596 So.2d 106 (1992)
SOUTHPOINTE PHARMACY, Appellant,
v.
DEPARTMENT OF HEALTH AND REHABILITATIVE SERVICES, Appellee.
No. 91-451.
District Court of Appeal of Florida, First District.
March 11, 1992.
William M. Furlow, Katz, Kutter, Haigler, Alderman, Davis, Marks & Rutledge, P.A., Tallahassee, for appellant.
David G. Pius, Sr. Atty., Dept. of Health and Rehabilitative Services, Tallahassee, for appellee.
WIGGINTON, Judge.
We have for review the final order of the Department of Health and Rehabilitative Services (Department or HRS) concluding that Southpointe Pharmacy (Southpointe), as a Medicaid provider, received overpayments from the Medicaid program in the amount of $7,742.93. In its final order, the Department demanded repayment of said amount, imposed a fine of $250, and suspended Southpointe from the Medicaid program for a period of three months. In doing so, it rejected in part certain findings of fact made by the hearing officer, as well as the hearing officer's conclusions of law drawn therefrom. We reverse.
It is undisputed that Southpointe Pharmacy is a Medicaid provider of prescription medications located in Dade County, Florida, and receives reimbursement for that medication from HRS. In turn, HRS' Office of Program Integrity has the responsibility for insuring that the goods and services billed to the Medicaid program are those that are actually provided to Medicaid recipients. To that end, at the times pertinent to this proceeding, HRS contracted with Professional Foundation for Health Care (PFHC) to conduct audits of a number of pharmacies. The audit period pertinent to this proceeding was from November 1, 1987 through July 31, 1988.
In his recommended order, the hearing officer made the following findings of fact:
7. In early 1989, PFHC conducted a routine audit of Southpointe. This routine audit is a general audit of a number of factors related to the provision of medications by a pharmacy and includes an examination of prescriptions in the pharmacy, an analysis of prices that are charged, and a review of the inventory of certain drugs to determine whether the volume of drugs on the shelf is consistent with the quantity of drugs that have been previously billed to Medicaid. Three computer printouts were generated for use of PFHC in the routine audit of Southpointe. The first computer printout was a random sampling listing 200 Medicaid claims out of the several hundred Medicaid claims that had been submitted by Southpointe during the audit period. The second computer printout and the third computer printout listed all of the drugs that had been paid for by the Medicaid program to Southpointe during the audit period. The second printout listed these drugs according to the total dollar amount of the claims paid for each respective drug while the third listed the drugs alphabetically.
8. After the routine audit was concluded, the results were sent to the Medicaid Program Integrity Office where it was reasonably determined that further action would be warranted. The routine analysis had been for screening purposes only, and did not serve as the basis for the position asserted by DHRS in this proceeding.
9. DHRS decided to conduct a further audit, referred to as an aggregate analysis. DHRS based its conclusion that Southpointe had been overpaid based on the results of the aggregate analysis. Prior to the date of the formal hearing in this case, DHRS had not promulgated a rule specifically authorizing the aggregate analysis auditing method, but instead utilized incipient, non-rule policy, and then sought to explicate that policy at the hearing.
10. An aggregate analysis is a more detailed audit that is typically done on behalf of DHRS by a contractor such as PFHC. The aggregate analysis does not review the Medicaid payments to a provider on a claims by claims basis. Instead, it is a statistical analysis in which claims that have been billed to and paid by the Medicaid program are reviewed to determine whether the pharmacy purchased a sufficient quantity of each respective drug to cover the provider's billings to the Medicaid program for that drug. The aggregate analysis is designed to determine whether the pharmacy's inventory levels are consistent with its Medicaid billings. If the billings to Medicaid exceed the amount of the drug that the provider had available for distribution during the period being audited, DHRS presumes that the pharmacy's Medicaid claims are not justified to the extent of the excess and that the pharmacy has been overpaid.
