A self-administered, health history questionnaire prepared for epidemiological studies or routine use in a medical clinic was completed twice (with different order), and subjects were interviewed with same questions. The reliability of the responses of ´45 subjects, all male college students, were tested statistically.
In this setting, 45 subjects averaged 6.7 minutes to complete the questionnaire. Test-retest reliabilty of responses to the questionnaire; was 0.89(by each respondent), 0.90(by each question) by agreement index and 0.54(by each respondent), 0.47(by each question) by kappa index. The study results showed a low incidence of false positive (5.8 percent) and false negative (3.9 percent) responses. By categorical classification of history contents, `habit´ was the most reliable part, and the next order was past medical history, family history, present symptoms, and past symptoms. The value of agreement and kappa index was significantly correlated.
Test-retest reliability of responses to questionnaire and interview by a paramedical personnel was 4.86(by each respondent), 0.86(by each question) by agreement index, and 0.28(by each respondent), 0.27(by each question) by kappa index, which was lower than that of responses to questionnaire only. Both levels of reliability were significantly correlated. The incidence of false positive responses was 4.4 percent and false negative response was 8.9 percent.
|