Clear Recommendation:
Based on two RCTs (Figure 4),(Bernard, Dinh et al. 2015, Tone, Nguyen et al. 2015) and concordant observational studies, we recommend a maximum of 6 weeks of antibiotic therapy for hematogenous or contiguous pyogenic osteomyelitis (including DFO), assuming adequate source control (i.e., no undrained abscesses too large to be treated with antibiotics alone, possibly ≥ 2-3 cm in diameter) and no retained prosthetic implant (Table 6). Insufficient data are available to establish a Clear Recommendation for durations shorter than 6 weeks (see Clinical Review below).
Clinical Review (insufficient quality of evidence to enable a Clear Recommendation):
Based on small RCTs, 3 or 4 weeks may be a reasonable duration of antibiotics for debrided osteomyelitis, whether hematogenous or contiguous (including DFO); however, confirmatory data are desired. Based on observational studies and one small RCT, it is reasonable to refrain from antibiotic use after total resection of infected bone if the treating physicians are confident that all infected bone has been resected. If administered, we do not recommend exceeding 2-5 days of therapy if there is no complicating soft tissue infection.
Clinical Review (insufficient quality of evidence to enable a Clear Recommendation):
Based on the Duration of Antibiotic Treatment in Prosthetic Osteo-articular infection (DATIPO) RCT, participating experts unanimously agree that 12 is preferred to 6 weeks of antibiotics for PJI treated with debridement, antibiotics, and implant retention (DAIR).(Bernard, Arvieux et al. 2021) Some experts also clearly prefer 12 weeks of antibiotics for PJI treated with prosthetic exchanges. However, others believe that equipoise remains between 6 vs. 12 weeks for these patients, particularly if S. aureus is not the etiologic pathogen, or for 1-stage exchanges, or 2-stage revisions with negative cultures prior to implantation.
Duration of therapy for other infected implants is not clear. A reasonable strategy, without evidence for or against, may be to treat with antibiotics until the bone heals sufficiently enough that the implants can be removed, such as in cases of fracture. Finally, chronic oral suppressive therapy may be considered for patients for whom the risk:benefit of curative surgery is deemed unacceptable; however, available data do not well-define the risks:benefits of this approach.
The duration of therapy for osteomyelitis has long been based on anecdote, case series, and tradition. However, multiple observational studies and seven RCTs have now begun to provide evidence to resolve this question. We reiterate that this guideline refers only to pyogenic osteomyelitis and does not consider durations of therapy for atypical causes (e.g., TB, fungal, Brucella).
Roblot et al. conducted a retrospective study of 120 patients with vertebral osteomyelitis to evaluate the impact of duration of therapy.(Roblot, Besnier et al. 2007) Receipt of ≤6 weeks of antibiotics was not associated with an increased risk of clinical failure compared to >6 weeks. Indeed, at a mean of 3.5 years of follow up, there was no significant difference in relapse or mortality for patients receiving ≤ 6 vs. >6 weeks of therapy.
Park et al. evaluated duration of antimicrobial therapy and outcomes among 345 patients with hematogenous vertebral osteomyelitis.(Park, Cho et al. 2016) Source control was obtained in more than half of patients, either by surgery or by needle drainage. In the pre-planned multivariate analysis, end stage renal failure, infection with MRSA, and undrained abscess (paraspinal or psoas) were the only predictors of recurrence/treatment failure; receipt of < 6 weeks of antibiotic therapy was not associated with recurrence/failure. Having seen the results of their first multivariate analysis, the authors then constructed a post-hoc multivariate model to attempt to refine their results. In this post-hoc model, they found that receipt of <6 weeks of antibiotics was associated with increased risk of recurrence. However, the post-hoc model raises concerns about variable selection bias and multiple comparisons issues. Based on the pre-planned multivariate analysis, prolonging antibiotics for > 6 weeks was not associated with a decreased risk of clinical failure, while undrained abscesses and MRSA infection were. These factors might indicate the need to treat for > 6 weeks if source control cannot be achieved, and the desirability of finding alternative therapies for MRSA in lieu of vancomycin (see Section 5). In the context of osteomyelitis, there are no specific data to define inadequacy of source control; we suggest that the presence of undrained abscesses that are too large to be treated with antibiotics alone (e.g., more than 2-3 cm in diameter) is a reasonable definition.
