Date of Award

Spring 3-14-2019

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Public Health

First Advisor

Daniel J. Whitaker, PhD

Second Advisor

Shannon Self-Brown, PhD

Third Advisor

Richard Rothenberg, MD, PhD


Background: Child maltreatment is a significant public health problem affecting more than 3.4 million children per year in the U.S. Child welfare systems, like public health and mental health systems, have moved to adopt evidenced-based programs, and many have adopted BPTs. One challenge is ensuring that when models are scaled-up, they are delivered with fidelity and positive outcomes are maintained.

Purpose: There is broad agreement that implementation of evidence-based practices is a key strategy to reducing child maltreatment. However, it is also important to examine outcomes and variation in outcomes among disseminated programs because (1) program impacts can be inconsistent in research studies and (2) program impacts tend to decrease with program dissemination. Moreover, there may be variation in program impacts associated with the providers who implementing EBPs and the families who receive them. The goals of this research are to examine program impacts of the SafeCare model across several implementations, and to explore variation in program impacts according to characteristics of providers and families.

Method: This data was derived from multiple implementations of a behaviorally based parenting program. The data utilized here were collected as part of several routine SafeCare implementations between 2008 and 2014. Data were collected on (1) family behavior change for each SafeCare module, (2) family characteristics, (3) provider characteristics, and (4) provider fidelity to the SafeCare model. The final analytic dataset consisted of 493 families served by 170 providers who had at least one complete SafeCare assessment.

Results: Change scores were analyzed via t-tests. The change for each module was significant. Health improved by 29.4 percentage points (sd = 26.8), t (214) =12.75. Safety hazards decreased by 12.54 hazards (sd = 13.9), t (176) = 19.9 and PCI/PII scores improved by 25.0 percentage points (sd = 23.9), t (290) = 17.85. Effect sizes were all very large, ranging from 1.7 to 3.0. For health, only race emerged as statistically significant. Findings from this study showed mostly null results with regard to family-level predictors of behavior change, and thus there was no evidence that SafeCare is more or less effective for families according to the variables analyzed. From the perspective of a child welfare system, one would want a program with uniform effects across client ages, races, income levels, etc. Provider demographics were generally not related to parent behavior change. Among work experience and attitudes, prior provider experience in delivering structured interventions was related to lower changes in home safety hazard reduction, and attitudes toward evidence-based practices were related to greater changes in health skill, but no consistent patterns of relationships across parent outcomes was observed. Among fidelity related predictors, having a failed fidelity session was related to lower changes in parent health knowledge, but this did not hold true for other modules. In the overall behavior change models fidelity variables were inconsistent, with the number of sessions a provider has delivered associated with positive behavior change, but mean provider fidelity associated with lower parent behavior change.