OTC CAFFEINE JAMA ANALGESIC STUDIES: SHALLOW DOSE RESPONSE CURVE

OTC CAFFEINE JAMA ANALGESIC STUDIES: SHALLOW DOSE RESPONSE CURVE indicates that the ingredient has no clinical effects, McNeil maintained in a critique of a study of caffeine as an analgesic adjuvant published in the April 6 issue of the journal. The critique, submitted to FDA for inclusion in the OTC Internal Analgesic Monograph record, commented on the study by Laska, Sunshine, et al, which evaluated pooled data from a series of 30 clinical trials conducted over the last 20 years. McNeil asserted that "a major shortcoming of the paper is the failure of [Laska, Sunshine et al] to indicate that a relative potency estimate of 1.41 is not associated with a meaningful clinical effect due to the shallow dose response curve associated with the data." McNeil maintained that "an analysis of the dose-response relationship" would translate "the calculated clinical benefit from a 41% increase in dose . . . into an increase of less than 2 units of %SPID [sum of pain intensity differences]. Such a small increment in %SPID is lacking in clinical significance." McNeil asserted that the conclusion that "an apparently significant relative potency does not equate to a significant treatment effect . . . is supporte by both a weighted and an unweighted analysis of covariance of the mean values which indicate there is no significant caffeine effect (both P values > .26) even though both analyses yield significant slopes (both P values < .009)." In critiquing the analytical methodology, McNeil noted that eight of the studies cited by the authors "were not designed to assess dose response relationships since they include only one or two dose levels per treatment group . . . Furthermore, the remaining studies have only three dose levels per treatment groups. This makes it difficult to validate the assumptions necessary for the proper pooling of studies." McNeil also declared that no justification was given by the authors for the exclusion of "two studies from the pooling for which a negative dose response relationship was obtained." McNeil further questioned the legitimacy of the pooling by asserting that the inclusion of two crossover studies with the parallel studies "is clearly not appropriate . . . because of their dissimilar variance-covariance structures." McNeil explained that "by separating the data into three groups: (1) less than 650 mg of analgesic drug, (2) between 650 and 1000 mg . . . and (3) greater than 1000 mg . . . and performing either a weighted or an unweighted analysis of the mean values, a negative dose relationship is obtained within each dose group. Thus a paradox arises since there are negative dose response realtionships within each dose group, but a positive dose relationship for all combined data." In August FDA concluded that "Caffeine as an analgesic adjuvant remains classified in Category III," in response to 11 studies submitted by Bristol-Myers to demonstrate that caffeine enhances the analgesic effect of acetaminophen ("The Pink Sheet" Sept. 3, T&G-4).