Improving the Health Assessment Questionaire: The HAQ-100

PURPOSE: The Health Assessment Questionnaire Disability Index (HAQ, HAQ-DI), over 27 years, has evolved into an extremely widely used outcome measure. Item improvement techniques and Item Response Theory (IRT) now permit improved items and instruments: (1) measurement characteristics may be improved,(2)information content may be increased and greater statistical power achieved, (3)new instruments may be cross-calibrated with the classic to allow valid interpretation of results from both instruments, and, while ongoing NIH-Roadmap outcome assessment systems extending this work will in the future yield dynamic computerized adaptive testing (CAT) with even greater precision, an improved HAQ-100 (scored on a 0-100 scale)can replace current HAQ usage with transparency and cross-calibration. />METHODS: 1860 disability items from 165 instruments, including the classic HAQ, were examined for clarity, patient importance, redundancy, and floor and ceiling effects using qualitative inputs from patients and health professionals. Desirable item features were identified and modified items constructed so that a classic and a revised HAQ could be compared. A reduced item bank of 208 physical function/disability items was administered to over 18,000 subjects including over 6,000 osteoarthritis and 3,000 rheumatoid arthritis subjects. Data on classic and revised HAQ items were extracted,instruments scored, and compared using IRT to assess information content. Instruments were compared using means, standard deviations, and correlation matrices.
RESULTS: Items were improved when the present tense was used rather than 7-day recall. 5-point response options had fewer ceiling and floor problems than 4-point scales. Clarity was improved by rewording and assessment of reading levels.

The Figure shows the information content of a typical HAQ item (taking a tub bath) in classic and revised versions; information content is a function both of the highest point of information and of the breadth (area under the curve)which relates to reduction in ceiling and floor effects. Revision and simplification of 'aids and devices' items improved content validity. Information content was improved by from 25 to 40 % in revised items. Mean values of instruments correlated closely (r >0.90);sensitivity was greater with the HAQ-100 (p<0.001), and item standard deviations smaller, representing increased precision.(p<0.001)
CONCLUSIONS: The HAQ-100 is more sensitive and more specific than the classic HAQ and has no disadvantages compared with that scale. It is currently available for use as a PROMIS instrument, and may be substituted for the classic HAQ in any application, with advantage.

 J.F. Fries, None; B. Bruce, None; M. Rose, None.