Fig. 1 Illustration of the model developed to predict two-year functional outcomes after total shoulder arthroplasty
Courtesy of Thomas W. (Quin) Throckmorton, MD, FAAOS


Published 9/2/2021
Rebecca Araujo

Model Predicts Two-year Functional Outcomes after Shoulder Replacement

Using a new model incorporating patients’ preoperative characteristics, clinicians were able to predict postoperative pain and functional outcomes after total shoulder arthroplasty (TSA), according to results of a pooled analysis from three high-volume centers. The poster will be on display today and tomorrow in Academy Hall.

“Clinical outcomes after shoulder arthroplasty can be highly variable given the heterogeneity among patients. One of the most important aspects of patient selection for shoulder replacement is to provide patients accurate information regarding their potential outcomes,” presenting author Thomas W. (Quin) Throckmorton, MD, FAAOS, of the University of Tennessee–Campbell Clinic, told AAOS Now Daily Edition.

In this study, Dr. Throckmorton and colleagues sought to create and validate a tool “that surgeons could use to counsel patients preoperatively about what they can expect following the operation. In that way, reasonable expectations can be set and managed.”

To develop their model, intended to predict two-year scores on the American Shoulder and Elbow Surgeons (ASES) tool, the researchers pooled data on 1,947 patients. Twenty-seven preoperative variables were included, including age, sex, race, BMI, comorbidities, employment status, social history, preoperative visual analog score (VAS) for pain, ASES score, preoperative diagnosis, and arthroplasty type. Those factors were evaluated for their association with two-year postoperative ASES scores. Any variables with a significant association (P <0.20) were included in the model (fig. 1)>

The following variables had the strongest associations with two-year postoperative outcomes: age, preoperative ASES score, presence of a disability, chronic obstructive pulmonary disease (COPD), alcohol use, anatomic versus reverse TSA, and primary versus revision TSA.

“As expected, disability claims and COPD correlated with worse outcomes in our study,” Dr. Throckmorton said. His team was surprised to find, however, that alcohol use—not abuse or dependence—was associated with a higher ASES score.

The model was externally validated in a cohort of 233 patients. After external validation, predicted versus actual ASES scores at two years differed by a mean 12.7 points, which fell within the prespecified accepted minimal clinically important difference (MCID) of 21 points after TSA. Specifically, the model accurately predicted two-year ASES scores within the MCID for 85 percent of patients.

“Our model provides an easy-to-use clinical tool that surgeons can use to set reasonable expectations for patients regarding their potential outcome after shoulder arthroplasty,” Dr. Throckmorton concluded. He added that he and his colleagues partnered with the AAOS Registry Program to offer this tool as a member benefit to participants in the Shoulder and Elbow Registry.

The study authors noted the limitations of predictive modeling. “While we tried to provide robust data by pooling outcomes from three high-volume shoulder replacement centers to create the model, the results may not be generalizable to all practices and settings,” Dr. Throckmorton explained.

Another limitation of this analysis is that variables outside of the 26 included in this model analysis may impact patient outcomes, and this predictive model could not account for those factors. “This study examined patient-specific risk factors, many of them related to general health, as outcome predictors after shoulder replacement,” said Dr. Throckmorton. “However, anatomic and surgical factors such as glenoid deformity, fatty infiltration of the rotator cuff, and deltoid volume provide other potential areas of study to refine and improve the model.”

The study will be on display as P0778 today and tomorrow in Academy Hall, Sails Pavilion, from 7 a.m. to 5 p.m.

Dr. Throckmorton’s coauthors of “Development and Validation of a Predictive Model for Outcomes in Shoulder Arthroplasty: A Multicenter Analysis of Over 2,000 Patients” are Tyler J. Brolin, MD; Saunak Sen, PhD; Lawrence V. Gulotta, MD; Surena Namdari, MD; Howard D. Routman, DO; and Frederick M. Azar, MD.

Rebecca Araujo is the associate editor of AAOS Now.