Prosthetic Joint Infection After Hip or Knee Arthroplasty: Treatment trends and outcomes

Prosthetic joint infection (PJI) is a feared complication after joint replacement surgery. The incidence of PJI has been reported to range between 1% and 2%. Prosthetic joint infections are treated with revision surgery, where the infectious tissues are debrided, and prosthetic components are removed or exchanged either partly or totally. In general, PJIs are associated with multiple surgeries, inferior patient-reported outcomes, as well as increased comorbidity and mortality.

The aims of this dissertation were to examine the epidemiology of PJI and trends in the treatment of patients who have the infection, to compare different treatment strategies, and to examine how outcomes after revision surgery due to early PJI can be better predicted.

The data of this study were collected retrospectively from electronic patient records at the Coxa Hospital for Joint Replacement for the period January 1, 2008, to September 12, 2021. Patients were identified by searching the ICD-10 (International Classification of Diseases 10th revision) code T84.5 (Infection and inflammatory reaction due to internal joint prosthesis). Thereafter, the 2013 International Consensus Meeting diagnostic criteria were applied to confirm the diagnosis of PJI. Infections were further classified as early, acute hematogenous, and chronic.

In studies I and III, descriptive statistics were used to examine the temporal trends and epidemiology of PJI. In studies II and IV, the Kaplan-Meier method and cumulative incidence functions were used for the survival analyses. In addition, Cox proportional hazards regression and Fine-Gray regression were used for regression analyses in studies II and IV. To examine the prediction of failure after early PJI, logistic regression and decision-curve analysis were used for the statistical analyses in study V. In studies I to IV, STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) guidelines were followed for the reporting of the results. In study V, the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) guidelines were followed in the reporting of the results.

The incidence of early PJIs of the hip increased from 0.11 per 100 primary THAs in 2008 to 1.09 in 2021. Among PJIs of the knee, no clear trends were observed, but the yearly changes in incidences were large. Among patients with PJI of the hip, the comorbidity burden increased during the study period. However, among patients with PJI of the knee, no change in the comorbidity burden was found.

The use of one-stage revisions increased remarkably during the study period. The incidence of one-stage revisions increased from 0.10 per 100 primary total hip arthroplasties (THAs) in 2010 to 0.91 per 100 primary THAs in 2021. A similar trend of increased use of one-stage revision was also observed among knee PJIs, as 12.1% of the revisions in 2008–2009 were one-stage, whereas the proportion had risen to 43.8% in 2020–2021. In contrast, the use of two-stage revisions for knee PJIs decreased remarkably from 57.6% in 2008–2009 to 6.3% in in 2020–2021.

At 1-year follow-up, 26.6% (confidence interval [CI], 22.2%–31.2%) of hip PJI patients had undergone a reoperation and 7.9% (CI, 5.2%–10.9%) had died. The lowest risk for reoperation was after one-stage revision (20.2%, CI, 13.4%–28%), and highest if debridement, antibiotics, and implant retention (DAIR) was performed (36.6%, CI, 28.5%–44.7%). Among early hip infections, one-stage revision was associated with a lower risk for reoperation (hazard ratio [HR] 0.51, CI, 0.31–0.84) with no added mortality risk (HR 1.05, CI, 0.5–2.2) when compared to DAIR. The results after knee PJI were better. After 1-year follow-up, 22.8% (CI, 18.6%–27.3%) of patients had to be reoperated and 3.6% (CI, 2.0%–5.9%) died. Furthermore, after 1-year follow-up the failure rates favored one-stage revision over two-stage revision for all knee infection types. In addition, the results after DAIR were good; for early infections, the risk of failure within the 1-year follow-up was the lowest when DAIR was performed (26.1%, CI 15%–35.8%).

In study V, DAIRs and one-stage revisions were analyzed, and a preoperative prediction score, the KLIC-score, was externally validated. After DAIR, the KLIC- score had a moderate predictive ability (odds ratio [OR] 1.45 per one-unit increase, CI, 1.13–1.90) for early failure, but after one-stage revision the predictive ability was inferior (OR 1.20, CI, 0.93–1.56). After 60 days, the discriminative ability of the KLIC-score was poor both after DAIR (area under curve [AUC] 0.63, CI, 0.55–0.72) and one-stage revision (AUC 0.56, CI, 0.46–0.66). Results from the decision-curve analyses were similar, and the KLIC-score offered no remarkable net benefit to clinical decision-making.

In conclusion, the findings of this dissertation reveal that the incidence of early PJIs has not decreased, but rather it has even increased during the previous decade. In addition, this study has also shown that one-stage revision is a viable treatment option for both early hip PJIs and chronic knee PJIs. As the number of PJIs will likely increase in future, the findings of this dissertation should be used during discussions on future treatment strategies. The results of this dissertation also show that the prediction of failure after PJI treatment is difficult, and that current prediction models are not valid for clinical use.

Finally, as the number of PJIs will increase in future, but the literature on PJIs remains divergent and limited, further high-quality studies on this important topic are warranted. Therefore, future research should focus on large, high-quality, prospective trials that compare different treatment approaches for PJIs, especially for cases of early PJIs.