Challenges using patient reported outcomes for functional abilities and quality of life in older adults

Tags: Oral
Lin J1, Whitlock E2, Eckstrom E3, Perdue L1
1Oregon EPC, Kaiser Center for Health Research, Portland, Oregon, United States, 2Kaiser Center for Health Research, Portland, Oregon, United States, 3General Internal Medicine, OHSU, Portland, Oregon, United States

Background: Research in older adults increasingly focuses on important patient reported outcomes (PRO). While these outcomes often better capture net health impact than traditional disease-specific outcomes, they pose methodological challenges for systematic reviewers. Objective: To describe methodological challenges for reviewers of primary evidence using health-related quality of life (HRQOL) or functional status measures in older adults. Methods: We conducted a systematic review of multi-factorial assessment and management strategies in older adults to prevent functional decline. We describe the challenges in measurement properties, comparability, and standardization in conducting meta-analyses for PRO. Results: Among 62 RCTs, 41 self-reported measures of functional status (20 ADL, 13 IADL, 8 combined ADL/IADL) were reported. Instruments were occasionally misclassified or not reported; measurement properties for instruments (e.g., development population, validity, independent validation, responsiveness to change) were never reported. Population considerations further complicated outcome interpretation, as measures of functional status and HRQOL have different measurement properties in different risk populations and approaches to defining and reporting patient risk across trials were inconsistent. Inability to determine when measures are appropriate hampers separating true null effects from measurement error. Pooling functional status or HRQOL measured by multiple instruments required use of unitlessmeasures and assumptions that measurement tools cover the same domains. Hand-checks of domains addressed by instruments illustrates that these assumptions may be questionable. Inconsistent reporting of data and issues of directionality required substantial data manipulations to allow pooling. Finally, there were challenges in the interpretation of clinically meaningful differences. Although methods on how to define clinically important differences in PRO exist, application of these methods was complicated by the pooling of multiple different measures, differences in participants’ baseline risk, and the variability in the natural history of functional decline. Conclusions: We caution systematic reviewers in pooling PRO measures in meta-analyses across studies of older adults without paying attention to the risk status of populations, validity of measurements, measurements properties in different risk populations, and the ability to interpret the clinical significance of pooled estimates.