Patient utilities in ankylosing spondylitis and the association with other outcome measures

To compare in patients with ankylosing spondylitis (AS) utilities derived by rating scale and standard gamble method, to relate these values to other outcome measures, and to assess the sensitivity to change of utilities relative to changes in other outcomes.

Patients with AS were randomly allocated to either weekly sessions of supervised group physical therapy for a period of 9 months or daily exercises at home. Analysis was restricted to the 59 patients who completed the Maastricht Utility Measurement Questionnaire (MUMQ) at baseline and after 9 months' followup and who were seen by the same interviewer. Reliability was assessed by intraclass correlation coefficient and change scores for marker states of disease. Construct validity was evaluated by correlation and multiple regression of baseline values with a variety of disease outcomes (pain and stiffness, patient's and physician's global assessment, Sickness Impact Profile, Health Assessment Questionnaire for the Spondyloarthropathies, Arthritis Impact Measurement Scale, functional, articular, and enthesis indices and spinal mobility measures). Sensitivity to change was assessed against changes in these outcome measures at followup.

The test-retest intraclass correlation coefficients for patient utilities were 0.95 (rating scale) and 0.79 (standard gamble), and for the marker state of mild disease 0.70 (rating scale) and 0.77 (standard gamble). A multiple regression analysis with the baseline rating scale or standard gamble utilities as dependent variable showed that patient's global assessment explained 59 and 11% of the total variance respectively. By multiple regression analysis 10% of the variance of change in rating scale utilities was explained by changes of patient's global assessment. In contrast, variance in change in standard gamble utilities was not explained by changes in other disease outcomes.

Findings obtained by rating scale and standard gamble differ considerably. Standard gamble utilities seem to address different aspects of health status than do rating scale utilities and more traditional outcomes. Utility measurement is sensitive to the method chosen to elicit patient well being.