Another study of 49 patients with chronic osteomyelitis who underwent debridement evaluated the impact of duration of therapy after surgical intervention.(Rod-Fleury, Dunkel et al. 2011) The median number of debridements was 2 (range 1-10), and the median duration of antibiotic therapy post-debridement was 8 weeks (range 4 to 14 weeks). At a minimum of two years of follow up, 80% of patients had persistent treatment success. By multivariate analysis, neither administration of 1 vs. 2 vs. 3 weeks of intravenous antibiotics, nor administration of ≤6 or > 6 weeks of total antibiotics post-debridement, was associated with treatment success.
Other studies also suggest that courses < 6 weeks may result in similar efficacy to > 6 weeks of antibiotic therapy. For example, Meißner et al. (first-author’s last name indexed as Meissner on PubMed) treated 53 evaluable patients with chronic, post-traumatic (including motor vehicle accidents, gunshot wounds, war wounds) osteomyelitis of the long bones with oral fosfomycin (not the sachet formulation available in the US) for between 5 and 28 days.(Meissner, Haag et al. 1989) These patients had very complex histories, with a mean of 37 months of persistence of osteomyelitis before treatment with fosfomycin was initiated. Etiologic pathogens were highly varied, including S. aureus, coagulase-negative staphylococci, streptococci, enterococci, and a variety of Gram-negative bacilli, including P. aeruginosa. Yet, the ≤ 4 weeks of fosfomycin therapy achieved a 73% long term success rate at follow up. Similarly, Shcherbin et al. studied 33 patients who had an average of seven years of osteomyelitis following gunshot injuries.(Shcherbin, Makhson et al. 1990) In addition to debridement, patients were treated with lincomycin and gentamicin for ≤22 days (range 7-22), resulting in an 88% (23/26) long term success rate at a mean of four years of follow-up.
Finally, Haidar et al. systematically reviewed the literature for reports of patients treated with shorter durations of antibiotics for osteomyelitis.(Haidar, Der Boghossian et al. 2010) In their summary of these small, uncontrolled case series, they describe a total of 21 other patients (non-redundant with the studies described above) treated with 1-4 weeks of a variety of antibiotics, of whom 18 (86%) achieved clinical success at last follow-up.
Several observational studies have evaluated the potential impact of therapy duration on DFO, with or without resection of infected bone. In a large observational study of 1,018 patients with DFI, 392 patients had confirmed osteomyelitis who underwent debridement or amputation, with antimicrobial therapy.(Gariani, Lebowitz et al. 2019) There was no difference in the mean duration of antimicrobial therapy for patients who developed recurrent osteomyelitis or not (mean 31 days vs. 34 days). Furthermore, there was no difference in the proportion of patients who received <3 weeks of total antimicrobial therapy vs. more among patients who developed recurrent osteomyelitis (27% vs. 32%). By multivariate analysis, total duration of therapy or receipt of <3 weeks of antibiotics vs. more were not associated with risk of recurrence of osteomyelitis.
In a study of 184 patients with DFO who underwent surgical resection, administration of <7 days of antibiotics in the presence of a positive post-operative margin by histopathology (indicating infection extended to the margin) was independently associated with a significant increase in risk of additional resection or amputation by multivariate analysis.(Barshes, Mindru et al. 2016) In a smaller, prospective observational study, 15 patients with diabetic foot osteomyelitis who underwent amputation or resection of infected bone were identified that had negative culture from the margin of the resection.(Julien, Francois et al. 2020) These patients were considered to have “clean margins”, and were administered antibiotics for 8 +/-6 days post-operatively. Eighty percent (n = 12) achieved healing without osteomyelitis recurrence by six months of follow up.
