The Society for Social Work and Research

2014 Annual Conference

January 15-19, 2014 I Grand Hyatt San Antonio I San Antonio, TX

Using the Ncands Data to Test Three Models of Fatal Child Maltreatment

Thursday, January 16, 2014: 2:30 PM
HBG Convention Center, Room 001B River Level (San Antonio, TX)
* noted as presenting author
Emily M. Douglas, PhD, Associate Professor, Bridgewater State University, Bridgewater, MA
Background and Purpose:

About 2,000 U.S. children die annually from maltreatment. Our nation’s policies and programs have increasingly paid attention to maltreatment fatalities but we know little about which child/family factors increase risk and if models which predict non-fatal maltreatment are relevant to fatal maltreatment: (1) parent psychopathology (Gelles, 1979), (2) social stress (Belsky, 1980), and child behavior (Chance & Scannapieco, 2002). The current study tests these three models of fatal maltreatment. This is the first study to examine child fatalities using a national data set and the first study to test models to predict fatal maltreatment.


This study uses the 2010 National Child Abuse & Neglect Dataset to examine fatal child maltreatment. This dataset captures information from all children and families who encountered the child welfare system in 2010. The final sample size for the current analyses is 1,446,712 cases, with 746 fatalities. The dependent variable is whether the child died from maltreatment. The main predictor variables are whether the caretaker has any mental health problem, whether the caretaker experiences any social stress, and whether the child is considered to have a behavioral/emotional problem. Control variables included the child age, child sex, child race/ethnicity, and child living arrangement. Analyses were conducted with Stata, using the add-on command for rare-event logistic regression (relogit).


The results showed support for one of the three models tested: social stress model. Children who comes from homes where social stress is present were 1.35 times more likely to suffer fatal maltreatment (p<.000). The findings for the other two models were not significant. Among the control variables, younger children are more at risk for fatality (OR=.76, p<.000), as were males (OR=1.46, p<.000), African American children (OR=1.56, p<.000) and Asian American children (OR=2.3, p=.003). Children who lived with a single parent (OR=.59, p<.000), with a single parent and another adult (OR=.59, p<.000), or in a residential setting (OR=.53, p=.046) were less likely to suffer a maltreatment fatality than children in other living arrangements – namely, with both of their parents.  Children who were the alleged victim of physical abuse (OR=1.93, p<.000) were more at risk for fatality than children who were the alleged victim of neglect (OR=.71, p=.001).

Conclusion and Implications:

The findings provide support for the social stress model of child maltreatment, which is the primary model used today to predict and service/treat families experiencing maltreatment. There was no support for the other models. The analyses provide additional information which has never been assessed before on a large-scale dataset: children living with both parents may be at higher-risk for maltreatment fatality and children who were initially reported for physical abuse, as opposed to neglect, are at a higher-risk for fatality. Other findings are consistent with existing literature –younger children and some racial minorities are at increased risk for fatal maltreatment. The findings provide new information for those involved in all levels of child welfare policy and practice, from screening to investigations, regarding risk factors for maltreatment fatality.