KneeDaily

16.2

Antibiotic Infection Prophylaxis in Anterior Cruciate Ligament Reconstruction: Preoperative, Intraoperative, and Postoperative

Yongtao Maov1.0Published 47 references
Evidence through 2026.09.08Chinese original
Contents

The author's practice for primary anterior cruciate ligament reconstruction (ACLR) is as follows: begin infusion of cefazolin 2 g (3 g if body weight ≥120 kg) within 60 minutes before incision and complete the infusion before tourniquet inflation; soak the graft in 5 mg/mL vancomycin solution for 10–15 minutes before implantation (most studies wrap the graft in soaked gauze); do not use antibiotics postoperatively. For patients who report β-lactam allergy, first identify the culprit drug and the type of reaction, then decide whether to switch to vancomycin (Section 2).

The three measures rest on different kinds of evidence. A single preoperative dose of cephalosporin is the consistent recommendation of the guidelines, but ACLR has no comparative evidence of prophylaxis versus no prophylaxis, and the "81% relative risk reduction" from arthroplasty that the guidelines cite is not the effect of systemic prophylaxis alone. Graft vancomycin soaking, added on top of systemic prophylaxis, shows a consistent reduction in infection in pooled analyses dominated by Level III retrospective cohorts (Figueroa, after deduplication, RR 0.067); to date there is no adequately powered randomized trial with clinical infection as the end point. No ACLR data show benefit from extending antibiotics postoperatively; results of extended oral regimens in arthroplasty are inconsistent, and the infection pattern, population, and baseline risk all differ from ACLR, so those findings cannot be applied.

1. Infection Rates, Pathogens, and Sources of Infection

Infection Rates Reported in Different Studies

The figures in the table below hold only under each study's own outcome definition and observation window:

Source Unsoaked group Infection rate Outcome and observation window
Naendrup 2019 pooled analysis[7] 44/2,099 2.1% Early septic arthritis, follow-up 6–52 weeks
Figueroa 2022 forest plot[8] 121/15,249 0.79% Postoperative infection, definitions varied across studies
Hu 2023 control group[9] 146/19,713 0.74% Postoperative infection
Carrozzo 2022 institutional cohort[32] 11/3,228 0.34% Septic arthritis
Kottek 2025 historical cohort[10] 26/1,891 1.4% Septic arthritis, 2007–2017, hamstring autograft
Offerhaus 2019 full cohort[42] 22/926 2.4% Deep knee infection, assessed at 6-week postoperative visit

The highest and lowest differ by more than sixfold. Part of this difference comes from differences in outcome ascertainment, length of follow-up, era of surgery, and population composition; real risk differences among studies may also exist. These figures have not been subjected to heterogeneity analysis, so the contribution of each factor cannot be separated.

Pathogens

ESSKA and EBJIS data show that 60%–90% of infections after ACLR are staphylococcal, with coagulase-negative staphylococci the most common, followed by Staphylococcus aureus[2]. Of the 22 infections in Offerhaus, 16 (73%) were coagulase-negative staphylococci[42].

Hematogenous Infection: Rare, but Reported

ESSKA/EBJIS writes: "To our knowledge, no cases of hematogenous joint infection after ACLR have been reported in the literature"[2]. This statement does not hold; at least two published cases exist:

  • Eranki 2013: a 19-year-old man developed knee infection 4 months after hamstring-tendon ACLR; blood culture, knee fluid, and an iliacus abscess all grew Staphylococcus aureus; the authors judged hematogenous spread from the iliacus abscess to the reconstructed knee[46].
  • Neri 2021: a 23-year-old patient presented 26 days after hamstring-tendon ACLR and was diagnosed with Salmonella Typhimurium knee infection; gastrointestinal symptoms had begun 10 days before the joint symptoms. The authors inferred hematogenous spread from an enteric source; blood cultures were negative, and the route of dissemination was not confirmed[47].

Case reports can show only that such events have occurred; they cannot be used to estimate incidence. The great majority of infections after ACLR are early postoperative infections; hematogenous infection is extremely rare, but it has been reported.

Graft Contamination and Clinical Infection

Bacterial contamination during graft harvest and preparation is the prevailing hypothesis for the source of infection and the mechanistic rationale for vancomycin soaking. Direct evidence comes from three studies:

  • In Pérez-Prieto 2018, 7 of 50 cases (14%) had positive cultures at harvest or preparation; all converted to negative after soaking in 5 mg/mL vancomycin (P<0.001)[29]. This conclusion is limited to the strains tested and to what culture can detect.
  • The meta-analysis by Thiagarajah 2025 included 12 prospective studies and 678 grafts, with a pooled contamination rate of 12.3% (95% CI 7.8%–19%); Staphylococcus epidermidis was the organism most often isolated. Of 583 cases with postoperative follow-up, none of the 78 with culture-positive contamination developed clinically apparent infection, whereas the 3 infections that occurred (0.51%) all had negative intraoperative cultures[30]. This observation sits on top of existing prophylaxis: 11 of the 12 studies used intravenous prophylaxis, and 109 of the 678 grafts also received local soaking, so "contamination without infection" does not mean contamination is harmless.
  • In Scheu 2025, a prospective study of 31 cases all received vancomycin soaking; 16S sequencing showed bacterial DNA in 11 (35.4%), and there was no early clinical infection at a mean follow-up of 150 days[31]. DNA positivity includes dead organisms and environmental sources and is not equivalent to culture of viable bacteria.

The ESSKA/EBJIS wording is consistent with this: "contamination as a source of infection is a theory that has not been fully validated; in fact colonization rates far exceed infection rates"[2].

Contamination during harvest and preparation does occur and can be cleared by vancomycin; that much is supported by direct observation. Existing clinical data are not yet sufficient to establish it as the source of most postoperative infections.

Graft Type

Bansal's meta-analysis included 21 Level 1 and 2 studies; 5 of them with comparable events yielded a pooled result: deep infection after bone–patellar tendon–bone autograft was 77% lower than after hamstring tendon (RR 0.23, 95% CI 0.097–0.54)[12]. ESSKA/EBJIS adopted this conclusion[1]. That soft-tissue grafts involve more preparation steps and longer extracorporeal exposure is a reasonable explanation, but no study has directly measured the relationship between number of preparation steps or exposure time and infection; the explanation remains an inference.

Timing of Prophylaxis

Contamination during harvest, preparation, and implantation occurs intraoperatively; prophylaxis directed at this source must cover the operative period: adequate serum concentration at incision, and local treatment completed before graft implantation. The basis for not giving antibiotics postoperatively is the comparative duration-of-therapy studies in Section 4.

