BROWNING, C.J.

The Agency for Health Care Administration (AHCA) appeals a final order of the Division of Administrative Hearings that held that a formula used by AHCA to calculate overpayments to Medicaid providers violated section 120.54(1)(a), Florida Statutes (2007). Because we find that the formula does not meet the statutory definition of a rule, we reverse the final order and remand the case for further proceedings consistent with this opinion.

AHCA is the state agency responsible for administering Florida's Medicaid program, for auditing Medicaid providers, and for recouping overpayments made to Medicaid providers. §§ 409.902, 409.913, Fla. Stat. (2007). Appellee, Custom Mobility, is a Medicaid provider. AHCA conducted an audit of the Medicaid claims submitted for payment by Custom Mobility between January 1, 2001, and December 31, 2003. In a Final Audit Report (FAR) dated December 19, 2005, AHCA advised Custom Mobility that it had determined that Custom Mobility was overpaid in the amount of $245,317.83. The overpayment calculation was performed using the statistical formula for cluster sampling that is at issue here. Custom Mobility was notified in the FAR that it had the right to request a formal or informal hearing with respect to that overpayment determination; Custom Mobility requested a formal hearing. Custom Mobility set forth evidence intended to show that it is substantially affected by the formula, in that the formula was used as the basis for calculating the amount that AHCA is seeking to recover from Custom Mobility as alleged Medicaid over-payments.

When AHCA audits a Medicaid provider, it "must use accepted and valid auditing, accounting, analytical, statistical, or peer-review methods, or combinations thereof. Appropriate statistical methods may include, but are not limited to, sampling and extension to the population, parametric and nonparametric statistics, tests of hypotheses, and other generally accepted statistical methods." § 409.913(20), Fla. Stat. (2007). Statistical sampling methodologies are used to permit the auditors to analyze a random sample from the population of Medicaid recipients and/or claims, determine the findings in the sample, and extend the sample findings to the population of recipients and/or claims.

In using the statistical sampling methodology of cluster sampling, the AHCA auditor draws a random sample of Medicaid recipients who have received goods or services from a particular Medicaid provider and evaluates the claims for each recipient who is included in the sample. The amount of overpayment is determined for each claim for each recipient, and the statistical formula for cluster sampling is used to extend the overpayment found in the sample to the entire population of Medicaid claims to determine the total overpayment. The formula at issue is the only statistical formula used by AHCA to calculate Medicaid overpayments when the statistical sampling methodology of cluster sampling is used.

In approximately 10% of the audits conducted between June 1, 2004, and May 31, 2007, AHCA used the cluster sampling methodology. The cluster sampling formula has been used by AHCA for approximately twenty years to calculate Medicaid overpayments. Counsel stipulated that the formula has not been adopted as a rule pursuant to the procedures set forth in Section 120.54, Florida Statutes (2007).

The administrative law judge issued a final order on August 23, 2007, holding that the cluster sampling formula violates Section 120.54(1)(a), Florida Statutes, and ordering AHCA to immediately discontinue all reliance on the statement or any substantially similar statement as a basis for agency action. AHCA timely filed a notice of appeal on September 7, 2007.

The legislature defines a rule as an "agency statement of general applicability that implements, interprets, or prescribes law or policy or describes the procedure or practice requirements of an agency. . . ." § 120.52(15), Fla. Stat. (2007). Florida imposes rulemaking procedures only upon "those statements which are intended by their own effect to create rights, or to require compliance, or otherwise to have the direct and consistent effect of law." McDonald v. Dep't of Banking Fin., 346 So.2d 569, 581 (Fla. 1st DCA 1977). The formula here does not by its own effect create rights, require compliance, or have the direct and consistent effect of law, because it is a mere formula and does not give the service provider any rights, or require compliance. The most the formula does is to calculate the amount of overpayment, and it is subject to discretionary application because AHCA has discretion to use or not use the corresponding methodology in any given audit.

