UNITED STATES COURT OF APPEALS FOR THE D.C. CIRCUIT


CTY LOS ANGELES

v.

SHALALA, DONNA E.


98-5254a

D.C. Cir. 1999


*	*	*


Wald, Circuit Judge: Brought by the owners of Medicare- provider
hospitals ("Hospitals") and the Secretary of Health  and Human
Services ("Secretary"), these cross-appeals pres- ent two issues.
First, under the Medicare statute, must the  Secretary provide
hospitals with retroactive reimbursements  to ensure that aggregate
outlier payments during any given  fiscal year meet minimum statutory
targets? And second,  has the Secretary adequately explained why, when
calculating  outlier thresholds for fiscal years 1985-1986, she relied
on a  1981 database instead of more contemporaneous records from  1984
Medicare discharges? Finding that Congress had spo- ken directly and
unambiguously to the first question, the 


district court granted partial summary judgment to the Hos- pitals.
With respect to the second issue, however, the court  perceived
nothing unreasonable in the Secretary's choice of  data, and entered
judgment accordingly for the Secretary.  Because we disagree with the
district court on both points, we  now reverse.


I. Background


Through a "complex statutory and regulatory regime,"  Good Samaritan
Hosp. v. Shalala, 508 U.S. 402, 404 (1993),  the Medicare program
reimburses qualifying hospitals for the  services that they provide to
eligible patients. See Social  Security Act, Pub. L. No. 89-97, tit.
XVIII, 79 Stat. 286, 291  (1965) (codified as amended at 42 U.S.C. ss
1395-1395ggg  (1994 & Supp. III 1997)). From its inception in 1965
until  October 1983, Medicare compensated hospitals for the "rea-
sonable costs" of the inpatient services that they furnished.  See 42
U.S.C. s 1395f(b). Experience proved, however, that  this system bred
"little incentive for hospitals to keep costs  down" because "[t]he
more they spent, the more they were  reimbursed." Tucson Med. Ctr. v.
Sullivan, 947 F.2d 971,  974 (D.C. Cir. 1991).


To stem the program's escalating costs and perceived inef- ficiency,
Congress fundamentally overhauled the Medicare  reimbursement
methodology in 1983. See Social Security  Amendments of 1983, Pub. L.
No. 98-21, s 601, 97 Stat 65,  149. Since then, this new regime, known
as the Prospective  Payment System ("PPS"), has reimbursed qualifying
hospi- tals at prospectively fixed rates. By establishing pre-
determined reimbursement rates that remain static regard- less of the
costs incurred by a hospital, Congress sought "to  reform the
financial incentives hospitals face, promoting effi- ciency in the
provision of services by rewarding cost/effective  hospital
practices." H.R. Rep. No. 98-25, at 132 (1983),  reprinted in 1983


Calculating prospective-payment rates begins with deter- mining the
"federal rate," a standard nationwide cost rate  based on the average
operating costs of inpatient hospital 


services. See 42 U.S.C. s 1395ww(d)(2)(A)-(B); 49 Fed. Reg.  234, 251
(1984). To account for regional variations in labor  costs, the
Secretary then establishes a wage index that aug- ments the adjusted
standardized payment depending on the  location of a qualifying
hospital. s 1395ww(d)(2)(H),  (d)(3)(E). The final variable is an
additional weighting factor  that reflects the disparate hospital
resources required to treat  major and minor illnesses. s
1395ww(d)(4). For each of 470  medical conditions--known as diagnosis
related groups or  "DRGs"--the Secretary assigns particular weights by
which  the federal rate is to be multiplied. The more complicated  and
costlier the treatment is, the greater the weight assigned  to that
particular DRG will be. To calculate the final "DRG  prospective
payment rate" for a patient discharge, the Secre- tary takes the
federal rate, adjusts it according to the wage  index, and then
multiplies it by the weight assigned to the  patient's DRG. By
statutory mandate, the Secretary must  publish the weights and values
that she will factor into the  prospective-payment calculus before the


Despite the anticipated virtues of PPS, Congress recog- nized that
health-care providers would inevitably care for  some patients whose
hospitalization would be extraordinarily  costly or lengthy. To
insulate hospitals from bearing a  disproportionate share of these
atypical costs, Congress au- thorized the Secretary to make
supplemental "outlier pay- ments." During the years at issue in these
cross-appeals, the  outlier-payment provisions were set forth in four
clauses of  the Medicare statute. 42 U.S.C. s 1395ww(d)(5)(A)(i)-(iv) 
(Supp. IV 1986). With the first two clauses, Congress estab- lished
two classes of outlier payments: day outliers and cost  outliers. s
1395ww(d)(5)(A)(i)-(ii). A hospital could qualify  for a day-outlier
payment if the patient's length of stay  exceeded the mean length of
stay for that particular DRG by  a fixed number of days or standard
deviations.  s 1395ww(d)(5)(A)(i). Along the same lines, the Secretary
 would make cost-outlier payments when a hospital's cost- adjusted
charges surpassed either a fixed multiple of the  applicable DRG
prospective-payment rate or such other fixed 


dollar amount that the Secretary established.  s 1395ww(d)(5)(A)(ii).
In the third clause, Congress provided  that outlier payments "shall
be determined by the Secretary  and shall approximate the marginal
cost of care beyond the  cutoff point applicable" to the day or cost
outlier.  s 1395ww(d)(5)(A)(iii).


It is the fourth and final clause, however, that forms the  textual nub
of the present controversy.  s 1395ww(d)(5)(A)(iv). During 1985 and
1986, paragraph  (5)(A)(iv) provided:


The total amount of the additional payments made under  this
subparagraph for discharges in a fiscal year may not  be less than 5
percent nor more than 6 percent of the  total payments projected or
estimated to be made based  on DRG prospective payment rates for
discharges in that  year.


Id. Traditionally, the Secretary has read paragraph  (5)(A)(iv) to mean
that at the start of each fiscal year, she  must establish the fixed
thresholds beyond which hospitals  will qualify for outlier payments
at levels likely to result in  outlier payments totaling between five
and six percent of  projected DRG payments for that year. In making
this  estimation, the Secretary first settles on the per-diem outlier 
payment, which pursuant to s 1395ww(d)(5)(A)(iii), must ap- proximate
the marginal cost of care. She then examines  historical
Medicare-discharge data to determine which thresh- olds, when
multiplied by the per-diem payment rate, would  probably yield total
outlier payments falling within the five- to-six-percent range in
paragraph (5)(A)(iv). As the Secre- tary observed during the
rulemaking, however, "given the  data available, forecasts of probable
future outlier payments  are inexact." 50 Fed. Reg. 35,646, 35,710
(1985). If it turns  out that the Secretary overestimated the mean
length of stay  for DRGs, the actual total outlier payments at the end
of the  year may amount to less than five percent of estimated DRG-
related payments. Conversely, underestimating the mean  length of stay
might produce outlier payments in excess of six  percent of estimated


