American Journal of Public Health, Vol. 75, No. 1, January 1985, 56-60.

Child Abuse Incidence and Reporting by Hospitals:
Significance of Severity, Class, and Race


Abstract: Estimates from the National Study of the Incidence and Severity of Child Abuse and Neglect suggest that hospitals recognized over 77,000 cases of child abuse between May 3, 1979, and April 30, 1980. Compared to other agencies in the sample, hospitals identified children who were younger, Black, lived in urban areas, and had more serious injuries. Hospitals failed to report to child protection agencies almost half of the cases that met the study’s definition of abuse. Discriminant analysis revealed that income, mother’s role in abuse, emotional abuse, race, maternal employment, and sexual abuse distinguished the reported from the unreported cases. Disproportionate numbers of unreported cases were victims of emotional abuse and carne from families of higher income. Their mothers were more often White and more often alleged to be responsible for the injuries. (Am J Public Health 1985; 75:56-60.)

* For ease of presentation, the term “abuse” will be used in this paper to refer to the full range of child maltreatment cases, including physical abuse, sexual abuse, emotional abuse, physical neglect, and emotional neglect. The specific forms of maltreatment are noted when appropriate in the text.


Child abuse*, as reported to state welfare agencies, has become more visible over the past decade. In 1981, according to the American Humane Association, over 850,000 reports were received,1 more than a tenfold increase in the course of a decade. Estimates of inflicted injuries to children, based not on case reports but on household surveys, range from one to four million per year.2 The true incidence of child abuse is unknown.

Hospitals and medical personnel have played important roles in the diagnosis, reporting, and treatment of child abuse and neglect. This paper examines the variables associated with the initial diagnosis of child abuse by hospitals and with the proportion of cases subsequently reported to child protective service agencies.


Data for this analysis were drawn from the National Study of the Incidence and Severity of Child Abuse and Neglect (NIS). The methodology and overall research design of the NIS have been described elsewhere.3,4 The study was conducted in a stratified random sample of 26 counties in Arizona, California, Georgia, Illinois, Kansas, Missouri, New Hampshire, New York, Ohio, and South Carolina. Data were collected on case reports received between May I, 1979 and April 30, 1980.

Overall the sample contained to urban, six suburban, and 10 rural counties. The counties were selected from a sampling frame which included all counties within the 48 coterminous states stratified by population size and distribution. Each county had a known probability of selection, which provided the basis for computing national estimates. 

The estimation weight for hospital cases was the product of three components: the “county selection weight”, the “hospital selection weight,” and an “annualization weight” (see Appendix).

In most of the sample counties, all short-stay general hospitals that contained an emergency room and/or a pediatric department (unit or ward) were included in the study design and were asked to take part in the study. Seventy of the 92 eligible hospitals (76 per cent) agree to participate.4 To protect confidentiality, no effort was made to distinguish among types of hospitals.

Within hospitals, those who reported directly to the study included all professional staff members in emergency rooms, pediatric departments, pediatric outpatient departments, and social service departments; as well as any other professional who might be attached to a child abuse team or program.

Hospital personnel could report cases either on a data form or via a WATS line. In all instances, the reporter provided an account of what happened and some basic nonidentifying information about the child and the family involved in the abuse. To ensure maximum participation, the study staff protected the identities of participating agencies and professional staff and of the families and children involved. Reporting an incident to the study did not constitute an official report of abuse or neglect; the data forms were not made available to the local Child Protective Service (CPS) unit or to anyone else at the local, state, or federal level. In addition, all participants were assured that the published findings of the study would not detail results by county or state, but only on the national level.

Hospital personnel were given concrete guidelines describing the kinds of child abuse situations which were of interest to the study and were asked to provide certain narrative and demographic information concerning the child. Based on the reporter’s narrative information and using detailed specifications, a second assessment was made by the study staff and the final categorization of abuse with respect to type and severity was made. In independent assessment of a random sample of 300 children, inter-rater agreement in the classification of the suspected form of maltreatment was 99 per cent.

For the purpose of this study a child abuse situation was defined as: “one where, through purposive acts or marked inattention to a child’s basic needs, behavior of a parent/substitute or other adult caretaker caused foreseeable and avoidable injury or impairment to a child, or materially contributed to unreasonable prolongation or worsening of an existing injury or impairment.”3

In implementing the general study definition, seven additional specific criteria were established. All of these had to be met for a child to be considered within the scope of the study (see Appendix).