2. Preoperative: Systemic Antibiotic Prophylaxis

Drug, Dose, and Timing of Administration

ESSKA and EBJIS jointly recommended in 2023 a single dose of cephalosporin 30–60 minutes before incision[1]; the companion document specifies cefazolin 2 g (3 g if body weight >120 kg) or cefuroxime 1.5 g (3 g if body weight >120 kg)[2]. The ASHP/IDSA/SIS/SHEA joint guideline specifies cefazolin 2 g, or 3 g if body weight ≥120 kg, with infusion begun within 60 minutes before incision; vancomycin and fluoroquinolones require longer infusion and should be started within 120 minutes before incision[3].

The wording of the Chinese 2015 guiding principles differs: intravenous infusion should be given within 0.5–1 hour before incision or at the start of anesthesia, and surgery should begin after the infusion is complete[25]. Dosing at the start of anesthesia does not necessarily fall within 60 minutes before incision; the Chinese text also does not address the timing of tourniquet inflation. A practice that satisfies both ASHP and the Chinese principles is to give the dose within 0.5–1 hour before incision; completing the infusion before tourniquet inflation is an additional requirement of the author's, for reasons given below.

Tourniquet

The author advocates completing the full dose before tourniquet inflation, on pharmacologic grounds: once inflated, drug can no longer enter the operative field. This practice has no clinical end-point evidence behind it. ASHP regards completing the full dose before inflation as "appearing to make sense," while noting that the only study with a direct comparison was in total knee arthroplasty: one group received cefuroxime 10–30 minutes before tourniquet inflation and the other 10 minutes before deflation, with no difference in surgical-site infection rates between groups[3]. The Chinese 2015 principles require only that "surgery begin after the infusion is complete," without mentioning the tourniquet, and cannot serve as a regulatory basis for this practice.

Intraoperative Redosing

Intraoperative redosing depends on two conditions—time and blood loss—and ASHP and the Chinese principles time them differently:

  • Time: ASHP uses an interval of twice the drug half-life, counted from the start of the first infusion; for cefazolin this is about 4 hours[3]. This interval applies to patients with normal renal function; when renal function is reduced the half-life lengthens and the redosing interval changes accordingly. The Chinese 2015 principles write "when operative time exceeds 3 hours or exceeds twice the half-life of the drug used"[25]; the 3 hours is counted as operative time and cannot be treated as the same rule as counting from the first dose.
  • Blood loss: redose when adult blood loss exceeds 1,500 mL; both documents agree[3,25]. Vancomycin and aminoglycosides are not redosed in this setting[5].

Most ACLR procedures do not meet these conditions. Waiting from induction to incision, multiligament reconstruction, or concomitant osteotomy can lengthen the interval from the first dose; whether to redose is judged by the actual interval and blood loss.

Lack of Direct Comparative Evidence for Systemic Prophylaxis in ACLR

ESSKA/EBJIS writes: "Studies evaluating the efficacy of intravenous prophylaxis in ACLR are lacking. However, Armstrong et al. and Carney et al. found fewer infections in the prophylaxis groups"[2]. In both original papers the infection figures in the prophylaxis groups were indeed lower, but neither can demonstrate the efficacy of systemic prophylaxis in ACLR:

  • Armstrong 1992 [43]: a case–control analysis of infections after knee arthroscopy over 4 years (the Abstract writes 4,256 cases; Methods and Results write 4,264; the original is internally inconsistent). In Table 5, 2 of 18 infected cases and 64 of 434 sampled controls received prophylaxis, OR 0.7. The original states explicitly that this "mild protective effect" was not statistically significant, and notes in the Discussion that because the infection rate is so low, demonstrating prophylactic efficacy would require an extremely large sample. This is not a positive comparative study.
  • Carney 2018 [41]: a review of 19 studies, of which 7 entered the pool: 5 compared prophylaxis versus no prophylaxis in procedures without a graft, and 2 compared soaking in ACLR. In non-graft knee arthroscopy, the prophylaxis group had 27/34,487 (0.08%) and the no-prophylaxis group 16/10,911 (0.15%), RR 0.53 (95% CI 0.29–0.99, P=0.05); a subgroup excluding bony procedures showed no significant difference. All included ACLR studies used prophylaxis, so the paper cannot provide a comparison of prophylaxis versus no prophylaxis in ACLR—a limitation of the data themselves, not an oversight of the paper. Its ACLR portion compared soaking versus no soaking (19/1,095 or 1.74% versus 0/2,034, RR 0.01, 95% CI 0.001–0.229).

Arthroplasty Efficacy Data Cited by the Guidelines

ESSKA/EBJIS writes: "A pooled analysis by AlBuhairan et al. of 7 studies with 3,065 arthroplasty cases showed that a single dose of cephalosporin 30–60 minutes before incision reduced relative risk by 81%"[2]. Returning to the AlBuhairan original[26]:

  • The pooled comparison was antibiotic prophylaxis versus none, RR 0.19 (95% CI 0.12–0.31, P<0.00001), absolute risk reduction 8%, NNT 13, I²=0%. The original did not stratify by timing of administration; "30–60 minutes" is a condition added when the guidelines paraphrased it.
  • Among the 7 pooled studies were two knee-arthroplasty trials by Chiu et al. that used cefuroxime-impregnated bone cement; the comparator was not preoperative systemic prophylaxis alone, so RR 0.19 cannot be taken as the pure effect of a single preoperative intravenous cephalosporin.
  • The absolute risk reduction of 8% and NNT of 13 reflect the high baseline infection rates of these early trials, which are not in the same order of magnitude as contemporary ACLR infection rates.

The basis for a single preoperative cephalosporin is therefore the consistent recommendation of the guidelines, biologic mechanism, and data from adjacent populations; there is no direct comparative evidence in ACLR.

β-Lactam Allergy

For patients who report allergy, first identify the culprit drug and the type of reaction; this step has clear evidence of benefit. In a Mayo Clinic analysis of 29,695 primary TKA/THA procedures (22,705 patients) from 2004–2017, 1,521 procedures (5.1%) used a regimen other than cefazolin; after adjustment for ASA class, procedure type, and BMI, those who received cefazolin had a 32% lower risk of PJI (HR 0.68, 95% CI 0.48–0.96, P=0.027); the NNT to prevent one PJI with cefazolin prophylaxis was 164 at 1 year and 84 at 10 years. Of 2,576 patients who underwent preoperative allergy evaluation, 97% were cleared for cephalosporin use, and the authors therefore recommended preoperative allergy evaluation for all who report penicillin or cephalosporin allergy[4]. These are arthroplasty data; ACLR has no corresponding study. Extending this result to ACLR is an extrapolation, based on the absence of any known difference between the two procedures in the mechanism by which switching drugs worsens antimicrobial coverage. The Wyles study itself can show only association.

