The Issue The issue to be decided is the amount to be paid by Petitioner to Respondent, Agency for Health Care Administration (Agency), out of her settlement proceeds as reimbursement for past Medicaid expenditures pursuant to section 409.910, Florida Statutes (2018).
Findings Of Fact On April 21, 2011, Ms. Puzanskas gave birth to her son. After birth, Ms. Puzanskas began experiencing symptoms of nervousness, panic attacks, and being overwhelmed. On June 21, 2011, she called her doctor's office and described her symptoms to her midwife. Her midwife concluded that Ms. Puzanskas was depressed or experiencing "baby blues." Based on this telephonic diagnosis, the midwife arranged for a prescription of the anti-depressant psychotropic drug, Zoloft, to be called into Ms. Puzanskas' pharmacy. The next day after taking the Zoloft, Ms. Puzanskas again called her doctor's office with complaints that the Zoloft was causing her to feel strange and jittery. Ms. Puzanskas was instructed to continue taking the medication. On June 24, 2011, Ms. Puzanskas began suffering from severe depression and hallucinations. That same day, she went into her back yard and doused herself with gasoline and set herself on fire. She suffered third-degree full thickness burns over 30 percent of her body requiring multiple skin grafts, with scarring over 60 percent of her body from all burns and grafts. Ms. Puzanskas' medical care for the injuries was paid by Medicaid, which provided $54,171.70 in benefits associated with her injuries. This amount constituted her entire claim for past medical expenses. As a condition of her eligibility for Medicaid, Ms. Puzanskas assigned to the Agency her right to recover from liable third-party medical expenses paid by Medicaid. Ms. Puzanskas brought a medical malpractice action against the medical staff responsible for her care to recover all of her damages associated with her injuries. During the pendency of the lawsuit, the Agency was notified of the action. Although it did not dispute the ultimate settlement received by Petitioner or otherwise participate in any aspect of the litigation, the Agency asserted a $54,171.70 Medicaid lien against Ms. Puzanskas' cause of action and settlement of the action. In preparation for the trial, Petitioner's counsel used mock jury panels to evaluate their trial strategies, value of damages, and the likelihood of a defense verdict. Mock jurors split. Some would have returned a verdict for the defense, finding no liability, while others would have returned a verdict for Ms. Puzanskas and given her some limited damages. Still others would have given her a very high amount of damages. See Pet'r Ex. 9. Eleven mock jurors provided verdicts from approximately $16,554,000 down to approximately $554,000. The remaining six jurors would have returned zero-dollar verdicts. The average award in the 17 verdicts was $3,741,000. Nine of the 11 jurors who produced a verdict for Petitioner included approximately $54,000 in their verdict, and then added amounts ranging from $500,000 to $16,500,000. The $54,000 is representative of Petitioner's rounded hospital bills. The insurance policy covering the incident had limits of $250,000 and the medical providers had no collectable assets. After the first day of trial, the medical providers offered $500,000 to settle the case, and this was accepted. However, this amount did not fully compensate Petitioner for her injuries. Mr. Moore, an experienced trial attorney who represented Petitioner, testified that based on his training and experience, Petitioner's damages had a value in excess of $3,700,000. However, using a conservative number for purposes of this case, he valued her damages at $3,000,000. Thus, the $500,000 settlement represented a recovery of 16.6 percent of the value of her damages, and a similar percentage for past medical expenses. Therefore, he testified that an allocation of $8,992.50, or 16.6 percent of $54,171.70, would be a reasonable and conservative portion of the settlement for past medical expenses. Based on his training and experience and review of the medical records and file, Mr. Barrett, a trial attorney, valued Petitioner's damages between three and five million dollars. He also opined that $3,000,000 would be a very conservative figure. Using the same allocation method advocated by trial counsel, Mr. Barrett applied a 16.6 percent ratio to the Medicaid expenses, and concluded that an allocation of $8,992.50 of the settlement to past medical expenses is reasonable, rational, and appropriate. This testimony was not rebutted by the Agency, and the Agency did not present any evidence proposing a differing valuation of damages or contest the methodology used to calculate the $8,992.50 allocation to past medical expenses. The testimony from Mr. Moore and Mr. Barrett is compelling and persuasive. Accordingly, the undersigned finds that Petitioner has proven by a preponderance of the evidence that $8,992.50 of the settlement represents reimbursement for past medical expenses.
The Issue The following are the issues presented: Whether Respondent, Leeland ER SVCS Partnership (“Leeland”), is liable to the Agency for Health Care Administration (“AHCA”) for Medicaid overpayments in the amount of $12,377.17, during the audit period of March 1, 2009, through August 31, 2011; Whether Leeland should be required to pay an administrative fine of $2,475.43, pursuant to Florida Administrative Code Rule 59G-9.070(7)(e); and Whether Leeland is liable to AHCA for the agency’s investigative, legal, and expert witness costs pursuant to section 409.913(23)(a), Florida Statutes.
