Abstracts 111S

60P ASSOCIATION BETWEEN FOOD FREQUENCY QUESTIONNAIRE ESTIMATES AND SERUM CONCENTRATIONS OF B-CAROTENE, RETINOL, RETINYL PALMITATE, AND a-TOCOPHEROL AMONG

PARTICIPANTS IN THE CARET LUNG CANCER CHEMOPREVENTION TRIAL

Gary Goodman, Ina Gylys-Colwdl, Garnet Anderson, Mark Kestin, Gilbert Omenn, and CARET Co-Investigators

Fred Hutchinson Cancer Research Center Seattle, Washington

We compared dietary intakes estimated from a serf-administered food frequency questionnaire with serum concentrations of B-carotene, retinol, retinyl palmitate, and a- tocopherol in over 1,000 individuals at high risk for lung cancer. Data were obtained at baseline from two populations randomized to CARET, a multi-center chemoprevention trial of B-carotene and refinyl palmitate: male and female current or former heavy smokers and men with occupational exposure to asbestos. Correlations between serum B-carotene and dietary intakes of total vitamin A and of B-carotene (adjusted for age, BMI, years quit smoking, c~dories and alcohol intake) were each r--0.23. We found weaker associations between these nutrients and both serum retinol and refinyl palmitate (ranging from r--0.12 to r--O.17) and also between dietary and serum a-tocopherol (r--O.l l , r--0.19 when controlling also for serum cholesterol). In multivariate analyses, supplemental vitamin use was the strongest predictor of each of the four analytes. We found weak negative associations between the number of cigarettes per day and serum retinol and between daily alcohol consumption and serum B-carotene. Positive interactions were found between years quit smoking and daily intake of B-carotene and between years quit and dietary retinol in predicting serum B-carotene and retinol, respectively. These analyses confirm previous reports showing that dietary intakes (measured here by a food frequency questionnaire) can only moderately predict serum concentrations of B-carotene, retinol, retinyl palmitste, and a- tocopherol.

61P PLACEBO-EMERGENT ADVERSE EVENTS DIFFER

AMONG INDICATIONS

Michael G. Wilson, David J. Goidstein Lilly Research Laboratories

Indianapolis, Indiana

Placebo-emergent signs and symptoms (PESS) are adverse events that first occur or worsen with p ~ e n t after randomization with p ~ n t . Regardless of disease process or comparative agent, plac.ebo-treated patients are often assumed to report a similar adverse event profile across trials.

We retrospectively calculated PESS rates from randomized, double-blind, controlled clinical trials to describe PESS profiles for different diseases. PESS profiles for fluoxefine studies of major depression (MD, n -- 1008), obesity (n ffi 1108 ), obsessive compulsive disorder (OCD, nffi89), bulinda nervosa (BN, n=267), smoking cessation (n--524), and alcoholism (n--25) and profiles for fluoxetine (nffi 1008) or tomoxetine (n=668) in MD were compared using principal component analysis.

PESS profiles for alcohol and OCD patients were most different from each other; (r <0.16); MD, obesity, and smoking were most similar (r >0.8), BN was intermedlat~