General Medical Company v. United States Food and Drug Administration, and Margaret M. Heckler, Secretary Department of Health & Human Services

*216 McGOWAN, Senior Circuit Judge: The pace of proceedings at the Food and Drug Administration (FDA), for whatever reasons, does not rival that of, say, a turn-of-the-century sweatshop in New York City. See Mcllwain v. Hayes, 690 F.2d 1041 (D.C.Cir.1982) (affirming FDA’s decision, under statute passed in 1960 setting forth 2-and-V2-year grace period, to extend deadline for statutory showing until 1984); Corn Products Co. v. FDA, 427 F.2d 511, 513 n. 5 (3d Cir.) (affirming 1968 order, original version of which issued in 1959, that mandated peanut butter should be 90%, not 87%, peanuts by weight), cert. denied, 400 U.S. 957, 91 S.Ct. 354, 27 L.Ed.2d 265 (1970). Almost ten years ago, Congress passed the Medical Device Amendments of 1976 (Amendments), Pub.L. No. 94-295, 90 Stat. 539 (codified at 21 U.S.C. §§ 360c-360k (1982)). The first cases reviewing the Amendments are now trickling into this court. See Contact Lens Manufacturers Association v. FDA, 766 F.2d 592, 594 (D.C.Cir.1985) (CLMA v. FDA). Here, we review the FDA’s order denying the petition of the General Medical Company (General Medical) for reclassification of its “Drionic” antiperspirant device. This case presents us with some sticky matters of statutory interpretation and some evidence of procedural sloppiness on the part of the FDA. We hold that the FDA has rightly guarded the consumer against the general marketing of a device not yet shown to be both safe and effective, and has adopted an allowable interpretation of the Amendments in so doing. We also hold that, while the FDA was less than punctilious in adhering to some procedures that safeguard manufacturers of medical devices against arbitrary action by the FDA, the FDA’s position in this case is strong enough that we see no utility in a remand on procedural grounds. We therefore uphold the FDA’s decision to deny General Medical’s petition for reclassification of the Drionic device. I The general scheme of the Amendments has been ably summarized by Judge Ginsburg in CLMA v. FDA. See CLMA v. FDA, at 592-595. Under the Amendments, the FDA is armed with a different set of regulatory weapons for each class of devices. In general, Class I products may be set upon the stream of commerce subject only to “general controls,” such as those necessary to prevent misbranding or adulteration. See 21 U.S.C. § 360c(a)(l)(A) (1982); see also id. §§ 351, 352, 360, 360f, 360h, 360i, 360j (describing general controls). Class III products, in contrast, must traverse the costly rapids of a “premarket approval.” See 21 U.S.C. § 360c(a)(l)(C) (1982); see also CLMA v. FDA, at 596 (noting that premarket approval for rigid gas-permeable contact lenses generally costs $750,000-$l,000,000). Class II products must negotiate the barrier, presumably intermediate in height between general controls and a premarket approval, of a “performance standard.” See 21 U.S.C. § 360c(a)(l)(B) (1982). This litigation concerns only the showing necessary for the FDA to allow reclassification from Class III to Class I. Neither party seems interested at the moment in attempting to reclassify the Drionic device into Class II. In addition, we note that the petitioners do not here directly contest the FDA’s initial decision to place the Drionic device in Class III. See generally CLMA v. FDA, at 594, 595 (discussing Act’s placement of devices outside scheme of Act or into Class III according to whether they were introduced before or after date when statute became effective). The petitioner has, however, mounted in another lawsuit a challenge of sorts to the FDA’s classification of the device in Class III. There, the petitioner has exacted from the FDA the acknowledgment in a settlement agreement that the Drionic device is “substantially equivalent” to a device marketed by the Fischer Company before the effective date of the Act. See Stipulation and Order of Dismissal, General Medical Co. v. FDA, Civ. Action No. 83-3314 *217 (D.D.C. filed Nov. 19, 1984) (reproduced as Exhibit 3 of Petitioner’s Post-Argument Brief on the Question of Mootness (D.C.Cir. filed Jan. 7, 1985)). Such an admission would ordinarily be of significant use to a manufacturer of a good introduced on the market after the effective date of the Amendments. See 21 U.S.C. § 360c(f)(l) (1982) (goods introduced after effective date are automatically in Class III unless “substantially equivalent” to goods introduced before effective date). General Medical is not, however, content with the FDA’s admission here. The FDA considers the “substantial equivalence” to be between the Fischer device and “the Drionic device as sold to physidans for prescription,” rather than between the Fischer device and “the Drionic device as sold