REINHARDT, Circuit Judge:
One hundred and sixty-three hospitals (the "Hospitals") seek payment for alleged shortfalls in Medicare reimbursements for fiscal years 1985 (FY 1985) and 1986 (FY 1986).
STATUTORY AND REGULATORY FRAMEWORK
The Secretary of the United States Department of Health and Human Services (the "Secretary") reimburses health-care providers for medical services provided to Medicare patients under the Health Insurance for the Aged Act ("the Act"). 42 U.S.C. § 1395f(b)(1). Before October 1, 1983, hospitals were reimbursed the lesser of their "customary charges" or the "reasonable costs" of their services. In an attempt to lower Medicare expenditures, Congress created the Prospective Payment System ("PPS"). Under PPS, the Secretary normally reimburses hospitals at a fixed amount for each patient discharged regardless of the costs incurred by the hospital. 42 U.S.C. § 1395ww(d)(1). Congress intended this method of payment to encourage hospitals to cut costs and increase efficiency. See Methodist Hosp. of Sacramento v. Shalala, 38 F.3d 1225, 1227 (D.C.Cir.1994) (citing Congressional Reports).
The fixed amount paid to hospitals under PPS is based on a standardized amount that is multiplied by a weighing factor. 42 U.S.C. § 1395ww(d)(2)(G), (3)(D). The standardized amount is a base amount that equals the average Medicare allowable cost per discharge for all hospitals participating in the Medicare program in a base year, which is adjusted according to regional wage variations, indirect medical education costs, and hospital case mix. 42 U.S.C. § 1395ww(d)(2). The weighing factor is a multiplier based on the diagnosis related group ("DRG") in which the discharged patient's illness falls. The Secretary has created 470 DRGs, each with a weight derived from the relative cost to treat a patient in that DRG. 42 U.S.C. § 1395ww(d)(4).
Hospitals first became subject to PPS in fiscal year 1984, beginning on October 1, 1983. 42 U.S.C. § 1395(d)(1)(A)(i). Thereafter, the Secretary phased in PPS over four years. During the phase-in years, Medicare reimbursements comprised two portions. The "hospital specific portion" was calculated under the prior system, based on the hospital's actual costs. 42 U.S.C. §§ 1395ww(d)(1)(A), (C). The other portion, the "federal rate," was determined under PPS. Id. The hospital specific portion of the reimbursements was 75% in FY 1984, 50% in FY 1985, 45% in 1986, and 25% in 1987, while the federal rate increased correspondingly. 42 U.S.C. § 1395ww(d)(1)(C). Thereafter, all reimbursements were calculated under PPS.
Under PPS Congress also provided for additional payments for patient discharges that qualified as "outlier" cases, which involved unusually costly or lengthy patient treatments. "Day outliers" occur when a patient's length of stay ("LOS"), measured in days, exceeds the mean LOS for a particular DRG by a fixed number of days or standard deviations. 42 U.S.C. § 1395ww(d)(5)(A)(i). "Cost outliers" occur when the cost exceeds a fixed multiple of a particular DRG's payment rate or exceeds the rate by a fixed dollar amount. 42 U.S.C. § 1395ww(d)(5)(A)(ii). The amount of additional payments for these cases "shall be determined by the Secretary and shall approximate the marginal cost of care" beyond the applicable cut-off point. 42 U.S.C. § 1395ww(d)(5)(A)(iii). Finally, the statute provides that
42 U.S.C. § 1395ww(d)(5)(A)(iv) ("Clause (iv)") (emphasis added).
The Secretary derives the standardized amount from the cost of all hospitals' discharges in a given year, including the outlier cases. In order to avoid overpaying hospitals generally for their expenses, the Secretary reduces the average standardized amount "by a proportion equal to the proportion
The Secretary publishes the methodology and data used in computing all DRG rates, including the thresholds to be used for calculating outlier payments, prior to the beginning of each fiscal year. See 42 U.S.C. § 1395ww(d)(6). For FYs 1984 and 1985, the Secretary used the 1981 Medicare Provider Analysis and Review file ("1981 MEDPAR"), a random sample of 20% of all hospital discharges during 1981, to set the thresholds so that total estimated outlier payments would equal 5.0% of total payments. 49 Fed.Reg. 34728, 34769, 34795 (Aug. 31, 1984).
