Manuscript Detail

View Revisions

Shaw, D. S., Connell, A., Dishion, T. J., Wilson, M. N., & Gardner, F. (2009). Improvements in maternal depression as a mediator of intervention effects on early childhood problem behavior. Development and Psychopathology, 21, 417–439.

Model(s) Reviewed: Family Check-Up® For Children
Manuscript screening details
Screening decision Screening conclusion
Passes screens Eligible for review
Study design details
Rating Design Attrition Baseline equivalence Reassignment Confounding factors
High Randomized controlled trial Low Established on race/ethnicity and SES; not established on baseline measures of the outcomes. None None
Notes:

footnote269

Submitted by user on Fri, 03/15/2019 - 14:29

Even though baseline equivalence was not established on baseline measures of the outcomes, this study rated high because it had low attrition, established baseline equivalence on race/ethnicity and SES, and controlled for baseline measures of some outcomes in the analyses. Those outcomes rated high. In addition to the outcomes that rated high, several outcomes in this study rated low: Eyeberg problem behavior at ages 3 and 4, externalizing behavior at ages 3 and 4, parent involvement at age 3, positive parenting at age 3, parent neutral at age 3, and Proactive parenting at age 3. These outcomes rated low because we could not assess attrition or baseline equivalence based on information reported in the study, nor was this information available from the author. Also, some outcomes in the structural equation models reported in this study were not eligible for review because the model did not estimate the direct, total effect of the intervention on the outcome.This study is part of a large RCT described by Dishion et al. (2008).

This study review was updated in 2017. Specifically, HomVEE identified additional outcomes within this study that were eligible for review in the course of the 2017 annual review. Those additional outcomes that received a moderate or high rating are reported for the study. The review of additional outcomes did not affect the overall study rating.

Study characteristics
Study participants The study included 731 families that met two criteria. First, they participated in the Supplemental Nutrition Program for Women, Infants, and Children (WIC) when their son or daughter was between 2 years 0 months old and 2 years 11 months old. Second, they met the study’s criteria for being at risk for behavior problems—defined as one standard deviation or more above normative averages in at least two of three domains: (1) child behavior problems (such as conduct problems or high-conflict relationships with adults); (2) primary caregiver problems (such as maternal depression, daily parenting challenges, self-reported substance or mental health diagnosis, or status as a teen parent at first birth); and (3) socioeconomic status (a caregiver with low educational achievement or low family income based on WIC criteria). Screening was conducted in 2002 and 2003. Of the 731 primary caregivers who agreed to participate, 41 percent had a high school diploma or GED, 32 percent had one or two years of post-high school training, and 24 percent had less than a high school diploma or GED. More than two-thirds of the randomized sample had an annual income below $20,000. Of the 731 children in the study, 50 percent were European American, 28 percent were African American, 13 percent were biracial, and 9 percent were from another racial group. Thirteen percent were Hispanic. The children were 29.9 months old on average at the time of the age 2 assessments. Forty-nine percent of the children were female, and 58 percent lived in two-parent households.
Setting Families were recruited from WIC program sites in and around Pittsburgh, Pennsylvania (37 percent of sample); Eugene, Oregon (37 percent of sample); and Charlottesville, Virginia (26 percent of sample).
Intervention services The Family Check-Up program typically involves three meetings: an initial contact meeting (a “get to know you” meeting); an assessment meeting, during which families participate in a comprehensive assessment of child and family functioning; and a feedback meeting to discuss the results of the assessment. After the feedback meeting, families can choose to participate in additional follow-up meetings. For this study, the order of the meetings was changed. All families participating in the study were given the comprehensive assessment. The researchers then randomized families into intervention and comparison groups. Following randomization, families in the intervention group participated in the initial contact and feedback meetings, which were led by parent consultants. These consultants discussed family issues and family functioning during the initial contact meeting and, during the feedback meeting, used motivational interviewing techniques to discuss the results of the assessment, areas of strength, areas for improvement, and recommended services that might help the family. After the feedback meetings, families could choose to participate in additional follow-up meetings. Families assigned to the intervention group received the intervention once yearly when their children were 2, 3, and 4 years old.
Comparison conditions Families in the comparison group received the Family Check-Up's comprehensive assessment but did not receive any other interventions or services.
Staff characteristics and training Parent consultants for this study had either a master’s or doctorate degree; had previous experience in conducting family-based interventions; and were of diverse ethnicities, including Latino, African American, European American, and mixed ethnicity. Consultants were trained for two-and-a-half to three months in strategies that included didactic instruction and role-playing as well as ongoing videotaped supervision of intervention activity. Consultants were certified by lead parent consultants at each site; the lead consultants were certified by a member of the research team. Certification was repeated annually and was established by reviewing videotapes of feedback and follow-up sessions. Weekly conference calls were held to discuss cases, and annual consultant meetings were held to update training, discuss possible changes in the intervention, and address issues related to families’ needs.
Funding sources Not reported.
Author affiliation The authors are associated with the University of Pittsburgh, Case Western Reserve University, the University of Oregon, the University of Virginia, and Oxford University. In addition, Thomas Dishion, a study author, is a developer of this model.
Study Registration:
Clinicaltrials.gov Identifier: None found. Study registration was assessed by HomVEE beginning with the 2014 review.

