The Fight for Medicaid Coverage of Evidence-Based Behavioral Treatments for Autism

By Miriam Harmatz, Betsy Havens & Monica Vigues-Pitan
Published in the September 2015 issue of the Clearinghouse Community (Clearinghouse Community is part of the Sargent Shriver National Center on Poverty Law)

K.G. seemed like a typical baby. He was happy, eating well, and communicating with his parents. But some months after his first birthday, everything began to change. By the time he was 2, he was barely communicating and was eventually diagnosed with autism. When we met K.G. at 5 years old, he could not speak at all, and, to the extent he interacted with others, he was kicking, biting, or hitting them. His mother had scars up and down her arms from his biting and scratching, and she spent her days (and nights) trying to stop him from hurting himself. His doctor prescribed a therapy known as applied behavior analysis, but children who relied on Medicaid, such as K.G., had no hope of receiving such therapy. When we filed K.G. v. Dudek in 2010, Florida’s Medicaid program (like that of virtually all other states at the time) did not cover any behavioral treatment for children with autism. Two significant events, however, helped lay the foundation for challenging Florida’s exclusionary policy. First, the 2008 Florida state legislature mandated coverage of applied behavior analysis by commercial insurance. Second, shortly thereafter, Ohio advocates won a tremendous victory when the Sixth Circuit upheld a preliminary injunction enjoining the state from terminating such coverage.

Our goal in this case was to ensure that applied behavior analysis became a Medicaid-covered treatment for eligible children with autism in Florida. Success required both prohibitory and affirmative injunctive relief: an order deleting the state rule that excluded Medicaid coverage of any behavioral treatment for autism and an order that applied behavior analysis be included in the state rule as a covered treatment. After our plaintiffs prevailed, the federal agency responsible for administering the Medicaid program (the Centers for Medicare and Medicaid Services) issued an amended regulation and new guidance that should render the need for Medicaid litigation for applied behavior analysis in other states unnecessary. Nonetheless, K.G.’s story is worth sharing because some of the challenges, issues, practice tips, and pitfalls are relevant in other health care advocacy—particularly regarding “evidence-based medicine” disputes and individual plaintiffs seeking systemic relief.


Prelitigation Discovery and Settlement Efforts

Before considering litigation, we filed extensive public-records requests with the state Medicaid agency for documents identifying potential avenues by which low-income children in Florida might access applied behavior analysis. The records confirmed that Medicaid-enrolled children with autism could not receive applied behavior analysis unless they were in the state’s home and community-based developmental disabilities waiver program. But, unlike regular Medicaid, waivers have a capped enrollment, and Florida’s waiver had a waiting list of 20,000. Given the recently mandated coverage of applied behavior analysis by commercial insurance, the door to such coverage opened by the Sixth Circuit, and our clients’ desperate need for the treatment, we contacted state Medicaid officials to amend their exclusionary rule and cover medically necessary applied behavior analysis for children with autism.


District Court Litigation

When the state failed to respond, we identified experts and filed a complaint alleging that the state’s rule violated the federal Medicaid Act. At the same time we filed a preliminary injunction motion for K.G., the oldest of the three plaintiffs and the most severely disabled. After a year of intensive litigation, we prevailed at a four-day bench trial.

Legal Claims. First, we claimed that the state’s refusal to cover prescribed applied behavior analysis violated the Medicaid Act’s Early and Periodic Screening, Diagnosis, and Treatment requirements. Under these requirements, which pertain to Medicaid beneficiaries under 21, states must cover any service that fits within one of the general categories of medical services enumerated in the Medicaid Act when correcting or ameliorating a child’s condition is necessary. Because the Medicaid Act lists no category for “behavioral services,” we argued that applied behavior analysis could be covered as a “rehabilitative service,” which requires, inter alia, that a physician or other licensed practitioner recommend the service “for the maximum reduction of physical or mental disability, and restoration of an individual to the best possible functional level.”

While children with other mental and behavioral health diagnoses could get coverage for their prescribed treatments, children with autism could not.

Proving that applied behavior analysis met the definition of a rehabilitative service under the Medicaid Act was a risky challenge. That the treatment would reduce a child’s disability (the first prong of the critical clause) was never in dispute. Rather, proving the second prong of the clause—restoring the child “to the best possible functional level”—posed the risk. This prong had never been definitively addressed in a reported Medicaid case. Parents’ League for Effective Autism Services v. Jones-Kelley settled after the preliminary injunction was upheld. And the Sixth Circuit’s opinion upholding the preliminary injunction in that case, while representing a tremendous victory, was vulnerable to a statutory-construction challenge because the opinion effectively treated as disjunctive, rather than conjunctive, the “and” between the two prongs requiring that the service both reduce and restore.

