I. Introduction

Wikipedia makes it sound simple:

The FDA requires that OTC products are labeled with an approved Drug Facts label to educate consumers about their medications. These labels comply to a standard format and are intended to be easy for typical consumers to understand.¹

Practical experience shows, though, that over-the-counter labeling is anything but simple: What information must be on the label? Should the label include additional information? Who qualifies as a “typical consumer?” How does one measure ease of comprehension? Those designing and seeking approval for an OTC label have to tackle not only these questions, but also many others implicit in the requirements of the Act, rules and regulations, as well as the many sub-issues each question generates.

The Federal Food, Drug, and Cosmetic Act requires that a label be written “in such terms as to render it likely to be read and understood by the ordinary individual under customary conditions of purchase and use.”² Under the auspices of this and other mandates, the FDA has sought for decades to improve consumers’ understanding of OTC labels. Since the 1970s and before, the government has crafted rules and regulations with the intention of increasing comprehension and compliance. The elusive goal of designing an OTC label to achieve the optimal level of understanding remains a focus of the FDA. This article discusses the 2006 Nonprescription Drugs Advisory Committee (NDAC) meeting that addressed these OTC label issues and the 2009 Draft Guidance to Industry issued afterwards.

II. The Regulations

A. The Early Years

It is ironic that industry is to derive its guidance for crafting a pharmaceutical product’s label — at least in some part — from the Federal Register, in the words of Dr. Eric Brass,³ “the single most unreadable thing in history.”⁴ Nevertheless, industry must navigate a labyrinth of governmental pronouncements when drafting an OTC label and subsequently seeking FDA approval.

In 1972, the FDA set out that the consumer should, “upon reading the label, […] be able to determine the uses for the drug, any warning against use, and any other pertinent information which will allow him to use the drug adequately.”⁵ The first two “determinations” impact accurate consumer self-selection. The third — other information necessary for adequate drug use — is the obvious catch-all that impacts every­thing else.

Over time, the FDA attempted to provide industry additional guidance. For instance, in 1974, the agency pronounced that “the purpose of OTC medication is to permit consumers to engage in self-medication without medical or other professional supervision, or in any event with the least amount of supervision feasible.”⁶ In essence, if consumers are to self-medicate safely without the input of a traditional learned intermediary such as a physician, then the label must function as a learned intermediary.⁷ Even then, the FDA recognized the limits of the OTC label:

[I]f labeling contains too many required statements, especially general statements of common sense, the impact of all warning statements on the label will be reduced. In addition, there is a space limitation on the number of statements that can appear on the labeling.⁸

B. Drug Facts

Twenty years later, in 1996, and still focusing on complexity, the FDA proposed a rule to provide “more simplified and understandable information” to consumers.⁹ The following year, the FDA noted once more that consumers were having difficulty understanding information contained in OTC labeling, and the agency proposed still another rule to improve comprehension by simplifying and standardizing labels.¹⁰ This 1997 proposed rule eventually became the 1999 Drug Facts regulation that governs OTC labels today.¹¹

One impetus for this renewed activity in the late 90s was studies indicating that consumers simply were not comprehending the information contained in OTC labels at levels acceptable to the FDA.¹² The agency described labels as “often printed in small type with a crowded layout and minimal white space.”¹³ As a result, it conducted two studies that looked at different formats and determined that OTC labels should “use […] more concise and easy to understand language,”¹⁴ as well as uniform headings and format.

[R]esearch on reading behavior and document simplification shows that the use of less complex terminology, presented in shorter sentences with an organized or ‘chunked’ structure, is likely to improve consumer processing of the information […]. Research also shows that consumers are more likely to engage in behavior that they believe they can successfully complete than in behavior that appears overwhelming […] or that presents a ‘cognitive load,’ such as the task of reading densely worded consumer information […].¹⁵

The FDA then set out specific templates for different pharmaceutical products, noting again that somehow all this important information must fit on the outside of the container.¹⁶ The story, though, did not end in 1999.

III. Label Comprehension Studies

A. 2006 Nonprescription Drugs Advisory Committee Meeting

The focal topic of the 2006 NDAC meeting was the “design, analysis, and interpretation of consumer behavior studies used to support OTC switch applications.”¹⁷ Because a switch typically does not involve pharmacology as the main point of interest, the more important question is a behavioral one: “How will consumers use the drug in the OTC environment and ultimately, will consumer behavior lead to safe and effective use?”¹⁸ The answer requires more than intuition.

