Objective. To test how prevalence estimates and characteristics of children vary by the way that disability is defined. Specifically, to determine 1) the proportions of children identified as disabled by one particular operationalization of disability based on parental reports of three types of consequences (ie, functional limitations [FL]), dependence on compensatory mechanisms (CD), and service use or need beyond routine [SU/N]), and 2) whether children identified as disabled by these three types of consequences differ by type of disorder or condition, age, socioeconomic status, or race.
Method. We analyzed a national dataset representing a random sample of 712 households with 1388 children. The Questionnaire for Identifying Children with Chronic Conditions (QuICCC) was used to identify children with disabling conditions. We divided the QuICCC items into three discrete sets, reflecting three definitional components of disability, and compared the proportions and characteristics of children fitting these components separately and in combination.
Results. Using the QuICCC definition of disability, SU/N identified the largest proportion of children (72%), followed by CD (55%) and FLs (49%). Forty-four percent of children were identified by only one component, 36% by two components in any combination, and 20% by all three components. The type of disorder or condition generally did not vary by the three definitional components, although the FL component may be more effective at identifying children with sensory impairments. Children identified by two or more components were more likely to have multiple conditions and had more pervasive disorders than those identified by only one component. The youngest children (0 to 3 years old) may be less likely to be identified as disabled than children of other ages, especially by FLs. FLs also were more likely to identify children from the poorest and least educated families.
Conclusions. Although the specific findings reported here pertain to a single definitional approach (the QuICCC), the data highlight that who will be classified as disabled (and who will not) may be dependent on how disability is defined. The implications of using different definitions and definitional components on both the prevalence and the characteristics of children with disabilities need to be considered before data can be applied responsibly and appropriately.
Significant budget cuts to Medicaid, widespread implementation of managed care programs, and other changes in the health care system have recently intensified the need to plan for services and benefits to children with chronic and disabling conditions. One of the major obstacles to this endeavor has revolved around the issue of how to best identify disability in children. The well-established ways to assess disability in adults (ie, assessment of the ability to work and to perform household chores and other activities of independent daily living) are not appropriate criteria for children. The ability to play, learn, and grow are concepts that do not lend themselves easily to operationalization.
Despite the inherent difficulties, there is no dearth of definitions. At the federal level alone, more than 40 different definitions of disability have been identified1 and the range is enormous. For example, some definitions serve specific programmatic purposes, whereas others were designed for planning and systems development or research. Some definitions only capture children with the most severe disabilities, whereas others identify disabled children within a wide range of severity levels. Disability in children tends to be defined narrowly. Common components include 1) having particular service needs; 2) carrying a diagnosis of a specific physical or mental condition; and 3) exhibiting specified functional deficits.
Given the diverse methods to determine disability status in children, it is not surprising that little is understood about the implications of using different definitions of disability. Knowing which disabled children may be more or less likely to be identified with any one definition will be critical to the health care planning process and ultimately to services and programs for these children. The purpose of our study was to address some of the consequences and health policy implications of applying different definitions by examining the ways in which the components of one particular operationalization of disability identify children.
In earlier work, we developed a noncategorical definition of serious ongoing health conditions2 that included disabilities. We conceptualized disabilities in the broadest sense and followed the language and perspective of the Americans with Disabilities Act (ADA) of 1990 (PL 101-336). The ADA states that disability refers to ongoing conditions (whether physical, cognitive, behavioral, or psychological) that currently have functional consequences, but also includes conditions and illnesses that do not currently cause functional consequences because of accommodation or other type of compensation occurring at the level of the person or environment. This definition then was operationalized in the Questionnaire for Identifying Children with Chronic Conditions (QuICCC), a measure that reflects the consequences and impairments associated with chronic disabling conditions in children younger than 18 years.3
The consequences specified by this particular definition includes three types: functional limitations (FL), dependence on compensatory mechanisms or assistance (CD), and service use or need beyond routine care (SU/N). Although FLs and utilization of medical and other related services typically are used at least to some extent in measures to identify and describe children with disabilities, the concept of compensated function with accommodations is relatively new4and has only recently been conceptualized for children.5
These domains have been described in detail elsewhere.3 In brief, FL is conceptualized as limitation of function, activities, or social role compared with healthy age peers in the general areas of physical, cognitive, emotional, and social growth and development. To ensure that the activities addressed are age-appropriate, there are age or school status restrictions for some of the QuICCC questions that assess consequences in this domain.
