Brief Report: Healthcare Utilization and Expenditure Trends Among Autistic Transition Age Youth

We used national Medicaid Analytic eXtract Files (MAX and Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF) for analyses. We selected a cohort of individuals aged 16–19 in 2015 who were enrolled in Medicaid in both 2015 and 2019 (aged 20–24). We limited our analysis to 2015–2019 to avoid the impact of period effects related to COVID-19 on longitudinal change in healthcare utilization. We included those enrolled for at least 10 months in a given calendar year to account for administrative churning and in alignment with previous claims-based research focused on autism (Grosse et al., 2022). We selected the claims records for enrollees with ASD or ID using validated algorithms. ASD was identified using 2 outpatient or 1 inpatient claim associated with an ASD diagnosis code (299.xx; F84.x). ID was identified as any use of claims codes 317.xx-391.xx, or F70-F79. Consistent with other published analyses of Medicaid claims data from our study’s time period (Bettenhausen et al., 2018), we only included fee for service (FFS) and primary care case management claims (PCCM; managed care plans paid through fee for service). Although the quality of Medicaid comprehensive managed care claim data has improved over time (Samples et al., 2025), there was a temporary and modest decline in quality from 2014 to 2016 as states transitioned from MSIS to T-MSIS systems. Considering that our primary objectives were to examine longitudinal change in healthcare from this time to 2019, and to draw comparisons to prior analyses (2001–2005) that were limited to FFS claims, we limited our analyses to FFS claims.

In Shea and colleagues’ (2018) original analysis of 2001–2005 claims data, there were two comparison groups: ASD only and ID only; we added a third group: ASD + ID. The sizes of each group were ASD only: N = 23,460; ID only: N = 21,256; ASD + ID: 10,115. Our primary service type categories were outpatient, inpatient, long-term, and pharmaceutical care. These types are pre-defined in the Medicaid analytic files as described in Table S1 (Christensen et al., 2021). For outpatient, inpatient, and long-term care utilization and expenditures, we distinguished psychiatric care from non-psychiatric (i.e., medical) to be able to draw comparisons to Shea et al. (2018), and because of the high risk for psychiatric conditions among ASD and ID populations (Buckley et al., 2020; Lai et al., 2019). We identified psychiatric care using codes 290.xx-319.xx and F01-F99 (excluding codes for ASD and ID) in any position on the claim, consistent with the approach used by Shea and colleagues. All claims not identified as psychiatric were considered “medical.” Medications were classified as psychiatric using the American Hospital Formulary Service (AHFS) Pharmacologic-Therapeutic Classification©. Drugs categorized within the central nervous system (CNS) agents that fell under the classification of anticonvulsants; psychotherapeutic agents; anxiolytics, sedatives, and hypnotics; antimanic agents; and amphetamines were considered as psychiatric use. A complete list of CNS agents and other medications are available in the AHFS Pharmacologic-Therapeutic Classification©. We defined utilization as at least one claim for a particular type of care during the year of analysis (2015 or 2019). Yearly healthcare expenditures were adjusted for inflation to 2019. We examined both overall per capital expenditures as well as per capita expenditures excluding those with zero expenditures, and Winsorized expenditures within each expense category at the 95th /5th percentile in order to minimize the influence of outliers. The overall expenditures provided a sense of population-wide HCRU, while the expenditures among those with >$0 expenditures provided a better sense of average expenditures among those receiving care. We calculated percent utilizing each service, mean (SD) expenditures, and 25th, 50th, and 75th percentiles for expenditures, for 2015 and 2019. Owing to the substantial sample size, statistical significance testing was not performed; rather, interpretation emphasized clinically meaningful differences. All members of our target population who met inclusion criteria were included in our analysis, and with extremely large sample sizes p-values provide little valuable information. Rather, we interpreted differences between estimates based on their relative size and meaningfulness in the context of prior findings and real-world context.

This study provides updated estimates of healthcare resource utilization (HCRU) among Medicaid enrollees with ASD or ID diagnoses longitudinally across the transition to adulthood, from 2015 to 2019. Notable patterns emerged from the analyses, including overall higher indicators of HCRU among the two ID groups (ID only and ASD + ID); especially high indicators of HCRU for psychiatric outpatient care among the ASD + ID group in adulthood; and overall declines in service utilization rates and expenditures across the transition to adulthood with two exceptions. First, we found increasing utilization rates for inpatient psychiatric care. Second, we found increasing per capita expenditures for outpatient and long-term care, driven by substantially increasing per capita expenditures among those with >$0 expenditures, which more than offset declining utilization rates.