Kowalski et al. evaluated outcomes in 111 patients who underwent bone resection treatment for DFO.(Kowalski, Matsuda et al. 2011) Of these patients, 39 had positive margins by histopathology, indicating residual osteomyelitis at the resection site. There was considerable overlap between the durations of antibiotics administered post-resection in patients with positive vs. negative margins, mean 19 days (range 10-134 days) vs. 14 days (range 2-63 days). There was no difference in long-term clinical failure between the two cohorts. However, more patients with positive margins required subsequent, more proximate amputation (44% vs. 15%). Of the patients with positive margins, the duration of antibiotics did not differ between those who progressed to failure and/or required further amputation vs. those who did not. Overall, these results suggest that the primary predictor of failure, including need for re-amputation, is surgical, rather than medical, management. This study does not support the practice of prolonging therapy post-amputation.
Finally, in the largest observational study of amputations specifically, Rossel et al. followed 482 patients with DFI who underwent amputation of various types.(Rossel, Lebowitz et al. 2019) Of these, 239 patients had a diagnosis of DFO. Amputation sites varied and included the metatarsals (n = 155), midfoot (n = 280), and hindfoot (n = 47). A median of 7 days (range 1-16 days) of antibiotics were administered post amputation. The investigators conducted a multivariate analysis and reported that neither duration of antibiotics, nor receipt of any antibiotics at all post amputation were associated with risk of clinical failure. They recommended that no antibiotics be administered after amputation with negative margins.
Cumulatively, these observational studies have not indicated an increased risk of clinical failure when ≤ 6 weeks of antibiotics were administered for various types of osteomyelitis, presuming source control is adequately achieved. Indeed, after resection of infected bone with clean margins (defined histopathologically or by negative cultures), it is not clear that any antibiotics are required. When administered in this setting, multiple observational studies have found no difference in outcomes in patients treated with a week or less of antibiotics. Even with debridement that does not remove all infected bone, observational data suggest 3 weeks may result in similar long-term success vs. longer courses. However, one study suggested that, when infected bone is retained, treatment courses of <1 week may result in higher failure rates.
Bernard et al. conducted an open-label RCT of vertebral osteomyelitis in which 351 patients were randomized to 6 vs. 12 weeks of antimicrobial therapy at 71 centers in France.(Bernard, Dinh et al. 2015) The specific antimicrobial agents used were left to investigator discretion, from a list of acceptable options. A wide variety of pathogens were etiologic in the infected patients, including S. aureus, coagulase negative staphylococci, streptococci, enterococci, Enterobacterales, and anaerobes. One year following treatment, the clinical success rates were 90.9% (160/176) vs. 90.8% (159/175) in the 6- vs. 12-week arms. All subgroup analyses were concordant.
Tone et al. conducted an open-label RCT of 6 vs. 12 weeks of antibiotic therapy for 40 patients with biopsy-proven DFO across 5 centers in France.(Tone, Nguyen et al. 2015) The patients were all managed without amputation. Again, a wide variety of antimicrobial regimens were used. Staphylococci and gram-negative bacteria were the predominant pathogens encountered. At a mean of 12 months of post-treatment follow up, treatment success rates were 60% (12/20) vs. 70% (14/20) in the 6- vs. 12-week groups. Fewer patients in the 6-week cohort experienced adverse events related to antimicrobial therapy compared with patients treated for 12 weeks.
Lazaro-Martinez et al. randomized 46 patients with DFO to 10 days of antibiotic therapy plus conservative surgical debridement (defined as removal of infected bone without amputation) vs. 90 days of antibiotics without debridement.(Lazaro-Martinez, Aragon-Sanchez et al. 2014) Treatment success, defined as persistent ulcer healing, occurred in 86% (19/22) of patients treated with 10 days of antibiotics plus debridement vs. 75% (18/24) of patients treated with 90 days of antibiotics alone.
Similarly, Gariani et al. randomized 93 patients with DFO to treatment with 3 vs. 6 weeks of antibiotics.(Gariani, Pham et al. 2020) All patients underwent debridement to remove necrotic tissue, but in contrast to the study by Lazaro-Martinez et al.,(Lazaro-Martinez, Aragon-Sanchez et al. 2014) there was no intent to remove all infected bone. Indeed, in the Methods the authors specifically comment that removal of all infected bone was an exclusion. After a median of 11 months of follow up, treatment success rates were 84% (37/44) vs. 73% (36/49), while antibiotic-related adverse events were reported in 9% (4/44) vs. 14% (7/49) for the 3- vs. 6-week therapy arms, respectively.
