group that consented for participation in the evaluation, the populations assigned to the intervention and control groups, and the participating intervention and control groups. These variables were used as covariates in analyses. Measures of the intervention impact were taken from administrative records.

    Intervention

    The FC program adheres to a manualized protocol8 consisting of the following 3 core components: (1) the direct intervention consists of engagement with families shortly after birth, (2) assessment and intervention with families through 1 to 3 home visits with a registered public health nurse, and (3) a final telephone contact 4 weeks after the home visit(s). During in-home visits, nurses provide brief educational interventions for all families (eg, feeding practices) and, using a high-inference approach combining parent self-report, direct observation, and nurse clinical judgment, systematically assess family need across 12 empirically derived factors associated with child health and well-being (health care: maternal health, infant health, and health care plans; parenting and childcare: childcare plans, parent-child relationship, and management of infant crying; family material resources and safety: material supports, family and community safety, and birthing parent history of parenting difficulties; and parent well-being: mental health, substance abuse, and social-emotional support).

    Nurses address family needs based on their ratings of each factor on a 4-point scale: 1 indicates low risk, and there is no subsequent intervention; 2 indicates moderate risk, with short-term, nurse-delivered education; 3 indicates significant risk, with collaborative connections to community services and resources tailored to address particular needs (eg, postpartum depression treatment, food assistance, and long-term home visiting program); and 4 indicates imminent risk, with emergency intervention (<1% of cases). With family consent, summary reports are sent to parent and infant health care professionals to support medical home connections. The final telephone contact 4 weeks after case closure ascertains family-consumer satisfaction and confirms outcomes for community referrals.

    The second core component is a comprehensive community alignment process that includes engagement of key community services and stakeholders, creation of a comprehensive electronic directory (“agency finder”) of community resources (allowing home visiting nurses to match services to family need), and a community-program “feedback loop” with bidirectional communication supporting alignment of community services. The third component is a comprehensive electronic database that houses the electronic agency finder and serves as the clinical record of the intervention.

    As reported by Dodge et al,10 80.1% of all eligible families (1863 of 2327) scheduled an FC home visit; 85.7% of these families (1596 of 1863) successfully completed the program (net completion rate, 68.6%). Nurse adherence to the home visit protocol was 83.6% (5267 of 6304 program elements checked); interrater agreement across all 12 risk factors was substantial (κ = 0.69).15 Of 1596 families, 50 (3.1%) stopped assessment because of family choice. A total of 1453 of 1546 families (94.0%) had 1 or more nurse-identified risks targeted for intervention, 681 of 1546 families (44.0%) had 1 or more risks best addressed by community connections, and 579 of 730 referred families (79.3%) reported at least 1 successful community connection during the follow-up telephone contact 4 weeks later.

    Covariates

    Hospital administrative records were coded for the presence of child medical risks at birth (any of: birth weight, <2500 g; gestational age, <37 weeks; and any International Classification of Diseases, Ninth Revision codes indicating birth complications or trauma17) and child sex (0 = boy and 1 = girl). Single-parent household status (0 = no and 1 = yes), child Medicaid or no health insurance status (0 = no and 1 = yes), and self-identified race/ethnicity (0 = nonminority and 1 = minority) were coded from maternal report.

    Outcome Measures

    As in the pretrial registry, the primary study outcome was CPS reports for suspected maltreatment. State administrative records were examined through 5 years of age for the 531 evaluation study families. Records were coded for the total number of investigations for suspected maltreatment per child. Records were also coded for the total number of substantiated per-child investigations if evidence of maltreatment was confirmed; this outcome was not tested because only 10 of 531 families (1.9%) had 1 or more substantiations through 5 years of age. The low base rate is attributed to reforms in state child welfare laws that prioritized engaging investigated families in clinical services rather than substantiating maltreatment.18

    Hospital administrative billing records were examined through 5 years of age and coded for the total number of per-child ED visits and overnight stays in hospital. The 2 scores were summed to measure total per-child emergency medical care use.

