Generic and condition-specific measures of quality of life are often used in parallel. Despite extensive evidence of question ordering effects in the general survey literature, there is no consensus on which type of measure should be administered first and little previous conclusive research into instrument ordering effects.
To investigate the effects of instrument ordering on response rates, speed of response, and response patterns to questions on health-related quality of life.
Subjects were randomized to two different versions of a self-completion questionnaire; in the first, condition-specific measures of quality of life preceded generic instruments; in the second version, the relative positions were reversed.
Adults with asthma or angina from 62 family practices in northeast England.
Instruments were the generic Medical Outcomes Study Short Form 36-item questionnaire, the EQ-5D, the Newcastle Asthma Symptoms Questionnaire, the Asthma Quality of Life Questionnaire, and the Seattle Angina Questionnaire. Effects were assessed in terms of questionnaire response rates, speed of response, item nonresponse rates, internal consistency, and domain scores on the quality of life measures.
Instrument ordering had no effect on questionnaire response rates or response speed. Only condition affected item nonresponse rates. Some ordering effects in respect of quality of life scores were observed, but these were inconsistent within and between conditions, and none of the differences were clinically significant.
There is little effect of instrument ordering on responses to self-completed measures of quality of life. Further research is required to test whether this finding extends to other methods of administration.