This study replicated and extended a prior analysis of the 2001–2005 Medicaid claims data (Shea et al., 2018). There were both similarities and differences when comparing across studies. For example, a key finding from the 2001–2005 data was a decline in psychiatric outpatient service utilization rates across the transition to adulthood for those with ASD or ID diagnoses, accompanied by increasing average expenditures among those utilizing services. This same pattern emerged in the 2015–2019 data. In the 2001–2005 data, the ASD only group tended to have higher utilization rates than the ID only group across psychiatric care types (except for long-term care), while the ID group had greater utilization rates for medical care. In 2015–2019, the ID groups tended to have greater utilization rates across both psychiatric and medical care. A likely explanation for this change over time is the evolution of both ASD and Medicaid. Changes to the ASD diagnostic criteria, increasing numbers of individuals seeking diagnoses (especially in adulthood), and expansion of Medicaid eligibility pathways for those with an ASD diagnosis (e.g., HCBS waivers; general Medicaid expansion) have increased both the prevalence of individuals with an ASD diagnosis and the proportion of these individuals enrolled in Medicaid. This period effect was evident when comparing the sample sizes for our analysis to the earlier study; in 2001–2005 there were 1,637 individuals included in the ASD sample; in 2015–2019 there were 23,460. This increase is consistent with prior work documenting increasing ASD prevalence in Medicaid over time (Rubenstein & Bishop, 2019; Rubenstein et al., 2023). Numbers for ID only also increased, although to a lesser extent. It is also possible these changes resulted from less intensive healthcare support needs in the ASD Medicaid population over time. For example, a prior study of the Wisconsin Medicaid population from 2008 to 2018 found increasing prevalence of ASD in Medicaid claims over time, but declines in number of healthcare visits and amounts paid per year in this population (Rubenstein & Bishop, 2019). Consistent with this interpretation, in comparison to 2001–2005 the 2015–2019 data showed substantially lower utilization rates for psychiatric care across types for the ASD only group. It is also possible that changes in Medicaid eligibility over time have resulted in more ASD only youth and young adults receiving care through managed care or private insurance. Long-term psychiatric care utilization rates decreased substantially between the 2001–2005 and 2015–2019 data, potentially reflecting an overall trend towards de-institutionalization of disabled adults in the U.S. (Larson et al., 2022), and greater access to Home and Community-Based Services (Shea et al., 2021) through Medicaid.

A concerning finding in the 2015–2019 data is that the average per capita cost of psychiatric outpatient care increased substantially for those utilizing services as youth transitioned to adulthood, for all groups and most dramatically for the ASD + ID group. This pattern resulted in overall increases in per capita resource utilization, despite a smaller proportion utilizing services. One possible explanation of this trend is the noted shift away from long-term care for Medicaid-enrolled adults with ID or ASD, resulting in individuals with more intensive mental health care needs utilizing outpatient services. The likelihood that mental health concerns are identified or addressed in the primary care setting has increased significantly over time (Rotenstein et al., 2023). As a result, specialist psychiatric care providers may have seen a proportional increase in visits for more significant mental health concerns, resulting in increased median spending per patient per year. Additional speculative interpretations include greater portions seeking less intensive care through managed care structures or private insurance in adulthood, or increasing expense of adult compared to child psychiatric care.

For both psychiatric and medical care, those with an ID diagnosis (only or with ASD) tended to have higher utilization rates, and utilizers had greater mean expenditures, in comparison to those with an ASD diagnosis only. These differences may reflect more intensive care needs or more expensive care provision among those with an ID diagnosis. For example, autistic adults with an ID diagnosis may have substantially greater rates of hospital readmission than adults with an autism diagnosis alone (Rast et al. 2025a, b). Specialized inpatient clinics for individuals with ID may be more expensive, and those who receive care in these specialized settings tend to have more complex care needs (Melvin et al., 2022).

Both medication utilization rates and per capita expenditures declined across the transition to adulthood for all three groups, for both psychiatric and medical care. These results contrast with prior findings of increasing medication use across the transition to adulthood among autistic youth and young adults (Davis et al. 2023; Esbensen et al. 2009; Rast et al. 2025a, b; Shea et al. 2019). General declines in medication prescriptions in the Medicaid population or changes in available prescription medications may have contributed to our findings; however, prior studies have still found increased medication use across the transition to adulthood for autistic patients despite these period effects (Davis et al., 2023). It is also possible that increased understanding of the need for behavioral supports as a front line treatment have reduced psychiatric medication use across the transition to adulthood for the Medicaid population; however, we saw both declines in psychiatric and non-psychiatric medication use across all three cohorts. Additional untested hypotheses include greater portions of adults receiving medications through alternative reimbursement sources (e.g., managed care or private insurance), and fewer seeking pharmaceutical care. Considering the deviation from prior findings, these novel trends warrant further exploration from multidisciplinary teams with expertise in Medicaid policy governing medication availability and formulary changes and prescribers.

