跳到正文
原文
Nature 健康护理主题·· 28 天前AI 评分40

美国精神卫生治疗系统的互操作性与功能性数字成熟度

Beyond Digitization: Interoperability and Functional Digital Maturity Across U.S. Behavioral Health Treatment Systems

AI 导读

基于2024年国家物质使用与心理健康服务调查的27,957家设施数据,美国精神卫生系统数字成熟度存在显著差异。50.7%的设施仅使用电子病历,但HIO参与度低(13.2%),且私立非营利设施混合使用电子与纸质记录的比例更高(RRR 1.22)。

正文

Introduction

The rapid expansion of electronic health records (EHRs) over the past two decades has transformed health care delivery in the United States, driven in part by federal investments through the Health Information Technology for Economic and Clinical Health (HITECH) Act1 enacted in 2009 as part of the American Recovery and Reinvestment Act (ARRA) and broader national interoperability initiatives. Although national policy efforts have accelerated the adoption of digital health technologies, concerns remain regarding fragmentation in interoperability, variation in functional capability, and unequal distribution of digital infrastructure across health care sectors2 Prior studies have shown that measures of “meaningful use” and EHR functionality vary considerably across organizations, particularly among safety-net and resource-constrained providers2,3,4,5. EHR systems differ substantially in the range of clinical and administrative functions implemented, including decision support, medication management, and patient engagement tools3. Adoption alone does not ensure interoperability or effective exchange of clinical information, reinforcing the distinction between availability of digital systems and mature integration into care delivery workflows2,5. Internationally, health systems have increasingly recognized that digital transformation extends beyond implementation of electronic records alone. Digital maturity frameworks developed across health systems emphasize progression from foundational digitization toward interoperability, integrated information exchange, and advanced data use capabilities6,7. The WHO Global Strategy on Digital Health and WHO–ITU National eHealth Strategy Toolkit further emphasize that this progression depends on enabling conditions beyond technology itself, including governance, sustained investment, infrastructure, interoperability standards, workforce capacity, and organizational readiness6,8. These perspectives are consistent with evidence that organizational and implementation contexts strongly influence whether health information technologies become effectively integrated into routine care7. Together, international experience suggests that digital maturity is better understood as a multidimensional health-system capability than as the adoption of any single technology.

Despite some advances, behavioral health systems have historically lagged behind other sectors of health care in digital infrastructure, interoperability, and participation in health information exchange organizations (HIOs), in part because behavioral health providers were less likely to benefit from early federal health information technology incentive programs and often operate within fragmented financing and regulatory environments6,7. Recent policy efforts by the Office of the National Coordinator for Health Information Technology (ONC) have increasingly emphasized interoperability, functional capability, and exchange readiness as central dimensions of digital health system maturity8. Behavioral health information exchange also occurs within a distinct regulatory environment, as certain substance use treatment records are subject to additional confidentiality protections under 42 CFR Part 2, alongside broader protections for protected health information. These requirements create important considerations for designing interoperable systems that support appropriate data sharing while maintaining patient privacy.

Using nationally representative facility-level data from the 2024 National Substance Use and Mental Health Services Survey (N-SUMHSS), we examined digital maturity in U.S. behavioral health systems using a multidimensional framework integrating three domains: EHR adoption, interoperability through participation in HIOs, and functional EHR capability. Building on existing descriptive summaries of behavioral health technology adoption, this study constructs a facility-level EHR capability index and applies multivariable analyses to evaluate organizational, financing, licensing, and facility-level factors associated with variation in digital maturity.

Results

A total of 27,957 behavioral health treatment facilities were included in the 2024 N-SUMHSS. Private non-profit facilities represented the largest ownership category (39.5%), followed by private for-profit facilities (29.0%) (Table 1). Combined mental health and substance use treatment facilities accounted for 33.2% of facilities. Table 1 also shows that telehealth services were common, particularly for substance use treatment (46.6%). Approximately 61.7% of facilities accepted Medicaid, and 41.7% accepted Medicare. Forty percent of facilities reported three or more licenses or accreditations. Half of the facilities reported exclusive use of electronic health records (50.7%), whereas 19.1% continued to use both EHR and paper records. Active participation in HIOs was low at 13.2%, and nearly one in five facilities reported not knowing whether an HIO was available in their area. Additional cross-tabulation analyses demonstrated significant variation in facility characteristics by ownership and treatment focus (Supplement Materials Tables 1 and 2). Ownership type and treatment focus differed by Medicaid and Medicare participation, licensing/accreditation intensity, and service integration patterns (all χ² p < 0.001).