over the counter to the lay person.” See Letter from Robert G. Britain, Director, Office of Device Evaluation, Center for Devices and Radiological Health, to Steve R. Trost, Esq. (Sept. 29, 1984) (reproduced as Exhibit 4 of Petitioner’s Post-Argument Brief on the Question of Mootness (D.C.Cir. filed Jan. 7, 1985)). General Medical wishes to gain Class I status for the Drionic device for over-the-counter sales. They, we, and the FDA all seem to agree — albeit with varying degrees of discomfort 1 — that the FDA’s current stance on substantial equivalence makes that status impossible to gain for Drionic *218 devices sold over-the-counter without our ruling here that the FDA must reclassify the device from Class III to Class I regardless of the Drionic device’s similarity vel non to other devices. It is to the question of what showing Congress intended to be necessary before such a reclassification could occur that we now turn. II The Amendments define a Class III device as A device which because— (i) it (I) cannot be classified as a class I device because insufficient information exists to determine that [general] controls ... are sufficient to provide reasonable assurance of the safety and effectiveness of the device and (II) cannot be classified as a class II device because insufficient information exists for the establishment of a performance standard to provide reasonable assurance of its safety and effectiveness, and (ii) (I) is purported or represented to be for a use in supporting or sustaining human life or for a use which is of substantial importance in preventing impairment of human health, or (II) presents a potential unreasonable risk of illness or injury, is to be subject ... to premarket approval to provide reasonable assurance of its safety and effectiveness. 21 U.S.C. § 360e(a)(l)(C) (1982). General Medical does not seek reclassification of the Drionic device into Class II. Section 360c(a)(l)(C)(i)(II) is therefore not at issue here. The FDA does not claim that the device, which is used to prevent excessive perspiration, is purported to be for use in sustaining human life or for a use of substantial importance in preventing impairment of human health. Section 360c-(a)(l)(C)(ii)(I) is therefore not at issue here. The FDA asserts rather that the Drionic device falls into Class III because insufficient information exists to show that general controls will provide reasonable assurance of the device’s effectiveness and because the device presents a potential unreasonable risk of injury. The FDA asserts that in this case it may require General Medical to demonstrate the safety and effectiveness of the Drionic device before it will concede that the device does not present an unreasonable risk of injury. We uphold as supported by substantial evidence the FDA’s findings of fact that General Medical has not shown the Drionic device to be safe and effective, and that the device presents a potential unreasonable risk of injury. See Rhone-Poulenc, Inc. v. FDA, 636 F.2d 750 (D.C.Cir.1980) (upholding as supported by substantial evidence an FDA decision that a new animal drug was not “safe and effective”); Edison Pharmaceutical Co. v. FDA, 600 F.2d 831 (D.C.Cir.1979) (same outcome in case involving drug designed for human use). We uphold as a permissible statutory interpretation the FDA’s position that the Amendments allow it to require a showing of safety and effectiveness before conceding that the particular device at issue here does not present a potential unreasonable risk of illness or injury. See Chevron U.S.A., Inc. v. Natural Resources Defense Council, — U.S. -, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). After a brief discussion of the device at issue, we examine the clinical study that is the main battleground between General Medical and the FDA, and the FDA’s interpretation of the Amendments allowing it to consider that study insufficiently persuasive evidence for General Medical’s contention that the Drionic device should be reclassified into Class I. A. The FDA’s Findings of Fact The Drionic device is designed as a substitute for chemical antiperspirants, 2 or *219 for the extensive medical treatments sometimes recommended for those who suffer from greatly increased perspiration (“hyperhydrosis”). The device consists of a housing for two wool felt pads and a battery. The pads are soaked in ordinary tap water and placed against the treated area, typically the hands, feet, or underarms. An electrical current is passed through the pads and the area of skin between the pads for about twenty minutes. Ions — atoms carrying an electrical charge — are thereby transmitted across the skin in a process called “iontophoresis.” This iontophoresis is thought to be the cause of whatever effectiveness the device may have. General Medical also proposes to include with the device a