In general, as noted by FDA participant Dr. Andrea Leonard-Segal, three types of studies are used to attempt to predict consumer behavior in the OTC arena: label comprehension studies (LCS), self-selection studies (SSS), and actual use studies (AUS).¹⁹ The Committee discussed different aspects of each type, offered suggestions for conducting better studies, and posed questions that should be addressed in the future.

1. Self-Selection Studies

“The purpose of a self-selection study is to determine if the consumer can correctly decide whether or not the product is appropriate for him or her to use based on the label information.”²⁰ Thus, as the starting point, information required for accurate self-selection must appear on the Drug Facts label.²¹ It is not helpful if information critical to a self-selection decision is tucked away inside a package insert and unavailable at the point of purchase.²²

This kind of study may be conducted as a stand-alone endeavor, or as part of either an LCS or AUS.²³ One question arising with SSS is whether participants who incorrectly say the product is not right for them to use should be considered when determining the success of the study. After all, if the participant is not going to use the medicine, he or she will not be harmed by the medicine.²⁴ How these participants’ responses will be analyzed (e.g., counted as incorrect or simply disregarded) must be ironed out with the FDA beforehand as the responses could skew the numbers and result in a failed study.²⁵ Another issue to consider is whether “correct” self-selectors should answer additional questions to verify the accuracy of their decision — or whether the sponsor should bear the expense of independent verification.²⁶

A different issue arises with a subpopulation at particular risk if members incorrectly self-select. Because the risk is higher for this subpopulation, the FDA may consider setting a higher threshold for compliance than that for the general population. The question then becomes: “When should the majority who could benefit from access to an OTC drug be denied that access because of self-selection errors made by a subpopulation at risk from drug use?”²⁷

2. Label Comprehension Studies LCS have two main purposes:

One is to test how the label communicates information to the consumer, and the other […is] to test the ability of the consumer to apply label information in hypothetical settings in which the drug should or should not be used.²⁸

These studies typically allow the participant to have a copy of the label while answering questions either from a questionnaire or an interviewer.²⁹ They generally enroll about 300 subjects of normal literacy and 150 of low literacy.³⁰ Difficulties arise when the answers are “partly” correct (e.g., the correct answer is “stop use and ask a doctor,” but the participant responds with “call a doctor”).³¹ Agreement on how to categorize such answers beforehand could be critical in rating the success of the study.

LCS should be conducted only after the sponsor and the agency agree on the critical information. Otherwise, studies indicate that information simply will overwhelm people.³² As one participant put it, “Somebody has got to define what are the three things that every single person who takes this medicine must understand.”³³

3. Actual Use Studies AUS mimic the OTC use of a product.³⁴

We sample consumers who we hope will represent the OTC population, people who are interested in the treatment, but who haven’t been screened for medical suitability, and then finally, we essentially step back and let the consumer make the decisions about buying the drug, about using it, whether to buy it again, when to discontinue or stop […].³⁵

This type of study can become quite difficult, depending on whether the medicine is indicated for short term or chronic use.³⁶ Analyzing compliance also is a tricky issue: What is the threshold for overuse, underuse, long-term adherence for chronic indications?³⁷

With respect to whether an actual use study “trumps” a comprehension study, the Committee consensus seemed to be that AUS are more important.³⁸ The point of the LCS, though, is that a “well-designed label comprehension program is in the sponsor’s best interest to increase the likelihood of success of an actual use study.”³⁹

4. Miscellaneous Comments of the Committee

In addition to discussing these three types of consumer behavior studies, the Committee opined on a number of additional issues. Some of the ideas broached during this meeting could have lasting consequences for industry— both as to the initial outlay that may be required to support a switch and the impact that implementation may have on litigation.

• Publish study results

Several participants questioned why the results of a sponsor’s consumer behavior study should not be made public.⁴⁰ This idea of publication paralleled the oft-mentioned idea that consumer behavior studies have not been — but should be — an iterative process.⁴¹ The rationale for publication: If the good and not-so-good studies are made available, the information will be valuable as a learning tool. In other words, industry should be designing better and better studies by utilizing “good” concepts from earlier studies and discarding “bad” concepts ⁴²:

[…W]e have never had an iterative process that has carried us forward from one experience to another, so that each time we see a study we are comparing it to something that went before, and we have gradually improved our performance as we do this.⁴³

