Context: Accurate assessment of a patient's palliative care needs is essential for the timely provision of treatment and support. The Integrated Palliative Care Outcome Scale (IPOS) is an ordinal measure possessing acceptable psychometric properties, but its ability to discriminate precisely between individual symptom levels has not been rigorously investigated.
Objectives: The study aimed to conduct Rasch analysis of the IPOS to evaluate and enhance precision of the instrument.
Methods: Responses of 300 community-dwelling palliative care patients were subjected to Rasch analysis using the partial credit model.
Results: Initial analysis supported the use of the Rasch model and acceptable reliability (person separation index = 0.77) was observed; however, unsatisfactory model fit was found. Local dependency between items was resolved through the creation of super-items, which increased model fit, reliability (person separation index = 0.80), and unidimensionality. There were no misfitting super-items or differential item functioning by age, rater, sex, or ethnicity. The IPOS showed satisfactory coverage of symptoms within the present clinical sample, with the ability to assess higher severity patients.
Conclusion: The modified IPOS showed excellent reliability for a clinical measure in assessing the overall palliative care needs of a patient. The provided ordinal-to-interval conversion table accounts for unique contribution of each symptom to the overall symptom burden and easy to use without the need to modify the original IPOS format.
Keywords: Integrated Palliative Care Outcome Scale (IPOS); Rasch analysis; assessment; psychometrics; validation.
Copyright © 2018 The Authors. Published by Elsevier Inc. All rights reserved.