Megan Wall, esq
Summary of the iBudget Rule Challenge Final Hearing before DOAH
Earlier this month, I attended the three day rule challenge hearing before the Division of Administrative Hearings (“DOAH), at which Holland & Knight attorneys Gigi Rollini, Karen Walker and Matthew Mears represented developmentally disabled consumers challenging the third version of proposed rules to implement the iBudget. Like so many other consumers receiving services through Florida’s Developmentally Disability Services Waiver, all four Petitioners received significant proposed reductions under the iBudget further depleting their already inadequate funding. Because this case impacts every Waiver consumer, I am summarizing what occurred for those who could not attend.
In response to the rule challenge, APD hired five private attorneys, in addition to two in-house lawyers. The team of seven lawyers had significant resources at their disposal to defend the rule.
In Florida, rule challenges are fast track cases meant to be heard within 30 days from the filing of the petition. This hearing was scheduled the very first full week in July. The Holland & Knight lawyers fought hard (and succeeded) against APD’s attempts to delay the hearing. While it is difficult to go to a full-scale hearing so quickly, the Holland & Knight team was concerned that to give APD more time to prepare would drive up the expense of the case. Amazingly, despite the holiday weekend, the hearing began as originally scheduled, which was difficult but strategically important.
The case at its core is simple. The iBudget statute directs APD to set each client’s funding for services at an amount based on a statistically valid algorithm, and then consider if the person needs an increase to that amount because of extraordinary, one-time, temporary or changed needs. Florida law prohibits APD from enacting rules that run contrary to, or expand the implementing statute. Despite these legal limitations, the evidence in the case showed that:
1. APD used the algorithm only as a starting point for setting funding because, in APD’s own words, the algorithm APD designed produced “harsh results” for many. Because the algorithm does not result in an equitable allocation of the funds for DD waiver clients, APD created an elaborate and subjective “methodology” to decide whether or not to use the algorithm to set a client’s iBudget. This supplemental “methodology” treats some groups of people differently, even where clients in those groups have the exact same needs. Most glaringly, APD’s “methodology” targets existing DD waiver clients and singles them out to receive lower budgets than a new client with identical needs. Given that the law requires an “equitable allocation” of funds based on “the client’s level of need,” the evidence showed at the hearing that APD’s actual system is not consistent with the system described in the iBudget statute.
2, APD told the Florida Legislature that if it adopted an iBudget system for Florida, then 64% of existing Waiver clients would see an increase in their budgets as a result of the iBudget. However, APD later made a “policy decision” that funding increases simply would not be given and the algorithm would be used only to reduce, not increase a client’s allocation of funds.
3. Former University of Florida Professor Dr. McClave (an expert in statistics and econometrics who has written text books on these types of models and created successful Florida Medicaid models) reviewed and tested the APD algorithm and testified on behalf of the Petitioners that it was statistically unreliable, including that APD’s chosen algorithm had a margin of error of over 40%! He said that this is not a model based on a valid statistics approach, and that the model was unacceptable in his field. He memorably testified that APD’s chosen model is as accurate as “throwing darts at a wall of numbers.”
4. APD identified no specific legal authority that allowed it to implement the iBudget the way it is being done. APD could point to no legal authority to use a statistically unreliable algorithm, to reduce amounts determined by the algorithm, to use a “decision tree” process to reduce existing client’s budgets, to single out existing clients and assign them lower budgets, or to reduce budgets without any final rules at all. Recognizing the shortcomings of its own position, APD admitted it asked the Florida Legislature in 2012 to change the iBudget statute to conform to how APD was actually implementing the program. The Legislature declined to make the revisions. Regardless, APD simply continued to implement the system using its system rather than one consistent with the statute.
5. APD’s statistics expert (Dr. Niu), who designed the algorithm, testified that he did what his client (APD) asked him to do and he did his best. He did not consider other state’s algorithms. His reason was that he did not need to because he “only relies on [him]self.” He used a very few variables (about seven), and assumed he could drop out some variables (such as the entire physical status portion of the QSI) on the basis that other, similar variables (such as the behavioral status portion of the QSI) tended to measure about the same thing—not realizing that every DD client doesn’t have both physical and behavioral needs, or does not have them in equal force. The result—those with significant physical needs, but no significant behavioral needs, will get substantially lower budgets than those with equally costly behavioral needs.
Also, it was shocking to learn that Dr. Niu “meant to do testing” to ensure the validity of the model, but then just did not get around to performing the type of testing necessary to ensure the validity of the model. He did no testing against other years’ budgets (as is commonly done in statistics, apparently), or “log” method testing (as is typically done in statistics, apparently, when the model is a budget allocation or distribution model). This was the case even though Dr. Niu has done a much simpler distribution model for the Florida Department of Transportation, and he did do “log” method testing for that model.
