Post-operative nausea and vomiting(PONV) is known to affect over two-thirds of patients after bariatric surgery. Washington University Weight Loss Surgery(WUWLS) currently uses a validated tool, the Rhodes Index of Nausea and Vomiting, to objectively evaluate PONV. However, this tool is impractical for clinical use because it consists of 8-questions. The objective of the study was to use item response theory(IRT) to identify the key items of the questionnaire and generate a mini-Rhodes Index to facilitate easy assessment of patient reported outcomes regarding PONV.
WUWLS surveyed patients post-operatively from 11/01/2017-11/01/2018. We used IRT to analyze these data with a probabilistic framework for modeling the interaction of a respondent with an item. We compared the performance of the original 8-item questionnaire to versions consisting of varying subsets of items. Performance of these shortened versions of the instrument were compared to the original version based on how much information they provide.
We collected data from 309-patients and 696-questionnaires. Analysis of all possible two-item subsets from the survey identified short forms that retained reasonable measurement precision over the same range as the original instrument. A two-item form containing items about distress from vomiting and the number of times someone retched was optimal amongst shortened versions.
The present study demonstrates a novel use of IRT to develop a shortened PONV questionnaire, the mini-Rhodes Index. Shortening the tool to two-items makes the tool more feasible for clinical use. WUWLS is using the mini-Rhodes index to further understand PONV and optimize post-operative management