Findings Of Fact At all times relevant to these proceedings, the Petitioner was an applicant for licensure as a life including variable annuity and health insurance agent. On or about February 2, 1994, the Petitioner was scheduled to take the life including variable annuity and health insurance agent examinations at Florida Technical College in Orlando, Florida. On February 2, 1994, the Petitioner chose to take only the life including variable annuity examination. He did not receive a passing grade on this examination. On February 4, 1994, the Petitioner returned to the Florida Technical College after the test had started. He attempted to use another test admittance authorization card to be able to re-test. At that time, the Petitioner was advised by Tom Bastedo, Test Site Computer Operator, that he could not use the authorization card to gain admittance to the examination since the Petitioner had already taken the life including variable annuity examination and the computer scoring system would report that the examination had already been scored. The Petitioner became angry and tore up the card. The Petitioner then discussed his personal situation with Tom Bastedo, which included a reference to some "emeralds" that the Petitioner owned. The Petitioner removed from his briefcase what appeared to be real emeralds and offered them as a bribe to Mr. Bastedo if he would make it look like the Petitioner had passed the examination which he had previously failed on February 2, 1994. Mr. Bastedo refused the offer and advised the Petitioner that he must reapply to be tested. The Petitioner hung around the testing site for awhile and then left the area. This incident was reported immediately by Mr. Bastedo to his supervisors. On or about March 9, 1994, the Petitioner's application for license as a life including variable annuity and health insurance agent was denied because of the events which had occurred at the Florida Technical College on February 4, 1994. Petitioner's explanation of the events that occurred on February 4, 1994 is not credible.
Recommendation Based on the foregoing Findings of Fact and Conclusions of Law, it is RECOMMENDED that the Department of Insurance and Treasurer enter an Order denying the license application of Dario Albert Alvarez for licensure as a life including variable annuity and health insurance agent. DONE AND ENTERED this 27th day of September, 1994, in Tallahassee, Florida. DANIEL M. KILBRIDE Hearing Officer Division of Administrative Hearings The DeSoto Building 1230 Apalachee Parkway Tallahassee, Florida 32399-1550 (904) 488-9675 Filed with the Clerk of the Division of Administrative Hearings this 27th day of September, 1994. APPENDIX Petitioner did not submit proposed findings of fact. Respondent's proposed findings of fact. Accepted in substance: paragraphs 1-7. COPIES FURNISHED: Dario Albert Alvarez 13329 Laver Lane Orlando, FL 32824 John R. Dunphy, Esquire Division of Legal Services 612 Larson Building Tallahassee, FL 32399-0333 Tom Gallagher State Treasurer and Insurance Commissioner Department of Insurance and Treasurer The Capitol, Plaza Level Tallahassee, FL 32399-0300 Bill O'Neil, Esquire General Counsel Department of Insurance and Treasurer The Capitol, PL-11 Tallahassee, FL 32399-0300
Findings Of Fact Respondent is Hooshang Hooshmand, a licensed physician at all times relevant to these proceedings. He was issued license number ME 0021496 by the State of Florida. Medicare is a program of the U.S. Department of Health and Human Resources which is administered by the Health Care Financing Administration (HCFA). The program allows for third party payment, by the federal government, for diagnostic programs and medical treatments administered on an inpatient and outpatient basis to individuals eligible for medicare coverage. Among providers of medical services, only licensed physicians may be paid by the program for rendition of services. Others who may be reimbursed include health care professionals, such as durable medical equipment suppliers, as well as patients themselves in the instance of medical services rendered by a nonparticipating physician. On October 9, 1987, in the United States District Court for the Southern District of Florida, Respondent was convicted, after a jury trial, of ten counts of submission of fraudulent medicare claims in violation of Title 18, U.S.C. Section 287 and Section 2. He was also convicted on 11 counts of devising a scheme to defraud by mail, the U.S. Department of Health and Human Resources in violation of Title 18, U.S.C., Section 1341 and Section 2. Respondent's sentence upon his conviction included a term of 18 months imprisonment and five years probation upon completion of imprisonment and any parole period; payment of restitution in the amount of $3,101.24; payment of a fine of $250,000; payment of an assessment of $300; performance of 5,000 hours of community service during the five year probationary period following imprisonment. The verdict and sentence are presently on appeal to the United States Circuit Court of Appeals for the Eleventh Judicial Circuit. The expert testimony of Michael Gutman, M.D., a specialist in forensic psychiatry in the State of Florida, establishes that the practice of medicine in Florida encompasses not only a physician's technical competence; but also the relationship between a physician and the patient. Such a relationship is based upon trust and honesty. While the physician's expectation of payment for services is part of the patient/physician relationship, fraudulent billing for those services by the physician to either the patient or a third party payor directly affects the practice of medicine through its impact on that relationship. A fraudulent billing scheme, such as that of which Respondent was convicted, introduces dishonesty to the physician/patient relationship and prevents a proper evaluation of the patient in favor of a methodology permitting fraudulent billing. Such methodology would necessarily be one chosen to permit fraudulent billing in a way which would escape detection; a choice not necessarily in the best interest of the patient. Gutman's testimony also provides an adequate record upon which to find that fraud, such as that reflected by Respondent's criminal conviction, also directly relates to the ability to practice medicine because the physician's professional judgement and ethical standards are involved. Such judgement has a direct bearing on the ability to practice medicine. How that judgement is exercised could very well affect the life of the patient in some situations. While it is found Respondent's conviction of fraud in the use of the billing apparatus in his practice directly relates to professional judgement and the ability to practice medicine, there has been no showing that the Respondent's judgemental aberration at that time detrimentally affected his patients' health or his technical competence. In mitigation of the charge in the administrative complaint, Respondent provided testimony of witnesses establishing his technical competence and expertise in his areas of specialization; his extremely impressive professional credentials; the high regard in which he is held by certain of his peers and patients; and his previously unblemished record in the practice of medicine. Respondent also provided testimony of witnesses establishing the complexity of medicare billing and the fact that many physicians, while holding ultimate responsibility for the accuracy of such billing, delegate this task to subordinates. Testimony of Respondent establishes that the complexity of medicare procedures played a major role in his violation of the legal requirements in that system of reimbursement and is partially to blame for his criminal conviction. The testimony of Eleanor Breckner, offered by Petitioner to rebut Respondent's testimony, is not credited. In addition to Beckner's demeanor while testifying, her testimony is diminished in view of her admission that she committed perjury and embezzlement on previous occasions. Beckner also admitted to incidents of attempted suicide indicative of mental instability. Her testimony is not credited with any probative value.
