How does a change in the administration method affect the reliability of the COOP/WONCA Charts?

Abstract

Background. An interviewer is often needed to administer the COOP/WONCA Charts to Chinese patients, and this may affect the reliability of results. Objectives. We aimed to find out the reliability of the COOP/WONCA Charts administered by an interviewer, and whether a change in the interviewer or administration method would affect the results. Methods. We carried out a cross-sectional test-retest study on 487 Chinese adult patients attending a family medicine clinic in Hong Kong. The COOP/WONCA Charts were administered by the same interviewer, two different interviewers or self-completion and interviewer administration, on test and retest. The random, inter-observer and inter-method variances were compared with the inter-subject variance. The reliability coefficient of each COOP/WONCA Chart was calculated for each method of administration. Results. Random errors could change the scores by 0.57-1.04, inter-observer variations could change the scores of four charts by 0.72-0.80, and a change in the method could change the physical fitness score by 1.79 and the daily activities score by 1.31, on a five-point scale. The reliability coefficients of the six COOP/WONCA Charts were 0.68-0.92 for one interviewer, 0.59-0.82 for two interviewers and 0.46-0.81 for two methods. Conclusion. The Chinese COOP/WONCA Charts were reliable in detecting real differences when administered by an interviewer. A change in the method of administration significantly decreased the reliability of the results. The use of more than one method of data collection in the same survey should be discouraged.