Whether the Secretary's projections prove to be correct  will depend,
in large part, on the predictive value of the  historical data on
which she bases her calculations. For fiscal  year 1984, the Secretary
relied on data culled from the 1981  Medicare Provider Analysis and
Review ("1981 MEDPAR")  file, a database containing 1.6 million
Medicare discharges  from 1981. As a product of the old
reasonable-cost system,  however, the 1981 MEDPAR file obviously did
not reflect one  of "[t]he most commonly accepted expectation[s] about
the  PPS at the time of its inception[:] that it would result in 
shorter stays for Medicare patients." Office of Research & 
Demonstrations, Health Care Fin. Admin., U.S. Dep't of  Health & Human
Servs., Pub. No. 03231, Report to Congress:  Impact of the Medicare
Hospital Prospective Payment System  6-13 (1984). By 1984, however,
preliminary data indicated  that the mean length of stay for virtually
all DRGs had, as  anticipated, declined dramatically under PPS. The
Secre- tary, nevertheless, chose to rely again on the 1981 MEDPAR 
file in setting outlier thresholds for fiscal years 1985-1986.  During
those years, though the Secretary set thresholds at a  level projected
to result in outlier payments at or above  paragraph (5)(A)(iv)'s
five-percent floor, actual outlier pay- ments in 1985 constituted only
3.0 percent of estimated DRG- related payments and in 1986 they
amounted to 4.4 percent.1  Given the enormity of the Medicare program,
these seemingly  modest percentage differences represent substantial
sums of  money: $241 million for fiscal year 1985 and $101 million for


Because actual outlier payments for fiscal years 1985-1986  amounted to
less than five percent of projected DRG-related  payments, the
Hospitals petitioned the Secretary for retroac- tive reimbursements to
satisfy the difference. According to  the Hospitals, paragraph
(5)(A)(iv) does more than instruct  the Secretary about where she
should set outlier thresholds 




__________

n 1 Note that these percentages were derived from the actual total  DRG
prospective payments in 1985 and 1986, not the projected  payments as
set forth in the statute. Both parties have agreed that  this is the
only practical method of calculating shortfalls at this  point. See
Br. of Sec'y at 13 & n.4; Br. of Hosps. at 4 & n.3.


at the beginning of each fiscal year; it affirmatively com- mands her
to recalibrate retroactively the outlier thresholds  if, after the
fiscal year concludes, actual outlier payments do  not equal at least
the five-percent statutory target. More- over, the Hospitals claimed
that the Secretary had acted  arbitrarily and capriciously by relying
on the 1981 MEDPAR  file when she forecast outlier thresholds for
fiscal years 1985- 1986. The Secretary rejected both claims, and the
Provider  Reimbursement Review Board authorized the Hospitals to  seek
expedited judicial review pursuant to 42 U.S.C.  s 1395oo(f).


In an opinion and order dated January 20, 1998, the district  court
granted in part and denied in part both the Secretary's  and the
Hospitals' cross-motions for summary judgment. In  granting a portion
of the Hospitals' motion, the court pro- ceeded no further than the
first step of Chevron U.S.A. Inc.  v. Natural Resources Defense
Council, Inc., 467 U.S. 837,  842-43 (1984), concluding that paragraph
(5)(A)(iv) unambigu- ously requires the Secretary to adjust outlier
payments retro- actively to ensure that total actual outlier payments
fall  within the statute's five-to-six-percent range. County of Los 
Angeles v. Shalala, 992 F. Supp. 26, 31-33 (D.D.C. 1998).  Holding,
however, that the Secretary's decision to favor the  1981 MEDPAR file
over the more recent, though preliminary,  1984 data when determining
outlier thresholds was "a rational  choice between two imperfect
databases," the district court  granted the Secretary's motion for
summary judgment on the  Hospitals' claim of arbitrary and capricious
agency action.  Id. at 34-36.


Having determined that paragraph (5)(A)(iv) required the  Secretary to
make retroactive outlier payments to the Hospi- tals, the district
court instructed the parties to meet and  confer on how to structure
the final remedy. On April 30,  1998, based largely on a joint
stipulation filed by the parties,  the district court entered an order
granting judgment to the  parties on each of the claims on which they
respectively  prevailed. While the April 30 order also instructed the 
Secretary to tender retroactive outlier payments to the Hos- pitals,
it did not specify a sum certain to be paid; rather, the 


court left it to the Secretary to calculate the amount owed. 
Thereafter, the Secretary noted a timely appeal and the  Hospitals
noted timely cross-appeals.


II. Analysis


A. Jurisdiction


Before reaching the merits, it is necessary to examine our 
jurisdiction to entertain these cross-appeals. Section 1291 of  the
Judicial Code provides that the courts of appeals may  review "all
final decisions of the district courts of the United  States." 28
U.S.C. s 1291 (1994). In the proceedings below,  the district court,
managing this case as it would any garden- variety civil suit,
adjudicated not only the respective legal  rights of the parties but
also took steps toward decreeing a  proper remedy. Thus in its January
20, 1998 order, the court  resolved the merits of the Hospitals'
claims, and with its April  30, 1998 order, directed the Secretary to
calculate the amount  of outlier payments due to the Hospitals and to
make pay- ment accordingly. This latter order has spawned some confu-
sion about our jurisdiction because of the general rule appli- cable
to civil actions that "where assessment of damages or  awarding of
other relief remains to be resolved," a district  court's judgment is
not " 'final' within the meaning of 28  U.S.C. s 1291." Liberty Mut.
Ins. Co. v. Wetzel, 424 U.S.  737, 744 (1976); see also A & S Council
Oil Co. v. Lader, 56  F.3d 234, 238 (D.C. Cir. 1995) (holding that an
order estab- lishing liability but referring the issue of damages to
arbitra- tion is not final). For it is clear that neither of the
district  court's orders resolved the precise quantum of payments to 


This rule of finality does not apply here, however, because  this is
not an appeal from an ordinary civil judgment ren- dered by the
district court. With both of their claims, the  Hospitals challenged
the Secretary's actions under section  10(e) of the Administrative
Procedure Act, 5 U.S.C. s 706(2).  As we have often observed, "[w]hen
a final agency action is  challenged under the APA in district court,
if the relevant  substantive statute does not provide for direct


court of appeals, the district court does not perform its  normal role"
but instead "sits as an appellate tribunal." PPG  Indus., Inc. v.
United States, 52 F.3d 363, 365 (D.C. Cir.  1995) (quoting Marshall
County Health Care Auth. v. Shala- la, 988 F.2d 1221, 1225 (D.C. Cir.
1993)); accord James  Madison Ltd. v. Ludwig, 82 F.3d 1085, 1096 (D.C.
Cir. 1996)  ("Generally speaking, district courts reviewing agency
action  under the APA's arbitrary and capricious standard do not 
resolve factual issues, but operate instead as appellate courts 
resolving legal questions."), cert. denied, 519 U.S. 1077 (1997). 
Whether it is a court of appeals or a district court, "[u]nder 
settled principles of administrative law, when a court review- ing
agency action determines that an agency made an error of  law, the
court's inquiry is at an end: the case must be  remanded to the agency
for further action consistent with the  corrected legal standards."
PPG Indus., 52 F.3d at 365; see  also South Prairie Constr. Co. v.
Local No. 627, Int'l Union  of Operating Eng'rs, 425 U.S. 800, 806
(1976); SEC v. Chen- ery Corp., 318 U.S. 80, 94-95 (1943). Once,
therefore, the  district court held that the Secretary had
misinterpreted  s 1395ww(d)(5)(A)(iv), it should have remanded to the
Secre- tary for further proceedings consistent with its conception of 
the statute. Not only was it unnecessary for the court to  retain
jurisdiction to devise a specific remedy for the Secre- tary to
follow, but it was error to do so. See Ommaya v.  National Insts. of
Health, 726 F.2d 827, 830 (D.C. Cir. 1984)  ("If MSPB relied on
incorrect legal grounds, it would be error  for this court to enforce
without first remanding for agency  examination of the evidence and
proper fact-finding.") (quot- ing White v. United States Dep't of the
Army, 720 F.2d 209,  210 (D.C. Cir. 1983)). Accordingly, because that
was all that  the district court had the power to do, we construe its 
January 20, 1998 order as a remand to the Secretary, and  ignore, for