Table 1 Characteristics of behavioral health treatment facilities and digital maturity indicators, N-SUMHSS 2024

Full size table

Multivariable Regression Results

Table 2 presents factors associated with EHR adoption, using facilities with exclusive EHR use and no paper records as the reference group. Compared with private for-profit facilities, private non-profit facilities were more likely to report using both EHR and paper records (RRR 1.22, 95% CI 1.10–1.36), whereas state government facilities demonstrated substantially higher likelihood of using both EHR and paper records (RRR 2.50, 95% CI 1.86–3.36). In contrast, federal facilities were substantially less likely to report using both EHR and paper records (RRR 0.12, 95% CI 0.06–0.26), and very few reported the absence of EHR systems. Compared with substance use–only facilities, combined mental health and substance use facilities were less likely to report using both EHR and paper records (RRR 0.61, 95% CI 0.53–0.69) and less likely to report having no EHR and no implementation plans (RRR 0.60, 95% CI 0.42–0.87). Facilities offering substance use telehealth services were also less likely to report use of both EHR and paper records (RRR 0.58, 95% CI 0.48–0.71), no EHR with implementation plans (RRR 0.31, 95% CI 0.21–0.47), or no EHR without implementation plans (RRR 0.38, 95% CI 0.25–0.57). Facilities accepting Medicare were consistently less likely to report non-EHR categories, including use of both EHR and paper records (RRR 0.55, 95% CI 0.49–0.61), no EHR with implementation plans (RRR 0.44, 95% CI 0.33–0.59), and no EHR without implementation plans (RRR 0.38, 95% CI 0.27–0.54). OTP-certified facilities were more likely to report using both EHR and paper records (RRR 1.37, 95% CI 1.20–1.57) but less likely to report having no EHR and no implementation plans (RRR 0.36, 95% CI 0.20–0.66). Overall, these findings suggest that facilities with greater organizational integration—including combined mental health and substance use treatment facilities, telehealth-enabled organizations, and facilities participating in Medicare—demonstrated more advanced EHR adoption, whereas substance use–only facilities were more likely to rely on less integrated digital systems.

Table 2 Multinomial logistic regression of factors associated with EHR adoption among behavioral health treatment facilities, N-SUMHSS 2024

Full size table

Table 3 presents factors associated with HIO participation, using facilities actively participating in HIOs as the reference group. Compared with private for-profit facilities, local or county government facilities were more likely to report that HIOs were available but that they were not participating (RRR 1.66, 95% CI 1.19–2.33), while tribal government facilities were substantially more likely to report non-participation despite HIO availability (RRR 3.32, 95% CI 1.42–7.77). In contrast, federal facilities were substantially less likely to report HIO non-participation (RRR 0.15, 95% CI 0.05–0.41), lack of HIO availability (RRR 0.10, 95% CI 0.04–0.28), or unfamiliarity with HIOs (RRR 0.18, 95% CI 0.11–0.28). Compared with substance use–only facilities, combined mental health and substance use facilities were less likely to report HIO non-participation despite availability (RRR 0.56, 95% CI 0.43–0.72), lack of HIO availability (RRR 0.64, 95% CI 0.49–0.83), or unfamiliarity with HIOs (RRR 0.46, 95% CI 0.38–0.56), suggesting stronger interoperability participation among integrated treatment facilities. Mental health–only facilities, however, were more likely to report that HIOs were unavailable in their area (RRR 2.60, 95% CI 1.55–4.35). Facilities accepting Medicare were consistently less likely to report barriers to HIO participation, including HIO non-participation (RRR 0.59, 95% CI 0.49–0.71), lack of HIO availability (RRR 0.54, 95% CI 0.44–0.66), and unfamiliarity with HIOs (RRR 0.43, 95% CI 0.37–0.50). FQHCs were also less likely to report HIO non-participation (RRR 0.26, 95% CI 0.18–0.37) or lack of HIO availability (RRR 0.48, 95% CI 0.35–0.65). Overall, these findings suggest that integrated behavioral health facilities, Medicare participation, and stronger organizational oversight were associated with greater interoperability participation, whereas substance use–only and tribal facilities demonstrated comparatively lower levels of HIO integration.