Additional research has examined healthcare costs and utilization for autistic youth across the transition to adulthood beyond Medicaid. Ames and colleagues reported a cross-sectional study of youth aged 14–25 years old using Kaiser Permanente claims data from 2014 to 2015, which includes individuals across insurers in one region of California (Ames et al., 2021). Autistic youth had higher rates of psychiatric care compared to those with attention deficit hyperactivity disorder and diabetes mellitus across age groups. Notably, outpatient psychiatric care utilization was substantially lower in this private payer population than our estimates from Medicaid enrolled autistic youth. It is unclear whether rates of mental health conditions among autistic youth are more prevalent for those enrolled in Medicaid versus private insurance, although this may be true considering many autistic youth are enrolled in Medicaid through income-based eligibility, and socioeconomic status is broadly associated with mental health conditions (McLaughlin et al., 2012). Further, autistic youth enrolled through disability-related eligibility mechanisms may have generally more intensive support needs than those who are not enrolled, due to level-of-support criteria utilized in determining eligibility. It is also possible that increased access to mental and behavioral health services for those enrolled in Medicaid drives greater rates of diagnosis and service utilization. Additional research has found that among privately insured autistic individuals during 2000–2013, overall healthcare utilization rates declined across the transition to adulthood (Nathenson & Zablotsky, 2017). We similarly found declines in utilization rates, except for increases in inpatient psychiatric care utilization.

Limitations of our research include that we did not directly test hypothesized mechanisms that may have driven the patterns we observed, as well as differences between our study and Shea and colleagues’ earlier parallel analysis. We interpreted longitudinal changes in healthcare utilization and expenditures from 2015 to 2019 to result from the changing age of enrollees across this time-period as they transitioned to adulthood (i.e., age effects). It is possible that period effects that generally influenced Medicaid-funded healthcare from 2015 to 2019 contributed to the patterns that we presented, independent of age; therefore, our findings may not generalize to transition-aged youth beyond the time-period that we studied. We only analyzed claims that were fee for service or primary care case management (PCCM; paid via fee for service) due to varying completeness and quality of comprehensive managed care (CMC) Medicaid claims over our study period. The utilization of managed care through Medicaid has expanded from 2001–2005 to 2015–2019, potentially impacting comparisons to Shea and colleagues’ earlier analysis, and limiting the generalizability of our findings. The majority of national Medicaid beneficiaries are currently enrolled through comprehensive managed care (CMC), although rates vary widely state-to-state. Enrollment through CMC is generally substantially lower for those eligible for Medicaid through disability status, potentially limiting impact on the generalizability of our findings. Historically, those with more complex medical needs are more likely to be enrolled through FFS or PCCM, potentially skewing our findings towards more intensive care needs, although in recent years more states are using comprehensive managed care to address complex medical needs among Medicaid enrollees. It will be important for future work to compare FFS versus managed care utilization and expenditures in this population, especially as the quality of managed care data continues to improve and the number of post-COVID-19 years with available data increases. During the study period (2015–2919) Medicaid claims data converted from the MAX to T-MSIS (TAF) system, potentially influencing longitudinal comparisons between 2015 and 2019 data. We limited our years of analysis to 2015–2019 to avoid period effects resulting from the COVID-19 pandemic; it is possible that more updated analyses would demonstrate different trends. We calculated mean expenditure estimates for all within a cohort, as well as the subset that excluded individuals with zero expenditures. We distinguished between types of care, including psychiatric and non-psychiatric care, using the data structure provided by CMS and standard codes; however, there remains the possibility of misclassification.

Consistent with prior analyses of 2001–2005 Medicaid claims data, we continued to identify substantial changes in Medicaid healthcare utilization across the transition to adulthood for youth and young adults with ASD or ID diagnoses. From 2015 to 2019, these changes were marked by greater resource utilization for outpatient and long-term care, driven by greater per capita costs among care utilizers; and reductions in medication and inpatient care resource utilization, driven by both lower utilization rates and lower per capita expenditures among utilizers in adulthood. Further work can help elucidate forces driving these changes, including changes in healthcare funding across the transition to adulthood (e.g., transition from fee for service to comprehensive managed care; acquisition of private insurance sources); changes in care need; changes in care access; and changes in cost of care. There have been extensive efforts to use Medicaid as a mechanism to improve access to care generally and specifically for youth with developmental disabilities, as well as efforts to reduce needs for long-term care utilization over time. In comparison to prior analyses from 2001 to 2005 and consistent with prior work, our findings demonstrate increases in youth and young adults with developmental disability diagnoses in the Medicaid population, especially for ASD. We also found a marked decline in utilization rates for psychiatric long-term care in comparison to the earlier data, potentially representing the impact of de-institutionalization. Further research directly testing these speculative mechanistic interpretations and hypotheses will more fully inform understanding of the reach and impact of relevant policies and programs.