Rhodia, Inc., Hess & Clark Division (Rho-dia) sought to amend approved new animal drug applications (NADA’s) in effect for three products containing the active drug furazolidone so as to include new sources of supply of the bulk drug in addition to the one named in the original NADA’s. The Director of the Bureau of Veterinary Medicine of the Food and Drug Administration (FDA) denied approval of Rhodia’s supplemental NADA’s. Finding the action arbitrary and capricious, we set aside the FDA order and remand for further proceedings.
I. BACKGROUND
Rhodia currently holds three approved NADA’s1 2for products used in the treat*1377ment of infectious and parasitic conditions in poultry and swine that contain the active ingredient furazolidone. At the time of Rhodia’s2 initial applications in the 1950’s, the Norwich Pharmacal Company held the patent on furazolidone and was listed by Rhodia in its applications as the sole source of supply of the bulk drug.
Evidence casting doubt on the safety of furazolidone prompted FDA to propose the withdrawal of approval of the drug for use in food-producing animals.3 On August 4, 1971, the agency published a notice of opportunity for hearing on the proposed withdrawal, alleging that furazolidone had been shown to cause tumors in laboratory animals, and that, in the absence of “appropriately sensitive” methods of establishing the absence of furazolidone residues from food produced by treated animals, the drug was no longer shown to be safe. 36 Fed.Reg. 14,342 (1971). Rhodia requested a hearing, but none was held, and no action was taken on that proposal to withdraw approval.
On May 13, 1976, the FDA withdrew its original notice of opportunity and issued a new notice. The new notice cited evidence that furazolidone induced cancer in laboratory animals and concluded that, in the absence of adequate, reliable and practicable methods of analysis to detect the drug’s presence in food derived from treated animals, approval of the drug must be withdrawn.4 41 Fed.Reg. 19,906, 19,907 (1976). Again, Rhodia requested a hearing; though FDA issued a notice of intent to hold a hearing, 42 Fed.Reg. 18,660 (1977), that hearing has yet to be held.
Rhodia first sought to supplement its NADA’s to add an alternative supplier of bulk furazolidone in December, 1972. The change was necessary because Norwich had terminated its supply agreement with Rho-dia. The Director of the Division of New Animal Drugs of the Bureau of Veterinary Medicine refused to approve the supplemental NADA’s. While acknowledging that the bulk drug to be supplied by the alternate supplier would meet the quality specifications for furazolidone, the Director stated that Rhodia’s applications would remain incomplete “pending FDA’s acceptance and validation of the more sensitive methods of analysis to establish the absence of the drug in food derived from treated animals.” J.A. at 4.
In reaching this decision, the Director evidently applied an FDA policy that approval of a supplemental NADA constituted an affirmation that all safety and effectiveness information contained in the NADA as a whole was scientifically accurate by current standards. Thus, submission of a supplemental NADA triggered a review by the FDA of all the safety and effectiveness data underlying the original application. As a result, approval of a supplemental NADA was precluded in situations like that of furazolidone, until questions of the safety of the drug were resolved. See 41 Fed. Reg. 50,003 (1976).
However, FDA policy in practice accepted some supplements, even though the validity of a NADA was under review, at least where safety would be enhanced by a supplement.5 Apparently Rhodia was of the *1378belief that the FDA would acquiesce in its change of suppliers, and put that change into effect. On March 8,1977, however, the Cincinnati district office of the FDA informed Rhodia that continued marketing of products containing furazolidone not supplied by Norwich (the supplier named in the approved NADA’s) was unlawful. Rhodia ceased marketing products containing fura-zolidone supplied by alternate suppliers and negotiated an agreement with Norwich to resume supplying it bulk furazolidone.
On March 11, 1977, Rhodia again submitted supplemental NADA’s naming alternate suppliers of bulk furazolidone. The Acting Chief of the Division of Drugs for Avian Species in the Bureau of Veterinary Medicine refused to approve the supplements. On May 20, 1977, following protest by Rhodia, the Director of the Bureau affirmed.
The Director invoked a revised FDA policy on supplemental NADA’s, adopted by the agency on November 12, 1976. 41 Fed. Reg. 50,003 (1976). Under that revised policy, approval of a supplemental NADA was permitted without a full safety and effectiveness review if approval of the supplement would not “entail any increase in potential risk of human exposure to drug residues.”6 Pointing to the pending proceedings to withdraw approval of furazolidone as carcinogenic, the Director noted that the FDA had to be “especially alert” to increases in potential risk until the safety questions were resolved. In essence, the Director found that to add new approved sources of supply would likely increase the aggregate quantity of furazolidone marketed and consumed, with resultant increase in the potential risk of human exposure to its residues.7 J.A. at 15 — 17.
Rhodia asks this court to set aside the order as “plainly unlawful.”
II. ANALYSIS
We find it unnecessary to decide whether the FDA is within the law in adopting a policy that requires a full safety and effectiveness review before a supplemental NADA may be approved. Assuming that it has such authority, we find the FDA’s application of its policy to deny approval of Rhodia’s supplemental NADA’s to be arbitrary and capricious.
*1379The Director of the Bureau of Veterinary Medicine concluded that Rhodia’s proposed changes to add approved suppliers of bulk furazolidone would entail an “increase in potential risk of human exposure to drug residues.” The theory was not that the changes would affect the quality of the new animal drug in such a way as to call into question the drug’s safety, but rather that safety was threatened because the addition of suppliers might result in an increase in the quantity of furazolidone on the market.
