This matter comes on before this court on appeal from an order entered August 27, 2002, denying appellant Shire U.S. Inc.'s ("Shire") application for a preliminary injunction in its case against Barr Laboratories, Inc. ("Barr") charging Barr with trade dress infringement and trade dress dilution under sections 43(a) and (c) of the Lanham Act, 15 U.S.C. §§ 1125(a) and (c), and state unfair competition laws with respect to Shire's rights in Adderall, an unpatented drug Shire manufactures and sells. Adderall is a central nervous system stimulant used in treating attention deficit hyperactivity disorder (ADHD) available only by prescription and dispensed to patients in pharmacy vials labeled "prescription-only" as required by law. Adderall is composed of the mixed salts of a single-entity amphetamine and is a controlled substance. Shire first placed Adderall on the market in 1996 and since that time it has enjoyed substantial success so that by 2001 it had a 32% market share in the United States ADHD prescription market.1

Adderall originally came in two dosage strengths and colors, 10 mg. (blue, round) and 20 mg. (orange, round). The tablets are currently either blue or pale orange/peach and either round or oval. Color and size vary with the tablet's strength, seven of which currently are prescribed: 5 mg. (blue, round), 7.5 mg. (blue, oval), 10 mg. (blue, round), 12.5 mg. (orange/peach, round), 15 mg. (orange/peach, oval), 20 mg. (orange/peach, round), and 30 mg. (orange/peach, round). Adderall tablets are scored and stamped with the mark "AD" on one side and the dosage size, e.g., "10" on the other.

Recently, Shire sought a trademark for the Adderall trade dress. Shire's trademark applications sought protection for the overall configuration of the round tablet shape and the colors blue and orange used in conjunction with the "AD" marking. The United States Patent and Trademark Office (USPTO) initially has refused to register Shire's configuration.2

Shire's product literature, promotional materials, and mailings, which its sales staff distributed to physicians, feature color pictures of the Adderall tablets and sometimes direct patients to examine the tablets to ensure that they have received exactly the drug prescribed. Shire does not advertise its products in general consumer publications, but pictures of Adderall tablets appear in the Physician's Desk Reference and in certain consumer books. While Shire continues to sell Adderall, it altered its marketing strategy for 2002 and discontinued promoting Adderall, promoting instead a patented, sustained-release version of the drug, Adderall XR.

Barr, a public company that develops and manufactures generic and proprietary pharmaceuticals, was the first manufacturer of a generic equivalent to Adderall. It began developing a generic amphetamine salt alternative in 1998 and started marketing it in February 2002 after submitting an Abbreviated New Drug Application ("ANDA") to the United States Food and Drug Administration ("FDA") and obtaining its approval. The FDA has approved Barr's generic amphetamine salts as safe and effective, and has classified Barr's product, which it manufactures in accordance with FDA regulations, as therapeutically equivalent to Adderall. Barr's product is the bioequivalent3 of Adderall, for which it thus may be interchanged freely. According to Shire, however, the products contain different inactive ingredients, and, in particular, Barr's tablets contain saccharin, a once controversial ingredient the FDA only recently removed from its list of banned substances.

Barr manufactures its generic amphetamine salts in 5 mg. (blue, oval), 10 mg. (blue, oval), 20 mg. (orange/peach, oval), and 30 mg. (orange/peach, oval) tablets.4 Barr's generic amphetamine salts are oval5 and convex in shape. Both the size and the color of Barr's tablets are linked to dosage. The face of the tablets has a "b" mark or the trade name Barr, and contains a numerical product code. The district court, on the basis of its physical examination of the tablets and the record before it, determined that while Barr's tablets, like Shire's, are blue and peach/light orange and those colors are keyed to dosage amounts, their shape and markings are different and "[j]uxtaposed against one another, the products are similar though not identical." JA 25.

On April 30, 2002, Shire filed this action against Barr, alleging that Barr's sale of generic amphetamine salts copying Adderall's appearance constituted unfair competition and diluted Shire's rights under federal and state law.6 On May 3, 2002, Shire filed a motion seeking a preliminary injunction precluding Barr's use of a tablet with an appearance similar to that associated with the appearance of Shire's Adderall tablets.

