Arthritis & Rheumatism, Volume 60,
October 2009 Abstract Supplement
The 2009 ACR/ARHP Annual Scientific Meeting
Philadelphia October 16-21, 2009.
Converting MHAQ, MDHAQ and HAQII Scores Into HAQ Scores Using Models Developed with 59,281 RA Patients
Anderson1, Jaclyn K., Wolfe2, F., Michaud1, Kaleb D.
The Stanford Health Assessment Questionnaire Disability Index (HAQ) is the gold standard functional status questionnaire in rheumatology, but is lengthy with 41 questions. Three prominent and much shorter versions, multidimensional HAQ (MDHAQ), modified HAQ (MHAQ), and HAQII, are often used in outcomes research as HAQ substitutes without demonstrated equivalence. Due to differing psychometric properties of each measure, we seek to develop a method of conversion between these three versions and the original HAQ.
Utilizing previously collected data from the National Data Bank for Rheumatic Disease (NDB) long-term outcomes study from 19982008, analysis was limited to comparison in rheumatoid arthritis (RA) patients at a random observation at which the HAQ was asked in conjunction with the MHAQ, MDHAQ, and HAQII. Univariate linear regression analyses were performed with 30 explanatory variables believed to be important indicators of the HAQ score with variables not contributing to the model excluded from further analysis. The developed models were limited to 80% of the data with the remaining 20% used to test the model fit. Predicted values were constrained to the 0 to 3 range of the HAQ. Graphical fits are presented in lowess curves.
The number of unique RA patients completing the HAQ and alternatives were: 29,686 MHAQ, 13,666 MDHAQ, and 19,117 HAQII. Table 1 displays coefficients (standard error [SE]) for the final models. The square root of MHAQ was more closely correlated with HAQ than the untransformed variable (0.881 vs. 0.857), improved the R2 modestly (DR2=0.042), and was included in the final model. Addition of pain VAS, number of years smoked, and patient global severity improved R2 by 0.013 but were not included in favor of parsimony. Transformed versions of the MDHAQ and HAQII did not significantly improve model fit. Including age and sex in the HAQII model produced a nonsignificant improvement (DR2=0.002). For each measure both the 80% development sample and 20% validation sample closely approximated the fitted HAQ values and demonstrate nearly identical lines (Figure 1).
We have developed conversion formulas between the MDHAQ, MHAQ, and HAQII and HAQ in a large sample of RA patients. We feel the models we have developed are useful for conversion of the measures in the research setting with application to the individual patient inappropriate.
Table 1. Model Coefficients.
|Model||Measure (SE)||AGE (SE)||SEX (SE)||Constant (SE)||[radic]MHAQ (SE)||R2 20% Validation Sample|
|MHAQ||0.380 (0.012)||0.006 (0.000)||-0.240 (0.005)||-0.068 (0.010)||1.085 (0.015)||0.816|
|MDHAQ||1.113 (0.005)||0.004 (0.000)||-0.206 (0.007)||0.053 (0.014)||||0.802|
|HAQII||0.993 (0.004)||||||0.044 (0.005)||||0.838|
Figure 1. Fitted vs. Predicted MHAQ, MD-HAQ, and HAQ-II.
To cite this abstract, please use the following information:
Anderson, Jaclyn K., Wolfe, F., Michaud, Kaleb D.; Converting MHAQ, MDHAQ and HAQII Scores Into HAQ Scores Using Models Developed with 59,281 RA Patients [abstract]. Arthritis Rheum 2009;60 Suppl 10 :79