There is no indication that the FDA has previously considered changes in a NADA that will increase available quantities of a new animal drug to bear on the safety of the drug. Indeed, the contrary appears to be the case. The FDA’s own regulations affirmatively exclude from the approval requirement certain NADA changes that appear to possess a potential for increasing quantities on the market similar to, or even greater than, that of changes in approved suppliers. Changes in equipment, including the enlargement of plant capacity that do not alter the method of manufacture, and changes in commercial batch size, are permitted without the submission or approval of a supplemental NADA. 21 C.F.R. § 514.-8(a)(5)(iii), (iv) (1978). An applicant may add new distributors of an approved drug automatically upon submission of a supplemental NADA for which no approval is required. Id., § 514.8(a)(6). Further, there is no check on the ability of a supplier of an applicant’s bulk drug to increase its manufacturing capacity (and thus the quantity of drug available to the applicant).
Furthermore, Rhodia asserts that the FDA’s regulations do not require that bulk suppliers be named in the NADA at all, but only that the NADA state that materials of specified quality or purity will be used. If so, the FDA’s control over Rhodia’s supplemental NADA’s would stem from a fortuity that Norwich was named as supplier in the original NADA’s, a circumstance probably ascribable to the fact that Norwich was at the time the sole supplier of bulk furazoli-done. The FDA disputes Rhodia’s construction of the applicable regulations. We do not parse the regulations to determine which position is correct, for the FDA does not refute Rhodia’s assertion that it has in practice approved NADA’s that do not specify a particular bulk drug supplier, but merely state that the ingredients obtained, from whatever source, will meet prescribed standards.
The FDA has the authority, and it has the responsibility, to define those changes that bear on safety so as to invoke the full safety and effectiveness review required by its supplemental NADA policy. Thus far, the FDA has not structured its regulations to define quantity as a factor triggering invocation of the safety review. Once it channels its discretion in a certain manner (here by excluding from the approval requirement certain types of NADA’s, and by a practice of approving original NADA’s that do not name a specific supplier), the agency should follow that course consistently or articulate reasons for departure. In view of its previous course, bypassing quantity as a determinative criterion, the FDA may not now latch onto this factor as the basis for rejecting an otherwise-unobjectionable supplemental NADA because of pre-existing doubts as to the safety of the product. The FDA did not offer, nor can we discern from the record, any rational basis for distinguishing the proposed change of suppliers here from other changes that would result in increased quantities on the market or from the FDA’s earlier approval of supplemental NADA’s not naming a specific supplier. Therefore we vacate the FDA order denying the supplemental application.
So ordered.
additional statement:
On several occasions I have appended concurring opinions to majority opinions I have written for the court.1 I have found *1380this practice useful for observations not essential to the judgment.
Speaking for myself alone, I would have preferred to rule on the merits of the issue whether the FDA has statutory authority to require a full safety and effectiveness review of the underlying NADA as a condition for approval of a supplemental NADA. Practically, the FDA’s policy appears to foreclose approval of a supplemental NADA where questions of the drug’s safety exist, and where the proposed change will “entail any increase in the potential risk of human exposure to drug residues.” While discussion of this question is not essential to a holding that the FDA’s action is invalid, in the absence of such discussion, the FDA would be entitled to conclude that it could lawfully implement the objective it has already announced and sought to carry out in its conception of the public interest, by means of modification of its approach and regulations — so as to bring changes bearing on quantity within the supplemental NADA requirements and the supplemental safety review policy.
In my view, the case is ripe for deciding the validity of the FDA interpretation of its statutory authority. The appellant’s challenge to that claim of authority is plainly before the court; the point is not academic or moot; and it has been fully briefed and argued. If appellant is correct, then a decision to that effect would obviate a futile expenditure of time and energy.2
The question is not an easy one, and in general I agree that courts should not undertake to decide difficult questions not necessary to be decided. See United States v. Hooper, 139 U.S.App.D.C. 171, 432 F.2d 604 (1970). But insofar as that is a rule of economy of judicial administration, it is offset by the interest of economy of agency administration. And here there is no problem of unnecessary decision of constitutional issues.
On the merits of the issue, I suggest that the FDA construction is reasonable,3 and consistent with the language of § 512(e)(1)(E) of the Food, Drug and Cosmetic Act (the Act), which provides the primary source of FDA authority to approve supplemental NADA’s: “The supplemental application shall be treated in the same manner as the original application.” 21 U.S.C. § 360b(e)(l)(E) (1976).
The Act is to be construed liberally to effectuate its overriding purpose to protect the public health. United States v. An Article of Drug . . . Bacto-Unidisk, 394 U.S. 784, 798, 89 S.Ct. 1410, 22 L.Ed.2d 726 (1969). The FDA’s broad mandate to safeguard the public health thus affords it the flexibility to shape its administrative actions when it has reason to doubt the safety of a new animal drug. For example, it may have proper grounds for denial of a supplement because of doubts even though these doubts may not amount to the “new evidence” required for withdrawal of approval. Similarly, the law permits the view that concerns may exist that justify the FDA in declining to lend its approval to a strengthening of the NADA, even though the FDA is not yet warranted in terminating the NADA.
Rhodia contends that rejection of a supplemental NADA permits backhanded withdrawal of approval. That may present problems, but in this case at least, Rhodia *1381has not demonstrated that this is the situation. Rhodia has been able to regain Norwich as its source of supply, though apparently at a higher price, under the original NADA.
There are bounds to the FDA’s power to require a full safety and effectiveness review. One cannot discern the basis for such an authority where the application demonstrates that the change has no bearing whatever on issues of safety or effectiveness.4