The Effect of Poverty and Caregiver Education on Perceived Need and Access to Health Services Among Children With Special Health Care Needs

Abstract

Objectives. We examined the association between several variables and the use of specialist physician services, developmental therapies, and prescription medications among children with special health care needs (N=38866).

Methods. We used a bivariate probit model to estimate whether a given child needed specialized services and whether that child accessed those services; we controlled for activity limitations and severity of special needs. Variables included family income, mother’s (or other caregiver’s) educational level, health insurance coverage, and perceived need for specialized services. We used data from the 2001 National Survey of Children with Special Health Care Needs.

Results. Lower-income and less-educated parents were less likely than higher-income and more-educated parents to say their special needs children needed specialized health services. The probability of accessing specialized health services—when needed—increased with both higher family income and insurance coverage.

Conclusions. Children with special health care needs have less access to health services because their parents do not recognize the need for those services. An intervention in the form of information at the family level may be an appropriate policy response.

Nearly 1 in 5 families in the United States have a child with special health care needs.1 Children with special health care needs, both physical and mental, are significantly more likely to live with poor families compared with children in general.2^–5 Although poverty is a risk factor for poor health among children, low parental education adds to this risk.6^–10 Lack of education is highly associated with low family income, and many families may not realize that children who “take their time” to gain particular skills have a disability or a delay that may be helped by a medical or therapeutic intervention. A survey of adults who accompanied special needs children during appointments with specialist physicians found that only half these caretakers (82.5% were parents) were able to provide even a lay description of their child’s diagnosis.11 Information, including available intervention strategies, was the greatest expressed need in a recent survey of parents of children with disabilities.12 This suggests that support services should be tailored to meet the needs of intended recipients—both children and their caregivers.13

Although children with special health care needs are as likely as other children to have health insurance coverage,14^,15 previous research has found lower use of specialty care among children whose parents were low income or less educated.8^–10 We hypothesized that information about special needs plays a key role in seeking specialty care—i.e., if parents do not think that their child needs a particular health care service, they will not seek access to that service. Therefore, even though a child with special health care needs may have insurance coverage for specialty care, service use will only occur if the parent seeks treatment.

We examined the association between several variables—family poverty, caregiver education level, health insurance coverage, perceived need for specialized health care—and the use of 3 types of specialty health care services—specialist physicians, developmental therapies, and prescription drugs—among children with special health care needs. We also examined caregiver-reported reasons for why some children’s service requests were not met. The perceived need for specialty health care services was assumed to be based on information provided to the parents.

Information has typically been characterized as an asymmetry problem in health services research—i.e., the physician has more medical knowledge than the patient does.16 We hypothesized that asymmetry also occurs when physicians (and possibly insurers) lack information about the efficacy of possible interventions7 and when physicians perceive immigrant, minority, less-educated, and lower-income patients to be less able to act on medical information.17^–19 In either case, physicians may not pass critical information on to the parents of their patients. Furthermore, higher-income and more-educated parents may have access to additional sources of information that less-educated and lower-income families generally do not have access to.20 As a result, poverty and lack of education may reduce a parent’s access to information about interventions for their child. Poverty also may contribute to a lack of health insurance, or underinsurance, and thus reduce access to needed services.

We therefore hypothesized that use of health services would be influenced by both perceived need (information) and access to care. We predicted that higher-income and more-educated parents would be more likely to report that their special needs child needed a specialized health service and that this child would also be more likely to access this service compared with a similar child with lower-income and less-educated parents.

We included additional variables in our study: (1) race/ethnicity, because discrimination may affect perceived need for specialized health care services; (2) number of adults in the household, because it may influence income status, and it may have some influence on levels of information; (3) native language, because parents who speak English as a second language may have difficulty gaining information and using the health care system; (4) child’s age, because health care needs change as children age; and (5) health insurance, because it affects access to health services.