Table 3 Multinomial logistic regression of factors associated with HIO participation status among behavioral health treatment facilities, N-SUMHSS 2024

Full size table

Table 4 presents factors associated with EHR capability, using facilities with high EHR capability as the reference group. Compared with private for-profit facilities, private non-profit facilities were more likely to demonstrate low capability (RRR 1.36, 95% CI 1.20–1.53) and moderate capability (RRR 1.15, 95% CI 1.04–1.26). Tribal government facilities also demonstrated substantially higher likelihood of low capability (RRR 3.60, 95% CI 2.25–5.76), whereas federal facilities were substantially less likely to report low capability (RRR 0.07, 95% CI 0.03–0.17) or moderate capability (RRR 0.27, 95% CI 0.19–0.38). Compared with substance use–only facilities, combined mental health and substance use facilities were less likely to demonstrate low capability (RRR 0.51, 95% CI 0.43–0.60) or moderate capability (RRR 0.59, 95% CI 0.51–0.67), suggesting greater functional digital maturity among more integrated treatment organizations. Mental health–only facilities were also less likely to demonstrate moderate capability relative to high capability (RRR 0.64, 95% CI 0.47–0.88). Facilities accepting Medicaid were more likely to demonstrate low capability (RRR 1.98, 95% CI 1.69–2.31) and moderate capability (RRR 1.62, 95% CI 1.43–1.84), whereas facilities accepting Medicare were less likely to demonstrate low capability (RRR 0.44, 95% CI 0.39–0.50) or moderate capability (RRR 0.81, 95% CI 0.73–0.90). FQHCs were also substantially less likely to report low capability (RRR 0.32, 95% CI 0.26–0.38) or moderate capability (RRR 0.42, 95% CI 0.37–0.48), indicating higher levels of functional digital maturity among federally supported safety-net organizations. Facilities with three or more licenses or accreditations were more likely to demonstrate moderate capability relative to high capability (RRR 1.74, 95% CI 1.43–2.12). Overall, these findings suggest that integrated treatment settings, Medicare participation, and federally supported organizational infrastructure were associated with greater functional EHR capability, whereas substance use–only and tribal facilities demonstrated comparatively lower levels of digital maturity. Internal consistency of the 22 EHR functionality items was assessed using Cronbach’s alpha. The capability items demonstrated good reliability (Cronbach’s α = 0.84), supporting aggregation of individual EHR functions into a composite facility-level capability index .

Table 4 Multinomial logistic regression of factors associated with EHR capability among behavioral health treatment facilities, N-SUMHSS 2024

Full size table

Figure 1 demonstrates substantial geographic variation in digital maturity across behavioral health treatment systems in the United States. Higher proportions of low EHR capability were observed in several Mountain West and Northern Plains states, including Alaska, South Dakota, Utah, and Maine, whereas lower levels were observed across portions of the Mid-Atlantic and southeastern regions. Moderate EHR capability was more common in several Midwestern and southern states, including Arkansas, Hawaii, South Carolina, and Oklahoma. High EHR capability was concentrated in portions of the Northeast, Mid-Atlantic, and selected southeastern states, with Puerto Rico, Maryland, North Carolina, Delaware, and Louisiana demonstrating some of the highest proportions of high-capability facilities. Overall, these findings suggest that digital maturity across behavioral health systems is unevenly distributed nationally and may reflect broader differences in infrastructure investment, organizational integration, and state-level health information technology capacity.

Fig. 1: State-level variation in EHR functional capability among U.S. behavioral health treatment facilities.

Maps show the percentage of facilities within each state classified as having low EHR capability (top left), moderate EHR capability (top right), and high EHR capability (bottom). Darker shading indicates a higher percentage of facilities within the respective capability category. Alaska (AK), Hawaii (HI), the District of Columbia (DC), Puerto Rico (PR), and other U.S. territories are presented separately with their corresponding percentages of facilities in each capability category.

