Formula v. Margaret M. Heckler, Secretary of Health and Human Services

Opinion for the court filed by Circuit Judge STARR. STARR, Circuit Judge: This case presents a challenge to portions of the Food and Drug Administration’s regulations pertaining to infant formula, 1 published at 5 C.F.R. §§ 106.1-106.-120 (1984) and promulgated pursuant to the Infant Formula Act of 1980, Pub.L. No. 96-359, 94 Stat. 1190 (codified principally at 21 U.S.C. § 350a (1982)). The regulations at issue address (1) quality control procedures by manufacturers of infant formula for assuring the nutrient content of such products, (2) quality factor requirements aimed at ensuring nutritional quality during the formula’s shelf life, and (3) record-keeping requirements. Appellants challenge these regulations as violative of both the Infant Formula Act and the Administrative Procedure Act. The complainants consist of FORMULA, a non-profit organization; Public Citizen Health Research Group; Consumer Federation of America; and two couples representing themselves and their children (referred to collectively hereinafter as “FORMULA”). FORMULA brought suit in United States District Court for the District of Columbia, seeking a declaratory judgment that the Food and Drug Administration’s final rule “violates the Infant Formula Act of 1980, 21 U.S.C. § 350a, et seq., and the Administrative Procedure Act, 5 U.S.C. § 706.” Complaint at 1-2. FORMULA sought “an order requiring the FDA to issue regulations that prescribe mandatory quality control and testing procedures, including adequate record-keeping and audit procedures, in accordance with the Infant Formula Act at 1980.” Id. Both FORMULA and the FDA moved for summary judgment. In a published opinion, the district court denied both motions *745 without prejudice. FORMULA v. Schweiker, 572 F.Supp. 862, 866 (D.D.C.1983). At a subsequent status conference, the district court directed the FDA to provide further materials explaining the “industry framework” to aid the court’s understanding of the manner in which the rules would operate and whether the regulations were lawful. The FDA supplied the affidavit of Taylor M. Quinn, Associate Director for Compliance of the FDA’s Bureau of Foods; FORMULA thereafter was permitted, over the FDA’s objection, to take Mr. Quinn’s deposition. 2 On the parties’ subsequent renewed cross-motions for summary judgment, the district court granted the FDA’s motion and dismissed appellants’ suit. FORMULA v. Schweiker, 593 F.Supp. 346, 348 (D.D.C.1984). This appeal ensued. After careful review of the administrative record, we affirm the district court’s judgment. We hold that the district court correctly determined that there were no remaining genuine issues of material fact, that summary judgment was properly awarded to the FDA on the basis that the FDA’s final regulations were not unlawful under the Infant Formula Act of 1980, and that the regulations constituted a proper exercise of the agency’s discretion in effectuating the purposes of the statute. I For many infants in this country, infant formula provides the sole source of nutrition. Deficiencies in an infant formula— for example, the absence of an essential nutrient — may therefore have a seriously deleterious effect on infants’ nutrition and development. In 1979, precisely this development took place, when a number of infants were exposed over a significant period to two infant formula products of Syn-tex, Inc., namely Neo-Mull-Soy and Cho-Free. These two products were deficient in chloride, a nutrient essential to infants’ proper development. Staff of the Sub-comm. on Oversight and Investigations of the Senate Comm, on Interstate and Foreign Commerce, 96th Cong., 2d Sess., Infant Formula: Our Children Need Better Protection 6 (Comm. Print 1980) (hereinafter “Committee Print”). The chloride deficiency arose when the manufacturer stopped adding salt — sodium chloride — to Neo-Mull-Soy and Cho-Free in the spring of 1978 in order to reduce the sodium content of the two products. Notwithstanding this substantial change in the formulas’ ingredients, Syntex failed to conduct a single chloride assay on these products for approximately eighteen months after this reformulation. Committee Print at 19. Thus, the deficiency went undetected until after infants began to experience severe problems. Id. Indeed, the consequences of Syntex’s conduct for some infants were highly serious. Over 100 infants developed hypochlo- *746 remic metabolic alkalosis, a rare and potentially lethal chemical abnormality caused by the insufficient levels of chloride in their diets. H.R.Rep. No. 936, 96th Cong., 2d Sess. 4 (1980) (hereinafter “House Report”); S.Rep. No. 916, 96th Cong., 2d Sess. 3 (1980) U.S.Code Cong. & Admin. News 1980, p. 2858, (hereinafter “Senate Report”). The serious medical problem was further exacerbated, the Senate Subcommittee found, by the FDA’s inappropriate handling of the recall of the deficient formulas 3 and Syntex’s failure to cooperate in effectuating the recall. 4 Committee Print at 3-4. These unfortunate incidents provided the backdrop for substantial legislative activity that quickly ensued. Beginning in the fall of 1979, several members of Congress introduced bills addressing infant formula nutritional standards. 5 The Subcommittee on Health and the Environment of the House Committee on Interstate and Foreign Commerce held hearings on the proposed legislation in February and March 1980. Nutritional Quality of Infant Formula: Hearings on H.R. 6590, H.R. 6608, H.R. 5836, and H.R. 5839 Before the Sub-comm. on Health and Environment of the House Comm, on Interstate and Foreign Commerce, 96th Cong., 2d Sess. (1980). On March 23,1980, the Subcommittee approved a working draft of a bill that consolidated the proposed legislative approaches and ordered the version approved reported out as a clean bill. House Report at 3-4. On March 26, 1980, twelve members of Congress introduced H.R. 6940, which reflected the Subcommittee’s actions to date. 126 Cong.Rec. 6779 (1980). A similar scenario unfolded in the United States Senate. 6 Thereafter, both the House of Representatives and the Senate approved a bill, and the President signed the measure into law on September 26,1980. 126 Cong.Rec. 27,-798 (1980). As enacted, the Infant Formula Act departs substantially from the traditional regulatory mode with respect to food products. The statute provides that infant formula “shall be deemed to be adulterated” if (1) it fails to provide twenty-nine specific nutrients listed in 21 U.S.C. § 350a(g) (1982), id. at § 350(a)(1)(A); (2) it fails to “meet the quality factor requirements prescribed by the Secretary under this section,” id. § 350(a)(1)(B); or (3) “the processing of such infant formula is not in compliance with the quality control requirements prescribed by the Secretary under this section.” Id. § 350a(a)(l)(C). Of especial importance to this case, the statute further provides as follow. The Secretary may by regulation— (A) revise the list of [required] nutrients ...; *747 (B) revise the required level for any [required] nutrient ...; (C) establish requirements for quality factors for such nutrients; and (D) establish such quality control procedures as the Secretary determines necessary to assure that an infant formula provides nutrients in accordance with this section and establish requirements respecting the retention of records of procedures required under this clause (including maintaining necessary nutrient testing records). Quality control procedures prescribed by the Secretary shall include the periodic testing of infant formulas to determine whether they are in compliance with this section. Id. § 350a(a)(2). Subsection (e) of the statute prescribes recordkeeping requirements for manufacturers and defines the FDA’s oversight and enforcement role with respect to record-keeping. It provides as follows: (1) Each manufacturer of an infant formula shall make and retain such records respecting the distribution of the infant formula through any establishment owned or operated by such manufacturer as may be necessary to effect and monitor recalls of the formula. No manufacturer shall be required under this subsection to retain any record respecting the distribution of an infant formula for a period of longer than 2 years from the date the record was made. (2) To the extent that the Secretary determines that records are not being made or maintained in accordance with paragraph (1), the Secretary may by regulation prescribe the records required to be made under paragraph (1) and requirements respecting the retention of such records under such paragraph. Such regulations shall take effect on such date as the Secretary prescribes but not sooner than the 180th day after the date such regulations are promulgated. Such regulations shall apply only with respect to distributions of infant formulas made after such effective date. Id. § 350a(e). Subsection (g) sets forth in tabular form the twenty-nine nutrients that infant formula must contain, along with an appropriate minimum or maximum for each nutrient. Id. § 350a(g). 7 II A Pursuant to its authority under the Infant Formula Act, the FDA published proposed regulations on December 30, 1980. 8 45 Fed.Reg. 86,362 (1980). The proposed regulations reflected a highly extensive and elaborate regulatory approach that monitored “the entire infant formula manufacturing process, from the receipt of ingredients until the finished infant formula is released for shipping ... to ensure that each batch of the formula is uniform in its composition and conforms to the requirements of section 412 of the Act.” 45 Fed. Reg. 86,363 (1980) (emphasis added). The proposed rule, as the final rule itself was destined to do, employed quality control checks at three manufacturing stages — (1) upon the manufacturer’s receipt of ingredients; (2) during the manufacturer’s processing of ingredients; and (3) after completion of the final product (e.g., shelf life *748 testing). The proposed quality control requirements would have required infant formula manufacturers who used nutrient premixes supplied by other manufacturers to analyze a composite of random samples of the premixes — even those that are generally stable through shipping and storage and even if the premixes were guaranteed or certified by the supplier. 