Findings details

Child development and school readiness
Rating Outcome measure Effect Sample Timing of follow-up Sample size Intervention group Comparison group Group difference Effect size Statistical significance Notes
High Child Behavior Checklist (CBCL) Externalizing, Age 3, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 651 children Not reported Not reported Not Reported Not available Not statistically significant, p ≥ 0.05
High Child Behavior Checklist (CBCL) Externalizing, Age 4, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 4 619 children Not reported Not reported Not Reported Not available Statistically significant, p<0.05
High Child Behavior Checklist (CBCL) Internalizing, Age 3, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 651 children Not reported Not reported Not Reported Not available Not statistically significant, p ≥ 0.05
High Child Behavior Checklist (CBCL) Internalizing, Age 4, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 4 619 children Not reported Not reported Not Reported Not available Statistically significant, p<0.05
High Eyeberg Child Behavior Inventory Problem Score, Age 3, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 642 children Not reported Not reported Not Reported Not available Not statistically significant, p ≥ 0.05
High Eyeberg Child Behavior Inventory Problem Score, Age 4, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 4 616 children Not reported Not reported Not Reported Not available Statistically significant, p<0.05
High Growth in Child Behavior Checklist (CBCL) Externalizing from ages 2 to 4, Latent growth model
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Ages 3 and 4 731 children Not reported Not reported Not Reported Study reported = 0.23 Statistically significant, p<0.051

footnote270

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used a latent growth model to estimate the impact, and reported a coefficient, standard error, and effect size.

High Growth in Child Behavior Checklist (CBCL) Internalizing from ages 2 to 4, Latent growth model
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Ages 3 and 4 731 children Not reported Not reported Not Reported Study reported = 0.21 Statistically significant, p<0.051

footnote270

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used a latent growth model to estimate the impact, and reported a coefficient, standard error, and effect size.

High Growth in Eyeberg Child Behavior Inventory Problem Score from ages 2 to 4, Latent growth model
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Ages 3 and 4 731 mothers Not reported Not reported Not Reported Study reported = 0.23 Statistically significant, p<0.051

footnote270

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used a latent growth model to estimate the impact, and reported a coefficient, standard error, and effect size.

Maternal health
Rating Outcome measure Effect Sample Timing of follow-up Sample size Intervention group Comparison group Group difference Effect size Statistical significance Notes
High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 2
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 3
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 4
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 5
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 6
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 7
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.052

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, Autoregressive model
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Study reported = 0.18 Statistically significant, p<0.051

footnote271

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used an autoregressive model to estimate the impact, and reported a coefficient, standard error, and effect size.

High Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, Correlation
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 651 mothers Not reported Not reported Not Reported Not available Not statistically significant, p ≥ 0.05
Positive parenting practices
Rating Outcome measure Effect Sample Timing of follow-up Sample size Intervention group Comparison group Group difference Effect size Statistical significance Notes
High Positive Parenting, Age 3, SEM, Figure 5
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.051

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Positive Parenting, Age 3, SEM, Figure 6
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.051

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

High Positive Parenting, Age 3, SEM, Figure 7
FavorableUnfavorable or ambiguousNo Effect
WIC sites in Pittsburgh, PA, Eugene, OR, and Charlottesville, VA Age 3 731 mothers Not reported Not reported Not Reported Not available Statistically significant, p<0.051

footnote272

Submitted by user on Fri, 03/15/2019 - 14:29

Authors used structural equation models to estimate the impact, and reported a coefficient, standard error, and p-value.