Finally, Benkabouche et al. randomized 123 patients with osteomyelitis and various orthopaedic implants to 4 vs. 6 weeks of antibiotic therapy.(Benkabouche, Racloz et al. 2019) All infected implants were surgically removed.(Benkabouche, Racloz et al. 2019) Prosthetic materials included 44 orthopaedic plates, 11 orthopaedic nails, 39 prosthetic joints, and 30 miscellaneous orthopaedic hardware. A wide variety of antibiotics were used. The overall treatment success rate at a median of 2.2 years of follow up was 94% (58/62) vs. 95% (58/61) in the 4- vs. 6-week therapy arms. Removing the 39 patients who had PJI, the success rates in the 4- vs. 6-week therapy arms were 94% (44/47) and 100% (37/37), respectively.
Thus, the available RCT data are concordant with observational data in demonstrating that six weeks of therapy is adequate for osteomyelitis, irrespective of hematogenous or contiguous routes of infection and presuming adequate source control. Furthermore, specifically in the setting of DFO, two small RCTs indicated that shorter regimens may be effective, ranging from 10 days after complete removal of infected bone to three weeks with debridement of necrotic tissue. Additionally, one RCT suggested that with removal of implanted materials with debridement, four weeks of antibiotics may be adequate to treat osteomyelitis. These data are concordant with observational studies suggesting that four weeks may be adequate for routine cases presuming source control is achieved, and it is conceivable that no therapy—and certainly no more than 2-10 days of therapy—is needed after total resection of infection with clear margins.
Therefore, we make a Clear Recommendation that no more than six weeks of antibiotic therapy should be administered for hematogenous or contiguous pyogenic osteomyelitis, presuming adequate source control. Some clinicians may prefer to use a 3- or 4-week regimen for appropriate cases with adequate debridement of bone, with small RCTs in support. However, sufficient data are lacking to definitively establish a Clear Recommendation between 4 vs. 6 weeks of therapy. We are also unable to make a Clear Recommendation regarding the duration of therapy after complete resection of infected bone with clear margins. Available data suggest that it may be reasonable to administer no antibiotics in this setting if the treating clinicians are confident all infected bone was removed, and if antibiotics are administered, durations beyond 10 days have not been shown to be of benefit, and we do not recommend exceeding 5 days of therapy based on consensus.
Observational data on durations of IV antibiotic therapy, total duration of therapy, or use of chronic suppression after completion of initial therapy, have demonstrated mixed results in patients with PJI, resulting in equipoise without a clear evidence of consistent benefit of longer therapy.(Brandt, Sistrunk et al. 1997, Crockarell, Hanssen et al. 1998, Silva, Tharani et al. 2002, Berdal, Skramm et al. 2005, Laffer, Graber et al. 2006, Marculescu, Berbari et al. 2006, Byren, Bejon et al. 2009, Rodriguez, Pigrau et al. 2010, Lin, Vasudevan et al. 2016, Tornero, Morata et al. 2016, Chaussade, Uckay et al. 2017, Becker, Kreitmann et al. 2020, Escudero-Sanchez, Senneville et al. 2020, Pansu, Hamoui et al. 2020)
Among several larger studies, Chaussade et al. retrospectively reported the outcomes of 87 patients with PJI who underwent DAIR and were treated with 6 or 12 weeks of total antimicrobial therapy.(Chaussade, Uckay et al. 2017) They reported no difference in long-term treatment success between these two groups at a median of more than 3 years of follow up. Similarly, Tornero et al. evaluated 143 patients who underwent DAIR.(Tornero, Morata et al. 2016) They, too, reported no relationship between duration of therapy (median 77 days) and treatment failure. Specifically, patients who experienced treatment success vs. failure had similar durations of antibiotic therapy, and a similar proportion of patients who were treated for >75 or >100 days with antibiotics. Furthermore, duration of antibiotic therapy was not associated with treatment failure by multivariable analysis. Rodriguez et al. described 50 patients with hematogenous PJI.(Rodriguez, Pigrau et al. 2010) Neither antibiotic treatment for >8 weeks or >12 weeks was associated with differences in treatment outcomes by multivariable analysis, either in the overall cohort (including the 16 patients treated with a two-stage joint replacement), or the DAIR cohort of 34 patients.
