An Expanded Approach to the Ascertainment of Children and Youth With Special Health Care Needs

Data were pooled from the 2016 to 2021 NSCH, which is an annual, cross-sectional, address-based, self-administered survey using web- and paper-based methodologies.⁹  The NSCH employs a 2-phase design in which a screener questionnaire is initially used to identify households with children, roster the household, collect demographic information, and identify CYSHCN using the CSHCN Screener. Next, a randomly selected child is then the subject of a more detailed and age-specific questionnaire (known as the topical) that collects information about health, health care, and related family- and community-level factors that can impact health. The design and administration of all 6 survey years have been described previously; data and questionnaires are publicly available.¹⁰  The overall weighted response rates in each year were 40.7%, 37.4%, 43.1%, 42.4%, 42.4%, and 40.3%, respectively, and the weighted interview completion rates (reflecting the probability that a household with children that started the survey completed it) were 69.7%, 70.9%, 78.0%, 79.5%, 81.2%, and 79.5%.^{9 ,11 –15} 

SHCN status was determined on the basis of parent or caregiver responses to the CSHCN Screener,²  which is a 5-item tool used to assess the need or use of prescription medications, above-routine medical services, and specialized therapies or functional limitations that are due to any health condition expected to last at least 12 months, as well as a specific criterion assessing the need or use of treatment or counseling for an emotional, behavioral, or developmental condition (Supplemental Table 5). Anchored to the federal CYSHCN definition, the CSHCN Screener does not require reports of a diagnosed condition, although the majority of children identified do. Children who did not meet any of these 5 criteria are henceforth referred to as non-CYSHCN.

Separate from the CSHCN Screener, the NSCH topical has 22 survey items to ascertain diagnosed health conditions and 12 survey items to ascertain functional difficulties (Supplemental Table 6). Children with at least 1 diagnosed condition of the 22 assessed were categorized as ≥1 conditions. Similarly, children with at least 1 functional difficulty of the 12 specific functional difficulties assessed were categorized as ≥1 difficulties.

Outcomes were selected on the basis of their salience to models of a well-functioning system of services for CSHCN^16 ,17  and reflect child- (≥2 emergency department [ER] visits, unmet health care needs, flourishing, and activity limitations), family (out-of-pocket costs >$1000, family members employment impacts due to child’s health, and frustration obtaining services) and systems-level indicators (inadequate health insurance and medical home access; Supplemental Table 7).

Sociodemographic characteristics associated with either the presence of SHCN or outcomes of interest were selected a priori and included child’s age, sex, race and ethnicity, health insurance, family income-to-poverty ratio as measured by Census Bureau poverty thresholds,¹⁸  household educational attainment, number of children in the family, and primary household language.

The combined analytic sample (n = 225 443) was split on the basis of the presence of ≥1 diagnosed condition of those assessed, ≥1 functional difficulty of those assessed, and the presence of SHCN (qualify as CYSHCN on the screener) such that the sample was stratified into 4 mutually exclusive groups. These groups include (1) CYSHCN with ≥1 condition and ≥1 difficulty, (2) non-CYSHCN with ≥1 condition and ≥1 difficulty, (3) CYSHCN without ≥1 condition and ≥1 difficulty (child had either ≥1 diagnosed condition only, ≥1 difficulty only, or neither conditions or difficulties), and (4) non-CYSHCN without ≥1 condition and ≥1 difficulty (child had either ≥1 diagnosed condition only, ≥1 difficulty only, or neither a condition or difficulty; Fig 1). CYSHCN without both conditions and difficulties and non-CYSHCN without both conditions and difficulties are intentionally inclusive of all other subgroups of children (conditions only, difficulties only, and those with neither conditions or difficulties) to test our first hypothesis that non-CYSHCN with ≥1 condition and ≥1 difficulty may experience adverse impacts that are similar to those experienced by children identified as CYSHCN with ≥1 condition and ≥1 difficulty. Our second hypothesis is that non-CYSHCN with ≥1 condition and ≥1 difficulty will have elevated adverse impacts compared with non-CYSHCN without both conditions and difficulties.

First, prevalence estimates were calculated for each subgroup. Next, the percentage of children with SHCN was calculated among each of the 22 conditions and 12 functional difficulties examined. Bivariate associations between each subgroup, sociodemographic characteristics, and health and related outcomes were examined by using χ-square tests and t tests. Adjusted multivariable logistic regression was used to calculate the adjusted prevalence ratios (APRs) and 95% confidence intervals (CI) to determine the associations with 9 health and related impacts across subgroups. Finally, national- and state-level estimates were calculated for CYSHCN by using both the traditional scoring methodology and an expanded approach, which includes children with at least 1 condition and 1 difficulty, regardless of meeting CSHCN Screener criteria.

