ASPIRIN/ALCOHOL INTERACTION IS "INCONCLUSIVE" IN EPIDEMIOLOGY DATA

ASPIRIN/ALCOHOL INTERACTION IS "INCONCLUSIVE" IN EPIDEMIOLOGY DATA, FDA epidemiologist Ray Alderfer, MD, told a joint meeting of FDA's Nonprescription Drugs and Arthritis Advisory Committees on Sept. 8. In a summary of the epidemiology data submitted by McNeil Labs to demonstrate an increased risk of gastrointestinal bleeding among alcohol users taking aspirin, Alderfer noted that the agency does "not think that the studies presented were entirely conclusive." In addition, Alderfer, a medical officer in FDA's Office of Epidemiology and Biostatistics, pointed out that "the magnitude of any excess upper G-1 bleeding caused by [the] interaction [between alcohol and aspirin] is uncertain." Alderfer's presentation was based on an agency analysis of five studies submitted by McNeil, including: a prospective observational study, a retrospective review study, the SUNY-Health Science Center Study, the Slone Epidemiology Unit Study and a study conducted by D. Henry et al. entitled "Major G-I Complications and Non-Aspirin Nonsteroidal Anti-Inflammatory Drugs (NANSAIDs). In focusing much of his presentation on the Slone study, Aiderfer noted that, of all the studies submitted by McNeil, the Slone study "has all the ingredients to be a study that establishes interaction." The study, conducted by Boston University's Slone Epidemiology Unit, is an ongoing multicenter, case-control study that has evaluated 1,004 cases of upper G-I bleeding in Hungary, Sweden and the U.S. ("The Tan Sheet" Sept. 6, p. 10). The study found that the relative risk of major upper gastrointestinal bleeding (UGIB) for regular aspirin users taking more than 325 mg/week and who consume more than 24 drinks a week is 8.2. The study also found that progressively higher levels of alcohol consumption among high-dose aspirin users correlated with an increased risk of UGIB episodes. However, the Slone study also found that the relative risk for regular aspirin users taking less than 325 mg/week decreased with increased alcohol consumption. Data on occasional aspirin use and alcohol consumption were inconclusive. Despite the evidence that suggests alcohol and aspirin interaction, Alderfer felt "the level of certainty for interaction and magnitude [of interaction] are indeterminate from this study . . . given the limitations of the data and some of the assumptions which need to be made." In his presentation, Alderfer noted that for regular aspirin users consuming more than 325 mg in addition to at least 24 drinks per week, the relative risk "is very dependent on the number of controls in the study who are aspirin users. If we assume a little misclassification . . . [and] one nonuser was actually an aspirin user the relative risk [drops from 8.2 to] approximately 6.5," he pointed out. Also, Alderfer showed that the relative risk for heavy drinkers changes according to the reference group. Persons non- exposed to alcohol "have a slightly higher risk of UGIB" than light drinkers -- 1.4 compared to 1. The reference group used to determine relative risk for heavy drinkers, he explained, is persons consuming less than one drink/week. But, if the reference group is non-users, he continued, the relative risk for heavy drinkers decreases from 2.3 to 1.6. The question Alderfer put forth then is "which reference group is more appropriate?" To determine incidence rates, Alderfer suggested that "some assumptions [had] to be made." For instance, he pointed out, the study authors assume that "drinking patterns of study cases of the case-control study are comparable to cases from the Massachusetts Health Data Consortium." In addition, Alderfer continued, they "assume that drinking patterns in the other countries [involved in the case control study] are comparable to Massachusetts." Other studies reviewed by the joint committee included 10 published epidemiologic studies, eight studies on the additive effects of NSAIDs and alcohol on G-I tract irritation, three studies on the ability of alcohol to potentiate prolonged bleeding in aspirin users and four studies on the effect of aspirin on alcohol pharmacokinetics. The joint committee found that many of the studies either lacked control groups, included "potential bias" in the controls or made conclusions based on assumptions. Some studies, the committee also concluded, were irrelevant to the issue of the additive effect of alcohol with aspirin/NSAIDs. George Ehrlich, MD, University of Pennsylvania, asserted that the committees, in looking too closely at the data, were "losing sight of the issue." He suggested that "no one" on the panel would "deny that there probably is a relationship between NSAIDs and some changes in the stomach . . . and [that] the addition of alcohol in larger quantities . . . probably increases the risk of this happening." But, he continued, "the question really is ultimately going to be labeling." The studies, he asserts, are "good food for thought" but that the joint committee must "keep in mind how . . . best [to] help the public understand what happens when they take medication." The Slone study authors, Judith Kelly, Boston University, et al. also found no UGIIB correlation with regular ibuprofen use and increased alcohol consumption, however. In addition, the data show that occasional ibuprofen users who drink moderately (1-20 drinks/week) or excessively (greater than 21 drinks/week) were at lower risk of UGIB events. The data on naproxen use and alcohol consumption are inconsistent -- showing an 8.2 relative risk for regular naproxen users who consume less than one drink/week, 11 for naproxen users who consume levels of alcohol between one and six drinks/week, and 2.6 for consumers of between seven and 20 drinks/week. The joint committee also heard presentations from McNeil, Whitehall and the Aspirin Foundation of America that focused on the Slone study and drew different conclusions from the data.