The study staff also received information about cases reported to CPS. Any particular child may have been reported to CPS more than once during the study year and/or may have been described to the study by any number or combination of sources. Enough identifying data were collected, however, to permit a reliable determination of whether or not any two data forms were describing the same child. Duplicate records were purged from the analysis file so that the case was counted only once. Whenever a particular child was identified to the study by a hospital, or when another non-CPS source also appeared in the CPS file, the CPS record was retained. In effect, non-CPS sources were given credit only for children who had not been reported to CPS.


Nationally, about 652,000 children are estimated to have met the operational definitions of abuse and neglect during the study year; of this number, 212,400 would have been known to the local child protective service agencies.3

Compared to other agencies in the sample, hospitals identified children who were younger, who had younger, parents who were black (25 per cent vs 16 per cent), and who lived in urban areas (65.8 per cent vs 42.1 per cent). There were no major differences between cases reported by hospitals and those reported by other agencies with respect to income, mode of medical payment (public or private), proportion of single-parent families, the child’s sex, or other demographic variables.

Hospitals identified many more cases of physical abuse than did other agencies. The proportion of cases in this category alone exceeded the proportion of physical, sexual, and emotional abuse cases recognized by all other agencies; over half the hospital cases fell into some category of abuse.

In the sampled counties, 805 cases of child abuse and neglect came to the attention of hospital staff members during the study year. Black and Latino families were more likely to be reported than white families (Table 1).

Families with incomes in the two lowest categories had the highest reporting rates, while among families with annual incomes of $25,000 or more, recognized child abuse was reported in only 36 per cent of the cases.

Children five years of age and younger constituted 55 per cent of our sample; 72 per cent of these cases were reported. Adolescents were less likely to be reported.

Serious injuries accounted for only 28 per cent of the reported cases from all sources and 40 per cent of all unreported cases. Although hospitals identify more serious cases than other agencies, they report less than half of them to CPS.

A case report was more likely to be filed by hospitals when the mother was not the alleged perpetrator than where she was the alleged perpetrator.

Physical abuse cases were more likely to be reported while cases of emotional abuse and neglect tended to be underreported.

Multivariate Analyses

A stepwise discriminant function procedure was used to select variables for inclusion in a multivariate analysis; the variables in table 2 appear in the order of their discriminating power. In our analysis, the following four variables seemed to have the strongest effect on hospital case reporting: type of abuse, household income, the mother’s role in abuse, and race. Disproportionate numbers of unreported cases were victims of emotional abuse, in White families, of higher income, and whose mothers were alleged to be responsible for the injuries.


The national study from which this sample was drawn was a unique effort to identify cases before reports were made and to ascertain the differences between cases which were reported and those which were not.

The findings suggest that for the hospital sample, social class and race are the most important perpetrator characteristics that distinguish reported from unreported cases of abuse. Degree of severity becomes statistically significant only when income is excluded from the analyses.5

Although hospitals reported cases within the scope of the study definitions more frequently than did other agencies, they failed to report large numbers of the cases that should have been reported. We believe this is attributable, at least in part, to the fact that child abuse is neither theoretically nor clinically well defined, even though it has become part of the medical taxonomy.6 Thus, there is much subjective judgment in professional evaluation.

Social distance is inherent in the provider-patient relationship: the provider (physician, nurse, or other professional) is always in a superordinate position because of his or her expertise.7 The label “child abuser” may be less likely to be applied if the diagnostician and suspected abuser share similar characteristics, especially socioeconomic status, particularly when the injury is not serious or a manifest result of abuse.8 Although professional practice claims to be valuefree, personal prejudices and judgments may affect the typing of individuals who are defined as deviant.

