Role of Routine Suppressive Antibiotic Therapy After Debridement, Antibiotics, and Implant Retention for Acute Periprosthetic Joint Infections

Abstract Background The first-line management strategy for acute periprosthetic joint infections (PJIs) is debridement, antibiotics, and implant retention (DAIR). Suppressive antibiotic therapy (SAT) after DAIR is proposed to improve outcomes, yet its efficacy remains under scrutiny. Methods We conducted a multicenter retrospective study in patients with acute PJI of the hip or knee who were treated with DAIR in centers from Europe and the United States. We analyzed the effect of SAT using a Cox model landmarked at 12 weeks. The primary covariate of interest was SAT, which was analyzed as a time-varying covariate. Patients who experienced treatment failure or were lost to follow-up within 12 weeks were excluded from the analysis. Results The study included 510 patients with 66 treatment failures with a median follow-up of 801 days. We did not find a statistically significant association between SAT and treatment failure (hazard ratio, 1.37; 95% CI, .79–2.39; P = .27). Subgroup analyses for joint, country cohort, and type of infection (early or late acute) did not show benefit for SAT. Secondary analysis of country cohorts showed a trend toward benefit for the US cohort (hazard ratio, 0.36; 95% CI, .11–1.15; P = .09), which also had the highest risk of treatment failure. Conclusions The utility of routine SAT as a strategy for enhancing DAIR's success in acute PJI remains uncertain. Our results suggest that SAT's benefits might be restricted to specific groups of patients, underscoring the need for randomized controlled trials. Identifying patients most likely to benefit from SAT should be a priority in future studies.

Periprosthetic joint infections (PJIs) are challenging infections to manage, even among experienced orthopedic surgeons and infectious disease doctors.A widely accepted management strategy for acute PJI is debridement, antibiotics, and implant retention (DAIR).DAIR is less invasive than complete resection of the prosthesis, although the outcomes are poorer.Treatment success rates hover around 60% to 67%; this is typically defined as clinical infection eradication, no need for further surgery, no PJI-related mortality, and no use of suppressive antibiotic therapy (SAT) [1].The poor outcomes are partly due to the presence of biofilm, rendering them refractory to the host immune system and antibiotic therapy [2].Therefore, improving DAIR's success rates is paramount among clinicians and patients.
A persistent question in DAIR is the role and efficacy of SAT.SAT is generally a long-term antimicrobial treatment intended to prevent the recurrence of infection.Practices differ around the world, as previous studies have shown mixed results regarding its effectiveness in preventing PJI recurrence post-DAIR [3].The ambiguity surrounding the benefits and potential drawbacks of SAT stems from a lack of comprehensive highquality randomized controlled trials.Furthermore, the existing literature predominantly consists of studies with a small sample size and single-center studies with varied definitions of infection and outcomes [4,5].The best evidence in the duration of antibiotic therapy in DAIR is guided by the DATIPO trial, which showed that 6 weeks of therapy is not noninferior to 12 weeks [6].In this analysis, patients treated with 12 weeks of antibiotic therapy after DAIR still had a failure rate of Suppressive Antibiotic Therapy After DAIR for Acute PJI • OFID • 1

Open Forum Infectious Diseases
14.5%, which raises the question whether patients might benefit from an even longer duration of treatment.
The primary aim of our research is to investigate the utility of SAT following acute PJI managed with DAIR in a cohort of patients in which the prescription of SAT is routine practice (United States) and in a cohort in which SAT is prescribed only in a select population (Europe).Our objective is to evaluate the association of SAT with risk of treatment failure.

Study Design and Population
We performed a retrospective cohort study of patients aged ≥18 years with PJI who underwent DAIR.We utilized previously collected data from 3 international cohorts: the United States, Spain, and the Netherlands (NED).The PJI cases from Europe occurred between 2005 and 2016, while the US cases were from 2008 to 2018.We included patients with acute hip or knee PJI who received at least 12 weeks of systemic antibiotic therapy.We excluded patients with chronic infections, especially with sinus tracts; those not treated with DAIR; and those for whom follow-up was <12 weeks from debridement (eg, due to death, loss to follow-up, treatment failure).We also excluded those who refused research authorization.These studies were deemed exempt by institutional review boards.