FACTS AND PROCEDURAL HISTORY
Each of the Hospitals appealed its Medicare reimbursements for FYs 1985 and 1986 to the Provider Reimbursement Review Board ("PRRB") pursuant to 42 U.S.C. § 1395oo. The PRRB accepted jurisdiction over each appeal and granted all of the Hospitals' requests for expedited judicial review, expressing no views on the merits.
The Hospitals then brought this action in federal district court. The Hospitals first contended that the Secretary breached her statutory duty under Clause (iv) by failing to make outlier payments equalling five to six percent of the total estimated or projected total PPS payments in FYs 1985 and 1986. The parties agree that for FY 1985, outlier payments totaled three percent, $241 million less than five percent of total estimated federal rate payments; and for FY 1986, they totaled 4.4 percent, $101 million less than five percent of total estimated federal rate payments. The Hospitals also contended that the FY 1985 thresholds, which were based on the 1981 MEDPAR file and were used for FY 1985 and the first eight months of FY 1986, were determined in an arbitrary and capricious manner in violation of the Administrative Procedure Act (APA), 5 U.S.C. § 706(2)(A). This contention is based on the ground that the MEDPAR 1981 file failed to reflect the anticipated decreasing average LOS under PPS and that as a result too few patient discharges were deemed "outliers." The Hospitals requested declaratory relief with respect to the Secretary's asserted violations of both the Act and the APA and an order awarding each of them a pro-rata share of sums due.
On cross-motions for summary judgment, the district court held with respect to the Act that "it cannot be clearly determined that Congress intended the Secretary to recalculate outlier payments after final payment data became available," and that the Secretary's construction of the statute was permissible.
We discuss the APA claim first. The Hospitals contend that the Secretary acted arbitrarily and capriciously when setting the outlier thresholds for FY 1985 because she relied on the 1981 MEDPAR data instead of preliminary data available from FY 1984 and failed to offer an adequate explanation for doing so. The Secretary contends that she properly decided not to rely on incomplete and preliminary data and adequately explained her decision. Although the FY 1984 thresholds are not at issue, the Secretary relies, in part, on explanations from the 1984 rulemaking record, as well as on explanations that are outside both the 1984 and 1985 records.
For FY 1985 the Secretary established the day outlier and cost outlier thresholds so that total estimated outlier payments for FY 1985 would be 5.0 percent of total estimated payments. 49 Fed.Reg. 34728, 34768 (Aug. 31, 1984). Day thresholds were set so as to equal the mean LOS within each DRG plus the lesser of 22 days (as compared to 20 days for FY 1984) or 1.94 standard deviations. 49 Fed.Reg. 34728, 34768. All thresholds were set according to the MEDPAR 1981 data. See 49 Fed.Reg. 27447 (July 3, 1984).
The Secretary twice addressed the decision to continue using the 1981 MEDPAR data for the 1985 calculations. First, in response to comments criticizing the outlier thresholds on the ground that they would result in outlier payments "well below the 6.0 percent expected outlier payment level," the Secretary stated:
49 Fed.Reg. 34769 (Aug. 31, 1984).
Second, in response to comments suggesting that the mean LOS for individual DRGs be recalculated based on actual PPS experience in order to accommodate decreasing LOS, the Secretary responded that
Id. at 34768-69.
The Secretary did, however, use preliminary FY 1984 data to adjust all DRG weighting factors by 1.05 percent to respond to increases in hospital case-mix values.
49 Fed.Reg. 34728, 34771 (Aug. 31, 1984).
Under the APA, we determine if the Secretary's action was "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with the law." 5 U.S.C. § 706(2)(A); Mt. Diablo Hosp. v. Shalala, 3 F.3d 1226,
A rule may also be invalidated under the APA if an agency fails to explain the rule adequately. Mt. Diablo Hosp., 3 F.3d at 1233-34; Yerger v. Robertson, 981 F.2d 460, 463 (9th Cir.1992). However,
Mt. Diablo, 3 F.3d at 1234 (quoting South Carolina ex rel. Tindal v. Block, 717 F.2d 874, 886 (4th Cir.1983)).
