SETH, Circuit Judge.

This appeal presents a challenge to the "Malpractice Rule," a regulation promulgated in 1979 by the Department of Health and Human Services. 42 C.F.R. § 405.452(b)(1)(ii). The rule is a substantial departure from the reimbursement methods previously used to compensate hospitals for care of Medicare patients. With the large increase in malpractice insurance costs, the Department had become convinced that hospitals were overpaid by Medicare for the costs of such insurance. See 44 Fed. Reg. 15,744, 15,745 (1979). Hospitals across the country unanimously opposed the change. This action and many others followed in the rule's wake.

Prior to the Malpractice Rule, a hospital's malpractice insurance costs were part of the "General and Administrative" cost center (G A), a catchall category for many of the indirect costs of providing medical services. When it came time to apportion those G A costs between Medicare and non-Medicare populations, the hospital's overall Medicare utilization ratio was applied to the G A costs.

From the outset of the Medicare program, the Secretary realized certain G A costs might be allocated disproportionately to Medicare or non-Medicare patients. It was concluded that any imprecision would be cured by the "averaging principle." Thus in an early policy statement the Department declared that "it is presumed that where a particular cost might be allocated disproportionately to or from the program, there will be other costs disproportionately allocated in the other direction which will compensate for the first cost." Intermediary Letter No. 234 (June 2, 1967). The averaging principle had been the guiding rationale. The new Malpractice Rule was a radical departure from the Department's historical practice. Cf. Boswell Memorial Hospital v. Heckler, 749 F.2d 788 (D.C. Cir.).

In its Notice of Proposed Rulemaking (NPRM), the Department declared that the utilization method caused the Medicare program to pay an excessive portion of malpractice insurance costs. 44 Fed.Reg. 15, 744, 15,745(1979). To buttress this conclusion the Secretary referred to an unnamed "consultant's study" which supposedly established that malpractice damage awards were significantly lower for Medicare patients than for the general patient population. This difference stemmed from the shorter life expectancy of Medicare patients and their reduced earnings potential.

During the comment period public reaction was overwhelmingly negative. Hospitals attacked the statistical validity of the Department's study. Additionally, they elaborated upon the bizarre results that would follow from applications of the proposed rule. Nonetheless, the rule in its final form differed little from that proposed in the NPRM.

The rule excludes malpractice insurance costs from the G A cost center and treats them as a separate item. Reimbursement is instead "based on the dollar ratio of the provider's Medicare paid malpractice losses to its total paid malpractice losses for the current cost reporting period and the preceding 4-year period." 42 C.F.R. § 405.452(b)(1)(ii). This result obtains regardless of the extent of Medicare patient use of the hospital. The rule also makes provision for some compensation in the event a hospital has no paid claims during the five-year period. A "national ratio" of 5.1 per cent applies. This figure is a product of the same study that provided the impetus for the rule.

The appellant is a 200-bed acute care medical facility. In its 1980 cost report for Medicare the appellant made two claims for compensation which are now in dispute. Like many hospitals across the country. Humana did not apportion its malpractice insurance costs on the basis of the new Malpractice Rule. Instead, it presented a claim based on the old utilization method. Humana also presented a claim for other G A costs it claimed were exclusively incurred on behalf of Medicare patients.

After following the administrative route the appellant filed suit in the United States District Court for the District of Colorado, seeking damages and declaratory and injunctive relief. Humana claimed the Malpractice Rule violated the substantive provisions of the Medicare Act because it failed to compensate the provider for its reasonable costs. 42 U.S.C. § 1395f(b). The appellant alleged the rule resulted in illegal "cross-subsidization," in contravention of section 1395x(v)(1)(A)(i) of the Act. Humana also claimed the rule was arbitrary and capricious and an abuse of the Secretary's discretion. 5 U.S.C. § 706(2)(A). Finally, Humana made a claim in the alternative for total reimbursement of other indirect G A costs it viewed as exclusively attributable to the Medicare program. The crux of this claim is that if the Malpractice Rule is valid, then the averaging principle can no longer be a defense to attempts by providers at "discrete costing" other G A items.

The parties filed cross-motions for summary judgment. The trial court ruled in favor of the Secretary. As to the substantive issues the court recognized that there might be "problems" with the Westat Report, the "consultant's study" used as a basis for the new rule. Yet the court found that the Secretary had authority to employ different methods in calculating reimbursement. The rule, therefore, had "some basis" and was not arbitrary and capricious. The court found no fundamental discrepancy between using the averaging principle for all other G A costs while parsing out malpractice insurance costs for separate treatment under the new rule. The court also rejected Humana's cross-subsidization argument. The court found nothing improper in the Department's refusal to provide complete reimbursement for the other G A costs which Humana wished to treat as solely attributable to Medicare. The court held that the Secretary had a rational basis for altering the reimbursement formula.

Appellant asserts the contention that the new rule is an arbitrary and capricious exercise of the Secretary's informal rule-making authority under 5 U.S.C. § 706(2)(A). While the scope of review is narrow we must nevertheless act within the scope directed by the Supreme Court. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 91 S.Ct. 814, 28 L.Ed.2d 136. The Supreme Court restated the standard of review in Motor Vehicle Manufacturers Association of the United States v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 103 S.Ct. 2856, 77 L.Ed.2d 443:

"Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress had not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise."

A rational connection must be found between the facts before the agency and the rule-making choice made. Id. 103 S.Ct. at 2866-67. See also Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 245, 9 L.Ed.2d 207.

