Community Prevention of Child Maltreatment

1. Community Prevention of Child Maltreatment Kenneth A. Dodge Presentation to the University of California Irvine April 14, 2014 Support is appreciated from The Duke Endowment, the Pew Center on the States, NIDA, and NICHD. Colleagues are Robert Murphy, Karen O’Donnell, Ben Goodman, Jeannine Sato, and Sue Guptil.

2. Antecedents of Chronic Youth Violence Gustav Vigeland sculpture, Oslo, Norway Total annual burden of child maltreatment: > $100 billion ($114,000/case) Maternal, Infant, and Early Childhood Home Visiting Program: $1.5 billion Case stories:1. Working single mother; 2. Confused mother

3. Plan for This Presentation 1. Propose model of how child maltreatment occurs 2. Describe Durham Connects Intervention and RCT 3. Discuss financing of Connects and other programs

4. The Challenge to Change Community Rates of Child Maltreatment • The Duke Endowment’s interest and ten-year commitment • Requirements in a response: • Replicable model based in developmental science • Rigorous evaluation of impact • Community rate of maltreatment / child well-being as the dependent variable • No program had ever successfully changed community rates. • Initial plan: • Formulate a model of child maltreatment based on study of risk and processes • Pilot several intervention and policy ideas • Test through a randomized controlled trial

5. Project MOM Representative sample of 500 pregnant Durham residents, interviewed in mid-trimester and followed through age 36 months. 17% had been physically abused or neglected. 9.6% reported substance-use problems. 44% reported at least one mental health problem. 62% reported poverty or Medicaid. By age 26 months, 8% of offspring had been reported as maltreated by official record review.

6. Mothers’ Prenatal Hostile Attributions about Infant Intentions Predict Later Child Maltreatment (Berlin, Reznick, & Dodge, 2013, JAMA Pediatrics)

7. Empirically Identified Risk Factors for Early Maltreatment(Berlin, Appleyard, & Dodge, 2011, Child Development) Maternal Social Isolation .14* .19* Maternal History of Abuse Maltreatment of Offspring by Age 26 Months • Maternal • Processing • of Cues • Hostile • Attributions • Aggressive • Problem Solving .15* .23*

8. Empirically Identified Risk Factors for Early Maltreatment(Appleyard, Berlin, Rosanbalm, & Dodge, 2011, Prevention Science) Maternal Substance Use Problems .20* Prenatal Mental Health Problems Maltreatment of Offspring by Age 26 Months .16* Socioeconomic Status -.51*

9. General Model of Child Maltreatment Behavior Deviant Parental Processing of Infant Information -- lack of knowledge -- hostile attributions -- poor problem solving Maltreatment of Offspring

10. General Model of Child Maltreatment Behavior Deviant Parental Processing of Infant Information -- lack of knowledge -- hostile attributions -- poor problem solving Parental Lack of Connectedness --Social -- Professional Maltreatment of Offspring

11. General Model of Child Maltreatment Behavior RISK FACTORS Healthcare: 1. parent healthcare 2. infant healthcare 3. health insurance Parenting/childcare: 4. childcare plans 5. par-inf relationship 6. manage infant cry Family safety: 7. material supports 8. family violence 9. parenting difficulties Parent mental health: 10. depression 11. substance abuse 12. emotional support Deviant Parental Processing of Infant Information -- lack of knowledge -- hostile attributions -- poor problem solving Parental Lack of Connectedness --Social -- Professional Maltreatment of Offspring

12. General Model of Prevention of Child Maltreatment Identify families at risk Healthcare: 1. parent healthcare 2. infant healthcare 3. health insurance Parenting/childcare: 4. childcare plans 5. par-inf relationship 6. manage infant cry Family safety: 7. material supports 8. family violence 9. parenting difficulties Parent mental health: 10. depression 11. substance abuse 12. emotional support Improve parents’ processing patterns Connect families to social and professional resources as needed Healthy Child

13. Traditional Institute of Medicine Modelof Intervention Development (Mrazek & Haggerty, 1994) • Basic science inspires intervention • Intervention is developed in a university setting under pristine circumstances • Randomized controlled trial (efficacy) with volunteer sample • Replicate in community setting (effectiveness trial) • Scale up to other communities

14. Reasons Why Scaling Up Small Programs Has Failed • Rarely intend to have policy impact at the outset • Selection bias in who participates in university study • Heterogeneity of population • When scaling up, penetration and retention low (35-50%) • Degradation of intervention fidelity and quality (“scale-up penalty” of 50%) • Over-estimate of community capacity to meet needs • NFP relies on nurse to assert competitive advantage for resources

15. Model of Universal Parent Intervention 1. Top down policy: -- Preventive System of Care -- Align resources -- Screen all families 2. Bottom up with families: -- Assess to identify risks/needs -- Improve community connectedness -- “Teach” cognitions

16. Three Steps to Durham Connects 1. Connect with family • Universal recruitment at birthing hospital • Home visit(s) by public health nurse • Assess 12 risk factors, quantify risk 2. Connect family with community, as needed • Professional, paraprofessional, and natural 3. So that parents can connect with infant • Improve cognitions, parent-infant relationship

17. Durham Connects • Piloted for three years before RCT. • Per-family cost of about $700, delivered universally. • 4-7 intervention contacts with triaging. • Birthing-hospital visit • 1-3 home visits between 3-8 weeks of infant age • 1-2 contacts with a community service provider • Follow-up one month later • Community resources aligned to improve capacity (e.g., Cribs for Kids, Mentors, bus routes, DSS worker) • Agency MoA to follow a Preventive System of Care.