Recommendation Based on the foregoing findings of fact and conclusions of law, it is RECOMMENDED that Respondent pay an administrative fine of $2,500 and that his license be placed on probation for a period of two years upon terms and conditions to be established by the Board of Medicine. It is further recommended that a condition of such probation require the satisfactory completion by Respondent of a course of study designed to provide him the information and skills necessary to properly comply with medicare reimbursement procedures. DONE AND ENTERED this 3rd day of March, 1989, in Tallahassee, Leon County, Florida. DON W. DAVIS Hearing Officer Division of Administrative Hearings The Desoto Building 1230 Apalachee Parkway Tallahassee, Florida 32399-1550 (904)488-9675 Filed with the Clerk of the Division of Administrative Hearings this 3rd day of March, 1989. APPENDIX The following constitutes my specific rulings, in accordance with section 120.59, Florida Statutes, on findings of fact submitted by the parties. RESPONDENT'S PROPOSED FINDINGS 1.-4. Addressed. 5.-6. Unnecessary to result. 7. Addressed in part, remainder unnecessary. 8-10. Unnecessary to result. 11. Addressed in part; remainder unnecessary. 12.-14. Unnecessary to result. Addressed in part; remainder unnecessary. Adopted by this reference. Rejected, not supported by the evidence. Unnecessary to result. 19.-20. Not supported by the weight of the evidence. Unnecessary to result; also cumulative. Adopted by this reference. Rejected as cumulative. Not supported by the weight of the evidence. Unnecessary to result and not relevant. 26.-27. Unnecessary to result. Unnecessary to result and cumulative. Addressed in part; remainder unnecessary to result. 30.-31. Unnecessary to result; cumulative. 32. Reject, not supported by weight of the evidence. 33.-36. Rejected, not relevant. 37.-41. Unnecessary to result. 42.-43. Addressed in part, remainder unnecessary. 44.-45. Unnecessary to result reached. 46. Addressed in part, remainder unnecessary. 47.-50. Unnecessary to result. Unnecessary and cumulative. Unnecessary to result. Rejected on basis of relevancy. 54.-56. Addressed in part, remainder unnecessary. Unnecessary to result reached. Rejected, not relevant. Unnecessary to result reached. Rejected, not relevant. 61.-67. Unnecessary to result reached. Rejected, not credible and not supported by the weight of the evidence. Also a legal conclusion. Addressed. Unnecessary to result. Adopted by this reference. Adopted in substance. Rejected as a legal conclusion. PETITIONER'S PROPOSED FINDINGS 1.-7. Adopted in substance. 8.-9. Addressed. COPIES FURNISHED: Jonathan King, Esq. Department of Professional Regulation 130 North Monroe Street Tallahassee, FL 32399-0750 Joesph C. Jacobs, Esq. Melissa Fletcher Allaman, Esq. 305 South Gadsden Street P.O. Box 1170 Tallahassee, FL 32302-1170 Roy L. Glass, Esq. 3000-66th Street North Suite B St. Petersburg, FL 33710
The Issue The issue in this case is whether proposed rules 65G-4.0210 through 65G-4.027 (the “Proposed Rules”) are an invalid exercise of delegated legislative authority as defined in section 120.52(8), Florida Statutes. (Unless specifically stated otherwise herein, all references to Florida Statutes shall be to the 2012 codification.) Specifically, Petitioners assert that the Proposed Rules (1) enlarge, modify, and contravene the specific provisions of the law they purport to implement; (2) contain vague and inadequate standards that vest unbridled discretion in the Agency for Persons with Disabilities (the “Agency” or “APD”); (3) are arbitrary and capricious; and (4) exceed the grant of rulemaking authority in section 393.0662(9), Florida Statutes. Petitioners further argue that, (5) APD failed to follow applicable rulemaking procedures required by sections 120.54(3) and 120.541, Florida Statutes, because APD failed to provide a Statement of Estimated Regulatory Costs (“SERC”) as a part of the rulemaking process.
Findings Of Fact Each of the Petitioners is a recipient of services under the DD Waiver Program. For example, Petitioner Z.L. is a 26- year-old male who was born with Cri-du-Chat syndrome, a fifth chromosome abnormality. As a result, Z.L. is low-functioning, with a non-measurable IQ level (but likely well below the level designating mental retardation). Z.L. speaks only a few words and communicates with some sign language. He is ambulatory, but he is totally dependent on others for all activities of daily living. Z.L. also has some extreme behavioral issues, including self-abuse and physical abuse of others. He lives in a private residence with two other developmentally disabled men. The home where they reside belongs to the family of K.L. (Z.L.’s father and legal guardian). K.L. rents the home for Z.L. and the other two men at less than its actual market value. (The home is a 1,500 square foot home located on 15 acres. K.L. pays about $600 per month rent; the home could rent for two or three times that much.) Z.L. receives the following services under the DD Waiver Program: 24-hour assistance with activities of daily living; behavioral analysis through a certified behavior analyst; and personal care assistance. The cost of his care plan for the previous year was $61,824.22 (i.e., that was the amount paid by the DD Waiver Program). Z.L.’s father and mother are unable to care for Z.L. in their home. The father is CEO of a bank and is involved in other businesses as well. The mother recently suffered closed head injuries as a result of a bicycle accident. She must be cautious about any further head injuries and fears that Z.L.’s aggression could result in physical harm to her. As a result of the implementation of the iBudget process, APD is proposing to reduce Z.L.’s care plan by the sum of $8,175.98. Under the iBudget process, Z.L. has the right to challenge the reduction of his care plan amount in a Fair Hearing before a Department of Children and Families Hearing Officer, which he has done. K.L. has expended about $6,000 in legal fees to contest the reduction of Z.L.’s care plan amount under the new iBudget system. He expects that if the matter goes to appeal, he will expend as much as $70,000 more in legal fees. K.L. has also hired a lawyer for one of Z.L.’s roommates.1/ APD is the state agency responsible for distributing funds from the DD Waiver Program. Prior to implementation of the iBudget process, APD used a four-tier system to provide the level of funds each client would receive.2/ The tier system was more rigid in its application than the iBudget system. Under the tier system, there were strict funding policies in place. For example, if dollars were allocated toward a specific service, e.g., transportation, those dollars could not be used for any other service, such as companion care or personal care. As will be discussed more fully below, the funds provided in the iBudget process are more flexible regarding services they can purchase. The DD Waiver funds administered by the Agency are the funds of last resort. If a service received by a client can be paid for by another agency or source of payment, those must be utilized before the Agency can allocate funds for the service. Development of the iBudget System The 2010 Florida Legislature mandated creation of an iBudget process for distributing funds from the DD Waiver Program. Section 393.0662(1) states in pertinent part: The agency shall establish an individual budget, referred to as an iBudget, for each individual served by the home and community- based services Medicaid waiver program. The funds appropriated to the agency shall be allocated through the iBudget system to eligible, Medicaid-enrolled clients . . . . In developing each client’s iBudget, the agency shall use an allocation algorithm and methodology. The algorithm shall use variables that have been determined by the agency to have a statistically validated relationship to the client’s level of need for services provided through the home and community-based services Medicaid waiver program . . . . The allocation methodology shall provide the algorithm that determines the amount of funds allocated to a client’s iBudget. The agency may approve an increase in the amount of funds allocated, as determined by the algorithm, based on the client having one or more of the following needs that cannot be accommodated within the funding as determined by the algorithm and having no other resources, supports, or services available to meet the need: An extraordinary need that would place the health and safety of the client . . . in immediate, serious jeopardy . . . . A significant need