Oct 17, 2023
Submitted via email
Office of Health Plan Standards and Compliance Assistance,
Employee Benefits Security Administration, Room N–5653,
U.S. Department of Labor, 200 Constitution Avenue NW, Washington, DC
20210
Re: [Docket No. FR–51552] RIN 1210-AC11 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA); Request for Comment
On behalf of the Autistic Self Advocacy Network (ASAN), the leading national advocacy organization run by and for Autistic Americans, we write to urge important revisions to the Departments of Labor, Treasury, and Health and Human Services’ proposed rule, FR–51552 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA); Request for Comment.. While ASAN supports the Department’s efforts to better enforce the Paul Wellstone and Pete Domenici Mental Health Parity and Addiction Equity Act of 2008 (MHPAEA)’s requirements regarding Non-Quantitative Treatment Limitations (NQTLs),
First, we are concerned regarding several proposed examples referencing insurer obligations to cover Applied Behavioral Analysis (ABA) as an intervention for autism. Although we agree that benefits for autistic persons should be covered under MHPAEA, by referring to a specific intervention rather than speaking more broadly regarding services to autistic people, the Departments would impose a frozen regulatory standard unable to adjust to an evolving evidence base regarding ABA and autism. As we discuss below, a growing number of autistic people, researchers, and families have expressed serious concerns regarding the elevation of ABA over other intervention alternatives, noting concerns regarding potential iatrogenic harm from ABA interventions and the relatively weak evidence base for ABA interventions. Recognizing these concerns, other federal agencies have typically referred to autism services more generally, explicitly clarifying ABA as only one intervention option among many. We urge the Departments to take a similarly broad-based approach when and if they discuss autism service provision.
Second, we urge the Departments to clarify that plans and issuers may take into account risk of iatrogenic harm, including from provider quality issues and risks associated with coercion, in a neutral fashion across both medical/surgical and mental health and substance use disorder (MH/SUD) benefits. Third, we believe the Departments must provide additional clarity regarding interactions between MHPAEA’s requirements and the integration mandates requiring that people with disabilities be served in the most integrated setting included within the Americans with Disabilities Act, Section 504 of the Rehabilitation Act of 1973 and Section 1557 of the Affordable Care Act. Fourth, we urge the inclusion of a set of broad criteria indicating that comparative analyses of NQTLs and subsequent enforcement may take into account relevant factors regarding the evidence base, practice guidelines, concerns regarding provider quality, iatrogenic harm risks, and other issues highlighted within this comment.
Finally, we urge the Departments to strengthen the regulation by requiring plans and issuers to practice data transparency by providing access to claims data in a HIPAA-compliant fashion to independent health services researchers upon request.
We discuss each of these issues in more detail below.
Autism Examples Should Not Reference or Be Limited To Applied Behavioral Analysis (ABA)
Although ABA has grown very prominent as an autism intervention over the last decade through the adoption of state insurance mandates, many autistic people, researchers, and families have expressed serious concerns regarding the intervention.
Unpacking ABA’s Evidence Base
Autism service-provision has suffered from low quality standards that are unusual compared to other fields. As recently pointed out by Bottema-Beutel (2023), in the autism context, “a practice could be considered ‘established’ even if there were no available studies that used randomized designs and masked assessors, and all of the available evidence was produced by researchers who were also intervention providers.” As these low standards have been challenged, researchers, autistic people and families have increasingly recognized the need to critically reevaluate the evidence base for ABA. A 2018 Cochrane review of early intensive behavioral intervention for young autistic children found that the evidence base in favor of ABA was “weak” and that the strength of such evidence “is limited because it mostly comes from small studies that are not of the optimum design. Due to the inclusion of non‐randomized studies, there is a high risk of bias and we rated the overall quality of evidence as ‘low’ or ‘very low’…meaning further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.”
Until recently, the literature in support of ABA relied heavily on single subject experimental designs. These studies compare the same person in a baseline period to subsequent observations after the person receives the intervention. Such studies are deeply limited in that they are unable to differentiate progress from treatment from developmental progress that would have emerged without regard to whether an individual receives an intervention. In addition to single-subject experimental designs, the ABA industry has relied heavily on clinical trials without random assignment. For example, Lovaas (1987), still heavily relied upon to make the claim that high-intensity ABA intervention is necessary for autistic children, made use of non-random assignment, selecting children for the control group due to staff availability and distance from the treatment center. Researchers have noted this limitation and argued that it should seriously call into question the use of Lovaas’s research to claim that ABA is “scientifically proven”.
Similarly, other recent clinical trials of ABA have relied on waitlist controls, even as they erroneously characterize themselves as randomized clinical trials (RCTs). This is a major methodological flaw. As Beaujean & Farmer (2021) point out in a recent critique of the use of this practice in a study regarding ABA, “This greatly compromises the integrity of the design. A key aspect of a randomized controlled trial (RCT) is that investigators form all groups through some process involving random assignment… Without assigning children to groups using some randomization process, there is no reason to believe that the groups are equivalent before implementing any intervention… Since children in the WC group were not statistically or functionally equivalent to children in the other groups before administering intervention, it is inaccurate to refer to this study as a RCT.” Studies that are purely observational with treatment assignment determined by parental and education team preference leave major potential confounders unaddressed, dramatically limiting the conclusions that can be drawn from them. Despite this, such research is often cited in the ABA literature to argue that the intervention’s long-term effects are scientifically proven.
In recent years, a small number of randomized clinical trials have been conducted evaluating the impact of a particular type of ABA, the Early Start Denver Model (ESDM), which incorporates elements of traditional ABA with other types of autism interventions (namely, developmental interventions quite distinct from standard ABA). This approach has sometimes been referred to (along with other similar ones) within the rubric of natural developmental behavioral interventions, reflecting the integration of behavioral and developmental intervention approaches. Dawson et al. (2010) found that ESDM resulted in some improvements in IQ and adaptive behavior. However, a subsequent RCT using a larger sample size was only able to partially replicate these results.
One serious limitation of these studies is their reliance on a comparison group of autistic children receiving treatment-as-usual in the broader community. Because ABA is generally delivered as a high-intensity intervention within a home environment, some gains may accrue solely by virtue of the provision of paraprofessionals delivering the service acting as informal free childcare for families. Some of the benefits shown in these studies may be the result of these hours of service serving a respite function for families, relieving stress, and increasing the ability of family members to be employed. If so, these gains may be more appropriately achieved through social care interventions rather than medical ones. Further research is needed to evaluate the mechanism behind these treatment effects, should they replicate in future studies. This should also call into question the wisdom of enshrining ABA in regulatory text.
Almost all high-quality studies regarding ABA have focused on the ESDM variant alone, rather than ABA as a whole. Consistent with this, a recent comprehensive systemic review and meta-analysis found that “When effect size estimation was limited to studies with randomized controlled trial (RCT) designs, evidence of positive summary effects existed only for developmental and NDBI [natural developmental behavioral interventions] intervention types. This was also the case when outcomes measured by parent report were excluded. Finally, when effect estimation was limited to RCT designs and to outcomes for which there was no risk of detection bias, no intervention types showed significant effects on any outcome.”
ASAN believes that further concerns articulated below regarding ABA still apply with respect to ESDM. However, in so far as the literature does document any gains from the ESDM model of ABA, it seems inappropriate to generalize that evidence base to any ABA intervention modality when the bulk of the research literature supporting ABA rests on a far weaker evidence base. This reinforces the argument for excluding the broad references to ABA included within the proposed regulation.
