Patients want to be healthy. They want to live longer, function better and have higher wellbeing. How patients feel is best measured by asking them. Few cancer patients these days have not been asked to fill out standardized questionnaires about their symptoms, functioning, and well-being. While these patient-reported outcomes (PROs) are not new to healthcare or clinical trials, their use to drive treatment decisions is relatively new and relatively scarce. How should the scores be interpreted? What do high and low values mean? How can we change treatment to affect these outcomes?
PRO scores can be challenging to interpret for both patients and physicians. Figuring out how to act on the results can be even more challenging. A new toolkit published in Medical Care provides some answers–different methods for interpreting the scores and different ways for acting on the results. A handy table describes the systems referenced in the toolkit [image_download]. Just looking at the table gives you an idea of the diversity of measures, frequency of collection, and target patients in cancer care.
Interpreting PRO scores is the first issue addressed in the supplement. Because of the diversity of measures, there isn’t any consistency in how scores are scaled or their directionality. Physicians and patients aren’t likely to want to spend hours figuring out if individual scores are relatively good or relatively bad. Authors in the supplement convened a panel of experts to brainstorm solutions to this problem. A few recommendations emerged from this Delphi panel for PRO measures:
- Provide descriptive y-axis labels.
- Indicate scores that are concerning.
- Provide scores for reference populations.
Anyone who has looked at the various PRO measures knows that these things are easier said than done. Often the threshold values, descriptions, and reference values are lacking–often because we just don’t know. Researchers could do more to improve measure interpretability in their own work, but collaboration is necessary to provide meaningful descriptions and data to create reference values. Several papers in the supplement address different ways to develop cutpoints, link scores to useful descriptions (and reference values) and improve understanding of changes over time. Additional articles focus on how to interpret the scores in the context of decision-making in clinical practice and whether methods used to create group interpretations can be applied to individuals.
Of particular interest to practitioners, the remaining 8 articles cover different systems that were developed to collect PROs and use them to aid decision making. These practical examples show many different ways of collecting PROs, how they can be added to existing workflows, and lessons learned. Just developing standardized PROs is not enough. Helping physicians and patients learn how to interpret and use the scores is the only way they could affect patient care.
Jess Williams