Finally, Byren et al. evaluated 112 patients with PJI who underwent DAIR.(Byren, Bejon et al. 2009) During a mean follow up of 2.3 years, 20 (18%) patients experienced treatment failure. There was no relationship identified between the duration of the initial course of antibiotic therapy and the risk of treatment failure. Treatment failures were then placed on chronic oral suppressive antibiotic therapy (mean of 1.5 years of therapy during follow up). After chronic suppressive therapy stopped, there was an increase in the risk of relapsed infection over the ensuing several months. Nonetheless, the total proportion of failures was relatively low (<15%), indicating that most patients did not fail after stopping chronic suppression.
However, other studies have suggested an advantage of longer therapy courses. One case series of 60 patients with PJI found that post-debridement antibiotic durations of < 3 months were associated with an increase in treatment failures by multivariable analysis.(Letouvet, Arvieux et al. 2016) Other observational studies have reported that the clinical success of chronic oral suppression after a 4 to 6 week course of “induction” antibiotic therapy can result in higher rates of treatment success at long-term follow up.(Segreti, Nelson et al. 1998, Rao, Crossett et al. 2003, Siqueira, Saleh et al. 2015)
Shah et al. conducted multivariable analysis for treatment success among 108 patients with PJI treated with DAIR, of whom 47% received chronic oral suppression after an initial IV course of antibiotics.(Shah, Hersh et al. 2020) Use of chronic suppression was associated with a significantly higher rate of long-term treatment success than not (HR 2.5; p < 0.009). However, there was no difference in treatment success rates for patients treated with <1 year or > 1 year of oral suppression. There was also no difference in the rate of adverse events associated with antibiotic use between those who were on antibiotics for < 6 weeks vs. > 6 weeks. In another study, 89 patients with infected orthopaedic implants, including spinal hardware, internal fixation devices, and PJIs, were studied to determine if chronic oral antibiotic suppression altered risk of long-term treatment failure.(Keller, Cosgrove et al. 2016) By multivariate analysis, receipt of 3 months of chronic oral suppressive antibiotics was associated with a reduced risk of relapse, however extending chronic suppression to 6 months was not.
Finally, a larger study from 2020 described the outcomes of 302 patients with PJI treated with chronic antibiotic suppressive therapy.(Escudero-Sanchez, Senneville et al. 2020) The mean patient age was 75 years, and more than a quarter of the patients were aged >85 years. Suppressive therapy was administered for a median of 3 years, with an interquartile range of 1.7 to 5 years. Tetracyclines and TMP-SMX were the most commonly administered agents, followed by b lactams and fluoroquinolones. The overall success rate was 59%; the 2-year follow-up success rate was 75% and declined to 50% at 5-year follow-up. By multivariable analysis the primary predictors of failure were age >70 years and infection caused by gram-positive cocci compared to other pathogens. Twenty-seven percent of patients suffered from adverse events, primarily gastrointestinal and cutaneous; 1% of patients developed C. difficile colitis. Adverse events were severe enough to necessitate cessation of suppression in 6% of patients and change to an alternative antibiotic in 15%.