Intervention Intensity
Although ABA providers have frequently argued that autistic children require between 25-40 hours a week of ABA-based intervention, this claim is not supported by rigorously conducted research. In fact, a recent multisite randomized controlled trial found that higher-intensity ABA intervention did not yield different results than lower-intensity ABA intervention, further calling into question claims from providers that high-intensity ABA is medically necessary. What’s more, there is reason to believe that high-intensity interventions may pose a substantial risk of harm to autistic children and their families by displacing other life experiences and segregating children from the normal contexts of childhood. To quote Sandbank, Bottema-Beutel & Woynaroski (2021), “highly intensive interventions that separate children from their siblings, peers, and family members for extended periods may have unintended adverse developmental and social consequences.” In the word of another prominent autism researcher, “receiving 40 hours of weekly therapy is essentially a full-time job for a 3-year-old.”
In so far as the Departments do choose to specifically reference ABA in its illustrative examples (something we recommend against below), we urge the Departments to explicitly clarify that plans and issuers may set limits on the number of hours of ABA provided, both because of concerns regarding iatrogenic harm and because rigorous research literature does not support the claim that higher-intensity interventions yield better outcomes for autistic children.
Failure to Report Conflicts of Interest and Monitor Adverse Events
A recently published comprehensive meta-analysis of all group-design, nonpharmacological early intervention autism research conducted between 1970 and 2018 found that 70% of studies had at least one conflict of interest, but only 6% contained conflict of interest statements accounting for all conflicts later found to exist. Common conflicts include the researcher receiving payments or royalties relating to the intervention, studies funded by intervention providers, the study being conducted by a researcher employed by a clinical provider, and other conflicts that suggest substantial financial conflicts in the autism research literature. Further research documents that these issues are also widespread in the ABA literature specifically.
Undisclosed conflicts of interest present serious concerns for these research findings’ integrity, trustworthiness, and, potentially, validity. Researchers with conflicts of interest may choose not to publish findings inconsistent with their financial interests, raising concerns over publication bias. Similarly, researchers with conflicts of interest may not adequately monitor for adverse events emerging from an intervention, a problem that -as we later discuss- is also widespread in the autism literature. These challenges support our request for the exclusion of explicit discussion of ABA from the regulation. We also urge the Departments to explicitly clarify that plans and issuers may take into account the presence of reported and unreported conflicts of interest when evaluating evidence in support of an intervention, provided that they do so in a neutral fashion across both medical/surgical and MH/SUD benefits.
Similarly, another recent review found that autism research frequently failed to incorporate adequate adverse event reporting, limiting conclusions regarding safety that can be drawn from the existing research literature. Unfortunately, many of the mechanisms through which adverse events might take place in autism interventions come from approaches common in ABA-based interventions. Dawson & Fletcher-Watson (2022) identify a number of “common plausibly harmful intervention practices”, including “the use of functional analyses unnecessarily subjecting autistics to repeated contrived provocations until harms ensue…the use of junk food reinforcers and related practices requiring junk food consumption (e.g. preference assessments)…the imposition of impoverished environments depriving autistics of information they need…the targeting of anything autistics show (distress, pain, fear, sadness, excitement, amusement, concentration, perseverance, curiosity, resourcefulness, etc.) as noncompliant “problem” or “challenging” behavior…[and] the punishment (response blocking, response cost, etc.) of autistics for being autistic (flapping their hands, discussing their interests, etc.).” They also note that “wider harms may derive from interventions displacing important activities and opportunities”, an area we discuss in further detail in our discussion of intervention intensity.
Harm from Clinical Passing Demands
Autistic adults have long expressed serious concerns about ABA’s insistence that the objective of intervention should be to make autistic children appear as much like non-autistic as possible, with intervention providers frequently emphasizing things like eye contact, the reduction of stimming, and the discouraging of non-harmful mannerisms and other traits solely because they are common to autistic children. A growing research literature has substantiated that such “passing demands” are associated with serious mental health side effects, including suicidality. Because ABA providers enforce such passing demands by targeting the removal of non-harmful autistic traits as intervention goals, ABA has the potential to be a source of significant harm to autistic children and adults.
As discussed above, autism intervention research has not adequately monitored for adverse events in the past. However, the presence of these ethical concerns and a growing interest on the part of researchers to evaluate them should reinforce the need to maintain a flexible posture in federal regulation. The Departments should not reference ABA by name within its regulation, including in illustrative examples, as doing so would make it difficult for understanding of MHPAEA obligations with respect to autism services to evolve in the face of future research findings on potential harms associated with ABA.
Provider Qualifications
To a far greater degree than other intervention modalities, ABA providers frequently rely primarily on minimally trained paraprofessionals for the delivery of services. ABA provider agencies generally deliver services primarily through Registered Behavior Technicians, who must receive only 40 hours of training and need not have more than a high school diploma to deliver services. While these technicians are nominally delivering service under the supervision of a licensed provider, this supervision is often no more than nominal and rarely takes place on-site. In contrast, occupational therapy and speech-language pathology services are generally delivered through the use of highly trained and licensed professionals.
In the words of a 2017 article written by ABA providers critiquing the field’s reliance on paraprofessionals for the delivery of service, ““previous literature on staff training brings into question whether an 18-year old with just a high school diploma or equivalent has enough of an educational background and level of maturity” to deliver effective services.” It is not unreasonable for plans and issuers to critically evaluate the ability of minimally trained personnel to deliver effective service-provision and to set provider participation requirements that require appropriate training and licensure for ABA providers consistent with those applied to other intervention methodologies, such as occupational therapy and speech pathology services.
We ask that the Departments explicitly clarify that plans and issuers may take into account the extent of provider training and licensure on the part of front-line staff, including paraprofessionals, actually delivering services when evaluating the compliance of NQTLs under MHPAEA, provided that such analyses are applied in a neutral fashion across both medical/surgical and MH/SUD benefits.
The Departments Should Not Elevate a Single Intervention
Federal agencies have historically been careful to avoid endorsing particular intervention modalities in autism service-provision. For example, in 2015 the Department of Education’s Office of Special Education Programs (OSEP), which oversees implementation of the Individuals with Disabilities Education Act (IDEA), issued a dear colleague letter highlighting concerns that “some IDEA programs may be including applied behavior analysis (ABA) therapists exclusively without including, or considering input from, speech language pathologists and other professionals who provide different types of specific therapies that may be appropriate for children with ASD when identifying IDEA services for children with ASD.” OSEP went on to add that “ABA therapy is just one methodology used to address the needs of children with ASD and remind States and local programs to ensure that decisions regarding services are made based on the unique needs of each individual child with a disability.”
Similarly, in 2014, the Centers for Medicare & Medicaid Services (CMS) noted in a Dear Colleague letter to state Medicaid directors that “While much of the current national discussion focuses on one particular treatment modality called Applied Behavioral Analysis (ABA), there are other recognized and emerging treatment modalities for children with ASD.” In short, there is a long history of federal policymakers framing policymaking regarding autism service-provision in terms of services that are responsive to the needs of autistic persons rather than particular intervention modalities, recognizing that the evidence base for any given intervention evolves over time and that it is not the role of the federal government to freeze in place a static understanding of autism services.
Urged Modification of Examples Within the Proposed Rule
Under § 54.9812–1(2)(ii)(C)(5), the Departments propose to add in an example (“Example 5”) of a plan that generally covers treatment for autism and covers outpatient, out-of-network developmental evaluations for autistic persons but excludes all other benefits for outpatient treatment, including ABA, when provided on an out-of-network basis, despite covering the full range of outpatient treatments and treatment settings for medical conditions and surgical procedures when provided on an out-of-network basis. The Departments write that “because the plan only covers one type of benefit for ASD in the outpatient, out-of-network classification and excludes all other benefits for ASD in the classification, but generally covers the full range of medical/surgical benefits in the classification, it fails to provide meaningful benefits for treatment of ASD in the classification.”