bottle of 100% sodium iodide solution and some sheets of paper impregnated with starch. When properly combined and placed against the skin, these materials produce a “print” of the skin with a darkness roughly equal to the amount of perspiration emitted from the area. By comparing prints of the treated area made before and after treatment, the user is said to be able to obtain an idea of the effectiveness of the treatment. General Medical carries the burden of proving that the device meets the requirements for reclassification set up by the Amendments. See CLMA v. FDA, at 599-GOO. Its main “valid scientific evidence” 3 showing the device’s risks and benefits consists of a study initially involving 169 human subjects. See I Joint Appendix (J.A.) 22. The subjects received several treatments over a period of a few weeks on one or more of three sites: hands, feet, and underarms. Id. Each set of treatments on a particular site was called a “case”; some subjects received treatments on more than one site. The study began with 225 “cases” under investigation. Id. at 23. Only 60 cases were still under investigation at the end of six weeks. There is no explanation given of how those 60 were chosen. Of those 60 cases, after two weeks almost 90% showed some decrease in their sweating from pre-treatment levels, as measured by a rough, qualitative measure similar to the “print” method proposed to be included with the device when marketed. Id. at 25. Of the 1,946 instances in which a patient received some treatment on some site, there were only 72 complaints. Id. at 26. Twenty-six of these complaints were not specifically identified. Id. at 28. The identified complaints were generally of mild tingling or burning, or minor skin irritations. 4 Id. The FDA has listed a number of concerns that it had with the study in its denial of. the petition for reclassification. See Letter to Larry Pilot, II J.A. 592 (letter serves as order denying reclassification). First and foremost is the failure of the study to explain how the treating physicians chose the 60 “follow-up” cases from the initial 225 cases. II J.A. at 614-15. Well over half of the patients were not *220 seen for a two-week or six-week follow-up. One can reasonably assume that patients who obtain no benefit from the treatment, or who are positively harmed by the treatment, are more likely to drop out of the study. If such selection occurs, then those cases still in the study for follow-up after a number of weeks will obviously constitute a sample of the initial population that is skewed towards patients for whom the treatment is safe and effective. The accuracy of the study’s results on safety and effectiveness are thereby obviously reduced. The FDA had a number of other concerns with the study. The “print” method is qualitative and subjective. Because the study was conducted at four different institutions but no attempt was made to standardize the use of the “print” method, the subjective and qualitative nature of the “print” method is even more likely to have reduced the accuracy of the study’s results on effectiveness. Id. at 614-17. The FDA was also concerned because a number of patients experienced bilateral reductions in perspiration although they had been treated on only one side. Id. at 615. A number of theories could explain such a result. Perspiration might be under the control of the central nervous system in such a way that the device affected that central control even though it was applied to only one side of the body. Or the treatment might have only a “placebo” effect: the reduction in perspiration could be a purely psychological byproduct of the ambience of a medical experiment, an effect that might have been produced with equally good results by “treating” the subjects with a completely inert device. The bilateral reduction in perspiration in patients treated only on one side would be consistent with this hypothesis. Or the reduction in perspiration may have stemmed from the subjects’ growing familiarity with the device and the events surrounding its use, with a concomitant reduction in perspiration-causing anxiety over time. As with the second explanation, the observed reduction in perspiration would occur regardless of the true effectiveness of the device. Obviously, only the first of these explanations is support for the effectiveness of the device. Because General Medical was unable to dispel the FDA’s reasonable doubts about which of these theories explained the bilateral reduction in perspiration in patients treated on one side only, the FDA properly considered the bilateral reductions as of little use in showing the effectiveness of the Drionic device. The FDA was also concerned even by the relatively small number of complaints noted by the study, particularly since the study involved following the subjects for only six