Discussion

In this national analysis of behavioral health treatment facilities in the United States, we found substantial variation across multiple dimensions of digital maturity, including EHR adoption, interoperability through HIO participation, and functional EHR capability. Although approximately half of facilities reported exclusive use of EHR systems, active participation in HIOs remained limited, and meaningful differences persisted in advanced EHR functionality across organizational settings. These findings suggest that implementation of EHR systems alone does not necessarily translate into interoperability, advanced digital capability, or integration into broader health information exchange ecosystems.

The gap between EHR adoption and interoperability has important implications for behavioral health care delivery. Facilities lacking interoperability or advanced EHR functionality may face challenges coordinating care, exchanging clinical information, integrating with broader health systems, monitoring medications, and supporting continuity of care across providers and settings. Nearly one-fifth of facilities reported unfamiliarity with HIO availability, suggesting that barriers may reflect not only infrastructure limitations but also limited technical readiness, organizational fragmentation, and lack of integration into regional health information exchange systems.

Organizational characteristics demonstrated strong and consistent associations across all three domains of digital maturity. Federal facilities and federally qualified health centers (FQHCs) generally demonstrated greater digital integration, including lower likelihood of limited EHR adoption, low EHR capability, and HIO non-participation. In contrast, tribal facilities demonstrated substantially higher likelihood of low EHR capability and HIO non-participation despite HIO availability, highlighting potential inequities in access to digital infrastructure and interoperability resources. Combined mental health and substance use treatment facilities were consistently more digitally mature than substance use–only facilities across adoption, interoperability, and capability measures, suggesting that more integrated treatment settings may benefit from broader financing structures, organizational scale, and participation in larger health system networks.

Telehealth use was associated with more advanced EHR adoption, particularly among facilities providing substance use telehealth services, consistent with broader digital transformation observed during and after the COVID-19 pandemic. However, telehealth integration demonstrated weaker associations with interoperability and advanced EHR capability, suggesting that adoption of remote care technologies alone may not ensure broader digital integration. Financing participation also demonstrated complex relationships with digital maturity. Medicare participation was consistently associated with greater interoperability and EHR capability, whereas Medicaid participation was associated with lower interoperability barriers but higher likelihood of low or moderate EHR capability. These findings may reflect differences between participation in financing systems and organizational capacity to support advanced digital infrastructure.

Facilities with multiple licenses or accreditations were less likely to report unfamiliarity with HIOs, suggesting greater engagement with reporting and oversight systems. However, facilities with greater licensing complexity also demonstrated higher likelihood of low and moderate EHR capability relative to high capability, indicating that regulatory oversight alone may not ensure advanced digital functionality.

Geographic analyses further demonstrate substantial state-level heterogeneity in digital maturity across behavioral health systems. Higher proportions of low-capability facilities were observed across several western and northern states, whereas higher proportions of high-capability facilities were concentrated across portions of the Northeast, Mid-Atlantic, and selected southeastern states. These findings suggest that behavioral health digital maturity may also reflect broader differences in infrastructure investment, interoperability ecosystems, workforce capacity, and state-level health information technology integration.

Our findings have important implications for ongoing federal behavioral health modernization and interoperability initiatives. Although national policy has increasingly emphasized interoperable health information exchange through the 21st Century Cures Act, the Trusted Exchange Framework and Common Agreement (TEFCA), the United States Core Data for Interoperability (USCDI), and related behavioral health interoperability efforts, our results suggest that these objectives are unlikely to be achieved through EHR adoption or technical interoperability standards alone. Facilities with greater organizational integration, including federal facilities and federally qualified health centers, consistently demonstrated higher levels of interoperability participation and functional EHR capability, whereas substantial geographic variation and lower digital maturity among many other facilities indicate uneven organizational readiness across behavioral health systems. Because a large proportion of behavioral health services are financed, regulated, and administered through state systems, achieving nationwide interoperability will require coordinated federal and state investment in digital infrastructure, workforce technical capacity, organizational readiness, and participation in health information exchange networks. Without strengthening these foundational capabilities, many behavioral health providers may remain unable to fully participate in interoperable health information ecosystems despite widespread EHR adoption, limiting the potential benefits of national interoperability initiatives for coordinated care and information sharing. Future behavioral health modernization strategies should therefore extend beyond promoting EHR adoption to strengthening the organizational and technical capacity required for meaningful interoperability across federal, state, and community behavioral health systems. Supplementary Material Table 3 translates these findings into policy and implementation priorities.