METHODS

The 2001 National Survey of Children with Special Health Care Needs (NSCSHCN) was primarily funded by the Maternal and Child Health Bureau and was conducted by the National Center for Health Statistics. The survey was designed to provide state and national estimates of both prevalence and health services use of children aged 0 to 17 years with special health care needs.21 In the survey, children with special health care needs were defined as those who had “a chronic physical, developmental, behavioral, or emotional condition and who also required health and related services of a type or amount beyond that required by children generally.”22^(p41) Subjects were screened into the survey through a random digit-dialing sampling procedure that was stratified by state. There were 38 866 special needs interviews, 96% of which were completed by a parent of the child in question. The large sample size allowed for the identification of rural and urban residence in 34 states.10

Dependent Variables

We examined the perceived need for and access to specialist physician services, therapy services, and prescription medications. The NSCSHCN asked the following questions: “During the past 12 months, was there any time when CHILD needed care from a specialty doctor?” If the answer was yes, the follow-up question was, “Did CHILD receive all the care from a specialty doctor that he/she needed?” If the answer was no, the next question was, “Why did CHILD not get the care from a specialty doctor he/she needed?” The first 2 questions in this sequence were used to determine the values of the dependent variables in the estimated models.

Independent Variables

Health status.

The NSCSHCN did not ask parents to identify their child by diagnosis; therefore, for the purposes of our study, children were divided into groups on the basis of a severity-of-special-needs scale. Parents were asked, “Overall how would you rank the severity of CHILD’S condition(s) or problem(s)? Please pick a number between zero and ten where zero is the mildest and ten is the most severe.” Similar to previous research, we collapsed this scale into 4 categories: very low, low, medium, and high severity of the special needs condition.10 We also used a yes-or-no screener question to indicate whether the child was “limited or prevented in any way in (his or her/their) ability to do the things most children of the same age can do.”

Demographics.

Parents’ levels of information about their children’s condition and needs were not directly available. We used 2 variables—family income and caregiver’s educational level—as proxy measures for information. Family income, which has been shown to have an effect on access to health services and perceived need, was included as an independent variable when we estimated both dependent variables. We divided family income into 3 categories: incomes below the federal poverty level (calculated by dividing respondent’s household income by the Department of Health and Human Services’ guidelines for poverty by household size), incomes between 100% and 200% of the federal poverty level, and incomes above 200% of the federal poverty level. We controlled for the educational level of the mother (or the respondent if the mother’s information was not available), which was categorized as having less than a high-school education, having completed a high-school education, or having at least some post–high-school education or training.

Several control variables were included. Race and ethnicity were examined separately, with categories for White, Black, other (including multiracial), and Hispanic. Because respondents were not asked their marital status, children were categorized as having only 1 adult in the household or having 2 or more adults in the household. Categorical variables indicated whether there was more than 1 child with special health care needs in the household.7 Dichotomous variables included whether the interview was conducted in a language other than English and whether the residence was in a metropolitan statistical area.

To capture the change in health care needs as children age, we created an age spline with kinks in the spline that corresponded roughly to the ages at which public interventions change. The first segment of the spline included children aged 0 to 2 years—i.e., those who conceivably might be eligible for and participating in a public birth-to-3 program, such as those funded through Title V (Maternal and Child Health). The second segment included children aged 3 to 5 years, which are the preschool years during which children become the responsibility of their local school district under the Individuals with Disabilities Education Act.23 The third segment included children aged 6 through 13 years (traditional elementary-school years), and the fourth segment included children aged 14 through 17 years (high-school years).

Health insurance.

We asked questions about health insurance coverage during the past 12 months. Nonexclusive categorical variables indicated whether the child was uninsured at all during the past year and what types of insurance the child had during the months of coverage. Insurance categories included private coverage, Medicaid, the State Children’s Health Insurance Program (SCHIP), Title V coverage, and other public insurance (Medicare, military, and Indian Health Service).