11. The aggregate analysis examines the top 100 drugs dispensed by Southpointe during the audit period based on the dollar value of all claims for those drugs. The auditors then attempt to determine the extent of the overpayment for each individual drug by determining whether the quantity of the drug billed to Medicaid exceeds quantity of the drug available for sale by the pharmacy. If the quantity of the drug billed to Medicaid exceeds the quantity of the drug available for sale by the pharmacy, the differences in quantity is multiplied by the price of the drug which produces the dollar amount of the overpayment. The overpayment for all drugs are [sic] then added to derive the total of the alleged overpayment. In conducting the aggregate analysis, DHRS' auditors presumed that all drugs available to Southpointe were dispensed to Medicaid patients. It had been calculated that 71% of Southpointe's business was comprised of Medicaid recipients. The initial calculations (which included the Gulf Distributors information that was not shown to be reliable) were based on the presumption that only 71% of the drugs available to Southpointe would have gone to Medicaid recipients and produced an initial determination that the overpayment was $60,201.02. A revised calculation (which also included the Gulf Distributors information) presuming that all of the drugs available to Southpointe would have gone to Medicaid recipients produced a determination that Southpointe had been overpaid in the amount of $50,247.25.
At that point, the record shows HRS demanded of Southpointe repayment of the overpayment, and advised that it intended to impose an administrative fine in the amount of $2,000 and intended to terminate Southpointe in the Medicaid program for a period of two years. At Southpointe's request, a formal hearing was held on May 22, 1990. However, prior to the rendition of a recommended order, a third letter was received by Southpointe showing a further revised amount of alleged overpayment of $49,728.82.
At the time of the audit, the main supplier for Southpointe was Gulf Distributors, Inc. HRS relied on information in the form of a computer printout it had received directly from Gulf Distributors in conducting its aggregate analysis. In addition, Southpointe was requested to provide all invoices for drugs it had purchased from any supplier other than Gulf Distributors during the pertinent period. Again, prior to the rendition of the recommended order, on September 19, 1990, HRS once more revised the amount of overpayment to $7,301.68. It did so after the hearing officer had held to be unreliable the computer printouts from Gulf Distributors. On this point, the hearing officer made these additional findings of fact:
14. The aggregate analysis of 12 drugs purchased by Southpointe from Upjohn Company and from Merck, Sharpe, and Dohme Company reflected an overpayment to Southpointe in the total amount of $7,301.68. The analysis assumed that the invoices Southpointe gave Mr. Kopotowsky [the auditor] constituted all of the drugs manufactured by Upjohn and MSD that were available to Southpointe for distribution during the audit period. There was no evidence that the auditors asked whether Southpointe acquired additional drugs manufactured by Upjohn and MSD by means other than purchase from these manufacturers. It is common practice for pharmacies to exchange or barter inventory, and such exchanges would not be reflected by an examination of the invoices of purchases directly from the manufacturers. There was no evidence that the auditors asked to see records of such acquisitions. There was no evidence as to the shelf life of any of these drugs or that DHRS, through its auditors, accurately determined Southpointe's beginning or ending inventory for the Upjohn and MSD drugs. Without this information, the aggregate analysis, upon which DHRS bases its claim for overpayment, is flawed and does not accurately reflect an overpayment to Southpointe.
Based on the foregoing findings, the hearing officer concluded that HRS had failed to prove any wrongdoing on the part of Southpointe.
In its final order, however, the Department granted HRS' exception to the hearing officer's finding of fact in paragraph 9, supra, that the Department relied on non-rule incipient policy in utilizing the aggregate analysis method to audit Southpointe. In rejecting that finding, the Department, instead, found that the relevant rule, Rule 10C-7.061(4)(b), Florida Administrative Code, authorized the use of a "statistical analysis" to determine overpayment, and that aggregate analysis is a statistical analysis. The Department also rejected the hearing officer's finding in paragraph 14, supra, that the aggregate analysis was flawed, and instead relied on the testimony of the expert witnesses presented by HRS to conclude that that finding was not based on competent and substantial evidence. In doing so, we must agree with Southpointe that the Department transgressed the fundamental rule of law that an agency may not disregard findings of fact of the hearing officer based on competent and substantial evidence, and substitute its own findings of fact. Where the hearing officer's findings of fact and reasonable inferences drawn therefrom are based upon competent and substantial evidence, it is a gross abuse of discretion for the agency to disregard those findings. See Heifetz v. Department of Business Regulation, 475 So.2d 1277 (Fla. 1st DCA 1985); Reese v. Department of Professional Regulation, 471 So.2d 601 (Fla. 1st DCA 1985).