    Missing Data

    Four of the 531 evaluation study participants (0.8%) had at least 1 missing value, representing 0.2% of all data points (12 of 6903). Following guidance by Schafer and Graham,19 single imputation was used to account for missing values from other family and demographic characteristics prior to estimating intervention effects.

    Statistical Analysis

    Statistical analysis was conducted from November 6, 2020, to April 25, 2021. Analysis was conducted on an intent-to-treat basis. We used SAS, version 9.4 (SAS Institute Inc) to estimate the effect of randomizaton to the FC program (or not) on CPS child maltreatment investigations and child emergency medical care use, regardless of intervention participation or adherence. Negative binomial regression models were used because the maltreatment and emergency medical care outcomes were count variables with skewed distributions.20 First, main effect models were estimated with infant birth risk, child health insurance status, birthing parent race/ethnicity, single-parent status, and child sex as covariates. Next, moderation analyses were estimated to examine whether intervention effects differed based on child and family characteristics. Moderators were examined individually with all covariates included in the model; a Holm-Bonferonni sequential correction was applied to account for increased type I error risk resulting from 20 individual moderation tests.21 Post hoc tests of significant interactions remaining after correction were conducted following the method of Aiken and West.22 Results are reported with 2-tailed P values and 95% and 90% CIs, with P < .10; a percentage decrease is reported as ([count per 100 control children − count per 100 intervention children]/count per control children) ×100.

    Results

    Among the children in the full study population, 2380 (49.8%) were female, 2397 (50.2%) were male, and 3359 (70.3%) were from racial/ethnic minority groups; of the 531 children included in the impact evaluation follow-up, 284 (53.5%) were female, 247 (46.5%) were male, and 390 (73.4%) were from racial/ethnic minority groups (Table 1). A total of 46 of 260 children (17.7%) in the intervention group and 59 of 271 children (21.8%) in the control group were the subject of 1 or more CPS investigations; 151 of 260 children (58.1%) in the intervention group and 193 of 271 children (71.2%) in the control group experienced 1 or more emergency medical care encounters. Descriptive statistics for child maltreatment investigations and emergency medical care, by intervention group and subgroup, are presented in Table 2.

    Intervention Effect on Child Maltreatment Investigations

    Results from negative binomial models indicated that random assignment to FC was associated with a 39% decrease in mean total per-child CPS investigations for suspected maltreatment (95% CI, −0.80 to 0.06; 90% CI, −0.73 to −0.01; control = 44 total investigations per 100 children, intervention = 27 total investigations per 100 children) (Table 3). Results from moderation models indicated that intervention effects on total maltreatment investigations did not differ across subgroups.

    Intervention Effect on Emergency Medical Care

    The results from negative binomial models indicated that randomization to the FC program was associated with a 33% decrease in the mean total per-child emergency medical care use (95% CI, −0.59 to −0.14; 90% CI, −0.55 to −0.18; control = 338 visits and overnight hospital stays per 100 children, intervention = 227 visits and overnight hospital stays per 100 children) (Figure 2).

    The results from moderation models indicated that intervention effects differed based on racial/ethnic majority or minority status of parent. Post hoc analyses revealed a positive effect of intervention assignment for both nonminority and minority families but a larger decrease for nonminority families. Among nonminority families, randomization to the FC program was associated with a 1.51-unit mean decrease in the mean total use of emergency medical care (95% CI, −1.68 to −0.32; 90% CI, −1.57 to −0.43; control = 217 visits and overnight hospital stays per 100 children, intervention = 96 visits and overnight hospital stays per 100 children) compared with control group nonminority families. Randomization to the FC program was also associated with a 0.99-unit decrease in the mean total use of emergency medical care for minority families (95% CI, −0.43 to 0.03; 90% CI, −0.39 to −0.01; control = 380 visits and overnight hospital stays per 100 children, intervention = 281 visits and overnight hospital stays per 100 children).

    Examining each component of total emergency medical care individually, we found that randomization to the FC program was associated with a 17% decrease in the mean total number of ED visits (−0.38 to 0.03; 90% CI, −0.35 to −0.002; control = 243 visits per 100 children, intervention = 202 visits per 100 children) and a 73% decrease in the mean total number of hospital overnight stays (95% CI, −1.89 to −0.14; 90% CI, −1.75 to −0.28; control = 95 overnight stays per 100 children, intervention = 25 overnight stays per 100 children). No significant interaction effects were observed for either construct.