Analyses

When modeling the use of health care services, a widely used technique is a 2-stage model: a dummy variable is the dependent variable in the first equation, which indicates whether or not the person accessed health services, and a continuous variable in the second equation measures the amount of health services accessed (measured through total cost of services).24 In this study, however, we did not have a continuous measure of service use among those who accessed services. Therefore, we used a 2-stage model with dummy variables as the dependent variables for both equations (1 = needed/accessed services, 0 = did not). Nevertheless, the disturbance (error) terms were correlated between the equations. The bivariate probit model handles this type of estimation,25 where each equation is a probit estimation equation with the error terms distributed with the normal distribution. Maximum likelihood estimation is used for the 2 equations. To estimate the bivariate probit model, different variables must be included in the different equations; therefore, some variables are indicated as “not included” in the tables.

The estimation of the second stage of the model—whether or not the child accessed the needed service—was modeled as conditional on the child’s need for service. Therefore, the model corrects for any selection bias that might be present, because those who indicated they needed services would access the services.25

RESULTS

Study Population

Characteristics of the sample are shown in Table 1 ▶. The most common special health care need identified was prescription medications (87.9%). More than half the families reported that their child needed specialist physician services (51%), and 23.5% reported that their child needed therapy services.

TABLE 1—

Sample Characteristics From 2001 National Survey of Children With Special Health Care Needs, Weighted Percentage of Sample

	Mean	No.
Dependent variables
Specialist physician services, %
Needed	51.0	38 755
Receiveda	92.4
Therapy services, %
Needed	23.5	38 780
Receiveda	88.2
Prescription medications, %
Needed	87.9	38 796
Receiveda	98.2
Independent variables
Income relative to federal poverty level, %		35 229
< federal poverty level	15.0
100%–200% of federal poverty level	22.0
> 200% of federal poverty level	63.0
Severity of functional limitations, %		38 655
Very low	17.0
Low	37.9
Medium	31.7
High	13.4
Education of mother or caregiver, %		38 866
< high school	14.4
High-school diploma	29.1
Some post–high-school education/training	52.7
Age of child, y	9.9	38 839
0–2	6.4
3–5	13.0
6–13	54.1
14–17	26.6
Insurance, %		38 866
Private	72.2
Medicaid	25.8
SCHIP	6.3
Other	7.8
Uninsured at least 1 month during past year	11.0
Has Title V coverage, %	3.0	38 866
Interview conducted in language other than English, %	3.5	38 866
Family composition, %		38 866
Total adults	2.1
Total children with special needs, %	1.4
Total children without special needs, %	0.9
Race/ethnicity, %		38 866
White	74.7
Black	14.9
Other	6.5
Hispanic	11.5

Note. SCHIP = State Children’s Health Insurance Program. Totals may add to more than 100% because respondents may report multiple responses to questions.

^aPercentage of those that needed specialist services.

Most children were described as having low to medium special health care needs (37.9% and 31.7%, respectively), and only 13.4% were described as having high needs. A little more than one third of the parents of children included in the sample had annual incomes below 200% of the federal poverty level (37%), which was similar to the proportion of mothers whose formal education had not extended beyond high school (43.5%). More than 70% of the children in the sample had been insured through a private policy sometime during the past year, and 11% had been uninsured for at least 1 month during the past year.

Information and Access to Specialty Health Care Services

The results of the multivariate analysis that was estimated with a bivariate probit model were somewhat difficult to interpret because of their nonlinearity; thus, they were converted into probabilities for more direct interpretation (full regression results are available from the authors). Table 2 ▶ shows the probability that a base case, or typical child with special health care needs, would both need services and then access services if the child needed the services. A typical child was defined as a child with the mean characteristics of all children in the data set; therefore, to compute the probabilities in Table 2 ▶, we used the mean of the variables in the regression models.