CD is conceptualized as dependency on an accommodating mechanism to compensate for or minimize limitation of function, activities, or social role. When abilities are maximized and symptoms are reduced successfully, many children who have disabilities will be functionally indistinguishable from healthy peers. But for successful compensation to occur, other consequences are imposed. It is these secondary consequences that are tapped in this domain. For example, the child with epilepsy whose seizures are fully controlled with medication would be identified with the QuICCC item on medication use. The child with phenylketonuria who can live an ordinary life as long as a special diet is maintained rigorously would be identified with the QuICCC item on special diet. In both cases, absence of treatment would be expected to produce functional impairments.
Finally, SU/N is conceptualized as the use of or need for medical care or related services, psychological services, or educational services beyond the norm for the child's age, or use of or need for special ongoing treatments, interventions, or accommodations at home or in school. This domain reflects current service use as well as expressed needs for services that are not presently being met.
For the purposes of this study, disability was defined as having a consequence meeting the Stein et al2 definition as measured by the QuICCC. Recognizing that the generalizability of our analyses would be limited by the way we operationalized the definition, we separated disability into the three conceptual components or types of consequences probed by the QuICCC. We tested empirically how prevalence estimates of disability among children varied according to how disability was defined using the approach exemplified by the QuICCC. We also explored whether disabled children identified by FL, CD, and SU/N (or any combination of these domains) differed in 1) type of disorder or condition, 2) age, 3) socioeconomic status, or 4) race.
We conducted analyses on a national dataset representing a random sample of households with children in the United States. Data collection for the study was performed in 1992 by a large metropolitan market and public opinion research firm with considerable experience in conducting health-related surveys. Any household with children younger than 18 years with a telephone was eligible to be in the sample. Respondents, usually the parents, were adults who lived in the households and who were the most knowledgeable about the health of the children.
A multistage sampling design was used. First, the distribution of households with children by region of the country (New England, Middle Atlantic, East North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific) was calculated using the 1990 Census Population and Housing Summary. The sample was selected to reflect these proportions. In the next stage, a sample of assigned telephone banks was randomly selected from an enumeration of the Working Residential Hundred Blocks (a block of 100 potential telephone numbers within an exchange that includes three or more residential listings), and from that, telephone numbers were randomly selected for dialing.
Six attempts were made to reach a person at each telephone number before discontinuing. A total of 7998 numbers were actually dialed; 69% of those (n = 3639) were to households. There were 730 households that included children younger than 18 years and agreed to participate (refusal rate was ∼7%). Complete disability data were collected on 712 households, representing 1388 children. Age and sex data were available on all children. However, other sociodemographic information (household income, parent education level, and race) was collected from all households in which there was a disabled child, but only from a sample of approximately half of the households that did not have a child identified as disabled. The total number of children available for analyses on these variables was 811.
Table 1, column 1, presents the characteristics of the total sample. Fifty percent of the children were female. There was a generally even distribution among infants and toddlers (0 to 3 years old), preschoolers (4 to 6 years old), school-age/latency (7 to 11 years old), and adolescents (12 to 17 years old). The majority of children were white, and >75% of the adult respondents had at least a high school education (88%). Forty-five percent of household incomes were <$30 000 per year; 18% were <$15 000.
The QuICCC3 was used to identify children with disability in this study. As noted previously, the QuICCC reflects the noncategorical definition that we developed in earlier work and identifies children with chronic conditions using a consequence-based approach.2
There are 39 item sequences in the QuICCC. The items are generally structured in three parts. The first part operationalizes a concept from one of the domains (eg, special diet—reliance on a compensatory modality). The second part queries whether the consequence is due to a medical, behavioral, or other health condition, and the third part reflects the duration criteria (ie, whether the consequence has lasted or is expected to last for at least 12 months). The child must meet each of these three criteria in at least one question sequence to be identified. Table 2 gives an example of the question sequence using an actual QuICCC item.