Second, we claimed that Florida’s policy violated the Medicaid Act’s “comparability” requirement, which essentially prohibits states from discriminating on the basis of a person’s diagnosis. In this case, Florida was discriminating against children with autism. While children with other mental and behavioral health diagnoses could get coverage for their prescribed treatments, children with autism could not.
The State’s Unexpected Defense. Because K.G.’s condition was very severe, we sought a preliminary injunction for the state to be ordered to cover his prescribed applied behavior analysis right away. The court did issue the injunction, but that was only the beginning.

From the start, we expected that the Medicaid agency would attack the Parents’ League for Effective Autism Services decision and argue that applied behavior analysis was not a “rehabilitative service,” and it definitely did so. What we had not expected was the agency’s argument that applied behavior analysis is “experimental.” As noted, coverage of applied behavior analysis was mandated by the Florida legislature for commercial insurance and was covered under the state’s developmental disabilities waiver, which similarly excludes “experimental” treatments.

By virtue of the state’s unforeseen defense, we learned a great deal about what is and is not “experimental” under Medicaid case law, what constitutes a “reasonable” determination of “experimental,” and the critical import of this evolving issue in case law and public policy. While the Medicaid statute and regulations contain no definition of “experimental,” controlling circuit precedent in Rush v. Parham established the following standard: if a treatment is considered proven and effective by the medical community, it is not experimental. If, however, the treatment is new or lacks consensus in the medical community, the treatment must still be covered; it is then incumbent on the beneficiary to give evidence of safety and efficacy to demonstrate that the treatment is not experimental.

The email … stated outright that the agency could not find an expert to testify on its behalf. It felt like Christmas had come early.

Under the controlling legal standard set forth in Rush, applied behavior analysis is not experimental; in the undisputed consensus of the medical community, it is an effective treatment for autism. Indeed, under Rush, one never even gets to the second prong of the inquiry, that is, needing to give evidence of safety and efficacy. Rush, however, was decided in 1980, before the prominence of “evidence-based medicine,” which requires that the treatment be supported by literature from peer-reviewed journals demonstrating efficacy. It was decided before the emergence (in the insurance industry, if not yet in the governing case law) of the so-called gold standard for proof of efficacy—large-scale randomized, double-blinded, controlled trials. Accordingly, notwithstanding Rush’s unequivocal support that applied behavior analysis is not experimental, we introduced a “plethora” of evidence supporting the treatment’s efficacy and demonstrated that the state had not applied a reasonable standard in determining that it was experimental.

The Medicaid agency insisted that applied behavior analysis was experimental because the evidence demonstrating the treatment’s efficacy did not meet the “gold standard.” Throughout most of the trial court proceedings, including the briefing on cross motions for summary judgment, the agency relied solely on reports that critiqued the strength and size of studies supporting the efficacy of applied behavior analysis. The agency was unable to locate an expert who would support its position that applied behavior analysis was experimental. Our favorite exhibit was an email forwarded to our rebuttal expert, Dr. Jon Bailey, by one of his former graduate students. The email, which originated from the state Medicaid general counsel’s office, stated outright that the agency could not find an expert to testify on its behalf. It felt like Christmas had come early.

Then, at the eleventh hour—months after the end of an extended expert discovery period and the conclusion of summary judgment briefing and only three weeks before the scheduled trial date—the state’s SOS was answered. Its lawyers suddenly declared that they had identified two experts and moved for leave to use their deposition testimony at trial. The court simultaneously granted the state’s motion and denied our motion for summary judgment. Whether applied behavior analysis was experimental had now become a disputed material fact.

Courtroom Drama. The fundamental factual dispute at trial was whether applied behavior analysis was experimental. Under the still good, albeit old, Rush case, the first inquiry is whether there is consensus in the medical community that the treatment is proven and effective, and on that issue there could be no doubt. We prepared an exhibit of consensus reports from the relevant medical entities supporting the efficacy of applied behavior analysis as a treatment for autism. It included reports from the Centers for Medicare and Medicaid Services, the surgeon general, the Centers for Disease Control and Prevention, the National Institute of Child Health and Human Development, the National Institute of Mental Health, and the American Academy of Pediatrics. Introducing the exhibit was our expert, Dr. James Mulick. Within the first five minutes of direct examination describing highlights in his 47-page curriculum vitae, there was no doubt that one of the world’s foremost experts in the treatment of autism was sitting in the witness chair.

After introducing the consensus exhibit and describing the overwhelming evidence supporting the efficacy of applied behavior analysis, Dr. Mulick poignantly described what autism treatment consisted of in the 1970s. Tragically, at that time, autism was considered untreatable. Children and adults with autism were routinely institutionalized where they were subject to segregation, sterilization, sedation, and restraints. He described how the emergence of applied behavior analysis revolutionized the treatment and prognosis of autism and resulted in significant public savings. According to studies described by Dr. Mulick, applied behavior analysis increases the likelihood that a child with autism will finish school and find employment rather than being institutionalized at taxpayer expense.

