Partial Oral versus Intravenous Antibiotic Treatment of Endocarditis
Iversen K, Ihlemann N, Gill SU, et al.. N Engl J Med 2019. PMID 30152252
Background
Left-sided infective endocarditis is classically treated with up to 6 weeks of IV antibiotics. The authors note that in-hospital mortality ranges from 15% to more than 45% and that most complications happen early, so for many patients the main reason to stay in hospital after the first phase is just to finish the IV course. Outpatient IV therapy is an option, but it brings catheter logistics and monitoring burdens. Evidence for oral step-down in left-sided endocarditis was sparse, so POET tested it head to head.
Methods
| Design | Nationwide, investigator-initiated, multicenter, randomized, unblinded noninferiority trial at cardiac centers in Denmark (July 2011 to August 2017). Events were judged by an adjudication committee unaware of treatment assignment. 1:1 web-based randomization, stratified by site. |
|---|---|
| Population | 400 stable adults (18+) with left-sided endocarditis (native or prosthetic valve) meeting modified Duke criteria, with blood cultures positive for streptococci, Enterococcus faecalis, Staphylococcus aureus, or coagulase-negative staphylococci. Out of 1954 screened, 20% were enrolled. |
| Intervention | Switch to oral antibiotics (201 patients), discharged to outpatient care if feasible and seen 2 to 3 times a week. Every oral regimen paired two drugs from different classes with different mechanisms and metabolism, chosen by species and MIC testing (EUCAST); drug levels were measured on day 1 and day 5. The specific combinations are in the supplement ⚠ data needed. |
| Comparator | Continue IV antibiotics in hospital (199 patients) per ESC guidelines with Danish modifications. |
| Primary endpoint | Composite of all-cause mortality, unplanned cardiac surgery, clinically evident embolic events, or relapse of bacteremia with the primary pathogen. Noninferiority margin: 10 pp (risk difference), 90% power. |
| Follow-up | From randomization until 6 months after antibiotics ended; visits at 1 week and 1, 3 and 6 months. No patients were lost to follow-up. |
| Key eligibility | Stable condition with a good response to at least 10 days of IV antibiotics (and at least 7 days after valve surgery, if done); no abscess or valve problem needing surgery on transesophageal echo before randomization; at least 10 days of planned treatment left. |
| Sites | Cardiac centers across Denmark; the article text doesn't give the number of sites ⚠ data needed. |
| Who was enrolled | 77% men, mean age 67 yr; 35% had a major coexisting condition; a prosthetic valve was affected in 27%; 38% had valve surgery before randomization. No patient had MRSA. Median 17 days from diagnosis to randomization. |
| Pathogens | Streptococci were most common (52.3% IV / 45.8% oral), then E. faecalis (23.1% / 25.4%), S. aureus (20.1% / 23.4%), and coagulase-negative staph (5.0% / 6.5%). |
Results
After randomization, antibiotics ran a median 19 days (IV) vs 17 days (oral), P=0.48. But hospital stay after randomization (not a prespecified outcome) dropped from 19 days to 3 days, and 80% of the oral group were treated partly or entirely as outpatients.
The primary composite outcome happened in 12.1% on IV vs 9.0% on oral: a difference of 3.1 pp (95% CI −3.4 to 9.6; P=0.40). The lower bound of the CI stayed well inside the 10-point margin, so oral was noninferior. The odds ratio was 0.72 (95% CI 0.37 to 1.36).
By component: deaths 6.5% vs 3.5% (HR 0.53, 95% CI 0.21 to 1.32); unplanned surgery 3.0% vs 3.0%; embolic events 1.5% vs 1.5%; relapse of bacteremia 2.5% vs 2.5%. Six patients (three per arm) had two events.
Robustness: per protocol, 9.1% of oral patients had an event (difference 3.0 pp, 95% CI −3.2 to 9.2). Counting the 4 oral-to-IV crossovers as failures, the difference was 1.2 pp (95% CI −5.6 to 7.5), still noninferior.
Safety: adverse antibiotic effects in 6% (IV) vs 5% (oral), P=0.66, mostly allergy, bone marrow suppression, and GI effects. In 7 oral patients one of the two drugs ran below target (rifampicin, moxifloxacin, linezolid, or dicloxacillin), but the partner drug was adequate and none had a primary outcome event. Regimen changes were more common on IV (22%) than oral (12%).
The paper doesn't define formal secondary efficacy outcomes in the main text. The extra outcomes above are components, sensitivity analyses, and safety data.
Figure 2 is a Kaplan-Meier plot with a numbers-at-risk table, but the text gives no cumulative event percentages at intermediate time points, so no curve was recreated (points would have to be read off the image).
Primary composite outcome
Data table
Between-group difference vs the noninferiority margin
Data table
The four components of the primary outcome
Data table
Median hospital stay after randomization
Data table
Median days of antibiotics after randomization
Data table
Adverse antibiotic effects
Data table
Pathogens at baseline
Data table
Conclusions
In stable patients with left-sided endocarditis, switching to oral antibiotics was noninferior to continuing IV treatment, and it cut the hospital stay sharply. Noninferior means "not meaningfully worse" by the pre-set 10-point margin, not "better", even though the oral group's rate was numerically lower.
Limitations
Only left-sided endocarditis caused by four bacterial groups; results may not apply to the 25 to 30% caused by other bacteria or to culture-negative cases. Only five people who inject drugs were enrolled, only 22% had S. aureus, and nobody had MRSA or other resistant phenotypes. Strict entry criteria (only 20% of screened patients were randomized), possible referral bias, an unblinded design, and outpatient discharge left to patient and physician choice. Done in Denmark, a low-resistance setting.
What changed in practice
Within the article, the authors say clinicians should use the trial's strict entry criteria when deciding to shift a patient from IV to oral therapy, and that a larger share of left-sided endocarditis patients may be candidates for partial oral therapy. They don't consider pharmacokinetics a barrier as long as two well-absorbed, susceptibility-matched drugs are used and gut uptake is normal. Guideline changes after 2019 aren't in this article ⚠ data needed.
Source: Iversen K, Ihlemann N, Gill SU, et al.. Partial Oral versus Intravenous Antibiotic Treatment of Endocarditis. N Engl J Med 2019. doi:10.1056/NEJMoa1808312. Figures recreated from published data. Educational summary, not medical advice.