The following are the limitations of the analysis. First, analyses were cross-sectional and cannot establish causal relationships between organizational characteristics and digital maturity outcomes. Second, because the survey uses modular administration with substantial logical skips and outcome-specific eligibility, analytic denominators varied across measures. Although outcome-specific complete-case approaches were used, residual selection effects may remain. Findings for smaller ownership categories, including tribal and federal facilities, should be interpreted cautiously because these groups represented a smaller proportion of facilities in N-SUMHSS. Observed differences may reflect broader organizational, jurisdictional, and infrastructure contexts and should not be interpreted as differences attributable solely to ownership status. Finally, geographic analyses were descriptive and not intended to support causal inference regarding state policy environments or infrastructure investments. Geographic differences in digital maturity should be interpreted within the broader context of jurisdictional variation in health information technology infrastructure, information-sharing environments, regulatory frameworks, and historical investments in digital modernization. Future research may evaluate how these state-level and regional factors influence interoperability participation and functional digital capability across behavioral health systems. N-SUMHSS does not directly measure coding and terminology standards, semantic interoperability, HL7 FHIR, APIs, or transaction-level information exchange. Limitations also include that the digital maturity construct does not comprehensively capture patient-facing functions such as portal access, multilingual functionality, or self-scheduling.

Despite these limitations, this study provides one of the first national evaluations of digital maturity across behavioral health treatment systems in the United States using a comprehensive facility-level census. By simultaneously examining EHR adoption, interoperability participation, and functional capability, the findings demonstrate that digital maturity in behavioral health systems is multidimensional and unevenly distributed across organizational settings. As behavioral health systems become increasingly integrated into broader health care delivery and data exchange ecosystems, policies focused solely on EHR adoption are unlikely to achieve meaningful digital integration without parallel investment in interoperability infrastructure and advanced functional capability.

Methods

We conducted a cross-sectional analysis using data from the 2024 National Substance Use and Mental Health Services Survey (N-SUMHSS), administered annually by the Substance Abuse and Mental Health Services Administration (SAMHSA) as part of the Behavioral Health Services Information System. N-SUMHSS is a national census-based survey of publicly and privately operated behavioral health treatment facilities in the United States and its territories and serves as the primary federal system for enumerating treatment facilities and monitoring behavioral health service infrastructure. Facilities included in N-SUMHSS are derived from SAMHSA’s National Inventory of Treatment Facilities (I-TF), which is maintained through state behavioral health reporting partnerships and serves as both the operational survey frame for N-SUMHSS and the federal FindTreatment.gov treatment locator system. Because N-SUMHSS functions as both a national facility census and a public treatment locator, participating facilities may elect to be included or updated within FindTreatment.gov. The 2024 survey included 27,957 facilities across all states, the District of Columbia, Puerto Rico, and U.S. territories and collected information on organizational characteristics, services offered, financing, licensing, and health information technology infrastructure. The survey uses a modular design in which selected questions are administered only to facilities eligible for specific service domains, including substance use treatment, mental health treatment, or combined services, resulting in outcome-specific analytic denominators. All analyses were conducted using de-identified publicly available data and were exempt from institutional review board oversight.

Data and statistical analysis

Digital maturity was conceptualized as comprising three related but distinct domains: electronic health record (EHR) adoption, interoperability participation, and functional EHR capability. EHR adoption reflects whether facilities have implemented electronic health record systems and the extent to which paper records continue to be used. Interoperability participation reflects engagement in broader health information exchange infrastructure through participation in health information exchange organizations (HIOs), distinguishing implementation of electronic systems from active information exchange. Functional EHR capability reflects the implementation of advanced clinical and administrative EHR functions, including clinical documentation, treatment planning, medication management, laboratory integration, clinical decision support, secure messaging, and patient engagement tools. Together, these domains represent progression from foundational digitization toward increasingly integrated and functionally mature digital systems. Although N-SUMHSS does not directly measure standards-based interoperability technologies such as HL7 Fast Healthcare Interoperability Resources (FHIR), application programming interfaces, semantic interoperability, or real-time bidirectional exchange, these measures align conceptually with national digital maturity frameworks that emphasize advancement from basic EHR adoption to interoperable and functionally integrated health information systems.