The Stanford perioperative prophylaxis guideline (version dated 2025-07-02) stratifies by side chain and reaction type[5]:

  • Patients with a Type I reaction (including a history of anaphylaxis) to penicillin, amoxicillin, amoxicillin–clavulanate, ampicillin, or cephalexin can still safely receive cefazolin. Cephalexin itself is a cephalosporin and can be listed in this group because it does not share an R1 side chain with cefazolin.
  • For severe delayed Type IV reactions to β-lactams (SJS/TEN, DRESS, AGEP), cross-reactivity is unclear and cefazolin may be contraindicated. The Type II–IV reactions listed in that guideline also include serum sickness, acute interstitial nephritis, hemolytic anemia and other cytopenias, and drug fever; its surgical classification table specifies that patients with a documented Type II–IV β-lactam reaction do not receive cefazolin and that orthopedic procedures switch to vancomycin. The 2022 drug-allergy practice parameter likewise lists severe cutaneous adverse reactions as circumstances in which challenge testing is not performed[28].
  • When a cephalosporin other than cefazolin is used (for example cefoxitin), side-chain cross-reactivity with other β-lactams must be assessed separately.

Patients with a severe immediate reaction (anaphylaxis) to a cephalosporin other than the drugs listed above cannot be given the drug simply because "the side chains differ." The pathway in the 2022 drug-allergy practice parameter is: first judge whether the prior reaction was a severe immediate reaction, then examine the side-chain relationship between the culprit and the intended drug, then perform skin testing or graded challenge as needed, and decide on the basis of the result[28]. Differing side chains are only one step and cannot replace evaluation.

The author manages reported allergy as follows:

Prior reaction Management
Type I reaction to penicillin, amoxicillin, amoxicillin–clavulanate, ampicillin, or cephalexin, including a positive penicillin skin test Use cefazolin
Type I reaction to cefazolin itself Switch to vancomycin
Positive cefazolin skin test Evaluate first; do not give directly
Confirmed severe delayed β-lactam reaction (SJS/TEN, DRESS, AGEP) or other Type II–IV immune reaction Switch to vancomycin
Severe immediate reaction to another cephalosporin Decide after evaluation per the 2022 practice parameter; do not give directly; if evaluation cannot be completed in time, switch to vancomycin

"Confirmed" means the reaction course is clearly documented. When the history is unclear, first verify the culprit drug and the course of the reaction, then classify by the table above.

ESSKA/EBJIS is more conservative than Stanford and lists "Type I allergy to penicillins or cephalosporins" as a whole as an indication to switch to vancomycin[1,2]. Adopting that uniform switch rule is also an acceptable choice.

When a glycopeptide is needed, ESSKA/EBJIS gives vancomycin 1 g, or 15 mg/kg up to a maximum of 2,500 mg, as a single dose[1,2]; the ASHP adult dose is 15 mg/kg[3]. Stanford provides weight bands and corresponding infusion durations that are easier to implement[5]:

Body weight Dose Infusion duration
<80 kg 1 g 60 minutes
80–99 kg 1.25 g 75 minutes
100–120 kg 1.5 g 90 minutes
>120 kg 2 g 120 minutes

Vancomycin should be started 60–120 minutes before incision[3,5]; the Chinese 2015 principles write 1–2 hours before surgery[25]. A 2 g dose requires a 120-minute infusion and should be started 120 minutes before incision; scheduling must also account for the time of tourniquet inflation.

Clindamycin 900 mg is another alternative listed by international guidelines[3]; the problem is staphylococcal resistance. Stanford 2024 hospital-wide susceptibility data showed MSSA sensitivity of 77% to clindamycin and 100% to vancomycin, and MRSA sensitivity of 63% to clindamycin; that institution therefore makes vancomycin first choice for β-lactam-allergic patients[5]. These are single-institution data and cannot represent resistance patterns in China. The author places clindamycin last and requires checking the institution's own susceptibility before use.

Teicoplanin is not among the alternative agents for perioperative prophylaxis in Chinese standards. The Guiding Principles for Clinical Use of Antibacterials (2015 edition), under perioperative prophylaxis, lists vancomycin, norvancomycin, and clindamycin for Gram-positive coverage in patients allergic to cephalosporins[25]. At the level of Chinese standards, the glycopeptide alternatives are therefore vancomycin and norvancomycin; the latter is an option specific to China. As of the search date, no direct comparative study of teicoplanin versus vancomycin for prophylaxis in ACLR was retrieved. Clinical use of teicoplanin is individual practice and cannot be treated as a standard recommendation.

MRSA Colonization, Prior Knee Infection, and Recent Antibiotic Therapy

The three documents differ in the populations they address and in their wording:

  • ESSKA/EBJIS: for known MRSA colonization, prior knee infection, or recent hospitalization or antibiotic therapy, the prophylactic regimen may be changed to a single dose of vancomycin[1,2]. The original uses "may," not a mandate; "recent" hospitalization or antibiotic therapy also has no defined time window, and any unit-specific window should be labeled as a unit rule.
  • Stanford: for known MRSA colonization, consider adding vancomycin or clindamycin to cefazolin[5]. This recommendation addresses only MRSA colonization, not the other history items above, and the wording is likewise "consider."
  • Chinese 2015 principles: for procedures with severe infectious consequences such as joint arthroplasty, when MRSA colonization is suspected preoperatively or the institution has a high MRSA rate, vancomycin or norvancomycin may be chosen for prophylaxis, with strict control of duration[25]. The orthopedic entry lists first- and second-generation cephalosporins as first choice for "arthroplasty, osteotomy, internal fixation, and spinal surgery (with or without implants or fixation)"[25]; ACLR is not listed separately and is extrapolated from this implant-surgery entry.

None of the three is a mandate for uniform application; when a unit adopts one, it should state which document it follows.

Preoperative Intra-articular Injection and Aspiration

Intra-articular corticosteroid injection is a risk factor for infection after ACLR[1]. Two bodies of evidence separated by more than thirty years point in the same direction:

  • In the Armstrong 1992 knee-arthroscopy case–control study, intra-articular long-acting corticosteroid (methylprednisolone acetate) had an infection OR of 27.4 and was the strongest risk factor in that study[43]; Armstrong 1994, investigating an outbreak of infection among 352 knee, shoulder, and ankle arthroscopies, likewise pointed to intra-articular steroid and environmental factors[44]. The exposure in these two studies was intraoperative or end-of-procedure intra-articular injection, which is not the same exposure as the "injection within 8 weeks before surgery" in Wainwright et al.; the two cannot substitute for each other, nor can they be used to calculate how long to wait after a preoperative injection. Both concerned mixed arthroscopic procedures, not ACLR specifically.
  • Wainwright 2025 used the TriNetX database for 1:1 propensity-score matching, 2,439 cases in each group. Infection at 90 days was 1.2% (30/2,439) after injection within 8 weeks before surgery versus 0.6% (14/2,439) in controls, OR 2.1 (95% CI 1.1–4.0, P=0.015); at 180 days, 1.3% versus 0.6%, OR 2.2 (95% CI 1.2–4.1; the Abstract writes P=0.009 and the Results body P=0.006; the original is internally inconsistent)[6]. Infection was defined as a formal infection diagnosis or an operation requiring irrigation. The database was extracted by diagnosis and procedure codes and cannot distinguish primary from revision procedures, nor confirm the specific joint or side of the injection.