Findings Of Fact Based on the oral and documentary evidence adduced at the final hearing, and the entire record in this proceeding, the following Findings of Fact are made: ACHA is designated as “the single state agency authorized to make payments for medical assistance and related services under Title XIX of the Social Security Act,” i.e., the “Medicaid program.” § 409.902(1), Fla. Stat. Among its duties as the Medicaid agency, AHCA is required to conduct audits of medical providers participating in the Medicaid program, and to “recover overpayments and impose sanctions as appropriate.” § 409.913, Fla. Stat. Section 409.913(1)(e) defines "overpayment" to include "any amount that is not authorized to be paid by the Medicaid program whether paid as a result of inaccurate or improper cost reporting, improper claiming, unacceptable practices, fraud, abuse, or mistake." The Medicaid provider agreement is a voluntary contract between AHCA and the provider. An enrolled Medicaid provider must comply fully with all state and federal laws pertaining to the Medicaid Program, including the Medicaid provider handbooks incorporated by reference into AHCA’s rules, as well as all federal, state, and local laws pertaining to licensure to receive payment from the Medicaid program. This case involves an AHCA Medicaid audit conducted of Leeland’s paid Medicaid claims as to the dates of service from March 1, 2009, through August 31, 2011, hereinafter referenced as the “audit period.” Leeland was randomly selected for audit and had no prior violations of Medicaid law. Therefore, any sanction imposed on Leeland in this proceeding would constitute a “first offense” under the operative rule discussed in the Conclusions of Law below. During the audit period, Leeland was an enrolled Medicaid provider and had a valid Medicaid provider agreement with AHCA. As an enrolled provider, Leeland was subject to all relevant federal and state statutes, rules, policy guidelines, and Medicaid handbooks incorporated by reference into rule. AHCA issued a PAR, dated June 20, 2013, alleging that Leeland was overpaid $200,349.16 for certain claims that in whole, or in part, were not covered by Medicaid. AHCA later issued a FAR, dated August 16, 2013, alleging that Leeland was overpaid $33,111.52 for certain claims that in whole, or in part, were not covered by Medicaid. The FAR further informed Leeland that AHCA intended to impose a fine of $6,622.30 (20% of the total overpayment) as a sanction for violation of rule 59G-9.070(7)(e) and to impose costs pursuant to section 409.913(23). Leeland received the FAR on August 23, 2013. Leeland timely filed a Petition for Formal Administrative Hearing on September 24, 2013. On October 9, 2013, Leeland tendered payment to AHCA in the amount of $33,111.52, as requested in the FAR, to be held in escrow pending the administrative hearing. The FAR set forth the basis for the overpayment determination as follows: Medicaid policy defines the varying levels of care and expertise required for the evaluation and management procedure codes for office visits. The documentation you provided supports a lower level of office visit than the one for which you billed and received payment. This determination was made by a peer consultant in accordance with Sections 409.913 and 409.9131, F.S. The difference between the amounts you were paid and the correct payment for the appropriate level of service is considered an overpayment. The FAR also stated that the overpayment calculation was based on a statistical formula by which a random sample of the claims submitted by Leeland was selected and extrapolated to the total number of claims in order to arrive at the amount of the total overpayment: A random sample of 63 recipients respecting whom you submitted 134 claims was reviewed. For those claims in the sample, which have dates of service from March 1, 2009, through August 31, 2011, an overpayment of $308.96 or $2.30567164 per claim, was found. Since you were paid for a total (population) of 26,060 claims for that period, the point estimate of the total overpayment is 26,060 x $2.30567164 = $60,085.80. There is a 50 percent probability that the overpayment to you is that amount or more. We used the following statistical formula for cluster sampling to calculate the amount due the Agency:[1/] All of the claims relating to a recipient represent a cluster. The values of overpayment and number of claims for each recipient in the sample are shown on the attachment entitled “Overpayment Calculation Using Cluster Sampling.” From this statistical formula, which is generally accepted for this purpose, we have calculated that the overpayment to you is $33,111.52 with a ninety-five percent (95%) probability that it is that amount or more. After issuance of the FAR, Leeland provided additional information and documentation to MPI, which conducted a peer review of the new material. AHCA subsequently reduced the alleged overpayments in the sample to $171.38. Overpayments were found on claims involving seven of the 63 recipients.2/ AHCA concluded that this overpayment amounted to 2.45 percent of the total payments of $6,987.99 made to Leeland for the claims in the sample. The overpayment amount of $171.38 was extrapolated to the entire population of claims using the formula set forth above. AHCA concluded that the total amount of overpayments to Leeland for all Medicaid recipients in the population was $12,377.17, with a 95 percent confidence level. This reduction in the alleged overpayment led AHCA to make a proportional reduction in the proposed fine, to $2,475.43. Leeland