    Discussion

    Study results demonstrate that random assignment to receive short-term, universal postpartum home visits from a nurse is associated with reduced child maltreatment rates and emergency medical care use through 5 years of age. Random assignment to the FC program was associated with a 39% decrease in the mean number of CPS investigations for suspected maltreatment and a 33% decrease in the mean rate of emergency medical care use. To our knowledge, these findings are the first reported on the 5-year effect of early home visiting on a community-wide population.

    The effect of randomization to the FC program held across every subgroup tested, including families with high and low child medical risk at birth, Medicaid or no health insurance and private insurance, single-parent and 2-parent families, racial/ethnic nonminority and minority status, and child sex. These findings of reduced rates of maltreatment investigations and emergency medical care use support the value of offering the FC program universally. Unlike some programs that target specific demographic subgroups, the FC program features community-wide implementation; these findings support that policy decision. The effect on child maltreatment investigations was observed at P < .10, rather than the typical P < .05, likely owing to lower base rates than anticipated before the trial. Nevertheless, the observed 39% decrease in the mean number of CPS investigations for suspected maltreatment is practically important for efforts to promote community-wide maltreatment prevention. In addition, the magnitude of the effect on emergency medical care was robust for both racial/ethnic minority and nonminority families, but larger for nonminority families compared with minority families. It is unclear why this racial/ethnic difference emerged; future studies should attempt understand both the reasons for the difference and whether similar patterns occur across diverse minority populations.

    Limitations

    This study has some limitations. This evaluation was conducted in only 1 medium-sized urban community with high socioeconomic diversity; findings may or may not be generalizable to communities that differ markedly in size, community resources, sociodemographic characterisitcs, or base rates of child maltreatment and emergency medical care use. We suggest similar investigations of the effect in other communities.

    Findings from a quasi-experimental field study in 4 low-resource, rural communities found that the FC program was associated with multiple short-term positive benefits for families during infancy, including reductions in emergency medical care use23; however, the effects on child maltreatment are yet to be examined. Like the present study, the FC program was implemented with high quality in these rural communities. It is not clear whether the positive effect would hold in contexts of lower-quality implementation. In addition, although the findings suggest possible positive returns for communities through avoidance of costly outcomes, a comprehensive study of the benefits and costs of the FC program is needed.

    Another limitation applies only to the outcomes studied here (child maltreatment and emergency medical care use); it is not known whether the 5-year effects of the FC program hold for the other outcomes found at 6 months of age, including community connections, mental health, and parenting. Although data on emergency medical care use were collected from both county hospitals, some families may have sought care elsewhere. The choice of care should not differ based on intervention assignment; the FC program neither endorses nor directs families toward specific medical organizations. Finally, the program completion rate among eligible families participating in the evaluation study was greater than that among eligible families in the full population (78% vs 69%). The present results may represent a high-end estimate of full-population effects.

    Conclusions

    This study found that, when implemented with high quality and broad reach, the FC program is associated with reduced rates of child maltreatment investigations and emergency medical care use through 5 years of age. These findings argue that such a public health prevention approach has benefits that extend throughout early childhood, making such programs a worthy community investment.

    Back to top
    Article Information

    Accepted for Publication: May 4, 2021.

    Published: July 7, 2021. doi:10.1001/jamanetworkopen.2021.16024

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Goodman WB et al. JAMA Network Open.

    Corresponding Author: W. Benjamin Goodman, PhD, Sanford School of Public Policy, Duke University, 2024 W Main St, Bay C, PO Box 90539, Durham, NC 27708-0539 (ben.goodman@duke.edu).

    Author Contributions: Dr Goodman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Goodman, Dodge, Murphy, O’Donnell.

    Acquisition, analysis, or interpretation of data: Goodman, Dodge, Bai.

    Drafting of the manuscript: Goodman, Dodge.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Goodman, Bai.

