Capturing health care utilization after occupational low-back pain: development of an interviewer-administered...

The aim of this study was to develop and test the feasibility and validity of a patient questionnaire to assess health care utilization after occupational low-back pain (LBP). Items generated after a literature search were revised and refined on the basis of their face and content validity (judged by a group of practitioners) and pretested with six lay subjects who had LBP. The 73-item questionnaire was then tested in interviews with subjects with acute, subacute, or chronic LBP. Its validity was judged by comparison with a prospective patient diary and with care-provider reports. Chance-corrected agreement was estimated using the kappa statistic. Response rates were 78%, 70%, and 59% for interview, diary, and provider reports, respectively. Eighty of 102 eligible workers completed the interview in an average of 45 minutes (SD = 17.7). Most LBP subjects (90.1%) found it easy to answer. In the opinion of the interviewer, 94.7% of subjects showed adequate comprehension and ability to recall. With a few exceptions, there was moderate to substantial agreement between the interview and the patient diary (most K values between 0.38 and 0.78). Overall, subjects reported more health care services to the interviewer than they recorded in the diary. Owing to the low response rate from providers, comparison with provider reports had to be restricted to 48 subjects and to physicians' reports only. Agreement between interviews and physicians' reports was substantial in use of plain X-rays (kappa = 0.79) and computed tomography scans (kappa = 0.85), but physicians often reported referrals not volunteered by the subjects. Agreement on prescription medications was fair (kappa = 0.29-0.46) with no systematic over reporting or under reporting. Our interviewer-administered questionnaire had better return rate than the patient diary and provider reports. It was easy to administer and understand. On the basis of our comparison with patient diaries and physicians' reports, we conclude that this questionnaire is a sufficiently valid source of health care utilization data in subjects with LBP.