Using administrative claims data, this study derived the prevalence of diagnosed mental health (MH) conditions in Medicaid-insured and commercially-insured children, for all MH conditions, and separately for attention deficit hyperactivity disorder (ADHD), depression disorders, and anxiety disorders, at census tract and state levels. The prevalence reflected current diagnosed-MH conditions, not capturing MH conditions not diagnosed within the corresponding health insurance program [15]. The study applied a two-step process to generate census tract-level rates: geographic projection from zip codes using population-weighted imputation, followed by spatial smoothing based on tract similarity. Statistical analyses included two-way ANOVA to assess state and rurality differences, and identification of high- and low-prevalence communities using quartile thresholds and confidence bands. This approach enabled fine-grained geographic comparisons across insurance types and MH conditions.
This study population consisted of over 32 million Medicaid-insured and 6 million commercially-insured children aged 3–17 across the U.S. The demographic similarity between Medicaid and commercial populations by age, sex, and geographic distribution (eTable C7) supports the prevalence comparisons between insurance types, as differences in diagnosed MH conditions are less likely to be confounded by these demographic factors. Medicaid-insured children showed significantly higher diagnosed MH prevalence than those with commercial insurance across nearly all states. ADHD, depressive disorders, and anxiety disorders followed similar patterns. Suburban areas consistently had higher diagnosed MH rates than rural areas, particularly in the Medicaid population. Statistical analysis revealed substantial variation in prevalence across states and census tracts, with notable disparities in both median rates and variability. High-prevalence and low-prevalence census tracts were identified using quartile thresholds, showing geographic clustering and differences by urbanicity. These findings highlight the importance of insurance type and location in shaping access to mental health diagnosis and care.
Overall, the national-level prevalence for diagnosed-MH conditions was 11.8% and 5.5% for Medicaid-insured and commercially-insured children, respectively, as compared to 20% population prevalence from a 2009 report [33], or 16.5% parental-reported prevalence from the 2016 National Survey of Children’s Health [34]. A review study covering children from high-income countries reported prevalence of 12.7% for children from high-income countries [35]. These studies reported higher overall prevalence. The differences in the estimates were due to the approach for identifying MH diagnosis (experienced vs. self-reported versus MH-provider diagnosed), the use of datasets (survey vs. medical claims records), the sampling approach (sampled vs. whole population) and the study population (age group, insurance status, residence). Survey data capture symptoms or impairment rather than confirmed diagnoses. In contrast, administrative claims data reflect both diagnosis and reimbursed treatment, offering a more direct measure of healthcare utilization. Moreover, estimates using administrative claims records are potentially biased due to access barriers [15]. This distinction is particularly important for commercial insurance, where families often pay out-of-pocket for MH services, leading to underrepresentation in claims data. Such out-of-pocket care is less common in Medicaid, making comparisons between insurance types critical for understanding disparities in access and service delivery.
For Medicaid-insured children, the diagnosed condition-specific prevalence was ADHD: 5.5%; depression: 2.7%; and anxiety: 0.9%, in contrast to the self-reported prevalence of ADHD: 8.7%; depression: 3.4%; and anxiety: 7.8% estimated using the 2016 − 2019 National Survey of Children’s Health (NSCH) [2]. A review study covering children from high-income countries reported prevalence for ADHD: 3.7% and anxiety: 5.2% [35]. The condition-specific prevalence for commercially-insured children in our study was lower than for the Medicaid-insured population, ADHD: 2.3%, depression: 0.2%, and anxiety: 1.1%. However, they were in line with estimates from a WHO study, reporting ADHD prevalence of 3.1% in 10–14 year-olds and 2.4% in 15–19 year-olds, and depression prevalence of 1.1% in 10–14 years, and 2.8% in 15–19 year-olds [36]. The comparison for individual conditions is more nuanced because of different diagnosis practices across countries. For example, ADHD was diagnosed at a higher rate in the Medicaid population than other populations (including those from other high-income countries).