In these proceedings, HRS, as the party seeking to establish a Medicaid overpayment, had the burden of proving the allegation by a preponderance of the evidence. Florida Department of Transportation v. J.W.C., Co., 396 So.2d 778 (Fla. 1st DCA 1981); David's Pharmacy v. Department of Health and Rehabilitative Services, 11 FALR 2935 (HRS 1988). Nevertheless, we agree with Southpointe and the hearing officer that the aggregate analysis was not contemplated by Rule 10C-7.061(4)(b), and was therefore incipient policy. A reading of the relevant section of the rule clearly shows that the rule, as it existed at the time of the audit and hearing, allowed a certain form of statistical analysis as described therein, but did not authorize every kind of statistical analysis.[1] Subsection (4) of the rule in relevant part reads as follows:
(4) Determination of overpayments.
(a) Analytic Calculation. The overpayment to a Medicaid provider may be determined analytically by comparing the amount actually paid for each claim with the amount that should have been paid for the claim based on an applicable Medicaid policy and then summing the overpayment for all claims reviewed. The analytic process may be computer assisted.
(b) Statistical Calculation. If it is not reasonably possible to examine every paid claim of a provider for a given period of time in order to compare the amount actually paid with the amount that should have been paid, generally accepted statistical methods may be employed to determine the overpayment for the total population of claims. The total number of provider's claims of a specific type or types paid by Medicaid during a given period of time, or such claims having dates of services extending over a given period of time, will be taken to be the population of claims analyzed. From this population of claims, or appropriate subset of the population, a statistically representative sample will be taken. The overpayment for the claims and the sample will be determined and then, using generally accepted statistical methods, the overpayment applying to the total population of claims, or subset of the population, will be calculated.
(c) Peer Review ... [not applicable to pharmacy audits].
In the instant case, HRS presented testimony from Robert Peirce, administrator of the Medicaid Investigative Unit in the Office of Program Integrity, who stated that the aggregate analysis was contemplated by the rule as a form of statistical analysis. Although we would generally defer to such an opinion, as we are required to give great weight to an agency's interpretation of its own rule, that deference is not absolute. Woodley v. Department of Health and Rehabilitative Services, 505 So.2d 676 (Fla. 1st DCA 1987). In short, simply because the Department says that the aggregate analysis is a form of a statistical analysis contemplated by the rule, does not make it so; the deference does not extend to unreasonable interpretations.
In this case, as pointed out by Southpointe, the rule in question allowed under certain circumstances a statistical sample of claims to be reviewed, rather than requiring HRS to look at each and every claim. Clearly, the rule is claims oriented. In contrast, the aggregate analysis focuses on inventory and purchasing patterns. As Robert Peirce testified, the aggregate analysis is an examination, over the period comprehended by the audit, of the top 100 drugs that have been billed to and paid for by the Medicaid program in order to determine whether for each of those drugs the pharmacy purchased a sufficient quantity to cover its billing to the Medicaid program. The underlying premise to this analysis is that before a drug is dispensed and billed to and paid for by the Medicaid program, the pharmacy should have the drug in its possession. Clearly, Rule 10C-7.061(4) does not embody that premise. Therefore, as found by the hearing officer, the Department was proceeding not under any existing rule but rather under incipient policy. That finding was based upon competent and substantial evidence, and we hold it was a gross abuse of discretion for the Department to reject that finding of fact.
It is, of course, well-settled that there is no impediment to an agency's utilizing a non-rule policy on a case-by-case basis, so long as its policy is explicated by an adequate record foundation in the proceeding. See St. Francis Hospital, Inc. v. Department of Health and Rehabilitative Services, 553 So.2d 1351 (Fla. 1st DCA 1989). In St. Francis, we held that
[w]hen an agency seeks to validate agency action based upon a policy that is not recorded in rules or discoverable precedents, that policy must be established by expert testimony, documentary opinions, or other evidence appropriate to the nature of the issues involved and the agency must expose and elucidate its reasons for its discretionary action. .. . [Citations omitted.] The agencies may apply incipient or developing policy in a section 120.57 administrative hearing, provided the agency explicates, supports and defends such policy with competent, substantial evidence on the record in such proceedings. [Citation omitted.]