Outcome measure summary

Child development and school readiness
Outcome measure Description of measure Data collection method Properties of measure

Child Behavior Checklist (CBCL) Externalizing, Age 3, Correlation

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.86 at age 2, Cronbach’s a = 0.89 at age 3, Cronbach’s a = 0.86 at age 4

Child Behavior Checklist (CBCL) Externalizing, Age 4, Correlation

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.86 at age 2, Cronbach’s a = 0.89 at age 3, Cronbach’s a = 0.86 at age 4

Child Behavior Checklist (CBCL) Internalizing, Age 3, Correlation

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.82 at age 2, Cronbach’s a = 0.86 at age 3, Cronbach’s a = 0.91 at age 4

Child Behavior Checklist (CBCL) Internalizing, Age 4, Correlation

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.82 at age 2, Cronbach’s a = 0.86 at age 3, Cronbach’s a = 0.91 at age 4

Eyeberg Child Behavior Inventory Problem Score, Age 3, Correlation

The Eyeberg Child Behavior Inventory is a 36-item behavior checklist that assesses two factors: (1) the perceived intensity, and (2) degree a behavior is a problem for caregivers. The researchers focused on the problem factor, which asks caregivers to report on the extent the behavior is a problem for the parent using a 7-point scale. Parent/caregiver report

Internal consistency = 0.84 at age 2, 0.90 at age 3, and 0.94 at age 4

Eyeberg Child Behavior Inventory Problem Score, Age 4, Correlation

The Eyeberg Child Behavior Inventory is a 36-item behavior checklist that assesses two factors: (1) the perceived intensity, and (2) degree a behavior is a problem for caregivers. The researchers focused on the problem factor, which asks caregivers to report on the extent the behavior is a problem for the parent using a 7-point scale. Parent/caregiver report

Internal consistency = 0.84 at age 2, 0.90 at age 3, and 0.94 at age 4

Growth in Child Behavior Checklist (CBCL) Externalizing from ages 2 to 4, Latent growth model

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.86 at age 2, Cronbach’s a = 0.89 at age 3, Cronbach’s a = 0.86 at age 4

Growth in Child Behavior Checklist (CBCL) Internalizing from ages 2 to 4, Latent growth model

The CBCL for ages 1.5–5 is a 99-item assessment of behavioral problems in young children. The CBCL includes two broadband factors—internalizing and externalizing—that were used to evaluate the frequency of problem behavior during the study period. Parent/caregiver report

Cronbach’s a = 0.82 at age 2, Cronbach’s a = 0.86 at age 3, Cronbach’s a = 0.91 at age 4

Growth in Eyeberg Child Behavior Inventory Problem Score from ages 2 to 4, Latent growth model

The Eyeberg Child Behavior Inventory is a 36-item behavior checklist that assesses two factors: (1) the perceived intensity, and (2) degree a behavior is a problem for caregivers. The researchers focused on the problem factor, which asks caregivers to report on the extent the behavior is a problem for the parent using a 7-point scale. Parent/caregiver report

Internal consistency = 0.84 at age 2, 0.90 at age 3, and 0.94 at age 4

Maternal health
Outcome measure Description of measure Data collection method Properties of measure

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 2

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 3

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 4

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 5

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 6

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, SEM, Figure 7

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, Autoregressive model

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Maternal Depression (Center for Epidemiological Studies Depression Scale), Age 3, Correlation

The Center for Epidemiological Studies-Depression Scale is a 20-item assessment of depressive symptoms. Maternal report

Cronbach’s a = 0.76 at age 2 years, Cronbach’s a = 0.75 at age 3 years

Positive parenting practices
Outcome measure Description of measure Data collection method Properties of measure

Positive Parenting, Age 3, SEM, Figure 5

A combination of four measures (parent involvement, caregiver prompting and reinforcing positive behavior, engaged parent child interactions, proactive parenting) of how well the caregiver supported the child's positive behaviors. In-person ratings and coding from videotaped interactions

Cronbach’s a = 0.61

Positive Parenting, Age 3, SEM, Figure 6

A combination of four measures (parent involvement, caregiver prompting and reinforcing positive behavior, engaged parent child interactions, proactive parenting) of how well the caregiver supported the child's positive behaviors. In-person ratings and coding from videotaped interactions

Cronbach’s a = 0.61

Positive Parenting, Age 3, SEM, Figure 7

A combination of four measures (parent involvement, caregiver prompting and reinforcing positive behavior, engaged parent child interactions, proactive parenting) of how well the caregiver supported the child's positive behaviors. In-person ratings and coding from videotaped interactions

Cronbach’s a = 0.61