Accurate Ascertainment of Child-Abuse Mortality

To the Editor: The article by Dr Herman-Giddens and colleagues¹ contributes to the literature on the scope of the child abuse problem by providing an estimate of underreporting in vital statistics. Similar studies² provide estimates of reliability for other causes of death.

Two reasons account for differences in estimates of child abuse deaths between the vital statistics system and those of the authors' comprehensive case fatality review: (1) the availability of information on circumstances of the death and (2) definitional differences. Herman-Giddens et al reviewed a broad range of information from which child abuse could be inferred, including medical examiners' reports, autopsy reports, toxicology reports, and case notes. In contrast, ascertainment of child abuse in vital statistics depends entirely on the cause of death reported by the medical examiner, coroner, or attending physician. This is the only information available to the mortality medical coder to make judgments or inferences about the circumstances of death.

Differences in definition are also critical. The authors propose a definition of child abuse that depends on information about the caregiver who was responsible for the child's health or welfare. Such a definition does not lend itself to vital statistics, because information about the caregiver is rarely reported on death certificates. In 1997, such information was reported on death certificates for only 1 of 10 deaths classified as due to child abuse.³

The authors question the vital statistics coding practice that, in the absence of the terms "abuse" or "maltreatment," relies on terms consistent with prior or repeated abuse, beating, or other maltreatment. These coding procedures were established by the National Center for Health Statistics in the 1980s in consultation with child abuse experts from the US Department of Health, Education, and Welfare to address the absence of uniform national coding guidelines for child abuse. Prior to those rules, each state coded child abuse according to its own criteria, resulting in statistics that could not be compared among states.

Opportunities exist to bring vital statistics data into closer alignment with the case review estimates through educating medical examiners about certifying these deaths.⁴ However, it is unrealistic to expect a broad-based data system such as vital statistics to provide the same estimates as an intensive case review system.⁵ The strengths of the vital statistics system are in providing timely, consistent, and uniform information on cause of death to illuminate geographic differences, trends over time, and sociodemographic differences in risk of death. Studies such as those by Herman-Giddens et al are a useful complement to mortality data from vital statistics.