Of course, properly characterizing the district court's order  as a
remand does not, without more, resolve our jurisdictional  quandary,
for a "remand order usually is not a final decision."  NAACP v. United
States Sugar Corp., 84 F.3d 1432, 1436 


(D.C. Cir. 1996). We have recognized "an exception to this  general
rule, however, where the agency to which the case is  remanded seeks
to appeal and it would have no opportunity to  appeal after the
proceedings on remand." Occidental Petro- leum Corp. v. SEC, 873 F.2d
325, 330 (D.C. Cir. 1989).  Animating this principle is a pragmatic
concern. Because an  agency must conduct its proceedings and render
its decision  pursuant to the legal standard that the district court
articu- lates in its remand order, "[u]nless another party appeals
[the  agency's subsequent] decision, the correctness of the district 
court's legal ruling will never be reviewed by the court of  appeals,
notwithstanding the agency's conviction that the  ruling is
erroneous." Id. Here, were the Secretary unable  to appeal the
district court's decision at this point, on remand  she would have to
interpret paragraph (5)(A)(iv) as the dis- trict court has construed
it, and disburse millions of dollars in  retroactive outlier payments
to various Medicare-provider  hospitals. Absent an appeal from that
decision by one of the  Hospitals, the Secretary would have no
opportunity to chal- lenge the legal basis for the disbursements. Our
jurisdiction  is therefore proper because the Secretary's appeal falls
within  the exception recognized in Occidental Petroleum. Addition-
ally, vested with jurisdiction to review the Secretary's appeal  under
s 1291, we may also consider the Hospitals' cross- appeal of the
district court's grant of summary judgment to  the Secretary on their
arbitrary and capricious agency-action  claim. See United States Sugar


B. The Secretary's Appeal


Turning first to the Secretary's challenge, we face a ques- tion of
statutory interpretation that the district court resolved  in the
affirmative: whether, under s 1395ww(d)(5)(A)(iv), the  Secretary must
make retroactive reimbursements to ensure  that aggregate outlier
payments during any given fiscal year  constitute at least five
percent of estimated or projected  DRG-related payments. Because we
may set aside agency  action only if it is "arbitrary, capricious, an
abuse of discre- tion, or otherwise not in accordance with law," 5
U.S.C.  s 706(2)(A), we accord no particular deference to the judg-
ment of the district court when conducting our review. See 


HCA Health Servs. v. Shalala, 27 F.3d 614, 616 (D.C. Cir.  1994);
Hennepin County v. Sullivan, 883 F.2d 85, 91 (D.C.  Cir. 1989) ("Our
review proceeds as if this case were an  immediate appeal from a
decision reached after an adminis- trative hearing on the record.");
Biloxi Reg'l Med. Ctr. v.  Bowen, 835 F.2d 345, 348-49 (D.C. Cir.


We initiate statutory analyses of the sort presented here by  first
asking whether "Congress has directly spoken to the  precise question
at issue." Chevron U.S.A. Inc. v. Natural  Resources Defense Council,
Inc., 467 U.S. 837, 842 (1984).  For if after "exhaust[ing] the
traditional tools of statutory  construction," Natural Resources
Defense Council, Inc. v.  Browner, 57 F.3d 1122, 1125 (D.C. Cir. 1995)
(quoting Chev- ron, 467 U.S. at 843 n.9), we ascertain that Congress's
intent  is clear, "that is the end of the matter." Chevron, 467 U.S.
at  842. But "if the statute is silent or ambiguous with respect to 
the specific issue, the question for the court is whether the 
agency's answer is based on a permissible construction of the 
statute." Id. at 843. Where judicial review is refracted  through this
analytic prism, the "view of the agency charged  with administering
the statute is entitled to considerable  deference; and to sustain it,
we need not find that it is the  only permissible construction that
[the agency] might have  adopted." Chemical Mfrs. Ass'n v. Natural
Resources De- fense Council, Inc., 470 U.S. 116, 125 (1985).


Contending that the statutory language boasts an ambigui- ty, the
Secretary maintains that she has reasonably construed  paragraph
(5)(A)(iv) as prescribing a specific methodology  that she must follow
when setting outlier thresholds at the  beginning of each fiscal year.
Under the Secretary's inter- pretation, paragraph (5)(A)(iv) mandates
that she must select  outlier thresholds which, when tested against
historical data,  will likely produce aggregate outlier payments
totaling be- tween five and six percent of projected or estimated DRG-
related payments. Because, under the Secretary's construc- tion,
paragraph (5)(A)(iv) speaks only to how she is to calcu- late outlier
thresholds for the forthcoming year, the Secretary  posits that she
has no obligation to ensure that actual outlier 


payments for the year total five percent of projected DRG- related
payments.


Advocating a results-oriented approach, however, the Hos- pitals argue,
and the district court agreed, that the Secre- tary's interpretation
contradicts the unambiguous meaning of  paragraph (5)(A)(iv). The
textual lodestar guiding their  plain-meaning critique is the single
phrase "payments made."  According to the Hospitals, by indicating in
paragraph  (5)(A)(iv) that "the total amount of the additional
payments  made ... may not be less than 5 percent" of total DRG-
related payments "estimated or projected to be made," 42  U.S.C. s
1395ww(d)(5)(A)(iv) (Supp. IV 1986) (emphasis add- ed), Congress
unmistakably meant that the total amount of  additional payments
actually made during a fiscal year must  meet the five-percent target.
During years in which the  Secretary's chosen thresholds yield outlier
payments that fall  short of the statutory mark, the Hospitals'
interpretation  would require the Secretary to bridge the difference
by  recalibrating outlier variables and making retroactive pay- ments


Standing alone, however, the phrase "payments made"  hardly conveys a
single meaning, much less the one that the  Hospitals advance. As it
is employed in paragraph (5)(A)(iv),  "payments made" is "simply an
adjectival phrase, not a  verbial phrase indicating the past tense,
and hence allows  alternative temporal readings." United States Dep't
of the  Treasury v. FLRA, 960 F.2d 1068, 1072 (D.C. Cir. 1992). It  is
not unlike the phrase "recognized as reasonable," which the  Supreme
Court, quoting favorably from our decision in Ad- ministrators of the
Tulane Educational Fund v. Shalala, 987  F.2d 790 (D.C. Cir. 1993),
held "does not tell us whether  Congress means to refer the Secretary
to action already  taken or to give directions on actions about to be
taken."  Regions Hosp. v. Shalala, 522 U.S. 448, ---, 118 S. Ct. 909, 
916 (1998) (quoting Tulane Educ. Fund, 987 F.2d at 796).  Evincing the
same syntactical equivalence, the phrase "pay- ments made" in
paragraph (5)(A)(iv), though certainly capable  of accommodating the
Hospitals' interpretation, can just as  easily be read to reflect
Congress's intent to "give directions 


on actions about to be taken." Id. In other words, instead of 
embodying a retrospective inquiry into the amount of outlier  payments
that have been made, the phrase "payments made  under this
subparagraph" might just as plausibly reflect a  prospective command
to the Secretary about how to structure  outlier thresholds for
payments to be made in advance of each  fiscal year. Cf. Regions
Hosp., 522 U.S. at ---, 118 S. Ct. at  916 ("[T]he phrase 'recognized
as reasonable' might mean  costs the Secretary (1) has recognized as
reasonable ..., or  (2) will recognize as reasonable....").
Ultimately, whether  the phrase is "recognized as reasonable,"
"adversely affect- ed," or "payments made," it is difficult to divine
with much  confidence the pellucid intent of Congress because "[t]he 
language, in short, is ambiguous." United States Dep't of the 
Treasury, 960 F.2d at 1072 (describing as ambiguous the  phrase
"adversely affected"); accord Tulane Educ. Fund, 987  F.2d at 796.