18. Evaluation Design for Durham Connects • Randomly assign by even-odd birthdate • 4,780 births between 7-1-09 and 12-31-10 • Recruit even birthdates into intervention • No contact with controls • Analyze by intent-to-treat • Administrative record review of all births • Random sample (n=686, 80.0% participation) from birth records for in-home interview at age 6 mos.

19. Implementation Findings • Penetration • 80.0% of families agree • Of these, 85.9% complete • Fidelity to protocol • Independent rater for 11% • 85% compliance by nurse

20. Scoring of Risk • Nurse scores each of 12 risk factors on 4-point scale • “1” indicates no risk • “2” indicates minor risk, resolved by nurse • “3” indicates considerable risk, referral • “4” indicates imminent risk, emergency • -- Inter-rater reliability of scoring of risk: Kappa = .69 • 45% of families score at least one “3” • 49% of families score “2”s but no higher • 6% of families score all “1”s • 39% of all families connected to community service

21. Number of Community Connections Reported at Age 6 months (Dodge et al., 2013, Amer J Pub Health) Effect Size = .28, p < .01

22. Impacts at Age 6 Months(Dodge et al., 2013, Pediatrics) • Mother-reported positive parenting behaviors • -- higher for intervention than control • (ES = .25, p < .01) • Blinded observer-rated mother parenting quality • -- higher for intervention than control • (ES= .23, p < .05) • 3. Child care center quality rating (when in care) • -- higher for intervention than control • (ES= .85, p < . 01)

23. Impacts at at Age 6 Months(Dodge et al., 2013, Pediatrics) • 4. Mother-rated father-infant relationship • -- better for intervention • (ES=.21, p<.07) • 5. Observer-rated home safety • -- better for intervention • (ES=.22, p<.05) • 6. Prob of mother clinical anxiety • -- lower for intervention • (OR=.65,p<.04)

24. Overall Emergency Health Care ServicesReported at Age 6 months (Dodge et al., 2013 Pediatrics) ES= .26, p < .001 Score is sum of # of hospital nights plus # of emergency visits

25. Intervention C = -0.36* -------------- Cʹ = -0.29† Mediation of Intervention Effect on Emergency Health Care Services(Goodman, Dodge, et al., 2012) 0.11** -0.24† 0.11** Mother Mental Health Child EM Care Use -0.29† Parent-Child Relationship Quality

27. Intervention effect is 59% reduction ˅------------------------------------------------˅

28. Intervention effect is 59% reduction ˅------------------------------------------------˅ Intervention effect is 31% more reduction ˅--------------------------------------------------˅

29. Cumulative Emergency Care at Age 12 Months (Dodge et al., 2013, Pediatrics) ES=.19 Interaction Effect p<.001

30. Cumulative Emergency Care at Age 12 Months (Dodge et al., 2013, Pediatrics) ES=.22

31. Cumulative Emergency Care at Age 12 Months (Dodge et al., 2013, Pediatrics) ES=.36 Interaction Effect p<.01

32. DC Impact at Age 24-Months

33. Benefit-Cost Analysis of Intervention Impact at Age 12 Months • Durham Connects intervention costs: $700/assigned family • Emergency Care Outcome Costs: CONTROL DC • $ 423 per emergency visit x .83 = $ 351 x .68 = $288 • $3,722 per hospital night x .74 = $ 2,754 x .11 = $409 • (OCC - OCI) ($3,105 - $697) • BCRDC = ___________ = ____________ = $ 3.44 • (ICI - ICC) $700 • For Durham, NC: • 3,187 resident births/year • Total emergency care costs without DC: $ 9,895,635 • Durham Connects would cost: $ 2,230,900 • Durham Connects would yield savings of: $ 7,674,296

34. Financing Early Intervention Barriers to Medicaid payment - Risk - No upfront capital Private investor solution: Social Impact Bonds - started in UK - 3 in U.S., more coming - NYC by Bloomberg - Utah by Goldman/Pritzker - SC by Goldman/Pritzker $250mil fund by Goldman Sachs

35. Current Plans 1. Continue follow-up of RCT through age 66 months. 2. New RCT now ongoing. 3. Disseminate to 4 rural NC counties. 4. Arrange financing in NC.