To substantiate this argument, the Departments write in the proposed rule’s preamble that “independent standards of current medical practice, ABA therapy is one of the primary treatments for ASD in children”. While ABA is commonly used for autism treatment at present, we have previously discussed how the evidence base for ABA is, in fact, mixed and relatively weak as compared to standards of interventions applied for medical/surgical benefits, offering a potential rationale for taking a different approach to ABA that is rooted in the application of neutral evidentiary standards applied equally to medical/surgical and MH/SUD benefits. In addition, the concerns regarding provider quality and reliance on paraprofessionals discussed above offer a further rationale for why plans and issuers may need to undertake particular scrutiny of ABA providers by covering the service only through in-network providers. We ask the Departments to remove this example or to revise it to remove reference to ABA. If maintained as specific to autism, we suggest that it may be better suited to apply to occupational therapy or speech-language pathology services.
Under § 146.136(c)(4)(viii)(I), the Departments propose to add in an example (“Example 9”) of a plan whose “medical necessity criteria for coverage of ABA therapy requires evidence that the participant’s or beneficiary’s primary caregivers actively participate in ABA therapy, as documented by consistent attendance in parent, caregiver, or guardian training sessions.” The Departments indicate that such a requirement would be prohibited as “the plan deviates from independent professional medical or clinical standards, and there are no similar medical necessity criteria requiring evidence of primary caregiver participation in order to receive coverage of any medical/surgical benefits.”
However, ABA clinical practice standards frequently themselves emphasize the essential nature of parental participation in successful practice. Multiple accreditation and professional organizations dedicated to the delivery of ABA identify parental and caregiver familiarity, training, and reinforcement of ABA techniques as integral to successful intervention, and cite literature identifying parental involvement as critical to ensuring long-term retention of therapeutic benefits –a domain where evidence for ABA efficacy is notably weak. Indeed, the same research base that supports the use of ABA as an autism intervention is unequivocal on the role of parent training and support, noting that parental involvement is a fundamental component of effective ASD intervention. If the Departments are to consider the evidence basis for ABA adequate to require coverage as a widely accepted clinical standard of care for ASD, they also need to permit insurers to consider what those same independent professional clinical standards assert are best practices for delivery of services– to do otherwise is to compel insurers to permit delivery of ineffective -if not potentially deleterious- care.
Under § 146.136(c)(4)(viii)(J), the Departments propose to add an example (“Example 10”) of a plan that “generally excludes experimental treatment of medical conditions and surgical procedures, mental health conditions, and substance use disorders when no professionally recognized treatment guidelines define clinically appropriate standards of care for the condition or disorder and fewer than two randomized controlled trials are available to support the treatment’s use with respect to the given condition or disorder. The plan provides benefits for the treatment of ASD, which is a mental health condition, but, in operation, the plan excludes coverage for ABA therapy to treat children with ASD, deeming it experimental.” The Departments state that this would violate MHPAEA as “more than one professionally recognized treatment guideline defines clinically appropriate standards of care for ASD and more than two randomized controlled trials are available to support the use of ABA therapy to treat certain children with ASD.”
As discussed above, the evidence base for ABA is substantially weaker than commonly understood, a point made clearly in the recent Cochrane review of behavioral interventions for young autistic children. Although randomized controlled trials exist to support the use of ABA therapy, they are generally limited to a specific subset of ABA interventions and cannot be generalized to ABA as a whole, as the proposed example would do. In addition, widespread concerns regarding the lack of adverse event reporting, undisclosed conflicts of interest, and the use of inappropriate control groups call into question the reliability of this evidence base. Plans and issuers should be able to take these factors into account when making decisions regarding what constitutes clinically appropriate care, provided they do so in a neutral fashion that applies such factors to both medical/surgical and MH/SUD benefits.
In addition, plans and issuers should be able to take into account both evidence in favor of and against the efficacy of a given intervention. For example, since 2014, Tricare has provided ABA services through the Autism Care Demonstration (ACD), serving more than 16,000 autistic individuals. Yet, outcomes in the ACD’s patient population have been mixed; in 2020, a Department of Defense review of ABA interventions and reported PDDBI outcome scores found “the majority of beneficiaries experienced little to no change in symptom presentation based on parent report. Additionally, a small percentage of beneficiaries were noted as having worsening of symptoms and a similar small percentage demonstrated symptom improvement” following 18 months of ABA. Following criticism of this report by ABA practitioners and advocates, the Department of Defense’s Comprehensive Autism Care Demonstration Annual Report for FY 2021 used a revised, combined measure for ABA services through 2 years of service and still found that 43% demonstrated no improvement or increased intensity of symptoms. Both ACD annual reports additionally note that they could not assess the clinical significance of changes from ABA interventions or attribute them to ABA rather than other interventions or developmental changes. The report additionally notes the weakness of the evidence base for ABA as we have discussed above, noting both the lack of high-quality evidence discussed previously but also the lack of evidence as to the long-term durability of any improvements, and has repeatedly concluded that “ABA services do not meet the TRICARE hierarchy of evidence standard for medical and proven care”.
We strongly urge the Departments to remove from the proposed examples referenced above all language specifically referring to Applied Behavioral Analysis (ABA) and replace them with other illustrations of the general principles being discussed that do not reference ABA. Similarly, we urge the Departments to consider revisions to prior MHPAEA subregulatory guidance explicitly requiring plans and issuers to cover ABA. Although we agree that benefits for autistic persons should be covered under MHPAEA, by referring to a specific intervention rather than speaking more broadly regarding services to autistic people, the Departments would impose a frozen regulatory standard unable to adjust to an evolving evidence base regarding ABA and autism. It would also ignore fundamental flaws in the existing evidence base for ABA, effectively placing the Departments in the position of making a static clinical judgment privileging a particular intervention over other autism intervention modalities that would be difficult to update in the future in response to an evolving evidence base.
We suggest the Departments consider applying these examples to discuss the application of MHPAEA to other conditions, such as the application of MHPAEA to medication-assisted treatment for persons with Opioid Use Disorder. Where the Departments do wish to retain autism-specific examples, we urge the Departments to either reference other intervention modalities that remain areas of broad consensus within autism contexts (such as occupational therapy or speech-language pathology services) or refer broadly to autism services rather than naming ABA specifically.
Should the Departments retain one or all of these examples with language regarding ABA, which we strongly advise against, we ask that the Departments explicitly indicate that plans and issuers retain the ability to revise their coverage of ABA in response to new evidence regarding risks of iatrogenic harm or updates to the research literature that shed new light on the efficacy and appropriateness of the intervention. We also ask that the Departments explicitly recognize that the existing high-quality evidence with respect to ABA is limited to only a small subset of ABA providers and that plans and issuers should not be required to generalize that evidence to all providers of ABA. While we recognize that this represents a level of detail not typically provided for in an illustrative example within a regulatory text, we believe this only reinforces the importance of removing explicit references to ABA prior to finalizing the proposed rule. Failure to clarify these points while discussing ABA would create requirements that plans and issuers ignore relevant evidentiary standards that they apply to medical/surgical benefits, effectively mandating ABA coverage without regard to the evolving state of the literature supporting it.