Hoping to stave off judicial review under Chevron's defer- ential
second step, the Hospitals attempt to resuscitate their  plain-meaning
interpretation by contrasting the two ways in  which Congress modified
the word "made" in paragraph  (5)(A)(iv). When it first appears,
"made" is used without  modifiers to describe the "total amount of the
additional  payments made under this subparagraph"; later, the word 
materializes to indicate that the total amount of outlier pay- ments
just described may not be less than five percent "of the  total
payments projected or estimated to be made" for DRG- related payments.
42 U.S.C. s 1395ww(d)(5)(A)(iv). Be- cause, the Hospitals reason,
Congress employed words of  estimation and projection to modify the
total amount of DRG- related payments "to be made" but neglected to do
so when  describing the total amount of outlier payments "made," it 
must have intended that total outlier payments actually made  during a
fiscal year would equal at least five percent of  estimated or
projected DRG-related payments.2




__________

n 2 The Hospitals cite only to this court's decision in Washington 
Hospital Center v. Bowen, 795 F.2d 139 (D.C. Cir. 1986), to buttress 
their argument that the plain meaning of "made" can be inferred 


Whatever logic this internal construction of paragraph  (5)(A)(iv)
enjoys, to prevent statutory interpretation from  degenerating into an
exercise in solipsism, "we must not be  guided by a single sentence or
member of a sentence, but  look to the provisions of the whole law."
United States Nat'l  Bank v. Independent Ins. Agents of Am., Inc., 508
U.S. 439,  455 (1993) (quoting United States v. Heirs of Boisdore, 49 
U.S. (8 How.) 113, 122 (1849)). Under Chevron step one, "we  consider
not only the language of the particular statutory  provision under
scrutiny, but also the structure and context of  the statutory scheme
of which it is a part." Illinois Pub.  Tele. Ass'n v. FCC, 117 F.3d
555, 568 (D.C. Cir. ), modified,  123 F.3d 693 (1997), cert. denied,
--- U.S. ----, 118 S. Ct.  1361 (1998); accord Conroy v. Aniskoff, 507
U.S. 511, 515  (1993) ("[T]he meaning of statutory language, plain or
not,  depends on context."); Davis v. Michigan Dep't of Treasury,  489
U.S. 803, 809 (1989) ("[W]ords of a statute must be read  in their
context and with a view to their place in the overall  statutory
scheme."). By examining paragraph (5)(A)(iv) in its  immediate
statutory context, any putative clarity that that 




__________

n from the different language that Congress used to modify that word 
in paragraph (5)(A)(iv). That decision misses the mark, however.  In
Washington Hospital Center, we presumed that when Congress  amended a
pre-existing section of the Medicare statute by adding  and deleting
certain words, it must have intended the amended  provision to have a
different meaning from its predecessor provi- sion. Id. at 146. More
on point for the Hospitals, though they did  not cite it, would be the
canon of construction that posits that  "where Congress includes
particular language in one section of a  statute but omits it in
another section of the same Act, it is  generally presumed that
Congress acts intentionally and purposely  in the disparate inclusion
or exclusion." Russello v. United States,  464 U.S. 16, 23 (1983);
accord Independent Bankers Ass'n of Am. v.  Farm Credit Admin., 164
F.3d 661, 667 (D.C. Cir. 1999). Of  course, "[c]anons of construction
are, after all, only aids in the  process of statutory construction,
nothing more, nothing less."  Eagle-Picher Indus. v. United States
EPA, 759 F.2d 922, 927 n.6  (D.C. Cir. 1985). As we demonstrate,
infra, this canon does not  resolve the ambiguity in paragraph


provision might arguably have quickly recedes to ambiguity  once
again.


Preceding paragraph (5)(A)(iv) by two paragraphs,  s 1395ww(d)(3)(B)
provided during the time relevant to this  litigation:


The Secretary shall reduce each of the average standard- ized amounts
under subparagraph (A) ... by a propor- tion equal to the proportion
(estimated by the Secretary)  of the amount of payments under this
subsection based  on DRG prospective payment amounts which are addi-
tional payments described in paragraph (5)(A) (relating  to outlier
payments)....


42 U.S.C. s 1395ww(d)(3)(B) (Supp. IV 1986) (emphasis add- ed). Not
only does this provision expressly indicate that total  outlier
payments are to be estimated by the Secretary, but it  also employs
language that closely parallels the language that  later appears in
paragraph (5)(A)(iv). In our endeavor to  determine whether the "total
amount of the additional pay- ments made under this subparagraph"
contemplates outlier  payments actually made or those estimated to be
made, we  find it significant that in paragraph (3)(B) Congress
provided  that the "amount of payments ... which are additional pay-
ments described in paragraph (5)(A)" are to be "estimated by  the
Secretary." s 1395ww(d)(3)(B). Given that in paragraph  (3)(B) it had
already indicated that the Secretary would  estimate the amount of
outlier payments described in subpar- agraph (5)(A), Congress could
have reasonably concluded that  there was no need to provide expressly
in paragraph  (5)(A)(iv) that the phrase "payments made" referred to
pay- ments estimated to be made. Thus, whatever can be said for 
Congress's disparate modification of the word "made" in  paragraph
(5)(A)(iv), when we open the analytic aperture to  examine that clause
in its proper statutory context, the  inherently ambiguous phrase


Nor does a passing observation in the Conference Report  that "the
Secretary would be required to provide additional  payments for
outlier cases amounting to not less than 5  percent, and not more than
6 percent, of total projected or  estimated DRG related payments,"
compel us to adopt the 


Hospitals' construction of the statute under Chevron step one.  H.R.
Conf. Rep. No. 98-47, at 189 (1983), reprinted in 1983  U.S.C.C.A.N.
404, 479 (emphasis added). Ambiguous in its  own right, this passage,
if given the gloss that the Hospitals  advance, would chafe against
the commentary in the Senate  Report. Suggesting that paragraph
(5)(A)(iv) reflects the  prospective inquiry that the Secretary
advocates, the Senate  Report provides that the "[t]otal expected
payments resulting  from this policy will be not less than 5 percent,
nor more than  6 percent, of total medicare payments to hospitals." S.
Rep.  No. 98-23, at 51, reprinted in 1983 U.S.C.C.A.N. 143, 191 
(emphasis added). The only conclusion that we can safely  draw from
these seemingly contradictory passages is that  "the little
legislative history that exists for [paragraph  (5)(A)(iv)] is as
ambiguous as the statute itself." Deaf Smith  County Grain Processors,
Inc. v. Glickman, 162 F.3d 1206,  1212 (D.C. Cir. 1998).


Ultimately, neither the text, structure, nor legislative histo- ry of
paragraph (5)(A)(iv) illuminates Congress's unambigu- ous intent.
Although the Hospitals' interpretation, and the  one that the district
court adopted, is plausible, it is not the  "only possible
interpretation," Sullivan v. Everhart, 494 U.S.  83, 89 (1990), and it
certainly is not "an inevitable one."  Regions Hosp., 522 U.S. at
----, 118 S. Ct. at 917. Because  we find the statute ambiguous, we
proceed to assess whether  the Secretary's interpretation of paragraph
(5)(A)(iv) is "rea- sonable and consistent with the statutory scheme
and legisla- tive history." Cleveland v. United States Nuclear
Regulatory  Comm'n, 68 F.3d 1361, 1367 (D.C. Cir. 1995).