Further, the cluster sampling formula is not an agency statement of general applicability. In Department of Revenue v. Vanjaria Enterprises, Inc., 675 So.2d 252, 255-56 (Fla. 5th DCA 1996), the court held that a tax assessment procedure was a statement of general applicability because it was the sole guide for auditors, it applied "for all audits performed" and was not applied on a case-by-case basis, as here, and the auditors had no discretion to act outside of the procedure. In Department of Commerce v. Matthews Corporation, 358 So.2d 256 (Fla. 1st DCA 1978), this Court held that wage rate guidelines were not statements of general applicability because they were "applicable only to the construction of the particular public building or other work specified in the determination," and because they have "no prospective application to any other contract — only the specific project involved in the particular location." Id. at 258. In Department of Highway Safety v. Schluter, 705 So.2d 81, 82 (Fla. 1st DCA 1997), this Court held that three of six agency statements in question, policies concerning officer discipline, were not statements of general applicability "because the record establishes that each was to apply only under `certain circumstances,'" and so did not have the "consistent effect of law" mentioned in McDonald. Similarly, the formula here applies only to some of the service providers being audited, and thus does not have the consistent effect of law.

Appellee argues that some agency statements have been held to have general applicability even though they apply only to a subgroup, because in Department of Administration v. Harvey, 356 So.2d 323 (Fla. 1st DCA 1977), the statement held to be a rule applied to all state employees seeking career service positions, rather than to all state employees; in Schluter the policies applied to all police under investigation, rather than to all policemen; and in Balsam v. Department of Health Rehabilitative Services, 452 So.2d 976 (Fla. 1st DCA 1984), the policy found to be a rule applied only for a finite period of time. However, these cases are materially distinguishable, because here the formula does not apply to all service providers, or even to all service providers being audited; some providers being audited are subject to other formulas under other methodologies. And, in fact, Harvey and Balsam do not explicitly address the factor of general applicability.

Finally, the cluster sampling formula at issue here does not implement, interpret, or prescribe law or policy. The formula does not create any rights or adversely affect others, because it does not itself establish that the service provider owes money; per section 409.913(20), Florida Statutes (2007), "the agency may introduce the results of such statistical methods as evidence of overpayment" (emphasis added). Section 409.913(21), Florida Statutes (2007), provides that, "[w]hen making a determination that an overpayment has occurred, the agency shall prepare and issue an audit report [FAK] to the provider showing the calculation of the overpayments." Thus, the results of the application of the formula provide evidence of the amount of overpayment, and this evidence is set forth in the FAR. Accordingly, neither the formula nor, to the extent it is somehow different, the methodology, creates rights or adversely affects others.

The Harvey progeny also include cases that emphasize that agency statements must undergo rulemaking procedures if they require compliance. See Vanjaria, 675 So.2d at 255-56; Dep't of Transp. v. Blackhawk Quarry, 528 So.2d 447, 450 (Fla. 5th DCA 1988); McCarthy v. Dep't of Ins. Treasurer, 479 So.2d 135, 137 (Fla. 2d DCA 1985). The tax assessment procedure in Vanjaria was held to be a rule because it "requires that the auditors review relevant documents such as purchase orders, lease payments, and legal descriptions, in addition to measuring the property in an effort to develop a taxable factor which is then applied to property." 675 So.2d at 256. The Standard Operating Procedure in Blackhawk Quarry was held to be a rule because it required that specific criteria be met in order to participate in state construction projects. See 528 So.2d at 450. And the statement in McCarthy, a letter setting qualifications for eligibility to take the exam to be a fire safety inspector, was held to be a rule because it required compliance with "categoric requirements as a prerequisite for obtaining certification." 479 So.2d at 137. In the instant case, unlike Vanjaria, Blackhawk Quarry, and McCarthy, neither the formula nor the audit statistical methodology sets forth a "categoric requirement" of specific criteria or a mandatory action with which to comply. Accordingly, by this measure also, the formula is not a rule.

For the foregoing reasons, we hold that the cluster sampling formula used by AHCA to calculate Medicaid overpayments is not an unpromulgated rule.

REVERSED and REMANDED.

PADOVANO and POLSTON, JJ., concur.