For quality of life (QOL) data to be useful in the clinical management of prostate cancer, they must be accurate, responsive to meaningful change and interpretable. Several QOL and symptom scales have been developed for men with prostate cancer but these scales are clinically underutilized, in part because they have not been constructed for individual diagnosis. Individual diagnosis requires precise measurement of well-defined constructs. Computerized adaptive testing (CAT), with item selection algorithms using item response theory (IRT) is one way - perhaps the only feasible way - to precisely measure QOL and its component dimensions. We propose to bring this technology to the measurement of prostate cancer quality of life (P-QOL). This project has two specific aims: To construct and extend an item bank for P-QOL outcomes; and to develop and pilot P-QOL CAT in clinical settings with prostate cancer patients. To accomplish these aims, we will conduct three projects designed to employ cutting edge measurement science and computer technology. Project 1: Use IRT models to construct an initial item bank of questions measuring urinary function, sexual function, bowel function, pain, appetite/weight, MWB, MWB, and overall QOL. Project 2: Using clinical experts, measurement experts, and patients, acquire items into the bank to fill current item difficulty gaps in measured dimensions and overall QOL. These new and rewritten items will then be administered to men with prostate cancer. Expert review of the summarized data will allow us to refine items in the bank and create new sub-banks as indicated. Project 3: Develop a CAT platform and pilot the use of CAT with prostate cancer patients in clinical settings. This will include a determination of the effect of item context, and measurement precision by comparing CAT-derived measures with measures obtained using a fixed subset of items drawn from the bank, and evaluation of preliminary data on sensitivity to meaningful change.
Showing the most recent 10 out of 209 publications
