In this multicentre, randomized controlled trial involving 48 patients with acute staphylococcal PJI treated with a DAIR procedure, the addition of rifampin to standard treatment with cloxacillin or vancomycin did not improve the cure rate. To our knowledge, this is the second randomized controlled trial to examine the effectiveness of adjuvant rifampin therapy in acute PJIs, and our findings are in contrast to previous findings. Zimmerli et al. published the first study back in 1998 [23]. It was a single-centre trial involving 24 patients, of which 15 were PJIs and 9 were infected osteosyntheses. This study was prematurely discontinued because all the failures occurred in the control group. It has been criticized for small numbers and limited statistical power, 33% drop-out rate in the rifampin group, as well as the choice of ciprofloxacin as monotherapy in the control group. The risk of emergence of ciprofloxacin-resistant staphylococcal strains is high when ciprofloxacin is given as monotherapy [29], which also was seen in the actual study. Due to the concerns by using ciprofloxacin in monotherapy, we chose to add rifampin to the standard treatment at the time, which was cloxacillin, or vancomycin in case of methicillin resistance.
Several retrospective observational studies and case series have been published in the last decades, evaluating different rifampin combinations. These studies were not controlled studies and the success rates have never reached 100% as in the Zimmerli study, but their findings have although favoured the use of rifampin [25, 26, 28]. There are difficulties in interpretation of these studies, including considerable differences in baseline characteristics between treatment groups, surgical methods not described in detail, and varying MRSA rates. Barberán et al. found a success rate of 65% in staphylococcal PJIs treated with rifampin and levofloxacin following DAIR, seemingly a more common and expected result from these infections [30]. In a recently published study, a significant higher failure rate was found in rifampin combinations with linezolid, co-trimaxazole and clindamycin. This is explained by the fact that rifampin reduces serum concentration of these drugs, and also for fusidic acid [31, 32]. Two review articles from 2008 and 2010, respectively, both conclude that the use of rifampin is based mostly on noncomparable in vitro and in vivo data and retrospective case reviews. Because of the biases of these, there are not sufficient data to support rifampin combination therapies [33, 34]. However, it suggests that it could be effective in infections containing biofilm-producing agents, such as staphylococcal PJIs, but its use must be evaluated against the probability of drug interference and toxicity for each individual patient. A retrospective study from 2017 proved no advantage in treating streptococcal infections with an addition of rifampin [35]. These infections also form biofilm and rifampin should theoretically improve the outcome. In vitro studies are inconsistent, but most have shown that the combination of vancomycin and rifampin promotes antagonism or indifference [36]. An in vivo case report showed a higher failure rate in PJIs when combining vancomycin and rifampin after debridement and retention of the prosthesis [30]. This is consistent with our results and the combination vancomycin-rifampin should be used cautiously.
The preferred treatment in early PJI is DAIR. This option reduces morbidity, improves function, is cost effective compared to 1- and 2-stage revisions, and has equal results when it comes to eradicating infection[11, 37]. It has been postulated that the biofilm has increased to such degree that cure with DAIR is less achievable after 1 month [38]. Only patients with infections within 4 weeks after surgery, and acute haematogenous infections, were included in our study. The reported results following DAIR have been varying considerably, ranging from 21–100% [13, 19, 23, 28, 39]. There are many limitations when interpreting the literature, as several factors are varying and the surgical procedures are often poorly described. Both the definition of acute PJI, the number of procedures, the type and duration of antibiotics, and even the definition of success vary. Our results of approximately 75% success at 2 years follow-up without suppressive antimicrobial therapy, is comparable with recent literature [40, 41]. Regarding the number of procedures, some advocates that repeated DAIR is effective [42]. On the other hand, some authors have found the need for additional DAIRs is associated with increased risk of failures [43]. In this present study, we chose to regard an additional DAIR procedure as a failure.
We found S aureus to be more frequent than CoNS in this material, which is in contrast to most reports [12, 14, 31]. This may reflect the challenges of defining PJI. We used the Tsukayama definition, including patients with a short duration of symptoms shortly after the index surgery (30 days)[13]. Many definitions include PJI within the first 3 months after index surgery as early postoperative infections. This may explain the findings of more virulent bacteria (S aureus) in our material.
The difference we found in preoperative CRP-levels is consistent with the current literature with lower CRP in the low-virulent infections[44]. We believe that the CRP should primarily be used to monitor the course of the disease and as minor criteria in the diagnosis of PJI[45].
PJI caused by methicillin resistant S aureus (MRSA) is reported to worsen the prognosis of DAIR, and a rifampin-containing antimicrobial regimen is often used. There were no MRSA infections in our cohort, reflecting the still very low prevalence of these infections in the Nordic countries. A prevelance < 1% is reported in Norway over the last 20 years [46].
In this study, we collected 8 specimens in total for culturing. According to the MSIS Definition of Prosthetic Hip and Knee, the recommended number of specimens should be 5. Our protocol was written approximately 7 years before The MSIS Definition was published. It is therefore not in line with the current guidelines[45].
There are several limitations to the interpretation of the data. First of all, even though this is the highest number of PJIs included in an RCT on adjuvant rifampin-treatment, the sample size is relatively small. Recruitment rates were lower than expected at all participating sites, illustrating why conducting randomized clinical trials in clinical settings with few eligible patients is difficult. The study was stopped after enrolment of 99 patients, without knowledge of study outcome. Small sample size increases the risk of type II error, and then often showing high intervention effects. In this study, there was an independent reason for stopping the trial, which may lower the risk of bias. One may argue that statistical power is a pre-study tool, and because our study ended due to slow inclusion rates, the actual results may be interpreted as they are, based on the 95% CIs [47]. Secondly, gentamicin collagen sponges were placed in the wound before closure. This adds an additional antimicrobial agent to the equation and was performed according to the guidelines at the time of initiating the study. However, all patients received these sponges, and it is difficult to see how it could have affected the final outcome. Third, the type of antibiotics combined with rifampin seems to affect the outcome. Our results may therefore not be comparable with other rifampin combinations.
In conclusion, our findings indicate that adding rifampin to cloxacillin or vancomycin treatment in patients with acute staphylococcal PJIs, does not affect the cure rate of DAIR procedure in hip or knees. The results must however be cautiously interpreted due to low number of patients, but the study still adds important knowledge to defining the benefit of rifampin.