• Objectives set out beforehand

A common theme throughout the meeting was that the objectives of the study (i.e., what information was being evaluated) and the goal of the study (i.e., what percentage of comprehension will indicate “success”) need to be agreed upon with FDA prior to the study, and that there must be consequences if objectives and goals are unmet.⁴⁴ Inasmuch as the FDA is making the risk-benefit decision, one participant noted that there must be agreement on the “substantial risk”:

This means the sponsors and the agency have to agree explicitly and beforehand what the core issues are for the OTC switch for real risk.⁴⁵

Studies then are designed around that risk. The more substantial the risk, the higher the level of necessary compliance.⁴⁶ In other words, there are no cookie-cutter study designs or set-in-stone benchmarks that apply to every switch, but each is dependent on the particulars of the medicine.⁴⁷

• Low literacy

Participants discussed at length the realities of literacy. As an initial matter, the average American reads at an 8th grade level but comprehends at a 6th grade level.⁴⁸ In other words, if consumers of low literacy are to be tested, the study must include subjects below those levels. One participant questioned whether labels even could be designed for lower literacy consumers.⁴⁹ Others questioned whether the target comprehension levels should be different for those of low literacy (e.g., 90% for average, but 70% for low).⁵⁰

• Other sources of information

One participant, after noting that reliance generally is placed almost exclusively on actual-use studies, suggested looking at other sources of information.⁵¹ For instance, information obtained from the prescription environment might be useful. Looking at patterns of use from other countries where the drug has been approved (even if the regulatory system requires behind-the-counter use) could provide insight. And, finally, a participant posited reviewing information from similar OTC products already on the market — post-marketing surveillance.⁵²

B. 2009 Draft Guidance for Industry: Label Comprehension Studies for Nonprescription Drug Products

With those comments from the NDAC meeting in mind, fast forward three years. The FDA’s April 2009 Draft Guidance “is intended to provide recommendations to industry on conducting label comprehension studies.”⁵³ The Draft Guidance provides no recommendations for the other types of study discussed at the 2006 NDAC meeting: self-selection studies and actual-use studies. That omission is notable inasmuch as the FDA deemed it “important” to advise the reader that the data gleaned from LCS is no predictor of actual consumer behavior.⁵⁴ In other words, consumer comprehension does not equate to consumer compliance. Despite this recognition, the FDA suggests that a sponsor conduct such LCS before conducting actual-use studies to determine whether “literate and low literate individuals can understand a drug product label.”⁵⁵

The Draft Guidance also sets out circumstances in which conducting LCS may not be a suggestion, but a requirement:

• Switches
• Substantive SNDAs, such as new indications
• New warnings
• Potential confusion over brand names and/or active ingredients
• Inclusion of a package insert⁵⁶

1. Study Design

As repeatedly stressed during the NDAC meeting, the Draft Guidance sets out that the study’s objectives must be established before the study begins. Primary communication objectives — information deemed most important to the safe and effective use of the product — should have a 90% or greater level of comprehension. These objectives can include understanding of indications, contraindications, or dose. Secondary communication objectives should have a target comprehension level of 80% or more. Self-selection may be tested as one of the communication objectives, or it could be tested in a separate study.

The Draft Guidance provides that the study “should include all subjects who could potentially use the drug product.”⁵⁷ Since the study participants will not ingest the medicine, exclusions should be rare and justified. In addition to testing comprehension in the general population, the study could be made more robust by targeting specific groups of particular interest as a result of indication (i.e., age) or contraindication (i.e., underlying contraindicated conditions). Finally, the Draft Guidance specifically notes that LCS should evaluate the comprehension of those individuals of low literacy — below an 8th grade level of reading.⁵⁸ This point probably requires evaluating literacy levels at screening to ensure a sufficient enrollment of the population for statistical analysis.

2. Data Analysis

Just as the study’s objectives should be established a priori, so should the criteria for determining the success of the study.⁵⁹ In other words, the “success criteria should be related to the predefined target level of comprehension for the primary communication objectives.”⁶⁰ The Draft Guidance suggests utilizing a confidence interval of 95% (two-sided) where the lower limit for comprehension is above the 90% comprehension target level (or higher, depending on the design of the study). If the study has more than one primary communication objective, results for all should be evaluated.