for one-time or temporary support or services . . . . A significant increase in the need for services after the beginning of the service plan year . . . . The agency shall reserve portions of the appropriation for the home and community- based services Medicaid waiver program for adjustments required pursuant to this paragraph . . . . A client’s iBudget shall be the total of the amount determined by the algorithm and any additional funding provided pursuant to paragraph (b). A client’s annual expenditures for home and community-based services Medicaid waiver services may not exceed the limits of his or her iBudget. The total of all clients’ projected annual iBudget expenditures may not exceed the agency appropriation for waiver services. In response to the statutory mandate, the Agency sought input from “stakeholders,” i.e., individuals and families receiving services, family care counsel groups, various provider groups, and organizations such as the Association of Retarded Citizens and the like. APD also looked at how other states had addressed the issue of fund distribution to developmentally disabled individuals. The Agency hired consultants to help make the process as equitable and fair as possible within the limits of its finite budget. One of the Agency’s hired consultants was Dr. Xufeng Niu, chair of the statistics department at Florida State University. Dr. Niu is a recognized expert in the field of statistics and had used his expertise in many areas, including transportation issues such as railroad crossing safety and environmental issues for the Department of Environmental Protection. Dr. Niu has been an academician and consultant since obtaining his Ph.D. in statistics from the University of Chicago in 1991. Dr. Niu’s testimony was extremely credible. APD hired Dr. Niu to develop an algorithm which would be the key feature to any individual budget calculation. APD’s goal in developing the algorithm was to create a formula fitting data patterns of past expenditures, then to mathematically replicate decisions that were made to establish a client’s prior budget amount. Dr. Niu, by way of statistical modeling techniques, developed certain factors which could be utilized by the Agency in determining which clients would receive funds for specific services. Using a catalogue of predictors or variables derived from information provided to him by the Agency, Dr. Niu built a tool to predict what each client’s cost for needed services would be. A Bell Curve was used to keep the application of the variables more symmetrical. In order to effectuate this desire, Dr. Niu utilized a form of “transformation” referred to as the Box-Cox Transformation Family. The Box-Cox Method involved raising data to a different mathematical power as a means of analyzing and applying the data. Dr. Jim McClave, who operates a statistical consulting firm, is an expert statistician and econometrician. His work involves regular stints as an expert in legal proceedings such as this rule challenge matter. His testimony was credible, but less persuasive than that of Dr. Niu.3/ Dr. McClave would have used a log transformation method rather than the Box-Cox method relied upon by Dr. Niu. However, while not discounting the log transformation method, Dr. Niu competently testified that the Box-Cox worked best in this particular case. After the transformation process, it was necessary to narrow down the number of variables to be used. Dr. Niu ultimately decided to use nine specific variables, including: the client’s living setting; whether the client is an adult; the client’s score on the six elements set forth in the Questionnaire for Situational Information (“QSI”) which was provided to all potential recipients of services; the client’s score on the 11 elements in the functional summary section of the QSI; and the client’s score on each of three specific elements in the QSI related to transfers (ability to transfer or change position), hygiene, and capacity for self-protection. Not all variables are necessarily useful and having too many variables causes over- fitting, i.e., trying to fit every situation into a perfect model, which simply is not possible. In fact, it is better to have fewer variables as long as sufficient data can be captured. A statistician must reach a balance on the number of variables in order to find the best model for each project. Dr. Niu’s affirmation of the variables he used is credible. Dr. Niu utilized the Generalized Information Criterion (“GIC”), a method of finding the best set of predictors when creating an algorithm. GIC is a criterion that tries to balance the model by carefully adding more variables without overpopulating the model with too many variables. GIC was used by Dr. Niu in conjunction with the concept of R-squared. That concept is a statistical measure of how well an algorithm fits the data in order to test how well the model predicts. The algorithm developed for use in the Proposed Rules has an R- squared value of .6757, meaning that it accounts for about 68 percent of the variation in the population of APD clients’ DD Waiver expenditures. By contrast to the GIC and R-squared approach, there is in the field of statistics a tool referred to as Residual Standard Error. This tool helps determine whether a model is predicting within two standard deviations and thus has a measure of certainty. The algorithm proposed by APD did not utilize the Residual Standard Error tool, relying instead on the combination of GIC and R-squared. Based upon Dr. Niu’s testimony, APD’s reliance on those tools is reasonable. Dr. Niu developed a number of models for possible use in the iBudget process, settling at last on Model 7b. The model was then applied to the pool of clients who would be affected by the new iBudget system. The client pool contained a large number of different situations and scenarios, as each client and client family is unique despite some similar developmental issues. As a result of these differences, there were cases in which a particular client -- because of his or her needs, or those of his or her family -- did not fit the model. These cases were called “outliers” and had to be treated differently by the Agency. Of the total group of some 26,000 clients, 9.37 percent, or about 2,400 clients, were deemed outliers. Dr. McClave criticized this percentage of outliers, but Dr. Niu's substantiation of the percentage is credible. Dr. Niu utilized actual expenditures by APD for DD Waiver Program clients during the 2007-2008 fiscal year as an indicator of what APD had faced in the past. Those data were recent enough in time to be linked to current assessment data for the clients and to be assigned scores from the QSI. APD also found that the 2007-2008 data more accurately reflected service needs compared to other recent years because the data pre-dated the implementation of the more restrictive Tier system. Dr. Niu did not use clients with less than one year of claims because they may not project the client’s actual annual expenditures. Dental services, environmental services, and durable medical equipment purchases were excluded because they are generally a once-a-year purchase. Four of Florida's 67 counties were excluded from the calculations because they had a much higher cost of living than the rest of the state. Mismatches and clerical errors in clients’ records were also taken into consideration. Age was used as a predictor, but after trial and error Dr. Niu decided upon a single division, i.e., persons under 21 years of age versus persons 21 or older. The rationale was that people under 21 receive services from other sources, like the public school system, for example. Persons over 21 begin to require more services as they age. Dr. Niu considered more factors than just the mathematical statistical accuracy. His extensive work resulted in the best model out of many possibilities. Transportation needs and costs were considered during the stakeholder meetings as a factor to be considered when discussing possible variables. Dr. Niu attempted to use a transportation index in his models, but that resulted in a negative coefficient which is less valid statistically. Applying the current year’s transportation costs did not work. It was also impossible to apply a portion of a year’s transportation costs as an indicator of the entire year’s transportation costs. And, because transportation costs constitute only about 1.5 percent of overall expenses, it was reasonably determined that such costs could be handled by way of an extra needs review. Upon completion of the iBudget system, it was implemented and