We conclude that the Secretary failed to offer an adequate explanation for the decision to rely on the 1981 MEDPAR data rather than more recent data that would reflect decreasing LOS. The Secretary's response that the outlier thresholds would not be adjusted because standardized amounts would also need to be adjusted is simply no explanation at all. The fact that other variables would need to be adjusted hardly identifies any of "the major issues of policy" involved in the Secretary not adjusting either the thresholds or the standardized amounts. Further, the response indicates that the Secretary was quite aware that LOS had decreased after the implementation of PPS.
The Secretary's other explanation for using the 1981 MEDPAR file, that "[b]ased upon outlier and DRG payment data received through July 27, 1984, there is no evidence to suggest that total outlier payments are below the levels intended," was simply not supported by evidence before the Secretary. The Secretary does not dispute that "[the data] consisted of raw billing data suggesting that outlier payments reported to the agency during the first half of the year were less than the target for fiscal year 1984," Sec. Br. at 54, but only argues, first, that the preliminary data was incomplete, and, second, that the final figures for outlier payments were higher than the preliminary data indicated. The Hospitals' experts, on the other hand, state that the preliminary data clearly indicated outlier payments were going to fall below their projected amounts. Even resolving all ambiguities and drawing all inferences in favor of the Secretary, her explanation that there was no evidence of an outlier shortfall was simply not supported by the record before her and did not explain her failure to use the more recent data.
The Secretary's use of the preliminary 1984 data to adjust for increasing case-mix values further undercuts her failure to use it with respect to the outlier thresholds. The Secretary defended the use of such data because it "reflected potential behavioral changes" caused by PPS. Given that decreased LOS was a primary goal of PPS, and given the prominence of LOS values in calculating outlier thresholds, the Secretary's failure to explain why the preliminary 1984 data
The "larger" administrative record does not, contrary to the Secretary's contentions, help to identify the major issues considered by her in reaching her decision. The Secretary points to two statements made in the 1984 rulemaking record, which stated that the 1981 MEDPAR file would be used to determine thresholds for FY 1984. Medicare Program; Prospective Payments for Medicare Inpatient Hospital Services, 48 Fed.Reg. 39752, 39843 (Sept. 1, 1983).
Medicare Program; Prospective Payments for Medicare Inpatient Hospital Services, 49 Fed.Reg. 234, 255 (Jan. 3, 1984). Second, in a preliminary impact analysis of the expected behavior changes under PPS, the Secretary observed that cost-cutting through reduced LOS would vary from hospital to hospital, depending on the backlog and occupancy rates, and the ratio of the cost of ancillary services used to shorten hospital stays as weighed against the savings from reductions in length-of-stay. Id. at 304.
Even assuming that we may examine this "larger" record and that statements made in the 1984 record could provide explanations for a decision made the next year, neither the Secretary's response nor impact analysis explains the decision not to use the partial 1984 data for FY 1985. The Secretary's response on "conditions that could not be quantified" is of doubtful relevance because none of the commenters was referring to decreasing LOS, and further, because at the time PPS had been implemented for just three months. Second, the Secretary's prediction that LOS reductions would vary from hospital to hospital according to certain factors does not explain how such variations would cast doubt on the validity of the preliminary 1984 data reflecting decreasing LOSs. Thus, even according to the "larger" rulemaking record, the Secretary never adequately explained why the preliminary 1984 data, used to adjust other values for that year, was improper to use with respect to determining outlier thresholds.
Connerton Decl. ¶¶ 10, 12, 15.
"We review [a] district court's decision to exclude extra-record evidence for an abuse of discretion." Southwest Ctr. for Biological Diversity v. U.S. Forest Svc., 100 F.3d 1443, 1447 (9th Cir.1996). A district court may go outside the administrative record for the purposes of background information or "for the limited purposes of ascertaining whether the agency considered all the relevant factors or fully explicated its course of conduct or grounds of decision." Asarco, Inc. v. United States Envtl. Protection Agency, 616 F.2d 1153, 1160 (9th Cir.1980); see Idaho Conservation League v. Mumma, 956 F.2d 1508, 1520 n. 22 (9th Cir.1992). Of course, explanatory materials cannot be used to offer new rationalizations for agency action. See Bunker Hill Co. v. EPA, 572 F.2d 1286, 1292 (9th Cir.1977). That is precisely what the Connerton affidavit appears to do. It offers explanations for the Secretary's decision that easily could have been part of the administrative record had they been a basis for the Secretary's decision. The Connerton explanation does not, however, appear in the record. Thus, the district court improperly considered portions of the Connerton Declaration, and we will not consider them on appeal. See Asarco, 616 F.2d at 1160-61 (finding district court to have improperly considered expert testimony outside of the administrative record).
