Our study demonstrated that the general incidence of infection recurrence in the treatment with DAIR was 18%. Controlling for other variables, only two risk factors among all evaluated were significant: multidrug-resistant bacterial infection (MRSA or Gram negative) and patients with previous rheumatic diseases. This is the main finding of this study.
The treatment of periprosthetic hip and knee infections is a challenge for both the orthopedist and the patient and is one of the most feared complications after arthroplasty . In acute infections, the main objective is to avoid biofilm formation around the prosthetic material through a careful procedure that includes irrigation, debridement, the exchange of modular components and broad-spectrum antibiotic therapy . The prognosis in these cases can be discouraging when associated with risk factors such as multidrug-resistant bacteria, rheumatic diseases, diabetes and smoking [2,9,10,12,14,15].
The recurrence of infection after the initial treatment of a periprosthetic joint infection with DAIR varies in the literature between 14% and 69% [10,16,17]. The highest number of cases study, as described by Cochran et al , presented a reinfection rate of 22.7% in 3069 cases. The incidence rate of 18% (10/57) found in our study agrees with the values described in the literature and is acceptable for this type of treatment.
Regarding the type of microorganism causing the infection, our study showed a treatment failure rate of only 8.3% (3/37) when the bacterium found in the culture was multidrug-sensitive. In the MRSA group, we observed a failure rate of 18.2% (2/11), with an odds ratio for recurrence of infection of 58.5 (p=0.01). More outstanding, the group of patients with Gram negative multidrug-resistant bacteria had a failure rate of 55.6% (5/9), with an odds ratio of 49.9 (p=0.048). Most studies in the literature corroborate our findings pointing to multidrug-resistant bacteria as a factor of poor prognosis in the success of PJI treatment [18,19]. The reasons for worst outcomes in resistant organism are multifactorial, such as limited choose of effective antibiotics, more aggressive biological behavior related to biofilm production and greater virulence [2, 20]. Bradbury et al  found an infection recurrence rate of 84% after irrigation and debridement treatment when the bacteria found in the cultures was MRSA compared to the overall failure rate of 18%. Based on these findings, some authors have gone so far as to recommend a two-stage revision in acute infection caused by MRSA [18,19]. In contrast, our findings demonstrate a reasonable chance of success in MRSA compared to multidrug-resistant Gram-negative bacteria, which raise a question regarding the role DAIR in these patients. Hsieh et al  compared the infection recurrence rate after irrigation and debridement of Gram-positive bacteria with Gram-negative bacteria and obtained a success rate of 47% vs. 27%, respectively. Even when compared to other types of initial surgical procedures, non-MRSA multidrug-resistant bacteria tend to have worse outcomes than MRSA. Vaso et al. showed a failure rate of 33% for non-MRSA multidrug-resistant bacteria versus 10% for MRSA in two-stage revision knee arthroplasty. However, some authors did not find a significant difference in results when comparing multidrug-sensitive and multidrug-resistant bacteria [22,23]. Most studies, however, have used small sample sizes, which hinders further conclusions.
Regarding antibiotic therapy management, in only 6/57 cases (all multidrug resistance bacteria) the initial antibiotic protocol was not suitable based on bacteria resistance profile. Half of those cases (3/6) had failed DAIR procedures. We analyze the risk and benefit of expanding the spectrum of initial antibiotic therapy protocol for PJI, but due to the concern in increasing bacteria resistance and the fact that nowadays all patients with suspected PJI are submitted to joint aspiration previously to any surgical procedure (with very few incidence of negative cultures and dry aspiration ), we find it unsuitable to modify our existent protocol.
Another risk factor that was statistically significant in our study was the presence of associated rheumatic disease with an odds ratio for treatment failure of 45.1. It is already well established in the literature that rheumatic diseases is an independent risk factor for PJI . Bongartz et al  demonstrated an increased risk of PJI in patients with rheumatic disease, with an odds ratio of 4.08 (95% CI 1.35-12.33).
The present study has limitations regarding its retrospective design and sample size. Despite the increase in bacterial drug profile resistance in the last decade, the low incidence of PJI, particularly related to multidrug-resistant organism, results in the absence of large studies involving these types of microorganisms in medical literature.
The prevention and treatment of acute prosthetic infection is extremely important today because of its severity, difficulty to treat, and its large contribution to the total number of failed joint replacements. As the number of joint replacements is expected to rise, this problem will only become more relevant in the next years. It is important to identify multi-drug resistant bacteria and inflammatory disease as factors of poor prognosis in the treatment of acute infection with DAIR. These data may warrant future clinical trials comparing DAIR and revision arthroplasty in this high-risk situation, and at present allow for a more informed treatment decision of surgeons, infection specialists and patients.