But the state still attacked evidence supporting applied behavior analysis as purportedly weak. Its attorneys relied primarily on a report by a for-profit health technology assessment entity whose services include proprietary reports that grade evidence supporting a treatment’s efficacy on a scale from A to D. It graded the evidence demonstrating the efficacy of applied behavior analysis as a “C” because of the lack of large-scale trials as required by the “gold standard” and critiqued the strength of studies supporting applied behavior analysis because the studies had few participants and were neither blinded nor controlled.

As a legal matter, a state cannot have a service that covers only three children. As a practical matter, without systemwide billing codes and an established process for reimbursement, clinicians will not provide services.

After the state’s experts opined on video that a Medicaid agency’s reliance on such reports was “reasonable,” our rebuttal expert, Dr. Jon Bailey, thoroughly dismantled that premise. In response to questions on direct and cross examination and from the judge herself, Dr. Bailey carefully explained why testing the efficacy of behavioral treatments such as applied behavior analysis simply could not be done through the “gold standard’s” large-scale blinded trials. Moreover, according to Dr. Bailey, the reports relied on by the state inappropriately excluded a large number of well-conducted “small-number” studies, including meta-analyses, which demonstrated the significant positive effect of applied behavior analysis on children with autism.

We were fortunate that the plaintiffs’ treating pediatric neurologists—Dr. Roberto Lopez Alberola and Dr. Elza Vasconcellos—happened to be the two foremost autism experts in Miami. They had each treated thousands of children with autism. And because applied behavior analysis had now been covered by commercial insurance for several years, they were able to describe the disparity in prognoses and outcomes between their patients with private insurance and those with Medicaid.

Dr. Lopez Alberola, chief of child neurology at the University of Miami School of Medicine, was an unflappable witness. He listened intently to each question and responded in a calm, measured manner. He testified that applied behavior analysis essentially restored the “building blocks” that had gone awry with autism and explained how this restoration returned the child to the developmental curve. While his style was unemotional, there was undeniable drama when the doctor offered his own judgment that giving the treatment to children with commercial insurance while denying it to those who relied on Medicaid was discriminatory. By contrast, Dr. Vasconcellos, director of the Autism Clinic at Miami Children’s Hospital, looked directly at the judge and, with her voice cracking, described how painful it was for her as a mother and doctor to have to tell parents who were already devastated by their child’s autism diagnosis that the treatment that could most help was not covered by Medicaid. She talked about how badly she felt for these parents. She knew they had very little money and could not afford applied behavior analysis, but she had to tell them about the treatment and write a prescription because she felt it would be malpractice not to do so.

K.G.’s therapist, a young, articulate behavior analyst, testified next. By the time of trial, K.G. had received only four weeks of therapy. But even in that short time, his improvement was profound. When he began applied behavior analysis, his behavior was out of control. He was kicking, hitting, throwing objects, banging his head against the wall, biting and scratching himself and others, and screaming incessantly. Using charts and graphs, the therapist showed the court the dramatic reduction in K.G.’s aggressive behavior.

Finally K.G.’s mother took the stand. In her soft-spoken Spanish, she told the court how applied behavior analysis was changing her son’s life and what their life was like before treatment. She shared the constant fear over what would happen to her son without treatment—especially after she was gone. She was afraid he would hurt himself or someone else; she was afraid that he would end up in an institution or jail. She rolled up her sleeves and showed the judge the scars on her arms from his biting and scratching. With tears streaming down her face, she testified that she felt applied behavior analysis was restoring her child to her, saying, “A light had been shut off in my son, and now it was being turned back on.”

At the end of the four-day trial, Judge Joan A. Lenard ruled from the bench:

The decision in this case cries out for immediate ruling.... I’ve sat on the federal bench for 16 years [and] this case, if not the most important, is one of the most important cases that I have ever heard.… Plaintiffs have established through their expert witnesses that there exists in the scientific and medical peer-reviewed literature a plethora of meta-analyses studies and articles that clearly establish Applied Behavior Analysis as an effective and significant treatment to prevent disability and restore children to their best possible functional level and restore their developmental skill.... We are now in 2012. How many children were lost?... The Medicaid population of children diagnosed with autism and/or autism spectrum disorder are deserving and will be given Applied Behavior Analysis treatment in the State of Florida.

Judge Lenard’s words, spoken late on a Friday afternoon, left everyone in the courtroom (with the exception of the judge herself and defense counsel) in tears.

