Comparison of self-completion vs personal interview methods in a survey of respiratory symptoms

Abstract

The level of pollutants, especially SO_X, are becoming lower in the air. The prevalence rates of respiratory symptoms such as cough and phlegm are also decreasing, but the sampling size must be increased to know the effects of air pollution. Self-completion methods have the merit of increasing the sampling size, but the reliability of self completion method has been argued. In previous studies, agreement between the prevalence rates of self-completion methods and those of personal interview was tested. But this agreement is not considered to be always necessary, if a specific relationship between the two methods is observed in the range of the prevalence rate.
From this point of view, the correlation and regression analyses of cough, phlegm, dyspnea and asthma symptoms were made, and the reliability of the self-completion method discussed.
The data of 19 districts in Okayama prefecture was used, where surveys using both methods were attempted. The questionnaire for the personal interview method was the BMRC's standard questionnaire, and the questionnaire for self-completion method was made by referring to the CMI's and other self-completion questionnaires.
The results were obtained as follows:
1. On regression analyses, the regression coefficients were statistically significant in many cases (p<0.01 for eight cases and p<0.05 for three out of twelve cases).
2. The correlation coefficients between the two methods were large, especially for the analyses of cough and phlegm. For phlegm, the lower limits of 95% confidence interval of correlation coefficients ranged from 0.833 to 0.982. For cough, it ranged from 0.630 to 0.941.
3. From the results of 1 and 2, it was considered that the self-completion method using a simple questionnaire was as useful as the personal interview method to study the prevalence of cough and phlegm in the districts.