TABLE 2—

Results of Bivariate Probit Models of the Effect of Individual Characteristics on Probabilities of Service Need and Use, by Service Type: National Survey of Children With Special Health Care Needs, 2001

	Specialist Physician Servicesa		Therapy Servicesb		Prescription Medicationsc
Characteristics	Probability of Needed Services	Probability of Accessed Services	Probability of Needed Services	Probability of Accessed Services	Probability of Needed Services	Probability of Accessed Services
Base case, mean	0.536	0.925	0.202	0.730	0.896	0.993
Income relative to federal poverty level
< federal poverty level	0.465 ***	0.875 ***	0.231 *	0.730	0.863 ***	0.984 ***
100–200% of federal poverty level	0.508 ***	0.899 ***	0.208 *	0.722	0.875 ***	0.987 ***
> 200% of federal poverty leveld	0.562	0.940	0.198	0.733	0.910	0.995
Severity of functional limitations
Very lowd	0.433	0.965	0.137	0.762	0.900	0.994
Low	0.500 ***	0.932 ***	0.184 ***	0.759	0.898	0.994
Medium	0.591 ***	0.906 ***	0.242 ***	0.710 *	0.887 **	0.990 ***
High	0.669 ***	0.894 ***	0.294 ***	0.706 *	0.906	0.987 ***
Child has activity limitations	0.653 ***	. . .e	0.413 ***	. . .e	0.871 ***	. . .e
Education of mother/caregiver
< high school	0.439 ***	0.912	0.197	0.728	0.851 ***	0.993
High-school diploma	0.488 ***	0.931	0.202	0.783 ***	0.884 ***	0.994 **
Some post–high-school education/trainingd	0.566	0.924	0.203	0.705	0.905	0.992
Age of child, y splinef
1	0.719 ***	0.953	0.233 **	0.790 *	0.927 ***	0.993
4	0.559 ***	0.939	0.318 ***	0.740	0.897	0.995
9	0.509 ***	0.928 *	0.230 ***	0.733	0.892	0.994 **
15	0.538 ***	0.910 **	0.142	0.699 *	0.896	0.990
Insurance
Private	. . .e	0.932 ***	. . .e	0.733	. . .e	0.993 **
Medicaid	. . .e	0.944 ***	. . .e	0.743	. . .e	0.995 ***
SCHIP	. . .e	0.941 **	. . .e	0.769 *	. . .e	0.993
Other	. . .e	0.911 *	. . .e	0.747	. . .e	0.992
Uninsured at least 1 month during past year	. . .e	0.806 ***	. . .e	0.637 ***	. . .e	0.957 ***
Has Title V coverage	0.659	0.964 ***	0.460 ***	0.802 ***	0.880	0.992
Non-English interview	0.654 ***	0.908	0.213	0.592 ***	0.807 ***	0.993
Household composition
Single parent	0.540	0.902 ***	0.194 *	0.730	0.898	0.990 ***
More than 1 adultd	0.535	0.929	0.204	0.730	0.896	0.993
> 1 child with special needs	0.528	0.914 **	0.216 ***	0.741	0.900	0.989 ***
Race/ethnicity
White, non-Hispanicd	0.547	. . .e	0.204	. . .e	0.903	NI
Black, non-Hispanic	0.415 ***	. . .e	0.183 ***	NI	0.867 ***	NI
Other, non-Hispanic	0.513 ***	. . .e	0.206	NI	0.875 ***	NI
Hispanic	0.508 ***	. . .e	0.208	NI	0.880 ***	NI
Metropolitan status
Lives in a MSA	0.545 ***	0.924	0.199 *	0.710 ***	0.897	0.992
Lives in non-MSAd	0.513	0.928	0.208	0.764	0.895	0.993

Note. SCHIP = State Children’s Health Insurance Program; MSA = metropolitan statistical area. Probabilities were computed with all variables at their mean, except for the variable shown; probabilities were determined from regression results (not shown but available from the authors).

^ar = 0.2678** (equation error terms are correlated).

^br = 0.4800*** (equation error terms are correlated).

^cr = −0.8281 (not significant).

^dReference group.

^e Not included in model.

^f The age spline is computed as an interaction term between the age category indicated and the continuous variable age of the child.