Previous research suggests that physicians’ judgments of possible child abuse are affected by socioeconomic and ethnic status. Socially marginal families may be victimized by a process in which their personal characteristics, rather than their behavior, define them as deviant.7 With respect to child abuse, a physician’s stereotype of an “abuser” may determine which parents of an injured child will be seen as possible abusers. A latent consequence of this biased reporting may be a failure to address the needs of the many children in middle-class families who are at risk of abuse.2

It has been proposed that the CPS classification of cases as “substantiated” or “not substantiated” assesses the occurrence of abuse more accurately than the suspicion of someone reporting the case to the system.9 This proposition rests on the assumption that the reporting process reflects a “screening” and the CPS investigation represents a “determinant test.” Our data suggest that biases at the point of recognition and reporting are important in determining which cases are channeled into the protective services system for investigation. Case substantiation, therefore, may not merely define whether the case passes a certain threshold of validity, but may also be affected by the reporting process. Unreported cases, by definition, cannot be substantiated by CPS. 

The status of the reporter may serve to intensify such a bias even after reporting. Carr and Gelles, in a review of case reports to the Florida central registry, found that physicians’ and hospitals’ reports were far more likely to be admitted as valid than reports of others regardless of the clinical severity of the cases.10

There remain important questions that these data do not answer: Are the findings reported here similar across types of hospitals? Are there reporting differences associated with professional status among hospital personnel? To what extent is the reporting decision associated with other efforts to obtain services for families, and how this is perceived by patients and professionals? What other factors may influence decision making by clinicians in cases of child abuse and neglect?

If the reporting of child abuse is as biased by class and race as these data suggest, there is a clear need for a critical review of the system as well as the process of reporting. To the extent that we selectively invoke agents of the state to police the lives of poor and non-White families, we may be inappropriately and unfairly condemning these families as evil. In selectively ignoring the prevalence of child abuse in more affluent, majority homes, we may be perpetuating a myth that child abusers are out there, and that homes like ours are free of violence.11

Table 1 – Per Cent of Reported and Unreported Abuse Casesby Characteristics of Subject, Family, and Injury
Attributes (%) in Sample Reported Not Reported
White (66.7) 60.5 39.5
Black (25.7) 74.3 25.7
Latino (7.6) 91.2 8.8
Less than $7000 (49.3) 77.7 22.3
7000-14,900 (36.8) 80.0 20.0
15,000-24,999 (9.1) 70.2 29.8
25,000+ (4.8) 36.7 63.3
Severity of Injury Impairment      
Fatal (.3) 100.0
Serious (33.2) 46.8 53.2
Moderate (35.7) 55.2 47.8
Probably (30.8) 63.9 36.1
Child’s Age (years)      
0-5 (54.9) 72.0 28.0
6-12 (8.9) 70.8 29.2
13-17 (16.0) 42.5 57.5
Sex of Child      
Male (51.6) 65.3 34.7
Female (48.3) 68.0 32.0
Role of Mother      
Maltreator (49.6) 48.7 51.3
Permitted (8.3) 71.3 28.7
Not Involved (34.5) 93.5 6.5
Don’t Know (7.5) 56.6 43.4
Father in Household      
Biological Father (43.4) 64.2 35.8
Father Substitute (12.7) 74.4 25.6
No Father in House (44.1) 66.9 33.1
Mother’s Age (years)      
19 or less (10.8) 70.9% 29.2%
20-24 (21.4) 83.0 17.0
25-29 (15.5) 76.9 23.1
30-35 (18.1) 80.1 19.9
36+ (14.1) 71.0 29.0
Don’t Know (20.0) 28.5 71.5
Physical Abuse (35.5) 75.6 24.4
Sexual Abuse (9.4) 80.8 19.2
Emotional Abuse (5.7) 36.1 63.9
Physical Neglect (26.3) 65.8 34.2
Educational Neglect (1.4) 5.5 94.5
Emotional Neglect (13.7) 42.6 57.4
Miscellaneous Maltreatment (7.8) 85.4 14.6
SMSA Over 200,000 (65.8) 68.2 31.8
Other SMSA (18.7) 70.5 29.5
Non SMSA (15.5) 55.3 44.7
TOTAL   66.6 33.4
Unweighted N = 805Weighted N= 77379