Neutral Accounting for Risk of Iatrogenic Harm is Consistent with MHPAEA
Physicians have long recognized the risk of iatrogenic harm in medical/surgical treatment. Such risks also exist in the context of MH/SUD treatment. While expansions in insurance coverage create opportunities for effective substance use treatment, they have also resulted in the scaling up of low-quality inpatient providers, facilitated by an unethical industry of “patient brokers” channeling people with substance use disorders into a revolving door of low-quality programs simply looking to fill beds. Unlike other forms of care, mental health service-provision can legally incorporate involuntary treatment, potentially traumatizing patients and highlighting the importance of providers that practice trauma-informed care. Plans and issuers should carefully monitor the services they deliver and the quality of the providers that deliver them to address these and other risks of iatrogenic harm. When the processes used to address such considerations are applied equally to medical/surgical and MH/SUD benefits, plans and issuers, plans and issuers are not in violation of MHPAEA. We ask the Departments to explicitly clarify this through both regulatory text and within the preamble to the regulation.
Although insurers may not impose more restrictive standards on MH/SUD benefits than on medical/surgical benefits in the same classification, they may apply neutral practices taking into account risk of iatrogenic harm to both types of benefit, even as different outcomes may result from the application of the same standard owing to different risks in different types of service-provision. For example, services whose delivery can involve the use of coercion may pose different risks of harm than medical/surgical benefits in the same category. Similarly, we note that settings in which patients are typically placed by third parties without their involvement in the selection of a provider and/or in which a competitive market among providers does not exist are likely to pose greater risk of iatrogenic harm, as providers have less of an incentive to avoid harm than they would in contexts in which providers are selected from a range of choices by the consumer receiving service-provision.
As an illustration of this point, we note that plans and issuers might reasonably choose to impose more rigorous utilization management requirements on the use of injectable medications that have a history of use for the application of chemical restraint in institutional settings for persons with MH/SUD needs than for other medications for medical/surgical services in the same classification that do not have such a history. This distinction emerges not because of a more restrictive standard being applied to MH/SUD benefits, but because the injectable medication in question has a greater risk of abuse by MH/SUD providers acting out of step with clinical guidelines. Given the evidence base showing the harms of chemical restraint, such restrictions are necessary to address risks of iatrogenic harm. We ask the Departments to explicitly clarify that such an application of utilization management techniques would not violate MHPAEA and to permit plans and issuers to take such factors into account in their comparative analyses.
Similarly, a plan or issuer might reasonably apply closer scrutiny to quality issues in inpatient psychiatric providers than they would to inpatient providers for physical rehabilitation as entrance to the former setting is more frequently the result of emergency placements that may take place against the will or with minimal involvement of the patient. In contrast, selection of an inpatient provider of physical rehabilitation services generally entails some degree of active choice on the part of the consumer or their authorized representative, meaning that consumer preference and a competitive market for services among providers may reduce the risk of iatrogenic harm. We ask the Departments to explicitly clarify that plans and issuers may take into account such factors in provider oversight activities, including setting provider qualifications and establishing provider networks. We further ask the Departments to clarify that in such an instance a plan or issuer that imposes more restrictive provider requirements on inpatient MH/SUD facilities that meet the criteria articulated above would not be in violation of MHPAEA, provided that they do so with the purpose of addressing a legitimate risk of iatrogenic harm.
It is important that the Departments explicitly indicate that evaluating the risk of iatrogenic harm that may emerge from treatment and acting to address provider quality issues that pose risk of such harm will remain a permissible activity under the revised MHPAEA regulations. Failing to do so would be out of step with MHPAEA and would risk significant harm to people with MH/SUD needs.
Interactions Between Integration Mandates from Federal Civil Rights Law and MHPAEA
While this rule seeks to address the disparities experienced by individuals accessing mental health care, it is important that they do not inadvertently contribute to other discriminatory practices inconsistent with federal law. As such, it is essential for the Departments to ensure that parity efforts are consistent with the Americans with Disabilities Act (ADA) and Section 504 of the Rehabilitation Act (Section 504), which impose nondiscrimination requirements on many entities covered by the MHPAEA, and they must be informed by Olmstead considerations. The ADA and Section 504 require covered entities to serve individuals with disabilities in the “most integrated setting.”. As the Department of Health and Human Services’ (“HHS”) recent Notice of Proposed Rulemaking for Section 1557 of the Affordable Care Act acknowledges, the benefit design of health care coverage can discriminate on the basis of disability when “in providing or administering health insurance coverage or other health-related coverage… have or implement benefit designs that do not provide or administer health insurance coverage of other health-related coverage in the most integrated setting appropriate to the needs of qualified individuals with disabilities.”
Coverage decisions, reimbursement rates, and other forms of benefit design can only be provided in parity when they do not discriminate based on disability and hence do not result in individuals being served unnecessarily in segregated settings such as hospitals, nursing homes, or board and care homes. In order to align with the priorities identified in the proposed rulemaking released earlier this year for Section 1557 of the Affordable Care Act and Section 504 of the Rehabilitation Act, the MHPAEA rule should also incorporate a comparable “integration mandate”. HHS must consider how parity measures align with the integration goals evidenced in these rules. Otherwise, insurance benefit packages may discriminate against disabled people by incentivizing segregated settings.
General Considerations for Comparative Analysis
In evaluating whether a health plan has applied a treatment limitation to mental health or substance use disorder benefits that is more restrictive than the predominant financial requirement or treatment limitation of that type applied to substantially all medical/surgical benefits in the same classification, we recommend the Departments explicitly permit comparative analyses and subsequent enforcement to take into account:
- Additions to the evidence base regarding a given treatment, intervention or other benefit;
- The quality of the existing evidence base for a given treatment, intervention or benefit, including study design, sample size, disclosed and undisclosed conflicts of interest and whether studies incorporate tracking of adverse events;
- Practice guidelines specific to a given treatment, intervention, or other benefit, recognizing that they may vary based on the evidence base and/or clinical guidelines relevant to the treatment, intervention or benefit in question;
- Relevant provider quality concerns specific to a given treatment, intervention, or other benefit, recognizing they may vary based on the treatment, intervention or benefit in question;
- The risk of iatrogenic harm associated with a given treatment, intervention or other benefit; and
- Compliance with other legal requirements, including the integration mandate of the Americans with Disabilities Act, Section 504 of the Rehabilitation Act and Section 1557 of the Affordable Care Act.
We request the Departments explicitly include these factors in the proposed rule under § 146.137(c) and elsewhere within the proposed rule as appropriate.
Data Transparency
Under the proposed § 146.137(c)(5), the Departments propose to require plans and issuers to evaluate whether, in any classification, under the terms of the plan (or health insurance coverage) in operation, the processes, strategies, evidentiary standards, or other factors used in designing and applying the NQTL to mental health or substance use disorder benefits are comparable to, and are applied no more stringently than, the processes, strategies, evidentiary standards, or other factors used in designing and applying the limitation with respect to medical/surgical benefits. We agree with this requirement, provided that it incorporates the caveats regarding general considerations for comparative analysis described above. However, we note that a substantial conflict of interest exists when insurers conduct evaluations regarding their compliance.
Fortunately, in recent years a growing community of health services researchers have made use of claims data to study the practices of health plans and insurers, including with respect to MH/SUD benefits. Such research has made use of claims data from both public and private insurers and has demonstrated the ability of third-party researchers to access and make effective use of claims data in HIPAA-compliant computing environments maintained by universities and other entities without such conflicts of interest. We urge the Departments to include language within § 146.137(c)(5) requiring plans and issuers to:
- Make arrangements to permit third-party researchers to access claims data for purposes of evaluating compliance with MHPAEA upon request, provided that they do so in a fashion that complies with HIPAA and other applicable laws regarding the safeguarding of protected health information;
- Indicate to whom such requests should be made at the plan or issuer;
- Describe specific requests that have been made to access such data, and;
- Describe the results of such requests, including the results of such analyses (if data is made available) and the reasons why data was not provided (if data is not made available).