In marking off the metes and bounds of our review under  the second
step of Chevron, we accord particular deference to  the Secretary's
interpretation of paragraph (5)(A)(iv) "given  the tremendous
complexity of the Medicare statute." Appa- lachian Reg'l Healthcare,
Inc. v. Shalala, 131 F.3d 1050, 1054  (D.C. Cir. 1997); accord
Methodist Hosp. v. Shalala, 38 F.3d  1225, 1229 (D.C. Cir. 1994). The
Hospitals, however, urge us  not to defer in any way to the
Secretary's interpretation of  paragraph (5)(A)(iv) because, they
contend, that provision  does not delegate interpretative authority to


but explicitly limits her discretion. The problem with this  argument,
of course, is that it assumes the truth of the  proposition that we
just rejected. Were paragraph (5)(A)(iv)  truly "an explicit
limitation on the Secretary's discretion," Br.  of Hosps. at 40, there
would be no need to analyze the  provision under Chevron step two.
While paragraph  (5)(A)(iv) is certainly designed to regulate the
Secretary's  discretion to some extent, as we have already concluded,
the  precise contours of that provision are hardly explicit but are 
instead ambiguous. Nor is it problematic, as the Hospitals  suggest,
that Congress did not expressly delegate interpreta- tive authority to
the Secretary in paragraph (5)(A)(iv). Def- erence to agency
interpretation is warranted "when Congress  has left a gap for the
agency to fill pursuant to an express or  implied delegation of
authority to the agency." Chevron, 467  U.S. at 843-44 (internal
quotations omitted). Where, as here,  Congress enacts an ambiguous
provision within a statute  entrusted to the agency's expertise, it
has "implicitly delegat- ed to the agency the power to fill those
gaps." National Fuel  Gas Supply Corp. v. FERC, 811 F.2d 1563, 1569
(D.C. Cir.  1987); see also Chevron, 467 U.S. at 843-44.


Equally untenable is the Hospitals' argument that the  Secretary's
interpretation is not entitled to deference because  she did not adopt
it through either formal rulemaking or  adjudication. But as our
precedents make clear, "an agency  need not promulgate a legislative
rule setting forth its inter- pretation of a statutory term for that
term to be entitled to  deference." Association of Bituminous
Contractors, Inc. v.  Apfel, 156 F.3d 1246, 1251-52 (D.C. Cir. 1998).
In fact,  "[e]ven if the legal briefs contained the first expression
of the  agency's views, under the appropriate circumstances we  would
still accord them deference so long as they represented  the agency's
'fair and considered judgment on the matter.' "  United Seniors Ass'n
v. Shalala, 182 F.3d 965, 971 (D.C. Cir.  1999) (quoting Auer v.
Robbins, 519 U.S. 452, ----, 117 S. Ct.  905, 912 (1997)); see
National Wildlife Fed'n v. Browner, 127  F.3d 1126, 1129 (D.C. Cir.
1997) ("The mere fact that an  agency offers its interpretation in the


not automatically preclude deference to the agency."); Tax  Analysts v.
IRS, 117 F.3d 607, 613 (D.C. Cir. 1997).


There is no reason to suspect that the Secretary's interpre- tation of
paragraph (5)(A)(iv) embodies anything other than  her fair and
considered opinion. In a final rule published on  January 3, 1984, the
Secretary articulated the same interpre- tation of paragraph
(5)(A)(iv) that she has pressed before  both us and the district
court. See 49 Fed. Reg. 234, 265  (1984). With that rule, she not only
observed that under her  interpretation "there is no necessary
connection between the  amount of estimated outlier payments and the
actual pay- ments made to hospitals for cases that actually meet the 
outlier criteria," id., but she also admonished Medicare pro- viders
that, in the event she overestimated the amount of  outlier payments,
she would "not adjust the DRG rates to  compensate hospitals for funds
that were not actually paid for  outlier cases." Id. at 266. Even if,
as the Hospitals com- plain, the final rule failed to provide a
"cogent explanation" of  the policies undergirding the Secretary's
interpretation, the  fact remains that for the past fifteen years, the
Secretary has  never wavered from that interpretation. We are
confident  that the interpretation of paragraph (5)(A)(iv) under
review  embodies the Secretary's "fair and considered judgment on  the
matter." Auer, 519 U.S. at ----, 117 S. Ct. at 912. It,  accordingly,


Having settled on the scope of our review, we have no  difficulty
concluding that the Secretary has advanced a rea- sonable
interpretation of paragraph (5)(A)(iv). Congress es- tablished outlier
payments because it recognized that "there  will be cases within each
[DRG] that will be extraordinarily  costly to treat ... because of
severity of illness or complicat- ing conditions, and [will] not [be]
adequately compensated for  under the DRG payment methodology." S.
Rep. No. 98-23, at  51 (1983), reprinted in 1983 U.S.C.C.A.N. 143,
191. But as  the term "outlier" suggests, these payments were not
intend- ed to subsidize hospitals simply for treatments, which in 
absolute terms, were extraordinarily costly or lengthy. Rath- er,
Congress directed the Secretary to provide "additional  payments for
cases which are extraordinarily costly to treat, 


relative to other cases within the DRG." Id. (emphasis  added); accord
H.R. Rep. No. 98-25, at 134-35 (1983), reprint- ed in 1983
U.S.C.C.A.N. 219, 353-54 ("Your Committee is  concerned that under the
prospective payment system, there  will be cases within each [DRG]
that will be extraordinarily  costly to treat, relative to other cases
within that DRG....").  Thus the House version would have required the
Secretary to  tender outlier payments only when a patient's length of
stay  exceeded by thirty days the average length of stay for cases  in
that same DRG, see H.R. Rep. No. 98-25, at 135, reprinted  in 1983
U.S.C.C.A.N. at 354, while the Senate bill, which the  Conference
Committee adopted, authorized outlier payments  for cases in which a
patient's length of stay eclipsed the mean  length of stay for
discharges within that same DRG by a fixed  number of days or standard
deviations. See 42 U.S.C.  s 1395ww(d)(5)(A)(i) (Supp. IV 1986); S.
Rep. No. 98-23, at  51, reprinted in 1983 U.S.C.C.A.N. at 191; H.R.
Conf. Rep.  No. 98-47, at 188 (1983), reprinted in 1983 U.S.C.C.A.N.


The Secretary's interpretation of paragraph (5)(A)(iv)  evinces far
greater fidelity to Congress's conception of outlier  payments than
does the view espoused by the Hospitals.  Under the Secretary's
reading of the statute, if it turns out  that actual outlier payments
do not meet the five-percent  target at the end of the fiscal year, it
is because the lengths of  stay for DRGs in that year proved to be
shorter than the  historical averages reflected in the data on which
the Secre- tary based her threshold calculations. Whether it is
because  hospitals became more efficient or a miracle drug was intro-
duced during the year, the shorter lengths of stay mean that  there
were fewer extraordinarily costly cases during the year.  In other
words, there were fewer outliers--and therefore,  fewer outlier