EHR adoption was measured using survey item EHR1 and categorized into four mutually exclusive groups: exclusive EHR use, combined EHR and paper records, no EHR with plans to implement an EHR, and no EHR with no implementation plans. Interoperability participation was measured using survey item EHR6 and categorized as active participation in an HIO, HIO available but not participating, no HIO available, or unfamiliar with HIO availability. Functional EHR capability was derived from 22 survey items (EHR8_1–EHR8_19 and EHR10_1–EHR10_3) representing clinical documentation, information management, reporting, communication, and electronic exchange functions.

A facility-level capability score was calculated as the proportion of available EHR functions among all eligible EHR functionality items in Eq. 1:

$${{\mbox{EHR}}{\mbox{Capability}}{\mbox{Score}}}_{i}=\frac{\mathop{\sum }_{j=1}^{22}{EH}{R}_{{ij}}}{\mathop{\sum }_{j=1}^{22}{Eligibl}{e}_{{ij}}}$$

(1)

where \({EH}{R}_{{ij}}=1\) if facility i reported implementation of EHR function j and 0 otherwise, and \({Eligibl}{e}_{{ij}}=1\) if facility i had a valid response for function j and 0 for logical skips, missing values, or not applicable responses. Consequently, only eligible EHR functions contributed to the denominator, ensuring comparability across facilities with differing survey eligibility. The resulting score ranged from 0 to 1, with higher values indicating greater functional EHR capability.

A facility-level capability score was calculated as the proportion of available EHR functions among eligible items, excluding logical skips, missing values, and non-applicable responses. Based on the observed distribution and conceptual stages of digital maturity, facilities were classified as having low (<0.75), moderate (0.75 to <0.95), or high (≥0.95) functional capability. Internal consistency of the capability measure was evaluated using Cronbach’s alpha, and sensitivity analyses using alternative capability thresholds and ordinal logistic regression produced findings consistent with the primary analyses. These categories represent study-defined thresholds rather than externally validated cut points for EHR functional capability.

Facility characteristics selected a priori included ownership, primary treatment focus, telehealth services, Medicare and Medicaid acceptance, opioid treatment program certification, federally qualified health center designation, and licensing/accreditation intensity, representing organizational capacity, financing structure, regulatory oversight, and integration within broader health information systems. Ownership was classified as private for-profit, private nonprofit, state government, local government, tribal government, or federal government. Facility type was categorized as substance use treatment only, mental health treatment only, or combined substance use and mental health treatment services. Licensing and accreditation intensity was operationalized as the number of licensing or accreditation entities overseeing each facility and categorized as none, one, two, or three or more.

Associations between facility characteristics and each digital maturity outcome were examined using separate multinomial logistic regression models for EHR adoption, HIO participation, and functional EHR capability. For each model, the most digitally advanced category—exclusive EHR use, active HIO participation, and high EHR capability, respectively—served as the reference outcome. Exponentiated coefficients are reported as relative risk ratios (RRRs) with robust standard errors. Adjusted predicted probabilities were estimated using marginal standardization for selected facility characteristics. State fixed effects were evaluated during model development but were not included in the final multinomial models because sparse outcome-by-state cells resulted in model instability and convergence concerns. Geographic variation was therefore assessed descriptively by calculating state-level proportions of facilities within each EHR capability category and visualized using choropleth maps developed in ArcGIS Pro.

Because N-SUMHSS uses a modular survey design, missingness primarily reflected structural features of the survey, including logical skips, module-specific eligibility, not applicable responses, and item-level nonresponse, rather than random missing data. Accordingly, outcome-specific complete-case analyses were performed, excluding facilities with logical skips, missing values, not applicable responses, or don’t know responses for the relevant outcome. Patterns of missingness were evaluated descriptively across organizational characteristics, and analytic denominators therefore differed across outcomes and represent eligible facility populations rather than the full N-SUMHSS census. All statistical analyses were conducted using Stata version 18 (StataCorp, College Station, Texas). Large language model–assisted tools (ChatGPT and Grammarly) were used to support language editing, literature organization, figure development, and code troubleshooting; however, all study design, statistical analyses, interpretation of findings, and final scientific content were independently developed, verified, and approved by the authors.

来源:Nature 健康护理主题 · nature.com