Reproducibility of a diet history in older men in Hawaii.

A diet history method was tested for reproducibility among 106 older men from the five major ethnic groups (Japanese, Caucasian, Hawaiian, Filipino, and Chinese) of Hawaii. The questionnaire, administered by trained interviewers, was designed to estimate the dietary intakes of total and saturated fat, cholesterol, vitamin A, beta-carotene, vitamin C, and zinc. Subjects were asked to recall their usual frequencies and amounts of consumption during a usual month of more than 100 food items, along with the intake of seasonal foods with a high vitamin A content. A second interview was conducted 1-14 months later to obtain a diet history covering the same time period as the first interview. The same questionnaire was utilized in both interviews. The effects of ethnicity, age, and recall interval (period between the diet reference date and initial interview) on the reproducibility of nutrient intakes were examined. The intraclass correlation coefficient (rI) and the weighted kappa statistic (Kw) were used to assess agreement. The overall levels of reproducibility were relatively good. Neither ethnicity, age, nor length of recall period had a major effect on reproducibility. The data suggested that this diet history method is an appropriate instrument for estimating the usual dietary intakes of a healthy heterogeneous group of older men in Hawaii.