By incorporating these data transparency requirements, the Departments have the opportunity to significantly strengthen the proposed § 146.137(c)(5) and ease their own enforcement activities by permitting the health services research community to assist in the evaluation of compliance with MHPAEA.
Thank you for the opportunity to provide comments on the proposed rule FR–51552 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA). For any further questions concerning ASAN’s comments, please contact Gregory Robinson at grobinson@autisticadvocacy.org.
ASAN Comments Re: Mental Health Parity and Addiction Equity Act
Oct 17, 2023
Submitted via email
Office of Health Plan Standards and Compliance Assistance,
Employee Benefits Security Administration, Room N–5653,
U.S. Department of Labor, 200 Constitution Avenue NW, Washington, DC
20210
Re: [Docket No. FR–51552] RIN 1210-AC11 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA); Request for Comment
On behalf of the Autistic Self Advocacy Network (ASAN), the leading national advocacy organization run by and for Autistic Americans, we write to urge important revisions to the Departments of Labor, Treasury, and Health and Human Services’ proposed rule, FR–51552 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA); Request for Comment.. While ASAN supports the Department’s efforts to better enforce the Paul Wellstone and Pete Domenici Mental Health Parity and Addiction Equity Act of 2008 (MHPAEA)’s requirements regarding Non-Quantitative Treatment Limitations (NQTLs),
First, we are concerned regarding several proposed examples referencing insurer obligations to cover Applied Behavioral Analysis (ABA) as an intervention for autism. Although we agree that benefits for autistic persons should be covered under MHPAEA, by referring to a specific intervention rather than speaking more broadly regarding services to autistic people, the Departments would impose a frozen regulatory standard unable to adjust to an evolving evidence base regarding ABA and autism. As we discuss below, a growing number of autistic people, researchers, and families have expressed serious concerns regarding the elevation of ABA over other intervention alternatives, noting concerns regarding potential iatrogenic harm from ABA interventions and the relatively weak evidence base for ABA interventions. Recognizing these concerns, other federal agencies have typically referred to autism services more generally, explicitly clarifying ABA as only one intervention option among many. We urge the Departments to take a similarly broad-based approach when and if they discuss autism service provision.
Second, we urge the Departments to clarify that plans and issuers may take into account risk of iatrogenic harm, including from provider quality issues and risks associated with coercion, in a neutral fashion across both medical/surgical and mental health and substance use disorder (MH/SUD) benefits. Third, we believe the Departments must provide additional clarity regarding interactions between MHPAEA’s requirements and the integration mandates requiring that people with disabilities be served in the most integrated setting included within the Americans with Disabilities Act, Section 504 of the Rehabilitation Act of 1973 and Section 1557 of the Affordable Care Act. Fourth, we urge the inclusion of a set of broad criteria indicating that comparative analyses of NQTLs and subsequent enforcement may take into account relevant factors regarding the evidence base, practice guidelines, concerns regarding provider quality, iatrogenic harm risks, and other issues highlighted within this comment.
Finally, we urge the Departments to strengthen the regulation by requiring plans and issuers to practice data transparency by providing access to claims data in a HIPAA-compliant fashion to independent health services researchers upon request.
We discuss each of these issues in more detail below.
Autism Examples Should Not Reference or Be Limited To Applied Behavioral Analysis (ABA)
Although ABA has grown very prominent as an autism intervention over the last decade through the adoption of state insurance mandates, many autistic people, researchers, and families have expressed serious concerns regarding the intervention.
Unpacking ABA’s Evidence Base
Autism service-provision has suffered from low quality standards that are unusual compared to other fields. As recently pointed out by Bottema-Beutel (2023), in the autism context, “a practice could be considered ‘established’ even if there were no available studies that used randomized designs and masked assessors, and all of the available evidence was produced by researchers who were also intervention providers.” As these low standards have been challenged, researchers, autistic people and families have increasingly recognized the need to critically reevaluate the evidence base for ABA. A 2018 Cochrane review of early intensive behavioral intervention for young autistic children found that the evidence base in favor of ABA was “weak” and that the strength of such evidence “is limited because it mostly comes from small studies that are not of the optimum design. Due to the inclusion of non‐randomized studies, there is a high risk of bias and we rated the overall quality of evidence as ‘low’ or ‘very low’…meaning further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.”
Until recently, the literature in support of ABA relied heavily on single subject experimental designs. These studies compare the same person in a baseline period to subsequent observations after the person receives the intervention. Such studies are deeply limited in that they are unable to differentiate progress from treatment from developmental progress that would have emerged without regard to whether an individual receives an intervention. In addition to single-subject experimental designs, the ABA industry has relied heavily on clinical trials without random assignment. For example, Lovaas (1987), still heavily relied upon to make the claim that high-intensity ABA intervention is necessary for autistic children, made use of non-random assignment, selecting children for the control group due to staff availability and distance from the treatment center. Researchers have noted this limitation and argued that it should seriously call into question the use of Lovaas’s research to claim that ABA is “scientifically proven”.
Similarly, other recent clinical trials of ABA have relied on waitlist controls, even as they erroneously characterize themselves as randomized clinical trials (RCTs). This is a major methodological flaw. As Beaujean & Farmer (2021) point out in a recent critique of the use of this practice in a study regarding ABA, “This greatly compromises the integrity of the design. A key aspect of a randomized controlled trial (RCT) is that investigators form all groups through some process involving random assignment… Without assigning children to groups using some randomization process, there is no reason to believe that the groups are equivalent before implementing any intervention… Since children in the WC group were not statistically or functionally equivalent to children in the other groups before administering intervention, it is inaccurate to refer to this study as a RCT.” Studies that are purely observational with treatment assignment determined by parental and education team preference leave major potential confounders unaddressed, dramatically limiting the conclusions that can be drawn from them. Despite this, such research is often cited in the ABA literature to argue that the intervention’s long-term effects are scientifically proven.
In recent years, a small number of randomized clinical trials have been conducted evaluating the impact of a particular type of ABA, the Early Start Denver Model (ESDM), which incorporates elements of traditional ABA with other types of autism interventions (namely, developmental interventions quite distinct from standard ABA). This approach has sometimes been referred to (along with other similar ones) within the rubric of natural developmental behavioral interventions, reflecting the integration of behavioral and developmental intervention approaches. Dawson et al. (2010) found that ESDM resulted in some improvements in IQ and adaptive behavior. However, a subsequent RCT using a larger sample size was only able to partially replicate these results.
One serious limitation of these studies is their reliance on a comparison group of autistic children receiving treatment-as-usual in the broader community. Because ABA is generally delivered as a high-intensity intervention within a home environment, some gains may accrue solely by virtue of the provision of paraprofessionals delivering the service acting as informal free childcare for families. Some of the benefits shown in these studies may be the result of these hours of service serving a respite function for families, relieving stress, and increasing the ability of family members to be employed. If so, these gains may be more appropriately achieved through social care interventions rather than medical ones. Further research is needed to evaluate the mechanism behind these treatment effects, should they replicate in future studies. This should also call into question the wisdom of enshrining ABA in regulatory text.
Almost all high-quality studies regarding ABA have focused on the ESDM variant alone, rather than ABA as a whole. Consistent with this, a recent comprehensive systemic review and meta-analysis found that “When effect size estimation was limited to studies with randomized controlled trial (RCT) designs, evidence of positive summary effects existed only for developmental and NDBI [natural developmental behavioral interventions] intervention types. This was also the case when outcomes measured by parent report were excluded. Finally, when effect estimation was limited to RCT designs and to outcomes for which there was no risk of detection bias, no intervention types showed significant effects on any outcome.”
