CORR Insights®: Poor Knee-specific and Generic Patient-reported Outcome Measure Scores at 6 Months Are Associated With Early Revision Knee Arthroplasty: A Study From the Australian Orthopaedic Association National Joint Replacement Registry

Where Are We Now?

Joint replacement of the hip and knee can eliminate patients’ pain and improve their function. Although many studies go into great detail, patient reports regarding this procedure can often be boiled down to “satisfied or dissatisfied” or “improved or worsened.” Perhaps that’s an oversimplification, but we can learn the most about these procedures by listening to our patients.

The evolution toward a patient-centered approach to surgical decision-making has been mirrored by the increasing use of patient-reported outcome measures (PROMs). This transition from surgeon-reported measures such as the Harris Hip Score or the old Knee Society Score to patient-reported scores such as the WOMAC, Hip Disability and Osteoarthritis Outcome Score, Knee Disability and Osteoarthritis Outcome Score, or University of California Los Angeles scores allows a more granular look at the patient’s experience from before surgery to after. National and other large joint registries that capture multisurgeon data also continue to evolve. Although these PROMs captured in large databases may allow use to better define what a good outcome is, currently, these measures are incompletely reported (only 16 of 39 registries at the local, institutional, regional, or multicountry level have obtained PROMs in one study on the topic [2]). Estimates of how large an improvement on one or another PROM that needs to be identified in order for a patient to perceive whether an improvement as clinically important also continue to be refined [3-5]. Altogether, it is fair to say that as of now, the promise of PROMs is real but not fully realized.

In this issue of Clinical Orthopaedics and Related Research®, Ackerman et al. [1] suggest that surgeons can use PROMs to evaluate success and anticipate impending failure. Using the Australian Orthopaedic Association National Joint Replacement Registry, the authors found that a high pain score or low function score collected starting 6 months after surgery is associated with an increased risk of revision surgery (the decision to obtain the first postsurgical PROM at 6 months was based on the recommendation of an international working group). This paper is both important and practical: The authors defined effect sizes clinicians can use to screen their patients for an endpoint—premature revision—that patients and their surgeons care about. For example, patients who had not achieved what the authors defined as clinically important improvements on any of several commonly used outcome tools by 6 months after surgery had substantially increased relative risks of early revision. Those risks increased anywhere from more than 2 to nearly 9, depending on the outcomes score in question. The consistency of the finding across outcomes scores suggests it is robust; the thresholds the authors defined make this study easy to use in practice.

Where Do We Need To Go?

There are four distinct yet important potential needs raised by the findings of Ackerman et al.’s [1] study. First, data capture completeness and clearly defined comparable data acquisition times and datasets among all data collection organizations have not been delineated, and we need to determine what data to obtain and how to use the data. The second need is one of clinical practicality. Knowing from a PROM that a patient is doing poorly (compared with a similar patient) is an important action trigger. In the first year after surgery, one would assume that the surgeon would be more likely to respond to the patient’s complaint, which would most likely be pain, than to a PROM metric (raising the sensible non-numeric clinical approach that if a patient tells the surgeon he or she does not see improvement and is dissatisfied, the doctor should listen). However, given variances in follow-up and patient compliance, the thresholds described by Ackerman et al. [1] are of great clinical practicality because these thresholds allow surveillance of all surgical patients. The third need is a simple statistical one: Which method, anchor-based or distribution-based, should be used? Finally, we need improved data security regarding who controls and uses the data.

How Do We Get There?

Although it is possible to address the needs identified above, doing so will require national and international consensus. Clearly, non-numeric data such as “satisfied or dissatisfied” or “improved or worsened” are potentially as important to our understanding of the benefit the patient perceives and receives from treatment as the derived minimum clinically important difference. Given the large number of aggregate PROMs, the sensible advice would be for an international working group to establish minimal clinically needed PROM scores and advocate for their use by data collection bodies. This might lower collection costs (less is likely more) and allow for direct comparison of separate registry body outcomes. I suggest using current data, sequentially removing individual PROMs, and examining whether that removal changes the findings. In determining which statistical method to use (anchor-based or distributive), although the distributive method is simpler, the anchor approach better represents the patient experience. The collection and use of PROMs must be relevant to the patient experience. Adoption of the anchor method by all registries would facilitate comparisons by reporting entities. Overall, these improvements must be surgeon-led (we must control the data). We all recognize the value and danger of large data. Large datasets aid in identifying and understanding subtle differences between groups of patients, and these subtle differences can be consequential. Exploration of smaller details (individual PROMs) that are combined to yield a patient report of “satisfied or dissatisfied” could be highly important and lead to an understanding at the level of each patient. As mentioned by the authors [1], their article only pointed to PROM findings that indicate a potential problem. In the future, specific causes of revision may be presaged by PROM-level findings associated with those etiologies or modes of failure. As a caveat, though, these data should only be used to inform the patient (to aid in their presurgical decision-making) and the surgeon (for improvement in surgeon performance). These data should never be used to punish the surgeon by equating the findings to perceived beneficence (that is, attentiveness, compassion, and kindness) or lack thereof (dismissiveness, callousness, or meanness), with attached monetary incentives or disincentives.