On June 3, 2002, after the parties had engaged in limited, agreed-upon discovery, the district court heard oral arguments on Shire's motion. In a comprehensive opinion dated August 26, 2002, the court denied Shire's motion on the grounds that Shire had not credibly carried its burden of establishing that the color and shape of Adderall is non-functional. Therefore the court concluded that Shire was not likely to succeed on the merits of its case. The court entered the order denying the motion for a preliminary injunction on August 27, 2002,7 but subsequently altered parts of its opinion by an amended opinion dated September 16, 2002, which corrected inaccuracies in certain terminology used in the original opinion.8 Shire filed a timely appeal on September 16, 2002.

We approach this case recognizing that when ruling on a motion for preliminary injunctive relief, a district court must be convinced that consideration of the four following factors favors the granting of preliminary relief: (1) the likelihood that the moving party will succeed on the merits; (2) the extent to which the moving party will suffer irreparable harm without injunctive relief; (3) the extent to which the nonmoving party will suffer irreparable harm if the injunction is issued; and (4) the public interest. See Clean Ocean Action v. York, 57 F.3d 328, 331 (3d Cir. 1995); AT & T Co. v. Winback and Conserve Program, Inc., 42 F.3d 1421, 1427 (3d Cir.1994). As we have indicated, the district court determined that Shire had failed to satisfy the first prong of the test for preliminary injunctive relief, finding that Shire was unlikely to succeed on the merits of its trade dress claim because it failed to demonstrate that the product configuration of Adderall is non-functional, which is one of the factors Shire needed to prove in order to prevail in an action for trade dress infringement.9

On this appeal we recognize that "[t]he decision whether to enter a preliminary injunction is committed to the sound discretion of the trial court, and will be reversed `only if the court abused its discretion, committed an obvious error in applying the law, or made a serious mistake in considering the proof.'" Duraco Prods., Inc. v. Joy Plastic Enters., 40 F.3d 1431, 1438 (3d Cir.1994) (quoting Loretangeli v. Critelli, 853 F.2d 186, 193 (3d Cir.1988)). However, "[a]lthough terms of an injunction are normally reviewed for abuse of discretion, any determination that is a prerequisite to the issuance of an injunction ... is reviewed according to the standard applicable to that particular determination." John F. Harkins Co. v. Waldinger Corp., 796 F.2d 657, 658 (3d Cir.1986); see AT & T v. Winback and Conserve, 42 F.3d at 1427. Thus, we review the district court's factual determinations for clear error but exercise plenary review with respect to its legal conclusions. See A & H Sportswear, Inc. v. Victoria's Secret Stores, Inc., 237 F.3d 198, 210 (3d Cir.2000). Under the clearly erroneous standard, "a finding of fact may be reversed on appeal only if it is completely devoid of a credible evidentiary basis or bears no rational relationship to the supporting data." American Home Prods. Corp. v. Barr Labs., Inc., 834 F.2d 368, 370-71 (3d Cir.1987).10

The Lanham Act, 15 U.S.C. § 1125(a),11 establishes a cause of action for trade dress infringement. "Trade dress" refers to the design or packaging of a product which serves to identify the product's source. See TrafFix Devices, Inc. v. Marketing Displays, Inc., 532 U.S. 23, 28, 121 S. Ct. 1255, 1259, 149 L. Ed. 2d 164 (2001). The purpose of trade dress protection, like trademark protection, is to "secure the owner of the [trade dress] the goodwill of his business and to protect the ability of consumers to distinguish among competing producers." Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 774, 112 S. Ct. 2753, 2760, 120 L. Ed. 2d 615 (1992) (citation omitted).