How long to delay surgery after injection is something ESSKA/EBJIS explicitly says has "no data on optimal timing"; its suggestion of 6 months is an inference from systematic reviews of total knee arthroplasty and is stated to need more supporting evidence[1].

Preoperative joint aspiration: the infection risk after aspiration of a native joint is <0.01%. ESSKA/EBJIS considers that post-injury hematoma may carry a higher risk because of the inflammatory response, but gives no figure; its indications for aspiration are limited to restricted motion, inability to activate the quadriceps, or intolerable pain from the hematoma[1].

3. Intraoperative: Graft Vancomycin Soaking

Evidence for the efficacy of graft vancomycin soaking is dominated by Level III retrospective cohorts and their pooled analyses; the direction of effect across cohorts is consistent. After removal of overlapping populations, Figueroa's pooled analysis was 1/7,894 in the soaked group and 121/15,249 in controls, RR 0.067 (95% CI 0.029–0.159)[8]. The two published randomized trials have very small samples; to date there is no adequately powered randomized trial with clinical infection as the end point. Several reviews repeatedly include the same set of cohorts; the total numbers each reports are not the numbers that entered the pooled analysis and cannot be added together.

Pooled Analyses and Primary Cohorts

Vertullo 2012 [13] first described the technique: after graft preparation, wrap the graft in Raytec gauze soaked in 5 mg/mL vancomycin in saline (500 mg dissolved in 100 mL saline) until implantation; each gauze absorbs about 7 mL of solution and contains 35 mg of vancomycin; controls used identical gauze soaked in saline. The original total case count is internally inconsistent: Abstract, Methods, and Discussion all write 1,135 cases, whereas Methods, Table 1, and Results give 285 and 870 in the two groups, summing to 1,155, with a difference of 20 unexplained. Mean follow-up differed between groups (5.8 versus 2.6 years), so the reported graft failure rates of 13/285 (4.6%) and 24/870 (2.75%) cannot be compared directly.

Naendrup 2019 [7]: a systematic review of 8 studies, of which 4 entered the infection pool, totaling 5,075 cases (intravenous prophylaxis alone 2,099; intravenous prophylaxis plus soaking 2,976), follow-up 6–52 weeks. Results were 44/2,099 (2.1%) versus 0/2,976, OR 0.04 (95% CI 0.01–0.16). The Abstract calls 5,075 "the number of patients examined in 8 studies," which does not match the Results; that figure is in fact the number in the 4 infection comparisons. Revision rate, IKDC, Tegner score, tendon biomechanical properties, and cartilage integrity were summarized only descriptively without pooled comparison; some of the biomechanical and cartilage data came from in vitro experiments. The authors rated their own evidence level as IV (a systematic review of Level III and IV studies).

Figueroa 2022 [8]: 11 reports, 24,298 cases—this is the reported total. Adding the 11 rows of their Table 2 gives 1/8,764 soaked and 125/15,534 controls; this sum double-counts Vertullo's 870 soaked and 285 control cases (Table 1 already notes that Phegan added only 430 soaked and 0 control cases relative to Vertullo). The actual pooled analysis (Figure 2 forest plot) is the deduplicated 10 comparisons: soaked 1/7,894, controls 121/15,249, Mantel–Haenszel pooled RR 0.067 (95% CI 0.029–0.159), I²=0.0%, all Level III retrospective cohorts. Of the 11 studies, 10 used 5 mg/mL and 1 used 1 mg/mL; the single infection occurred in the 1 mg/mL study; 10 used soaked gauze wrapping and 1 used direct immersion; 8 reported soaking duration, ranging from 10–20 minutes.

Hu 2023 [9]: 13 studies, 31,150 cases, soaked 11,437 and unsoaked 19,713, reported infection rates 0.09% versus 0.74%, OR 0.17 (95% CI 0.10–0.30). There is an extraction error in the soaked-group events; 0.09% and OR 0.17 cannot serve as estimates of the soaking effect.

Hu's Table 2 and forest plot record Offerhaus 2019 as 8/853 infections in the soaked group; Offerhaus's original[42] Table 1 (full cohort of 1,779) has 0/853 infections in the soaked group; 8 is the number of graft reruptures in another sampled subgroup in Table 3 (257 primary four-strand semitendinosus reconstructions). Of the 10 total soaked-group events Hu lists, 8 come from this row—80%; the corresponding tables in Figueroa and Carrozzo correctly list 0/853.

Hu also included both Vertullo (0/870 versus 4/285) and Phegan (0/1,300 versus 4/285), so the same 285 control cases were counted twice. The control group's 146/19,713 (0.74%) is not invalidated by the soaked-group error but is still affected by this double-counting. Subtracting these 8 events from the total does not yield a usable corrected value; both deduplication and pooling would need to be redone.

Carrozzo 2022 (SANTI) [32]: an institutional primary cohort of 5,300 cases, unsoaked 11/3,228 (0.34%), soaked 1/2,072 (0.05%). The 29,659 in the title is the reported total of its systematic review: Table 4 lists 11 external studies and the institutional cohort, 12 comparisons in all; this figure is neither a single-center count nor a deduplicated independent patient count and cannot be added to other reviews.

Pooled results were OR 14.39 (95% CI 5.90–35.10) for all grafts and OR 13.50 (95% CI 4.03–45.21) for the hamstring subgroup, the latter with a fragility index of 16; the forest plot retains both Vertullo and Phegan, so this 16 cannot be read as robustness after excluding overlapping populations. Stratified by graft, the association was robust in the hamstring subgroup and only a trend for bone–patellar tendon–bone. The effect size for the institutional cohort has two values in the original: Abstract and Results give OR 5.13 (95% CI 1.16–48.30), and the corresponding row of Figure 3 gives OR 7.08; the original does not explain the difference in estimation methods.

Offerhaus 2019 [42]: a historical control study, with the same surgical team operating at two centers. Full cohort of 1,779 autograft ACLR (including different grafts and revisions), unsoaked 22/926 (2.4%), soaked 0/853 (0%); all infections occurred within 5 weeks postoperatively, and the full cohort was assessed for infection at the 6-week visit. The study also drew 424 primary four-strand semitendinosus reconstructions for a complication analysis (unsoaked 167, soaked 257, mean follow-up 37 months): infection 7/167 versus 0/257; graft rerupture 16/167 (9.6%) versus 8/257 (3.1%); arthrofibrosis 11/167 versus 27/257, difference not significant. Both groups received systemic prophylaxis. This subgroup was sampled rather than randomly allocated, and the original also discusses alternative explanations such as longer follow-up in controls; the between-group difference in rerupture cannot be attributed directly to vancomycin.

