2005D-0240 Guidance for Industry on Development and Evaluation of Drugs for Treatment or Prevention of Gingivitis; Availability
FDA Comment Number : EC6
Submitter : Dr. Karen Williams Date & Time: 10/27/2005 05:10:23
Organization : University of Missouri-Kansas City SOD
Category : Health Professional
Issue Areas/Comments
The following comments are made in response to Dr.Hyman request for feedback on the draft "Gingivitis: Development and Evaluation of Drugs for Treatment or Prevention." First, with respect to primary verus secondary endpoints, it is more logical to have a bleeding index (BI) as a primary endpoint along with gingival index (GI), as they are different measures of gingivitis and provide complementary information regarding product efficacy. Primary clinical outcomes for gingivitis should arguably be valid measures of the target disease, comprehensive in selection, sensitive to change and biologically sensible. Collectively BI and GI meet these requirements. Since the goal of an antigingivitis product is to maintain gingival health, the number of sites at the end of the trial that measure 0 or 1 for BI and GI in the test group, compared to the number in the placebo group, is an outcome measure that is meaningful to researchers and clinicians alike. While most gingivitis is indeed plaque-induced, having plaque index (PI) as a co-primary endpoint seems illogical. PI is a measure of potential etiology rather than change in the disease, and as such, is more appropriate as a secondary endpoint (along with stain and calculus assessments).

Second, the guidelines state "Of special consideration is the degree of gingivitis appropriate for enrollment, which will depend on the intended claim for the drug product and whether it would be marketed by prescription or OTC", yet provide no suggestions on a minimal level of gingivitis that would be deemed appropriate. Clearly some guidance on operationalizing "gingivitis" at a minimal level should be considered. Having participated in clinical antigingivitis trials over the past two decades, this is a topic of ongoing debate. Perhaps the omission of comments in this area is by intent; however, since efficacy is often a function of disease severity, it would be prudent to address this issue. Considerable discussion in the guideline is given to proportional reduction in GI and PI score (at least 15%), without having a concomitant discussion of a minimal score (either GI or BI or both) that represents "gingivitis".

Lastly, the guidelines state that statistical testing for GI and PI variables is usually performed with a comparison of means test through analysis of covariance, yet acknowledge earlier that both measures are ordinal in nature. Statistical analyses for ordinal-level measures are traditionally conducted by non-parametric tests. Although there is some evidence in the literature that parametric analyses often yield comparable results to non-parametric tests for ordinal measures, the underlying assumptions for parametric tests still need to be considered. No discussion is given in the guidelines regarding the distributional characteristics of ordinally scaled measures prior to conducting parametric analyses. Additionally, no mention is given to the issue of inflated family-wise error rate which seems to plague interpretation of results in the dental literature.

Overall, I believe that the guidelines are important, have great utility and should reflect an appropriate level of rigor.

Dr. Karen Williams
Director, Clincal Research Center
UMKC School of Dentistry