Dr. Niu also had no response to the UF Professor’s testimony that the algorithm was not statistically valid. In fact, he said complimentary things about the other expert, Dr. McClave, and hoped his model “would get better over time.”
Note: APD testified that it will analyze each client’s iBudget every 12 months to determine what further cuts can be made. When APD decides it is appropriate, it may (or may not) re-run the inaccurate algorithm in any year it wants. Under the proposed rules, APD has authority to use the proposed subjective methods in the rule to further reduce client services at its discretion. However, for future years, the processes which permit increases for extraordinary, one-time, temporary and changed needs will no longer be available to the person whose services are being reduced. Of course, if the algorithm dictates an increase for a DD client, APD will still have the ability to find that award is “really just not medically necessary,” since APD continues to be the sole determiner of what is and what is not medically necessary.
APD’s arguments in response to Petitioners’ case were as follows: We worked really hard on this system; the Tier system has expired and, if this new system is thrown out, APD cannot return to the Tier system; we did our best; we do have authority to do what we did (without specific reference to any law or statute); and, mainly, the Florida Legislature is very happy with APD because APD has substantially cut the cost of the DD waiver program with APD’s interpretation of the iBudget system! In fact, APD testified that the Legislature “rewarded” APD as a “thank you” for doing such a good job with two additional appropriations totaling $60 million, which APD used internally and to bring people off the waitlist. [The judge asked if APD should have put that into the iBudget calculations and reappropriated the money to the existing DD clients, and APD said no; the Judge asked if APD had re-run the algorithm since APD first did it (one time) based upon each years' new appropriation, and APD said, “No, and we are not going to.”]
And, finally, there was the real client example offered at the hearing: An amazing and articulate father of a DD 26-year-old came in and told his son’s story—with a 20 IQ, he lives in a home inherited by his father, which permits his son and two other DD clients to have extremely reduced rent. All need 24-hour supervision, which they can barely piece together between the three of them—with less and less money each year, less and less of a rate of payment each year. And, now, all three have been given iBudget reduction notices (his son’s, alone, is an over $8,000 reduction under iBudget). They were barely making it since their budget is already to the bone (even with the father’s substantial assistance). With the cuts now proposed, there is no hope and nowhere for them to live, except an institution were these cuts to be implemented.
With that, the story of this proposed iBudget system was told. The agency admitted it was not following the statute to implement the system and the model being used cannot be shown to be statistically reliable. They agreed that cuts may continue each year. Even if every client could bring into an individual hearing an expert statistician to object to algorithm used to calculate the proposed reduction, the DCF Fair Hearing Officers lack the authority to find the rules invalid. Only DOAH has that authority. Thus, the reason why this case is before DOAH for a decision, and why it is so crucial a test case.
A decision will be issued around the end of August or early September. The lawyers for the Petitioners agreed to file this case without sufficient cash in hand on the hope that once the DD community knew the truth about the proposed iBudget system, they would join together to support efforts to compel APD to use an algorithm that is statistically reliable and a process that follows the law. The Petitioners’ lawyers brought this David and Goliath case, did it in an unbelievably short period of time and, as a result, kept the costs to a bare minimum ($225,000). Our team, led by Gigi Rollini, presented a clear, articulate, and persuasive case. And, now, we have to pay her! [And did I mention that in addition to APD's seven attorneys and the work they all created, there was also surprise evidence sent by APD at 8:30 am the final day of trial, and surprise witnesses on the final day of witness disclosure—less than one week before hearing—including an expert APD brought in at the last minute and paid $85,000 for his service. And despite the additional work and expense tied to this surprise witness, he ultimately did not even testify because he had nothing to say after hearing Dr. McClave.]
Please contribute now to help us to meet our $125,000 goal! To get there, we are asking that you consider donating at least at the $10,000 level. We also specially thank those increasing this to the $20,000 level—we could not do this without you!
Please be a part of the solution! There are two easy ways to contribute to the APD iBudget Proposed Rule Challenge:
(1) Checks may be made out to Holland & Knight LLP, and sent to ATTN: Gigi Rollini, Post Office Drawer 810, Tallahassee, Florida32302; or
(2) Payment by wire transfer to:
Wells Fargo Bank N.A.
420 Montgomery Street
San Francisco, CA 94104-1207
ABA # 121000248
Account # 2090002390441
For Credit to: Holland & Knight, LLP
Please indicate “APD iBudget Proposed Rule Challenge” as the Beneficiary Reference in the wire.
–Megan Wall, Managing Attorney, St. Johns County Legal Aid