introduced to all eligible DD Waiver clients. The program was introduced in “waves,” i.e., not all DD waiver clients being served by APD received their iBudgets at one time. Rather, the new system was phased in over time. How the iBudget System Is Employed APD sends an information packet to each client, i.e., each person seeking services to be paid for under the DD Waiver Program. This information packet, called a Welcome Guide, is meant to help the client understand the new system. The Welcome Guide provides a large amount of information, plus education and training possibilities as well. It is understandably difficult to absorb all of the information contained in the packet, but APD opted for completeness rather than over-simplifying the information. Z.L.’s father, who is a licensed attorney and CEO of a bank, expressed difficulty understanding the information contained in the Welcome Guide. However, he testified that he has "some kind of memory block" about DD Waiver services. It is understandable that this would be a difficult thing for a parent to review. The first step of the process for requesting funds for services under the iBudget system is to have the client complete a QSI form. After the QSI assessment is done, the second step of the process is for the Agency to run its algorithm using the previously discussed variables such as age, living arrangement, behavioral status, functional status, and the responses to various personal questions concerning the client. Running the algorithm then creates a dollar value for the services deemed appropriate for the client. The cost of the services is then related back to the appropriation of funds received by APD from the Legislature for providing all needed services. Each client’s sum for needed services is then given a pro rata reduction (or, theoretically, an increase) based on the total funds available to APD. There are then adjustments which can be made to the algorithm amount. For example, if the algorithm amount for a client was greater than the amount set forth in the client’s existing care plan, that client’s “algorithm amount” was reduced to the existing care plan amount, at least temporarily pending further possible actions under the iBudget process. There are specific services identified in the Proposed Rule (at 65G-4.0212(b)(2)), which are indicative of certain health and safety needs. If a client needs any of those services and the cost of those services is greater than the algorithm amount, the greater sum will be substituted. If the algorithm amount was less than the client’s care plan amount but within $1,000 of the existing care plan amount, then the care plan amount was used as the “algorithm amount.” This $1,000 buffer will necessarily mean that a client whose care plan amount is $999 more than the algorithm amount may be treated differently from a person whose care plan amount is $1,001 more than the algorithm amount. Still, the decision to employ a $1,000 threshold is generally reasonable as APD attempts to maintain a sufficient care plan allocation despite the change in systems. APD reasonably believes it would be more time-consuming and costly to deal with changes of $1,000 or less than to simply accept the prior care plan amount (which was based upon the client's needs). If the algorithm amount is less than the amount in the client’s existing care plan, then APD determines whether the reduction is greater than 50 percent of the existing care plan amount. If so, the algorithm amount is raised to an amount equal to at least 50 percent of the existing care plan amount. After application of the above-reference factors and -- if warranted -- adjustments are made, the client is provided an amount which is referred to as the “Target Allocation.” The fourth step in the process is for APD to provide the Target Allocation amount to the client and WSC. Step five of the process is a review to determine whether, notwithstanding the algorithm amount, a client has extra needs that warrant an increase in their ultimate allocation of funds for services. This is called the Extraordinary Needs Review. The first phase of this step is an allocation implementation meeting (AIM), wherein the client is advised about the changes --if any -- to his/her care plan. The client and his or her waiver support coordinator (WSC) are given information about how the reductions may be handled, e.g., that under the iBudget it might be possible to utilize funds to pay for one service even if they are allocated for another service. Or, there may be ways under the iBudget system to merge two or more services into one. One example of that is that in-home personal service caregivers may be allowed to perform other tasks, e.g., they may be able to provide services outside the home setting. After almost a full year of implementing the iBudget system, this portability of funds from one service to another has proven to be one of the most appreciated functions of the new process by waiver support coordinators. If the client and WSC agree that the service needs can be met by the Target Allocation, that amount becomes the client’s iBudget Allocation amount. If the client and WSC do not think the Target Allocation amount is sufficient to meet the service needs, the AIM form is completed and sent to APD for further review. If the health and safety of the client, client’s caregiver, or the public is placed in immediate jeopardy without an increase in the allocation, then an increase will be approved. APD then gives the client notice as to its decision and the final iBudget Allocation is provided. This constitutes step six of the process. Subsequent to setting and providing notice of the final iBudget Allocation, a client may seek supplemental funding for significant one-time or temporary needs. If a significant increase in need for services arises after the beginning of a plan year, a process exists for further consideration of the client’s needs. For new clients, i.e., those who do not have an existing care plan when the iBudget is applied to them, the process is slightly different. First there is an eligibility determination (which has already occurred for existing clients). The client then responds to the QSI. The algorithm is calculated to form the target allocation for the new client. An extra needs review is then performed to make sure that all health and safety needs are being met. It is possible that a new client with exactly the same condition, circumstances, and needs as an existing client (albeit an extremely unlikely occurrence), could receive a larger amount under the iBudget than the existing client. If both clients were assigned exactly the same score under the algorithm, but the existing client’s allocation amount were larger than the care plan amount under the Tier system, then the existing client’s allocation would be reduced. There would not be a concomitant reduction of the new client’s allocation. Although Petitioners pointed out this alleged flaw, no remedy was suggested that would make it possible for APD to make the treatment of two similarly situated clients more equal. The iBudget system is not flawless, but it is an admirable effort toward equality of application to all “clients.” The Agency did not set aside or reserve any portion of their allocation from the Legislature as a Reserve Fund, per se. Rather, APD uses the reserve fund concept as a management tool to be used when making adjustments to an individual client’s final allocation of funds. Thus, during the AIM process or the Supplemental Cost Funding phase, APD might raise a client’s allocation based on funds it has “reserved” under the algorithm calculation. Statement of Estimated Regulatory Costs APD published the initial proposed rule on August 3, 2012. The publication included a statement that the Agency had determined there would not be an adverse impact on small business nor would it increase regulatory costs in excess of $200,000 within one year. Petitioners’ contention that clients may have difficulty understanding the welcome packet information and may challenge iBudget Allocations by way of fair hearings does not establish the necessity for SERC.
The Issue The issues for disposition in this case are whether: (1) proposed rules 65G-4.0213 through 65G-4.0218 of the Florida Administrative Code (“the Proposed Rules”) constitute an invalid exercise of delegated legislative authority as defined in section 120.52(8), Florida Statutes (2015)1/; and whether (2) the Agency for Persons with Disabilities (“APD”) failed to follow applicable rulemaking procedures in seeking to adopt the Proposed Rules.