The QuICCC has demonstrated content, construct, and criterion-related validity, and adequate test–retest reliability (2-week κ = .73). It is simple to administer and well received by respondents. A complete and detailed description of the development of the measure and its psychometric properties can be found elsewhere.3
The 39 items of the QuICCC were divided into three discrete sets to reflect three definitional components of disability it incorporates: FL (15 items), CD (12 items), and SU/N (12 items). Groups of children fitting the definitional components individually and in combination were compared to address the research questions. Analyses were performed using the microcomputer version of the Statistical Package for the Social Sciences software program.6
Proportion Identified as Disabled
Table 3 shows the proportions of children identified by each of the three conceptual components of disability examined in this study. Proportions are expressed both as a percent of the total sample (column 2) and as a percent of the identified (disabled) children (column 3). Both the type of consequence (FL, CD, or SU/N) and the number of consequences (1, 2, or 3) in all possible combinations are shown. This provides the most complete picture of some of the potential effects of using different definitional components of disability. Table 3 shows that overall, 256 of the 1388 children surveyed (or 18% of the total sample) were identified as disabled by the QuICCC.
Number and Types of Consequences
Table 3 also shows that of those identified as disabled, 49% had functional limitations, 55% relied on compensatory mechanisms, and 72% used or needed services beyond routine care. These percentages include all children meeting criteria for a given disability component, regardless of whether they were identified by that component only or in combination with one or both of the other two components. Table 3 also presents the distribution of children according to the number of definitional components that identified them as disabled. The largest proportion of disabled children (44%) were identified by one component only, nearly half were captured by SU/N alone. Thirty-six percent of disabled children were identified by two components in any combination, and 20% of disabled children met criteria for all three components (FLs + CD + SU/N).
Types of Disorders and Conditions
We examined the types of conditions found among children identified as disabled to determine whether they varied discernibly according to the three definitional components. Diagnostic information was obtained from a probe added solely for research purposes to the end of each question sequence.
More than 50 different conditions were reported, with considerable overlap among children identified by FL, CD, and SU/N. Children with the same diagnosis were reported to have a range of different consequences. For example, asthma, epilepsy, attention deficit disorder, and heart conditions were found among children identified by each of the three definitional components alone. Each component also appeared to be equally likely to pick up children with severe conditions.
Although far from clear-cut, some patterns also were suggested. Children with hearing impairments/deafness and/or visual impairments/blindness tended to be identified more often by the FL component relative to other definitional components; children with emotional and behavioral problems tended to be identified by the FL or SU/N components relative to CD. No other patterns were observed.
The diagnoses found among disabled children also appeared to vary by the number of definitional components that identified the child. Children identified by two or more definitional components were more likely to have multiple health conditions than children identified by only one (25% vs 4%). Furthermore, children identified by all three components tended to have more pervasive conditions such as Down syndrome and other forms of mental retardation, cerebral palsy, autism, and a variety of birth defects typically associated with disabilities.
Variations in Children's Characteristics by Definitional Components
The proportion of males to females in the total sample was exactly 1:1 (Table 1). However, significantly more of the disabled children than nondisabled were male (P < .05). Children were then stratified by type and number of definitional components. Among those identified as disabled, males were overrepresented among those with SU/N (58% vs 42%; P < .01) and CD (57% vs 43%; P < .05) compared with females. A nonsignificant trend for multiple (two or more) components to identify more males than females (59% vs 41%) was seen.
We divided children into four age groups reflecting different developmental stages (0 to 3 years, infants and toddlers; 4 to 6 years, preschoolers; 7 to 11 years, school-age/latency; and 12 to 17 years, adolescents). As shown in Table 1, children identified as disabled tended to be from the two older age groups; infants and toddlers appeared to be particularly underrepresented among children identified as disabled because there were fewer 0- to 3-year-olds than expected in the disabled group (11%) relative to their representation in the total sample (22%). These differences were not statistically significant.
When children in these groups were stratified by type and number of definitional component, smaller than expected percentages of 0- to 3-year-olds were likely to be found among children with FLs; only 8% of children identified by any FL were in this age group and there were no children in the 0- to 3-year age group among those identified by FLs alone. Slightly more young children (0 to 3 years old) than expected were included in the group identified by CD, but this was not a statistically significant difference. Finally, there were no statistically significant differences in age group by number of definitional components used to identify children.
Socioeconomic Status and Race
We have shown previously that estimates of the proportion of children defined as disabled by the QuICCC as a whole (ie, using the three types of consequences together) did not vary substantially across two samples with distinctly different sociodemographic characteristics.3 However, because prevalence rates derived from other measures, such as some of the disease-specific checklists, have been known to be vulnerable to just such reporting biases,7-9 we believed it would be important to examine socioeconomic status (and race) by the individual components of the definition.
