(The following is a close reading of the OVIVA trial, newly published in NEJM. If you’re not interested in that, skip to the bottom of this page for links to this month’s articles).
OVIVA, or Oral Versus IV Antibiotics for bone and joint infections, was a pragmatic, open-label randomized controlled trial that aimed to establish the non-inferiority of oral antibiotics for the treatment of bone and joint infections. The authors enrolled adults at 26 UK medical centers with native osteomyelitis (vertebral with and without discitis as well as non-vertebral), native joint infection, prosthetic joint infection (whether managed with DAIR or excision +/- reimplantation), and other orthopedic hardware infections, who the treatment teamed deemed would ordinarily receive 6 weeks of intravenous therapy. The exclusion criteria notably included S. aureus bacteremia, endocarditis, septic shock, or any other concurrent infection deemed to require IV antibiotics, non-bacterial (e.g. fungal) infection, and clinician judgement that the patient was unlikely to adhere to therapy. The predetermined threshold for non-inferiority was originally set at 5% (this was later expanded to 7.5% given a higher-than-expected failure rate in an interim analysis, but the change proved inconsequential).
Participants were randomized 1:1 within a week of either surgical intervention or starting antibiotics (if no surgery was planned) to either intravenous or oral antibiotic therapy. Infectious diseases specialists chose both the IV and oral regimens in every case. Investigators allowed use of oral rifampin in the IV arm and <5 day courses of IV antibiotics to treat other infections in the oral arm. The primary endpoint was definite treatment failure at 1 year, defined as clinical, microbiologic, or histologic evidence of relapsed or recurrent infection at the same site; all possible treatment failures were classified by consensus by a blinded committee of three ID specialists as not treatment failure, possible treatment failure, probable failure, and definite failure.
The authors enrolled a total 1054 patients, of whom 1015 could be included in the modified intention-to-treat analysis. Participants randomized to each arm were similar in baseline characteristics; 61% had a hardware infection, 8% were treated without surgical intervention, and 77% were treated based on microbiologic data (i.e. rather than empirically). Importantly, antibiotic therapy was extended in 77% of cases, but to similar degrees (median 11 total weeks in the IV group and 10 total weeks in the oral group). The outcome of definitive treatment failure occurred in 14% of all participants (15% in the IV group and 13% in the oral group). The difference in treatment failure risk with oral antibiotics was -1.4% (that is, favoring oral therapy; 95% CI -5.6 to 2.9), meeting both the 5% and 7.5% thresholds for non-inferiority. In a “worst case scenario” analysis, where all patients lost to follow-up were presumed to have had failure with oral treatment and success with IV treatment, oral antibiotics still proved non-inferior. In subgroup analysis by antibiotics given (or rather planned to be given) and causative pathogen, oral and IV antibiotic treatment failure rates were similar in every case.
With regards to secondary endpoints, there were no differences in the rates of probable or possible treatment failures with oral versus IV therapy (1.2% vs 2%). Early discontinuation of therapy occurred more frequently in the IV group (19% vs 13%; p=0.006), as did IV catheter complications (9% vs 1%, p<0.001 and no surprise). Overall adverse events rates were similar, but median hospital stays were longer with IV therapy (14 vs 11 days; p<0.001). Medication adherence monitoring in a subgroup of patients given oral therapy revealed that more than 90% of participants took more than 95% of the oral antibiotic doses prescribed.
What antibiotics were used? For IV drugs it was vancomycin (41%) and cephalosporins (33%); the most commonly used oral drugs were quinolones (37%) and combinations of multiple classes (17%). Oral rifampin was analyzed separately, having been given in 16% of IV and 52% of oral cases, and was not associated with treatment outcome. Taking a more granular look (see figure S2), oral antibiotics faired particularly well versus glycopeptides (OR 0.6, 95% CI 0.3-1.1) and particularly poorly versus carbapenems (OR 2.8, 95% CI 0.7-11) – and in fact the carbapenem subgroup is the only one whose confidence interval crosses their non-inferiority threshold of 7.5 (this comes with the caveat that the study wasn’t powered to prove the non-inferiority of oral drugs versus carbapenems). As for which oral antibiotics performed best? Here are the data in Fig S2 expressed as simple failure rates:
Failure rates (%)
Planned PO treatment: Ended up getting PO Ended up getting IV
Penicillins 23% 16%
Quinolones 10% 13%
Tetracyclines 15% 19%
Clinda/Macrolides 9% 12%
Other single drugs 17% 21%
PO combinations 13% 13%
Patients planned to receive oral penicillins were the only folks to have a higher failure rate in the oral vs IV arm. Out of curiosity, I ran a chi square test comparing the outcomes of patients planned to get oral penicillins who were randomized to oral therapy versus everyone randomized to IV therapy. This analysis boosts the sample size and hence statistical power, and its implicit argument is “maybe osteo is osteo and the causative pathogen doesn’t matter” – which might be a bad argument, but hey, my blog isn’t NEJM and I’m not beholden to Reviewer 2. Anyway, oral penicillins recipients’ 23% failure rate was higher than the 16% failure rate for patients given any IV drug, with p = 0.052. Understanding that the 0.05 cutoff for statistical significance is a bit arbitrary, this gives me pause about recommending oral penicillins for osteomyelitis until more data are available.