Under the Hospitals' interpretation, however, regardless of  actual
costs or inpatient lengths of stay during a fiscal year,  Medicare
providers are guaranteed a substantial and fixed  sum of outlier
payments. As they read the statute, even  during a fiscal year in
which the length of stay for every  inpatient discharge in every DRG
in every hospital equaled or 


exceeded by just a day the mean length of stay for each  respective
DRG, the Secretary would nonetheless have to  reward hospitals with
additional "outlier" payments totaling  five percent of the entire DRG
budget. One need not be well  versed in the discipline of statistics
to recognize that such  insignificant deviations from the mean do not
constitute outli- ers. To sanction the Hospitals' interpretation of
paragraph  (5)(A)(iv) would not only require us to assume that
Congress  did not appreciate the meaning of outlier--a term, it should
 be noted, that appears throughout both the legislative history  and
the text of the Medicare statute--but it would also  transform the
character of the outlier-payment regime from a  system intended to
insulate hospitals from aberrational and  extraordinary costs into
nothing more than an entitlement  program for Medicare providers. Such
was hardly Con- gress's intent, for if anything, Congress indicated
that it was  "equally concerned that adjustments may be required for 
cases which have an unusually short length of stay or which  are
significantly less costly than the DRG payment." H.R.  Conf. Rep. No.
98-47, at 478, reprinted in 1983 U.S.C.C.A.N.  at 478 (emphasis
added); see also H.R. Rep. No. 98-25, at 135,  reprinted in 1983
U.S.C.C.A.N. at 354 ("The Secretary would  be required to study ...
the appropriateness of, and necessi- ty for, adjustments in payment
rates for extremely short  lengths of stay within a DRG....").
Proposals like these  reflect a reluctance to reimburse Medicare
providers at rates  grossly disproportionate to the cost of treatment.
We find it  unlikely that Congress nevertheless would have wanted
hospi- tals to reap additional compensation over and above the 
standard DRG payment where treatment costs for a particu- lar
discharge were not extraordinarily costly relative to the  mean costs


Moreover, compared to the Hospitals' interpretation of  paragraph
(5)(A)(iv), the Secretary's reading better harmon- izes each of the
four clauses in paragraph (5)(A). As did the  district court, the
Hospitals struggle to reconcile their con- ception of the fourth
clause with the language of the third,  which provides that the amount
of each outlier payment "shall  approximate the marginal cost of care
beyond the cutoff point 


applicable" for each DRG. 42 U.S.C. s 1395ww(d)(5)(A)(iii)  (Supp. IV
1986). Under the Hospitals' construction of the  statute, outlier
payments might bear little relationship to the  marginal cost of care.
At the end of each fiscal year, if actual  outlier payments fell short
of the five-percent target, the  Secretary would be required
retroactively to supplement  those payments to satisfy the difference.
Depending on how  great that initial disparity was, by the time that
these cura- tive outlays were made, the newly computed outlier
payments  might not approximate anything close to the marginal cost of
 care as paragraph (5)(A)(iii) mandates. By contrast, outlier 
payments under the Secretary's interpretation will always  approximate
the marginal cost of care because when deter- mining where to set
outlier thresholds for DRGs at the  beginning of each fiscal year, she
directly factors the margin- al cost of care into her calculus.


Echoing the district court's holding, the Hospitals discount  paragraph
(5)(A)(iii) as merely a "guideline" while contending  that paragraph
(5)(A)(iv) operates "as a limitation on the  Secretary's discretion."
County of Los Angeles, 992 F. Supp.  at 32. Based on this view, the
Secretary is supposed to set  outlier thresholds at the beginning of
each year "at marginal  cost and then, when actual outlier data is
known, adjust[ ] the  final payments to ensure that the Secretary has
met her  statutory obligation to the providers." Id. at 31. Why this 
parsing of the statutory language is more reasonable than  that of the
Secretary's--much less compelled as an unambigu- ously plain reading
of the provision, as the Hospitals have  urged--is not at all clear.
After all, paragraph (5)(A)(iii)  employs mandatory language of the
sort not normally used if  all that Congress intended to do was to
offer a discretionary  guideline for the Secretary to follow. See  s
1395ww(d)(5)(A)(iii) ("The amount of such additional pay- ment under
clauses (i) and (ii) shall be determined by the  Secretary and shall
approximate the marginal cost of  care....") (emphasis added).
Nevertheless, even were the  Hospitals' synthesis of the third clause
into the remainder of  paragraph (5)(A) plausible, it would not be
enough to impugn  the otherwise reasonable interpretation that the


has advanced since "we need not find that it is the only  permissible
construction that [the Secretary] might have  adopted but only that
[her] understanding of this very 'com- plex statute' is a sufficiently
rational one to preclude a court  from substituting its judgment for
that of [the Secretary]."  Young v. Community Nutrition Inst., 476
U.S. 974, 981  (1986) (internal quotation omitted).


Moreover, the Secretary's interpretation avoids the sub- stantial
administrative burden attendant with the Hospitals'  vision of
paragraph (5)(A)(iv). It strains credulity to assume  that Congress
would have directed the Secretary to establish  outlier thresholds in
advance of each fiscal year, see  s 1395ww(d)(3)(B), (d)(6), and
process millions of bills based  on those figures, only to have her at
the end of the year  recalibrate those calculations, reevaluate anew
each of the  millions of inpatient discharges under the revised
figures, and  disburse a second round of payments. As we have held in
an  analogous context, "[u]nder these circumstances, retroactive 
corrections would cause a significant, if not debilitating, dis-
ruption to the Secretary's administration of the already- complex
Medicare program." Methodist Hosp., 38 F.3d at  1233. Nor is this
administrative process rendered less awk- ward and unwieldy if, as the
Hospitals suggest, the Secretary  actively monitors outlier payments
and adjusts the thresholds  as the fiscal year unfolds. Apart from the
tremendous re- sources that would be required to maintain such a
vigilant  watch over a program as expansive as Medicare, intermittent-
ly modifying outlier thresholds at various times during the  year
would mean that different hospitals would likely receive  different
outlier reimbursements for the same treatments  based on nothing more


By the same token, this uncertainty and fluidity in outlier- payment
amounts under the Hospitals' interpretation lead us  to the final
virtue of the Secretary's construction. One of the  touchstones of the
Prospective Payment System, as its name  suggests, is prospectively
determined reimbursement rates  that remain constant during the fiscal
year. In setting, prior  to each fiscal year, fixed outlier thresholds
and per-diem 


reimbursement rates that are not later subject to retroactive 
correction, the Secretary promotes certainty and predictabili- ty of
payment for not only hospitals but the federal govern- ment--concerns
that played a prominent role in Congress's  decision to adopt PPS. See
H.R. Rep. No. 98-25, at 132,  reprinted in 1983 U.S.C.C.A.N. at 351
("The bill is intended  to improve the medicare program's ability to
act as a prudent  purchaser of services, and to provide predictibility
[sic] re- garding payment amounts for both the Government and 
hospitals."). To be sure, we have previously speculated that  "the
real linchpin of the [PPS] system may not be that the  exact
reimbursement figure is known in advance, but rather  may be that the
hospital knows that nothing it does in  providing services will lead
to a higher reimbursement level."  Georgetown Univ. Hosp. v. Bowen,
862 F.2d 323, 330 (D.C.  Cir. 1988) (Georgetown II). Yet while we,
therefore, have  recognized that retroactive corrections may not
ultimately  undermine PPS, we have emphasized that that "does not 
establish that a prospective-only policy is unreasonable."  Methodist


In Methodist Hospital, we found the Secretary's decision  not to
recalculate retroactively the DRG wage index to be  reasonable, in
part, because the Secretary's prospective policy  advanced the
principles of PPS.3 With language applicable to  the present case, we




__________

n 3 The Hospitals cannot successfully distinguish Methodist Hospi- tal.
Admittedly, unlike the DRG wage index at issue in Methodist  Hospital,
outlier payments do not factor directly into every inpa- tient
discharge. But outlier payments do influence indirectly the  overall
DRG payment rate that governs all discharges. As already  discussed,
pursuant to s 1395ww(d)(3)(B), the Secretary must re- duce the
standard DRG payment rate by a factor equal to the  outlier payments
that she predicts she will have to disburse during  the forthcoming
year. Nor is it accurate to claim, as the Hospitals  do, that outlier
payments are entirely divorced from PPS. As an  initial matter, the
provisions relating to outliers are contained in the  same subsection
of s 1395ww as those establishing the PPS regime.  See s 1395ww(d).
Moreover, Congress established outlier pay- ments not as a distinct
reimbursement methodology but as a 


While retroactive adjustments might leave the "linchpin"  of PPS
intact, that does not mean that a prospective-only  policy would not
further, to some degree, the overall  goals of PPS.... [H]ospitals,
like other businesses, do  make projections about future costs and
service levels  based on their experience and historical patterns. To 
the extent that the Secretary's prospectivity policy per- mits
hospitals to rely with certainty on one additional  element in the PPS
calculation rate ... the Secretary  could reasonably conclude that it
will promote efficient  and realistic cost-saving targets.