Findings Of Fact The Parties Petitioners face profound mental and physical challenges, and they receive services through the Home and Community-Based Services Medicaid Waiver Program for individuals with developmental disabilities (“the Waiver”). APD’s mission is to help people with developmental disabilities live and receive services in their communities. § 393.066(1), Fla. Stat. The Waiver Medicaid is the primary funding source for services to income-qualified, developmentally disabled individuals. See Russell v. Ag. for Pers. With Disab., 929 So. 2d 601, 602 (Fla. 1st DCA 2006). See also § 393.063(9), Fla. Stat. (defining “developmental disability” to mean “a disorder or syndrome that is attributable to intellectual disability, cerebral palsy, autism, spina bifida, or Prader-Willi syndrome; that manifests before the age of 18; and that constitutes a substantial handicap that can reasonably be expected to continue indefinitely.”). States participating in Medicaid must comply with federal requirements governing the program, but the Medicaid Act allows states to obtain waivers from those requirements. Russell, 929 So. 2d at 603. Florida provides services to APD’s clients2/ through the Waiver, and APD administers the Waiver. APD’s goal is to enable its clients to live productive lives in their communities rather than in costly institutions. See § 393.066(1), Fla. Stat. (mandating that APD “shall plan, develop, organize, and implement its programs of services and treatment for persons with developmental disabilities to allow clients to live as independently as possible in their own homes or communities and to achieve productive lives as close to normal as possible.”). Even though the services provided to APD’s clients are funded through the Waiver, those services must still comply with certain federal requirements. For example, all Waiver services must be medically necessary. Also, Waiver funding is funding of last resort. Therefore, a Waiver recipient must exhaust all other available resources from his or her state, local community, and family prior to receiving waiver funds. See § 393.0662(4), Fla. Stat. (mandating that “[a] client must use all available services authorized under the state Medicaid plan, school-based services, private insurance and other benefits, and any other resources that may be available to the client before using funds from his or her iBudget to pay for support and services.”). APD currently serves approximately 30,600 clients through the Waiver. Due to budgetary constraints, approximately 21,000 people are on a “wait list” for receiving Waiver services. The iBudget Statute In 2007, the Florida Legislature directed APD to develop and implement a comprehensive redesign of the waiver system. Moreland v. Ag. for Pers. with Disab., 19 So. 3d 1009, 1010 (Fla. 1st DCA 2009). This new system, referred to as the “Tier System,” placed each client into one of four different “tiers” based on that client’s need for services and the cost of providing those services. See § 393.0661(3), Fla. Stat. See also Moreland, 19 So. 3d at 1011 (noting that “[t]iers two, three, and four each have a different annual monetary cap; thus, the level of services available for each individual is contingent upon the tier into which the individual is placed.”). APD responded by promulgating proposed rules to implement the Tier System. However, the First District Court of Appeal invalidated them. Moreland, 19 So. 3d 1009. In 2010, the Florida Legislature shifted to the iBudget System. As discussed in the preamble to section 393.0662, Florida Statutes, the iBudget System was intended to assist APD with its budgeting so that the agency could avoid the budget shortfalls that had plagued the Tier System. The Legislature also intended to provide APD’s clients with greater flexibility in choosing the services that would enable them to remain in their local communities: The Legislature finds that improved financial management of the existing home and community-based Medicaid waiver program is necessary to avoid deficits that impede the provision of services to individuals who are on the waiting list for enrollment in the program. The Legislature further finds that clients and their families should have greater flexibility to choose the services that best allow them to live in their community within the limits of an established budget. Therefore, the Legislature intends that [APD], in consultation with the Agency for Health Care Administration, develop and implement a comprehensive redesign of the service delivery system using individual budgets as the basis for allocating the funds appropriated for the [Waiver] among eligible enrolled clients. The service delivery system that uses individual budgets shall be called the iBudget system. § 393.0662, Fla Stat. The increased flexibility that clients have under the iBudget system enables them to obtain necessary services without getting prior approval from APD. The Legislature also expressed a commitment to keeping clients in their local communities rather than in costly institutions: The Legislature finds and declares that existing state programs for the treatment of individuals with developmental disabilities, which often unnecessarily place clients in institutions, are unreasonably costly, are ineffective in bringing the individual client to his or her maximum potential, and are in fact debilitating to many clients. A redirection in state treatment programs for individuals with developmental disabilities is necessary if any significant amelioration of the problems faced by such individuals is ever to take place . . . . Further, the greatest priority shall be given to the development and implementation of community- based services that will enable individuals with developmental disabilities to achieve their greatest potential for independent and productive living, enable them to live in their own homes or in residences located in their own communities, and permit them to be diverted or removed from unnecessary institutional placements. § 393.0662, Fla. Stat. In order to accomplish the goals described above, section 393.0662(1), provides that “[t]he agency shall establish an individual budget, referred to as an iBudget, for each [client]. The funds appropriated to the agency shall be allocated through the iBudget system to eligible, Medicaid- enrolled clients.” Also, APD must utilize an algorithm in establishing each client’s iBudget: In developing each client’s iBudget, the agency shall use an allocation algorithm and methodology. The algorithm shall use variables that have been determined by the agency to have a statistically validated relationship to the client’s level of need for services provided through the [Waiver]. The algorithm and methodology may consider individual characteristics, including, but not limited to, a client’s age and living situation, information from a formal assessment instrument that the agency determines is valid and reliable, and information from other assessment processes. §393.0662(1)(a), Fla. Stat. The algorithm referred to in section 393.0662(1)(a) (“the Allocation Algorithm”), is a multiple linear regression model. In very simple terms, a multiple linear regression model is an equation with independent variables and a constant on one side and a dependent variable on the other. A very simple regression model would follow this format: dependent variable = constant + coefficient1(independent variable1) + coefficient2(independent variable2) + coefficient3(independent variable3) The amount of the dependent variable is correlated to some degree with each of the independent variables, and the coefficients next to each independent variable quantify the correlation between the dependent variable and that particular independent variable. Therefore, if one has amounts for each of the independent variables, then the expected amount for the dependent variable can be calculated by: (a) plugging the amounts for the independent variables into the regression equation; (b) multiplying the independent variables by their coefficients; and (c) adding those results to the constant. The Legislature intends for the Allocation Algorithm to establish each client’s iBudget. However, the Legislature has recognized that there may be instances in which a particular client’s needs will not be met through the iBudget amount or other resources. These instances are referred to as “significant additional needs” (“SANs”)3/, and they will lead to an increase to a client’s iBudget. The first SAN is characterized in section 393.0662, as “[a]n extraordinary need that would place the health and safety of the client, the client’s caregiver, or the public in immediate, serious jeopardy unless the increase is approved.” This first SAN category is very broad and encompasses situations such as: (1) a documented history of life-threatening behaviors; (2) a complex medical condition that requires intervention by a licensed nurse; (3) a chronic comorbid condition; and (4) a need for total physical assistance with the activities of daily living. See § 393.0662, Fla. Stat. The second SAN category encompasses “[a] significant need for one-time or temporary support or services that, if not provided, would place the health and safety of the