* P < .10; **P < .05; ***P < .01.

Table 2 ▶ shows, for example, that the statistical probability was .536 that a base case child would need specialist physician services and was similar to the percentage who were identified during screening as having accessed more health care than their peers. However, a child with these characteristics would be very likely to get the services if their parents identified the child as needing the services, because the probability was .925 that this base case child would get specialist physician services. Similarly, the parents of the base case child would be very likely to indicate that the child needed prescription medications (P=.896) and to get medications if the parents indicated a need (P=.993). However, the parents of the base case child would be much less likely to indicate a need for therapy services (P=.202) and less likely to obtain these services even if the parents indicated that therapy was needed (P=.730). The finding that fewer than three quarters of children with special health care needs who needed therapy services actually obtained these services suggests a potential problem of access to care.

The power of the multivariate procedures is to isolate the effects of individual child characteristics on the probability of needing and accessing services while holding all other characteristics equal. Table 2 ▶ shows the probability of needing and getting services for a base case child with all the characteristics set equal to the base characteristics, except for the characteristic isolated and shown in the table. For instance, parents whose family income was below the federal poverty level would be less likely to indicate that their child needed specialist physician services (P = .465) and also less likely to get the needed services (P = .875) compared with a base case child (P = .536 and .925, respectively).

We tested whether access to information affects parents’ ability to know whether their child needs services and possibly their ability to access needed services, with both the education and the income variables. We found support for the hypothesis that education influences awareness of need for services, but education had less impact on access to services. For example, mothers who did not complete high school were about 13 percentage points less likely to indicate that their child needed specialist physician services (P = .439) compared with mothers who had college degrees (P = .566). The less-educated mothers also were about 5 percentage points less likely to report a need for prescription medications. There were mixed results with respect to the impact of mother’s education on actual use of services, with children of less-educated mothers more likely to access specialist physician services and therapy services.

Consistent with our hypothesis about the role of family income, Table 2 ▶ shows that a child whose family was poor was significantly less likely to use specialist physician services and prescription medications compared with a child whose family income was above 200% of the federal poverty level. However, lower-income parents also were less likely to report that their child needed specialized health services.

Parents of children with activity limitations were significantly more likely to report that their children needed specialist physician services and therapy services, although the actual use of specialist physician services was significantly lower among these children, and use of therapy services was significantly higher. As expected, access to health insurance played a key role in gaining access to health services, although our analysis suggests that type of health insurance makes little difference, particularly for use of therapy services (Table 2 ▶).

Why Children Did Not Receive Services

Parents of children who did not access all services needed were asked why services were not accessed. The 2 most common responses were that the services “cost too much” or that there was a “health plan problem” (Table 3 ▶). However, the patterns of responses differed quite a bit across the services. For example, although 60.5% of respondents reported that prescription medications cost too much and 29% reported a health plan problem, the responses were reversed for therapy services, with 19.1% reporting the services costs too much and 22.6% reporting a health plan problem. Therefore, adequate coverage for therapy services was a significant barrier to accessing care for children with special health care needs. In addition to cost and health plan problems, “lack of resources at school” was often cited as a reason for not accessing all necessary therapy services.

TABLE 3—

Reasons Why Needed Services Were Not Accessed and Frequency of Response: National Survey of Children With Special Health Care Needs, 2001

Service	Reasons Why Not Accessed	Frequency of Response, Mean (%)
Specialist physician services	Costs too much	27.60
	Health plan problem	24.80
	Not available in area or transportation a problem	9.70
	Inconvenient times	7.70
	Doctor did not know how to treat	10.50
	Some other reason	17.20
	No insurance	2.70
	Child refused to go	4.10
	Difficulty getting appointment	3.90
	Dissatisfaction with provider	2.20
	Treatment is ongoing	4.60
Prescription medications	Costs too much	60.50
	Health plan problem	29.00
	Not available in area or transportation a problem	2.50
	Doctor did not know how to treat	2.20
	Some other reason	11.80
	No insurance	5.30
Did not access all needed therapy services	Costs too much	19.10
	Health plan problem	22.60
	Not available in area or transportation a problem	12.00
	Not convenient times	7.50
	Doctor did not know how to treat	3.80
	Some other reason	19.10
	No insurance	2.20
	Difficulty getting appointment	2.40
	Treatment is ongoing	8.30
	Lack of resources at school	14.90

Note. Totals may add to more than 100% because of rounding or because respondents report multiple responses to questions.

