131 PROMIS in VWD, IPFDs and RBDs 5 Reliability Reliability is the degree to which the measure is free of error. In IRT, on which the PROMIS CATs are based, measurement precision varies across the different levels of the measured constructs which is reflected in the standard error (SE). IRT based SEs can be transformed to classical test theory based reliability coefficients 49,50. For example, an SE of 3.2 corresponds to a reliability of 0.90 26,27,49,51 and an SE of 5.5 corresponds with a reliability of 0.70 26. Using the SE of the PROMIS scores, reliability coefficients were calculated for all PROMIS measures 49,52. The reliability of the SF-36v2 was assessed using the greatest lower bound (GLB) based on a preliminary analysis of the data. The GLB represents the smallest reliability possible given the inter-item correlation between the items of the questionnaire. In skewed data and with unequal inter-item correlations, the GLB has been found to provide more precise estimates of reliability compared to the Cronbach’s alpha 53. The Cronbach’s alpha is reported in addition to the GLB to promote better reliability estimation practice 54. Both GLB and Cronbach’s alpha were computed using the “psych” package in R using, respectively, the alpha and glb.algebraic function. For both the PROMIS measures and the SF-36v2 we assessed its ability to draw reliable conclusions based on the test scores on individual (reliability ≥ 0.90) and group level (≥ 0.70) 54-56. Relevance The relevance of the PROMIS measures for patients with VWD, IPFDs and RBDs was determined based on a comparison with the general population. The mean T-scores of the PROMIS CATs were categorized into four groups according to the available Dutch thresholds: 1) patients with no limitations/ symptoms within normal limits, 2) patients with mild symptoms, 3) patients with moderate symptoms, and 4) patients with severe symptoms 36-38. The distribution of the Dutch general population across these groups was compared with the distribution of the study participants on the PROMIS CAT physical function, pain interference, anxiety, depression, ability to participate in social roles and activities, and satisfaction with social roles and activities 36,38. The distribution of the Dutch general population is unknown for the PROMIS CAT fatigue and the short form anger. A PROMIS measure was determined to be relevant for patients with VWD, IPFDs and RBDs if the PROMIS scores negatively deviated from the scores of the general population.
RkJQdWJsaXNoZXIy MTk4NDMw