weeks, while purchasers of the device might well use it for much longer periods of time. Id. at 617-19. The FDA feared that purchasers of the device who also used chemical antiperspirants might suffer from the transfer of aluminum ions, often present in chemical antiperspirants, across their skin along with the other ions in the current. Id. at 619. The FDA worried that some of the chemical byproducts of the “print”-solution reaction could cause dermatological problems after repeated use. Id. at 619-21. We find this congeries of concerns more than sufficient to uphold the findings by the FDA that, given General Medical’s burden of proof, the Drionic device was not “safe and effective” and presented a potential unreasonable risk of illness or injury. We discuss briefly Congress’s intent with respect to these definitions and then apply them to the facts of this case. The Amendments specify that “safety and effectiveness” is to be determined by “weighing any probable benefit to health from the use of the device against any probable risk of illness or injury from such use.” 21 U.S.C. § 360c(a)(2)(C) (1982); see also 21 C.F.R. § 860.7 (1985) (discussing how to show safety and effectiveness). The relevant House Report explains Congress’s intent with respect to this weighing in more detail: The key concept that safety and effectiveness of a device are to be determined “weighing any probable benefit to health *221 from the use of the device against any probable risk of injury or illness from such use” makes it clear that the proposed legislation recognizes that products having the power to be useful in the healing arts also have the potential to do harm and that the determination of safety and effectiveness is to carefully balance these considerations. Regulation cannot eliminate all risks but rather must eliminate those risks which are unreasonable in relation to the benefits to be derived. H.R.Rep. No. 853, 94th Cong., 2d Sess. 16-17 (1976). The Amendments’ use of “potential unreasonable risk of illness or injury,” 21 U.S.C. § 360c(a)(l)(A)(ii)(II) (1982), id. § 360c (a)(l)(C)(ii)(II), is similarly meant to allow a balancing of benefits against risks, and is intended to emphasize a particular caution where safety is involved: The phrase “presents a potential unreasonable risk of illness or injury” has two significant features. First, the requirement that a risk be unreasonable contemplates a balancing of the possibility that illness or injury will occur against benefits from use. Second, the risk need only be a potential one. The risk may be one demonstrated by reported injuries or it may simply be foreseeable. The fact that a device is being marketed without sufficient testing is an adequate basis for the Secretary’s conclusion that the device presents a potential unreasonable risk to health. H.R.Rep. No. 853, at 36. In view of the clinical study’s potentially biased sample of follow-ups and the frequent bilateral response to unilateral treatment, as well as to a less convincing extent the FDA’s other concerns, we think it within the FDA’s broad discretion to have ruled that the effectiveness of the device has not been demonstrated. As far as the safety of the device is concerned, the potentially biased sample of follow-up patients means that the risks of the device may be significantly underestimated by the study’s results. In any case, the actual complaints from some subjects are sufficient to show that the device is not without some risks. Where the benefits are essentially impossible to determine and the harms are small but clearly demonstrated, it is within the FDA’s broad discretion to find that a device is not “safe and effective.” Similarly, the FDA was within its broad discretion in weighing unproven benefits against small but proven harms and finding the balance tilted towards a finding of a “potential unreasonable risk of illness or injury.” We note that such a finding is especially reasonable in light of the legislative history when, as here, the FDA can plausibly identify some potential risks in addition to those actually demonstrated by the study, and when inadequate testing is among the FDA’s chief reservations about the device. B. The FDA’s Statutory Interpretation The FDA has interpreted the Amendments to allow it to require a showing of the safety and effectiveness of the Drionic device before reclassifying the device from Class III to Class I. General Medical argues that the FDA must reclassify any device that does not present a potential unreasonable risk of illness or injury, even if the device has not been shown to be safe and effective. We uphold the FDA’s interpretation of its statutory mandate on the facts of this case, but note that there is, at least hypothetically, a category of devices which the FDA must reclassify from Class III to Class I even if the manufacturer