We will be performing site maintenance on AAOS.org on February 8th from 7:00 PM – 9:00 PM CST which may cause sitewide downtime. We apologize for the inconvenience.

Fig. 1 Preoperative PROMIS score versus PROMIS score improvement: Patients below the calculated cutoff of 29.7 who met minimally clinically important improvements (MCID) are highlighted in blue, while patients above the calculated cutoff of 42 who did not meet MCID are highlighted in red.
Courtesy of Bryant Ho, MD.


Published 9/1/2016
Terry Stanton

Study: Preoperative PROMIS Scores Can Predict Outcomes for Foot and Ankle Patients

Those with lower function and higher pain benefit most from surgery
A study that examined how well preoperative patient-reported outcome scores would predict postoperative improvement after foot and ankle surgery found that patients with low preoperative function and high preoperative pain experienced outcomes superior to those with relatively high function and lower pain.

The authors of the study, which was the 2016 recipient of the J. Leonard Goldner Award at the annual meeting of the American Orthopaedic Foot & Ankle Society, say the results for this use of PROMIS (Patient Reported Outcome Measurement Information System) scores will help physicians and patients make informed decisions in treatment planning, specifically in choosing surgery versus nonsurgical management.

PROMIS scores "have been shown to be validated in the foot and ankle," said Bryant Ho, MD, of the University of Rochester, who presented the study. "They have been shown to be efficient and better than previous scales, with less floor/ceiling effect. Previous studies have shown that preoperative scores can predict outcomes in total hip replacement, total knees, and total shoulders. Therefore, we hypothesized that preoperative PROMIS scores can predict outcomes in foot and ankle patients and also can help us identify who will benefit from surgery."

Specifically, the authors hypothesized that all PROMIS scores would improve from preoperative to postoperative assessment and that preoperative scores for physical function, pain interference, and depression would accurately predict the extent of that improvement.

PROMIS, Dr. Ho explained, is a patient-reported outcome measure that employs computerized adaptive testing based on item response theory to decrease administration time while maintaining accuracy. It uses a T-score from zero to 100, with a mean of 50 and a standard deviation of 10. "For physical function scores, the higher the better, while for pain interference and depression, the lower the better," he said.

In setting up the study, the authors defined success as the minimal clinically important difference (MCID)—the smallest improvement after surgery that a patient would define as important. "For our purposes," Dr. Ho explained, "it was an improvement in the PROMIS score of approximately 4."

The study examined 16,023 visits to a foot and ankle clinic from February 2015 to April 2016. This included 7,996 patients from whom 3,611 new patients were selected. The investigators excluded patients with less than 7 months' follow-up, those with diagnoses and procedures not classified as related to the foot and ankle, those with nonelective procedures such as infections or fracture, and patients with missing data. This left a final cohort of 61 surgical patients who completed all PROMIS domains.

The most common diagnoses were Achilles tendonitis and posterior tendon dysfunction. The most common procedures were Achilles débridement, ankle arthroscopy, and flatfoot correction.

The authors applied receiver operative characteristic curves (ROC curves) to the results to determine how predictive preoperative scores were for achieving the MCID after surgery. "Then we defined cutoff PROMIS scores that predicted whether or not patients would achieve MCID," Dr. Ho explained. "We calculated our cutoffs for patients that would meet MCID, and we wanted to calculate a cutoff that would provide a 95 percent specificity for identifying patients that met MCID."

ROC curves demonstrated that preoperative physical function scores were predictive of a meaningful postoperative improvement in physical function. Patients with a preoperative physical function T-score below 29.7 had an 83 percent probability of achieving a clinically meaningful improvement in function as defined by MCID. Patients with preoperative physical function T-score above 42 had a 94 percent probability of not achieving MCID (Fig. 1). Patients with preoperative pain above 67.2 had a 66 percent probability of achieving MCID, while patients with preoperative pain below 55 had a 95 percent probability of not reaching MCID. Patients with preoperative depression scores below 41.5 had a 90 percent probability of not achieving MCID.

The authors commented that the patients they were treating surgically "had significant physical detriments and pain limitations compared to the normal population." Their average T-score for physical function was 34.2, and for pain interference it was 61, "indicating significantly low function and high pain levels." In contrast, they noted the mean PROMIS depression score was 47.8, "indicating minimal mean difference in depression levels in our patients with foot and ankle problems compared to the U.S. population."