45 Fed.Reg. 86,-367 (1980). Under the proposed rule, the manufacturers would have been required to analyze a sample from each lot of other ingredients and premixes. Id. Only ingredients, other than premixes, which are generally stable in shipping and storage and which are accompanied by an appropriate guarantee or label could be used without further analysis. Id. In addition, consistent with the approach of intensive federal regulation, the proposed rules would have required monitoring of each batch at several steps during the manufacturing process; moreover, the proposed regulations would have required a manufacturer to accumulate in-process batches in holding tanks for testing and verification of nutrient content. Id. at 86,-368. The rules, as proposed, required the manufacturer to analyze by several different types of tests random samples of the finished product from each packaging line. Id. Also required was testing of infant formula after the manufacturer had altered the formula ingredients or the processing method, id. at 86,368-69, the testing which Syntex, of course, had failed to conduct in 1979. In contrast to the procedures prescribed in the final rule, however, these quality control procedures were not mutually exclusive. That is to say, a manufacturer’s performing the required analyses on the finished product, for example, did not excuse compliance with regulations governing ingredient or in-process control. In addition, the proposed regulations incorporated extensive recordkeeping requirements. These provisions directed manufacturers to maintain records verifying compliance with FDA regulations promulgated under the Infant Formula Act and other provisions of the Food, Drug, and Cosmetic Act. The proposed regulations further required manufacturers to maintain those quality control records “necessary to ensure nutrient content of infant formulas, including periodic nutrient testing.” Id. at 86,369. These latter records were to contain “sufficient information to permit a public health evaluation of any batch of infant formula.” Id. B The FDA received approximately 300 separate comments on the proposed regulations, some of which are reproduced in the parties’ joint appendix (“JA”). Numerous commentators, including the American Medical Association (“AMA”) and several infant formula manufacturers, expressed concern that the regulations, as proposed, would significantly increase production costs for infant formula and render such products too expensive for many consumers. Comments of Ross Laboratories, JA 15-16; Comments of Mead Johnson, JA 20; Comments of Infant Formula Council, JA 34; Comments of Loma Linda Foods, JA 42; Comments of Dr. Buford L. Nichols, Baylor College of Medicine, JA 47; Comments of AMA, JA 63, 65. The comments attributed the increased costs to capital outlays for changes in the manufacturing process which would be necessitated; increased operating expenses; Comments of Mead Johnson, JA 20; Comments of Loma Linda Foods, JA 42-43; and capital expenditures for expanded production and storage facilities to enable manufacturers to continue to meet consumer demand while complying with regulations that delayed delivery of finished products to the market. Comments of Infant Formula Council, JA 34. Two manufacturers expressed concern that some formulas with low sales volume, designed for infants with special needs, would be rendered so uneconomical under the proposed regulatory regime that the manufacturers would simply cease production of them. Comments of Ross Laboratories, JA 16; see Comments of Gerber, JA 5 (discussing projected impact of proposed *749 regulations on Meat Base Formula for infants with special dietary needs). In addition to the prospect of increased cost and the specter of certain product lines being terminated, several commentators, including the AMA, projected that the contemplated regulatory scheme would result in diminished product quality because maintaining infant formula in a holding tank at elevated temperatures in preparation for the requisite testing procedures would permit bacterial growth and oxidation, leading to a deterioration of nutrients. Comments of AMA, JA 34; Comments of Mead Johnson, JA 19. The efficacy of some proposed testing steps for ensuring product quality was deemed questionable, some commentators opined. Indeed, FORMULA’S own comments in this respect stated: “Because of the importance of finished infant formula testing, we question the necessity for all of these tests and whether harmful delays in the processing of infant formula could occur while all testing results are being obtained.” JA 59. As if more were needed, some manufacturers argued that the regulations would stifle technological improvements in infant formula processing; in their view, the detailed proposed rule would circumscribe the range of what the FDA would define as acceptable product, Comments of Ross Laboratories, JA 14-15; indeed, even minor changes would trigger extensive testing requirements, id. at 15-16; and changes in the processing system would require the manufacturer to petition for an exception from the prescribed quality control system. Comments of Mead Johnson, JA 19. The exacting level of detail set forth in the proposed regulations was deemed objectionable and unnecessary by the AMA and several manufacturers. Comments of Ross Laboratories, JA 13; Comments of Mead Johnson, JA 18; Comments of AMA, JA 64. Also under attack were the requirements for redundant testing at various stages in the manufacturing process. Comments of Infant Formula Council, JA 34; Comments of AMA, JA 64. Some commentators suggested that the FDA revise the proposal to require analytical proof that infant formula nutrient content did not deteriorate unacceptably during its shelf life. Comments of Mead Johnson, JA 20; Comments of Cutler Laboratories, JA 50. 9 The list of grievances went on. Wyeth Laboratories objected to the proposed requirement that manufacturers test nutrient premixes even though those premixes had been certified or guaranteed by the supplier. Comments of Wyeth Laboratories, JA 28. The specific argument was advanced that the supplier’s certification or guarantee ought to suffice, rather than requiring duplicative testing of the same ingredients at the manufacturer’s facility. In this respect, FORMULA’S comments urged that sampling and analysis ought to be required for all ingredients unless the supplier’s guarantee or certificate stated that it had sampled and tested the ingredient according to appropriate techniques. Comments of FORMULA, JA 57. C Published on April 20, 1982, the final regulations differed substantially from the *750 proposed rules. 10 47 Fed.Reg. 17,016 (1982). The Federal Register entry stated that the FDA had “reevaluated the proposal in light of the comments,” and that the agency believed “that the proposal was unnecessarily detailed, contained some proposed requirements that are not compatible with all manufacturing systems currently in use, and would have required some unnecessary testing.” Id. The final rule eliminated many of the more detailed requirements contained in the proposed rule and embraced instead a “more flexible regulatory scheme.” Id. at 17,017. The final rule requires each infant formula manufacturer to have a quality control system. FDA recognizes that because manufacturing processes, practices, and equipment vary, no single type of system would be appropriate for all manufacturers. Therefore, the final rule does not require any particular provisions in the quality control system. A manufacturer may establish any reasonable system. In addition, § 106.20 (21 CFR 106.20) requires ingredient composition review, § 106.25 (21 CFR 106.25) requires verification of in-process control records, and § 106.30 (21 CFR 106.30) requires finished product compliance. FDA concludes that these controls are adequate and that it is unnecessary to require additional test methods and procedures or to have records reconfirmed on a 6-month basis. 47 Fed.Reg. 17,018-19. The FDA stated that the final rule “together with the procedures described in the current good manufacturing practice (CGMP) regulations[ 11 ] that apply to infant formulas (21 CFR Parts 110 and 113), are sufficient to provide assurance of the safety and nutritional adequacy of infant formulas.” Id. at 17,-018. The final quality control procedures regulations addressed ingredient control, 21 C.F.R. 106.20 (1984), in-process control, id. § 106.25, and finished product evaluation. Id. § 106.30. Picking up on the comments with respect to pre-mix ingredients, the quality control procedures with respect to ingredient control require no manufacturer testing for those ingredients that meet two requirements. First, they must be generally stable in shipping and storage. Second, they must be received under a supplier’s guarantee or certification that the supplier has analyzed the ingredient mixture for its nutrient composition or be “labeled as having nutrient compositions complying with” certain recognized standards. Id. § 106.-20(a). Further testing requirements apply unless the manufacturer tests the finished product pursuant to 21 C.F.R. § 106.