ASAN believes that further concerns articulated below regarding ABA still apply with respect to ESDM. However, in so far as the literature does document any gains from the ESDM model of ABA, it seems inappropriate to generalize that evidence base to any ABA intervention modality when the bulk of the research literature supporting ABA rests on a far weaker evidence base. This reinforces the argument for excluding the broad references to ABA included within the proposed regulation.
Intervention Intensity
Although ABA providers have frequently argued that autistic children require between 25-40 hours a week of ABA-based intervention, this claim is not supported by rigorously conducted research. In fact, a recent multisite randomized controlled trial found that higher-intensity ABA intervention did not yield different results than lower-intensity ABA intervention, further calling into question claims from providers that high-intensity ABA is medically necessary. What’s more, there is reason to believe that high-intensity interventions may pose a substantial risk of harm to autistic children and their families by displacing other life experiences and segregating children from the normal contexts of childhood. To quote Sandbank, Bottema-Beutel & Woynaroski (2021), “highly intensive interventions that separate children from their siblings, peers, and family members for extended periods may have unintended adverse developmental and social consequences.” In the word of another prominent autism researcher, “receiving 40 hours of weekly therapy is essentially a full-time job for a 3-year-old.”
In so far as the Departments do choose to specifically reference ABA in its illustrative examples (something we recommend against below), we urge the Departments to explicitly clarify that plans and issuers may set limits on the number of hours of ABA provided, both because of concerns regarding iatrogenic harm and because rigorous research literature does not support the claim that higher-intensity interventions yield better outcomes for autistic children.
Failure to Report Conflicts of Interest and Monitor Adverse Events
A recently published comprehensive meta-analysis of all group-design, nonpharmacological early intervention autism research conducted between 1970 and 2018 found that 70% of studies had at least one conflict of interest, but only 6% contained conflict of interest statements accounting for all conflicts later found to exist. Common conflicts include the researcher receiving payments or royalties relating to the intervention, studies funded by intervention providers, the study being conducted by a researcher employed by a clinical provider, and other conflicts that suggest substantial financial conflicts in the autism research literature. Further research documents that these issues are also widespread in the ABA literature specifically.
Undisclosed conflicts of interest present serious concerns for these research findings’ integrity, trustworthiness, and, potentially, validity. Researchers with conflicts of interest may choose not to publish findings inconsistent with their financial interests, raising concerns over publication bias. Similarly, researchers with conflicts of interest may not adequately monitor for adverse events emerging from an intervention, a problem that -as we later discuss- is also widespread in the autism literature. These challenges support our request for the exclusion of explicit discussion of ABA from the regulation. We also urge the Departments to explicitly clarify that plans and issuers may take into account the presence of reported and unreported conflicts of interest when evaluating evidence in support of an intervention, provided that they do so in a neutral fashion across both medical/surgical and MH/SUD benefits.
Similarly, another recent review found that autism research frequently failed to incorporate adequate adverse event reporting, limiting conclusions regarding safety that can be drawn from the existing research literature. Unfortunately, many of the mechanisms through which adverse events might take place in autism interventions come from approaches common in ABA-based interventions. Dawson & Fletcher-Watson (2022) identify a number of “common plausibly harmful intervention practices”, including “the use of functional analyses unnecessarily subjecting autistics to repeated contrived provocations until harms ensue…the use of junk food reinforcers and related practices requiring junk food consumption (e.g. preference assessments)…the imposition of impoverished environments depriving autistics of information they need…the targeting of anything autistics show (distress, pain, fear, sadness, excitement, amusement, concentration, perseverance, curiosity, resourcefulness, etc.) as noncompliant “problem” or “challenging” behavior…[and] the punishment (response blocking, response cost, etc.) of autistics for being autistic (flapping their hands, discussing their interests, etc.).” They also note that “wider harms may derive from interventions displacing important activities and opportunities”, an area we discuss in further detail in our discussion of intervention intensity.
Harm from Clinical Passing Demands
Autistic adults have long expressed serious concerns about ABA’s insistence that the objective of intervention should be to make autistic children appear as much like non-autistic as possible, with intervention providers frequently emphasizing things like eye contact, the reduction of stimming, and the discouraging of non-harmful mannerisms and other traits solely because they are common to autistic children. A growing research literature has substantiated that such “passing demands” are associated with serious mental health side effects, including suicidality. Because ABA providers enforce such passing demands by targeting the removal of non-harmful autistic traits as intervention goals, ABA has the potential to be a source of significant harm to autistic children and adults.
As discussed above, autism intervention research has not adequately monitored for adverse events in the past. However, the presence of these ethical concerns and a growing interest on the part of researchers to evaluate them should reinforce the need to maintain a flexible posture in federal regulation. The Departments should not reference ABA by name within its regulation, including in illustrative examples, as doing so would make it difficult for understanding of MHPAEA obligations with respect to autism services to evolve in the face of future research findings on potential harms associated with ABA.
Provider Qualifications
To a far greater degree than other intervention modalities, ABA providers frequently rely primarily on minimally trained paraprofessionals for the delivery of services. ABA provider agencies generally deliver services primarily through Registered Behavior Technicians, who must receive only 40 hours of training and need not have more than a high school diploma to deliver services. While these technicians are nominally delivering service under the supervision of a licensed provider, this supervision is often no more than nominal and rarely takes place on-site. In contrast, occupational therapy and speech-language pathology services are generally delivered through the use of highly trained and licensed professionals.
In the words of a 2017 article written by ABA providers critiquing the field’s reliance on paraprofessionals for the delivery of service, ““previous literature on staff training brings into question whether an 18-year old with just a high school diploma or equivalent has enough of an educational background and level of maturity” to deliver effective services.” It is not unreasonable for plans and issuers to critically evaluate the ability of minimally trained personnel to deliver effective service-provision and to set provider participation requirements that require appropriate training and licensure for ABA providers consistent with those applied to other intervention methodologies, such as occupational therapy and speech pathology services.
We ask that the Departments explicitly clarify that plans and issuers may take into account the extent of provider training and licensure on the part of front-line staff, including paraprofessionals, actually delivering services when evaluating the compliance of NQTLs under MHPAEA, provided that such analyses are applied in a neutral fashion across both medical/surgical and MH/SUD benefits.
The Departments Should Not Elevate a Single Intervention
Federal agencies have historically been careful to avoid endorsing particular intervention modalities in autism service-provision. For example, in 2015 the Department of Education’s Office of Special Education Programs (OSEP), which oversees implementation of the Individuals with Disabilities Education Act (IDEA), issued a dear colleague letter highlighting concerns that “some IDEA programs may be including applied behavior analysis (ABA) therapists exclusively without including, or considering input from, speech language pathologists and other professionals who provide different types of specific therapies that may be appropriate for children with ASD when identifying IDEA services for children with ASD.” OSEP went on to add that “ABA therapy is just one methodology used to address the needs of children with ASD and remind States and local programs to ensure that decisions regarding services are made based on the unique needs of each individual child with a disability.”
Similarly, in 2014, the Centers for Medicare & Medicaid Services (CMS) noted in a Dear Colleague letter to state Medicaid directors that “While much of the current national discussion focuses on one particular treatment modality called Applied Behavioral Analysis (ABA), there are other recognized and emerging treatment modalities for children with ASD.” In short, there is a long history of federal policymakers framing policymaking regarding autism service-provision in terms of services that are responsive to the needs of autistic persons rather than particular intervention modalities, recognizing that the evidence base for any given intervention evolves over time and that it is not the role of the federal government to freeze in place a static understanding of autism services.
