Background The Stanford Health Assessment Questionnaire (HAQ, 20 questions for 8 domains, 21 for devices/aids use)1 has been the gold standard functional status assessment questionnaire in rheumatology. HAQ modifications to improve feasibility by questionnaire shortening, or psychometric properties have been proposed. The modified HAQ (MHAQ, 8 items)2, the multidimensional HAQ (MDHAQ, 10 items)3, and the HAQ II (10 items)4 are most widely used modifications. Conversion formulas have been developed to predict HAQ score from each of these shorter versions to allow comparison across different study populations.5
Objectives To compare distribution of HAQ score with MHAQ, MDHAQ and HAQII scores in a multinational cohort of rheumatoid arthritis (RA) patients and assess the ability of conversion formulas to predict mean HAQ score from the shorter versions.
Methods A cross-sectional study of 8488 patients in the Quantitative Standard Monitoring of Patients with RA (QUEST-RA) database who received usual care from rheumatologists in 33 countries was done. Patients simultaneously completed HAQ, MHAQ, MDHAQ & HAQII questionnaires (all scored from 0-3). Mean (SD) and median (IQR) were used to compare score distributions. Proportion of patients with no disability (score = 0) and good functional status (score ≤ 0.5) were assessed. For each of the shorter HAQ versions, proportion that differed from HAQ score by > 0.22 (usually considered minimal clinically important difference) was assessed. Means of predicted HAQ derived from the reported shorter and longer conversion formulas (please see Reference # 5) were compared with the actual HAQ mean.
Results HAQ, MHAQ, MDHAQ & HAQII scores were available for 8027 (94.6%), 7979 (94%), 7937 (93.5%) & 6926 (81.6%) patients respectively. Compared to mean HAQ, mean scores were higher for MHAQ and MDHAQ and lower for HAQII (Table). Compared to HAQ, more patients were classified as having no disability (score = 0) with MHAQ and less patients with MDHAQ and HAQII. MHAQ led to score of ≤ 0.5 in most and HAQII in least number of patients. Scores differed from HAQ by > 0.22 in 66.1%, 43.6%, and 42.5% patients for MHAQ, MDHAQ and HAQII respectively. All predicted HAQ mean scores were higher with actual HAQ mean with best results from shorter formula based on MDHAQ and HAQII (difference 0.11, 10.8%).
Conclusions HAQ and its shorter versions have different distribution and cannot be used interchangeably for individual patients. Though higher, reasonable estimates for group level HAQ mean can be made from MDHAQ and HAQII thus enabling comparison of patients in different studies.
Arthritis Rheum 1980;23:137;
Arthritis Rheum 1983;26:1346;
J Rheumatol 2005;32:1432;
Arthritis Rheum 2004;50:3296;
Arthritis Care Res 2010;62:1481
Disclosure of Interest None Declared