Abstract
There is limited evidence for the effects of ordering and length on responses to questionnaires used in the health field. Multiple outcome measures used in back pain studies have implications for respondent burden and response quality. This randomised study assessed the effect of questionnaire ordering and length on missing data and internal reliability for two health outcome measures.
Back pain patients were recruited from 26 UK practices in the UK BEAM feasibility study. Patients were randomised to receive a 27 page self-completed questionnaire with the Roland Disability Questionnaire (RDQ) at the front and SF-36 at the back of the questionnaire, or vice versa.
The mean number of missing items for the SF-36 was 0.07 (sd=0.68) and 0.56 (sd=2.73) at the front and back of the questionnaire; this difference was statistically significant (p< 0.05) for the general health perception scale. The internal consistency (Cronbach’s Alpha) of the RDQ was unaffected by questionnaire positioning; but was generally higher when the SF-36 (mean difference = 0.03) was at the start of the questionnaire and statistically significant for the vitality scale (p< 0.01).
The positioning of instruments affects patients’ responses. Researchers should consider the influence of questionnaire design. Primary measures should be positioned at the front of questionnaires.
The abstracts were prepared by Dr C Pither. Correspondence should be addressed to him at the British Orthopaedic Association, Royal College of Surgeons, 35–43 Lincoln’s Inn Fields, London WC2A 3PN