Sample weights, developed by the US Census Bureau, that account for differences in the probability of household selection and nonresponse were applied to produce estimates representative of all noninstitutionalized children aged 0 to 17 years. Multiply imputed values for missing family income-to-poverty ratio (missing: 15.3% to 19.7%) were used. All analyses accounted for the complex survey design and were performed using SAS-callable SUDAAN, version 11.0.1 (Research Triangle Institute, Research Triangle Park, NC).

Nearly 7% of non-CSHCN had both ≥1 diagnosed conditions and ≥1 functional difficulty, 16.7% had ≥1 conditions alone, and 9.8% had ≥1 difficulties alone (Table 1). For CYSHCN, the majority (61.2%) had both ≥1 diagnosed conditions and ≥1 functional difficulty, 29.7% had ≥1 conditions alone, and 3.9% had ≥1 difficulties alone.

More than 50% of children whose parent reported that they had been diagnosed with ≥1 of the 22 conditions assessed and >50% of children that had ≥1 of the 12 difficulties did not meet the CSHCN Screener criteria for SHCN (Fig 2). The percentage of children with conditions and functional difficulties not identified as CYSHCN ranged from 6.3% (among those with cerebral palsy) to 59.1% (among those with allergies), with more than one-third of children with asthma (34.5%), cystic fibrosis (41.4%), heart conditions (44.6%), frequent headache (44.7%), and blood disorders (54.6%) not identified as CYSHCN. For the difficulties examined, the range was 12.1% (among those with difficulty dressing or bathing) to 84.2% (among those with difficulty using hands).

χ² and t tests conducted for CYSHCN with both conditions and difficulties compared with non-CYSHCN with both conditions and difficulties revealed that those who were younger, female, Hispanic, Asian, uninsured or privately insured, living in a household in which the parent’s highest education level was less than high school, high school or general educational development education, multiple children households, and a primary household language other than English were less likely to be identified as CYSHCN on the basis of the CSHCN Screener (Table 2).

Non-CYSHCN with ≥1 condition and ≥1 difficulty were more likely than CYSHCN without conditions and difficulties to have unmet health care needs (6.4% vs 4.8%) and equally likely to have ≥2 ER visits (Table 3). Conversely, non-CYSHCN with ≥1 condition and ≥1 difficulty were less likely than CYSHCN without conditions and difficulties to have frequent and significant activity limitations (3.5% vs 7.2%) or not to be flourishing (30.0% vs 32.9%). After adjustment, non-CYSHCN without both conditions and difficulties, non-CYSHCN with ≥1 condition and ≥1 difficulty were more likely to be negatively impacted across all child-level outcomes (Table 4). Non-CYSHCN with both conditions and difficulties were 2 to 3 times as likely to have ≥2 ER visits, unmet health care needs, frequent and significant activity limitations and not to be considered flourishing compared with non-CYSHCN without both conditions and a difficulties.

Non-CYSHCN with ≥1 condition and ≥1 difficulty were more likely to experience each of the family-level outcomes compared with non-CYSHCN without both conditions and difficulties (Table 3). Non-CYSHCN with ≥1 condition and ≥1 difficulty were less likely to report out-of-pocket costs >$1000 (15.9% vs 22.0%) than CYSHCN without both conditions and difficulties but were equally likely to report family members experiencing employment impacts and frustration acquiring services for their children. After adjustment, non-CYSHCN with ≥1 condition and ≥1 difficulty were consistently more likely than non-CYSHCN without both conditions and difficulties to report the 3 negative family-level outcomes examined (Table 4). For example, this group was more likely to have out-of-pocket costs >$1000 (APR: 1.53 95% CI:1.40–1.67), and 2 to 3 times more likely to experience employment impacts or frustration getting services for their children.

For both systems-level outcomes, a stepwise relationship was noted such that CYSHCN with ≥1 condition and ≥1 difficulty fared the worst, followed by non-CYSHCN with ≥1 condition and ≥1 difficulty, CYSHCN without both conditions and difficulties, and non-CYSHCN without both conditions and difficulties (Table 3). After adjustment, compared with non-CYSHCN without both conditions and difficulties, non-CYSHCN with ≥1 condition and ≥1 difficulty were more likely to have inadequate health insurance (APR: 1.43 95% CI:1.34–1.52) and more likely to lack a medical home (APR:1.12 95% CI:1.07–1.16) (Table 4).

Using an expanded approach that included both CSHCN Screener criteria and the presence of conditions and difficulties regardless of CSHCN Screener responses, the national prevalence of CYSCHN would increase from 19.1% to 24.6% and state-level estimates would range from 17.8% in Hawaii to 30.1% in Arkansas (Fig 3). Percentage point increases at the state level would range from ∼4% in Hawaii, Illinois, Minnesota, and Wisconsin to ∼7% in Idaho, Arkansas, New Mexico, Mississippi, and Alabama. State rankings and comparisons to the national average using the expanded CSHCN calculation were similar to those when the CSHCN Screener is used (Supplemental Table 8).

