Cross appeals from a judgment of the Supreme Court (Benza, J.), entered June 18, 2003 in Albany County, which, inter alia, partially granted petitioners’ applications, in six proceedings pursuant to CPLR article 78, to recalculate respondents’ Medicaid reimbursement rates for the period between April 1995 and October 1997.
These proceedings stem from earlier litigation in which numerous nursing homes located in New York successfully asserted that certain adjustments to the state’s Medicaid reimbursement rates made by a 1987 State Plan Amendment (hereinafter the 1987 Adjustment) were invalid for failure to comply with the Boren Amendment (42 USC former § 1396a [a] [13] [A]). As then existing, the Boren Amendment afforded health care providers certain substantive and procedural rights by requiring states to make findings and assure the Health Care Financing Administration (hereinafter the HCFA) that their reimbursement plans provided “reasonable and adequate” payment to “meet the costs which must be incurred by efficiently and economically operated facilities” (42 USC former § 1396a [a] [13] [A]; see Wilder v Virginia Hosp. Assn., 496 US 498, 509-510 [1990]). Due to insufficient findings, the 1987 Adjustment was found to be in violation of the Boren Amendment and judicially invalidated “until such time [as] proper findings [were] submitted and approved by HCFA” (Pinnacle Nursing Home v Axelrod, 928 F2d 1306, 1318 [1991]; see Matter of Avon Nursing Home v Axelrod, 83 NY2d 977, 983 [1994]).
Beginning in April 1995, respondents reimplemented the 1987 Adjustment through three State Plan Amendments (hereinafter SPAs): SPA 95-23, SPA 95-24 and SPA 96-24. In support of these SPAs, respondents made new findings (hereinafter the 1995 *922findings) using the three-step procedure mandated in Pinnacle Nursing Home v Axelrod (supra at 1314-1315) to meet the requirements of the Boren Amendment. Respondents first identified efficiently and economically operated facilities (hereinafter EEOFs) as those nursing home facilities with costs less than the median of the costs reported by all facilities. Respondents next calculated the “necessary” costs that must be incurred by EEOFs using, among other things, a Medicaid revenue adjustment (hereinafter the MRA) that effectively reduced a facility’s reimbursable costs in proportion to the amount of revenue received from sources other than Medicaid. The amount of the MRA was calculated using a formula derived from a regression analysis of the relationship between the costs reported by all facilities in 1993 and the percentages of their revenues received from Medicaid. Finally, respondents compared the necessary costs calculated using the MRA with the state’s reimbursement rates to show that those rates more than met the necessary costs of EEOFs.
When the 1995 findings were later held to be inadequate to support the SPAs because, among other things, the regression analysis underlying the MRA was found to be flawed, respondents were required to make new findings without the MRA (see Valley View Manor Nursing Home v De Buono, 1997 WL 855508 [US Dist Ct, WD NY, Telesca, J., 89 Civ 0706T, 89 Civ 0744T, Feb. 12, 1997]). Due to the repeal of the Boren Amendment, however, respondents’ appeal from that ruling was dismissed as moot and the underlying complaint was dismissed (see Hall v Sullivan, 129 F3d 113 [1997] [op published at 1997 WL 643921, 1997 US App LEXIS 37069 [1997]).
Petitioners, many of the same nursing homes, then commenced these consolidated proceedings challenging the three SPAs and seeking recalculation of their reimbursements for the period between April 1995 and October 1997. Petitioners alleged that, among other things, the 1995 findings were insufficient to meet the procedural requirements of the Boren Amendment. To resolve the factual issue of whether the MRA had a statistically valid basis in respondents’ regression analysis, Supreme Court reviewed the parties’ submission of the record of the hearing held in Valley View Manor Nursing Home v De Buono (supra) and found the regression analysis insufficient to support the MRA. The court then concluded that while the 1995 findings failed to support the three SPAs due to their inclusion of the MRA, respondents’ revised findings in 1997 (hereinafter the 1997 findings), which did not include the MRA, effectively cured the defects as to SPA 95-24 and SPA 96-24, but *923not as to SPA 95-23. Supreme Court ordered respondents to recalculate petitioners’ reimbursement rates for the relevant time period without using the SPA 95-23 adjustment. Respondents and petitioners cross-appeal.