client, the client’s caregiver, or the public in serious jeopardy ” A significant need for one-time services could include needs such as modifications to a home or services to address the temporary loss of a caregiver. Id. The third SAN category accounts for a significant increase in the need for services after the beginning of the plan year due to substantial changes in a client’s circumstances. This could encompass situations such as the permanent or long-term loss of a caregiver or a significant change in medical or functional status that requires the provision of additional services. Id. Section 393.0662, contains other important provisions. For instance, section 393.0662(1)(b), requires that APD “shall reserve portions of the appropriation for [the Waiver] for adjustments required pursuant to this paragraph and may use the services of an independent actuary in determining the amount of the portions to be reserved.” Also, section 393.0662(1)(c), mandates that “[a] client’s iBudget shall be the total of the amount determined by the algorithm and any additional funding provided pursuant to paragraph (b) [i.e. SANs]. A client’s annual expenditures for [Waiver] services may not exceed the limits of his or her iBudget.” The Florida Legislature gave APD the authority to adopt rules governing the implementation of the iBudget system: [APD] and the Agency for Health Care Administration may adopt rules specifying the allocation algorithm and methodology; criteria and processes for clients to access reserved funds for extraordinary needs, temporarily or permanently changed needs, and one-time needs; and processes and requirements for selection and review of services, development of support and cost plans, and management of the iBudget system as needed to administer this section. § 393.0662(9), Fla. Stat. APD’s First Attempt to Adopt iBudget Rules APD utilized its rulemaking authority by promulgating Proposed Rules 65G-4.0210 through 65G-4.027 (“the First Proposed Rules”) in 2012. The Petitioners in the instant case filed a petition alleging that the First Proposed Rules were an invalid exercise of delegated legislative authority. An Administrative Law Judge rejected their challenge, but the First District Court of Appeal held that the First Proposed Rules violated section 120.52(8)(c), by enlarging, modifying, or contravening section 393.0662: Here, the Legislature was clear: the algorithm is the sole mechanism to set a client's iBudget, save for three exceptions specifically delineated by statute. § 393.0662(1)(c), Fla. Stat. In contravention of this clear requirement, the Proposed Rules use the algorithm, instead, as merely a starting point. The algorithm amount is then put through various modification mechanisms—none of which are contemplated by the clear statutory mandate that the "iBudget shall be the total of the amount determined by the algorithm and any additional funding provided pursuant to paragraph (b)." Id. Further, the use of the review mechanisms to decrease the algorithm amount contravenes the iBudget Statute. Nowhere in the statutory language does the Legislature contemplate decreasing the algorithm amount. The Legislature directed the algorithm be the floor and then permitted increases to that algorithm amount based on three delineated circumstances. G.B. et al. v. Ag. for Pers. with Disab., 143 So. 3d 454, 457- 58 (Fla. 1st DCA 2014). Even though the First Proposed Rules were invalidated, APD transitioned its clients to the iBudget System on July 1, 2013, when the federal waiver for the Tier System expired. This was possible because APD enacted an emergency rule and the Legislature allowed APD to continue using the First Proposed Rules until a new set of rules was adopted. In order to comply with the G.B. ruling, APD increased the iBudgets for approximately 14,000 clients so that their iBudgets were at least the algorithm amount. Development of the Proposed Rules After the G.B. decision, APD published a notice of development of rulemaking on October 23, 2014. Over the next several months, APD held multiple public hearings and noticed multiple changes to the Proposed Rules based on public comment. APD staff analyzed whether a statement of estimated regulatory costs (“SERC”) was required by completing a standardized checklist form entitled “Proposed Rule: Is a SERC Required.” See § 120.541(1)(b), Fla. Stat. (mandating that “[i]f a proposed rule will have an adverse impact on small business or if the proposed rule is likely to directly or indirectly increase regulatory costs in excess of $200,000 in the aggregate within one year after the implementation of the rule, the agency shall prepare a statement of estimated regulatory costs as required by s. 120.54(3)(b).”); § 120.541(3), Fla. Stat. (mandating that “[i]f the adverse impact or regulatory costs of the rule exceed any of the criteria established in paragraph (2)(a), the rule shall be submitted” for legislative ratification). Denise Arnold, APD’s Deputy Director, testified at the final hearing that APD’s SERC analysis included an examination of rules that had already been adopted, an evaluation of public comments, and an analysis of whether the Proposed Rules would raise costs for those private entities providing services to APD’s clients. Using the SERC checklist, APD concluded that the direct and indirect regulatory costs associated with implementing the Proposed Rules would be “zero;” that “zero” entities would be impacted; and that there would be no adverse economic or noneconomic impact from the implementation of the Proposed Rules. During the final hearing, there was no persuasive evidence that regulatory costs incurred by those providing services to APD’s clients would be increased if the Proposed Rules were to be adopted. For example, no Waiver providers testified at the final hearing. In contrast, the greater weight of the evidence demonstrated that the Proposed Rules would not materially increase the costs of providing services to APD’s clients. Under both the current iBudget system and the Tier System, Waiver providers have substantially the same responsibilities and workloads. While the Proposed Rules may require Waiver providers to complete new forms, the evidence demonstrated that those new forms will simply replace ones Waiver providers are currently using. Accordingly, APD’s Director signed the SERC checklist on December 3, 2014, and thus certified that no SERC was required. APD published a version of the Proposed Rules on December 5, 2014. That notice stated that a SERC and legislative ratification were not required. However, the notice did not include a description of the information utilized by APD to make that determination. Accordingly, the Joint Administrative Procedures Committee (“JAPC”) addressed a letter to APD on May 13, 2015, advising APD that the legislative ratification statement was missing from the notice of proposed rulemaking. The letter continued by stating that because “a SERC was not prepared for the rules, please publish a notice of correction describing the information expressly relied upon by [APD] in determining whether the rules are expected to require legislative ratification.” APD responded by publishing a notice of correction on May 28, 2015 stating: The Notice of Proposed Rule did not state what information was relied upon in determining that the proposed rule is not expected to require legislative ratification. The information expressly relied upon is the analysis conducted by [APD] to determine whether a SERC was required and the nature of the rule. This information sufficiently described the information utilized by APD in its SERC analysis. However, even if that were not the case, there is no basis for inferring that any deficiency in the notice of correction impaired the Petitioners’ substantial interests or the fairness of the proceeding. Also, there was no evidence or testimony presented at the final hearing to suggest otherwise. The Proposed Rules A few key components of the Proposed Rules warrant discussion. For example, the completion of a Questionnaire for Situational Information (“QSI”) is the first step in establishing a client’s iBudget. Proposed Rule 65G-4.0213(18) defines a QSI as “[a]n assessment instrument used by [APD] to determine an individual’s needs in the areas of functional, behavioral, and physical status,” and Proposed Rule 65G- 4.0213(18) incorporates the QSI by reference. The QSI is a series of questions designed to gather information about a client’s functional ability (i.e., the client’s need for assistance with eating, bathing, walking, etc.), behavioral issues (i.e., aggression, elopement, self- injury, etc.), and physical issues (i.e., medical issues that require assistance). The answers each client provides to the QSI enable APD to identify that client’s level of need and the types of supports he or she requires. The University of South Florida verified that the questions set forth in the QSI elicit answers that enable APD to reliably evaluate each client’s level of need for services. Proposed Rule 65G-4.0213(18) identifies the QSI as a “valid and reliable assessment instrument.” Ms. Arnold testified that a QSI is to be completed for every client at least once every three years. However, Proposed Rule 