cannot show that the device is safe and effective. That class is likely to include at most only a small number of devices, but we can make no sense of Congress’s structuring of the Amendments without assuming that devices in such a category may exist. We note first that the primary concern of the Amendments is with safety and effectiveness. The need for such a showing is often the only showing mentioned in discussions of the Amendments. See, e.g., H.R.Rep. No. 853, at 3 (“[The bill] would provide significant new authority to the *222 Secretary of Health, Education, and Welfare to assure the safety and effectiveness of medical devices intended for human use.”); id. at 5 (“Legislation which would amend the Federal Food, Drug, and Cosmetic Act to provide for the safety and effectiveness of medical devices for human use was first considered ... in October of 1973....”); id. at 12 (“[The bill] would prohibit ... the marketing of a new device until the safety and effectiveness of the device has been established____”); id. (medical devices are to “be classified into one of three categories depending on the extent of regulation: necessary to assure safety and effectiveness”); id. at 31 (premarket approval “is to be denied if the Secretary finds a lack of showing of reasonable safety or effectiveness”). These general statements, many of which are unqualified, emphasize that Congress’s primary concern was to prevent the marketing of devices not both safe and effective. Nonetheless, Congress did pass specific provisions which, especially in light of the legislative history, appear to allow the marketing in certain narrowly defined circumstances of devices that may not be both safe and effective. Section 360c(a)(l)(A)(ii), for example, specifically allows into Class I “a device for which insufficient information exists to determine that [general] controls ... are sufficient to provide reasonable assurance of the safety and effectiveness of the device,” but which meets the preconditions of subsections 360c(a)(l)(A)(ii)(I) and (II). Section 360c(a)(l)(C) sets forth the corresponding conditions in defining which devices fall into Class III: it states that Class III devices are those for which the general controls of Class I or the performance standards of Class II are insufficient to assure safety and effectiveness, and are purported to be of substantial importance to human health or present a potential unreasonable risk of illness or injury. We must infer from the presence of conditions in addition to those governing safety and effectiveness that some criteria other than safety and effectiveness are potentially important in determining the classification of a device. The legislative history of the Amendments similarly supports the conclusion that Congress contemplated that a device not both safe and effective might be able to enter the stream of commerce, although such a device has to meet conditions so narrow that no actual device may fall into the category: The second type of device which is to be classified into class I consists of those devices for which insufficient information exists with which to determine that general controls are sufficient or to establish a standard to assure safety and effectiveness, but which (1) do not purport or are not represented to be for uses of substantial importance in supporting, sustaining, or preventing impairment of human life or health, and (2) do not present potential unreasonable risks of illness or injury. The Committee recognizes that no device may fall into this latter category. It is, however, necessary to include it in the proposed legislation in view of the requirements (described below) that two determinations be made prior to classification of a device into class III. H.R.Rep. No. 853, at 34. The legislative history’s discussion of Class III devices and the “two determinations” similarly implies that devices the manufacturer of which cannot make a showing of safety and effectiveness are not always relegated to Class III: [T]he bill adopts a two-pronged test to determine whether a device should be required to undergo premarket approval. The first criterion screens out devices which can be adequately regulated by general controls or standards: Premarket approval is required for a device if it cannot be classified into class I or II because insufficient information exists to determine the adequacy of general controls or standards to provide reasonable assurance of safety and effectiveness. The second requirement provides that premarket approval is to be required only for devices which either are for a *223 use which is of substantial importance in supporting, sustaining, or preventing impairment of human life or health, or which present a potential unreasonable risk of illness or injury. This requirement will assure that premarket approval does not become