does not challenge the agency’s conclusion that the actual overpayment found in the sample amounted to $171.38. Leeland does challenge the method by which AHCA used that actual overpayment to extrapolate an overall overpayment amount of $12,377.17 for the entire body of Medicaid claims submitted by Leeland during the audit period. AHCA is required by statute to use an “accepted and valid statistical calculation” to determine Medicaid overpayments. ACHA submitted its audit report and work papers into evidence. To support the validity of the cluster sampling method used in this case, AHCA presented the testimony of Dr. Fred Huffer, a professor in the Statistics Department at Florida State University, as well as the AHCA employees who provided the data to which the formula was applied. Robi Olmstead, supervisor of MPI’s Practitioner Care Unit, testified that Leeland was randomly selected for audit. Once the selection was made, Ms. Olmstead assigned the case to an investigator. Her office applied a computerized claim sampling program to select the recipients and claims to be audited. The program pulled all claims for the provider during the audit period. Ms. Olmstead sorted the claims, selecting only those that were fee-for-service, then generated the “seed” and selected the cluster sample. Ms. Olmstead testified that the program tells her how many recipients should be reviewed to make a statistically valid sample. In Leeland’s case, the program stated that 62.6 recipients should be used, so the number was rounded up to 63. Lisa Robinson, the MPI investigator who handled the Leeland audit, testified that the claim sampling program selected the list of 63 recipients to be audited. Ms. Robinson sent a request for medical records to Leeland. Once Leeland submitted the records for the 63 recipients, Ms. Robinson reviewed the records. The claim sampling program generated a worksheet listing each billed claim for each recipient. Ms. Robinson attached the worksheets to the records and prepared them for the nurse reviewer. The nurse reviewer reviewed and organized the records for a peer review by a physician. After the physician reviewed and determined any disallowed amounts, the records were returned to Ms. Robinson, who entered the disallowed amounts into the claim sampling program to determine the amount of the overpayment. Ms. Olmstead testified that she has no statistical expertise and that she relied on Dr. Huffer to review and validate the results obtained by the claim sampling program. Ms. Robinson likewise claimed no statistical expertise or any real knowledge of how the claim sampling program works. Ms. Robinson simply enters data into the program and accepts the results it generates. Dr. Huffer, who has consulted with MPI since 2004, testified that when he received the overpayment calculation results, he first checked the calculations. Next, he constructed hypothetical populations based on MPI’s sample to test the confidence level of 95 percent asserted in the FAR. Dr. Huffer explained that a confidence level is a probability attached to the correctness of some statement or procedure. The 95 percent confidence level in this case means that if MPI runs its audit procedure repeatedly, the number that it states as the overpayment from a sample of the population will be less than the “true” overpayment in the overall recipient population 95 percent of the time. The “true” overpayment value remains unknown, but the simulations performed by Dr. Huffer lead to a “reasonably confident” conclusion that the assessed overpayment is an underestimate of that “true” value. Dr. Huffer stated that the simplest type of sampling scheme is a simple random sample, in which units are selected at random and audited. He noted that sometimes the units are naturally grouped into clusters, and much sampling effort can be saved by sampling the clusters of units rather than the units individually. In this case, AHCA was interested in auditing a population of claims, but the claims were naturally grouped by recipients. Therefore, to conserve resources, AHCA used single- stage cluster sampling, with each selected resident constituting a cluster of claims to be audited. Dr. Huffer noted the practical advantages of this method: [T]here’s a lot less effort in accessing the records of a smaller number of recipients, and also there’s a lot less effort in making decisions about medical necessity for a small number of recipients versus, say, a large number of recipients. So there’s a lot of savings in sampling effort by doing a cluster sampling based upon clusters, which are the recipients. Dr. Huffer testified that a sample size of 63 was valid, independent of the size of the population from which the sample was taken. He stated that “it is a well-known fact in statistics that it is the sample size which primarily governs the accuracy of the result, not the population size.” He noted, for instance, that a sample size of 35 could be validly used for a population of one million. Dr. Huffer explained that he constructed a hypothetical population that is “like a large scaled-up version of the sample.” He “cloned” every recipient and every claim for all recipients about 208 times to make a hypothetical population of approximately 13,000 recipients. From this population, he sampled 63 recipients at random and performed the same calculation that AHCA did on its sample. He performed the calculation procedure on two million samples of 63 recipients drawn from his hypothetical population. Dr. Huffer’s two million simulations yielded an empirical confidence level of 97.7 