Fellow's quick read · Infectious Diseases
Prosthetic joint infection
IDSA 2012 (now archived) + Michigan dental-prophylaxis CPG 2026 · reviewed 2026-06-28
Personal study digest for a new ID fellow. Recommendations are from the IDSA 2012 PJI guideline (Osmon, CID 2013) and the Michigan dental-prophylaxis CPG; the "What's changed since 2012" section is reviewer synthesis of newer evidence, each claim cited. Not a substitute for the full guideline.
In one line
PJI is a biofilm disease — cure needs surgery + a pathogen-specific course, and the duration/route hinge on one question the orthopedist answers first: is the hardware staying in or coming out?◆When to suspect / diagnose
- Suspect PJI with: a sinus tract or persistent wound drainage over the prosthesis, an acutely painful prosthesis, or any chronically painful prosthesis — especially with no pain-free interval, early after implantation, or a history of wound-healing problems (rec 1).
- Get ESR + CRP in everyone when the diagnosis isn't obvious (the combination has the best sensitivity/specificity) plus a plain radiograph (recs 3–4).
- Arthrocentesis in suspected acute PJI: send synovial cell count + differential and aerobic/anaerobic culture (rec 5). Advanced imaging (bone/leukocyte scan, MRI, CT, PET) is not routine for diagnosis (rec 8).
- Definitive PJI (recs 12–16): a communicating sinus tract; OR purulence around the prosthesis; OR ≥2 cultures (or preop + intraop) growing the same organism; acute inflammation on periprosthetic histopathology is strongly supportive.
◆Workup the fellow drives
- Hold antibiotics ≥2 weeks before sampling the joint (and before intraoperative cultures) if the patient is stable — low-virulence biofilm organisms are easily suppressed and missed (recs 6, 11).
- At surgery, send ≥3 (optimally 5–6) periprosthetic tissue samples for culture — single CoNS/Cutibacterium growth may be a contaminant; matching organisms across samples make it real (recs 10, 15).
- Blood cultures if febrile, acute onset, or an organism (e.g., S. aureus) that makes bacteremia likely (rec 7).
◆Empiric & definitive therapy
Match the IV agent to the pathogen (Table 2): MSSA = cefazolin 2 g IV q8h or nafcillin; MRSA = vancomycin (AUC-guided); Pseudomonas = cefepime 2 g IV q12h; strep = penicillin G or ceftriaxone. Duration & rifampin depend on the surgery:
| Surgical strategy | Antimicrobial backbone | Duration · strength |
|---|---|---|
| Staph PJI, hardware RETAINED (DAIR or 1-stage) | IV antistaph 2–6 wk + rifampin 300–450 mg PO BID, then rifampin + a fluoroquinolone (cipro or levo) PO | 3 mo (hip) / 6 mo (knee) · A-I |
| Non-staph PJI, hardware retained | Pathogen-specific IV or highly bioavailable oral | 4–6 wk · B-II |
| Hardware REMOVED (2-stage / resection) | Pathogen-specific IV or highly bioavailable oral (no rifampin needed — no retained biofilm) | 4–6 wk · A-II |
| Amputation (all infected tissue out) | Pathogen-specific | 24–48 h only (longer if sepsis/bacteremia) · C-III |
◆Key decisions a fellow owns
- Is DAIR an option? Debridement + implant retention is appropriate only when the implant is well-fixed, there's no sinus tract, AND it's acute — within ~30 days of implantation OR <3 weeks of symptoms (rec 18). Outside that window, relapse is high → exchange.
- Use rifampin only with retained hardware (DAIR or 1-stage) and never as monotherapy (rapid resistance). It is the biofilm-active partner; pair it with a fluoroquinolone.
- Withhold antibiotics before cultures — the single most fixable cause of culture-negative PJI.
- Chronic oral suppression (Table 3) is optional after the defined course — reserved for patients unsuitable for/declining further surgery; rifampin is not used for suppression (recs 25, 28, 33, 36).
- Surgical strategy is the orthopedist's call with ID/plastics input (rec 17) — but you own the antibiotic plan that follows from it.
◆Special populations
- Shoulder / indolent PJI: think Cutibacterium acnes (renamed from Propionibacterium) — hold anaerobic cultures up to 14 days.
- Dental prophylaxis: routine antibiotic prophylaxis before dental work is NOT recommended for a prosthetic joint (AAOS/ADA). Michigan CPG carves out selected high-risk patients to consider it — prior PJI, severe immunocompromise, or someone unlikely to tolerate future revision: amoxicillin 2 g (or cephalexin 2 g; doxycycline 100 mg) PO once, 1 h pre-procedure. Clindamycin is no longer recommended (adverse effects/resistance).