Trade dress protection, however, is not intended to create patent-like rights in innovative aspects of product design. Eppendorf Netheler Hinz GmbH v. Ritter GmbH, 289 F.3d 351, 355 (5th Cir.2002). Thus, trade dress protection, unlike patent law which is not implicated here, does not foster innovation by preventing reverse engineering or copying of innovative product design features. Id. "Trade dress protection must subsist with the recognition that in many instances there is no prohibition against copying goods and products." TrafFix, 523 U.S. at 29, 121 S.Ct. at 1260. Therefore, trade dress protection extends only to incidental, arbitrary or ornamental product features which identify the product's source. Eppendorf, 289 F.3d at 355. To establish infringement of its unregistered trade dress a plaintiff must prove that (1) the allegedly infringing feature is non-functional, (2) the feature is inherently distinctive or has acquired secondary meaning, and (3) consumers are likely to confuse the source of the plaintiff's product with that of the defendant's product. See Wal-Mart Stores, Inc. v. Samara Bros., Inc., 529 U.S. 205, 210-211, 120 S. Ct. 1339, 1343, 146 L. Ed. 2d 182 (2000).

The functionality doctrine "accommodates the twin purposes behind the Lanham Act. It protects the manufacturer (and the consumer) from the copying of those features that signify a product's source (and quality) and encourages competition by preventing one manufacturer from acquiring a monopoly by attempting to trademark those features of a design essential to a successful product of that type." Standard Terry Mills, Inc. v. Shen Mfg. Co., 803 F.2d 778, 780-81 (3d Cir. 1986). However, the definition of "functionality" has not enjoyed such clarity. See 1 J. THOMAS MCCARTHY, MCCARTHY ON TRADEMARKS AND UNFAIR COMPETITION, § 7:67 (4th ed.2001) (discussing the "plethora of definitions" for functionality). In TrafFix, the Supreme Court set forth two tests for functionality. First, the Court recognized the "traditional" definition of functionality: "a product feature is functional, and cannot serve as a trademark, `if it is essential to the use or purpose of the article or if it affects the cost or quality of the article.'" TrafFix, 532 U.S. at 32, 121 S.Ct. at 1261 (quoting Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 165, 115 S. Ct. 1300, 1304, 131 L. Ed. 2d 248 (1995) (quoting Inwood Labs., Inc. v. Ives Labs., Inc., 456 U.S. 844, 850, n. 10, 102 S. Ct. 2182, 2187 n. 10, 72 L. Ed. 2d 606 (1982))). In addition to the traditional definition, TrafFix recognized a second test for functionality: "a functional feature is one the `exclusive use of [which] would put competitors at a significant non-reputation-related disadvantage.'" Id. (quoting Qualitex, 514 U.S. at 165, 115 S.Ct. at 1304).

The district court, applying the "traditional" definition of functionality12 found that "[i]n reviewing the proofs, the Court is not persuaded that the color and shape of Shire's product is non-functional."13 JA 24. Preliminarily the court noted that, contrary to Shire's contention, Barr did not adopt the "identical trade dress developed by Shire for its Adderall product." Id. After an examination of the actual tablets the court found that the products have different shapes and markings and therefore while the products are similar, they are not identical.

In addition, the district court found that "the record developed for purposes of this injunctive relief application fails to support the conclusion that Shire's alleged trade dress is non-functional" inasmuch as Shire "falls short on its burden to prove non-functionality because it has not credibly rebutted Barr's theory that the similar color-coding and shape of the products are particularly meaningful for ADHD patients and enhance efficacy." JA 25-26. In reaching its conclusion the district court quoted the declaration of Lawson F. Bernstein, M.D., the affidavit of Cheryl D. Blume, Ph.D., and the declaration of Gregory Drew.

Dr. Bernstein's declaration explains, inter alia, that because ADHD patients overuse visual cues, (1) when therapeutically equivalent ADHD products have similar visual recognition properties, adult ADHD patients will experience less confusion in correctly identifying the agent and/or its dosage strength; (2) given that almost all patients require some initial dosage titration and a subsequent substantial majority require intermittent dosage adjustment, the color coding of a particular preparation of mixed amphetamine salts tablets confers a substantial degree of clinical functionality for the patient in the titration/adjustment process; (3) many adult patients may take multiple daily dosages of different strength amphetamine salts tablets, also inferring the usefulness of similar color-coding.