a routine requisite for all devices, while still providing the Secretary with ample latitude to classify a device into class III in instances in which its use poses public health concerns. H.R.Rep. No. 853, at 35. Once more, it appears that the inquiry into safety and effectiveness does not end the classification determination. The material from which we are to infer Congress’s intent is in this case less than crystal clear. We believe that the fairest reading of Congress’s intent is that it contemplated an occasional device for which neither general controls nor performance standards were sufficient to guarantee safety and effectiveness, but which might nonetheless properly enter the stream of commerce without premarket approval, so long as the device was not purported to be of substantial importance in improving human health and presented no potential unreasonable risk of illness or injury. The exception is much narrower than might at first blush appear, since the line between “safety and effectiveness” and “no potential unreasonable risk of illness or injury” may be quite thin. That is to say, even a small risk of minor illness or injury may be an unreasonable one when the device in question is of no demonstrable benefit. That is the situation here. The effectiveness of the Drionic device is, at least when evaluated with the data currently available, dubious; and the studies have shown at least some risk of minor harm. Particularly since the device is intended for long-term use, we think FDA was well within its powers to find that the device presents a potential unreasonable risk of illness or injury and thus constitutes a Class III device. Ill The petitioners alleged a number of procedural violations by the FDA. We hold with respect to each alleged violation that either the FDA acted within the bounds of its discretion or that any procedural error is harmless given the strength of the FDA’s factual findings. General Medical alleges that the classification panel which assists the Secretary in reclassification, see 21 U.S.C. § 360c(b)(l) (1982), was insufficiently expert to evaluate the Drionic device. Section 360b(a)(2) provides that the Secretary shall appoint to such panels persons who are qualified by training and experience to evaluate the safety and effectiveness of the devices to be referred to the panel and who, to the extent feasible, possess skill in the use of, or experience in the development, manufacture, or utilization of, such devices. The Secretary shall make appointments to each panel so that each panel shall consist of members with adequately diversified expertise in such fields as clinical and administrative medicine, engineering, biological and physical sciences and other related professions. 21 U.S.C. § 360c(b)(2) (1982). General Medical requested that its petition for reclassification be heard by the General and Plastic Surgery Section of the FDA’s classification panels. When this panel evaluated the Drionic device, its members were a pathologist, a professor of surgery, and a professor of medicine and surgery. General Medical asserts that the panel’s composition therefore violated the mandate of 21 U.S.C. § 360c(b)(2) (1982). We disagree. In addition to its full members, the panel employed as consultants a dermatologist and a neurologist, specialties of some bearing upon problems of perspiration. Furthermore, the relevant valid scientific evidence was flawed in a manner perceptible to any physician familiar with validation studies. Indeed, even a court can understand that high attrition rates and the failure to explain away possible placebo effects undermine a study’s conclusions. The panel was sufficiently expert to comprehend and analyze the evidence relevant *224 to the FDA’s determinations about the device, and we therefore have no criticism of the FDA’s procedures in this aspect. The FDA provided a summary of relevant scientific literature and of the law to the panel members. See 21 C.F.R. § 14.33(c) & (f) (1985) (authorizing such communications). General Medical characterizes this summary as an illegal, ex parte communication. We note first that the panel members are quasi-governmental employees serving as advisors to the agency, see 21 U.S.C. § 360c(b)(3) (1982), and that ex parte communication between them and the agency is not statutorily proscribed except to the extent that it may produce a violation of the statutory requirement that the panels “provide an opportunity for interested persons to submit data and views,” 21 U.S.C. § 360c(c)(l) (1982). Petitioner may be correct, however, that the failure to provide it a copy of the communication before the panel meeting violated the provisions of the FDA’s rules, see 21 C.F.R. § 14.75(a) (1985). In the circumstances of this case, we cannot find that this failure produced prejudicial error, see 5 U.S.C. § 706 (1982). This is not a case where the alleged error in the scientific information was so blatant as to justify the expectation that a rebuttal would have altered the panel’s decision, cf. Portland Cement Association v. Ruckelshaus, 486 F.2d 375, 392-93 (D.C.Cir.1973), cert. denied, 417 U.S. 921, 94 S.Ct. 2628, 41 L.Ed.2d 226 (1974). And the alleged error in the summary of the law was overcome by the fact that the committee had the full text of the statute before it at the time of its deliberations. Moreover, the committee’s recommendation was in the last analysis only that — a recommendation — and petitioner had the opportunity, which it twice took advantage of, to point out the alleged errors to the FDA itself. Under the circumstances, we think whatever procedural deficiency existed was harmless. The petitioner’s final, and most meritorious, procedural complaint involves the transmission of the classification panel’s recommendation to the proper full-time governmental official for a final decision. The Amendments specify: A panel to which a petition has been referred shall not later than ninety days after the referral of the petition make a recommendation to the Secretary respecting approval or denial of the petition. Any such recommendation shall contain (I) a summary of the reasons for the recommendation, (II) a summary of the data upon which the recommendation is based, and (III) an identification of the risks to health (if any) presented by the device with respect to which the petition was filed. 21 U.S.C. § 360c(f)(2)(B)(i) (1982); see also 21 C.F.R. § 860.134(b)(4) (1985) (adopting same criteria for use in reclassification proceeding). The FDA essentially concedes that the panel has made no such recommendation. It informs us rather that, as a matter of course, the FDA accepts as a substitute a written copy of the transcript of the panel’s proceedings. We think it clear that such a transcript is not what Congress had in mind when it enacted section 360c(f)(2)(B)(i). A “summary” is not a “transcript,” even if, as here, the participants are asked to summarize their views at the end of the proceedings. The Secretary, or a delegatee, could clearly benefit from a concise statement of the panel’s reasons; the panel could clearly benefit from being forced to convert its multiple, spoken impressions to a single, written equivalent. The decision-making process as a whole, and thus the public, could benefit as a result. In the case before us, however, the FDA’s decisions are based on sufficiently strong evidence that it would be pointless to require the panel to complete a summary and submit it to the proper official. We are confident that the outcome would be exactly the same, and that the FDA would be correct in so ruling. We therefore simply warn the FDA that future panels of this court may look less favorably upon the failure of the FDA to require its classification panels to complete a summary, or upon *225 the classification panel’s failure to do so within the time required by the statute. IV We hold that the FDA was within its broad discretion to conclude that General Medical has failed to meet either the burden of showing the Drionic device to be safe and effective or the burden of showing that the device does not present a potential unreasonable risk of illness or injury. We hold that, while Congress contemplated that not all devices in Class I would have to have been shown safe and effective, the FDA may require the Drionic device to be shown safe and effective before the device is reclassified into Class I. We rule that the FDA committed no procedural errors of consequence in this case, but warn the FDA that it would do well in the future to adhere more closely to Congress’s directive that the classification panel submit a collective, written summary of its views to the Secretary or her delegatee. So ordered. Footnotes [1] General Medical, as one might expect, is the least happy with the situation. General Medical asserts that the FDA’s "purported authority” for its action, 21 C.F.R. § 801.109 (1985), is completely unsuited for justifying the FDA’s limitation on the "substantial equivalence” of the Drionic device to the Fischer device, and that the FDA therefore has "no precedent and ... no authority” for its action. Petitioner’s Post-Argument Brief on the Question of Mootness at 4 (D.C.Cir. filed Jan. 7, 1984). We agree with General Medical that the regulation cited by the FDA in its letter of September 29th to petitioner’s counsel appears designed to do no more than set forth the labelling requirements for devices available to consumers only through prescription. We can see in the regulation no hints for those attempting to determine the substantial equivalence of one product to another. We are not therefore entirely comfortable with the FDA’s interpretation of "substantial