Rheumatology 2000; 39: 783-790
© 2000 British Society for Rheumatology
Use of health status measures in patients with low back pain in clinical settings. Comparison of specific, generic and preference-based instruments
Department of Medicine, Baylor College of Medicine, Houston, TX, USA,
1 University of Alberta, Edmonton, Alberta and
2 Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Institute of Pharmaco-Economics, Edmonton, Alberta, Canada
Objective. To evaluate the discriminative performance over time of specific, generic and preference-based instruments in patients with low back pain (LBP) in clinical settings.
Methods. Forty-six consecutive patients with LBP participated in the study. Self-response questionnaires were administered at baseline and 3 and 6 months, including the following instruments: Oswestry (specific for LBP), SF-36 (generic), EuroQol (EQ-5D) and Health Utilities Index (HUI) (preference-based). EQ-5D and HUI weights were derived from previously published evaluations in the general population. Patients were asked to compare their health status with their baseline health and were categorized on the basis of an ordinal scale as: (a) improved; (b) stable; or (c) worse. Changes in the instruments were evaluated by rescaling the instruments over the same scale interval and by estimating standardized effect sizes between two time points for the three categories of change.
Results. Thirty-seven patients (80%) completed both the baseline and the 3-month questionnaire and 34 the baseline and 6-month questionnaires (74%). Overall, at both time points, approximately half of the patients reported no changes in their health status. Correlations between instruments were generally low, suggesting that they measure different health domains. The scales which discriminated best between patients who improved and those who deteriorated at 3 months were the Oswestry, the HUI, the EQ-5D and the SF-36 bodily pain and emotional role subscales. The SF-36 subscales appeared to have a floor effect for those patients who had deteriorated.
Conclusions. Most SF-36 subscales did not adequately reflect changes in the health status of patients with LBP, mostly for those who reported deterioration. Preference-derived quality-of-life scores appeared to discriminate among patients who improved and those who deteriorated, although not as consistently as the disease-specific measure (Oswestry). Additional research is needed to evaluate the role of generic measures of quality of life in the assessment of patients with LBP before they can be widely implemented in clinical settings or outcomes research.
Correspondence to: M. E. Suarez-Almazor, Health Services Research, Baylor College of Medicine, Veteran Affairs Medical Center (152), 2002 Holcombe Blvd, Houston, TX 77030, USA.