Reliability and validity of self and proxy reporting of morbidity data: a case study from Beirut, Lebanon.

We compared the self-reported illnesses (heart disease, back pain, rheumatoid arthritis, hypertension, and pulmonary disease) and smoking histories of 100 cases and 100 controls matched for age and sex with reports of this information from proxy informants from the same household in two areas in the city of Beirut. In addition, both cases and controls were given physical examinations to evaluate the accuracy of the responses. The level of agreement between the responses of subjects and of their informants varied from one condition to the other. Heart disease had the highest level of agreement, with the proportion of agreement greater than 93% for the cases and the controls and having chi values of 0.79 and 1.0, respectively. The report of back pain exhibited the lowest level of agreement, with responses showing a proportion of agreement of 74% for the cases and 90% for the controls, with chi values of 0.49 and 0.50, respectively. In comparing the responses of subjects and proxy informants with the results of physical examinations, heart disease had the highest level of agreement (J index ranged from 0.69 to 0.84), and back pain had the lowest level of agreement (J index ranged 0.42 to 0.48). These results show that proxy informants are good respondents for members of the same household and that health interview surveys are accurate for data collection of well defined chronic conditions.

Duke Scholars

Author Susan Halabi Biostatistics & Bioinformatics, Division of Biostatistics