Urged Modification of Examples Within the Proposed Rule
Under § 54.9812–1(2)(ii)(C)(5), the Departments propose to add in an example (“Example 5”) of a plan that generally covers treatment for autism and covers outpatient, out-of-network developmental evaluations for autistic persons but excludes all other benefits for outpatient treatment, including ABA, when provided on an out-of-network basis, despite covering the full range of outpatient treatments and treatment settings for medical conditions and surgical procedures when provided on an out-of-network basis. The Departments write that “because the plan only covers one type of benefit for ASD in the outpatient, out-of-network classification and excludes all other benefits for ASD in the classification, but generally covers the full range of medical/surgical benefits in the classification, it fails to provide meaningful benefits for treatment of ASD in the classification.”
To substantiate this argument, the Departments write in the proposed rule’s preamble that “independent standards of current medical practice, ABA therapy is one of the primary treatments for ASD in children”. While ABA is commonly used for autism treatment at present, we have previously discussed how the evidence base for ABA is, in fact, mixed and relatively weak as compared to standards of interventions applied for medical/surgical benefits, offering a potential rationale for taking a different approach to ABA that is rooted in the application of neutral evidentiary standards applied equally to medical/surgical and MH/SUD benefits. In addition, the concerns regarding provider quality and reliance on paraprofessionals discussed above offer a further rationale for why plans and issuers may need to undertake particular scrutiny of ABA providers by covering the service only through in-network providers. We ask the Departments to remove this example or to revise it to remove reference to ABA. If maintained as specific to autism, we suggest that it may be better suited to apply to occupational therapy or speech-language pathology services.
Under § 146.136(c)(4)(viii)(I), the Departments propose to add in an example (“Example 9”) of a plan whose “medical necessity criteria for coverage of ABA therapy requires evidence that the participant’s or beneficiary’s primary caregivers actively participate in ABA therapy, as documented by consistent attendance in parent, caregiver, or guardian training sessions.” The Departments indicate that such a requirement would be prohibited as “the plan deviates from independent professional medical or clinical standards, and there are no similar medical necessity criteria requiring evidence of primary caregiver participation in order to receive coverage of any medical/surgical benefits.”
However, ABA clinical practice standards frequently themselves emphasize the essential nature of parental participation in successful practice. Multiple accreditation and professional organizations dedicated to the delivery of ABA identify parental and caregiver familiarity, training, and reinforcement of ABA techniques as integral to successful intervention, and cite literature identifying parental involvement as critical to ensuring long-term retention of therapeutic benefits –a domain where evidence for ABA efficacy is notably weak. Indeed, the same research base that supports the use of ABA as an autism intervention is unequivocal on the role of parent training and support, noting that parental involvement is a fundamental component of effective ASD intervention. If the Departments are to consider the evidence basis for ABA adequate to require coverage as a widely accepted clinical standard of care for ASD, they also need to permit insurers to consider what those same independent professional clinical standards assert are best practices for delivery of services– to do otherwise is to compel insurers to permit delivery of ineffective -if not potentially deleterious- care.
Under § 146.136(c)(4)(viii)(J), the Departments propose to add an example (“Example 10”) of a plan that “generally excludes experimental treatment of medical conditions and surgical procedures, mental health conditions, and substance use disorders when no professionally recognized treatment guidelines define clinically appropriate standards of care for the condition or disorder and fewer than two randomized controlled trials are available to support the treatment’s use with respect to the given condition or disorder. The plan provides benefits for the treatment of ASD, which is a mental health condition, but, in operation, the plan excludes coverage for ABA therapy to treat children with ASD, deeming it experimental.” The Departments state that this would violate MHPAEA as “more than one professionally recognized treatment guideline defines clinically appropriate standards of care for ASD and more than two randomized controlled trials are available to support the use of ABA therapy to treat certain children with ASD.”
As discussed above, the evidence base for ABA is substantially weaker than commonly understood, a point made clearly in the recent Cochrane review of behavioral interventions for young autistic children. Although randomized controlled trials exist to support the use of ABA therapy, they are generally limited to a specific subset of ABA interventions and cannot be generalized to ABA as a whole, as the proposed example would do. In addition, widespread concerns regarding the lack of adverse event reporting, undisclosed conflicts of interest, and the use of inappropriate control groups call into question the reliability of this evidence base. Plans and issuers should be able to take these factors into account when making decisions regarding what constitutes clinically appropriate care, provided they do so in a neutral fashion that applies such factors to both medical/surgical and MH/SUD benefits.
In addition, plans and issuers should be able to take into account both evidence in favor of and against the efficacy of a given intervention. For example, since 2014, Tricare has provided ABA services through the Autism Care Demonstration (ACD), serving more than 16,000 autistic individuals. Yet, outcomes in the ACD’s patient population have been mixed; in 2020, a Department of Defense review of ABA interventions and reported PDDBI outcome scores found “the majority of beneficiaries experienced little to no change in symptom presentation based on parent report. Additionally, a small percentage of beneficiaries were noted as having worsening of symptoms and a similar small percentage demonstrated symptom improvement” following 18 months of ABA. Following criticism of this report by ABA practitioners and advocates, the Department of Defense’s Comprehensive Autism Care Demonstration Annual Report for FY 2021 used a revised, combined measure for ABA services through 2 years of service and still found that 43% demonstrated no improvement or increased intensity of symptoms. Both ACD annual reports additionally note that they could not assess the clinical significance of changes from ABA interventions or attribute them to ABA rather than other interventions or developmental changes. The report additionally notes the weakness of the evidence base for ABA as we have discussed above, noting both the lack of high-quality evidence discussed previously but also the lack of evidence as to the long-term durability of any improvements, and has repeatedly concluded that “ABA services do not meet the TRICARE hierarchy of evidence standard for medical and proven care”.
We strongly urge the Departments to remove from the proposed examples referenced above all language specifically referring to Applied Behavioral Analysis (ABA) and replace them with other illustrations of the general principles being discussed that do not reference ABA. Similarly, we urge the Departments to consider revisions to prior MHPAEA subregulatory guidance explicitly requiring plans and issuers to cover ABA. Although we agree that benefits for autistic persons should be covered under MHPAEA, by referring to a specific intervention rather than speaking more broadly regarding services to autistic people, the Departments would impose a frozen regulatory standard unable to adjust to an evolving evidence base regarding ABA and autism. It would also ignore fundamental flaws in the existing evidence base for ABA, effectively placing the Departments in the position of making a static clinical judgment privileging a particular intervention over other autism intervention modalities that would be difficult to update in the future in response to an evolving evidence base.
We suggest the Departments consider applying these examples to discuss the application of MHPAEA to other conditions, such as the application of MHPAEA to medication-assisted treatment for persons with Opioid Use Disorder. Where the Departments do wish to retain autism-specific examples, we urge the Departments to either reference other intervention modalities that remain areas of broad consensus within autism contexts (such as occupational therapy or speech-language pathology services) or refer broadly to autism services rather than naming ABA specifically.
Should the Departments retain one or all of these examples with language regarding ABA, which we strongly advise against, we ask that the Departments explicitly indicate that plans and issuers retain the ability to revise their coverage of ABA in response to new evidence regarding risks of iatrogenic harm or updates to the research literature that shed new light on the efficacy and appropriateness of the intervention. We also ask that the Departments explicitly recognize that the existing high-quality evidence with respect to ABA is limited to only a small subset of ABA providers and that plans and issuers should not be required to generalize that evidence to all providers of ABA. While we recognize that this represents a level of detail not typically provided for in an illustrative example within a regulatory text, we believe this only reinforces the importance of removing explicit references to ABA prior to finalizing the proposed rule. Failure to clarify these points while discussing ABA would create requirements that plans and issuers ignore relevant evidentiary standards that they apply to medical/surgical benefits, effectively mandating ABA coverage without regard to the evolving state of the literature supporting it.