65G-4.0214(1)(d) provides that a client may request that another QSI be completed if there has been a significant change in circumstances that could affect a client’s iBudget amount. The data gathered from a client’s QSI plays a substantial role in calculating that client’s iBudget. As discussed in more detail in a subsequent section, the answers from several of the QSI questions serve as inputs into the Allocation Algorithm. Proposed Rule 65G-4.0216(3) provides that after calculation of a client’s Allocation Algorithm amount, the client’s Waiver Support Coordinator meets with the client, the client’s representative, and/or the client’s advocate in order to evaluate whether the client has any SANs.4/ Proposed Rule 65G-4.0218(1) sets forth SAN categories that essentially mirror the three listed in section 393.0662. However, Proposed Rule 65G-4.0218(1) provides for an additional or fourth SAN category involving transportation services to a waiver-funded adult day training program or to a waiver-funded supported employment. This applies when the client’s need cannot be met through his or her iBudget amount without compromising the client’s health and safety and there are no other viable means of transportation. If the Waiver Support Coordinator in conjunction with the client, the client’s representative, and/or the client’s advocate determines that the client has a SAN, then Proposed Rule 65G-4.0216(3) provides that the Waiver Support Coordinator will complete an Amount Implementation Meeting (“AIM”) Worksheet and submit it to APD within 30 days of receiving a new algorithm amount. If no additional documentation is needed, then APD will have 30 days from receipt of the AIM Worksheet to issue a decision on the client’s iBudget amount. Proposed Rule 65G-4.0216(3) mandates that APD “shall approve an increase to the iBudget amount if additional funding is required to meet [SANs] subject to the provisions of the iBudget rules.” In contrast, if the Waiver Support Coordinator in conjunction with the client, the client’s representative, and/or the client’s advocate determines that the algorithm amount is sufficient to pay for all the services and supports the client will need, then the individual review is complete and no further action is necessary. The client or someone acting on the client’s behalf signs the AIM form and notifies APD that there are no SANs. Once a client’s iBudget amount is established, Proposed Rule 65G-4.0217(1) requires the client’s Waiver Support Coordinator to submit a cost plan proposal to APD reflecting the specific waiver services and supports that the client will utilize and the providers of those services and supports. APD then reviews the cost plan to ensure that it conforms with the iBudget rules and the rules governing Waiver services in general. The Allocation Algorithm Section 393.0662(1)(a) directs APD to use an algorithm with variables that have been determined by APD to have a statistically validated relationship to the client’s level of need for services provided through the Waiver. In 2009 to 2010, APD contracted with Dr. Xu-Feng Niu, the chair of the statistics department at Florida State University, to develop and recommend options for an algorithm for the First Proposed Rules. Dr. Niu has performed statistical work for other Florida state agencies, including developing models and algorithms for the Department of Environmental Protection and the Department of Transportation. Dr. Niu has an extremely long and distinguished career as a statistician. He earned his Ph.D. in statistics from the University of Chicago in 1991. After being hired by Florida State University to teach statistics in 1991, Dr. Niu became the Chair of the University’s statistics department in 2011 and still holds that position. Dr. Niu created the allocation algorithm that was incorporated into the First Proposed Rules. That allocation algorithm’s validity was at issue during the previous rule challenge, and the ALJ who considered that challenge found that Dr. Niu’s allocation algorithm was valid. G.B., Z.L. through His Guardian K.L., J.H., and M.R. v. Ag. for Persons with Disab., Case No. 13-1849 (Fla. DOAH Sept. 9, 2013). In the subsequent appellate decision that invalidated the First Proposed Rules, G.B., et. al v. Ag. for Persons with Disab., supra, the First District Court of Appeal in G.B. did not address the allocation algorithm’s validity. In 2014, APD contracted with Dr. Niu to update the Allocation Algorithm for the Proposed Rules. Dr. Niu developed the Allocation Algorithm in the Proposed Rules by using a linear regression statistical method. As explained above, a multiple linear regression model utilizes independent variables (predictors) to arrive at a response (dependent variable). The independent variables in the Allocation Algorithm include a client’s age, living setting, and certain individual characteristics and support needs specified in the QSI. To provide a baseline to begin constructing the allocation algorithm, Dr. Niu used the clients’ fiscal year 2013-14 waiver expenditures (also referred to as claims data) as the dependent variable. Use of that fiscal year was significant because that was the first year that the Tier system was no longer in place, and the iBudget statute had been in effect for a full year. According to APD, every client’s needs had been met during the 2013-14 fiscal year. In practice, the dependent variable will be unknown because the dependent variable is that particular client’s future expected need, and that is what the Allocation Algorithm is designed to calculate. Dr. Niu’s analysis did not include 2013-14 claims data for a significant number of APD’s clients. For instance, APD served approximately 30,600 clients during that fiscal year, but Dr. Niu justifiably did not rely on claims data for any clients who had not been on the Waiver for at least one year. Also, Dr. Niu excluded the claims date from approximately 2,410 clients who were deemed to be “outliers.” In this context, “outliers” were deemed to be clients whose total claims during the 2013-14 fiscal year were extremely high or extremely low. This exclusion was done partially because including such outliers in a regression model sometimes reduces the model’s predictive ability. In making the decision to exclude approximately 10 percent of the clients who had been receiving Waiver services for a full year or more as outliers, rather than five percent as suggested by Petitioners, Dr. Niu considered input from APD’s stakeholders. As discussed above, APD conducted several public meetings regarding the formulation of the Proposed Rules, and all stakeholders (i.e., interested parties such as parents or guardians of clients, service providers, and staff from the Agency for Healthcare Administration) had ample opportunity to provide input to APD. Dr. Niu presented the stakeholders with a choice of designating either five percent or 10 percent of clients as outliers and excluding their claims data from the Allocation Algorithm’s formulation. The stakeholders ultimately recommended that 10 percent be designated as outliers, and Dr. Niu agreed with their recommendation. According to Dr. Niu, designating either five percent or 10 percent of the clients as outliers would have been appropriate. However, designating 10 percent of the clients as outliers was more appropriate in this instance, and resulted in a better model, because the iBudget system had only been in place for one year. The 10 percent exclusion was also deemed appropriate because there was a great deal of variation in the claims data. Because the claims data was in dollars, Dr. Niu needed to transform the data in order to create a bell curve. He accomplished that by performing a square root transformation on the data. While Dr. Niu could have used a logarithmic transformation, he determined that the square root transformation made the data more bell-shaped. Though either transformation could have been used, the preponderance of the evidence indicated that the square root transformation was a reasonable and appropriate means for developing the bell curve. After transforming the claims data from the 2013- 14 fiscal year, Dr. Niu had a baseline or a dependent variable for the Allocation Algorithm, but he still needed independent variables. In order to satisfy section 393.0662’s requirement that any allocation algorithm utilized by APD use independent variables that “have a statistically validated relationship to the client’s level of need for services provided through [the Waiver],” Dr. Niu turned to the QSI. As noted above, the QSI questions have already been validated as having a statistically valid relationship to a client’s level of need. However, Dr. Niu still had to determine what data from the QSI to use as independent variables in the algorithm. While Dr. Niu could have constructed an algorithm that utilized every piece of data from a client’s QSI, such a model would have been overly complex. Accordingly, Dr. Niu tested over 100 different regression models, each with different combinations of QSI data points as independent variables. After considering several different regression models, Dr. Niu ultimately determined that a regression model he designated as “5B” was the best one at predicting a client’s