Table 2 – Standardized Canonical Discriminant Function Coefficientsand Summary Statistics for Hospital Reporting
    Reported Cases Unreported Cases
Variable Coefficient Mean S.D. Mean S.D.
Emotional abuse .65 .04 .21 .38 .48
Role of Mother -.45 .42 .49 .83 .37
Family Income .58 1.59 .76 2.08 1.07
Race -.30 .57 .49 .85 .35
Emotional Neglect .22 .10 .26 .18 .39
Mother Employed .26 .23 .49 .24 .51
Sexual Abuse .24 .08 .23 .08 .27
Number of Victims -.22 1.66 1.13 1.64 1.02
Mother’s Education -.20 2.25 .75 2.29 .83
Urbanicity -.15 1.52 .72 1.65 .82
Sex of Child -.13 .48 .49 .48 .49
Physical Abuse .10 .36 .42 .18 .48
Canonical R=.602
When emotional abuse is excluded from the analysis, Mother’s Role in Malreatment and Family Income remain the most important discriminators. The relative contributions of Physical Abuse, Physical Neglect, and Ethnicity also increase.


1. American Humane Association: Annual Report. Highlights of official child neglect and abuse reporting. Denver: American Humane Association, 1981.

2. Straus MA, Gelles RJ, Steinmetz, SK: Behind Closed Doors: Violence in the American Family. New York: Doubleday, 1980.

3. US Department of Health and Human Services: Study Findings: National study of the incidence and severity of child abuse and neglect. DHHS Pub. No. (OHDS) 81-30325. Washington, DC: DHHS, 1981.

4. US Department of Health and Human Services: Study Methodology: National study of the incidence and severity of child abuse and neglect. DHHS Pub. No. (OHDS) 81-30326. Washington, DC: DHHS, 1981.

5. Hampton RL, Newberger EH: Hospitals as gatekeepers: recognition and reporting in the national incidence study of child abuse and neglect. Report to National Center on Child Abuse and Neglect. 1983.

6. Newberger EH, Bourne R: The medicalization and legalization of child abuse. Am J Orthopsy 1978: 48:4,593-607.

7. O’Toole R, Turbett P, Nalepka C: Theories, professional knowledge, and diagnosis of child abuse. In: Finkhelhor D, Gelles RJ. Hotaling GT, Straus MA: The Dark Side of Families: Current Family Violence Research. Beverly Hills: Sage Publications, 1983; 349-362.

8. Pfohl S: The discovery of child abuse. Soc Prob 1977; 24:310-323.

9. Jason J, Andereck ND, Marks J, Tyler CW Jr: Child abuse in Georgia: A method to evaluate risk factors and reporting bias. Am J Public Health 1982; 72:1353-1358.

10. Carr A, Gelles RJ: Reporting child maltreatment in Florida: the operation of public child protective service systems. Report submitted to the National Center on Child Abuse and Neglect. 1978.

11. Gelles RJ: The social construction of child abuse. Am J Orthopsy 1975; 45:364-371.


This research was supported by a grant from the National Center on Child Abuse and Neglect (9OCA891). Department of Health and Human Services. Washington, DC. Dr. Hampton served as Rockefeller Foundation Postdoctoral Fellow during the period in which this research was conducted.


I. Study Definitions

In implementing the general study definition. seven more specific criteria were established, all of which had to be met for a child to be considered within the scope of the study. These criteria included:

a. Age of child: live-born and under 18 years of age at the time of the harm-causing acts or omissions.

b. Residence of child: lived in one of the 26 study counties at some time during the period May 1. 1979 to April 30, 1980.

c. Custody of child: child was a non-institutionalized dependent of parent(s)/substitute(s) at the time of the harm-causing acts or omissions.

d. Time of occurrence: the harm-causing acts or omissions occurred during the (four or twelve-month) study period which applied for the agency and county.

e. Person responsible for maltreatment: a parent/substitute or other adult caretaker: assault, exploitation, etc., by other persons (siblings. etc.) or developmentally maladaptive behavior by the child was in scope only if knowingly permitted by a parent/substitute.

f. Nature of maltreatment acts or omissions: The occurrence or prolongation of injury/impairment was a foreseeable consequence of purposive (non-accidental) acts or of marked inattention to the child’s basic needs.

g. Effect of maltreatment: there must have been reasonable cause to believe that maltreatment acts/ omissions during the study period caused or materially contributed to the occurrence or unreasonable prolongation of some actual injury or impairment; depending upon the form of maltreatment. the injury/impairment must have been of moderate or serious severity, at minimum.3,4