Definitions
The definition of PJI was based on criteria from the Musculoskeletal Infection Society [7].Early and late acute were included.We defined early acute PJIs as infections occurring within 3 months of the index or revision arthroplasty.Late acute PJIs were defined as infections occurring after 3 months but with a symptom duration <3 weeks prior to debridement.Arthroplasties were primary or revision.SAT was defined as antibiotic therapy after 12 weeks of therapy.In the US cohort, SAT was prescribed as a routine practice in every patient.In the European cohorts, SAT was prescribed at the discretion of the treating physician based on risk of relapse of infection.Some of the factors considered were persistently elevated C-reactive protein and delayed wound healing.The primary outcome was treatment failure, defined as recurrence of PJI, unplanned reoperation secondary to infection, or infection-related death.Covariates such as age, comorbidities, culture results, and surgical history were assessed at the time of debridement.

Statistical Analysis
The association between SAT and treatment failure was assessed in a multivariable Cox proportional hazards model.The start date for analysis was 12 weeks after surgery.To address potential immortal time bias, which can arise when measuring the effect of an exposure that varies over time in cohort studies, we assessed SAT as a time-varying covariate.We performed least absolute shrinkage and selection operator (LASSO) to reveal important covariates with the most potential for confounding.We selected the penalty parameter for the LASSO model using cross-validation and chose the lambda with the smallest cross-validation c-index.We also considered the largest lambda within 1 SD of the smallest cross-validation c-index, but we felt that this created too much regularization as it selected only a single variable into the model.In addition to our analysis with SAT as a time-varying covariate, we performed a Cox model analysis with country cohorts instead of SAT and with the rest of the covariates unchanged.This analysis allows us to determine if variations in treatment regimens across countries, including differences in how SAT is utilized, lead to different rates of treatment failure.We checked for violations of the proportional hazards assumption using the test proposed by Grambsch and Therneau [8].

RESULTS
There were 510 patients in the final analysis: 184 from the United States, 236 from NED, and 90 from Spain.The majority were female (58%, n = 296) with a mean age of 70.4 years (SD, 11).Diabetes mellitus was the most common comorbidity with a prevalence of 20% (n = 102).There were 254 knee and 256 hip PJIs.Primary arthroplasties represented 69% of the cases (n = 352).The majority of cases were early acute infections (62%, n = 365) while the mean duration of symptoms was 6.8 days (SD, 7).Staphylococcus aureus was the most common causative microorganism (38%, n = 194).For a complete overview of the clinical characteristics of the cohort, see Table 1 and Supplementary Table 1.

SAT and Outcomes
There were 66 treatment failures (13%), and the median followup was 801 days (range, 91-4100).Treatment failure occurred in 39 of 167 patients undergoing SAT, as compared with 27 of 343 patients not receiving SAT.The LASSO model revealed age, infection with S aureus, knee joint, late acute infection, chronic kidney disease, revision arthroplasty, and cemented arthroplasty as potential covariates.SAT did not demonstrate a significant impact on the treatment outcome after adjusting for several covariates identified by the LASSO model (hazard ratio [HR], 1.37; 95% CI, .79-2.39;P = .27;Figure 1).Subgroup analyses, segmented by joint and infection type (early or late acute), did not demonstrate benefit.The NED cohort showed increased treatment failure with SAT, likely reflecting selection bias (HR, 11.46; 95% CI, 2.08-63.03;P < .01).There was no significant relationship between SAT in the US or Spain cohort.Table 2 lists the rest of the HRs.

Country Cohort as SAT Proxy
Given the extreme differences in SAT duration among country cohorts, particularly the United States and NED, including SAT and country in the same model would introduce multicollinearity (Supplementary Table 2).As an example of this difference, 167 patients were undergoing SAT for the entire duration of follow-up, and 160 of these were from the US cohort.Additionally, 99 patients were not receiving SAT as of 12 weeks after surgery, and 98 of these were from the NED cohort.Furthermore, the majority of treatment failures were from the US cohort (36/66).Therefore, we conducted an additional sensitivity analysis using the country as a proxy for antibiotic exposure.Controlling for country cohort allowed us to see if differences in country treatment regimens, which include differences in SAT use, resulted in different rates of treatment failure.The NED cohort's HR was 31% lower as compared with the US cohort, though this was not statistically significant (95% CI, .35-1.38;P = .30).Similarly, the HR for the Spanish cohort was 11% lower but also nonsignificant (95% CI, .45-1.77;P = .74).In addition to reporting the analysis based on a multivariate Cox model, we present the Kaplan-Meier survival curve estimates for each country in Supplementary Figure 1.