In the case before us we must conclude that the evidence that was before the agency is contrary to the administrative action ultimately taken. The fundamental nexus between evidence and agency action is absent. This is demonstrated by the deficiencies of the Westat Report, which the Report itself emphasizes. This is the unidentified "consultant's study" in the NPRM that purportedly justified the rule's promulgation.

This court must apply added requirements when there has been a significant departure from long-standing policy. Motor Vehicle Manufacturers Association of the United States v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443. This court recently reiterated the requirements by the holding in Dimension Financial Corp. v. Board of Governors of the Federal Reserve System, 744 F.2d 1402 (10th Cir.), that: "[T]he agency must clearly articulate the basis for the change. This is not only an aspect of any procedural notice requirement in rule-making but is a substantive need." See also Midwestern Transportation, Inc. v. I.C.C., 635 F.2d 771, 777 (10th Cir.).

Finally, the Secretary's discretion is limited also by the statutory scheme of the Medicare Act; some of the Act's directives are mandatory. See, e.g., 42 U.S.C. § 1395x(v)(1)(A); St. John's Hickey Memorial Hospital v. Califano, 599 F.2d 803, 813 (7th Cir.).

We agree with the appellant that the rule-making record had to support two basic propositions. First, the utilization apportionment approach had to overstate Medicare's share of malpractice insurance costs. Thus the record had to demonstrate that the utilization method no longer worked with regard to malpractice premium costs. Secondly, any imbalance in that expenditure had to evade the operation of the averaging principle. The rule-making record fails to establish these propositions.

One of the assumptions underlying the Malpractice Rule is that Medicare patients recover smaller malpractice awards than those persons in the general patient population. Thus the insurance costs for Medicare patients should be less. However, insurance premiums are derived from a variety of considerations. Risk cannot be calculated on the basis of one variable. Indeed, the Secretary acknowledged when issuing the rule in its final form that "insurance companies generally do not determine insurance rates for malpractice insurance based upon the financial status of the patients." 44 Fed.Reg. 31,642 (1979). Insurance rates depend on a broad pool of data, not on the isolated claims experience of a hospital during a defined time period. The Westat Report makes no pretense of linking insurance rates and paid claims. See Mt. Carmel Mercy Hospital v. Heckler, 581 F. Supp. 1311, Medicare Medicaid Guide (CCH) 33,529 at 9563 (E.D.Mich.).

It was also necessary for the rule-making record to show that Medicare patients filed few malpractice claims in relation to their use of hospital resources. The Secretary proffers the Westat Report in support of that proposition. The study itself acknowledges that it was based on a statistically insignificant sample with limits thereby placed on its application. We agree with the court in Abington Memorial Hospital v. Heckler, 576 F. Supp. 1081, 1087, aff'd, 750 F.2d 242 (3d Cir.), that "an agency need not await perfect data before taking regulatory action. There are limits . . . to the degree of imperfection that is permissible. . . ." The Secretary exceeded those limits. When an agency adopts a regulation based on a study not designed for the purpose and which is limited and criticized by its authors on points essential to the use sought to be made of it, the administrative action is arbitrary and capricious and a clear error in judgment. Almay, Inc. v. Califano, 569 F.2d 674 (D.C. Cir.). Cf. Camp v. Pitts, 411 U.S. 138, 143, 93 S.Ct. 1241, 1244, 36 L.Ed.2d 106.

We base our conclusion not on our evaluation of the Report but on the limitations expressed in the Report itself. The Westat Report's authors warned that "[c]onclusions must be drawn very cautiously because of the possibility of biased data." Bias is apparent on several grounds. First, there is the problem of the size and limited time frame of the group studied. The Report examines the malpractice claims of nine insurance companies over a four-month period in 1976. Out of some 4,000 claims, only 166 are identified as originating with Medicare patients. Indeed, the Medicare status of a large portion of the sample could not even be identified. Report at 5-9, 5-10. The authors again declared that the utility of the information is limited to the "time frame for which the data were collected and only to the universe of claims closed by the nine participating companies." Report at 2-9.

Additional sources of bias exist. The Report makes no attempt to distinguish between malpractice claims against hospitals and those against doctors, dentists, nursing homes, or other insureds. Report at 2-2. Secondly, the Report has no means of taking into account medical facilities that are self-insurers. Id. A sizable minority of hospitals self-insure, and their claims experience entirely evades the Report's scrutiny. Any one of these problems taken alone might be enough to invalidate the rule. However, these flaws taken as a whole render reliance by the agency on this "evidence" as irrational. The connection between the rule-making record and the action ultimately taken is too attenuated to uphold the regulation.

Finally, we note that the Report in its most basic sense was never designed or intended by its makers to answer the questions or support the propositions the Secretary wished. See Report at 1-3. We hold, therefore, that the Malpractice Rule violates the provisions of the Administrative Procedure Act as an arbitrary and capricious exercise of the Secretary's informal rule-making powers.

Because of our disposition of the cause under the Administrative Procedure Act we need not reach the appellant's argument that the Malpractice Rule results in cross-subsidization and thereby violates the substantive provisions of the Medicare Act.

In its alternative claim the appellant seeks compensation for what it views as Medicare's actual share of other G A costs. In view of our disposition of the validity issue we do not reach this question nor the general adequacy of reimbursement.

The case is REVERSED and REMANDED to the District Court for further proceedings or for remand to the agency for resolution of appellant's claim for reimbursement. IT IS SO ORDERED.