David A Andrae1, Paul S Covington2, Donald L Patrick3. 1. Furiex Pharmaceuticals Inc, Morrisville, North Carolina. Electronic address: david.andrae@furiex.com. 2. Furiex Pharmaceuticals Inc, Morrisville, North Carolina. 3. Department of Health Services University of Washington, Seattle, Washington.
Abstract
BACKGROUND: In previous studies, the Irritable Bowel Syndrome Quality of Life (IBS-QOL) instrument has been determined to have good measurement properties for general irritable bowel syndrome (IBS) and the diarrheal IBS (IBS-d) subtype in clinical trials. OBJECTIVE: This article aims to extend the true-score analyses that have been previously conducted to evaluate the IBS-QOL in IBS-d patients. METHODS: Item response theory analysis was conducted by fitting models to responses from 753 patients with severe IBS-d from a recent clinical trial. Three item response theory models, the constrained graded response model (CGRM), the unconstrained GRM (UGRM), and the testlet response model (TRM), were fit to the 34 items of the IBS-QOL questionnaire. Subsequently, differential item functioning (DIF) for patient sex was assessed by fitting nested models by applying likelihood ratio tests. Model latent trait estimates were then compared with the IBS-QOL score and the IBS Symptom Severity Score. RESULTS: Model fits improved with complexity, with the TRM model fitting best compared with the CGRM and UGRM. The DIF evaluation for patient sex flagged 17 items for the CGRM and 9 items for the UGRM; however, these items were not found to have much effect on the overall estimation of the latent trait. Differential testlet functioning was not indicated, and no items exhibited potential DIF under the TRM because likelihood ratio tests were not statistically significant. Comparison of latent trait estimates to the IBS Symptom Severity Score and IBS-QOL questionnaire revealed high Spearman correlations (0.47 and ≥0.99, respectively). CONCLUSION: Previous true-score approach results were supported by the IBS-QOL item-level analysis. Further, the IBS-QOL total score was found to be a valid measure of perceived quality of life for IBS-d patients when compared with more sophisticated model-based estimates of perceived quality of life. ClinicalTrials.gov identifier: NCT01130272.
BACKGROUND: In previous studies, the Irritable Bowel Syndrome Quality of Life (IBS-QOL) instrument has been determined to have good measurement properties for general irritable bowel syndrome (IBS) and the diarrheal IBS (IBS-d) subtype in clinical trials. OBJECTIVE: This article aims to extend the true-score analyses that have been previously conducted to evaluate the IBS-QOL in IBS-d patients. METHODS: Item response theory analysis was conducted by fitting models to responses from 753 patients with severe IBS-d from a recent clinical trial. Three item response theory models, the constrained graded response model (CGRM), the unconstrained GRM (UGRM), and the testlet response model (TRM), were fit to the 34 items of the IBS-QOL questionnaire. Subsequently, differential item functioning (DIF) for patient sex was assessed by fitting nested models by applying likelihood ratio tests. Model latent trait estimates were then compared with the IBS-QOL score and the IBS Symptom Severity Score. RESULTS: Model fits improved with complexity, with the TRM model fitting best compared with the CGRM and UGRM. The DIF evaluation for patient sex flagged 17 items for the CGRM and 9 items for the UGRM; however, these items were not found to have much effect on the overall estimation of the latent trait. Differential testlet functioning was not indicated, and no items exhibited potential DIF under the TRM because likelihood ratio tests were not statistically significant. Comparison of latent trait estimates to the IBS Symptom Severity Score and IBS-QOL questionnaire revealed high Spearman correlations (0.47 and ≥0.99, respectively). CONCLUSION: Previous true-score approach results were supported by the IBS-QOL item-level analysis. Further, the IBS-QOL total score was found to be a valid measure of perceived quality of life for IBS-d patients when compared with more sophisticated model-based estimates of perceived quality of life. ClinicalTrials.gov identifier: NCT01130272.