On their appeal, respondents argue that the regression analysis upon which the MRA is based was a rational approach to calculating necessary costs. More specifically, they assert that the relationship between Medicaid revenue and costs, upon which the MRA is predicated, was suggested by anecdotal evidence of nursing home practices, supported by the economic principle that spending practices are influenced by revenue streams and quantified through the use of an accepted statistical tool.
The standard of review applicable to an administrative action such as that taken here by respondents is whether it had a rational basis in the record, and was not unreasonable, arbitrary or capricious (see CPLR 7803 [3]; Kuppersmith v Dowling, 93 NY2d 90, 96 [1999]; New York State Assn. of Counties v Axelrod, 78 NY2d 158, 166 [1991]). A rate-setting action of the Department of Health may be annulled only “upon a compelling showing that the calculations from which it derived were unreasonable” (Matter of Catholic Med. Ctr. of Brooklyn & Queens v Department of Health of State of N.Y., 48 NY2d 967, 968 [1979]; see New York State Assn. of Counties v Axelrod, supra at 166; Matter of Society of N.Y. Hosp. v Axelrod, 70 NY2d 467, 473 [1987]). Further, “we note that ‘[a] trial court’s findings are not to be lightly set aside unless its conclusions could not have been reached based upon any fair interpretation of the evidence’ ” (Silverman v Mergentime Corp./J.F. White, Inc., 252 AD2d 925, 926 [1998], quoting Osterhout v Mesivta Sanz of Hudson County, 226 AD2d 893, 894 [1996]).
The record of the hearing in Valley View Manor Nursing Home v De Buono (supra) contains ample evidence supporting Supreme Court’s finding that respondents’ regression analysis produced a statistically invalid and, thus, unreasonable formulation of the relationship between Medicaid revenue and nursing home costs. Petitioners’ expert, Jack Zwanziger, testified that while regression analysis is a well-accepted statistical tool, it was used improperly by respondents. Zwanziger reported that he performed an F Test, a standard statistical test which gauges whether the variances in two groups of data are sufficiently similar for the groups to be combined in a single regression analysis, and this test demonstrated that it was statistically inappropriate to combine the data from EEOFs and non-EEOFs. Zwanziger also stated that the inclusion of data from non*924EEOFs meant that the MRA reduced the necessary costs of EEOFs much more than the correlation between their revenue and costs would warrant. Zwanziger noted that it is also conceptually inappropriate to include the non-EEOF data in the regression because reimbursement rates are to be based upon the necessary costs of EEOFs only (see Public Health Law § 2807 [3]). In light of this testimony and the absence of empirical data from any EEOF to confirm the mathematical validity of the formula produced by the regression analysis, we find that the evidence supports Supreme Court’s conclusion that the MRA lacked a rational basis.
On their cross appeal, petitioners contend that none of the SPAs should have become effective prior to October 1, 1997 because they were not “approvable” until that date. However, since the 1997 findings constituted supplemental materials that were submitted to HCFA before it ruled on the sufficiency of SPA 95-24 and SPA 96-24, we agree with Supreme Court’s conclusion that those SPAs were “approvable” within the meaning of 42 CFR 447.256 (c) as of the first day of the calendar quarter in which those SPAs were submitted (see Independent Acceptance Co. v California, 204 F3d 1247, 1256 [2000]). Contrary to respondents’ argument, however, the 1997 findings do not provide a similar effective date for SPA 95-23. Inasmuch as that SPA was approved before the 1997 findings were issued, it was not then an “approvable” submission and it could not become effective prior to the submission of an “approvable” SPA (see State of New York v Shalala, 119 F3d 175, 181 [1997]; Matter of Avon Nursing Home v Axelrod, 195 AD2d 1046, 1047 [1993], affd 83 NY2d 977 [1994]; see also 42 CFR 447.256 [c]).
Finally, we find no merit in petitioners’ arguments that respondents failed to give adequate notice to the public and timely assurances to HCFA for SPA 96-24. The regulations do not require supplemental notice or assurances when the overall effect of the amendments is different because the legislation as enacted does not include all of the amendments listed in the notice (see 42 CFR 447.205, 447.253, 447.255), and it is within HCFA’s discretion to decide when additional notices or assurances are needed (see State of New York v Shalala, supra at 182-183).
Mercure, J.P., Peters, Mugglin and Lahtinen, JJ., concur. Ordered that the judgment is affirmed, without costs.