-30(b)(1) (1984). If the supplied ingredient is not generally stable in shipping or storage, the regulations require the manufacturer to “analyze that ingredient for each relied-upon nutrient that may be affected [adversely by shipping or storage conditions], using validated analytical methods.” For ingredients that lack a supplier’s certification or guarantee or a label signifying compliance, the manufacturer itself must sample and analyze the ingredient for each relied-upon nutrient. Ingredients used as a major source of protein or fat are exempt from this requirement if the manufacturer’s records show that the relied-upon nutrients are “present at a reasonably constant level.” Id. § 106.20(b)(2). The regulations similarly require that the manufacturer sample and analyze nutrient premixes received from outside suppliers unless the *751 supplier has certified that it has sampled and analyzed each batch of premix for each relied-upon nutrient. Id. 12 Regulations governing in-process control require the manufacturer to prepare a “manufacturing order” 13 governing manufacture of each different type of infant formula for approval by a "responsible official of the manufacturer.” Id. § 106.25(a). The final regulations also require the manufacturer to “establish a quality control system that assures and verifies the addition of each ingredient specified in the manufacturing order.” Id. As in the ingredient control regulations, the manufacturer’s election not to test the finished product pursuant to 21 C.F.R. § 106.30(b)(1) (1984) triggers additional in-process testing requirements. The manufacturer must analyze each in-process batch for a number of formula constituents including “the indicator nutrient(s) in each nutrient premix,” and “[e]ach nutrient added independently of nutrient premixes during formulation of the product, except for lineolic acid, vitamin D, vitamin K, choline, inositol, and biotin....” Id. § 106.25(b)(3), (4). The finished product evaluation regulations require the manufacturer to “establish criteria for sampling and testing to ensure each batch of infant formula meets the nutrient requirements” of the Act and the regulations. Id. § 106.30(a). The regulations mandate that “[b]efore release of product [sic] for commercial or charitable distribution, the manufacturer shall analyze representative samples of each batch,” id. § 106.30(b)(1), for nutrient content, including “[a]ll nutrients not previously analyzed for by the manufacturers, unless each in-process batch is analyzed for nutrients as specified in § 106.25(b) and the ingredients are analyzed as specified in § 106.20(b)”. Id. § 106.30(b)(l)(ii). The regulations do not, however, require batch-by-batch analysis for the six nutrients enumerated in the prior paragraph, namely, lineolic acid, vitamin D, vitamin K, choline, inositol, and biotin, nor for nutrients “added as a part of a nutrient premix analyzed by the manufacturer or having a supplier’s guarantee or certification and for which an indicator nutrient(s) was analyzed by the manufacturer.” Id. In addition, the regulations, as finally adopted, require periodic analysis and stability analysis of the product. Id. § 106.-30(b)(2), (3). The periodic analysis requirements mandate that the manufacturer sample and analyze “at least one newly processed finished product batch every 3 months” for all nutrients except those for which an analysis is done in the pre-distri-bution “immediate analysis” required under id. § 106.30(b)(1) (1984). To ensure the product’s quality during its shelf life, the manufacturer must conduct a stability analysis on “representative samples collected from finished product batches ... with sufficient frequency to substantiate the maintenance of nutrient content throughout the shelf life of the product.” Id. § 106.30(b)(3). Finally, the regulations direct the manufacturer to conduct certain tests on both new formulas and upon modifications in the formula itself, the level of nutrients, or the manufacturing process. The regulation specifies that formulas affected by a “major change,” which is defined to include new formulations and “any change of ingredients or processes where experience or theory would predict a possible significant adverse impact on levels of nutrients or availability of nutrients,” must be analyzed for a number of formula constituents and attributes. The manufacturer may release the formula for distribution, however, prior to receiving the results of vitamin D and protein biological quality bioassays, which are required if the formulation change is expected to have an adverse effect on the *752 biological quality of the protein, “provided that other tests for the presence of those two nutrients have already been completed.” Id. § 106.30(c)(2). The regulations also prescribe, in rather general fashion, recordkeeping requirements for manufacturers. Section 106.-100(a) provides as follows: The manufacturer shall maintain quality control records that contain sufficient information to permit a public health evaluation of any batch of infant formula. Subsection (b) sets forth record retention requirements and requires that the records be made available to the FDA for inspection and copying. 14 Ill On this appeal, FORMULA argues that the FDA’s final regulations fly in the teeth of both the Infant Formula Act and the Administrative Procedure Act. We analyze first FORMULA’S arguments with respect to the Infant Formula Act and then turn to the issues raised under the APA. A [1] In its wide-ranging attack on the regulations, FORMULA argues strenuously that the FDA's quality control regulations violate the Infant Formula Act because they fail, in FORMULA’S view, to include necessary testing requirements. The regulations are too broad and vague, FORMULA maintains, and leave entirely too much to self-regulation by industry, as opposed to the more elaborate regulatory regime envisioned by Congress and reflected in the initial, proposed regulations. Thus, a principal bone of contention between FORMULA and the FDA is the requisite level of detail that the regulations must contain in order to conform to Congress’ will as embodied in the statute. Specifically, the parties invite us to focus upon the meaning of the pivotal term, “periodic testing,” found in the final sentence of 21 U.S.C. § 350a(a)(2)(D) (1982), which is the provision authorizing the Secretary to promulgate certain types of regulations. That sentence, quoted in full context at p. 7 supra, provides as follows: “Quality control procedures prescribed by the Secretary shall include the periodic testing of infant formulas to determine whether they are in compliance with this section.” FORMULA argues that “periodic testing” requires the manufacturer to test each batch of formula in the manufacturing process for each of the statutorily prescribed twenty-nine nutrients prior to releasing the product for distribution to the public. To support this position, FORMULA points to three regulatory “exemptions” — as FORMULA sees it — granted by FDA from the testing program prescribed in the statute: (1) manufacturers are not required, as we have seen, to test nutrient premixes if they are properly labeled or received under a supplier’s certificate or guarantee to their content; (2) manufacturers are not obliged to sample and analyze infant formula for six ingredients identified in the statute — lineolic acid, vitamin D, vitamin K, cholin, inositol and biotin — prior to release for public distribution; and (3) products affected by a “major change” — as defined by the regulations — may be released for public distribution before completion of bioassays for vitamin D and protein. In contrast to FORMULA’S view of the FDA’s statutory obligations, the agency argues that the Infant Formula Act expressly affords the FDA broad discretion in determining how best to achieve statutory aims and, more specifically, that the term “periodic testing” was not intended to prescribe a particular regulatory scheme, such as batch testing for all nutrients. The agency maintains that the full analysis (as to all nutrients) required by the FDA every three months on at least one newly *753 processed batch of finished product, 21 C.F.R. § 106.30(b)(2) (1984), amply fulfills Congress’ intent in specifying periodic testing. In view of these competing views of the statute, we thus begin our inquiry by attempting to divine the intent of Congress with regard to the term “periodic testing.” In questions of statutory construction, our initial inquiry must, of course, be to “determine whether Congress ‘has directly spoken to the precise question at issue.’” General Motors Corp. v. Ruckelshaus, 742 F.2d 1561, 1566-67 (D.C.Cir.1984) (en banc) (quoting Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984)), cert. denied sub nom. General Motors Corp. v. Thomas, — U.S. -, 105 S.Ct. 2153, 85 L.Ed.2d 509 (1985). “If the intent of Congress is clear ... the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” See Chevron, supra, 104 S.Ct. at 2781-82. In that process, “reviewing courts ... must not ‘rubber-stamp ... administrative decisions that they deem inconsistent with a statutory mandate or that frustrate the congressional policy underlying a statute.’” Bureau of Alcohol, Tobacco and Firearms v. Federal Labor Relations Authority, 464 U.S. 89, 97-98, 104 S.Ct. 439, 444, 78 L.Ed.2d 195 (1983) (quoting NLRB v. Brown, 380 U.S. 278, 290-92, 85 S.Ct 980, 987-88, 13 L.Ed.2d 839 (1965)). Our analysis begins with the statute itself. See, e.g., Consumer Product Safety Comm’n v. GTE Sylvania, 447 U.S. 102, 108, 100 S.Ct. 2051, 2056, 64 L.Ed.2d 766 (1980) (“the starting point for interpreting a statute is the language of the statute itself”); Reiter v. Sonotone Corp., 442 U.S. 330, 337, 99 S.Ct. 2326, 2330, 60 L.Ed.2d 931 (1979) (“our starting point must be the language employed by Congress”); Center for Auto Safety v. Ruckelshaus, 747 F.2d 1, 3 (D.C.Cir.1984) (same, quoting Reiter). The Infant Formula Act employs the term “periodic testing” only once, in the sentence which we have before us for examination. The statute contains no definition of the term nor any other language to aid our understanding. The FDA urges that the plain meaning of the word “periodic,” namely, “happening or appearing at regular intervals,” American Heritage Dictionary (2d college ed. 1982), indicates that Congress intended for the regulations to prescribe testing at regular intervals such as the intervals prescribed in the post-release requirements set forth at 21 C.F.R. § 106.30(b)(2) (1984). On its face, however, this definition would permit testing at regular intervals of once a decade or once every fifteen years, an extraordinary result that no one could reasonably deem to effectuate Congress’ intent with respect to infant formula quality control. In addition, the literal language of “periodic testing” does not by itself exclude an interpretation that would require testing at regular intervals during the manufacturing process in a pattern that would approximate or match the batch testing scheme urged by appellants (and embraced by the discarded proposed regulations). In sum, the plain meaning of the word “periodic” is not so evident as to close our inquiry. To the contrary, the term “periodic testing” is susceptible to various interpretations, including the competing interpretations championed by the parties. Cf. Wisconsin Electric Power Co. v. Department of Energy, 778 F.2d 1, at 3-4 (D.C.Cir.1985). In view of this ambiguity, it is appropriate for us to turn to the legislative history to examine whether any light is shed there as to Congress’ intent in employing this particular term. See, e.g., United States v. American Trucking Ass’n, 310 U.S. 534, 543-44, 60 S.Ct. 1059, 1064, 84 L.Ed. 1345 (1940); Chemehuevi Tribe of Indians v. Federal Power Comm’n, 420 U.S. 395, 402-05, 95 S.Ct. 1066, 1071-73, 43 L.Ed.2d 279 (1975); State of Montana v. Clark, 749 F.2d 740, 745 (D.C.Cir.1984), cert. denied sub nom. Montana v. Hodel, — U.S.