    Obtained funding: Goodman, Dodge.

    Administrative, technical, or material support: Goodman, Murphy.

    Supervision: Goodman, Dodge.

    Conflict of Interest Disclosures: All the authors support Family Connects program dissemination to communities and are reimbursed for actual costs of training and dissemination but do not receive any financial profit. They occasionally give lectures about the program for a modest stipend. Dr Goodman reported receiving grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, The Duke Endowment, and the Pew Center on the States during the conduct of the study; honorarium and travel expenses from Japan Society for Prevention of Child Abuse & Neglect to present on Family Connects at the JaSPCAN annual conference; and honorarium and travel expenses from Princeton University to present at a conference and coauthor a chapter on Family Connects for Future of Children outside the submitted work. No other disclosures were reported.

    Funding/Support: Funding was provided by The Duke Endowment, the Pew Center on the States, and the Eunice Kennedy Shiver National Institute of Child Health and Human Development (grant R01HD069981).

    Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Disclaimer: The views expressed are those of the authors and do not necessarily reflect the views of the funders.

    Data Sharing Statement: See Supplement 2.

    Additional Contributions: We acknowledge the contributions of many staff members and Durham, North Carolina, community leaders in implementing Family Connects and its evaluation. The authors thank Nissa Towe-Goodman, PhD, FPG Child Development Institute, University of North Carolina, Chapel Hill, and 3 anonymous reviewers for their helpful comments on previous versions of this manuscript. Dr Towe-Goodman received no financial compensation for her contribution.