So where does that leave us? Some final thoughts:
1) Given the limited exclusion criteria, OVIVA’s results are highly generalizable, much more so than POET. In light of these data (and, yes, the trials that preceded this one), I see no reason the practice should not become the universal default approach to managing bone and joint infections - as I mentioned last month, it’s time to change the conversation to “why does this bone and joint infection need IV therapy?”
2) Speaking of which, for whom should we still consider IV therapy for bone and joint infections? Based on these data, I’d still consider IV in cases with S. aureus bacteremia, for patients who have significant barriers to adherence to oral therapy, and perhaps in the occasional case in which a carbapenem would be preferred (e.g. osteomyelitis due to Enterobacter with an inducible AmpC phenotype and resistance to TMP/SMX and quinolones).
3) Where were TMP/SMX and the oral cephalosporins in this study? TMP/SMX simply wasn’t used often enough for a separate analysis, and according to our UK ID colleagues on twitter, cephalexin and its cousins are not widely available or used in the United Kingdom. In addition, some UK ID folks I communicated with indicated that amoxicillin-clavulanate probably represented the bulk of the oral penicillins used in this study, based on their impression of colleagues’ prescribing habits. To me then, OVIVA leaves open the question of whether oral cephalosporins are appropriate for bone and joint infections. For one, clavulanate is not terribly bioavailable and for another, recent data suggest amoxicillin-clavulanate is not the most effective choice of beta-lactam for MSSA and that we might particularly expect poorer outcomes at sites of infection where drug levels are suboptimal (e.g. infected bone in a patient with microvascular disease due to diabetes). A counterpoint to this is that there are data showing comparable outcomes for amox-clav and linezolid in diabetic foot ifnections, but the samples sizes there were small, so who knows? Anyway, I wouldn’t count out the utility of high dose cephalexin or cefadroxil based on the uninspiring results of penicillins here.
4) Note that not all of the macrolide/lincosamide use was clindamycin – clarithromycin and erythromycin saw use in a half-dozen cases. I don’t typically think of macrolides as drugs for bone and joint infections, and their outcomes weren’t reported separately from clindamycin here , but this may be worth revisiting given that macrolides are starting to get renewed attention as add-on therapy for MDR gram-negative and other pathogens.
5) Finally, the tetracyclines performed well. Moreover, that was with the preponderance of tetracyclines prescribed being doxycycline – which some experts have suggested is inferior to minocycline. Minocycline is more robust against resistance and achieves marginally better levels in tissues than doxycycline, and some experts (e.g. Dr. Cunha at NYU) have opined that minocycline is the tetracycline of choice for MRSA. This is welcome news, as I’ve been using doxy and mino for osteomyelitis off and on with good anecdotal results but knowing the literature to support it is scant.
And now for January’s medical literature:
Antimicrobial agents literature this month covered the adequacy of ceftriaxone dosing in ICU patients with superb kidney function, the synergies of dalbavancin, daptomycin, and vancomycin with beta-lactams, and the prominence of erythromycin as a risk factor for cardiac death in CAP.
ID diagnostics research in January included the diagnostic utility of extended-duration blood cultures, the (lack of) utility of pan-scan CTs for endocarditis, culture data review for the prediction of antimicrobial resistance in gram-negative bacteremia, and the predictive power of the MRSA nasal swab for MRSA infection at other body sites.
General ID literature this month included risk factors and treatment strategies associated with P.vivax malaria, outcomes in ORIF hardware infections managed with device retention, the duration of sexual transmission risk after acute Zika infection, and a review of the duration of therapy for osteomyelitis.
HIV and STD research published this month concerns the use of antiretroviral monotherapy for maintenance of HIV suppression (don’t do it!), the pharmacokinetic acceptability of combining dolutegravir and darunavir-cobi, data on why the South(‘s viral load) will rise again, and a new, altruism-focused model for increasing STD testing among MSM in China.
Onc and transplant ID research this month described a new endemic fungus, Blastomyces helicus, in the western US and Canada, the risk of genital warts after renal transplantation, and a comprehensive review of the published experience with mucormycosis.
Antimicrobial stewardship and infection control literature this month included the inefficacy of contact isolation for preventing transmission of MDR E.coli, the safety of withholding antibiotics while working up most cases of complicated UTI, OPAT’s utility in people who use injection drugs, and the importance of hospital-wide cephalosporin use as a driver of C.difficile infection.