percent, meaning that “we’re even more confident in this case that the number we announce as the overpayment is less than the true overpayment . . . in the population.” Dr. Huffer explained the extrapolation of the sample to the population. By taking the $171.38 of total overpayments found in the 134 claims for the population of 63 residents in the sample, MPI derived an average overpayment per sample claim of $1.27.3/ There were 26,060 claims in the entire population. Multiplying the total number of claims by the $1.27 average overpayment yielded a “point estimate” of the total overpayment of a little more than $33,000. Dr. Huffer stated that while the overpayments in the population may be “in the neighborhood” of the point estimate, there is never an expectation that the point estimate will be exactly correct. Every random sample of recipients would yield a somewhat different total. Therefore, a standard error of the overpayment was introduced as an estimate of how far wrong the point estimate might be. The standard error in this case was $12,547.82. The true overpayment could be plus or minus some multiple of the standard error. Dr. Huffer testified that to reach the lower bound of the 95 percent confidence level, MPI subtracted about one and one-half times the standard error from the point estimate to arrive at an overpayment value of $12,377.17. Dr. Huffer concluded that there was “strong evidence” that the true overpayments exceeded $12,377.17, because that figure was an “intentional underestimate.” Counsel for Leeland questioned Dr. Huffer about the validity of the statistically derived overpayment, given that the actual overpayment drawn from the sample, $171.38, was so small compared to the total Medicaid payments for those recipients. Dr. Huffer testified that the 95 percent confidence rate is “totally unrelated” to the magnitude of the actual overpayments. To counter Dr. Huffer’s testimony on the irrelevancy of the size of the actual overpayment to the validity of the sampling method, counsel for Leeland presented a federal Medicare statute, 42 U.S.C. § 1395ddd(f)(3), which provides as follows, in relevant part: Limitation on use of extrapolation A medicare contractor may not use extrapolation to determine overpayment amounts to be recovered by recoupment, offset, or otherwise unless the Secretary determines that— there is a sustained or high level of payment error; or documented educational intervention has failed to correct the payment error . . . . Dr. Huffer responded that the federal statute does not imply that extrapolation is not allowed for statistical reasons. He believed that the reason for the Medicare law’s disallowance of extrapolation in smaller cases could be simply to forgive errors below a certain threshold. Counsel for Leeland offered another example, an “Open Letter to Health Care Providers” issued by the Office of Inspector General of the U.S. Department of Health and Human Services in 2001. The letter sets forth new claims review procedures, including a statement that if the net financial error rate in a discovery sample is below five percent, the provider is not required to perform any further audit work and only the actual identified overpayments must be refunded. Dr. Huffer pointed out that the letter, like the statute, does not question the statistical validity of extrapolation. “They do not give any statistical reason for saying that it would be wrong to proceed in this case. As far as I know, they’re just saying if you [have] a small error rate, we’ll forgive it.” Dr. Huffer agreed that there was not a “sustained or high level of payment error” in this case, but observed that this case was not being decided under the federal Medicare statute. Dr. Huffer opined that the sampling method used in this case was reasonable and comported with generally accepted statistical methods. His opinions and explanation were credible, were unrebutted, and are accepted. Leeland's attempt to undermine Dr. Huffer’s opinions through cross-examination was ineffective and lacked the support of contradictory expert testimony regarding generally accepted statistical methods. AHCA seeks to recover its investigative, legal, and expert witness costs pursuant to section 409.913(23)(a). AHCA has established its right to recover these costs. At the outset of the final hearing, the parties agreed that if AHCA prevailed in the case-in-chief, and was found to be entitled to costs, then this tribunal would retain jurisdiction for the limited purpose of allowing AHCA to document its costs in the manner provided by section 409.913(23)(b).
Recommendation Based on the foregoing, it is, therefore, RECOMMENDED that the Agency for Health Care Administration enter a final order requiring Leeland ER SVCS Partnership to repay the sum of $12,377.17 for overpayments on claims that did not comply with the requirements of Medicaid laws, rules, and provider handbooks, including interest. Jurisdiction is retained to determine the amount of costs and attorney's fees, if the parties are unable to agree to the amount, and either party may file a request for a hearing within 30 days after entry of the final order to determine the appropriate amounts. DONE AND ENTERED this 11th day of April, 2016, in Tallahassee, Leon County, Florida. S LAWRENCE P. STEVENSON Administrative Law Judge Division of Administrative Hearings The DeSoto Building 1230 Apalachee Parkway Tallahassee, Florida 32399-3060 (850) 488-9675 Fax Filing (850) 921-6847 www.doah.state.fl.us Filed with the Clerk of the Division of Administrative Hearings this 11th day of April, 2016.
