By Chad Shenk, Ph.D.
Child maltreatment, which includes acts of child sexual abuse, physical abuse, and neglect, affects 12.5% of all children under the age of eighteen in the United States.1 In 2014, the most recent year in which national data are available, there were over 700,000 new cases of child maltreatment.2 The impact of child maltreatment, as measured in child welfare and public health costs, translates into a life-long economic burden of $124 billion.3
One reason child maltreatment exerts such a large impact is because it increases the risk for a host of adverse health outcomes throughout subsequent development.4 However, variation in the significance and magnitude of effect size estimates reported across independent prospective studies, often considered the ideal methodology for generating causal inferences about the effects of child maltreatment,5 has led to a controversy about whether, and to what extent, child maltreatment truly affects different health outcomes.6 Identifying the sources of such variation is important because failing to do so increases the probability of Type II errors made by researchers examining the long-term effects of child maltreatment as well as healthcare providers faced with the decision to consider a patient’s history of child maltreatment when treating a particular health outcome.
To help address this issue, my colleagues and I led an investigation to determine the prevalence of child maltreatment in a comparison condition, a phenomenon known as contamination,7 and whether such contamination produces variation in the significance and magnitude of effect size estimates across distinct subsequent health outcomes in late adolescence: obesity, teenage births, major depression, and past month cigarette use.8 Results showed that demographic matching, a method often used to generate comparison groups in prospective research, produced a comparison condition where 44.8% of participants experienced at least one instance of child maltreatment. When this contamination was left uncontrolled, child maltreatment predicted only two of the four outcomes assessed, teenage births and past month cigarette use. Thus, exerting no control over contamination could lead to the conclusion that child maltreatment had no effect on the risk for subsequent obesity or major depression. Once contamination was controlled however, by using a novel, multi-informant strategy that screened the comparison condition for the presence of child maltreatment using both self-report and substantiated case records, effect size estimates increased dramatically with child maltreatment now significantly predicting all four outcomes. Controlling contamination in this manner also yielded a comparison condition that more closely reflected national prevalence estimates for each of the outcomes examined.
This study supports a viable method for identifying and controlling contamination, a potential source of variation in the significance and magnitude of effect size estimates reported across prospective studies. This can help reduce Type II errors and increase the likelihood of successful replication of findings in future research on the effects of child maltreatment. Replication studies examining the prevalence and impact of contamination in national samples of child maltreatment are currently underway in our lab and stand to make a sustained contribution to the adoption of experimental and statistical methods needed to generate more accurate and reliable estimates of the effect of child maltreatment. Ultimately, this line of research can strengthen causal inferences about the long-term impact of child maltreatment.
While continued research on the effects of contamination is clearly needed, there are a few helpful suggestions based on this study that can enhance the planning of future studies. First, the prevalence of child maltreatment varies considerably depending on the method of assessment, such as self-report or substantiated cases records. Each method has their own limitations and advantages, which is why a multi-informant strategy should be used to screen for contamination in comparison conditions at study entry and during each follow-up assessment. Second, the optimal experimental and statistical methods for controlling contamination is an open area of research. Both are likely needed to achieve the most reliable and accurate estimates and devising new methods for controlling contamination will greatly facilitate this line of research.
If accurate, the results from this study suggest that the effects of child maltreatment on subsequent health outcomes could be worse than initially thought, especially if rates of contamination are high. Knowing the accurate effect of child maltreatment will facilitate public policy and intervention efforts earlier and more confidently, helping a significant proportion of the pediatric population in the United States achieve the care and help they need for a variety of public health concerns.
Read the article:
Shenk, C.E., Noll, J.G., Peugh, J.L. Griffin, A.M. & Bensman, H.E. (2016). Contamination in the prospective study of child maltreatment and female adolescent health. Journal of Pediatric Psychology, 41, 37-45. DOI: 10.1093/jpepsy/jsv017
About the Author
Chad Shenk, Ph.D. is Assistant Professor of Human Development and Child Abuse Pediatrics and a member of the Child Maltreatment Solutions Network at The Pennsylvania State University. He is Principal Investigator and Co-Investigator on several institutionally and federally-funded grants examining the impact of child maltreatment on subsequent health outcomes as well as the pathways linking child maltreatment to the onset of various forms of psychopathology. This latter work focuses on the identification of mechanisms for psychological conditions across multiple levels of analysis (e.g. psychological, biological, familial) to translate basic science findings into the clinic context. As a licensed clinical psychologist working at Penn State’s Transforming the Lives of Children Clinic, Dr. Shenk’s research also involves the development of prevention programs and optimization of existing clinical interventions that target putative risk and protective mechanisms more directly and effectively for those affected by child maltreatment. http://www.hhd.psu.edu/hdfs/directory/Bio.aspx?id=Shenk
- Wildeman C, Emanuel N, Leventhal JM, Putnam-Hornstein E, Waldfogel J, Lee H. The prevalence of confirmed maltreatment among US children, 2004 to 2011. JAMA Pediatrics. 2014;168(8):706-713.
- U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth, and Families, Children's Bureau. Child Maltreatment 2014 2016.
- Fang X, Brown DS, Florence CS, Mercy JA. The economic burden of child maltreatment in the United States and implications for prevention. Child Abuse & Neglect. 2012.
- Felitti VJ, Anda RF, Nordenberg D, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. American Journal of Preventative Medicine. 1998;14(4):245-258.
- Widom CS, Raphael KG, DuMont KA. The case for prospective longitudinal studies in child maltreatment research: Commentary on Dube, Williamson, Thompson, Felitti, and Anda (2004). Child Abuse & Neglect. 2004;28(7):715-722.
- Gilbert R, Widom CS, Browne K, Fergusson D, Webb E, Janson S. Burden and consequences of child maltreatment in high-income countries. Lancet. 2009;373(9657):68-81.
- Cuzick J, Edwards R, Segnan N. Adjusting for non-compliance and contamination in randomized clinical trials. Statistics in Medicine. 1997;16(9):1017-1029.
- Shenk CE, Noll JG, Peugh JL, Griffin AM, Bensman HE. Contamination in the prospective study of child maltreatment and female adolescent health. Journal of pediatric psychology. 2016;41:37-45.