DISCUSSION
There is intense interest in improving the success rates of DAIR for the management of acute PJI.SAT after the initial 12 weeks   Abbreviations: HR, hazard ratio; PJI, periprosthetic joint infection.
Suppressive Antibiotic Therapy After DAIR for Acute PJI • OFID • 3 of antibiotic therapy is proposed as one of the strategies to improve patient outcomes [9].Unfortunately, the current literature is inconclusive and does not address immortal time bias [10].Adverse drug reactions and antibiotic resistance are also concerns with SAT [11].The primary objective of our study was to investigate the utility of SAT following acute PJI managed with DAIR.In our international multicenter retrospective study, there was no significant association between SAT and treatment failure (HR, 1.37; 95% CI, .79-2.39;P = .27).The subgroup analysis for joint, type of infection, and country cohort did not show benefit of SAT.
Our findings contribute to the ongoing debate regarding the role of SAT in acute PJI managed with DAIR.This study raises questions about routine SAT, especially considering the risks associated with prolonged antibiotic therapy.We found that the majority of patients remain infection free after 12 weeks without SAT, which is consistent with other studies on DAIR from countries without routine SAT use [12][13][14].While we did not see an outright benefit of routine SAT use, the trend is toward benefit for the US cohort (HR, 0.36; 95% CI, .11-1.15).The analysis on outcomes of cohorts revealed that the US cohort had the highest risk of failure when compared with European groups.A review of differences in the characteristics of the country cohorts showed that the US cohort had a longer duration of symptoms and greater body mass index, as well as higher proportions of bacteremia, rheumatoid arthritis, and chronic obstructive pulmonary disease.This suggests that the other uncontrolled and unmeasured covariates might have contributed to poorer outcomes in the US cohort in which SAT might be appropriate.
We note that the Cox model results for the model with SAT as a time-varying covariate and the model with country cohort as a covariate should not be interpreted as implying a causal effect of SAT on treatment failure.This is for 2 reasons.The first is that we believe that there are many differences in treatment regimens among countries that go beyond the variables in our data set and can act as potential confounders of the relationship between SAT and treatment failure.The second involves the well-known issues with trying to interpret parameters from the Cox model as showing causal associations [15].We do believe that the associations in our results are important because they suggest that patients from the NED cohort have better outcomes than those in the US cohort, even with shorter SAT duration.However, shorter SAT duration may not be the cause of this improvement.
Our subgroup analyses offered a glimpse into possible risk stratification for SAT.Patients without high-risk featuressuch as S aureus infection, knee PJI, late acute infection, chronic kidney disease, revision arthroplasty, and cemented arthroplasty-have the least benefit for SAT.This is consistent with previous literature reporting that these are risk factors for treatment failure in DAIR [16].Note that most of these studies analyzed the risk factors for patients whose treatment had already failed during antibiotic therapy.Therefore, more studies are necessary to assess the risk factors for failure after 12 weeks of antibiotic therapy.Some experts recommend shared decision making in recommending SAT to patients with factors that are more subjective.These include limited surgical options, recurrent infections, difficult-to-treat pathogens, or immunosuppression [17].
There are some limitations to our study.Ideally, we would exclude those patients whose indication for SAT was a concern for persistent infection, but this was not possible in the US cohort.Therefore, we included patients in the NED and Spanish cohorts whose indication for SAT was a concern for persistent infection, which introduced selection bias against SAT when applied to all cohorts collectively.While this is the largest study to date, the low number of patients limited the power of our analysis, and the few events precluded us from analyzing more covariates.For instance, we were unable to consider variations in initial antibiotic treatment or conduct subgroup analyses for different microbial causes, as the smaller sample sizes would not provide meaningful results.While we made efforts to adjust for potential confounders, differences in comorbidities among cohorts are likely to be residual confounders.Differences in the time frame of the cohorts also introduce variability that might affect the outcomes.Due to limitations in our data, we were unable to perform other statistical methods that can result in parameters with stronger causal implications, such as the G-formula and inverse probability weighting methods.
Given the nonconclusive nature of our findings and the mixed results of prior studies, there is a clear need for further research.Randomized controlled trials would provide more definitive evidence on SAT's role in acute PJI after DAIR management.Additionally, future research should focus on identifying patients who benefit the most from SAT, including the role of artificial intelligence in this regard.

CONCLUSION
While routine SAT has been viewed as a potential strategy to bolster the success rates of DAIR in managing PJI, our study suggests that its benefits might be limited.There might be a subset of patients who benefit from SAT and should be the focus of future research.

Figure 1 .
Figure 1.Estimated survival curve for treatment failure on and off Suppressive Antibiotic Therapy based on Cox model.

Table 1 . Clinical Characteristics of the Cohort (N = 510)
a Missing in 25 cases.b Denominator was 410.c Denominator was 411.

Table 2 . Analysis of Treatment Outcomes With Suppressive Antibiotic Therapy as a Time-Varying Covariate
Staphylococcus aureus infection, type of joint, type of infection, chronic kidney disease, type of procedure, and cemented arthroplasty as appropriate.