On the following Monday morning, the court entered a permanent injunction order ensuring that applied behavior analysis would immediately be available as a covered Medicaid benefit. The injunction required the Medicaid agency to delete the exclusion of any behavioral treatment for autism, to inform providers that applied behavior analysis was now a covered Medicaid service for children diagnosed with autism, and to supply authorization codes so that providers could bill for the service. The district court further ordered the Medicaid agency to

• “provide, fund, and authorize Applied Behavioral Analysis treatment to Plaintiffs … as well as to all Medicaid-eligible persons under the age of 21 in Florida who have been diagnosed with autism or Autism Spectrum Disorder, as prescribed by a physician or other licensed practitioner,” and
• “take whatever additional steps are necessary for the immediate and orderly administration of [applied behavior analysis] treatment for Medicaid-eligible persons under the age of 21 who have been diagnosed with autism or Autism Spectrum Disorder.”

Surprises on Appeal

On appeal, we were once again surprised by the state officials’ arguments. The state officials did not argue that applied behavior analysis was not a rehabilitative service within the Medicaid Act; nor did they argue that applied behavior analysis was experimental. They did not challenge the lower court’s finding that they had been arbitrary and capricious. Instead the state officials challenged the order as “overbroad” and asked the Eleventh Circuit to vacate and remand the lower court’s order to change the exclusionary policy as it applied statewide and to limit relief solely to the three plaintiffs.

This individual lawsuit was an appropriate and efficient vehicle for the affirmative injunction requiring the Florida Medicaid agency to provide applied behavior analysis as a covered treatment.

Seizing on the trial court’s language that applied behavior analysis be provided immediately both to the plaintiffs and to other children, the state officials first argued that the court had stripped them of their role in individual medical-necessity determinations by ruling that applied behavior analysis was medically necessary for every child. This argument was a red herring. That a Medicaid-covered service must be medically necessary for each child and that the state has a role to play in that determination were never in dispute. The district court did not decide otherwise. Thus, while ordering the lower court to make undisputed clarifications of two paragraphs of the order, the court of appeals found no abuse of discretion.

The state next argued that the district court abused its discretion when it granted affirmative injunctive relief to nonparties in the absence of a certified class. This second argument, in contrast to the first, was a huge surprise and posed a significant issue. The trial court record was replete with evidence and argument—from the state in particular—that granting the plaintiffs’ requested relief would require the Medicaid agency to include applied behavior analysis as a covered treatment for other children with autism. At the very outset, the joint scheduling report included the plaintiffs’ uncontested statement that the case involved “systemic issues affect[ing] numerous indigent children in Florida who suffer from autism [and] that the ... relief they sought would remedy the alleged violation without moving for class certification.” At multiple junctures throughout the trial court proceedings, the Medicaid agency had highlighted the expense of covering applied behavior analysis for nonparties. Yet on appeal the agency suddenly argued that the district court had abused its discretion by ordering relief benefiting nonparties and that the relief had to be limited to the three plaintiffs. The agency strenuously objected to being ordered to provide applied behavior analysis as a covered benefit. Overturning this aspect of the district court’s injunction was vital to the state. Indeed, when the agency issued the trial court’s ordered notice to providers that applied behavior analysis was now a covered benefit, the notice explicitly stated that the service was being provided pursuant to a federal court order and that the agency “intends to appeal this ruling. If the ruling is overturned, Medicaid will cease to cover these services for the treatment of autism spectrum disorders.”

The Eleventh Circuit agreed and ruled that “the district court did not abuse its discretion in issuing a permanent injunction that … requires Medicaid coverage of this treatment.” The court did not vacate the injunctive and declaratory relief or limit the relief only to the plaintiffs, as the state had asked. Instead the appeals court ordered, sua sponte, that the final declaratory judgment and permanent injunction be published in their entirety.

While the evidence and ruling in K.G. concerned a single current treatment for autism, our takeaway message is not simply that children with autism are entitled to Medicaid coverage of medically necessary applied behavior analysis. Rather, our point is that your state Medicaid agency must not categorically exclude covering treatment for autism—or any other condition. Moreover, assuming the absence of an equally effective and less expensive treatment, the state must cover for children under 21 any treatment that falls within the Medicaid Act’s listing of covered services, will correct or ameliorate the child’s condition, and is not experimental. The determination of whether any future and as-yet-uncovered treatment for autism (or any other condition) is experimental should hinge on a reasonable standard for evaluating effective treatments for that particular condition—rather than on the type of “gold standard” insisted on by the Florida Medicaid agency in K.G. v. Dudek.

Miriam Harmatz’s Acknowledgments

After 30 years as legal services lawyer, I had the great fortune to be lead counsel on this almost perfect case. The clients; legal services cocounsel (two of whom are coauthors on this article); our experts; trial cocounsel, Neil Kodsi; appellate cocounsel, Jane Perkins of the National Health Law Program; and Autism Speaks as amicus on appeal and its pro bono counsel, Greg Wallance of Kaye Scholer LLP, were perfect. Of course, parts of the appeal were frankly bizarre, and implementation remains a challenge.

See this article as originally published and with footnotes at the Clearinghouse Community website