- Penicillin allergy: MSSA → vancomycin or daptomycin if a true severe allergy; otherwise cefazolin is fine.
◆Duration & stopping
- The defining number: ~12 weeks total for staph PJI with retained hardware (3 mo hip / 6 mo knee); 4–6 weeks once the hardware is out.
- End therapy at the planned course if the joint is clinically quiet; there is no validated biomarker that defines "cure." Pre-reimplantation ESR/CRP help judge success before a staged second stage (rec 19).
What's changed since 2012
Reviewer synthesis of newer evidence — not the guideline. Each claim cited; flags where the 2012 IDSA guideline lags current practice.
- No newer IDSA PJI guideline — the 2012 document is now formally "Archived." The best contemporary synthesis is a 2023 NEJM review (Patel). NEJM 2023;388:251-262; doi:10.1056/NEJMra2203477 · PMID 36652356
- Duration: PJI is the exception to "shorter is better." DATIPO — 6 weeks was NOT noninferior to 12 weeks (persistent infection 18.1% vs 9.4%; risk difference 8.7 pp, 95% CI 1.8–15.6). 12 weeks stays standard; don't reflexively shorten. Bernard, NEJM 2021;384:1991-2001; doi:10.1056/NEJMoa2020198 · PMID 34042388 · NCT01816009
- Oral step-down is validated. OVIVA — oral noninferior to IV for the first 6 weeks of bone/joint infection incl. PJI (failure 13.2% PO vs 14.6% IV; diff −1.4 pp, 90% CI −4.9 to 2.2), fewer line complications. The 2012 text assumed prolonged IV. Li, NEJM 2019;380:425-436; doi:10.1056/NEJMoa1710926 · PMID 30699315
- Rifampin's A-I rec rests on thin evidence. It comes from one small 1998 RCT (Zimmerli, n=33; PMID 9605897). The only modern dedicated RCT was null (Karlsen 2020: remission 74% vs 72%, RR 1.03, 95% CI 0.73–1.45) but underpowered; it remains standard on observational support, and the RiCOTTA RCT (ongoing) is testing whether the oral phase can drop it. Karlsen, J Orthop Surg Res 2020; doi:10.1186/s13018-020-01877-2 · PMID 32859235 · RiCOTTA PMID 40456484
- Diagnosis modernized. The 2012 binary definition is operationally superseded by tiered frameworks — ICM 2018 and EBJIS 2021 (infection unlikely / likely / confirmed) — adding synovial WBC/PMN% thresholds (≥3,000 cells/µL or ≥80% PMN), alpha-defensin, leukocyte esterase, and sonication of explanted hardware. EBJIS, Bone Joint J 2021 (PMID 33380199); ICM 2018, J Arthroplasty (PMID 29551303)
- Still current: the DAIR eligibility window, surgical-strategy framework, and pathogen-specific agent choices hold. Recency-sensitive: confirm no new IDSA PJI guideline has published and whether RiCOTTA has reported.
◆Anki cards minted this run
- Staph PJI, hardware retained → rifampin + fluoroquinolone; 3 mo hip / 6 mo knee (rec 23, A-I).
- Hold antibiotics ≥2 weeks before joint/intraop cultures (recs 6, 11).
- PJI synovial thresholds (WBC ≥3,000/µL or PMN ≥80%) + tiered ICM 2018 / EBJIS 2021 definitions.
- Rifampin's A-I evidence is weak — Zimmerli 1998 vs null Karlsen 2020 RCT; RiCOTTA ongoing.
Held as already in deck: DAIR indications (note 1658682885261); DATIPO ≥12-week duration (1699233094924); Cutibacterium→shoulder (1658682644183); PJI microbiology timeline; single-stage indications; dental-prophylaxis cards; cefazolin surgical prophylaxis (CID 2026).
Sources: IDSA PJI guideline (Osmon, CID 2013;56:e1-e25; doi:10.1093/cid/cis803); Michigan dental-prophylaxis CPG 2026; DATIPO (PMID 34042388); OVIVA (PMID 30699315); Zimmerli rifampin RCT (PMID 9605897); Karlsen rifampin RCT (PMID 32859235); RiCOTTA protocol (PMID 40456484); EBJIS 2021 (PMID 33380199); ICM 2018 (PMID 29551303); NEJM PJI review (PMID 36652356); AAOS/ADA dental-prophylaxis guidance.