The district court quoted Dr. Blume's affidavit explaining that a generic drug's similar appearance to the branded product "enhance[s] patient safety and compliance with the medically prescribed dosing regimen" and that safety and compliance "would be particularly important for ADHD drugs when non-medical intermediaries (such as school secretaries) dispense mid-day doses to children [treated for ADHD]." JA 27. Dr. Blume's affidavit explains that "[d]osage form similarities enhance patient acceptance" and points to generic formulations of other central nervous system drugs that are identical or mirror the brand drug in color. Id.

The district court also noted that, concurring with both Dr. Bernstein and Dr. Blume, the declaration of Gregory Drew, a registered pharmacist and Vice President of Pharmacy Health Services for Rite Aid Corporation, explains that Rite Aid prefers that "the generic tablet look as similar to the branded tablet as possible" so as to "increase[] patient acceptance and comfort," as well as compliance and that "all other things being equal, Rite Aid will choose to stock the generic product that most closely resembles the branded product." Id.

The district court found that "Barr has not copied features signifying the Adderall source" and that "the color and shape of Barr's product are directly linked to the drug's efficacy in ADHD patients." JA 28. The court concluded that "Shire has failed to prove the requisite non-functionality as an initial step to gaining the desired relief here." Id. The court's functionality conclusion is a factual finding, see CIBA-GEIGY Corp. v. Bolar Pharm. Co., Inc., 747 F.2d 844, 850 (3d Cir.1984), that "may be reversed on appeal only if it is completely devoid of a credible evidentiary basis or bears no rational relationship to the supporting data." American Home Prods., 834 F.2d at 370-71.

Shire's primary argument is that the district court "committed an obvious error by failing to apply Third Circuit precedent which is contrary to the district court's ruling." Br. of appellant at 21. But as Barr points out, most of opinions on which Shire relies were district court opinions from the early 1980s14 which the court here was not bound to follow. Br. of appellee at 35. In addition, even when Shire's district court case law is considered for its persuasive value, it does not support its contention that the court committed reversible error as a court must decide each trademark case on its own facts, Dresser Indus., Inc. v. Heraeus Engelhard Vacuum, Inc., 395 F.2d 457, 461 (3d Cir. 1968), and the cases on which Shire relies are distinguishable on their facts. Most significantly, though the cases involved prescription drugs, none involved controlled substances15 and in all of the cases there was evidence of the passing off of the defendant's product by pharmacists, see Boehringer Ingelheim G.m.b.H. v. Pharmadyne Labs., 532 F. Supp. 1040 (D.N.J.1980); Hoffman La Roche, Inc. v. Premo Pharm. Labs., Inc., 210 U.S.P.Q. 374 (D.N.J.1980); Biocraft Labs., Inc. v. Merck & Co., 532 F. Supp. 1068 (D.N.J. 1980); American Home Prods. Corp. v. Chelsea Labs., Inc., 572 F. Supp. 278 (D.N.J.1982), aff'd, 722 F.2d 730 (3d Cir. 1983) (table), or of an intent to induce illegal substitution on the defendant's part. See Ciba-Geigy Corp. v. Bolar Pharm. Co., 547 F. Supp. 1095 (D.N.J.1982), aff'd, 719 F.2d 56 (3d Cir.1983).16 Shire also relies on our opinion in SK&F, Co. v. Premo Pharmaceutical Laboratories, Inc., 625 F.2d 1055 (3d Cir.1980), but SK&F also is distinguishable on its facts as in SK&F we found evidence of actual passing off by pharmacists. Id. at 1063. Shire does not make a comparable claim in this case.17

It is true that in SK&F and several of the other cases on which Shire relies, the defendant offered affidavits and declarations of pharmacists and physicians making claims relating to functionality similar to those Dr. Bernstein, Dr. Blume and Mr. Drew have made in this case and that the courts in those cases did not credit the evidence. For example, in SK&F Dr. Shafer, a physician, submitted an affidavit in which he supported the sale of similarly configured generic tablets as he believed this configuration would enable the patient to feel confident that there was no change in the chemistry of the medication and that patients might become uneasy, confused or react adversely if the generic medication looked different from the market innovator. Id. at 1060-61. But we explained that the district court nevertheless "apparently chose not to credit the assertion of the Shafer affidavit, crediting instead the affidavits of Drs. Meyerson and Tannenbaum that in their experience the appearance of a drug bears no established relationship to its therapeutic efficacy." Id. at 1061. SK&F does not dictate, however, that we overturn the district court's findings of fact in this case because, in contrast to the district court's conclusions in SK&F, the district court in this case did credit the assertions made by Drs. Bernstein and Blume and Mr. Drew that were comparable to those of Dr. Shafer. Just as in SK&F we deferred to the district court's findings of fact it is appropriate for us to do so in this case as well.18 After all, it is inherent in the very nature of the deferential appellate review of findings of fact that a court of appeals can and, indeed, should, depending on the records before it, uphold arguably inconsistent outcomes.