-, 106 S.Ct. 246, 88 L.Ed.2d 255 (1985). We thus return to the setting of the late 1970’s when this statute emerged from the Article I branch *754 in the wake of the unfortunate experiences of chloride-deficient Neo-Mull-Soy and Cho-Free. The infant illnesses arising out of the Neo-Mull-Soy and Cho-Free episodes provided the direct impetus for the legislative activity culminating in passage of the Infant Formula Act. As we recounted above, a number of bills were introduced in the fall of 1979 following the discovery of the chloride deficiencies in the two Syntex products during the prior spring. The legislative history of the bills contains abundant evidence of Congress’ awareness of these two incidents and of congressional sentiment that the incidents had demonstrated manifest shortcomings in existing legislative and regulatory controls over infant formula. Congressman Gore, one of the sponsors of the bill, stated in remarks on the floor of the House of Representatives: We found in the investigative hearing [before the Subcommittee of Health and the Environment of the House Committee on Interstate and Foreign Commerce] that the current law is incredibly inadequate. Under current law any individual or group of individuals can mass market virtually any concoction that they want to as the sole source of nourishment for infants in this country. This set of circumstances cannot be allowed to continue, in my judgment. Under current law, no company is required to pretest an infant formula before it is mass marketed, and under current law, the Food and Drug Administration does not have the authority to insure nutritional quality or to guarantee that a prompt and effective recall of adulterated infant formulas occurs. 126 Cong.Rec. 11,790 (1980) (emphasis added). Congressman Waxman, another of the bill’s sponsors, expressed the same theme: When I first learned about this [Neo-Mull-Soy and Cho-Free] episode, I was shocked to discover that there are no Federal statutes or regulations which require infant formulas to contain all nutrients recognized as essential. Federal regulations with respect to infant formula are currently limited to assurances that labels are accurate and that the formula is processed in sanitary facilities. The absence of more substantive requirements regarding formula content may have created a regulatory environment which permitted the marketing of the nutrient deficient formulas. Id. at 11,789. Both the House Report (at 4-5) and the Senate Report on the companion legislation before that body (S. 2490, at 3-4) refer to the Neo-Mull-Soy and Cho-Free incidents in describing the need for legislation. Both reports summarized the need for legislation in identical language: For parents to continue to have confidence in the quality of formula upon which their children depend, they must be assured that formula contains all essential nutrients and has been adequately tested prior to marketing. Mandatory Federal standards to assure the safety and quality of infant formula are in the public interest. The Committee believes the passage of [the legislation] will go far in restoring public confidence in the safety and nutritional quality of infant formula. House Report at 5; Senate Report at 4. Congressman Gore explained in remarks from the floor that the legislation sought to improve the existing controls over infant formula in three principal ways: First, it would require that any infant formula marketed in the United States as the sole source of nutrition for normal babies include minimum amounts of all essential nutrients. Second, it would require that infant formulas be tested before they are mass marketed in order to insure the nutritional adequacy of those formulas. Third, this legislation would require the Food and Drug Administration to establish mandatory procedures for the recall of deficient formulas. 126 Cong.Rec. 11,790 (1980). Fairly viewed, the legislative history indicates Congress’ recognition of the need for *755 legislation to prevent manufacturers from repeating Syntex’s error, namely, releasing for public distribution new or reformulated infant formula without testing to ensure the product’s nutritional soundness. 15 To promote expeditious enforcement of uniform nutritional standards, Congress took the unusual step of including in the statute a table specifying nutritional standards, 21 U.S.C. § 350a(g) (1982), as opposed to leaving that task to the FDA. House Report at 6. At the same time, however, Congress recognized that the FDA’s regulatory expertise was important in effectuating the purposes of the statute and in enforcing compliance with the statute’s nutritional standards. 21 U.S.C. § 350a(g) (1982). This recognition is evident from the expansive rulemaking authority that Congress conferred upon the agency, as well as from remarks throughout the legislative history pertaining to the Secretary’s authority with respect to, inter alia, quality control of infant formula processing, quality factors pertaining to shelf life of infant formula, and recordkeeping requirements. For example, the House Report calls attention to the provision, later codified at 21 U.S.C. § 350a(a)(2) (1982), which “provided authority to the Secretary to require that formula nutrients contain certain quality factors and that the formula be manufactured in accordance with effective quality control procedures. Quality factors pertain to the bioavailability of a nutrient and the maintenance of level or potency of nutrients during the expected shelf life of the product.” House Report at 6. In its section-by-section analysis of the bill, the House Report refers to the Secretary’s authority “to establish ... quality control ... requirements necessary to insure that infant formula is safe and will promote healthy development.” Id. at 15. The House Report also states that “[t]he Secretary is authorized to establish quality factors for nutrients and quality control procedures for formula processing.” Id. at 16. As to the particular regulatory scheme that would implement the legislation, the legislative history contains little more than general references to the rulemaking authority delegated to the Secretary. Among the rare, more specific references are remarks by Congressman Gore in support of a Senate amendment that altered the language of the provision now codified at 21 U.S.C. § 350a(a)(2) (1982) to its present form. Speaking on the House floor, Congressman Gore stated: The amendment provides the Secretary with clear authority to require the periodic testing of infant formula prior to marketing ... In combination with mandatory ingredient standards, conscientious and periodic testing of infant formula during the manufacturing process is the best protection against a repetition of last summer’s tragedy. 126 Cong.Rec. 24,817 (1980) (emphasis added). Congressman Gore’s remarks are noteworthy because they contain one of only a very few references to “periodic testing,” which is, of course, at the center of the dispute here. Although the Congressman’s remarks indicate that he envisioned the term as embracing testing at a stage prior to release of the infant formula for public *756 distribution, not merely afterwards as the FDA argues, nothing in these remarks suggests that “periodic testing” was meant to require the batch-by-batch testing regime urged by FORMULA. Indeed, Congressman Gore couched all of his remarks in terms of what the Secretary was authorized to do; he did not suggest in any way that the legislation required the Secretary to promulgate any rules governing periodic testing or any rules beyond those “necessary” in his or her judgment. See also id. (remarks of Congressman Waxman) (subsection (a)(2)(D) designed “to permit and encourage” Secretary to prescribe periodic testing regulations). 16 Although FORMULA asserts that the legislative history of the statute supports its argument that Congress intended to require batch testing, or even testing of “each container” of formula (appellants’ brief at 21-22, quoting district court’s memorandum opinion, FORMULA v. Schweiker, 572 F.Supp. at 865), for every nutrient prescribed in the statute, the legislative materials upon which FORMULA relies simply do not, upon analysis, support this sweeping understanding. The cited portions of the Subcommittee Report, for example, indicated Congress’ sentiment that legislation was necessary to prevent repetition of the Neo-Mull-Soy and Cho-Free incidents and that infant formula testing was “[a]n indispensable need.” Committee Print at 19, 20. The Subcommittee recommended that “Congress enact legislation requiring, at a minimum, that ‘infant formulas’ be tested for nutritional components before marketing and after any reformulation or other change in the manufacturing process.” Id. at 20. The Subcommittee Report added that “[cjonsideration should also be given to require similar testing at periodic intervals during marketing.” Id. at 20. We have found nothing in the Subcommittee Report, however, suggesting that “periodic testing” was meant to require the comprehensive batch-by-batch testing approach championed by FORMULA. Indeed, in this context — “periodic intervals during marketing” — it would not be unreasonable to conclude that the Subcommittee was proposing a testing scheme for infant formula already released for public distribution, wholly outside the manufacturing process. FORMULA thus has failed to cite to us, nor have we found in the course of our own research, any legislative materials supporting the understanding of “periodic testing” as requiring batch-by-batch testing for each essential nutrient. We note, moreover, that the final regulations do require the manufacturer to test each batch at some stage prior to distribution for twenty-three of the twenty-nine required ingredients if they were not added as part of a properly certified, guaranteed, or labeled nutrient premix. 