More important, in Mt. Diablo, the fact at issue that was not addressed by the Secretary was only of "minor significance at the time." 3 F.3d at 1234. In this case, the decreasing LOSs were highly significant — their decline was, as the Secretary has repeatedly acknowledged, a highly anticipated result of PPS. Nothing in the administrative record identifies the factors that motivated the Secretary to decide not to use the preliminary FY 1984 data that reflected that decline. See Beno v. Shalala, 30 F.3d 1057, 1074 (9th Cir.1994) (reversing and remanding to Secretary when administrative record did not indicate whether she considered important issues). Because the Secretary failed to explain her decision not to use the preliminary 1984 data with respect to the 1985 day outlier thresholds, and because the data was highly significant to an accurate determination of those thresholds, we find her to have abused her discretion in their determination.
Normally, if the "`record before the agency does not support the relevant agency action,'" the proper course of action is to remand to that agency for further investigation and explanation. UOP v. United States, 99 F.3d 344, 351 (9th Cir.1996) (quoting Florida Power & Light Co. v. Lorion, 470 U.S. 729, 744, 105 S.Ct. 1598, 84 L.Ed.2d 643 (1985)); see Beno v. Shalala, 30 F.3d at 1076 (remanding to Secretary for "additional consideration" of issues which administrative record shows she failed to consider). Therefore, on remand, the Secretary ordinarily would consider the preliminary 1984 data that was available when the original decision was made and either make appropriate use of the data now or provide a satisfactory reasoned explanation of her initial decision not to do so. It is now possible, however, for the Secretary to set the FY 1985 day-outlier thresholds according to the complete and more reliable data contained in the complete 1984 MEDPAR file, as she did in setting the thresholds for FY 1986. Because (1) the reimbursements at issue arose over thirteen years ago, (2) further recondite litigation might ensue if the Secretary merely attempts to provide further explanation for her decision, and (3) it is now possible not merely to estimate but rather to determine the precise thresholds necessary so that the outlier payments would equal five percent of the total projected payments, we direct the Secretary to use the final 1984 MEDPAR file to adjust the FY 1985 day-outlier thresholds so as to achieve the appropriate overall percentage payment and to pay the Hospitals the difference between what they are entitled according to the revised thresholds and what they were actually paid for FY 1985 and the pertinent portion of FY 1986.
The relief we grant on the APA claim renders moot the Hospital's claim under the Act. Because with the additional payments
We reverse the district court's decision granting summary judgment on the APA claim and grant the Hospitals' cross-motion for summary judgment on that claim. We remand to the Secretary for re-determination of the FY 1985 day outlier thresholds with instructions to recompute them using the 1984 MEDPAR file and to reimburse the Hospitals for any difference in outlier payments.
REVERSED AND REMANDED.
It is also undisputed that a decrease in LOS for FY 1984 actually occurred. A 1984 PPS Impact report issued on August 1986 indicated that actual outlier payments for FY 1984 was "largely due to the decline in length of stay in recent years and the decrease in charges per case under the PPS." Finally, an Impact Report for 1985 PPS concluded that for FY 1984 day outlier payments had been about 60% lower than anticipated due to decreasing LOS.
For FY 1986 the Secretary revised the day outlier and cost outlier thresholds using 1984 MEDPAR data. 50 Fed.Reg. 35646, 35709 (Sept. 3, 1985). In a response to a comment that FY 1985 outlier payments were lower than expected, the Secretary pointed out that a substantial decrease in LOS reduced the number of cases qualifying for outlier payments. 50 Fed.Reg. 35710 (Sept. 3, 1985).
In response to a comment suggesting that DRG relative weights be recalibrated, the Secretary responded,
50 Fed.Reg. 35646, 35654 (September 3, 1985). These observations with respect to the FY 1984 data help to explain why the Secretary might not have wanted to use it for FY 1985. Since they are not part of the 1985 administrative record, however, they cannot affect our analysis with respect to the APA claim.