A systematic review of the literature in 2020 found that the evidence in favor of chronic suppression with antibiotics after PJI treated with DAIR was limited and of low quality; therefore, it was insufficient to draw meaningful conclusions regarding safety and efficacy.(Malahias, Gu et al. 2020) They described a 15% rate of adverse events (the nature of which were not well delineated in the review), and a 75% long-term treatment success rate, which is not substantively different than other studies (such as those described above) that did not use chronic suppression. A second systematic review in 2021 reported an 8%-43% adverse event rate for chronic suppression depending on the study, and suggested that chronic suppression could be considered for patients whose implants cannot be removed, including unfavorable surgical risk:benefit ratio, short life-expectancy, or patient refusal.(Cobo and Escudero-Sanchez 2021)
For implants that are intended to be removed following bone healing, it may be possible to treat until the bone is stable enough for removal, and then complete a 4-6 week course of therapy after the implant is removed. Furthermore, providing the bone heals to stability, it may be reasonable to stop therapy even without removal of the implant. In the event of a relapse, the surgeon may then remove the implant without the need for complex reconstructions, after which a 6-week course of therapy could be applied. We did not find data evaluating the efficacy or tolerability of such a practice. Nevertheless, it may be an option to spare the potential for chronic oral suppression for patients who can safely have implants removed. Indeed, wherever possible, it is likely preferable to remove the implant to allow a higher chance of long-term remission as opposed to retaining the implant followed by prolonged chronic oral suppression or 'test of cure' with the implant still in place.
Ultimately, delaying relapse for patients who are too high-risk to tolerate repeat surgery may be an important goal. This goal must be weighed against the potential harm of the antibiotics and any suppressive treatment should be reviewed at regular intervals, since new treatment toxicities may tip the harm:benefit balance. As Byren et al. wrote, “One might conclude that most patients cured of PJI by DAIR are cured early on, and that prolonged antibiotic therapy does not prevent treatment failures in those who are not cured, but merely postpones them...Life-long antibiotics might simply postpone, rather than prevent, treatment failure, but this may be all that is required for older patients with limited life expectancy. For patients in whom further surgery might be limb- or life-threatening, postponing this outcome with indefinite antibiotic treatment is also justified”.(Byren, Bejon et al. 2009)
In an open label study, Lora-Tamayo et al. randomized 63 patients with early onset PJI managed surgically with DAIR to eight weeks vs. three months of treatment with levofloxacin and rifampin.(Lora-Tamayo, Euba et al. 2016) Early onset infection was defined as occurring within 30 days of implantation. They found very high success rates in both arms of the study in the per protocol analysis, with 92% (22/24) and 95% (19/20) success rates at a median of 862 days of follow up in the short vs. longer therapy arms. However, there were seven more dropouts from the intention-to-treat (ITT) population in the longer therapy arm, due to higher rates of adverse events (13% vs. 18%), orthopaedic failure of the implant (0% vs. 6%), and lost to follow up (3% vs. 12%). As a result, in the ITT population, the short course therapy regimen had a more favorable success rate than longer therapy at 73% (22/30) vs. 58% (19/33). Indeed, a meta-analysis of this RCT with nine other retrospective studies concluded that shorter course therapy regimens had similar outcomes to longer, and that eight weeks of therapy was adequate for hip PJIs and 75 days (just under 7 weeks) was adequate for knee PJIs treated with DAIR.(Yen, Hsieh et al. 2019)
As discussed above, an RCT of 4 vs. 6 weeks of antibiotics for patients with infected implants included 39 patients with PJI.(Benkabouche, Racloz et al. 2019) These PJI patients all underwent two-stage replacements, and hence removal of the initially infected implant. On inquiry with the corresponding author, the treatment success rates in the 4 vs. 6 weeks arms for these 39 patients were 93% (14/15) vs. 88% (21/24), respectively, at two years of follow up.
However, most recently Bernard et al. conducted a larger, open-label, multi-centered study (DATIPO) of 410 patients with PJI randomized to receive 6 vs. 12 weeks of antibiotic therapy.(Bernard, Arvieux et al. 2021) A wide variety of antibiotic therapy was used, with 70% of patients receiving rifampin therapy, 68% a fluoroquinolone, and 51% both. The median duration of IV therapy administered was only nine days, with the remainder of therapy in both arms administered orally. Approximately 41% of patients were treated surgically with DAIR, 37% with one stage prosthetic implant exchange, and 22% with two stage exchange. The primary endpoint of treatment success in the modified ITT population was significantly lower in the short course therapy arm, at 82% (158/193) vs. 91% (173/191), for an adjusted difference of 9% (2%-16%). Treatment success in the per protocol population was also significantly lower for the short-course therapy arm at 82% (136/165) vs. 93% (149/160), for an adjusted difference of 11% (4%-18%). Given the very high treatment success rate (91% in the ITT population, 93% in the per protocol) with a 12-week antibiotic course, there would likely be diminishing returns to prolonging therapy beyond 12 weeks.
