Esther Y. van Hulten¹, Lianne Messchendorp², Henk Scheper¹, Mark G. J. de Boer¹

¹ Department of Infectious Diseases, Leiden University Medical Center, Leiden, the Netherlands
² Department of Medical Microbiology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands

Unravelling treatment success in enterococcal prosthetic joint infection: a systematic review with individual patient data

---

Background

Enterococcal prosthetic joint infection (PJI) is often associated with polymicrobial co-infection, antimicrobial resistance and limited treatment options, resulting in a poor prognosis.

Aim
This review estimates the outcome of enterococcal PJI in different subgroups, focusing on methodological quality of included studies.

Methods

• PubMed and Embase until 8 January 2025
• Adults with PJI caused by Enterococcus spp.; two reviewers screened, extracted and appraised quality (Newcastle–Ottawa)
• Outcome: treatment success as defined by each study
• Pooled proportion — random effects, Freeman–Tukey, REML, Hartung–Knapp, with a 95% prediction interval

Snapshot

• 11 — studies included in the review
• 638 — enterococcal PJI episodes (10 studies)
• 194 — patients with individual patient data (5 studies)
• 53% — of infections polymicrobial

---

Figure 1 · Study selection (PRISMA 2020)
[PRISMA-diagram]
7,264 records identified · 5,009 screened · 235 full texts assessed · 11 studies included

Table 1 · Study population — all studies vs IPD cohort

N differs per row: only episodes for which the characteristic was reported are counted (available-case analysis).

---

Figure 2 · Pooled treatment success — primary analysis

74.0% pooled treatment success
95% CI 61.9–84.6% · 95% prediction interval 36.2–98.7% · I² = 84.9% · k = 10, n = 594

[forest plot]

Random-effects model (Freeman–Tukey, REML, Hartung–Knapp). Wouthuyzen-Bakker excluded from pooling (cohort overlap with Tai), retained in the IPD analysis. With heterogeneity this high, the prediction interval — not the point estimate — is what a clinician should read as the expected range in a new cohort.

Methodological considerations about included studies

• Treatment groups not mutually exclusive in 8 of 11 studies
• No comparative study exists: all 11 cohorts are observational and single-arm studies
• Confounding by indication present in all studies
• Newcastle–Ottawa quality assessment: of 7 studies assessed, 5 FAIR and 2 GOOD
• GRADE certainty for the comparative questions expected to be very low — observational design, high heterogeneity, small numbers
• Publication bias not formally tested: with 11 studies a funnel plot is not interpretable

Pre-specified contrasts — in progress

• B1 · Monotherapy vs combination therapy — El Helou, Kheir, Renz, Maurille (k = 4), pooled as a risk ratio
• B2 · With vs without rifampicin — Tornero, Thompson, Maurille, Davis (k = 4), pooled as a risk ratio
• Individual studies point in opposite directions and none reached significance

---

Used antimicrobial strategies targeting enterococcal PJI
[staafdiagram: β-lactam, Glycopeptide, Other, Aminoglycoside, Linezolid, Daptomycin — All studies vs IPD cohort]

Percentage of episodes in which each class was used. Denominators overlap — a patient on combination therapy appears in more than one bar — so these bars describe use, not effectiveness.

---

Conclusions

• 11 studies, including 638 enterococcal PJI episodes, with individual patient data for 194 patients from 5 studies.
• Pooled treatment success: 74.0% (95% CI 61.9–84.6%), with a 95% prediction interval of 36.2–98.7% reflecting substantial heterogeneity (I² = 84.9%) across studies.
• Significant methodological constraints across the included studies mean that no scientific conclusions can be drawn about the most adequate antimicrobial treatment strategy for enterococcal PJI.

e.y.van_hulten@lumc.nl