Table 1 shows that the disabled group had more children in the lowest income category (< $15 000) than either the total sample or the nondisabled group (P < .01). After stratifying by type and number of consequences, we found that family income was not significantly associated with the number of components used to identify a child, but income did vary significantly with type of consequence. Specifically, among children identified by FL, there were fewer children than expected from the highest income group (>$45 000) (21% vs 32%; P < .05).
Education level of the parents with disabled and nondisabled children did not differ (Table 1). However, when we examined education level by type and number of definitional components, we found that there were fewer children from the highest education group (more than high school) than expected among the children identified by FLs (P < .01) and more from the same group (more than high school) among those identified by CD (P < .05). No other significant findings emerged.
More nonwhite children than expected (Table 1) were identified as disabled (P < .05). The only other significant difference found for race emerged in the analysis that stratified by number of definitional components; children identified by all three components were more likely to be white relative to their proportion in the total disabled sample (74% vs 70%; P < .01). There were slightly more white children among those identified by the CD component (75% vs 70%), but this difference only approached significance (P = .10).
Previous research has generated widely discrepant prevalence estimates.10-14 However, few empiric studies have examined why and how estimates vary. We have found that some components of disability in the QuICCC definition may be more likely than others to identify children. We recognize that the specific findings reported here pertain to a single definitional approach (the QuICCC). Other definitions using different conceptualizations and measures of disability, as well as different data sources (eg, physicians or direct physical examinations) may generate different rates of identification. The exact rates for any specific measure of disability will need to be determined by individual studies.
However, the results of this study do suggest that care be exercised when interpreting and comparing disability rates and characteristics of children identified as disabled. Our findings highlight the need for increased awareness that including or excluding different components of disability in the definition of disability may substantially affect prevalence estimates and may differentially identify children with particular individual characteristics. Several broad implications for program and service funding, and for health policy for children with disabilities, are raised, which should be considered in future work.
First, the policy implications of using a particular definition of disability may extend beyond the effects on prevalence. Because fiscal planning of programs and services depends heavily on available prevalence estimates, how disability is defined also can ultimately affect the number of children served. Recognizing which identifying criteria may produce underestimates of disability in children can be useful in guiding fiscal and policy decision-making. Data using the QuICCC methodology reported here suggest, for instance, that using functional limitations as the criterion might seriously underidentify disabled children 0 to 3 years of age.
Second, definitions of disability based on a single conceptual element may risk significant underidentification. Data using the QuICCC showed that only 20% of the children in this study met criteria for all three types of consequences that have been conceptualized as specifying disability. In this case, many children could have been missed if any one component alone were applied.
The risk of underidentification with any single component of disability will depend to large extent on the operationalization of the specific definitional elements. It also will depend on the degree to which the components do or do not overlap with each other and with our own conceptualizations of these domains. For example, we chose to differentiate service use from other elements that reflect compensatory or accommodating mechanisms. Other definitions of disability do not make the same conceptual distinction between person-level and system-level accommodations. Using fewer but broader domains will certainly decrease the likelihood that disabled children will not be identified with any single component or with any definition that focuses on only one aspect of disability. However, valuable information may be obscured as a result of this tactic.
Whether conceptualized as a separate disability domain or included under a broader category, we believe that children who rely on mechanisms or assistance (such as medications, special diets, assistive devices, or help from others) to maintain daily functioning or to minimize or compensate for illness consequences are a strategic group to include in any disability census. From the perspective of disability policy, recent legislation relating to the ADA of 1990 mandates a broader perspective toward the disabled community by addressing issues relevant to people who have ongoing health conditions with consequences, including those not directly experiencing disability in the classic sense. We anticipate that use of compensatory mechanisms will play an important role in the future management of chronic conditions in childhood.
The data from this study support the contention that compensatory mechanisms may be differentially available to children with disabilities; we found that children from more affluent socioeconomic backgrounds were less likely to report functional limitations as measured by the QuICCC than were children from more impoverished households. As health care technology improves, there will be greater and more varied options for children that can decrease the functional limitations commonly associated with chronic and disabling diseases that cannot be cured. Failure to include this component of disability will bias substantially our ability to monitor this critical change or to justify the health services necessary to sustain the compensatory mechanisms.
