Reproducibility of reports of past history, smoking, drinking and dietary habits obtained by a personal interview for a case-control study on liver cirrhosis and hepatocellular carcinoma.

The reproducibility of responses on life style measurements observed during a personal interview was assessed by a test-retest method. Fifty-one apparently healthy male subjects were interviewed twice, mostly within a 10 week interval. Responses on past history, smoking and drinking habits yielded relatively high intraclass correlation coefficients between the two interviews, as compared to the responses on dietary habits for the previous 1 to 2 years, which varied depending on the food items.