The study's finding that foot and ankle patients with low preoperative function and high pain experienced superior outcomes to those with high function and low pain echoed results reported for anterior cruciate ligament reconstruction and arthroplasty for hip, knee, and shoulder.

The promise of PROMIS
This information can directly be applied to patient care, the authors wrote. According to Dr. Ho, "Preoperative patient reported outcomes [PROMIS] can help us predict who will benefit from surgery. Patients with high pain and low function are likely to benefit from elective foot and ankle surgery, while patients with low pain and high function are not likely to benefit. Furthermore, our reported cutoff PROMIS scores can provide pre- and post-test probabilities for success after surgery. Patients come to us to ask, 'Am I a good candidate for surgery?' or 'Will I benefit from surgery?' and now we have answers to both of these questions for them."

Dr. Ho explained that the study evolved from the practice at the University of Rochester orthopaedic service to "collect prospective data for every patient, every day, every visit, as the standard of care. We share this information with the patient during the visit, and this aids us in jointly deciding on treatment plans. It was a natural progression to examine if the initial PROMIS scores could indicate if a patient would benefit from surgery and if not, then other treatment options would be implemented."

Dr. Ho said that he and his colleagues expected initial function and pain scores to correlate with how patients fared after surgical treatment, "but we were surprised at how strong the relationship was." He continued, "We also expected depression scores to have a high impact on postoperative pain and function, which has previously been shown in the joint replacement literature. We were surprised that this was not the case in our patients."

The same PROMIS assessments used in the study are expanding across the entire University of Rochester Medical Center, Dr. Ho noted, and currently are being used as the standard of care for shared decision making in more than 30 departments/divisions by some 300 providers. "These assessments populate the same graphic display in our electronic record, and this means that the patient is getting follow-up data well beyond the time that we see the patient in orthopaedics. This allows long-term follow-up to be obtained. This is a great solution to a big challenge we have had in the past—long-term follow-up."

Patient-reported outcomes are evolving, Dr. Ho said. "They can be much more than just a research tool. They can help guide our treatment options. They can provide expectations for surgery. They can maximize healthcare resources by identifying who benefits from surgery and who does not. Our results have changed the way we practice. We now review the patient-reported outcomes with the patients to help them make informed and educated decisions together with the physician for surgery and also for nonsurgical care options such as physical therapy and orthotics."

Limitations to the study include potential selection bias, the follow-up length, and a wide spectrum of diagnosis, in addition to the generalization of the results beyond the patient population in the study. "Patients are not routinely asked to return at 1 year if they are satisfied with their outcome," the authors wrote. "Routine follow-up at 1 year is planned for future studies." The selection of only those patients with complete data sets might also be bias, the authors noted. "Other prognostic variables may also further refine clinical decision making, including demographics such as age and sex, or preoperative risk factors such as diabetes and ASA class. All of these limitations are opportunities for further stratification of prognostic patient-reported outcome data," they added.

Future studies, Dr. Ho said, might include gathering of longer follow-up data and breaking down the analysis into specific cohorts by diagnoses and procedures. "This information also helps us look at variations in treatment related to outcomes such as cost, provider, and other services. We have a lot of exciting research in front of us to benefit our patients and provide a healthier population."

Dr. Ho's coauthors are Jeff R. Houck, PT, PhD; Adolph S. Flemister, MD; John P. Ketz, MD; Benedict F. DiGiovanni, MD; and Judith F. Baumhauer, MD, MPH, MS.

The authors' disclosure information can be accessed at www.aaos.org/disclosure

Terry Stanton is senior science writer for AAOS Now. He can be reached at tstanton@aaos.org

Bottom Line

  • Preoperative PROMIS scores have been shown to be predictive in outcomes for total knee, hip, and shoulder arthroplasty.
  • This study of patients undergoing foot and ankle surgery found that those with high preoperative pain scores and low physical function scores reported greater improvements from surgery than those with initial low pain or high function.
  • Preoperative scoring can serve as a useful prognostic tool in treatment planning, including selection of surgical patients.