Neutral Accounting for Risk of Iatrogenic Harm is Consistent with MHPAEA
Physicians have long recognized the risk of iatrogenic harm in medical/surgical treatment. Such risks also exist in the context of MH/SUD treatment. While expansions in insurance coverage create opportunities for effective substance use treatment, they have also resulted in the scaling up of low-quality inpatient providers, facilitated by an unethical industry of “patient brokers” channeling people with substance use disorders into a revolving door of low-quality programs simply looking to fill beds. Unlike other forms of care, mental health service-provision can legally incorporate involuntary treatment, potentially traumatizing patients and highlighting the importance of providers that practice trauma-informed care. Plans and issuers should carefully monitor the services they deliver and the quality of the providers that deliver them to address these and other risks of iatrogenic harm. When the processes used to address such considerations are applied equally to medical/surgical and MH/SUD benefits, plans and issuers, plans and issuers are not in violation of MHPAEA. We ask the Departments to explicitly clarify this through both regulatory text and within the preamble to the regulation.
Although insurers may not impose more restrictive standards on MH/SUD benefits than on medical/surgical benefits in the same classification, they may apply neutral practices taking into account risk of iatrogenic harm to both types of benefit, even as different outcomes may result from the application of the same standard owing to different risks in different types of service-provision. For example, services whose delivery can involve the use of coercion may pose different risks of harm than medical/surgical benefits in the same category. Similarly, we note that settings in which patients are typically placed by third parties without their involvement in the selection of a provider and/or in which a competitive market among providers does not exist are likely to pose greater risk of iatrogenic harm, as providers have less of an incentive to avoid harm than they would in contexts in which providers are selected from a range of choices by the consumer receiving service-provision.
As an illustration of this point, we note that plans and issuers might reasonably choose to impose more rigorous utilization management requirements on the use of injectable medications that have a history of use for the application of chemical restraint in institutional settings for persons with MH/SUD needs than for other medications for medical/surgical services in the same classification that do not have such a history. This distinction emerges not because of a more restrictive standard being applied to MH/SUD benefits, but because the injectable medication in question has a greater risk of abuse by MH/SUD providers acting out of step with clinical guidelines. Given the evidence base showing the harms of chemical restraint, such restrictions are necessary to address risks of iatrogenic harm. We ask the Departments to explicitly clarify that such an application of utilization management techniques would not violate MHPAEA and to permit plans and issuers to take such factors into account in their comparative analyses.
Similarly, a plan or issuer might reasonably apply closer scrutiny to quality issues in inpatient psychiatric providers than they would to inpatient providers for physical rehabilitation as entrance to the former setting is more frequently the result of emergency placements that may take place against the will or with minimal involvement of the patient. In contrast, selection of an inpatient provider of physical rehabilitation services generally entails some degree of active choice on the part of the consumer or their authorized representative, meaning that consumer preference and a competitive market for services among providers may reduce the risk of iatrogenic harm. We ask the Departments to explicitly clarify that plans and issuers may take into account such factors in provider oversight activities, including setting provider qualifications and establishing provider networks. We further ask the Departments to clarify that in such an instance a plan or issuer that imposes more restrictive provider requirements on inpatient MH/SUD facilities that meet the criteria articulated above would not be in violation of MHPAEA, provided that they do so with the purpose of addressing a legitimate risk of iatrogenic harm.
It is important that the Departments explicitly indicate that evaluating the risk of iatrogenic harm that may emerge from treatment and acting to address provider quality issues that pose risk of such harm will remain a permissible activity under the revised MHPAEA regulations. Failing to do so would be out of step with MHPAEA and would risk significant harm to people with MH/SUD needs.
Interactions Between Integration Mandates from Federal Civil Rights Law and MHPAEA
While this rule seeks to address the disparities experienced by individuals accessing mental health care, it is important that they do not inadvertently contribute to other discriminatory practices inconsistent with federal law. As such, it is essential for the Departments to ensure that parity efforts are consistent with the Americans with Disabilities Act (ADA) and Section 504 of the Rehabilitation Act (Section 504), which impose nondiscrimination requirements on many entities covered by the MHPAEA, and they must be informed by Olmstead considerations. The ADA and Section 504 require covered entities to serve individuals with disabilities in the “most integrated setting.”. As the Department of Health and Human Services’ (“HHS”) recent Notice of Proposed Rulemaking for Section 1557 of the Affordable Care Act acknowledges, the benefit design of health care coverage can discriminate on the basis of disability when “in providing or administering health insurance coverage or other health-related coverage… have or implement benefit designs that do not provide or administer health insurance coverage of other health-related coverage in the most integrated setting appropriate to the needs of qualified individuals with disabilities.”
Coverage decisions, reimbursement rates, and other forms of benefit design can only be provided in parity when they do not discriminate based on disability and hence do not result in individuals being served unnecessarily in segregated settings such as hospitals, nursing homes, or board and care homes. In order to align with the priorities identified in the proposed rulemaking released earlier this year for Section 1557 of the Affordable Care Act and Section 504 of the Rehabilitation Act, the MHPAEA rule should also incorporate a comparable “integration mandate”. HHS must consider how parity measures align with the integration goals evidenced in these rules. Otherwise, insurance benefit packages may discriminate against disabled people by incentivizing segregated settings.
General Considerations for Comparative Analysis
In evaluating whether a health plan has applied a treatment limitation to mental health or substance use disorder benefits that is more restrictive than the predominant financial requirement or treatment limitation of that type applied to substantially all medical/surgical benefits in the same classification, we recommend the Departments explicitly permit comparative analyses and subsequent enforcement to take into account:
We request the Departments explicitly include these factors in the proposed rule under § 146.137(c) and elsewhere within the proposed rule as appropriate.
Data Transparency
Under the proposed § 146.137(c)(5), the Departments propose to require plans and issuers to evaluate whether, in any classification, under the terms of the plan (or health insurance coverage) in operation, the processes, strategies, evidentiary standards, or other factors used in designing and applying the NQTL to mental health or substance use disorder benefits are comparable to, and are applied no more stringently than, the processes, strategies, evidentiary standards, or other factors used in designing and applying the limitation with respect to medical/surgical benefits. We agree with this requirement, provided that it incorporates the caveats regarding general considerations for comparative analysis described above. However, we note that a substantial conflict of interest exists when insurers conduct evaluations regarding their compliance.
Fortunately, in recent years a growing community of health services researchers have made use of claims data to study the practices of health plans and insurers, including with respect to MH/SUD benefits. Such research has made use of claims data from both public and private insurers and has demonstrated the ability of third-party researchers to access and make effective use of claims data in HIPAA-compliant computing environments maintained by universities and other entities without such conflicts of interest. We urge the Departments to include language within § 146.137(c)(5) requiring plans and issuers to:
By incorporating these data transparency requirements, the Departments have the opportunity to significantly strengthen the proposed § 146.137(c)(5) and ease their own enforcement activities by permitting the health services research community to assist in the evaluation of compliance with MHPAEA.
Thank you for the opportunity to provide comments on the proposed rule FR–51552 Requirements Related to the Mental Health Parity and Addiction Equity Act (MHPAEA). For any further questions concerning ASAN’s comments, please contact Gregory Robinson at grobinson@autisticadvocacy.org.