The CSHCN Screener is used to identify children with any type of ongoing condition who need or use an above-routine amount of health and related services or experience functional limitations. Only approximately half of children with 1 or more diagnosed health conditions and half of children with 1 or more functional difficulties met CSHCN Screener criteria. A small but significant proportion of children with both diagnosed health conditions and difficulties did not qualify as CYSHCN using the CSHCN Screener (∼4 032 000 children). The authors of previous work have acknowledged that there may be children who meet the MCHB definition of having SHCN but do not qualify using the CSHCN Screener.^3 ,19 ,20  Consistent with the adoption of a noncategorical approach to defining CYSHCN, there has been a long-standing expectation that not all qualifying children may have a formal condition diagnosis, and not all children with diagnosed conditions will experience negative impacts. For example, it is plausible that nearly 60% of children with diagnosed allergies would not qualify as SHCN on the basis of the duration or impact of their symptoms. However, we were not able to make the same argument for the 40% of children with cystic fibrosis or for approximately one-sixth to one-quarter of children with mental, emotional, or behavioral conditions, particularly considering current trends in mental health problems among youth.^21 –24  This raises questions about sole reliance on the CSHCN Screener to identify children who meet the MCHB definition (not considering “at risk”) of SHCN. An expanded approach to quantify CYSHCN leverages presence of both conditions and difficulties regardless of CSHCN Screener results, therefore casting a wider net of CYSHCN. This subgroup of non-CYSHCN with both conditions and difficulties are more likely than non-CYSHCN without both conditions and difficulties to have health impacts at the individual, family, and systems levels, and these health impacts are similar to those of children that met the CSHCN Screener criteria.

Importantly, we observed sociodemographic disparities in the proportion of children who were not identified as CYSHCN despite having both health conditions and difficulties. These children were more likely to be younger, female, Hispanic, Asian, privately insured or uninsured, and in households with lower adult educational attainment, as well as to have families with more children and a primary household language other than English compared with children who met the criteria for SHCN. Many of these differences highlight potential biases and linguistic or cultural limitations of the current screening tool to identify children from marginalized backgrounds who may require additional support and services. Others, such as young age, may be a result of parents’ uncertainty to endorse duration qualifiers. Given the variation in complexity of SHCN, different populations, such as those who are privately insured, may be at an advantage for identifying a child as having elevated service needs, although children in poverty are more likely to be identified as CYSHCN using the CSHCN Screener. Ultimately, a complex interplay exists between the cultural and socioeconomic contexts in which a child’s health care needs are understood, the health care system in which they and their family are served, and how parents are able and willing to report on related experiences and impacts.

These analyses reveal a subgroup of children with elevated health and related impacts that are not accounted for in CYSHCN prevalence calculations. Children not identified as SHCN despite having health conditions and difficulties were more likely to experience impacts in daily activities and experience frustration in accessing health services compared with non-CYSHCN without both conditions and difficulties. An expanded approach to identifying CYSHCN may provide a more inclusive representation of the population. Continued use of different CSHCN Screener scoring models can be used to stratify groups of CYSHCN on the basis of variations in complexity or needs, including the use of the initial impact question without qualifying on duration.¹⁹  However, implications for funding and organizing related services should be considered.

Given the discordance between the presence of diagnosed conditions and difficulties and the lack of endorsement of CSHCN Screener items, various aspects of the CSHCN Screener administration, such as the condition duration, health, and health care need or use consequences included, may be areas for further study. This point is underscored by the findings that Hispanic children and children in lower educated households are less likely to be identified by the CSHCN Screener. These populations are less likely to qualify on a variety of health risk assessments pointing to cultural variations in line with previous work but may warrant further study.^25 ,26 

NSCH data are based on parent or caregiver report, and recall bias is possible. Additionally, because reports are not confirmed by health care professionals or medical records, there is a possibility of misreporting conditions. The small number of children identified on the CSHCN Screener with neither conditions or difficulties is in line with previous findings.³  It is possible that these children have conditions and difficulties not assessed on the NSCH or whose conditions have yet to be diagnosed but that, nonetheless, result in ongoing health consequences and need for services. Their inclusion’s impact on findings was not thoroughly investigated.

There are groups of children with both conditions and difficulties that have elevated health impacts and barriers to accessing services but are not currently identified as SHCN. Including children with conditions and difficulties in the calculation of CSHCN would result in a larger proportion of CYSHCN both nationally and in each state.

Thank you to Drs Ashley Hirai and Stephen Blumberg for their review of the manuscript and conceptual guidance, respectively.

APR

adjusted prevalence ratio

CI

confidence interval

CSHCN

children with special health care needs

CYSHCN

children and youth with special health care needs

ER

emergency department

MCHB

Maternal and Child Health Bureau

NSCH

National Survey of Children’s Health

SHCN

special health care needs