expected future need for services. That model’s independent variables include the client’s age, living setting status, and certain individual characteristics and support needs specified in the individual’s QSI. That QSI information included questions (among others) pertaining to a client’s ability to eat, dress, and protect himself or herself. Other questions assessed whether a client engages in inappropriate sexual behavior or experiences episodes of aggression. Dr. Niu determined that Model 5B was the best one by performing a bootstrapping analysis. Bootstrapping is a statistical method used for testing regression models. The method involves running a model with the original data while randomly repeating some claims and randomly dropping others. Dr. Niu performed this test approximately 10,000 times in order to test the reliability of Model 5B. Dr. Niu’s conclusion about Model 5B was also influenced by its R-squared value. The R-squared value quantifies how much of the variation from the average claim is explained by a particular model, and the R-squared value is a number between zero and one. If a particular model’s R-squared value is 1, then that model explains 100 percent of the deviation from the average claim. Model 5B had an R-squared value of .80. Therefore, that model explained 80 percent of the deviation. Dr. James McClave5/, Petitioner’s expert statistician, testified during the final hearing that the Allocation Algorithm’s margin of error was 90 percent when used to calculate actual claims totals by APD’s clients in 2013-14. In response, Dr. Niu testified that Dr. McClave was testing for the wrong thing. According to Dr. Niu, evaluating the Allocation Algorithm by its ability to predict past claims for the 2013-14 fiscal year was not a valid test of what the Allocation Algorithm was designed to accomplish. Specifically, Dr. Niu testified that the Allocation Algorithm is not attempting to predict past expenditures. Instead, it is trying to predict or calculate a claimant’s future expected need. A client’s future expected need is driven by that particular client’s individual characteristics and circumstances (i.e., the independent variables in the Allocation Algorithm). That same client’s future expected need can differ from his or her actual future need due to unexpected circumstances such as a drastic change in the client’s condition or the death of a caregiver. The evidence and testimony presented at the final hearing indicated that it would be extremely difficult to construct an algorithm that could account for such unexpected circumstances and thus be effective in calculating or predicting that client’s actual future need. Accordingly, the Allocation Algorithm is intended to calculate or predict a client’s future expected need based on the information that is known about that client. Dr. Niu’s testimony was credible and is accepted. To that point, Ms. Arnold persuasively explained why an individual client’s expenditures may unexpectedly change from one year to another. Ms. Arnold is the Deputy Director of Programs for APD and has been involved in implementing the iBudget System since its inception. She has worked for APD for over 30 years, and she was a service provider prior to her service at APD. Ms. Arnold has been working on the iBudget implementation and development for ten years, spending about 75 percent of her time on it. Ms. Arnold has spent numerous hours in meetings regarding the Proposed Rules and has spent many hours in meetings with Dr. Niu discussing the Allocation Algorithm. Thus, Ms. Arnold understands the Proposed Rules, what the Allocation Algorithm is supposed to do, and what the QSI does. As Ms. Arnold testified, the Allocation Algorithm is based on characteristics that identify a client’s future expected need. Therefore, clients with the same characteristics will receive the same Allocation Algorithm amount. However, this statistical process, which predicts need based on individual characteristics, does not account for all of the social variables that can create totally different situations for one client as opposed to another. For example, Ms. Arnold testified that you could have two individuals with similar characteristics; however, one client lives with a single mother and the other lives with a brother and sister who provide a great deal of support. As a result, the client with a single parent who works may need support during the day (such as a day program), while the client with sibling support may not have that same need. Ms. Arnold also testified that clients’ lives can change rapidly, and an algorithm relying on immature data cannot capture all of those changes. In addition, there are other reasons why a client’s future expected need may differ from his or her future actual need. Clients living in rural communities may receive services less often, a caregiver’s condition could change, an individual could go on vacation, or an individual may go into a hospital for an extended period. Also, Waiver funding is funding of last resort. § 393.0662(4), Fla. Stat. Therefore, a client’s need for Waiver funds may be accurately predicted by the Allocation Algorithm, but that client may not end up using all of the Waiver funds because his or her need is being met through other resources. As mentioned above, section 393.0662(1)(b), requires that the Allocation Algorithm “use variables that have been determined by [APD] to have a statistically validated relationship to the client’s level of need” for Waiver services. APD proved by a preponderance of the evidence that the Allocation Algorithm satisfies that requirement. During the final hearing, Petitioners attempted to undermine the Allocation Algorithm developed by Dr. Niu by noting that it was ineffective in “predicting” the Waiver claims made by APD’s clients during the 2013-14 fiscal year. This was evidenced by the 90 percent margin of error discussed above. Petitioners argued that if the Allocation Algorithm does a poor job of predicting claims made during the 2013- 14 fiscal year, then the Allocation Algorithm is not satisfying section 393.0662(1)(b)’s requirement to use variables having a statistically validated relationship to the client’s level of need. However, APD’s witnesses persuasively testified that the test put forth by Petitioners was not the appropriate way to ascertain whether the Allocation Algorithm uses variables having a statistically validated relationship to the client’s level of need. As discussed above, APD is not attempting to use the Allocation Algorithm to predict or calculate a client’s actual future need. Instead, APD is attempting to use the Allocation Algorithm to predict or calculate a client’s future expected need based on certain facts known about that client. Unexpected events may cause that client’s actual future need to substantially differ from that client’s future expected need. Moreover, Petitioners presented no other evidence that the information utilized in the Allocation Algorithm (such as the QSI information) did not have a statistically validated relationship to the client’s level of need. However, a cursory examination of the independent variables utilized in the Allocation Algorithm indicates they would have a substantial influence on a client’s level of need. Dr. Niu simply determined that a particular combination of variables had the strongest correlation to a client’s future expected need. As noted above, APD transitioned its clients to the iBudget System on July 1, 2013, and has been utilizing the First Proposed Rules. The finding that the Allocation Algorithm uses independent variables having a statistically validated relationship to the client’s level of need is corroborated by the fact that the iBudget System has been successful during the short amount of time that it has been implemented. The First Proposed Rules also had an Allocation Algorithm, and its R-squared value was .67 while the R-squared for the Allocation Algorithm at issue in this case is .80. Nevertheless, Ms. Arnold testified that only about 2,500 clients had to utilize the SANs process in a recent year. Also, Ms. Susan Chen, a statistician employed by APD, testified that APD’s clients generally spent only about 80 percent of their iBudgets during the 2013-14 fiscal year. If the iBudget system was not working, then the clients’ needs would have exceeded their iBudget allocation. Ms. Chen also testified that three of the Petitioners did not spend all of the money they had received via the Allocation Algorithm currently in effect. Specifically, two of the Petitioners spent only 80 percent of their allocations, and a third spent approximately 90 percent of what he or she was allocated. Finally, APD had budget surpluses during the 2013- 14 and 2014-15 fiscal years. This is additional evidence that the needs of those clients receiving Waiver services are being met. It is also evidence that the Allocation Algorithm (with a higher R-squared value than the one currently being utilized) has a statistically validated relationship to the client’s level of need. In sum, the evidence and testimony presented by APD was more persuasive than the evidence and testimony presented by Petitioners. Petitioners raised several other issues that will be addressed below.
