Development of a CAT-tool for PRO assessment in oncology

Background In oncology in the last two decades there has been growing awareness of the importance to complement the traditional physician-rated assessment of the patient`s health status by an assessment made by the patients themselves (patient reported outcomes - PRO). Computer adaptive testing (CAT) provides an innovative methodology for data collection that makes PRO assessment more patient-oriented, precise and effective. Cancer- related fatigue (CRF) is one of the most common unrelieved symptoms experienced by cancer patients. However, up to now CRF is often underrecognized and undertreated partly due to a lack of clinically sensitive assessment instruments that integrate easily into daily clinical practice and research. CAT allows to overcome these obstacles by enabling precise assessment of fatigue without requiring the administration of a large number of questions. Aim and hypotheses In the planned translational research project we aim at improving the methodology for the assessment of cancer- related symptoms and QOL by developing a CAT tool (software) and transferring it into oncological research and care. Fatigue, the most commonly reported symptom in cancer patients, was selected as a prototype for showing the advantages and the clinical usefulness of the new methodology. The following three hypotheses are tested: (A) The Fatigue-CAT requires less items to achieve the same or even higher precision of measurement than two widely used fatigue questionnaires, the Functional Assessment of Chronic Illness Therapy - Fatigue (FACIT-F) and the Multidimensional Fatigue Inventory (MFI-20). (B) Due to higher precision of measurement and less bottom and ceiling effects the Fatigue-CAT is more sensitive to change over time and more reflective of clinical group differences than the FACIT-F and the MFI-20. (C) The Fatigue-CAT (including computer-based graphical reports) improves identification of cancer patients suffering from fatigue and initiates fatigue-related clinical interventions. Methods To develop and test the CAT-tool four principal steps are planned: (1) Construction and validation of a German fatigue item bank based on an existing US version. (2) Development of the actual CAT procedure: Construction and programming of the CAT-algorithm. (3) Testing of the Fatigue CAT with cancer patients of the various departments (oncology, surgery, haematology, gynecology, etc.) of Innsbruck Medical University in comparison to the FACIT-F and the MFI-20. This includes precision testing and exploring the sensitivity of the CAT with respect to clinical change over time. (4) Implementation of the Fatigue CAT into a clinical setting including the development of computer-based clinically meaningful real-time reports (graphical feedback) and their evaluation with regard to clinical usefulness. Sample In total a sample size of 1400 cancer patients is required. For the calibration and psychometric testing of the item bank (step 1) a sample size of 1000 cancer patients is needed according to a power calculation. Feasibility testing (step 2) will include assessment of patients` reactions by using a feedback questionnaire. 100 patients will be included. For precision testing and sensitivity analysis of the Fatigue CAT (step 3) 300 cancer patients will be used. For the prospective clinical evaluation study (step 4) a subsample of 150 patients of step 3 will be assessed at four further time points. All patients will be recruited at Medical University of Innsbruck. Statistical analysis Statistical analysis will be performed using both item response theory (IRT) methodology and classical test theory. For step (1) the analysis will include the establishment of item hierarchies and step measures, the identification of gaps in the item hierarchy and a check for unidimensionality of the item bank. The Fatigue-CAT (step 2) algorithm will be programmed in Visual basic and incorporated into our Data management programme CHES (Computer based health evaluation system). For step (3) a main part of the analysis will be performed by means of general linear models. ) Within step (4) chi-squared tests, Mann-Whitney U test as well as analyses of covariance will be applied to compare the two groups (fatigue-reports based on Fatigue-CAT provided/ not provided) with regard to various outcome criteria.

The project aimed at developing computer-adaptive assessment instruments for important patient-reported outcomes (PROs) in oncology. PROs refer to any aspect of a patients health status that is reported directly by her/himself (e.g. pain, fatigue, depression, nausea). For the quantitative assessment of PROs in research and clinical practice a number of paper-pencil questionnaires are available. These present the same questions to each patient. The computer-adaptive measures developed in this study overcome several limitations of these conventional questionnaires. Most importantly, they provide different questions to different patients, so that each patient is asked precisely those questions that are most relevant to his/her current condition. Tailoring questions to the individual patient shortens the assessment, lowers patient burden and increases measurement precision. Our project focused on the development of computer-adaptive PRO instruments for key cancer symptoms (e.g. Fatigue, Emotional Functioning, Physical Functioning). All instruments were developed in close collaboration with the Quality of Life Group of the European Organisation for Research and Treatment of Cancer (EORTC), which has been conducting a large-scale CAT development study including cancer centres in Europe and Australia. The computer-adaptive fatigue instrument developed in this project has already been implemented into symptom monitoring at several hospital units at Innsbruck Medical University to improve treatment evaluation and monitoring in cancer patients with regard to this key symptom. At these units patients complete the CAT measure on a tablet PC and with the help of software the results can be immediately presented to the medical staff as graphical charts. In general, the project contributed to an improvement of PRO assessments by providing short and precise measures for key symptoms in cancer patients. The availability of such measures and their use in research and daily clinical practice reflects and increases the attention towards the patients individual perspective on his/her disease and treatment. Valid, reliable and efficient PRO measures are the cornerstone for the standardised assessment of perceived symptom burden and treatment (side-) effects which contributes to symptom management and screening in the individual patient, to comprehensive treatment evaluation in oncological trials and to quality assurance in hospitals.