A COMPARISON OF DIFFERENT INDEXES OF RESPONSIVENESS

The first purpose of this study was to determine if different indices of responsiveness provided similar rank orderings of scales in terms o f responsiveness. The second purpose was to compare the responsiveness of patient specific, disease-specific, and generic health status meas ures for patients undergoing total hip arthroplasty. All patients of o ne surgeon at a single institution were eligible for the study. Patien ts who did not speak English or did not return for post-operative eval uations were excluded. Patients completed two disease-specific scales (the Harris Hip Scale and the Western Ontario and McMaster osteoarthri tis scale or WOMAC), one generic health status scale (the SF-36), and two patient-specific scales (the McMaster-Toronto Arthritis questionna ire or MACTAR and the Patient Specific Index or PASI). All scales were administered on two occasions: before and 6 months after total hip ar throplasty. Responsiveness was measured using: (1) the responsiveness statistic; (2) standardized response mean; (3) relative efficiency sta tistic; (4) effect size; and also by (5) correlating each scale's chan ge score with the change in patients' global ratings of their ''hip fu nction.'' Seventy-eight sequential patients completed the study. The m ean age was 62 years (range 25-87), 55% were male, and 71% had osteoar thritis. Test-retest reliability of the scales ranged from 0.31 to 0.9 3. The correlation among scales was consistent with a priori hypothese s confirming construct validity of the scales. Although the disease sp ecific scales were generally rated as the most responsive scales, the different indices provided different rank orderings by up to 5 levels (p = 0.04). In conclusion, disease specific scales are the most respon sive scales. However, choosing among scales based on responsiveness mu st be done with caution because different indices of responsiveness pr ovide different rank ordering. (C) 1997 Elsevier Science Inc.