Label Comprehension “Yes/No” Questions Should Be Avoided, FDAer Says

Drug sponsors should avoid using "yes/no" questions in label comprehension tests for Rx-to-OTC switch candidates, FDA's Karen Lechter, PhD, advised at the Consumer Healthcare Products Association's Research & Scientific Development Conference in Washington, D.C. Nov. 8.

"Fifty percent of the time people, by chance, are going to get it right," noted Lechter, a social science analyst for the agency's Division of Drug Marketing, Advertising & Communications.

Lechter cited as an example the question: "According to the label, can someone using product X use nicotine-containing products?" Such a question is also somewhat leading, she said, because "why would you even be asked if there isn't a problem [with nicotine]?"

The solution might lie in a series of questions involving hypothetical "nicotine use" scenarios, which ask the participant whether or not, and how much of the product can be used, she suggested.

Also, the question might be embedded "in a series of other...similar questions in which the answer is opposite so they don't have any bias toward always answering a certain way," the FDAer said.

Lechter also cautioned against use of "leading" questions that prompt responses otherwise not considered.

For example, if a sponsor asks, "What should you do before using product X?" then the participant knows "there's something you should do [before using the product] and they'll go back and read the label more carefully," she explained.

Lechter again recommended use of "scenario" questions. For instance, she said, a label comprehension test for a cholesterol-lowering drug might ask, "If a person [is] very inactive and has a high-fat diet...what is the first thing he should do about his cholesterol?" The following question would then state: "What is the second thing?"

However, such questions might be too open-ended and fail to zero in on the answer sought, she cautioned.

Another option is a multiple-choice question, which includes the possibility that there is nothing a consumer should do before taking a particular product, she said.

Lechter offered similar suggestions to Merck and Bristol Myers-Squibb on label comprehension studies for their respective statin switch candidates Mevacor (lovastatin) and Pravachol (pravastatin) at a joint meeting of the Nonprescription and Endocrinologic & Metabolic Drugs Advisory Committees in July 2000 (¹ (Also see "OTC Statin Physician Follow-Up Label Comprehension Not Shown - FDA" - Pink Sheet, 24 Jul, 2000.), p. 12).

In conjunction with yes/no, multiple choice or leading questions, "it's important to ask why they answered as they did to be sure that if they answered correctly it was for the right reason, or if they answered incorrectly it was because they really didn't understand the label," Lechter advised.

However, both Lechter and Johnson & Johnson-Merck Director of Market Research Stephanie Levy maintained label comprehension alone is not enough, and that sponsors also should ask questions on self-selection to ensure the product's label is "convincing."

"Not only must consumers understand what the label is saying, but they must also believe that it makes sense and that it applies to their situation," Levy said. "The label must convince them that they need to follow the instructions and pay attention to the warnings."

In a separate session on actual use studies, NDAC consultant and former chairman Ralph D'Agostino, PhD, Boston University, said researchers should limit the number of objectives they attempt to reach so as not to overly complicate study results.

Researchers must state their objectives for a chronic disease study "explicitly and precisely" and "keep them focused and simple," D'Agostino advised.

Statistical issues that are accentuated in actual use studies include "multiple objectives, how you review target versus sample population biases and messy data, drop outs and missing data," he said.

When data from actual use studies is collected "you usually have a [huge] mess," D'Agostino explained, suggesting, "You should go back and see what individuals did. How did they behave."

In addition, chronic disease studies "should not have lots of endpoints and ways to analyze them" but instead, "should sort out primary and secondary objectives, avoiding efficacy as a primary objective," he said.