Development and Evaluation of GS-SF for Gastric Stuffiness (Wei PI) Patients Based on Item Response Theory, Computerized Adaptive Testing, and Data Simiuation / 中国中西医结合杂志

ABSTRACT

<p><b>OBJECTIVE</b>To develop and evaluate the short version of patient reported outcomes (PROs) questionnaire for gastric stuffiness (Wei Pi) patients with modern test theory and technologies, hoping to provide testing tools for related clinical practice and scientific researches with higher quality and less administrative and response burdens.</p><p><b>METHODS</b>Using descriptive study design, clinical data were collected with sociological questionnaire and previous developed full items version of PROs instrument for gastric stuffiness (Wei Pi) patients via field and online surveys between Sep 2011 and Mar 2012. The statistical analysis group identified the termination parameters firstly, and then selected items with discrimination, fitting residual, item information curve (IIC) , item characteristic curve (ICC), and the rank of computerized adaptive testing (CAT) select proportion, etc. After assumption evaluation of item response theory (IRT), IIC, ICC, difficulty coefficient distribution, items-response relation and thresholds, etc. were used for psychometric evaluation of instrument.</p><p><b>RESULTS</b>A total of 331 patients [Ages 31.99 +/- 10.29 yrs; Male 186 (56.3%)] were enrolled in statistical analysis. The test termination criterion was Max SE = 0.2 or Max items number =16. After items selection, a 15-item short version of instrument, which contains symptoms facet (8 items) and impact facet (7 items) was generated. With good unidimensionality, local independence, and monotonicity, the IC and ICC in IRT analysis showed good working capability of the questionnaire. The difficulty coefficient distribution and items-response relation were also rational, as well as response thresholds.</p><p><b>CONCLUSIONS</b>The short version of PROs instrument for adult gastric stuffiness (Wei Pi) patients was successfully developed and assessed. The instrument with good methodological and reporting quality could be used in clinical and scientific evaluating their symptoms and impact.</p>