3. Questionnaire

The Draft Guidance sets out that a sponsor should hire an expert when designing the study’s questionnaire⁶¹ and also lists other considerations:

• Drafting questions specifically to assess comprehension of primary communication objectives
• Utilizing simple, specific, non-compound and unambiguous questions
• Employing practical applications questions (e.g., John is 5 years old and weighs 40 pounds. How many teaspoons should he take?)
• Combining open-ended and close-ended types of questions (e.g., Mark has diabetes. Should he take the medicine?)
• Asking questions designed to elicit a narrative response (e.g., Why did you answer the previous question the way you did?)
• Recording verbatim responses, particularly for incorrect answers
• Avoiding leading or biasing questions, as well as the biasing ordering of questions
• Including “I don’t know” as a response, but excluding “Ask a doctor” as a response⁶²
• If self-selection is one of the primary communication objections, using questions to validate that selection at the end of the study (e.g., pointed questions about medical history)
• Pre-testing the questionnaire

4. Miscellaneous

The Draft Guidance suggests the use of comparator labels, either in different studies or within the same study⁶³ but notes that the “Drug Facts label format and content requirements should be used.”⁶⁴ LCS also should be open book tests. Because “customary conditions of purchase” usually do not require a consumer to memorize the label, study participants should have access to the label throughout the study. Finally, “[v]erbatim responses to all questions should be recorded.”⁶⁵

IV. Conclusion

The Draft Guidance appears to be only the tip of the iceberg compared to all of the matters discussed at the 2006 NDAC meeting. While one group suggested imposing additional requirements on industry, another group (or at least two members of the NDAC) admitted that they do not believe that labels matter when it comes to consumer behavior.⁶⁶ Two members, though, conceded one real area where labels do matter: litigation.⁶⁷

And some of the issues from the NDAC meeting and the Draft Guidance may be helpful to a sponsor subsequently embroiled in litigation. For instance, if the FDA sets out what should be tested as a primary communication objective, a sponsor’s not testing communication of some other language (e.g., an extremely rare adverse reaction) may be easier to justify to a jury. In other words, the FDA will have set out (perhaps in agreement with the sponsor) what it considers to be the “substantial risk” of the medication. Under the Draft Guidance, the sponsor will have successfully tested communication of that risk, and it will have an FDA-approved reason for why it did not test communication of other risks.

Nevertheless, despite decades of study and reams of regulations, the FDA continues to strive for better comprehension and compliance. And despite all the decades of work, there are only three takeaways. The first is the belief that “more research is needed.”⁶⁸ The second is that “we have to recognize that no one label is ever going to be perfect for everybody.”⁶⁹ And the third remains the same one iterated over several decades:

[T]he best way . . . to improve OTC labels [is] to find a way to say it clearly, standardize it, [and] don’t say more than you need to.⁷⁰

---

[1] Wikipedia. “Over-the-counter drug.” <http://en.wikipedia.org/wiki/Over-the-counter_drug>. Quoted as viewed June 29, 2010. Last accessed July 1, 2010.

[2] 21 U.S.C. 352(c).

[3] M.D., Ph.D., Acting Chair, Nonprescription Drugs Advisory Committee.

[4] Nonprescription Drugs Advisory Committee meeting, Sept. 25, 2006. Transcript available at <http://www.fda.gov/ohrms/dockets/ac/06/transcripts/2006-4230t.pdf>. (“NDAC Tr.”).

[5] 37 Fed. Reg. 9464 (May 1972).

[6] 39 Fed. Reg. 19880 (June 1974).

[7] NDAC Tr. at 74.

[8] 40 Fed. Reg. 11717 (1975); see also NDAC Tr. at 15-16 (noting OTC products like cholesterol-lowering drugs and NSAIDs have more complex labeling that “populate the label to a magnificently cramped extent […]. [A]t what point do we pack so much information into the label that people stop reading it […] ?”); id. at 353 (noting the speakers agreed that “listing all the side effects has no utility”).

[9] 61 Fed. Reg. 8450 (Mar. 1996).

[10] 62 Fed. Reg. 9024 (Feb. 1997).

[11] 64 Fed. Reg. 13254 (Mar. 1999).

[12] Id.

[13] Id. at 13277.

[14] Id. at 13254.

[15] Id. at 13255 (internal references omitted).

[16] Id. at 13258.

[17] NDAC Tr. at 10.

[18] Id. at 73.

[19] Id. at 14.

[20] Id. at 23.

[21] Id. at 15.

[22] Id. at 351 (“[…T]here is regulatory and philosophical consensus that the Drug Facts label must contain all the information that is required to make that initial self-selection material, and that other material is really supplementary.”).

