We present an investigation on the association between PJI after primary TKA and mortality rate in a large, national cohort. We found a higher mortality rate after PJI compared to patients without PJI in both short- and long-term. The association remained after adjusting for available confounders such as age, sex, diagnosis, and, in the latter time period, ASA-class. We chose to analyze patients with PJIs diagnosed within 90 days of primary surgery to be able to capture predominantly early postoperative infections and to exclude the effect of hematogenous infections and secondary septic seeding as far as possible. More than 70% of the included PJIs were also diagnosed within the first 30 days (Table 2), strengthening this assumption.
The mortality rates at 1- and 5 years (2.6 and 15.7%, respectively) are somewhat lower than the 4.33 and 21.6% found in the meta-analysis by Lum et al. [5]. This could partly be explained by variations in patient selection. The meta-analysis by Lum et al. included a heterogenous sample of studies on two-stage revision knee-PJIs, whereas our material is comprised mainly of early PJIs that were treated with a DAIR strategy. Further, we included only primary TKAs, excluding patients with previous infection or revision for other causes. Since it is reasonable to assume that mortality is even higher in this group, it can explain some of the difference between the studies. No comparable nationwide studies on TKA exist to our knowledge, but some works on hip arthroplasty can be found. In the recent Swedish study on primary THA by Wildeman et al. a 10-year mortality rate after PJI in THA patients of 45% was found, somewhat higher than the 38% found in our study [6]. Similarly, Gundtoft et al. found a 1-year mortality rate of 8% following revision for THA PJI, quite a lot higher than the 2.6% in our study [11]. Hip fracture patients were included in both studies and is a group with known fragility. This may explain some of the difference, as the present study only included elective procedures.
As of 2018 the 10-year survival rate for all cancers in Sweden was 69% [15], suggesting a worse prognosis following PJI than for many cancers. We can only speculate on the reason for the increased mortality rate after PJI, since our investigation offers no data on causes of death. Some deaths can surely be attributed to the deep infections of themselves, but a majority probably have other causes. In our data 374 out of 466 PJIs (80%) were considered cured at the last follow up visit, suggesting a minor proportion of direct infection-related deaths. Further, when comparing cured infections with non-PJIs, a higher mortality rate was seen for the PJI cohort at 5- and 10-years (12.8 and 33.8%, respectively) but not at 1 year (0.5%) (data not shown). It has been recently proposed that PJI could be regarded as a chronic condition even after cure of the infection [6]. The decreased joint function and mobility would, according to this proposition, inhibit physical activity leading to a general health deterioration. Future research is needed to investigate this hypothesis. It is reasonable to assume that increases in mortality after PJI is not solely due to the infection, especially not more than 2 years after PJI diagnosis. Therefore, the presence of other conditions, predisposing for both PJI and death must be considered.
Previous studies identify several comorbid conditions that predispose patients to PJI [16,17,18], leading to a generally worse health in PJI patients than in arthroplasty patients in general. This is reflected in our data where PJI patients had slightly higher ASA-scores than the non-PJI cohort. ASA-score was included in the SKAR as a crude marker for comorbidity from 2009 and, though suffering from reliability issues, has been demonstrated to correlate to both increased mortality following arthroplasty and to increased risk for infection in other studies [16, 19]. Several comorbidities have also been identified as independent risk factors associated with increased risk of death within 1 year of PJI, such as diabetes, chronic lung disease and congestive heart failure [12]. These risk factors, if present, also affect the ASA-score. The effect of ASA-score on mortality in our study was therefore expected.
We found no visible difference in mortality between the 2 time periods in the study suggesting that improvements in treatment practice of PJI in latter years does not influence over all mortality in PJI patients. This is further elucidated by the fact that no significant difference in mortality was observed between different treatment modalities. In a previous study by Fischbacher implant exchange was associated with lower mortality than DAIR [20], a finding that has not been reproduced elsewhere. The choice of surgical treatment strategy is usually made depending on type of infection, status of the implant and surrounding tissue. It could be hypothesized that DAIR treated patients would have lower mortality since most patients undergo only one surgery. On the other hand, most DAIR treated patients have early infections that tend to be caused by more virulent bacteria than chronic infections that need implant exchange to be cured, possibly counteracting the benefits of fewer operations. It is of interest that the few patients that were not operated at all had a similar mortality rate as surgically treated patients, suggesting that infection eradication is not a prerequisite for survival.
The presented study has a few limitations. First, we used retrospective review of medical records to determine whether PJI was present or not, and diagnosis of PJI was determined by the treating physician. This leads to some variability, and PJIs may have been missed or falsely classified as infected. Further, there is an issue of immortality time bias in the PJI group that we were unable to control for. We believe the impact of this to be minor in the long term, with a maximum of 90 days increased survivorship in the PJI group. In the short term, however, mortality in the PJI cohort may have been underestimated due to this issue. Major risk factors for PJI and death, such as smoking and comorbidities, were unavailable to us. ASA score was included in the SKAR from 2009 and therefore only available for part of our cases. We included ASA score in our analyses where possible to try to ameliorate this lack. It would be interesting to investigate the cause of death in the PJI cohort, since this would elucidate the increased mortality rate further.
The major strength of our work is its size and nationwide setting. The use of the Swedish tax agency accounts for accurate mortality data with approximately zero data loss or misclassifications.
