Discussion

Summary of findings

The PSR for primary TKA in patients with end-stage HA was 92.2% at 15 years and 90.9% at 25 years. The PSR for primary THA in patients with end-stage HA was 91.0% at 15 years and 78.6% at 25 years. In TKA, PJI during follow-up, which was more prevalent in patients with an inhibitor, was a strong predictor of prosthesis failure. In THA, cementation of the femoral stem, which was associated with a higher age, showed a trend in predicting prosthesis failure.

Comparisons with the literature

The 15-year PSRs for TKA and THA in our cohort of patients with bleeding disorders are similar to those of the general Dutch population (92.8% for TKA and 91.1% for THA) [26] and substantially higher than those previously reported in hemophilia populations (80%-84% for TKA and not available for THA) [8,14]. The hemophilia cohorts in previous studies are similar to the present cohort of people with bleeding disorders in terms of demographics, such as severity and inhibitor status, as well as peri- and postoperative coagulation protocols. Also, bleeding rates were similar, 32% vs 34%. Conversely, in comparison to the general Dutch population, our cohort of people with bleeding disorders consisted of younger, predominantly male patients who more often had multijoint involvement. This highlights the characteristics of HA, which starts at an early age and severely affects multiple joints. Besides demographic differences, infection rates and bleeding rate also differed between our cohort of people with bleeding disorder and the general population: 6% vs 1.2% in TKA and 3% vs 1.5% in THA for infections and 25% vs 3% for bleeding [39,40].

PJI and the presence of an inhibitor were significant predictors for prosthesis failure in TKA. In previous research, these factors were also associated with adverse outcomes. PJI in particular is known as the most frequent complication after TKA and the most common reason for revision [41]. In the general population, PJI is known to be more prevalent in young, male patients with preoperative anemia, prolonged operation duration and hospital stay, postoperative transfusions, and bilateral procedures [41]. All of these are characteristic features of people with bleeding disorders compared with the general population. The effect of having an inhibitor at the time of surgery is described in a few other studies, mostly in case reports, and did not lead to strong evidence or consensus on higher complication rates or shorter survival [42]. Therefore, the significant, higher risk for prosthesis failure in patients with an inhibitor in our study could possibly be explained by these patients having higher risk for PJI through the described risk factors and the higher risk for (subclinical) bleeding.

In THA, age has not been described as a predictor of prosthesis failure in previous hemophilia research. In the general population, a younger age is associated with an increased risk of revision surgery [43]. In the present study, however, a higher risk of prosthesis failure was observed among older patients and patients receiving cemented femoral stems. Recent studies also show improved survival rates in cementless hips, potentially due to improved osteointegration [44]. Cemented stems are more frequently used in older patients and in those with poor bone quality. Moreover, during the study period, cementation of the femoral stem was standard practice for all THA patients aged >65 years. Consequently, the observed association between cemented stems and PSR appears to be mediated by age, as demonstrated in the multivariable analysis. These findings suggest that cementation is a surrogate marker for age and bone quality, while also explaining why older patients have a higher risk for failure in this cohort of people with bleeding disorders: higher age correlates with worse bone quality, especially in individuals with bleeding disorders.

Higher PSRs compared with previous hemophilia populations and the reduction in annual bleeding rate after 2000 may indicate that outcomes improved over time due to advances in hemophilia care. However, year of surgery was not a significant predictor of prosthesis failure in this study, and, in contradiction with Fenelon et al. [8], the present study showed no decrease in long-term complication rates after 2000 compared with before 2000 (P > .05). Lastly, recent studies suggest that modern biomaterials, including oxidized zirconium femoral components with highly crosslinked polyethylene used in TKA and ceramic femoral heads in THA, may contribute to improved prosthesis survival. However, these factors were not included in our analyses and therefore could not be assessed in the present study [32,44]. As modern biomaterials were unavailable during the earliest inclusion period, participants enrolled at that time did not receive these prostheses. Consequently, prosthesis survival may improve further with the increasing use of these newer materials.

Strengths and limitations

One strength of the study is that it contains rich, single-center data of high quality due to standardized treatment, thorough registration through the years, and complete follow-up of every patient for lifelong care. Furthermore, the use of a strict inclusion criterion (factor levels <5%) ensured that only patients with defined HA were included. Another strength of this study is that we matched our endpoint (first revision) to the large Dutch Arthroplasty Register (LROI) osteoarthritis database, in combination with a clear separation between end-stage HA patients and osteoarthritis patients in the general population, allowing a strong comparison. Lastly, the high methodological quality of the study is confirmed by the fact that multiple sensitivity analyses gave very similar results. A limitation of our study is the limited statistical power of the multivariable analyses due to the low number of prosthesis failures. In the TKA model, 4 variables were included despite the rule of thumb [45,46] suggesting a maximum of approximately 2 variables. In the THA model, 2 variables were included, whereas the number of events would ideally only support univariable analyses. Nevertheless, the multivariable models provide additional clinical insight, although their results should be considered exploratory, and replication studies with different cohorts are needed to confirm the results. Lastly, although missingness was generally low, this was not the case for the BMI variable. Specifically, BMI data of an earlier date (before 2007) were largely missing due to the transition to electronic patient records (before 2007). As a result, the findings of the corresponding univariable analyses may be biased. However, these results were not statistically significant and do not affect our key message.

Clinical relevance

Given that current PSRs are comparable to those in the general population, TJR in people with bleeding disorders appears more feasible and should be approached with less concern regarding failure. Healthcare professionals can therefore provide more deliberate advice to patients by directly incorporating the identified predictors of failure into their consultation, especially by using results from the post hoc analysis with only preoperative variables. As people with bleeding disorders are typically younger and present with severe joint damage, these PSRs help address the clinical dilemma of whether to delay arthroplasty. Our findings support considering joint replacement at a younger age when clinically indicated. Earlier surgery may in particular be beneficial for THA, as it increases the possibility of using an uncemented femoral stem, and bone quality appears to play an important role in prosthesis survival, given that loosening is the primary cause of failure in THA at longer follow-up. In TKA, additional anti-infectious measures could be used, especially in patients who also have an inhibitor at time of surgery, as early failure is mostly infection-related, and late failure due to loosening is rare in this population.

Recommendations for future research

The present study reports reliable and long-term PSRs for TKA and THA in people with bleeding disorders. However, functional outcomes were not considered. Functional outcomes have been assessed in previous studies and appear promising, although these studies share the same limitations as the PSR literature: long-term data, extending beyond 2 decades and reflecting developments in hemophilia care, are lacking. Therefore, future research should use the data of the present cohort and connect it to functional outcome measures. This would further strengthen TKA and THA as earlier interventions and effective therapeutic options for improving quality of life in patients with end-stage HA.