[23] Id. at 23.

[24] Id. at 24-25. The participant may be harmed, however, by not having the benefit of the drug (e.g., cholesterol-lowering drugs).

[25] Id. at 26, 27 (noting sponsors often do not provide information with respect to “incorrect” deself selectors).

[26] Id. at 28-29 (noting verification can be “very difficult”).

[27] Id. at 33.

[28] Id. at 17.

[29] Id. at 20.

[30] Id. at 23.

[31] Id. at 22.

[32] Id. at 70.

[33] Id.

[34] Id. at 30.

[35] Id. at 74.

[36] Id. at 31-32.

[37] Id. at 32-33.

[38] Id. at 342-43 (“I think I care more about the actual use because that is closer to how it is actually going to be started in terms of consumer use.”); 343 (“I think actual use is much more informative in that result, and I think it does trump.”).

[39] Id. at 344.

[40] Id. at 254, 264.

[41] One participant opined that a distinct type of behav­ioral science — “methodological science of OTC switch” — should be created. Id. at 91-92.

[42] See id. at 152.

[43] Id. at 169.

[44] Id. at 293-94 (“… [T]here is going to be a pre-specified threshold that is going to be in the protocol, in the statistical analysis plan, and it will be in the sponsor’s best interest to reach prospective agreement with the agency as to what that threshold would be appropriate for that specific case.”); 380 (noting Dr. D’Agostino “said 100 times today, [the study’s threshold] has to be set out in advance a priori […].”).

[45] Id. at 78.

[46] Id. at 81.

[47] An exception for particular warning language could be switches within the same class that, for example, carry the same concerns of contraindication. It seems, however, that this matter already is addressed with class labeling.

[48] Id. at 372.

[49] Id. at 371. And if a label cannot be designed below these levels, is the FDA unfairly withholding medications that otherwise are safe and effective for the average American?

[50] Id. at 374-75

[51] Id. at 86.

[52] Id. at 86-87, 93 (“analyzing the switches that have already taken place”). The sponsor, of course, was volunteered as the entity to bear the financial onus for such research. Id. One participant suggested a limited market approval and conducting Phase IV studies in that limited market. Id. at 94. At this juncture, however, the FDA does not have regulatory authority for a “limited market approval,” although at least one FDA participant noted it was a “very good idea.” Id. at 95.

[53] April 2009 Draft Guidance for Industry: Label Comprehension Studies for Nonprescription Drug Products (“Draft Guidance”), at 1.

[54] Id. at 1. Comparatively, during the NDAC meeting, Dr. Brass specifically noted that the studies are not “academic exercises” because they have as their goal the objective to “model and predict consumer behavior in the real marketplace.” NDAC Tr. at 11. Apparently, that goal has yet to be reached.

[55] Id. at 1, 2.

[56] NDAC Tr. at 16 (noting consumer leaflets or package inserts can be a condition of approval and FDA “control[s] what is on the package insert”).

[57] Draft Guidance at 5.

[58] Id. The FDA also suggested testing numeracy if label comprehension includes understanding numbers (weight-based dosing).

[59] Id. at 6.

[60] Id.

[61] Id. at 7.

[62] Dr. Doug Bierer, the CHPA representative, presented a somewhat contrary position. NDAC Tr. at 241 (suggesting “I would talk to my doctor” should not be an “incorrect” response).

[63] Id. at 8. See also NDAC Tr. at 148-151 (discussing comparison of labels and use of control groups).

[64] Draft Guidance at 9.

[65] Id.

[66] Id. at 177 (Dr. Alastair Wood, non-voting consultant: “[…] I don’t believe labels work.”); id. at 346 (Dr. Richard Neill, voting consultant: “[…] I believe […] that labels don’t matter much.”). Of course, not everyone agreed. See id. at 39 (Dr. Ruth Parker, committee member: “I think everybody here would really agree that labels are necessary for drug safety.”).

[67] Id. at 179 (Dr. Richard Neill, voting consultant: “I would mention that labels do work well at providing defensibility in court for sponsors and pharmacists and physicians and sometimes for patients.”); see also id. at 180 (Dr. Wood agreeing and noting “that’s why they are in small print and so long”).

[68] Id. at 91.

[69] Id. at 98-99.

[70] Id. at 57. A fourth lesson to consider is that sponsors should be prepared for and attend future NDAC meetings focused on consumer behavior studies.

Finis