21 C.F.R. §§ 106.25, 106.-30 (1984). In the absence of evidence that the FDA’s regulations deviate from the intent of Congress, we defer to the agency’s considered expertise in exercising the authority delegated to it under the Infant Formula Act. Chevron, U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 104 S.Ct. 2778, 2782, 81 L.Ed.2d 694 (1984); Whirlpool Corp. v. Marshall, 445 U.S. 1, 11, 100 S.Ct. 883, 890, 63 L.Ed.2d 154 (1980). *757 2 Having rejected FORMULA’S principal argument (that the statute directs the FDA to adopt regulations requiring batch-by-batch testing for all twenty-nine ingredients), we need not dwell at undue length upon the several “exemptions” which FORMULA argues undercut what it deems to be the statutorily mandated batch-by-batch testing scheme. We begin this phase of our inquiry by reemphasizing the broad language Congress employed in authorizing the Secretary to promulgate regulations: “The Secretary may by regulation ... (D) establish such quality control procedures as the Secretary determines necessary to assure that an infant formula provides nutrients in accordance with this section and establish requirements respecting the retention of records of procedures required under this clause (including maintaining necessary nutrient testing records).” 21 U.S.C. § 350a(a)(2) (1982) (emphasis added). 17 The Secretary may promulgate regulations; indeed, the legislative history indicates that Congress hoped the Secretary would promulgate regulations governing the areas enumerated in § 350a(a)(2), including quality control, § 350a(a)(2)(D). Congress left it to the Secretary to determine which regulations were “necessary.” Having left this judgment entirely up to the Secretary, Congress offered virtually no guidance as to the specific content or methodology of any regulations that the Secretary might see fit to promulgate. The agency inscribed its regulations, in a word, on a clean slate. In this branch of its challenge, FORMULA maintains that the regulations do not protect against release of nutritionally deficient infant formula products for the following reasons: (1) manufacturers are not required to test for lineolic acid, vitamin D, vitamin K, choline, inositol or biotin at any stage (other than, of course, before a new or changed product enters the market) before release for distribution, and (2) manufacturers may rely on premixes to supply essential nutrients and thereby avoid sampling and analysis requirements. FORMULA points out that an infant formula manufacturer could, at least theoretically, rely solely on premixes for all twenty-nine nutrients designated by statute. If each nutrient is stable — a likelihood about which we have little if any information in this record — and is received under a supplier’s certificate or guarantee as to their composition, or meets the rule’s labeling requirement, then the manufacturer would not need to sample and analyze the premix. 18 *758 FORMULA also argues that regulations governing testing of new and reformulated products are inadequate because they allow the manufacturer to release the product for public distribution before completion of bioassays for vitamin D and protein. This lacuna, FORMULA argues, could permit nutritionally deficient formula to reach consumers and as much as three months might pass before the periodic analysis prescribed in 21 C.F.R. § 106.30(b)(2) (1984) would even detect the deficiency. This inadequacy is particularly egregious, FORMULA asserts, because Congress enacted the legislation to prevent, at a minimum, release of nutritionally deficient reformulated products, which was of course the situation with Syntex. The FDA persuasively responds to each of FORMULA’S concerns. With regard to the six “exempted” ingredients identified in § 21 C.F.R. § 106.25(a)(4) (1984), the FDA in the comments accompanying the final rule specifically stated that these ingredients are unlikely to be absent from an infant formula and that many infants receive some of these nutrients from alternative sources. 47 Fed.Reg. 17,021 (1982). The FDA therefore balanced the potential benefit to the public in requiring more frequent testing against the costs the manufacturer would incur in conducting such testing and then settled upon a requirement of testing as to all twenty-nine nutrients only (1) before release of a new formula, 21 C.F.R. § 106.30(c)(2) (1984), (2) after certain reformulations, id., and (3) on a newly processed finished batch every three months, id. § 106.30(b)(2). As to premixes, the FDA maintains persuasively, in the face of appellants’ general arguments to the contrary, that a supplier’s certificate or guarantee, coupled with the FDA’s authority to take action under other provisions of the Food, Drug and Cosmetic Act against suppliers who furnish false certificates or guarantees, provide sufficient assurances that the finished product will contain the required nutrients. The FDA also points out that the manufacturer must test every infant formula prior to initial release for marketing, every three months, and after certain reformulations. With regard to bioassays for vitamin D and protein, the agency observed in its comments accompanying notice of the final rule that bioassays are rather time-consuming and do not stand alone as safeguards of nutrient quality. This is the FDA’s primary rationale for allowing release of the formula before completion of the bioassays under certain circumstances. Indeed, early release is permissible only if the bioassays are already underway “and if another analysis for the vitamin D has been run and the protein content has been determined by a suitable method.” 21 C.F.R. § 106.30(c)(2) (1984). The FDA contends that this early-release provision is entirely sound since the bioassays constitute, at bottom, double checks against other test methods; the FDA further observes that nothing in the statute actually requires bioassays. In the face of the agency’s arguments, FORMULA protests that his light-handed regulatory scheme will not prevent manufacturers from releasing nutritionally deficient formula products for public distribution. In particular, FORMULA points out that a deficient formula that slips through the testing steps prescribed by the statute —if, for example, the missing nutrient is one of the six for which the manufacturer need not test prior to releasing the formula for public distribution, or if the missing nutrient is a non-indicator nutrient included in a premix — could be on grocery store shelves for three months before the required periodic analysis would detect the deficiency. While true, this argument misses the decisive point. The language of the statute, as we read it, leaves to the agency’s discretion the choice of the extent and pattern of regulation that will best ensure quality control. FORMULA’S arguments that the regulations thwart the statute’s purposes *759 would be better directed to the legislative body whose broad delegation of authority triggered the exercise of agency discretion in the first instance. [3] FORMULA also challenges the regulation’s stability testing requirements, promulgated pursuant to the Secretary’s delegated authority to issue quality factor regulations. 21 U.S.C. § 350a(a)(2)(C) (1982). The stability testing requirements are set forth at 21 C.F.R. § 106.30(b)(3) (1984), which provides as follows: (3) Stability analysis. Using representative samples collected from finished product batches, the manufacturer shall conduct stability analysis for selected nutrients with sufficient frequency to substantiate the maintenance of nutrient content throughout the shelf life of the product. FORMULA argues that the regulation leaves too much to manufacturers’ discretion. The regulation allows manufacturers to decide which nutrients to analyze and how often to conduct the analysis. The regulation, FORMULA further argues, does not require testing even for nutrients known to be unstable. The stability testing regulations therefore fail, in FORMULA’S view, to fulfill the statute’s directives. The statute provides on this subject only that “[t]he Secretary may by regulation ... (C) establish requirements for quality factors for such nutrients_” 21 U.S.C. § 350a(a)(2)(C) (1982). The statute contains no definition of quality factors, but the House Report states that quality factors “pertain to the bioavailability of a nutrient and the maintenance of level or potency of nutrients during the expected shelf life of the product.” House Report at 6; Senate Report at 5. The FDA responds to FORMULA’S arguments by pointing out that nothing in the statute requires stability testing and that not a word in the legislative history suggests that Congress intended to require such testing. The agency refers to its remarks accompanying the announcement of the final rule wherein the FDA explained that it elected to rely on the manufacturers’ application of scientific data and experience to determine which nutrients they should analyze and how frequently they should conduct the analysis. In its brief before this court, the FDA represents that it monitors manufacturers’ determinations under the stability analysis provision. Products of manufacturers who do not conduct adequate stability analyses, the agency indicates, may be deemed adulterated under 21 U.S.C. § 350a(a)(l)(B) and (C) (1982). The provisions of the stability analysis regulation are indeed broadly worded. We have no information in the record before us with respect to the FDA’s success vel non in enforcing this regulation, nor what impact this provision has had on the FDA’s administration of the statute as a whole. Given the broad delegation of authority in the statute, however, and the absence of expressions of congressional intent to require something more than or different from the scheme embraced by the FDA, we are unable to conclude that the regulation in this respect is at odds with the statute. [4] In its final line of attack under the statute itself, FORMULA argues that the regulation’s recordkeeping requirement violates