However, the treatment difference was most dramatic in the DAIR cohort, with a success rate 16% (3-30%) higher in the 12-week arm. Indeed, 23 of 32 of the treatment failures in the 6-week arm occurred in the DAIR cohort (with only six in the 2-stage and three in the 1-stage exchange cohorts). The differences in efficacy between long vs. short therapy for the one and two stage exchanges were not statistically significant, at 1.2% better (95% CI, 5% worse to 8% better) and 10% better (95% CI, 3% worse to 23% better) for longer therapy, respectively. Furthermore, among the patients treated with six weeks who underwent knee prosthetic exchanges, there were no failures, whereas the infection rates were higher for hip prosthesis exchanges. For patients undergoing 2-stage exchange, in addition to 6 vs. 12 weeks of antibiotics, the time between exchanges also varied at 6 vs. 12 weeks. Thus, the morbidity associated with a further six weeks prior to definitive prosthesis replacement should be considered. Thus, there may remain a role for six weeks of therapy in some patients undergoing prosthetic exchanges.
Another note of caution about the DATIPO RCT is that there were important imbalances in pathogens causing infections in the 6- vs. 12-week therapy arms. Specifically, the 6-week arm had 20 more infections caused by S. aureus which has, in general, higher rates of treatment failure. The clinical failure rate for S. aureus infection was 23% (21/90) vs. 13% (9/70) in the 6- vs. 12-week arms, suggesting that 12 weeks of therapy may improve outcomes of S. aureus infection, specifically. The 12 more clinical failures due to persistent S. aureus infections in the shorter therapy arm accounted for 71% (12/17) of all excess failures in the 6-week arm, across all surgical subtypes (DAIR, 1-, or 2-stage exchanges).
Overall, based on clear superiority in the largest RCT conducted, all participating experts unanimously prefer 12 weeks of therapy for PJI treated with DAIR; a Clear Recommendation is not made because WikiGuidelines evidentiary standards require two prospective, controlled studies, and a second study is not yet available to address this question.
Some experts also prefer 12 weeks of therapy for PJI treated with 1- or 2-stage exchanges given the trends to favorable outcomes in the DATIPO RCT, particularly for patients with hip prosthesis (as opposed to knee) or with S. aureus. However, because microbiological imbalances in the larger RCT could have disadvantaged the 6-week therapy arm, failures were not seen in patients undergoing knee exchanges, and very high cure rates were achieved with either 4 or 6 weeks of antibiotics in patients undergoing 2-stage exchanges in the smaller RCT,(Benkabouche, Racloz et al. 2019) some equipoise remains regarding antibiotic durations for PJI caused by pathogens other than S. aureus, particularly for patients managed surgically with 1-stage exchanges or 2-stage exchanges with negative cultures prior to reimplantation.
Of note, for patients undergoing 2-stage exchanges, duration may also be affected by the nature of the care intervening the exchanges. Surgical management without a spacer after explantation may rationally support a 4-6 week duration of therapy.(Benkabouche, Racloz et al. 2019) However, this procedure was not used in the DATIPO RCT, in which patients undergoing 2 stage revisions may have had a spacer between procedures. In DATIPO, patients undergoing such exchanges of the hip had an imbalance in failures favoring the 12-week regimen.
More RCT data are needed to definitively establish the optimal treatment duration of patients with PJI undergoing prosthetic exchange.
Clear Recommendation |
Clinical Review |
|
Osteomyelitis without retained implant (including DFO) | Maximum 6 weeks |
|
Osteomyelitis with total resection of infected bone |
|
|
PJI with DAIR |
|
|
PJI with Exchange |
|