Kottek 2025 [10]: among 1,329 soaked patients from January 2018 to November 2023, 1 infection (the original writes 0.07%; the precise value is 0.075%), follow-up 12 months; controls were 1,891 cases at the same center from January 2007 to December 2017 (26 cases, 1.4%). The two periods are adjacent and together span nearly 17 years. The original states that technique and protocol were otherwise identical apart from soaking; a letter about the paper raised concern about era confounding[39]—a concern, not confirmed evidence of a change in practice. The letter also notes that the pathogen in the single soaked-group infection was Enterobacter cloacae, which is outside the antimicrobial spectrum of vancomycin[39].

Bohu 2020 [11]: 1,674 cases (unsoaked 1,184, soaked 490), sample size calculated for a noninferiority test; infections 7 versus 0. Return to running at 1 year was 901 (76.1%) unsoaked versus 372 (75.9%) soaked, no difference (the Abstract reverses the order of the two numbers; figures here follow the grouping in Table 4). Return to preinjury sport was 52.2% versus 57.5% (P=0.04, unadjusted). In Table 1 the two groups did not differ in activity-level composition (competitive plus professional athletes: unsoaked 526/1,184 or 44.4%, soaked 216/490 or 44.1%, n.s.); what did differ significantly was graft type (hamstring 82.5% versus 71.8%, P<10⁻⁴, Table 2). The multivariate analysis included study group together with age, sex, activity level, prior surgery, preoperative scores, additional extra-articular reconstruction, medial meniscal injury, and serious complications within one year. This sport outcome must be read together with the original's baseline and technique differences and the adjusted analysis; it cannot be taken as proof that soaking has been shown to affect return to sport.

Randomized Trials: Sample Size Insufficient to Test the Infection End Point

Two published randomized studies were retrieved:

  • Mousavi 2023 [33]: 80 patients randomized into two groups of 40, both receiving preoperative intravenous cefazolin; the experimental group's graft was placed in 5 mg/mL vancomycin solution for 10–15 minutes; follow-up 1 year. Neither group had an infection, and clinical examinations did not differ. Sample size was calculated on functional and physical-examination indices; the design purpose was to check for adverse effects of soaking on the graft. A 0-versus-0 result can prove neither prophylactic efficacy nor equivalence of the two treatments.
  • Gourab 2026 [34]: 76 patients randomized into two groups of 38; controls soaked in saline; infection rates 0% versus 5.26% (2 cases), P=0.152. The primary outcome was judged by postoperative CRP, and the specific threshold for diagnosing infection is inadequately reported; follow-up time points are internally inconsistent (Abstract writes functional scores at 6 months; body and Tables 3 and 4 write "to 3 months"). The control infection rate of 5.26% is far above levels commonly reported in the literature.

A randomized trial comparing intravenous prophylaxis alone with intravenous prophylaxis plus soaking has been registered (NCT06166381, Duhok University, planned 288 patients, started 2024-01-25, estimated primary completion 2024-12)[27]. Queried on 2026-09-08, status was "Recruiting," last submitted update 2024-12-26 and posted update 2024-12-30; no results have been posted.

Soaking Is an Add-On to Systemic Prophylaxis

All comparisons are "intravenous prophylaxis" versus "intravenous prophylaxis plus soaking"; soaking cannot replace systemic prophylaxis[7,8,32,42]. Figueroa also states explicitly that "prophylactic intravenous antibiotics remain the standard of care, and replacing them with soaking is not recommended"[8]. The data are dominated by hamstring autograft; bone–patellar tendon–bone stratification shows only a trend[32], and allograft data are insufficient.

Technique

  • Concentration: 5 mg/mL, used by the great majority of studies and recommended by ESSKA/EBJIS[1,8].
  • Method and duration: ESSKA/EBJIS recommends soaking the graft in 5 mg/mL solution for 10–15 minutes[1]. Durations actually reported in clinical studies are 10–20 minutes[8]; Hu recorded about 10–15 minutes for the studies it included[9].
  • Origin of "needs 20 minutes": this wording comes from a Schüttler experiment paraphrased in Figueroa's Discussion: porcine tendons soaked with sterile dressings at 1, 2.5, 5, or 10 mg/mL for 10 or 20 minutes; the conclusion was that at 5 mg/mL, 20 minutes was needed to clear all tested tendons[8]. The same paragraph also paraphrases Grayson's experiment: after gauze wrapping for 10 minutes at rest, the amount of drug released already reached the staphylococcal MIC[8]. The Schüttler original could not be obtained and is an animal experiment; it cannot determine the optimal duration for human ACLR. Existing data do not support making 20 minutes an operative requirement; the durations given by different sources are not consistent.
  • In vitro concentration and duration experiments: Parinyakhup 2025 inoculated human semitendinosus tendons with 4 strains of 3 bacterial species (Staphylococcus aureus, two strains of Staphylococcus epidermidis with and without biofilm production, and Enterococcus faecalis), compared different regimens within 5 minutes, and reported that 5 mg/mL direct immersion or 10 mg/mL gauze wrapping completely cleared bacteria, whereas gauze wrapping at lower concentrations was insufficient; under all conditions the mean relative viability of tenocytes and mesenchymal stem cells remained above 70%—group means, not a guarantee that every graft reached that level[36]. The original is internally inconsistent on gauze concentrations: Abstract and Results (including Table 1) give 2.5/5/10 mg/mL, whereas Methods section 4.1.3 gives 5/10/20 mg/mL. This conflict directly affects the operative conclusion; the result can serve only as an in vitro finding and cannot be used to change clinical practice.
  • Intra-articular exposure: Pfeiffer 2023 wrapped grafts in 5 mg/mL gauze in 20 cases (10 semitendinosus and 10 quadriceps tendon) and aspirated synovial fluid through the anterolateral portal after the harvest site and medial portal were closed; sampling was at a mean of 15.40 ± 5.33 minutes after implantation; mean synovial vancomycin concentration was 23.23 ± 21.68 µg/mL (lowest 2.32, highest 71.56 µg/mL), far below the in vitro chondrotoxicity threshold of 1,000 µg/mL[35]. This is a single synovial measurement that may miss the peak and is not a serum concentration; no significant difference was seen between the two graft types, but the sample is small and insufficient to establish equivalence.
  • Serum concentration: as of the search date, no formal paper measuring human serum vancomycin concentration after soaking was found. A 2026 ACL Study Group conference abstract reports serum concentration measurement; it is a preliminary conference report, the corresponding full text was not obtained, and it is not used as evidence.
  • Allergy: it is necessary to distinguish which drug the patient is allergic to. Type I allergy to β-lactams is unrelated to local graft vancomycin and is not a contraindication to soaking. Whether patients allergic to vancomycin itself can undergo local soaking is unknown; low intra-articular concentration does not mean zero systemic exposure, and no study has observed reactions in such patients, so safety cannot be inferred from concentration.

