Background TAQRDis is an instrument to measure the therapeutic alliance in rheumatic patients developed and validated in our department. Despite TAQDRDis performs well, we found some limitations that may be overcome Rasch analysis.
Objectives To validate TAQRDis in patients with RA or gout using Rasch analysis.
Methods The questionnaire was administered to two cohorts of patients, one with gout (GRESGO) and another with RA (RECAR) to collect sociodemographic, clinical, and therapeutic data.
Instrument description: TAQRDis includes 21 questions (9 for therapeutic adherence and 12 for patient-physician's relationship) with 34 items. The correlation betweem TAQRDis and other adherence questionnaires (MASRI, HAQ-DI, and EuroQoL 5D) was satisfactory.
Estimation of Cronbach's α and major component analysis was performed. For Rasch analysis we estimated separation and reliability rates, fit statistics separation person-item and dimensionality statistics components analysis, infit and outfit rates and Wright map). We used Stata v11 and WinSteps v3.73 software.
Results 238 patients with gout (age 47.6±13 years, 95% males, education 9.2±4.2 years, socioeconomical level index 8.44±2.11) and 50 with RA (94% female, age 45±14 years, education 8.6±3 years, socioeconomic level index 14.0±1) were included. Cronbach's α value was 0.75. eighty percent of the variance accounted for 6 of the 34 items. We found an item separation index of 6.26 for gout and 2.19 four RA, and a person separation index of 0.17 for gout and 1.79 for RA. Item reliability was 0.98 for gout and 0.83 for RA. Mean-data adjustment-ranges comparing the observed vs. the expected pattern were 0.93 for gout and 0.97 for RA for infit (weighted item and person behavior), and 1.05 for gout and 0.94 for RA for outfit (sensitive to outliers). Uni-dimensional analysis showed an explained total variance of 148% in patients with gout and 62% with RA; unexplained variance was 2.5% and 6.5% for gout and RA. In the alliance tested for both groups, 88% of items showed loads between 0.5 and -0.5 (optimal), whereas the extent of difficulty lied between 3.98 and -4.5 logit for gout and 3.63 and -2.08 for RA.
Conclusions According to the ideal model out forward by Rasch, TAQRDis has good discrimination and adjustment in measuring the therapeutic alliance. Patients with RA were more consistent in their responses. The use of Rasch analysis is recommended to validate outcome measures from the patient's perspective.
Disclosure of Interest None declared