We also point out that while Shire is correct that the district court did not discuss most of the case law on which it relies, it concedes that the court did not ignore all precedents in its favor. Thus, after quoting at length from Dr. Bernstein's declaration explaining that there is clinical functionality when a generic drug is physically similar to the branded version, testimony which the district court credited, the court cited our opinion in CIBA-GEIGY, 747 F.3d at 850-51, contrasting it as a case "affirming finding of non-functionality where district court found no `medical or business considerations' compelled the size and shape of a drug." JA 26.

While district courts in this circuit have rejected functionality arguments similar to those the court credited in this case, other district courts, such as that in Ives Laboratories, Inc. v. Darby Drug Co., 488 F. Supp. 394 (E.D.N.Y.1980),19 have credited similar testimony bearing on functionality. In Ives the manufacturer of the prescription drug cyclandelate sought an injunction against manufacturers of generic cyclandelate claiming that the defendants' use of the same capsule colors was "a false designation of origin" or a "false description or representation" of defendants' product. Id. at 398. But the district court in Ives found that capsule colors were functional in several respects. "First, many elderly patients associate the appearance of their medication with its therapeutic effect.... Second, some patients co-mingle their drugs in a single container and then rely on the appearance of the drug to follow their doctors' instructions.... Third, to some limited extent color is also useful to doctors and hospital emergency rooms in identifying overdoses of drugs." Id. at 398-99.20

The Supreme Court noted the district court's finding of functionality in Ives in Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844, 853, 102 S. Ct. 2182, 2188, 72 L. Ed. 2d 606 (1982), after it granted certiorari and reversed the judgment of the court of appeals which had reversed the district court.21 In addition, citing its opinion in Inwood v. Ives in Qualitex, the Court commented on the functional nature of the color of medical pills stating:

[T]his Court has written that competitors might be free to copy the color of a medical pill where the color serves to identify the kind of medicine (e.g., a type of blood medicine) in addition to its source. See [Inwood v. Ives] at 853, 858, n. 20, 102 S.Ct. at 2188, 2190 n. 20 (`Some patients commingle medications in a container and rely on color to differentiate one from another'); see also J. Ginsberg, D. Goldberg and A. Greenbaum, Trademark and Unfair Competition Law 194-195 (1991) (noting that drug color cases `have more to do with public health policy' regarding generic drug substitution `than with trademark law').

Qualitex, 514 U.S. at 169, 115 S.Ct. at 1306.

Here the district court credited Barr's evidence that similarity in tablet appearance enhances patient safety by promoting psychological acceptance. We reiterate that we recognize that district courts in this circuit have rejected arguments of functionality similar to those credited by the trial court in Ives and by the district court in this case, and that we have affirmed several of those factual determinations as not clearly erroneous. But, as we have said, those cases are distinguishable on their facts. Moreover, we have the benefit of the Supreme Court's most recent trade dress decisions22 which caution against the over-extension of trade dress protection. In this regard we point out that there is language in several of the Court's opinions supporting a conclusion that the district courts in Ives and in this case did not err in their functionality conclusions. Overall we are satisfied that the district court, after it determined that by being physically similar to Adderall Barr's generic amphetamine salts tablets materially benefitted the patient population, did not clearly err in finding that Shire had failed to show that its product configuration was non-functional. Consequently, Shire did not demonstrate its likelihood of success on the merits and the court appropriately denied it a preliminary injunction.

For the foregoing reasons the order of August 27, 2002, will be affirmed.