Unresolved Safety Issues

Weninger 2026 single-center data raise a safety signal: among 468 primary ACLR (hamstring 421, quadriceps tendon 47), all soaked for 15 minutes in 1% solution (that is, 10 mg/mL), 22 early graft failures (graft resorption) occurred within 12 months, all in the quadriceps tendon group (46.8% versus 0%)[37]. The study had no unsoaked quadriceps tendon control group; the authors state that they did not separately evaluate the role of vancomycin, and the cohort also includes referrals from other institutions. On this basis one cannot conclude that 5 mg/mL causes quadriceps tendon graft failure, nor treat 46.8% as a general failure rate; but use of quadriceps tendon is increasing, and some units use concentrations above 5 mg/mL, so this signal needs to be followed.

Data in the opposite direction come from Offerhaus's sampled subgroup: graft rerupture was less frequent in the soaked group than in controls (3.1% versus 9.6%), with no significant difference in arthrofibrosis[42]. This is a sampled comparison within a historical control study; on that basis one cannot establish that the reduction in rerupture was caused by vancomycin, nor exclude an effect of vancomycin.

4. Postoperative: No Extended Antibiotics

ESSKA/EBJIS holds that neither intravenous nor oral antibiotics are necessary postoperatively[1]; oral agents should be avoided because of side effects, low bioavailability, and consequent inadequate efficacy; prophylaxis beyond 24 hours increases side effects (such as microbiota disruption) and may even raise the infection rate[2].

Direct Evidence

As of the search date, no ACLR-specific study of extended postoperative antibiotics was retrieved. The comparative study that included ACLR patients is the 2025 retrospective cohort by Mirahmadi et al.: 398 cases, mixed TKA (113), THA (61), ACLR (106), and hip internal fixation (118); short-course group 246 (cefazolin 1 g preoperatively and one dose each at 6 h and 12 h postoperatively), extended group 152 (oral cephalexin to 7 days)[14].

  • Overall 3-month SSI: short-course 2.8% versus extended 4.6% (P=0.35; the Discussion also writes 2.7% and 4.6%; the original is internally inconsistent).
  • ACLR subgroup: short-course 1/80 (1.3%) versus extended 1/26 (3.8%), P=0.432.
  • Overall postoperative complications: extended 36.2% versus short-course 22.8% (P=0.004, OR 1.9, 95% CI 1.2–3.0; Discussion writes 35.5%), the difference driven mainly by swelling; antibiotic-related adverse effects such as diarrhea and allergy did not differ significantly in any procedure stratum.

The "short-course" arm of this study itself included doses at 6 h and 12 h postoperatively; the comparison is therefore roughly 24-hour coverage versus 7 days of oral therapy, not complete postoperative omission versus extended use. The authors also acknowledge that indications for extended use were not recorded, so the difference in complications can be understood only as association. With a small sample and mixed procedures, the only conclusion that can be drawn is that the extended regimen showed no benefit.

Duration-of-Therapy Data from Arthroplasty

The meta-analysis by Siddiqi et al. 2019 contains two different comparisons[15]:

  • Preoperative single dose versus preoperative plus continued postoperative dosing: RR 0.96 (95% CI 0.73–1.26). The authors state explicitly that this confidence interval still accommodates about 27% difference in either direction and is underpowered; it cannot be read as equivalence.
  • Postoperative ≤24 hours versus >24 hours: no difference.

All included studies were underpowered, antibiotic type and duration varied, and the overall GRADE rating was low.

Harm Signals from Extended Duration

The multicenter cohort by Branch-Elliman et al. 2019 included 79,058 surgical patients, with prophylaxis <24 hours as the reference: after duration was extended, there was no additional benefit for surgical-site infection (SSI), while adjusted risks of acute kidney injury (AKI) and Clostridioides difficile infection (CDI) increased[38]. The authors summarize in the Abstract and Discussion as risk rising with each additional day. Observation windows were SSI 30 days, AKI 7 days, and CDI 90 days; results by duration group were:

Duration AKI (noncardiac surgery aOR) C. difficile infection aOR Unadjusted number needed to harm for CDI
24–<48 h 1.31 (1.21–1.42) 1.08 (0.89–1.31) 2,000
48–<72 h 1.72 (1.47–2.01) 2.43 (1.80–3.27) 90
≥72 h 1.79 (1.27–2.53) 3.65 (2.40–5.53; Table 3 gives 5.55) 50

The AKI analysis included 71,344 cases; duration was calculated as the interval from first to last dose; numbers needed to harm were calculated from unadjusted incidence differences. In the grouped results, AKI was elevated in all three duration groups among noncardiac surgery patients; CDI was not significantly elevated in the 24–<48 h group (aOR 1.08, 0.89–1.31) and rose significantly only in the 48–<72 h and ≥72 h groups. The authors' day-by-day summary therefore cannot be read as a significant rise with every additional day. The study is dominated by multispecialty surgery in a veterans system, mean age 64.8 years, 96.3% male—not an ACLR population; it provides an order-of-magnitude reference for "duration-related harm from extended prophylaxis" and cannot be treated as expected values for ACLR.

Chinese Standards

The Guiding Principles for Clinical Use of Antibacterials (2015 edition) stipulate that total duration of prophylaxis for clean surgery should not exceed 24 hours; cardiac surgery may be extended to 48 hours as appropriate[25]. The text does not state whether these 24 hours are counted from the first or the last dose; concluding from it that "a full postoperative day is invariably a violation" goes beyond the text. What can be stated is that routine postoperative intravenous dosing for 2–3 days has neither ACLR evidence nor support within the duration allowed by the principles; practice should check the timing of the first dose against total coverage time.

5. Extended Oral Antibiotic Studies in Arthroplasty Cannot Be Applied to ACLR

The designs of these studies differ substantially: some are before–after comparisons within an institution, some matched cohorts from different periods at the same institution, some propensity-score-matched national database analyses, and one a meta-analysis pooling these studies. Direction of results cannot substitute for appraisal of design: most studies are single-center; between studies that report benefit and those that do not, the main differences lie in how controls were constructed and in scale. The three primary cohorts listed here that report benefit all used historical controls before and after a time point within an institution; that is a shared feature of these three and does not represent all studies in the field.