Disability definitions that make it easy to analyze this component separately also will be essential to the effort to identify, count, and monitor these specific children for health planning purposes. As our health care system shifts perspective, this subpopulation may be extremely vulnerable to the effects of funding cuts on medical and related services and to the provisions of managed care programs. Should compensatory mechanisms be less available to children because of lack of funding, the number of children exhibiting functional impairments might increase as might the number eligible for services and programs for the disabled. It could reveal much about the quality of health care to compare over time the relative proportions of those who have functional limitations to those who rely on compensatory mechanisms.
These analyses can only be suggestive of the ways in which different elements of disability may affect prevalence estimates and the descriptive characteristics of the disabled children they identify. We were able to approximate at least two of the major definitional types extant in the disability arena: functional limitations and service use. However, other concepts relevant to children with disabilities may not have been included in the QuICCC. Moreover, some of the consequences we assessed may have been conceptualized differently using other classification schemes. Other methodologies that do not rely on parental reports also may generate different results.
Because of the relatively small sample size, we were restricted in the level of subgroup analysis we could conduct while maintaining the integrity of the statistical analyses performed. The numbers available for some of the subgroup analyses on sociodemographic characteristics were especially small. Thus, it would be premature to conclude that any patterns revealed in the analyses regarding socioeconomic status, race, and education are reliable or stable.
It will be important to replicate these analyses in larger, broader datasets. To follow our preliminary work, we are conducting a similar analysis of data from the Children's Section of the Disability Supplement to the 1994–1995 National Health Interview Survey. This dataset represents the largest and richest source of information that currently exists on our nation's disabled children, with information available ultimately on ∼45 000 children. This and other analyses will provide critical information to elucidate the full scope of conceptual, methodologic, and definitional issues around disability.
Data on children with disabilities will play a critical role in informing the process of health care change. Existing data are already being applied to make program and funding decisions, and to frame the health policy issues for this population of children. Who will be classified as disabled and who will not is largely dependent on the definition used. Therefore, it is essential that the implications of using different elements of definitions for children with disabilities are understood so that data can be applied appropriately to policy and funding decisions.
This work was supported by Grant MCJ 367007 from the Maternal and Child Health Bureau and Grant 94ASPE261A from the Office of the Assistant Secretary for Planning and Evaluation.
- Received March 19, 1997.
- Accepted October 2, 1997.
Reprint requests to (R.E.K.S.) Department of Pediatrics, Albert Einstein College of Medicine, 1300 Morris Park Ave, Bronx, NY 10461.
An earlier version of this paper was presented at the 25th national meeting of the Public Health Conference on Records and Statistics; July 17–20, 1995; Washington, DC.
Dr Westbrook's current affiliation: Department of Neurology, New York University–Hospital for Joint Diseases, New York, NY.
- ADA =
- Americans with Disabilities Act •
- QuICCC =
- Questionnaire for Identifying Children with Chronic Conditions •
- FL =
- functional limitations •
- CD =
- dependence on compensatory mechanisms •
- SU/N =
- service use or need beyond routine care
- ↵Adler M. Conceptual issues and definitions for persons with mental retardation or other developmental disabilities. Presented at the American Public Health Association Meetings, Disability Forum Program; November 1996; New York, NY
- ↵Stein REK, Bauman LJ, Westbrook LE, Coupey SM, Ireys HT. Framework for identifying children who have chronic conditions: the case for a new definition. J Pediatr. 1993;122:342–347. Special article
- Stein REK,
- Westbrook LE,
- Bauman LJ
- ↵Verbrugge LM. The iceberg of disability. In: Stahl SM, ed. The Legacy of Longevity: Health and Health Care in Later Life. Newbury Park, CA: Sage Publications; 1990
- ↵Stein REK. Measuring health status in children: children with chronic health conditions. In: Primary Care Research: Theory and Methods. Conference proceedings of the Agency for Health Care Policy and Research. Washington, DC: DHHS, AHCPR; 1991;25–29
- ↵Norusis MJ. SPSS/PC+ Version 5.0. Chicago, IL: SPSS Inc; 1992
- Newacheck PW,
- Stoddard JJ,
- McManus M
- ↵Zill N, Peterson JL. Guide to Federal Data on Children, Youth, and Families. Washington, DC: Child Trends; 1990
- National Center for Youth with Disabilities
- Mattsson A
- Copyright © 1998 American Academy of Pediatrics