Id. The same, quite reasonably, can be said of the Secre- tary's
interpretation of paragraph (5)(A)(iv). Under her con- struction of
the statute, at the outset of each fiscal year,  hospitals know the
point beyond which a patient's length of  stay will trigger outlier
payments and the corresponding rate  at which they will be reimbursed
for each day beyond the  threshold. A less determinate policy would
not only deprive  hospitals of the ability to make accurate
projections about  outlier payments for the forthcoming year but also
threaten  them at the end of each year with the prospect of actually 
having to forfeit a portion of those payments to the Secretary;  for
as the Hospitals concede, under their interpretation,  Medicare
providers collectively would be bound to repay any  portion of outlier
payments that exceeded six percent of  estimated DRG-related payments.
See Br. of Hosps. at 31- 32. We conclude that the Secretary has
advanced a reason- able interpretation of an ambiguous statutory
provision, and,  therefore, reverse the judgment of the district court
with  respect to the Secretary's appeal.


C. The Hospitals' Cross-Appeal


Because outlier thresholds are measured by their distance  from the
mean length of stay, accurately forecasting outlier 




__________

n carefully crafted supplement to PPS. For that reason, Georgetown  II,
which concerned retroactive adjustments under the pre-PPS  "reasonable
cost" system--clearly a payment methodology lacking  any relationship
to PPS--is inapposite.


payments depends, in large part, on how closely the mean  length of
stay reflected in the Secretary's historical data  reflects the actual
average length of stay for that particular  DRG. When setting outlier
thresholds for fiscal years 1985- 1986, the Secretary relied on the
1981 MEDPAR file, a  database containing patient-specific data for a
random sample  of 20 percent of all Medicare-hospital discharges
during 1981.  Compiled during an era when Medicare still reimbursed 
hospitals under the reasonable-cost system, the 1981 MED- PAR file
could not have predicted how, under PPS, the  average length of stay
for virtually all DRGs would eventually  decline dramatically. The
Hospitals observe, however, that  by July 27, 1984, the Secretary had
already collected data  from 2.5 million discharges under PPS that
indicated that the  average length of stay for all DRGs had declined
from 9.5  days to 7.5 days under the new payment methodology. Pursu-
ant to section 10(e) of the APA, 5 U.S.C. s 706(2)(A), the  Hospitals
claim that the Secretary acted arbitrarily and capri- ciously when she
calculated outlier thresholds for 1985-1986  based on the 1981 MEDPAR
file instead of the preliminary  1984 data and failed to explain
adequately her decision.  Rejecting the Hospitals' claim, the district
court agreed with  the Secretary that her decision to use the 1981
MEDPAR file  over the more contemporaneous but preliminary 1984 data 
"was a rational choice between two imperfect databases."  County of
Los Angeles, 992 F. Supp. at 36.


Under the APA, we may set aside agency action found to  be "arbitrary,
capricious, an abuse of discretion, or otherwise  not in accordance
with law." 5 U.S.C. s 706(2)(A). Foreclos- ed from substituting our
judgment for that of the agency, we  do not set aside agency action
lightly. See Motor Vehicles  Mfrs. Ass'n v. State Farm Mut. Auto. Ins.
Co., 463 U.S. 29,  43 (1983); Petroleum Communications, Inc. v. FCC,
22 F.3d  1164, 1172 (D.C. Cir. 1994). Nevertheless, we intervene to 
ensure that the agency has "examine[d] the relevant data and 
articulate[d] a satisfactory explanation for its action." State  Farm,
463 U.S. at 43. "Where the agency has failed to  provide a reasoned
explanation, or where the record belies  the agency's conclusion, we
must undo its action." BellSouth 


Corp. v. FCC, 162 F.3d 1215, 1222 (D.C. Cir. 1999) (citation  and
quotation omitted).


The only contemporaneous explanation that the Secretary  offered for
using the 1981 MEDPAR file consisted of two  sentences in the Federal
Register: "Based 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. Therefore, as discussed
above, we are  continuing to set the outlier thresholds on the basis
of the  1981 MEDPAR data." 49 Fed. Reg. 34,728, 34,769 (1984) 
(emphasis added). We agree with the Ninth Circuit, which  recently
considered this same issue, that the Secretary's  "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." Alvara- do Community Hosp. v. Shalala, 155
F.3d 1115, 1122 (9th  Cir. 1998). Data that the Secretary possessed as
late as July  27, 1984, indicated that the average length of stay for
prac- tically all DRGs had declined considerably under the nascent 
PPS program. More concretely, at that point during the  fiscal year,
outliers constituted only 1.9 percent of total PPS  discharges instead
of 5.0 percent as predicted. And while  these conclusions were drawn
from preliminary data, that  data reflected 2.5 million patient
discharges under PPS; the  1981 MEDPAR file, by contrast, contained
1.6 million dis- charge records. Failure to account for this trend is
all the  more perplexing insofar as the Secretary herself had antici-
pated that the average length of stay for DRGs would decline  under
PPS. In 1984 she observed that "[t]he most commonly  accepted
expectation about the PPS at the time of its incep- tion was that it
would result in shorter stays for Medicare  patients.... [R]educed
length of stay was to be one of the  major vehicles through which
hospital costs were to be con- trolled under the PPS." Office of
Research & Demonstra- tions, Health Care Fin. Admin., U.S. Dep't of
Health &  Human Servs., Pub. No. 03231, Report to Congress: Impact of 
the Medicare Hospital Prospective Payment System 6-13  (1984). At
bottom, for the Secretary to say that she had "no  evidence to suggest


the levels intended," 49 Fed. Reg. at 34,769, runs "counter to  the
evidence before the agency" and "is so implausible that it  could not
be ascribed to a difference in view or the product of  agency
expertise." State Farm, 463 U.S. at 43.


