Choosing Patient-Reported Outcomes and Measurement Methods for Team Based Health Care
March 7, 2018 | Integrated Health Care Conference, Arizona State University Doctor of Behavioral Health Program, Scottsdale, AZ
In a keynote lecture, Dr. Ware provided an overview of the history of the conceptualization and measurement of patient-reported outcomes (PROs), and noteworthy methodological advances and their implications for going forward. Applications of item response theory (IRT) methods have improved the quantification of health-related quality of life (QOL) domains and are the basis for more efficient computerized adaptive test (CAT) survey administrations and scoring. However, a consequence of the development of very homogeneous survey item banks preferred by IRT models is a conceptual shift favoring measures of very specific symptoms (e.g., depression) and activities (e.g., walking) as opposed to broader concepts of mental, physical and social functioning and well-being. In contrast, an emphasis on summary measures – the “tips of the QOL icebergs” that enable more efficient adaptive approaches – makes it practical to drill down to measure specific limitations in QOL only when are more likely to occur and to adapt automatically to multiple chronic conditions when they are present. This adaptive approach is the most practical and useful way represent PROs in “big data.”
Dr. Ware presented examples of published and forthcoming JWRG findings from evaluations of new generic “super” short-form items that improve survey efficiency over legacy tools and use standardized IRT-based metrics to cross-calibrate new and legacy PROs. New, more valid and responsive disease-specific PROs were also discussed. The latter QOL Disease Impact Scale (QDIS) measures, which are standardized across diseases and norm-based, yield a more global summary score that fills the gap between disease-specific symptoms that are not QOL and generic QOL measures that are not disease-specific. Once conceptualized and quantified, such higher-order (summary) concepts can be estimated more efficiently using fewer items in deciding when to drill down. Adaptive survey logic software, that automatically adapts to the presence of limitations in generic QOL and to the presence of multiple chronic conditions, was described as a more practical solution to integrating disease-specific and generic measures into a common “dashboard” of PROs.
In summary, Dr. Ware emphasized that:
- Comparisons of effectiveness across applications require QOL measurement standardization,
- To make results more actionable, generic and disease-specific PROs must be displayed on the same dashboard,
- To make data collection more practical, a new generation of single-item-per domain and single-item per disease super-short-forms are required to manage adaptive measurement,
- Improvements over legacy (Medical Outcomes Study, PROMIS, utility) items for common domains have been linked to single-item measures with:
- Broader representation of descriptive content,
- Response categories covering a wider range,
- Direct measurement of higher-order concept (“tip of the iceberg”),
- Better matching of operational definitions with the essence of a generic domain or disease-specific experience.
In conclusion, a more aggressive adaptive measurement system that monitors QOL at a higher conceptual level with “super SF items,” drills down to quantify specific symptoms or activities when necessary, and automatically adapts to the presence and severity of multiple chronic conditions is possible and is required for team-based health care.