In 2004, the NIH put forth its Roadmap for Medical Research, part of which focused on improvements in patient reported outcome measures. Out of this came the Patient-Reported Outcomes Measurement Information System (PROMIS), a series of instruments focused on measuring physical, mental and social health. While the NIH and other groups have been supporting the use of the PROMIS instruments, there is sparse literature regarding their use in the spine population. The PROMIS theoretically provides better psychometric properties compared to many of the widely used patient reported outcomes questionnaires, including improved reliability, fewer floor and ceiling effects, and less differential item functioning (i.e. subgroups with similar health states responding differently). While physical, mental, and social health are all important in spine patients, the PROMIS domains of physical function, pain intensity, and pain interference seem especially germane for this population. Given the interest in this new patient reported outcome measure, Dr. Brodke and his colleagues from Utah evaluated the psychometric properties of the PROMIS physical function item bank in 438 patients with neck or back problems. While the PROMIS physical function questions can be asked as part of a variety of short forms or with computer adaptive testing, the authors had every patient respond to all 124 items. They found the physical function items to be reliable, unidimensional (i.e. measuring a single domain, physical function), and with minimal floor and ceiling effects. The only shortcoming they noted was a tendency for differential item functioning, meaning that different subgroups responded to the questions differently (i.e. older patients responded differently from younger patients, back patients responded differently from neck patients).
With the NIH and other stakeholders strongly supporting the PROMIS initiative, it is good to see it being evaluated in the spine population. Given the results of this study, it seems as though it has excellent psychometric properties for spine patients. One of the limitations of this paper was that direct comparison to other patient reported outcome measures was not performed. Hopefully future work will determine how well the PROMIS measures correlate with traditional spine outcome measures such as the Oswestry Disability Index and Roland-Morris Disability Questionnaire. Additionally, it would be interesting to directly compare the psychometric properties of these questionnaires within the spine population. The PROMIS has the advantage of potentially using computer adaptive testing in order to arrive at a very precise measurement using a minimal number of questions, which could reduce questionnaire fatigue and improve response rates in clinical studies. It should also allow for comparison of health states across diseases, something not possible with spine specific outcome measures. While there is much work to be done to determine if the PROMIS is an improvement over the currently used spine outcome measures, it seems to be here to stay. One of the challenges to the spine community will be to determine the best set of outcomes measures to use that both optimizes psychometric properties of the questionnaires and also allows for comparison to past research. The entire field of patient reported outcome measures is still relatively young, so it will likely continue to evolve rapidly. Hopefully spine researchers will be able to use these new outcome measures effectively without creating a body of literature with so many disparate metrics that comparison across studies becomes impossible.
Please read Dr. Brodke’s article on this topic in the January 15 issue as well as his accompanying commentary on The Spine Blog. Does this article change how you feel about the PROMIS outcome measure? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor