Rx-to-OTC switch sponsors will be expected to work with FDA to establish predefined endpoints for each consumer behavior trial they plan, if the agency follows advice given by the Nonprescription Drugs Advisory Committee during a Sept. 25 meeting.

Trial developers should also be asked to determine before each trial is conducted what level of statistical variability will be allowed in deeming certain results close enough to those endpoints to be acceptable, NDAC panelists and guests said.

NDAC convened in Gaithersburg, Md., to discuss issues related to the design of label comprehension, actual use and self-selection trials.

Former committee chair Alastair Wood, MD, Emeritus Professor of Medicine, Vanderbilt University, framed much of the discussion on predefining trial endpoints. Wood was a guest speaker at the meeting, presenting on "Complexities of the Rx-to-OTC Switch."

Sponsors and the agency need to "define standards that we expect to be met, and we need to define these standards before we start the studies - and [need] the FDA to agree on these standards before the studies are initiated," Wood said.

One advantage to this approach would be to allow FDA input in developing questions for labeling and self-selection trials that will pinpoint the most crucial selection issues, Wood said.

"Nirvana would be that...pre-submission question-polishing would go on, and would also be conveyed to the FDA, so that there would be some understanding" of what types of questions and labeling are effective, he added.

In previous consumer behavior studies submitted for switches there was "seldom a pre-defined endpoint," Wood said. "We never know what the study goal was in these label comprehension studies."

Former NDAC chair Ralph D'Agostino, PhD, Boston University, discussed statistical issues related to the analysis of data from trials. D'Agostino participated in the meeting as a temporary member.

Predefined levels of acceptable statistical variability will help sponsors set thresholds for different outcomes on individual endpoints, D'Agostino said.

A trial designer should "take the issues that you are interested in, and set them up in [separate] domains," D'Agostino said.

This would require specifying in protocols the types of self-selection responses that may not be literally correct, but may represent acceptable results.

Data analysis has often been "a moving target" in the past Wood said, with decisions about acceptable results being made "on the seat of the pants."

"If we don't know going in what the passing grade is going to be, it's impossible to decide if it is reasonable or not," he added.

Former and acting Chair Eric Brass, MD, discussed the setting of thresholds for risk. Brass is director of the Harbor UCLA Center for Clinical Pharmacology, UCLA School of Medicine.

"I have a kind of simple perspective on this in terms of a committee member decision. What I would want to know is with what frequency a non-optimal behavior occurred, and the consequence of that non-heeding, then I could make some sort of qualitative assessment" of potential public health risk," he said.

D'Agostino and Brass supported the use of a 95% confidence interval for safety issues. "When it comes to safety, what you would like to be able to do is make an affirmative statement about the safety, not simply that you didn't observe any risk, or that you had some point estimate that suggested there was no risk," Brass said. A 95% CI "restricts what the risk will be."

Office of Nonprescription Products (ONP) Director Charles Ganley, MD, expressed some concern at the panel's unwillingness to set a similar general guideline for thresholds for endpoints. Ganley sought "some sense of what percentages we are talking about."

"In setting these thresholds, if we decide there's a certain threshold that has to be met, the last thing we want to do is come to the committee and have them say 'what the heck were you thinking?'" Ganley said.

Brass said NDAC would not likely bristle after the fact at thresholds set by FDA if the agency can "articulate the basis by which [the threshold] is derived."

"It's the thinking behind it that's important," Brass said.

"There will be some thresholds that will be over 90%, and that will be in targeted subpopulations on critical issues, and others will be more lenient - not 50/50, but maybe 80% - in terms of what's good enough," he said.

Andrea Leonard-Segal, MD, Director of ONP's Division of Nonprescription Clinical Evaluation, proposed that in some cases FDA "may try to bring threshold questions to NDAC ahead of time."

Brass noted the process of screening threshold questions through NDAC in advance could cause "considerable delay" to sponsors' programs.

FDA's policy on behavior studies should also feature "some expectation there will be consequences for unacceptable outcomes," Wood said.

For example, "if your comprehension study did really badly, there should be some expectation that that isn't going to result in an approval, so that people will strive to get studies done and data presented in a way that's comprehensible," he said.

The panel's discussion could help shape policy fairly quickly; ONP is developing several documents related to Rx-to-OTC switch.

The office is developing separate guidances on actual use and label comprehension studies, FDA announced in March (¹ Also see "Actual Use, Label Comprehension Guidances Slated For 2006" - Pink Sheet, 6 March, 2006., p. 5).

The agency is also revising its Manual of Policies & Procedures (MaPP) for Rx-to-OTC switch products (² Also see "ONP Has Actual-Use Trial Panel, Rx-To-OTC Switch MaPP In Mind – Ganley" - Pink Sheet, 8 May, 2006., p. 6.)

Presenting during the open public hearing on behalf of Consumer Healthcare Products Association, Douglas Bierer, PhD, provided industry perspective on behavior trials. Bierer served previously as the trade group's VP-regulatory & scientific affairs and is now a consultant.

"Decisions about whether consumer behavior is 'acceptable' should be based on actual risk to the consumer," Bierer said.

CHPA's message focused on the individual nature of switch candidates and of risk factors, stating for example that the risks related to noncompliance can vary for different products and label statements.

- Christopher Walker

NDAC Calls For Predefined Endpoints In Consumer Behavior Trials

More from Archive

More from Pink Sheet