    References
    1.
    U.S. Department of Health & Human Services, Administration for Children & Families, Administration on Children, Youth and Families, Children’s Bureau. Child Maltreatment 2019. Published February 2021. Accessed April 2, 2021. https://www.acf.hhs.gov/sites/default/files/documents/cb/cm2019.pdf
    2.
    Alpern  ER, Clark  AE, Alessandrini  EA,  et al; Pediatric Emergency Care Applied Research Network (PECARN).  Recurrent and high-frequency use of the emergency department by pediatric patients.   Acad Emerg Med. 2014;21(4):365-373. doi:10.1111/acem.12347 PubMedGoogle ScholarCrossref
    3.
    U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Early childhood home visiting models: reviewing evidence of effectiveness. OPRE report #2020-126. Updated December 2020. Accessed January 15, 2021. https://homvee.acf.hhs.gov/sites/default/files/2020-12/HomVEE_Summary_Brief.pdf
    4.
    Michalopoulos  C, Faucetta  K, Hill  CJ,  et al. Impacts on family outcomes of evidence-based early childhood home visiting: results from the Mother and Infant Home Visiting Program Evaluation. OPRE Report 2019-07. U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Published January 18, 2019. Accessed February 9, 2019. https://www.acf.hhs.gov/sites/default/files/documents/opre/mihope_impact_report_final20_508_0.pdf
    5.
    Lee  H, Crowne  SS, Estarziau  M,  et al. The effects of home visiting on prenatal health, birth outcomes, and health care use in the first year of life: final implementation and impact findings from the Mother and Infant Home Visiting Program Evaluation—Strong Start. OPRE Report 2019-08. U.S. Department of Health & Human Services, Administration for Children & Families, Office of Planning, Research, & Evaluation. Published January 18, 2019. Accessed February 9, 2019. https://www.acf.hhs.gov/sites/default/files/documents/opre/mihope_strong_start_final_report_final508_3.pdf
    6.
    Glasgow  RE, Vinson  C, Chambers  D, Khoury  MJ, Kaplan  RM, Hunter  C.  National Institutes of Health approaches to dissemination and implementation science: current and future directions.   Am J Public Health. 2012;102(7):1274-1281. doi:10.2105/AJPH.2012.300755 PubMedGoogle ScholarCrossref
    7.
    Health Resources & Services Administration. Maternal, Infant, and Early Childhood Home Visiting (MIECHV) Program. Accessed December 14, 2020. https://mchb.hrsa.gov/maternal-child-health-initiatives/home-visiting-overview
    8.
    Dodge  KA, Goodman  WB, Murphy  R, O’Donnell  K, Sato  J.  Toward population impact from home visiting.   Zero Three. 2013;33(3):17-23.PubMedGoogle Scholar
    9.
    Dodge  KA, Goodman  WB.  Universal reach at birth: Family Connects.   Future Child. 2019;29(1):41-60. doi:10.1353/foc.2019.0003 Google ScholarCrossref
    10.
    Dodge  KA, Goodman  WB, Murphy  RA, O’Donnell  K, Sato  J, Guptill  S.  Implementation and randomized controlled trial evaluation of universal postnatal nurse home visiting.   Am J Public Health. 2014;104(1)(suppl 1):S136-S143. doi:10.2105/AJPH.2013.301361 PubMedGoogle Scholar
    11.
    Goodman  WB, Dodge  KA, Bai  Y, O’Donnell  KJ, Murphy  RA.  Randomized controlled trial of Family Connects: effects on child emergency medical care from birth to 24 months.   Dev Psychopathol. 2019;31(5):1863-1872. doi:10.1017/S0954579419000889 PubMedGoogle ScholarCrossref
    12.
    Dodge  KA, Goodman  WB, Bai  Y, O’Donnell  K, Murphy  RA.  Effect of a community agency–administered nurse home visitation program on program use and maternal and infant health outcomes: a randomized clinical trial.   JAMA Netw Open. 2019;2(11):e1914522. doi:10.1001/jamanetworkopen.2019.14522 PubMedGoogle Scholar
    13.
    Crane  J, Barg  M. Do early childhood programs really work? Washington, DC: Coalition for Evidence-Based Policy. Published 2003. Accessed October 13, 2018. https://pdfs.semanticscholar.org/90f5/fe541b5641037518041e66fe6b5af6d4a51a.pdf
    14.
    Ludwig  J, Sanbonmatsu  L, Gennetian  L,  et al.  Neighborhoods, obesity, and diabetes—a randomized social experiment.   N Engl J Med. 2011;365(16):1509-1519. doi:10.1056/NEJMsa1103216 PubMedGoogle ScholarCrossref
    15.
    Cohen  J.  Statistical Power for the Behavioral Sciences. 2nd ed. Erlbaum; 1988.
    16.
    Stroup  WW.  Generialized Linear Mixed Models: Modern Concepts, Methods, and Applications. CRC Press; 2012.
    17.
    Alonso-Marsden  S, Dodge  KA, O’Donnell  KJ, Murphy  RA, Sato  JM, Christopoulos  C.  Family risk as a predictor of initial engagement and follow-through in a universal nurse home visiting program to prevent child maltreatment.   Child Abuse Negl. 2013;37(8):555-565. doi:10.1016/j.chiabu.2013.03.012 PubMedGoogle ScholarCrossref
    18.
    Lawrence  CN, Rosanbalm  KD, Dodge  KA.  Multiple response system: evaluation of policy change in North Carolina’s child welfare system.   Child Youth Serv Rev. 2011;33(11):2355-2365. doi:10.1016/j.childyouth.2011.08.007PubMedGoogle ScholarCrossref
    19.
    Schafer  JL, Graham  JW.  Missing data: our view of the state of the art.   Psychol Methods. 2002;7(2):147-177. doi:10.1037/1082-989X.7.2.147PubMedGoogle ScholarCrossref
    20.
    Coxe  S, West  SG, Aiken  LS.  The analysis of count data: a gentle introduction to poisson regression and its alternatives.   J Pers Assess. 2009;91(2):121-136. doi:10.1080/00223890802634175PubMedGoogle ScholarCrossref
    21.
    Holm  S.  A simple sequential rejective method procedure.   Scand J Stat. 1979; 6: 65-70.Google Scholar
    22.
    Aiken  LS, West  SG.  Multiple Regression: Testing and Interpreting Interactions. Sage; 1991.
    23.
    Goodman  WB, Christopoulos  C, Quinn  J. Evaluation of the Family Connects Northeast Program in the North Carolina Race to the Top Early Learning Transformation Zone: final report. Duke University; 2016.
    ×