Id. at 1354.
However, in the instant case, we agree with the hearing officer that HRS failed in its mission to support and defend the aggregate analysis with competent and substantial evidence. In an earlier final order issued by the Department, David's Pharmacy v. Department of Health and Rehabilitative Services, 11 FALR 2935 (HRS 1988), wherein aggregate analysis was utilized for the first time, the Department found HRS had not appropriately explicated this non-rule policy by its failing to produce evidence that would establish a rational, reasonable basis for the procedure.
In the instant case, despite rather pat testimony to the effect that the aggregate analysis is indeed contemplated by the rule, it was shown that HRS had not checked a single Medicaid patient to determine if the medication had been dispensed, or a single physician to see if the medication had been prescribed. Robert Peirce testified that the only thing HRS had done to validate the aggregate analysis auditing method since David's Pharmacy, was to delete the requirement of utilizing a "percentage of Medicaid sales" from the formula. As pointed out by Southpointe, none of the other shortcomings of aggregate analysis which were identified in the David's final order were remedied by HRS at the hearing below. For example, neither a beginning nor ending inventory had been taken into consideration, and no consideration was given as to whether Southpointe had acquired additional drugs to augment its inventory by means other than direct purchase from its manufacturers. Although the Department in its final order found that Southpointe was given an opportunity to provide evidence of any bulk purchases, including purchases made outside the audit period, or additional purchases not originally computed in the overpayment calculations, the record shows that the only witness who spoke directly with anyone from Southpointe stated specifically that the owner was asked only to produce invoices. Nowhere in the record does it show that the reason behind this request was explained to the owner, much less that he should provide evidence of other acquisitions, such as bulk purchases, trades, inventory acquisition, etc.[2] Consequently, the hearing officer's findings on the issue of whether or not HRS had satisfactorily validated and explicated its non-rule policy were based on competent and substantial evidence and should not have been rejected by the Department. See section 120.57(1)(b)10, Florida Statutes.
Because the Department improperly rejected the hearing officer's findings of fact and conclusions of law, we reverse the final order and direct that the Department adopt the hearing officer's recommended order. In so holding, we need not reach the other points raised by appellant Southpointe for reversal.
REVERSED and REMANDED for further proceedings consistent with this opinion.
BARFIELD and WOLF, JJ., concur.
NOTES
[1] The Department has since amended the rule to specifically include aggregate analysis. See Department of Health and Rehabilitative Services v. Westchester Pharmacy, 13 FALR 2728 (HRS 1991). As amended, the rule now reads in pertinent part as follows:
(4) Determination of Overpayments ...
(b) Statistical Calculation. If it is not reasonably possible to examine every paid claim of a provider for a given period of time in order to compare the amount actually paid with the amount that should have been paid, generally accepted statistical methods may be employed to determine the overpayment for the total population of claims... . From this population of claims, or appropriate subset of the population, a statistically representative sample will be taken... .
(c) Aggregate Analysis. Upon reasonable request, a provider of goods that are billed to and paid for by the Medicaid program must furnish to authorized auditors, investigators and agents invoices and other information directly related to inventory or acquisition of goods from suppliers and other documentation demonstrating that the provider had available during the audit period sufficient quantities of such goods to support billings to Medicaid. Such information and documentation shall include total sales data pertaining to each of the products that are sold and billed to Medicaid and to other customers, and shall include purchase, barter and inventory records and the names of major supplies... . Purchase and barter documentation will be considered to the extent furnished by the provider... .
[2] These factors were vitally important in validating the non-rule aggregate analysis methodology. Thus, Southpointe's failure to present any evidence at the hearing establishing that it had augmented its inventory by these alternative methods would not affect the agency's initial burden to explicate its incipient policy. But cf. Department of Health and Rehabilitative Services v. Westchester Pharmacy, supra, note 1.