The commercial MH-diagnosed prevalence was much lower than population or self-reported prevalence, potentially indicating the limited access to MH services available for some children reimbursed with commercial insurance, deemed as under-insured due to limited reimbursement of the full range of diagnostic codes outlined in the established clinical guidelines [37, 38].
We observed a consistent pattern of higher prevalence in suburban communities versus urban or rural communities in about 32 states for the Medicaid-insured population. This pattern, however, was found in only 14 states for the commercially-insured population. A similar pattern was also observed in a recent CDC study, which estimated overall higher prevalence in rural communities for ADHD, depression, and anxiety [2]. MH conditions were being diagnosed for Medicaid-insured children in both rural and urban communities, however commercially-insured children from rural communities had lower rates of MH diagnosis than those from urban communities, suggesting that children from rural communities with commercial insurance were less likely to be diagnosed with MH conditions, which in turn, could be due to less effective MH care access in rural settings.
The state-level estimates from our study were also widely different from the recent CDC study [2]. For example, our study estimated a lower MH prevalence in Florida (Medicaid-2.9%, commercial-4.6%), whereas the CDC reported a prevalence of 9.7%. Our study also estimated higher prevalence for the Medicaid-insured population than those in the CDC study in 41 states, with particularly large discrepancies observed in Vermont (Medicaid-30.9%, CDC-16.4%) and Maine (Medicaid-27.5%, CDC-12.9%). Noteworthy, our estimates consistently showed higher prevalence in the Medicaid-insured population than in the commercially-insured population across 49 states. These discrepancies were the largest in states like Maine and Vermont.
While there was not a consistent pattern in the prevalence estimates from our study for the Medicaid-insured and commercially-insured populations versus those provided by the CDC study, both studies identified wide variations between states. The high prevalence maps displayed a pronounced pattern, with many more high prevalence tracts in the east states, some in the northwest border for Medicaid-insured children only and almost none from the mid-country states. The low prevalence maps instead displayed a few states with low prevalence, but those states had most of the low prevalence communities. Practically, there were high prevalence states vs. low prevalence states, with very few in between. This extreme variation was not an indication of the children’s experience with MH conditions but instead state-level practices and policies, which play a role in healthcare utilization [34, 39, 40]. States administered their healthcare systems differently, some state Medicaid programs carving out MH services [41], coordinated at the county or state level, with potentially implications on the availability of MH providers [42]. While the estimates for the commercially-insured population were from the same commercial insurance program, hence practicing similar reimbursement and administration policies, the between and within-state variations were also large, impacted by the provider network available to treat children. Thus, the estimates in this study reflected healthcare delivery system barriers to mental healthcare utilization.
Our findings must be considered within the context of study limitations. This study was conducted using administrative claims data. Claims data may not fully capture MH diagnosis since these data were designed for billing purposes. While the DQ Atlas [20] indicated no major concerns with diagnosis data quality across all states, other dimensions of data quality could impact our prevalence estimates, including potential issues with eligibility and enrollment data accuracy, variations in claims submission practices, differences in encounter reporting completeness across states and providers, and inaccuracies in enrollee information that may affect diagnostic accuracy. The Medicaid claims data also included only claims that have been submitted for reimbursement and they only allow estimation of diagnosed MH conditions. Therefore, prevalence estimates may be biased where certain subgroups have difficulty in maintaining Medicaid coverage, lack access to care, or were susceptible to particularly disparate utilization. A small proportion of Medicaid-insured children might also have dual insurance, primarily those enrolled due to special healthcare needs or disability [43, 44], however most of them are using Medicaid MH services, thus not significantly impacting our results. By studying a maximum of one year of a child’s healthcare, prevalence could be underrepresented for those who sought care over prior years. The most recent years of Medicaid claims data may provide different estimates, particularly in the context of the recent pandemic and societal disruption, greatly impacting children’s mental health.
Additionally, the inclusion of children with partial-year enrollment enhances generalizability by reflecting real-world insurance coverage patterns. While enrollment duration differed between insurance types (Medicaid: 10.8 months; Commercial: 8.9 months), our prevalence estimates capture diagnosed mental health conditions among all enrolled children, providing a more comprehensive assessment than studies restricted to full-year enrollees.
link