In her brief, the Secretary now contends that what she  meant by "no
evidence" was "no reliable evidence." To  bolster this specific claim
and her broader argument that the  1984 data were too suspect and
incomplete to make accurate  outlier projections, the Secretary
appended to her summary  judgment motion in the district court an
affidavit from Rose  Connerton, an official with the Health Care
Financing Admin- istration ("HCFA") who helped develop the outlier
thresholds  for 1985-1986. Essentially, the Connerton affidavit claims
 that the 1984 data were not complete and did not represent a  random
sample of cases, that because they were based on a  partial year they
would not reflect seasonal and regional  variances, and that any
analysis drawn from them would be  skewed.4 See J.A. 90 (Aff. of
Connerton pp 10, 12, 15). The  Hospitals contend that the Connerton
affidavit, having sur- faced for the first time during litigation, is
an impermissible  post-hoc rationalization that the district court
should have  stricken. See SEC v. Chenery Corp., 318 U.S. 80, 87-88 




__________

n 4 Through the Connerton affidavit, the Secretary attempts to 
dramatize the unreliability of the partial 1984 data. As of April 27, 
1984, reported outliers constituted only 1.9 percent of total PPS 
discharges. See J.A. 71. By the end of fiscal year 1984, however, 
actual outlier payments ended up totaling 5.3 percent of total PPS 
payments, suggesting, Connerton avers, that the preliminary data  were
in fact unreliable. Although Connerton's calculations are  accurate,
the conclusions that she draws from them are subject to  debate.
During a portion of fiscal year 1984, the Secretary errone- ously
provided hospitals with additional outlier payments for non-
PPS-covered treatments, but never sought to recoup these surplus 
amounts. That outlier payments amounted to 5.3 percent that year  thus
may say less about the reliability of the 1984 data and more  about
the scope of the Secretary's clerical error. Whatever the  reason,
this dispute underscores the wisdom of Benjamin Disraeli's  sardonic
quip (attributed to him by Mark Twain) about the three  great


(1943); Reeve Aleutian Airways, Inc. v. United States, 889  F.2d 1139,
1144 (D.C. Cir. 1989). Indeed, faced with a similar  affidavit from
Connerton, the Ninth Circuit held that the  district court erred in
considering it. See Alvarado Commu- nity Hosp., 155 F.3d at 1124.
Ultimately, we need not reach  this question, for even were we
inclined to accept everything  in the Connerton affidavit, we would
still remand to the  Secretary for a more adequate justification for
her database  selection.


"A long line of precedent has established that an agency  action is
arbitrary when the agency offer[s] insufficient rea- sons for treating
similar situations differently." Transactive  Corp. v. United States,
91 F.3d 232, 237 (D.C. Cir. 1996); see  also State Farm, 463 U.S. 29,
57 (1983); Airmark Corp. v.  FAA, 758 F.2d 685, 691-92 (D.C. Cir.
1985); Local 777,  Democratic Union Org. Comm. v. NLRB, 603 F.2d 862,
872  (D.C. Cir. 1978). Although maligning the 1984 data as too 
unreliable to calculate outlier thresholds for fiscal years 1985-
1986, the Secretary nonetheless used those same data on  August 31,
1984, to reduce across-the-board all 470 DRG  weighting factors by
1.05 percent. See 49 Fed. Reg. 34,728,  34,770-71 (1984). Such an
adjustment was necessary, the  Secretary noted at the time, because
"[t]he emerging experi- ence under the prospective payment system"--an
experience  gleaned from the preliminary 1984 data--revealed that the 
different incentives that hospitals faced under PPS were  producing
unexpected distortions. See id. In making this  correction, the
Secretary expressly endorsed the reliability of  the 1984 data: "To
date, we have now analyzed 2.5 million  discharges under the
prospective payment system, which fully  reflect the effect of those
incentives, and we believe this  affords us a better measure of the
effect of coding improve- ments in the average case mix." Id. at
34,771. Moreover, in  responding to a question about the legitimacy of
the prelimi- nary data during a 1984 congressional oversight hearing,
the  HCFA Administrator responded that "[o]ur sample now is  based on
approximately 50 percent of all of the claims or the  admissions that
we had projected for this year. We think  that's a fairly


care's Prospective Payment System: Hearing Before the  Subcomm. on
Health of the Comm. on Fin., 98th Cong. 62  (statement of Mr. Davis,
Administrator, HCFA). In sum, the  Secretary has inadequately
explained why the 1984 data were  suitable for one significant
calculation but unreliable for  another. Her sole justification is
that preliminary data may  be used to make across-the-board
adjustments, as was done  to reduce all DRG weighting factors by 1.05
percent, but that  they may not be used for setting outlier thresholds
because a  unique standard deviation must be calculated for each of
the  470 DRGs. What renders this explanation inadequate is that  DRG
weighting factors, like outlier thresholds, are ordinarily  determined
on a DRG-by-DRG basis. Indeed, the very pur- pose of a DRG weighting
factor is to reflect the different  costs of treating minor and major
illnesses; to do so, each  DRG must be assigned its own unique weight
based on the  cost and complexity of treatment peculiar to that DRG.
The  Secretary's proffered distinction is thus not reasonable. She 
may in her discretion, of course, rely on preliminary data to  make an
across-the-board adjustment to variables that ordi- narily are
determined on a case-by-case basis. But when she  does so, she must be
prepared to explain why she cannot also  use that data to make a
similar adjustment to variables that  are also typically calculated on
an individual basis. As broad  as her discretion is, it "is not a
license to ... treat like cases  differently." Airmark Corp., 758 F.2d
at 691; accord Teva  Pharms., USA, Inc. v. FDA, 182 F.3d 1003, 1010-11
(D.C.  Cir. 1999); Transactive Corp., 91 F.3d at 237; Local 777, 603 


This case must therefore be remanded to the Secretary to  allow her
either to recalculate outlier thresholds for fiscal  years 1985-1986
or to offer a reasonable explanation for  refusing to use the 1984
data in setting outlier thresholds  during those years. In reaching
this conclusion, we necessar- ily part ways with the Ninth Circuit,
which, in Alvarado  Community Hospital, chose not to remand to the
Secretary,  but instead ordered her to adjust outlier thresholds for
fiscal  year 1985 based on final 1984 data. Alvarado Community  Hosp.,
155 F.3d at 1125. As the Supreme Court has instruct-


ed, however, where "the record before the agency does not  support the
agency action, ... the proper course, except in  rare circumstances,
is to remand to the agency for additional  investigation or
explanation." Florida Power & Light Co. v.  Lorion, 470 U.S. 729, 744
(1985); see also Dunlop v. Bachow- ski, 421 U.S. 560, 574-75 (1975)
("Where the statement inade- quately discloses his reasons, the
Secretary may be afforded  opportunity to supplement his statement."),
overruled on  other grounds by Furniture & Piano Moving v. Crowley,
467  U.S. 526, 549-50 n.22 (1984). We find no reason to depart  from
that course here. While we have identified significant 
inconsistencies and gaps in the Secretary's rationale for using  the
1981 MEDPAR file, bedrock principles of administrative  law preclude
us from declaring definitively that her decision  was arbitrary and
capricious without first affording her an  opportunity to articulate,
if possible, a better explanation.  See Bechtel v. FCC, 10 F.3d 875,
887 (D.C. Cir. 1993);  Philadelphia Gas Works v. FERC, 989 F.2d 1246,
1251 (D.C.  Cir. 1993); Sullivan Indus. v. NLRB, 957 F.2d 890, 905
n.12  (D.C. Cir. 1992); Tex Tin Corp. v. EPA, 935 F.2d 1321, 1324 
(D.C. Cir. 1991); see also Checkosky v. SEC, 23 F.3d 452, 463  (D.C.
Cir. 1994) (Silberman, J., concurring) (citing some of the  "many
instances where we have remanded to an agency for a  better
explanation before finally deciding that the agency's  action was
arbitrary and capricious"). Because we fail to  perceive any "rare
circumstances" that would warrant a  break with established
administrative practice, we adhere to  the proper course of remanding


III. Conclusion


For the foregoing reasons, we reverse the judgment of the  district
court with respect to the Secretary's appeal, and  remand with
instructions to enter judgment in the Secretary's  favor. As for the
Hospitals' cross-appeal, we reverse the  judgment of the district
court, and instruct it to remand the  case to the Secretary for
further proceedings consistent with  this opinion.


So ordered.