the Infant Formula Act because it does not require maintenance and retention of records documenting the actual contents of the infant formula product. The final rule requires manufacturers to maintain “quality control records that contain sufficient information to permit a public health evaluation of any batch of infant formula.” Section 106.100(b) of the proposed rule sets forth essentially the same requirement but was preceded by subsection (a), deleted from the final rule, which would also have required manufacturers to maintain “records of results of examinations and copies of suppliers’ guarantees or certifications that verify compliance with FDA regulations, guidelines, or action levels for raw materials, food-packaging materials, and infant formulas....” *760 The FDA no doubt relies on accurate records pertaining to infant formula production in monitoring compliance with the statute and the regulations. The central requirement defining which records the manufacturer is required to keep is broadly worded. Under 21 C.F.R. § 106.100(a) (1984), to all appearances, the manufacturer decides in the first instance whether the records it chooses to maintain contain “sufficient information” as required by the regulation. The FDA has an opportunity, during routine inspection or administration of a recall, to opine whether a manufacturer’s records do indeed contain “sufficient information.” While the regulation is broad, we do not agree, however, that the rule departs impermissibly from the statute. We reach this conclusion in view of the broad language of the statute itself, which provides that the “Secretary may by regulation ... establish requirements respecting the retention of records of procedures required under this clause (including maintaining necessary nutrient testing records).” 21 U.S.C. § 350a(a)(2)(D) (1982). Absent further guidance in the statute or evidence in the statute’s legislative history that indicates that the agency’s interpretation is not in keeping with congressional intent, we cannot conclude to the contrary. IV FORMULA also challenges the regulations as arbitrary and capricious under the APA insofar as the agency failed adequately to explain the extensive differences between the proposed rule and the final version. FORMULA assails as implausible the FDA’s explanations that it could not promulgate quality control regulations applicable to the different infant formula manufacturing processes, and argues that the agency improperly relied on cost considerations in rejecting the proposed rule. Three terms ago, the Supreme Court addressed the scope and application of the arbitrary and capricious standard 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 (1983). That statement of law bears repeating here: The scope of review under the “arbitrary and capricious” standard is narrow and a court is not to substitute its judgment for that of the agency. Nevertheless, the agency must examine the relevant data and articulate a satisfactory explanation for its action including a “rational connection between the facts found and the choice made.” Burlington Truck Lines v. United States, 371 U.S. 156, 168 [83 S.Ct. 239, 245-55, 9 L.Ed.2d 207] (1962). In reviewing that explanation, we must “consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment.” Bowman Transportation, Inc. v. Arkansas-Best Freight System, Inc., supra, [419 U.S. 281] at 285 [95 S.Ct. 438, 441-42, 42 L.Ed.2d 447]; Citizens to Preserve Over-ton Park v. Volpe, supra, [401 U.S. 402] at 416 [91 S.Ct. 814, 823-24, 28 L.Ed.2d 136]. Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has 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. The reviewing court should not attempt itself to make up for such deficiencies; we may not supply a reasoned basis for the agency’s action that the agency itself has not given. SEC v. Chenery Corp., 332 U.S. 194, 196 [67 S.Ct. 1575, 1577, 91 L.Ed. 1995] (1947). We will, however, “uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.” Bowman Transportation Inc. v. Arkansas-Best Freight System, Inc., supra, [419 U.S.] at 286 [95 S.Ct. at 442]. See also Camp v. Pitts, 411 U.S. 138, 142-143 [93 S.Ct. 1241, 1244, 36 L.Ed.2d 106] (1973) (per curiam). 463 U.S. at 43, 103 S.Ct. at 2866-67; see also International Ladies’ Garment *761 Workers’ Union v. Donovan, 722 F.2d 795, 814 (D.C.Cir.1983) (applying arbitrary and capricious standard to decision to rescind restrictions on homework in knitted outerwear industry), cert. denied sub nom. Breen v. International Ladies’ Garment Workers’ Union, — U.S.-, 105 S.Ct. 93, 83 L.Ed.2d 39 (1984). The agency action at issue in State Farm was, of course, the Department of Transportation’s decision to rescind the requirement of passive restraints for automobiles. State Farm, 463 U.S. at 34, 103 S.Ct. at 2862. In contrast to the regulations at issue in State Farm, the regulations involved here were, of course, only in proposed form; they had not yet enjoyed legal effect. The agency, to be sure, is still obliged to provide a rational explanation as to how it conceived and fashioned the new version of the rule. Applying the guidelines of State Farm here, we conclude that the agency did, in fact, provide a reasoned basis for the final rule and for its departure from the earlier, proposed regulatory approach. In its preamble to the notice of the final rule, the agency explained the rationale underlying its reconsideration of the proposed rule: FDA received 300 letters of comment from manufacturers of infant formulas, a trade association, universities, individuals, and a U.S. Senator. Most of the comments from individuals supported the proposal. These comments emphasized that the nutritional quality of infant formulas is paramount and that regulations safeguarding infant formulas are worth any additional product cost. However, a number of individuals, the American Medical Association, the American Academy of Pediatrics, and the Infant Formula and Nutrition Organization suggested that some of the proposed testing may not be necessary to assure nutrient content or formula quality and that the burden of such tests should be reduced so that the cost of the infant formulas would not increase dramatically. The manufacturers of infant formulas contended that the cost of implementing the proposal would be excessive, product quality might be adversely affected by delays resulting from waiting for test results and evaluation, and few, if any, consumer safety benefits would accrue. 47 Fed.Reg. 17,016 (1982). The agency went on to state: FDA has reevaluated the proposal in light of the comments. The agency believes that the proposal was unnecessarily detailed, contained some proposed requirements that are not compatible with all manufacturing systems currently in use, and would have required some unnecessary testing. FDA has revised the final rule so that it will be compatible with all manufacturing systems, removed unnecessary provisions, and eliminated unnecessary sampling and testing requirements. FDA believes that the final rule ... will provide assurance that infant formulas contain all required nutrients, as its testing requirements include requirement that a manufacturer test each manufacturing batch of infant formula for required nutrients. On the other hand, FDA believes the cost of compliance with the final rule should have no perceptible effect on the retail cost of infant formula. Id. The agency reiterated these principles in response to a comment suggesting that the FDA establish requirements for in-process test methods and procedures and that it require a quality assurance review every six months: The final rule requires each infant formula manufacturer to have a quality control system. FDA recognizes that because manufacturing processes, practices, and equipment vary, no single type of system would be appropriate for all manufacturers. Therefore, the final rule does not require any particular provisions in the quality control system. A manufacturer may establish any reasonable system. In addition, § 106.20 [21 CFR 106.20] requires ingredient composition review, § 106.25 [21 CFR 106.25] requires verification of in-process control *762 records, and § 106.30 [21 CFR 106.30] requires finished product compliance. FDA concludes that these controls are adequate and that it was unnecessary to require additional test methods and procedures or to have records reconfirmed on a 6-month basis. Id. at 17,018-19. These prefatory remarks set the stage for the agency’s explanation of the changes affecting various provisions of the regulations. The notice of the final regulation reflects a careful study and weighing the numerous comments that the agency received. See 5 U.S.C. § 553(c) (1982) (informal rulemaking should encompass “consideration of the relevant matter presented”); 1 K. Davis, Administrative Law Treatise § 6:12, at 505-06 (1978) (in informal rule-making, agency's statement of basis and purpose or statement or reasons should include responses to important comments); see also Batterton v. Marshall, 648 F.2d 694, 703-04, n. 47 (D.C.Cir.1980) (citing S.Doc. No. 248, 79th Cong., 2d Sess. 19-20 (1946), regarding importance of public participation in rulemaking). We set forth below several examples that illustrate the agency’s reasoned basis for making these changes. With regard to ingredient control, for example, the agency adopted a substantially more flexible approach in its final regulations than embodied in the proposed version. The agency explained that these changes eliminated duplicative testing that would have been required under the proposed rule and that the final version afforded manufacturers more flexibility to develop testing plans that suited their own production processes. The agency also explained that certain formula nutrients were exempted from tests at certain steps of the production process because the FDA had determined that less frequent testing would be adequate to ensure the nutritional quality of the formula. Specifically, the final rule required the manufacturer to test for the six nutrients of vitamin D, vitamin K, choline, biotin, inositol, and lineolic acid only once every three months for the stated reason that there were no documented cases of deficiencies of these nutrients in normal infants, and other properties and readily available alternative sources of these nutrients made it unlikely that infants would suffer from such deficiencies in infant formula. The agency also stated that it had determined that manufacturers need not sample and analyze ingredients relied on as sources of fat or protein because these ingredients do not, in the agency’s judgment, vary in their nutritional composition to any significant extent from shipment to shipment. With respect to regulations governing in-process control, the agency’s final rule incorporates significant changes that afford manufacturers more flexibility in testing procedures. The final regulations eliminate base blend testing and testing of samples from a holding tank in light of comments that pointed to significant practical problems with such requirements. The final regulations also eliminate the requirement that manufacturers test for homogeneity, osmolality, and sedimentation while the product is in-process in view of comments that these attributes are not related to the nutritional quality of the formula. 