equivalence." Congress did state that the safety and effectiveness of "a device” are to be determined with reference to the "conditions of use prescribed, recommended, or suggested in the labelling of the device.” See 21 U.S.C. § 360c(a)(2)(B) (1982). The FDA believes that this provision is sufficient to justify its subdivision, for purposes of determining "substantial equivalence,” of "a Drionic device” into "a Drionic device labelled for prescription use” and "a Drionic device labelled for over-the-counter use.” See Supplemental Brief for Respondents at 3 n. 4 (D.C.Cir. filed Jan. 9, 1985). The determinations of safety and effectiveness are clearly crucial considerations in the classification of devices, see 21 U.S.C. § 360c(a) (1982), but the FDA’s argument requires us to make a hop and a skip of faith from the safety-and-effectiveness provisions to the classification provisions to the substantial-equivalence provisions. Nonetheless, the FDA has broad discretion in implementing the definition of “substantial equivalence.” See Chevron USA, Inc. v. Natural Resources Defense Council, - U.S. -, 104 S.Ct. 2778, 2782-83, 81 L.Ed.2d 694 (1984). In addition, the Food and Drug Act should be construed to effectuate the overriding purpose of protecting the public health. See United States v. An Article of Drug ... Bacto-Unidisk, 394 U.S. 784, 798, 89 S.Ct. 1410, 1418, 22 L.Ed.2d 726 (1969); H.R.Rep. No. 853, 94th Cong., 2d Sess. 36 (1976) ("The Committee believes that the term ['substantially equivalent’] should be construed narrowly where necessary to assure the safety and effectiveness of a device____’’). We think that the FDA’s hesitancy in granting substantial equivalence to Drionic-as-over-the-counter-device is reasonable in this particular instance. In any case, General Medical was a party to the settlement agreement. It appears to have entered that agreement aware of the FDA’s position as set forth in the letter of September 29th. It appears to have accepted the proposition that interpretation of the settlement agreement is informed by that letter. It does not appear in this lawsuit to be challenging the legality of the FDA’s interpretation of "substantial equivalence.” Therefore, while we do not wish to constrain in any way the future analysis of the FDA’s substantial-equivalence policy by this or any other court, we could hardly do anything in this case but accept the FDA’s interpretation of "substantial equivalence.” Given the FDA’s interpretation of substantial-equivalence and given General Medical’s desire to market the Drionic device over the counter, we agee with the parties that General Medical has enough to gain from winning this lawsuit to make the case a live controversy even after the settlement agreement in the related lawsuit. See DeFunis v. Odegaard, 416 U.S. 312, 94 S.Ct. 1704, 40 L.Ed.2d 164 (1974); Flynt v. Weinberger, 762 F.2d 134 (D.C.Cir.1985). [2] General Medical claims that the FDA is holding its device to a higher standard of effectiveness than that enforced upon over-the-counter chemical antiperspirants. The FDA has proposed a rule allowing a manufacturer to show the effectiveness of a chemical antiperspirant by demonstrating a sweat reduction of at least 20% in at least 50% of the subjects. 43 Fed.Reg. 46,694, 46,717 (Oct. 10, 1978). Drionic notes that 88% of the subjects in its study at the *219 two-week mark experienced reductions in perspiration levels. As we discuss below, however, see infra pp. 219-220, the credibility of this statistic is dubious, both because of the study’s high attrition rate and because of the qualitative nature of the "print” measurement of perspiration levels. (The chemical antiperspirants must show reductions using a more precise, "gravimetric” method.) We note also that the rule with respect to chemical antiperspirants is not yet final. [3] The FDA requires manufacturers to show safety and effectiveness with "valid scientific evidence." 21 C.F.R. § 860.7(c)(1) (1985); see id. § 860.7(c)(2) (describing what constitutes valid scientific evidence). In addition to the clinical study, General Medical submitted a wide range of more anecdotal evidence on the Drionic device and other devices using iontophoresis. FDA discounted most of this evidence, a decision within its discretion. See CLMA v. FDA, at 600. [4] None of the identified complaints involved cardiac irregularities or serious electrical shock, two of the hazards associated with prior versions of iontophoretic devices. If none of the unidentified complaints involved such hazards, then the inventor of the Drionic device should be pleased that his conceptual innovation and hard work, cf. T. Edison, Life ch. 24 (1932) (“Genius is 1 per cent inspiration and ninety-nine per cent perspiration”), have paid off sufficiently to remove a class of serious hazards from the risks of iontophoresis.