Hospital Inpatient Stays for Autistic Youth and Youth With Other Disabilities

This study used data from hospital inpatient discharges from the Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality National Inpatient Sample (NIS) 2017.⁹  The NIS is a representative sample of an all-payer database of US hospital inpatient stays. NIS 2017 captured >7 million stays from 47 states plus the District of Columbia, covering 97% of the US population, excluding rehabilitation and long-term acute care hospitals.

Patients ages 0 to 17 were included in this study. Index stays may record up to 40 diagnostic codes by using the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM). Autistic youth were captured by codes F84.0, F84.5, or F84.9 in any position in the record. Patient demographic characteristics including age, race, and sex are captured by administrative hospital data. The inclusion of race is important in acknowledging and addressing racial and ethnic health disparities.

The MBND comparison group was created by using Clinical Classifications Software Refined (CCSR) for ICD-10-CM-coded diagnoses version 2020.3, housed and curated by HCUP. CCSR aggregates ICD-10-CM diagnosis codes into smaller, clinically meaningful categories that generally align with ICD-10-CM diagnosis chapters. The CCSR body system for MBNDs includes 32 subcategories, 1 of which includes autism. The MBND comparison group was created by using all 32 subcategories, excluding stays with autism in any diagnosis position, which generally correspond to ICD-10-CM codes beginning with “F.”

The NIS contains information on the total charge for each stay. Costs were calculated by using hospital-specific charge-to-cost ratios provided by HCUP generated on the basis of information collected by the Centers for Medicare and Medicaid Services. Costs were then evaluated for outliers and Winsorization was used to censor the distribution at the first and 99th percentiles to reduce variability at the ends of the distribution.

Index stay demographic and stay characteristics were compared between autistic youth and youth with MBND. Significant differences between groups were assessed by using logistic regression, as appropriate for complex survey data. The principal diagnosis is often considered the primary reason for hospitalization. The prevalence of the 10 most common principal diagnoses for index stay, categorized by using the CCSR taxonomy, were calculated for autistic youth and MBND youth. When appropriate for interpretation, CCSR categories were combined to create categories such as mood disorders, physical injury, stomach or intestinal diseases, and pregnancy. For the 10 most common principal diagnoses in the autism group, the average length of stay and cost per day were calculated in the autism and MBND groups. Analysis was conducted by using Stata 16 to account for the complex survey design.

In 2017, autistic youth had 40 000 inpatient stays and MBND youth had 350 000. Autistic youth were younger, more often male, White, from zip codes with higher median household incomes, and from the Northeast (Table 1). The mean cost per day was $400 higher in autistic youth than youth with MBND although the total cost per stay was $900 lower.

Of every 1000 stays, 7.3 were for autistic youth and 65.2 were for youth with MBND. The rate of autism stays was higher in zip codes with higher median household incomes: 6.7 per 1000 in zip codes in the 0 to 25th percentile compared to 7.9 per 1000 in zip codes in the 76th to 100th percentile. This was a reverse of the trend for MBND: 66.2 per 1000 in zip codes in the 0 to 25th percentile compared to 62.5 per 1000 in zip codes in the 76th to 100th percentile. Rates differed by region for autistic youth, with higher rates in the Northeast (10.1 per 1000) and Midwest (8.3) than the South (6.9) and West (5.3). The highest rates of MBND stays were in the Midwest (88.5 per 1000), versus Northeast (66.5), South (63.5), and West (45.6).

The most common principal diagnosis in autistic youth was mood disorders, as in youth with MBND (Table 2). All top 10 principal diagnoses for autistic youth were ACSCs with exception of physical injuries. Three non-ACSC conditions were on the top 10 list for MBND: physical injury, pregnancy, and antineoplastic therapies.

Figure 1 examines the cost per day and length of stay for the 10 most common principal diagnoses in autistic youth compared with youth with MBND. Many conditions were similar in cost and length of stay for autistic youth and MBND youth. Physical injuries resulted in longer stays for youth with MDNB, whereas schizophrenia resulted in longer stays for autistic youth. The highest average cost per day for autistic youth was for physical injuries ($4320 per day) followed by epilepsy ($3480 per day).

This study found a high rate of admissions for ACSCs for autistic youth. ASCSs were some of the longest (schizophrenia) and most expensive (epilepsy) stays. The high rates of ASCSs suggest that primary and specialty outpatient care may not currently adequately address the health and mental health needs of autistic youth. More than half of the ACSCs identified in autistic youth were related to mental health concerns. Escalation of health and mental health concerns are likely impacted by the management of mental health issues.

The long length of stay for schizophrenia, disruptive, impulse-control and conduct disorders, and mood disorders is also of note. Clinical complexity and high rates of condition cooccurrence are likely drivers of this finding. We lack evidence for effectively treating many disabilities when they cooccur, especially for autistic youth. In addition, the natural course of the development and experience of co-occurring conditions may result in changing levels of service needs over time. Although this study begins to describe cooccurrence of diagnoses in autistic youth, more research is needed to understand the rate and mechanisms of cooccurrence with the ultimate goal of expediting bench-to-bedside practice to improve care. Autism characteristics may also be related to longer stays; acclimation to a new setting, disruptions of routine, and lack of familiarity may result in emotional or violent behavior, adding complexity to the stay that results in additional days and even use of restraints. Hospital settings and staff that do not understand or accommodate neurodivergent patients will be limited in the quality of care they can provide. The difference in admission rates by region and income also highlight areas of concern for adequate treatment of ACSCs within primary care settings.

Two other considerations are illuminated by these findings. First, racial and ethnic disparities need to be investigated further, as does the higher rate of Black MBND versus Black autistic youth admissions. Racial and ethnic minority autistic youth are more often misdiagnosed and underdiagnosed than White youth, and behaviors associated with autism may be misinterpreted.¹⁰  Concerns of misattribution of autism characteristics may lead to more use of restraint, longer stays, and more inaccurate diagnoses associated with behavioral issues such as violence in autistic Black children. Second, Medicaid was the majority primary expected payer in autistic youth inpatient stays. This group likely faces multiple contributing factors that exacerbate health issues and barriers to getting needed care, including poverty and disability among other social determinants of health, including race. How care is provided to these vulnerable populations is of great concern.

Although the NIS provides a national snapshot of inpatient hospitalizations in a given year, the following limitations are important to consider. The use of principal diagnosis allows for a convenient way to study the reason for admission, but it likely differs by diagnosis, location, provider, and insurer, as diagnostic coding influences reimbursement. We also did not explore the clustering of diagnoses that might be related to certain reasons for admission, nor variation by race, ethnicity, sex, or other important social determinants of health. However, the NIS provides an unparalleled opportunity to examine all-payer inpatient hospitalizations in autistic youth in the United States. These findings provide new insight into the health care of this population, examining inpatient stays while highlighting the need for more comprehensive preventive and maintenance care.

ACSC

ambulatory care sensitive conditions

CCSR

Clinical Classifications Software Refined

HCUP

Healthcare Cost and Utilization Project

ICD-10-CM

International Classification of Diseases, 10th Revision, Clinical Modification

MBND

mental, behavioral or neurodevelopmental disabilities

NIS

National Inpatient Sample