Primary Cohorts Reporting Benefit

  • Inabathula 2018 [16]: a single-center series of 2,181 primary TKA/THA from 2011–2016; from January 2015, high-risk patients received 7 days of oral antibiotics after discharge. Whole-cohort 90-day infection rates were 1.0% for TKA and 2.2% for THA. In binary logistic regression, high-risk patients who did not receive extended oral therapy had PJI ORs of 4.9 (95% CI 1.1–22.4, TKA) and 4.0 (THA), with very wide confidence intervals. The P=0.009 and P=0.037 appended to these two figures in the Abstract are tests of the "study group" term in the model, not the P values for each of these two comparisons. Level III evidence.
  • Kheir 2021 [17]: 3,855 cases at the same center from December 2011 to March 2019, same protocol. One-year PJI rates: low-risk 1.29% (14/1,082), high-risk untreated 2.64% (23/870), high-risk treated 0.89% (P<0.001). This is a follow-on study that expanded enrollment and lengthened the observation window on the base of Inabathula; the two overlap substantially in time and population and are not independent replications.
  • Bundschuh 2024 [45]: 2,982 cases by a single Emory surgeon from 2016–2022 (primary 2,677, aseptic revision 305); from January 2020 all patients were treated, mainly with cefdinir 300 mg bid for 7 days, with 15 switched to doxycycline 100 mg for cephalosporin allergy. Three-month PJI: overall 0.41% versus 1.13% (adjusted model P=0.02); primary TJA 0.23% versus 0.74% (P=0.04; PJI OR for untreated versus treated 3.85); aseptic revision showed only a trend. This is an independent-center result, likewise using historical controls bounded by a protocol change in January 2020, and from a single surgeon.

The high-risk criteria often cited from Inabathula and Kheir are BMI ≥35, diabetes, Staphylococcus aureus nasal colonization, CKD, smoking, or autoimmune disease; the regimen was cefadroxil 500 mg bid, switched to trimethoprim–sulfamethoxazole for MRSA colonization[16,17]. These six items are not a complete list, and adjudication differed between the two periods: Kheir listed additional high-risk circumstances, and the inclusion rules for nasal colonization also changed over time[17]. Flynn's high-risk subgroup required at least two risk factors[19]. What the studies call "high risk" is not the same population.

Studies Showing No Benefit

Carender, Flynn, and Raju are likewise single-center data. The study periods of Carender and Flynn each spanned a protocol change; whether antibiotics were given may have changed with era or depended on clinical indication, so between-group differences likewise cannot be attributed entirely to the drug.

  • Carender 2021[18]: 650 patients with BMI ≥40 (205 THA, 445 TKA), 177 (27%) received extended oral therapy for 7–14 days; 90-day PJI 1.7% versus 0.6% (P=0.35). Wound-complication figures are internally inconsistent in the original (Abstract 11% versus 8%, P=0.41; Table 2, 17 cases 9.6% versus 35 cases 7.4%, P=0.36); the PJI figures are consistent.
  • Flynn 2024[19]: 4,576 cases (treated 1,769, untreated 2,807); 90-day PJI 1% versus 0.8% (P=0.6), 1-year both 1% (P=0.9); high-risk subgroup (254 versus 396) 0.8% versus 2.3% (P=0.2). The study also reports C. difficile infection within 90 days of 1/1,769 versus 2/2,807 (0.1% versus 0.1%, P>0.9); with so few events this can show only that no increase was observed, not that safety is equivalent.
  • Fuqua 2025 (hip) [20]: national database primary THA 2009–2022, 4,153 who received 7–14 days of oral therapy and were propensity-score matched; HRs for PJI-free survival at 90 days, 1 year, and 2 years were all nonsignificant.
  • Fuqua 2025 (knee) [40]: same design for primary TKA, 5,701 treated patients matched; no reduction in PJI risk at 90 days, 1 year, or 2 years.
  • Raju 2026[21]: single-center 2015–2025, after 1:4 matching 2,935 cases (treated 587, untreated 2,348); 90-day PJI 11/587 (1.9%) versus 30/2,348 (1.3%), P=0.366; 1-year 16/587 (2.7%) versus 44/2,348 (1.9%), P=0.254; in multivariate Cox regression, extended oral therapy was not an independent predictor.
  • Haziza 2026[22]: national database primary TKA, 34,584 propensity-score-matched patients; the oral group had higher 90-day PJI (1.99% versus 0.72%, P<0.0003) and higher revision (0.7% versus 0.44%, P=0.0039), with no difference in superficial infection; low-risk subgroup PJI 1.52% versus 0.50%. Exposure was identified by a postoperative day-1 oral antibiotic prescription or administration record, without verification of actual duration or adherence, and the reason for the prescription was not recorded. With this unmeasured indication confounding, one can conclude neither that "oral therapy is harmful" nor that "oral therapy at least confers no benefit"; confounding may equally mask a true protective effect. The study can show only that no protective association was observed; the net causal effect remains undetermined.

Dasari 2024 Meta-analysis

Dasari 2024[23] pooled 18 studies and 19,153 cases, comparing prophylaxis beyond 24 hours postoperatively (mean 6.6 days, range >1 to 14 days) with a standard regimen of ≤24 hours. Benefit was shown for all TJA, primary TJA, primary THA, aseptic revision TJA, and aseptic revision TKA; primary TKA (P=0.17) and aseptic revision THA (P=0.36) showed no significant difference. This analysis has three specific limitations: exposure was not limited to oral agents; the control arm was a ≤24-hour regimen whose duration was not uniform; and Inabathula and Kheir were both included and weighted separately, so 19,153 is not a deduplicated independent patient count.

2025 International Consensus

The 2025 International Consensus Meeting (ICM) Question 4[24]: "Current evidence is insufficient to recommend routine use of extended oral antibiotic prophylaxis in high-risk patients undergoing primary total joint arthroplasty," with a limited recommendation strength and 94.8% delegate agreement. "Not recommended for routine use" is not a prohibition and does not mean proven ineffectiveness.

Reasons It Cannot Be Applied

The infection pattern differs. PJI is prosthetic biofilm infection, driven by host factors, with a substantial fraction late hematogenous; ACLR infections are overwhelmingly early postoperative, and all 22 of Offerhaus's cases occurred within 5 weeks[42]. The two hematogenous infections described in Section 1 occurred at 4 months[46] and 26 days[47] postoperatively. They are single cases and cannot be used to evaluate the efficacy of 7-day oral prophylaxis; in temporal terms, 7 days of oral therapy after discharge also cannot cover an event at 4 months. ACLR likewise implants an avascular graft (which behaves like a foreign body) and fixation devices[2]; one cannot assert that biofilm or host-factor issues are entirely absent in ACLR.

The population differs. "High risk" in the studies above means obesity, diabetes, and CKD[16,17]; ACLR patients are predominantly young and athletic. Individuals with similar high-risk features may also exist in ACLR, and identifying them remains meaningful.