DISCUSSION

Our study documents the impact of information (measured by proxy through poverty status and mother’s education) on the perceived need for and access to 3 specialized health services for children with special health care needs. Access appears to be driven by the income and educational status of the parents and the severity and nature of the special needs. Both income and education affect access indirectly, because less-educated and lower-income parents have a lower perceived need for specialized health care services. Lower-income parents were more likely than higher-income parents to report that their child with special health care needs had severe functional limitations, but they were less likely than higher-income parents to say that their children needed specialized health care services. Some of these findings may be the result of parents who believed services were not affordable, because “high cost” was the reason most often reported for not accessing services (Table 3 ▶). However, the empirical results presented here point to another explanation: because income levels of parents affected their perceived need for specialized health care services—which should be there regardless of income—lower-income parents may lack the information or the resources to navigate the difficult terrain of accessing health care for their children. We saw a similar pattern when comparing families by mother’s education. The probability of accessing specialized health care services increased with family income.

Many parents who did not access services cited the cost of services as the reason. It is not surprising that respondents cited out-of-pocket costs of services as a barrier, because in 2004, 97% of families with private insurance coverage were responsible for office visit copays or coinsurance, and nearly 80% of families in preferred provider organizations who saw in-network providers were responsible for annual deductibles that averaged $287.26 A separate 2003 survey reported that 42% (up from 28% in 2001) of working-age Americans with chronic illnesses spent more than 5% of their income on out-of-pocket medical costs (excluding health insurance premiums), despite being covered by private health insurance.27

Insurance increases the probability of accessing health services, although insurance type did not appear to matter in our analysis. With the exception of therapy services, nearly everyone who reported a service need did indeed access the needed service. Interestingly, public insurance and private insurance appeared equally effective at providing access to specialist physician services and prescription medications, but they were equally ineffective at providing access to therapy services. However, we also found that parents of children with special health care needs were less likely to access health services if they had been uninsured for at least 1 month during the past year, which suggests that a broader policy response may be appropriate. Title V programs are available to only a small proportion of this population, but these programs appear to be particularly effective at providing access to specialized care.

Our results clearly point to the importance of targeted outreach to low-income and less-educated parents who have children with special health care needs. We found these children were less likely to access health services because their parents did not recognize the need for those services or did not know what services were available. Intervention in the form of information at the family level may be an appropriate policy response, particularly among harder-to-reach subpopulations, such as those with lower incomes or lower literacy.

The way parents obtain access to therapy services needs to be revised. Anecdotal evidence suggests that insurance plans often limit or restrict access to therapy services, although no systematic research appears to have been done in this area. Our results show that parents cited health plan problems more often than costs of care as a barrier to accessing therapy services. Although anecdotal evidence suggests that children with special health care needs benefit greatly from therapy services, the research evidence is mixed.28 Among school-aged children, therapy services are often accessed through the public schools (as mandated under the Individuals with Disabilities Education Act), yet many of the parents in our study reported that their child’s school-based therapy was “inadequate.”

Although children with special health care needs are disproportionately covered by Medicaid,29 the role of SCHIP in providing access to health care services for these children is being examined. In a 3-state study, 16% to 25% of new enrollees in SCHIP were children with special health care needs, and fewer than half of these children were covered by any insurance before SCHIP enrollment.30 Because of recent state proposals to cut SCHIP plans,31 it is imperative to know what impact the type of public health insurance has on the population of children with special health care needs.