Patients with AS achieve 2-year outcomes for pain and function that are clinically similar to patients with OA despite significantly worse preoperative pain and function and more comorbidities, factors recognized as significant risk factors for poor outcome in OA. Patients with AS achieved significant improvements in pain and function in a manner similar to other patients with chronic rheumatic diseases, for whom baseline status does not predict a poor 2-year outcome. We have reported the same pattern for patients with rheumatoid arthritis undergoing total knee replacement and those with systemic lupus erythematosus undergoing THR and total knee replacement, where significantly worse preoperative pain and function were not predictors of poor outcomes. Importantly for patients with AS, they are no more likely to have poor pain or function and are as likely to achieve clinically meaningful improvements in both pain and function as OA at 2 years. One can speculate that patients with chronic rheumatic diseases are better able to cope with the demands of painful postoperative physical therapy and are therefore able to achieve pain and function outcomes equivalent to those of patients with OA.
These results should be considered in the context of certain limitations, such as the 63% 2-year response rate for patients with AS, with nonresponders having worse baseline WOMAC function and higher expectations of surgical success. Others have also reported low response rates on follow-up questionnaires in patients with AS. Patients with poor baseline status and diseases are less likely to respond to questionnaires compared with others. However, given our extensive self-report data at baseline and our large numbers, we were able to additionally analyze our outcomes using multiple imputations for the missing data, increasing the validity of our analysis and decreasing the systematic ascertainment bias seen when follow-up is low. "Bootstrapping" is a more sophisticated method for imputing missing data than more commonly used methods such as last value carried forward or simply imputing the mean. Our analyses were conducted with the imputed data set with systematically worse outcomes for pain and function at 2 years for the nonresponders. Using imputed data showing poorer outcomes introduces a conservative bias for AS, increasing the validity of our positive results. Even using this conservative approach, AS was not an independent risk factor for poor pain or function after THR.
There are few existing studies to inform decision making by patients with AS who are contemplating THR. A large retrospective series compared function in 340 patients with AS who had undergone THR to patients with AS from the same cohort who had not undergone THR and reported that overall function was poorer for the patients with AS who had undergone THR. However, it is difficult to extrapolate from these results because of the unavoidable confounding by indication in a retrospective case series, as in other published series. By gathering data on our patients prospectively and comparing them to matched patients with OA, we were able to avoid this issue. Importantly, we were also able to control for back pain. As severe spine disease commonly accompanies severe hip disease, it is important to include this important potential confounder in outcome studies of THR in AS. Despite the high prevalence of clinical back pain and radiographic sacroiliitis in our cohort, back pain was not significantly associated with poor outcome after THR.
Additional retrospective studies have described excellent THR outcomes for both pain and function for AS. However, these studies were uncontrolled and included patients operated on over a 30-year period. Given the tremendous advances in surgical techniques, anesthesia, prosthesis design, and AS therapeutics, data from 30 years ago have minimal relevance to today's THR patients.
The presence of more comorbidities has been associated with poorer THR outcomes but that was not the case with our series of patients with AS. Patients with AS had overall worse health as estimated by the ASA score, but this was not a risk factor for worse postoperative pain or function. Poor preoperative function has also been described as a risk factor for poor functional outcome in OA. For our cohort, better preoperative function decreased the likelihood of a clinically significant improvement in function. Why this ceiling effect was observed is unclear.
Despite the significant improvement in WOMAC pain and function scores, patients with AS continue to feel more limited by their health status compared with OA, with significantly worse SF-12 PCS scores both at baseline and persisting at 2 years. Helping patients with AS have realistic expectations of THR is an important aspect of preoperative counseling. In fact, poor preoperative health-related quality of life on the baseline SF-12 PCS score was the most significant risk factor for poor pain at 2 years.
High-volume centers are associated with better outcomes after THR, so our results may not be generalizable to low-volume centers, where most THR surgeries are performed. However, this would not affect comparisons with control cases from our own institution. In addition, our short period of follow-up does not permit us to assess implant survival or long-term outcomes.
Strengths of this study include a well-validated cohort of patients with AS. A sophisticated imputation technique was used to control for missing outcome values. Use of a contemporaneous control group undertaking the same orthopedic procedure and matched for other potential confounders also lends strength to our observations. In addition, our cases were operated on within the short time frame of 2007 to 2010, making confounding due to secular changes in orthopedic and anesthetic techniques and implant design less likely.
Total hip replacement is an effective treatment for patients with AS. However, AS had lower baseline and 2-year SF-12 PCS scores than matched OA control subjects, which suggests that health-related limitations persist after hip replacement surgery. These data provide important information for contemporary patients with AS undergoing THR and can be used to ensure patients with AS have accurate expectations of hip arthroplasty.