Grant: Operating Grant - PA: Regional Partnership Program - Saskatchewan

Self-reported measures of health outcomes like quality of life, satisfaction with treatment, functional ability, and disease symptoms, can contribute to improved decision making in a patient-centred healthcare system. Insights into patients' own experiences with their health can be obtained by using these measures. However, there are a number of challenges associated with the development and analysis of patient-reported outcome measures. In longitudinal studies in which the outcomes are repeatedly measured over time, response shift, a form of measurement bias, can result in incorrect conclusions about change. Previous research has shown that when response shift is present in a set of data, it will often lead to the conclusion that no change exists, when in fact there is a real change within the patient population. Our research will develop statistical methods to detect the presence of response shift in longitudinal patient-reported outcome measures. These methods will be evaluated and compared using computer simulation techniques, to better understand how they perform under different data-analytic conditions. Then, we will apply them to existing datasets to test for response shift in patient populations with stroke and osteoporosis-related fractures, two health conditions that can result in substantial change in quality of life and patient functioning. This research will benefit clinicians and measurement scientists as they develop new patient-reported outcome measures. It will also contribute to improving the validity of studies about the effects of new treatments and healthcare interventions on patient outcomes.