19 The agency explained that each of these changes was made in light of comments received in response to the proposed rule. The FDA points out that FORMULA’S own comments were among those advocating less extensive testing during the in-process stage because of the “harmful delays” that might result. This comment was, in fairness, made in the context of the more extensive and somewhat duplicative testing scheme set forth in the proposed regulations. But the point is that even from those favoring a detailed, intensive regulatory approach, the proposed rules were deemed to impose excessive testing *763 requirements that could have had deleterious consequences. The agency also revised the final product evaluation regulations to allow for variations among manufacturers’ production processes. The final regulations required the manufacturer to analyze fewer samples of the finished product and gave the manufacturer flexibility to decide which formula nutrients should be analyzed to determine whether nutrients had been damaged during the processing of the formula. The agency also revised the rule to adopt the substance of a comment urging that manufacturers be required to test every three months for all nutrients not subjected to immediate analysis, including vitamin D, vitamin K, choline, inositol, biotin, and li-neolic acid. The agency cited comments suggesting that manufacturers be allowed to determine on the basis of “scientific data and experience” the appropriate frequency, within limits, of stability analysis in explaining its decision to require such analysis to be of “sufficient frequency to substantiate the maintenance of nutrient content throughout the shelf life of the product.” 21 C.F.R. § 106.30(b)(3) (1984). The agency also discussed at length changes in the requirements for vitamin D and protein bioassays, made in response to comments questioning the cost-benefit balance of the vitamin D bioassay requirement in the proposed regulation and the suitability of a particular bioassay method for protein. Thus, the record abundantly demonstrates that the agency carefully considered and evaluated the record before it in the decisionmaking process that culminated in the final rule. Still, FORMULA argues, the proffered rationale for the change — difficulty of promulgating a single rule to cover widely different manufacturing processes and the potentially burdensome cost to manufacturers of complying with a more detailed rule — lacks plausibility. FORMULA adverts to the detailed regulations governing the much more numerous class of pharmaceutical manufacturers and suggests that, in view of that much larger class being subjected to a single, standardized regulatory scheme, it should be even less difficult for the FDA to promulgate detailed, comprehensive regulations governing the much smaller infant formula industry. While the FDA does not respond specifically to this argument, we are simply not in a position to second guess the agency’s judgment in this matter. The specialized agency is far better suited than the generalist judiciary to assess the particular needs and problems of both industries and to determine which schemes of regulation are most appropriate in each instance. In any event, we do not view the agency’s stated rationale as “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, 103 S.Ct. at 2866-67. FORMULA also argues, as we have previously observed, that the FDA improperly based its decision to adopt the less-detailed final regulations on considerations of the costs that the industry would incur in complying with the proposed regulations. FORMULA argues that in the statute Congress announced the policy “that health and safety concerns are paramount to considerations of increased industry costs,” and that the FDA has elevated cost concerns to such a degree that the final regulation fails to implement Congress’ policy. It is indeed true that an agency rule may be fatally arbitrary and capricious “if the agency has relied on factors which Congress has not intended it to consider.” Id. That is not the case here, however. The language of the statute commits the regulatory responsibility to the FDA without any prohibitions as to what the agency may legitimately consider. Moreover, the legislative history indicates Congress’ concern that the regulatory scheme not overburden industry. In remarks from the floor on S. 2490, by that time amended to conform to the House version of the bill, Senator Metzenbaum, one of the sponsors of S. 2490, stated that the legislation will ensure the safety of formula products without placing an undue burden on formula manufacturers. *764 This is the way legislation should be written — taking into account the legitimate concerns of all interested parties. 126 Cong.Rec. 24,512 (1980). Senator Bau-cus, another sponsor of S. 2490, stated in remarks inserted after the floor debate on the amended version of S. 2490 that “this legislation does not impose burdensome regulatory requirements.” Id. The legislative history also suggests that Congress was concerned about the impact that increased manufacturing costs might have on the availability of infant formula. In his opening remarks at Senate hearings on S. 2490, Senator Hatch, who became a cosponsor of the bill after its introduction, stated that the legislation ought to respond to concerns about infant formula safety and nutrition without “destroying availability” of infant formula products. Senate Hearings at 41. While the evidence in this respect is by no means compelling, we are satisfied based upon our reading of the legislative history that Congress manifested an intent that cost considerations could, in fact, play a legitimate role in shaping the regulatory scheme. At a minimum, we have found nothing to suggest that this factor could not lawfully be taken into account by the FDA. Finally, FORMULA contends that the re-cordkeeping requirements of the regulations are arbitrary and capricious because the FDA failed to provide a reasoned explanation as to why it abandoned the more detailed recordkeeping requirements set forth in the proposed rule in favor of the less stringent, more general requirements of the final regulation. We are unpersuaded; upon analysis, we conclude that the agency provided a sufficiently reasoned explanation for this change. Essentially, the FDA eliminated subsection (a) of the proposed rule and revised subsection (b) of the proposed rule to eliminate a reference to “periodic nutrient testing.” The Federal Register notice announcing promulgation of the final rule stated that the change was intended to eliminate references to records pertaining to provisions other than § 350a(a). This explanation of an intent to organize the regulations in a more rational way reflects an exercise of judgment on the part of the agency that we are unable to conclude is arbitrary or capricious. V Having carefully evaluated all of appellants’ arguments, we hold that the FDA’s regulations contravene neither the Infant Formula Act nor the Administrative Procedure Act. Accordingly, for the reasons stated, we affirm the district court’s judgment. Affirmed. Footnotes [1] “Infant formula” is statutorily defined as "a food which purports to be or is represented for special dietary use solely as a food for infants by reason of simulation of human milk or its suitability as a complete or partial substitute for human milk”. 21 U.S.C. § 321(aa) (1982). [2] At the outset, we emphasize that we do not rely in any measure on Mr. Quinn’s, affidavit or deposition in reaching our decision. We conclude wholly independently of the affidavit and the deposition that there were no genuine issues of material fact in the case that precluded entry of summary judgment and that the agency was entitled to judgment as a matter of law. Byers v. Burleson, 713 F.2d 856, 859 (D.C.Cir. 1983) (setting forth appellate court’s standard of review in appeals from grant of summary judgment). Therefore, we are not required to pass on the propriety of the district court’s procedure in ordering the supplemental affidavit and the ensuing discovery. We find that procedure troubling, however, for two reasons. First, the affidavit and deposition that followed may be viewed as supplementing the administrative record in a manner allowing the FDA to offer post hoc rationalizations to support the challenged rule. Ordinarily, of course, a reviewing court is to confine its inquiry with respect to the validity of an administrative rule to the administrative record. Camp v. Pitts, 411 U.S. 138, 142, 93 S.Ct. 1241, 1244, 36 L.Ed.2d 106 (1973); Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 419, 91 S.Ct. 814, 825, 28 L.Ed.2d 136 (1971). Second, the district court’s references in its second memorandum opinion to factual information contained in Mr. Quinn’s affidavit suggests that the court may have employed that affidavit to resolve factual disputes that, in the court’s earlier holding, had precluded summary judgment on the parties’ initial motions. It is, of course, inappropriate for a district court to find disputed facts in the context of a summary judgment motion. Rodway v. United States Department of Agriculture, 482 F.2d 722, 727 (D.C.Cir.1973), citing Empire Electronics Co. v. United States, 311 F.2d 175, 179 (2d Cir.1962). [3] FDA compliance officials designated the recalls as Class II, signifying that illness from use of the formulas was reversible or remote, in the face of scientific experts’ recommendations that the recall ought to be designated as Class I, which signifies a life-threatening situation. Committee Print at 2-3. [4] Syntex, Inc. failed to provide pertinent information to the FDA with respect to the products subject to recall. Committee Print at 13. [5] Then-Congressman (now Senator) Gore introduced H.R. 5836 on November 8, 1979, 125 Cong.Rec. 31,565 (1979); Congressmen Gore and Mottl introduced H.R. 5839 on the same day, id.; Congressman Carter introduced H.R. 6590 on February 25, 1980, 126 Cong.Rec. 3661 (1980); and Congressman Gore introduced H.R. 6608 on February 26, 1980, 126 Cong.Rec. 3893 (1980). [6] Senators Metzenbaum, Baucus, and Leahy introduced S. 2490, a bill that also addressed nutritional standards for infant formula, on March 27, 1980. 126 Cong.Rec. 6917 (1980). The Subcommittee on Health and Scientific Research of the Senate Committee on Labor and Human Resources held hearings on S. 2490 on June 12, 1980. Infant Formula Act of 1980: Hearing on S. 2490 Before the Subcomm. of Health and Scientific Research of the Senate Comm, on Labor and Human Resources, 96th Cong., 2d Sess. (1980). On August 5, 1980, the Subcommittee considered a working draft reflecting modifications of H.R. 6940, which the House already had approved. The Subcommittee approved the draft with one amendment offered by Senator Metzenbaum, and its language was substituted for S. 2490. The committee ordered the bill reported on August 21, 1980. S.Rep. No. 916, 96th Cong., 2d Sess. 2 (1980), U.S.Code Cong. & Admin.News 1980, p. 2858. [7] In addition, subsection (b) imposes a requirement that manufacturers provide certain types of notice to the FDA; the requirements expired upon the FDA’s promulgation of final regulations. Id. § 350a(b). Subsection (c) requires a manufacturer that has "knowledge which reasonably supports the conclusion” that its product may be adulterated to notify the FDA. Id. § 350a(c). Subsection (d) prescribes procedures governing recalls of infant formula and defines the FDA’s oversight role, and subsection (f) exempts special formulas, e.g., those for infants with inborn metabolic errors, low birth weight, or other unusual medical or dietary problems, from the provisions of subsections (a) and (b). Subsections (b), (c), (d), and (f) of the statute are not implicated in FORMULA’S challenge to the regulations. [8] The Secretary delegated his authority to promulgate regulations under this statute to the Commissioner of Food and Drugs. 21 C.F.R. § 5.10(a)(1). [9] The proposed rule required "progressive analysis” of a new filling batch every three months during the batch’s shelf life. 45 Fed.Reg. 86,368 (1980). Manufacturers were required to test for "physical attributes (i.e., solids, homogeneity, osmolality, and sedimentation), and for all nutrients declared on the label.” Id. The rule also outlined special procedures for testing the formula’s Vitamin D and protein content. Id. The final rule, by contrast, requires the manufacturer to conduct a stability analysis on representative samples "for selected nutrients with sufficient frequency to substantiate the maintenance of nutrient content throughout the shelf life of the product.” 21 C.F.R. § 106.30(b)(3). The recordkeeping requirements under both the proposed and final rules required the manufacturer to “maintain quality control records that contain sufficient information to permit a public health evaluation of any batch of infant formula.” 21 C.F.R. § 106.100(a) (final rule); 45 Fed. Reg. 86,368 (1980) (proposed rule). Thus, the comments on this point appear to have been directed to the recordkeeping requirement, not to the substance of the required analysis, which was more extensive in the proposed rule than in the final version. [10] The FDA did not republish the new version of the regulations as proposed regulations but invited comment on the final regulations. 47 Fed.Reg. 17,016, 17,017 (1982). The agency stated that it would amend the rule if necessary in light of comments received on the final rule. Id. at 17,017. FORMULA elected not to challenge the regulations on the basis of this decision not to republish the rule in proposed form. Plaintiffs' Memorandum in Support of Summary Judgment Motion at 12 n. 2. That point is thus not before us. [11] Current Good Manufacturing Practices regulations outline procedures governing the handling of food products throughout the manufacturing process so that they will be safe for human consumption. See 21 C.F.R. §§ 110.1-110.99; id. §§ 113.3-113.100 (1984). [12] The manufacturer, it should be noted, must sample and analyze nutrient premixes that it prepares for its own use. 21 C.F.R. § 106.-20(b)(2). [13] A manufacturing order, appellees inform us, is a list of all the ingredients in a particular formula and the production steps to be followed in manufacturing the formula. Appellees’ Brief at 9-10. [14] Subpart D of the regulations, not challenged in this suit, prescribes notification requirements. The key provision requires a manufacturer who has knowledge that reasonably supports the conclusion that its product may be adulterated or nutritionally deficient to notify the FDA. 21 C.F.R. § 106.120(b) (1984). [15] The regulations also reflect this understanding insofar as they require extensive testing for new formulations or reformulations that involve "major changes” as defined in the regulation. 21 C.F.R. § 106.30(c) (1984). Appellees suggest, and appellants do not deny, that Syn-tex’s defective formula likely would not have . reached the market had these regulations been in effect at that time. Appellants point to two other formula deficiency incidents. In 1982, Wyeth mistakenly omitted vitamin B6 from two of its products. In 1983, Loma Linda Foods recalled one of its formula products when the FDA determined that the formula’s vitamin A had deteriorated below acceptable levels. The FDA represents, and appellants do not deny, that testing required by the final rule would have prevented the Wyeth incident. As for the Loma Linda incident, which appellants argue underscores the need for testing at regulatorily prescribed intervals, we note that the FDA has represented its intent to enforce the regulation’s "sufficient frequency" requirement in a way that will ensure continued nutrient content quality throughout the product’s shelf life. Appellee’s Brief at 30-31. [16] FORMULA also cites the remarks of Congressman Gore at Senate Hearings on S. 2490 for the proposition that premarket testing could have prevented the chloride deficiency incidents. To the extent that FORMULA means to suggest that Congressman Gore was endorsing the idea of batch-by-batch testing for all ingredients, that is simply not accurate. Congressman Gore stated Syntex had reformulated the two products and then mass marketed them without testing the new compositions in order to make sure they contained sufficient amounts of all the nutrients needed for healthy development of infants. We also found that formula manufacturers are not required by law to conduct such tests, nor are they prohibited from marketing a formula deficient in vital nutrients. Infant Formula: The Present Danger: Hearing on S. 2490 Before the Suhcomm. on Health and Scientific Research of the Senate Comm, on Labor and Human Resources, 96th Cong., 2d Sess. 42 (1980). In context, Congressman Gore's statement is more fairly understood as referring to the need for testing of new and reformulated products to avoid Syntex’s error, as opposed to periodic testing or batch-by-batch testing. [17] FORMULA draws our attention to House oversight hearings in 1982 to discuss vitamin deficiencies in certain infant formula products that arose after enactment of the Infant Formula Act of 1980. Infant Formula: The Present Danger: Hearing Before the Suhcomm. on Oversight and Investigations of the House Comm, on Energy and Commerce, 97th Cong., 2d Sess. (1982). In particular, FORMULA focuses on the testimony of FDA Commissioner Hayes that the FDA interpreted the permissive language of the statute “as a rather direct mandate from Congress in view of the legislative history.” Id. at 26. Although the word “may" ordinarily is a permissive word and not mandatory, Bennett v. Panama Canal Co., 475 F.2d 1280, 1282 (D.C.Cir.1973), this presumption may change in light of the relevant legislative history. United States v. Rodgers, 461 U.S. 677, 706, 103 S.Ct. 2132, 2149, 76 L.Ed.2d 236 (1983). Even if the word “may" is to be construed as mandatory in this statute, however, this does not undercut the breadth of the authority delegated to the Secretary who “may” promulgate whatever quality control regulations he or she deems "necessary.” 21 U.S.C. § 350a(a)(2)(D) (1982). [18] The regulations do require, however, analysis of each finished batch for "[t]he indicator nutrients) in each nutrient premix." 21 C.F.R. § 106.25(a)(3) (1984). Specifically, the regulations set forth the following definitions: (e) Nutrient premix. A nutrient premix is a combination of ingredients containing two or more nutrients. 21 C.F.R. § 106.3(e). (a) Indicator nutrient. An indicator nutrient is a nutrient whose concentration is measured during the manufacture of an infant formula to confirm complete addition and/or uniform distribution of a premix or other substance of which the indicator nutrient is a part. Id. § 106.3(a). (d) Nutrient. A nutrient is any vitamin, mineral, or other substance required in accordance with the table set out in section 412(g) of the act or by regulations promulgated under section 412(a)(2)(A) of the act. Id. § 106.3(d). Thus, the regulations require that the manufacturer test each finished batch before release *758 for distribution for at least one of the relied on nutrients listed in the statute, 21 U.S.C. § 350a(g), for every nutrient premix provided by an outside supplier. [19] The final regulations continued to require the manufacturer to test for proper dilution and for osmolality of new formulas or formulas affected by a "major change” as defined in the regulations. 21 C.F.R. § 106.30(c)(2) (1984).