Baseline infection rates are low, so obtainable absolute benefit is extremely small. Crude infection rates in soaked groups across studies were: Figueroa forest plot 1/7,894 (0.013%)[8], Carrozzo 1/2,072 (0.05%)[32], Kottek 1/1,329 (0.075%)[10], Offerhaus 0/853[42], Naendrup 0/2,976[7]. The calculation below assumes a baseline of 0.1%. That figure is higher than the point estimates above, but sampling uncertainty is large in zero-event and few-event studies, and populations and observation windows differ across studies, so one cannot assert that 0.1% is necessarily above the true risk. Under this assumption, even if extended oral therapy completely eliminated infection, the theoretical lower bound of the NNT would be 1,000; with a 50% relative reduction the NNT would be 2,000; the smaller the effect, the larger the NNT, and as efficacy approaches zero the NNT has no upper bound. This is arithmetic under a stated assumption, not a measured effect; no ACLR data have measured the actual efficacy of extended oral therapy in this population.

The harm side likewise has no measured ACLR values. Branch-Elliman's data show that extending duration itself carries duration-related harm[38]; Mirahmadi observed more complications in the extended group[14]; Flynn observed no increase in C. difficile infection[19]. These data come from different populations, exposures, and observation windows and cannot be placed on the same scale as the NNT above to calculate net benefit; the number needed to harm for extended oral therapy in ACLR has also never been measured. On one side is a benefit whose NNT is no lower than 1,000 even if completely effective; on the other is unquantified but real harm. In the absence of direct evidence, not using extended oral antibiotics is the more prudent choice; this is a trade-off under uncertainty, not a calculated net-benefit conclusion.

Within arthroplasty, knee results are the most inconsistent. Dasari's primary TKA subgroup showed no significant benefit (P=0.17)[23], and the TKA database analyses of Haziza and Fuqua likewise showed no benefit[22,40]. A nonsignificant subgroup shows only that the estimate in that subgroup was not significant; it cannot be used to infer that the true effects in knee and hip differ. What it reflects is that benefit of extended oral therapy in knee arthroplasty has not appeared consistently across studies.

The author does not recommend postoperative oral antibiotics as a research direction for ACLR, because obtainable absolute benefit is too small. Still under the assumed baseline of 0.1%, to prove a further halving of the infection rate (0.1%→0.05%), with two-sided α=0.05, 80% power, and 1:1 allocation under the normal approximation, about 47,057 per group would be needed (about 62,996 at 90% power), before accounting for loss to follow-up, continuity correction, and center clustering. A conventional randomized trial is therefore difficult to conduct; multicenter registries, cluster designs, or studies with graft contamination as a surrogate end point may still provide higher-level evidence. This is the author's value judgment about research priority, not an established fact.

6. Evidence Summary

Measure Practice Essential conditions Nature of evidence
Preoperative cefazolin Routine use 2 g; 3 g if body weight ≥120 kg (ESSKA: >120 kg); begin infusion within 60 min before incision; complete before tourniquet inflation Consistent guideline recommendation[1,2,3,25]; no direct comparative evidence in ACLR
Intraoperative redosing Redose when conditions are met Cefazolin about 4 h from start of first infusion (ASHP, normal renal function); Chinese: operative time >3 h or >2× half-life; redose if adult blood loss >1,500 mL, except vancomycin and aminoglycosides Guideline recommendation[3,5,25]
Reported β-lactam allergy First verify culprit drug and reaction type Type I reaction to penicillin, amoxicillin, amoxicillin–clavulanate, ampicillin, or cephalexin: still use cefazolin. Type I reaction to cefazolin itself, or confirmed severe delayed or other Type II–IV β-lactam reaction: switch to vancomycin. Severe immediate reaction to another cephalosporin: evaluate per practice parameter first; do not give directly Institutional guidelines and practice parameters[5,28]; association data from arthroplasty[4]
Vancomycin Dose by weight Dose by the weight bands in Section 2 with adequate infusion time; 2 g requires 120 min infusion; start 60–120 min before incision; complete before tourniquet inflation Guideline recommendation[1,2,3,5]
Known MRSA colonization Optional; state which document is followed ESSKA/EBJIS: may switch to vancomycin; Stanford: consider adding vancomycin or clindamycin; Chinese: may choose (nor)vancomycin when MRSA colonization is possible or institutional MRSA rate is high Guideline recommendation[1,2,5,25]
Preoperative intra-articular corticosteroid injection Delay surgery when possible Injection within 8 weeks before surgery, 90-day infection OR 2.1; no data on how long to delay; the 6-month suggestion is an inference from total knee arthroplasty data Database cohort and case–control[6,43]; guideline[1]
Graft vancomycin soaking Routine use 5 mg/mL, 10–15 min ("needs 20 min" is a second-hand paraphrase of a porcine tendon experiment); add-on to systemic prophylaxis, not a substitute; data dominated by hamstring tendon Retrospective cohorts and pooled analyses, consistent direction of effect (Figueroa after deduplication RR 0.067[8]); two small RCTs underpowered[33,34]
Postoperative antibiotics Do not use Neither intravenous nor oral; Chinese standards limit total prophylaxis for clean surgery to ≤24 h, without stating the start of the count Guidelines advise against[1,2,25]; no benefit shown in ACLR[14]; arthroplasty duration comparisons underpowered[15]; harm signals from other surgical populations[38]
Extended oral regimens from arthroplasty Do not apply to ACLR Infection pattern, population, and baseline risk differ Arthroplasty results inconsistent; the three primary cohorts reporting benefit listed in Section 5 all used within-institution historical controls[16,17,45]; ICM 2025 judged evidence insufficient to recommend routine use[24]

Data Scope

The evidence search date was 8 September 2026. The scope is antibiotic use in the perioperative period of primary anterior cruciate ligament reconstruction, including systemic prophylaxis and local graft treatment; it does not include the diagnosis or treatment of infection.

References

  1. [1]

    Pérez-Prieto D, Totlis T, Madjarevic T, Becker R, Ravn C, Monllau JC, Renz N. ESSKA and EBJIS recommendations for the management of infections after anterior cruciate ligament reconstruction (ACL-R): prevention, surgical treatment and rehabilitation. Knee Surg Sports Traumatol Arthrosc. 2023;31(10):4204-4212. DOI: 10.1007/s00167-023-07463-3. PMID: 37243789.

  2. [2]

    Renz N, Madjarevic T, Ferrari M, Becker R, Ravn C, Vogely C, Pérez-Prieto D. Recommendations on diagnosis and antimicrobial treatment of infections after anterior cruciate ligament reconstruction (ACL-R) endorsed by ESSKA and EBJIS. J Infect. 2023;86(6):543-551. DOI: 10.1016/j.jinf.2023.03.021. PMID: 37019288.

  3. [3]

    Bratzler DW, Dellinger EP, Olsen KM, et al. Clinical practice guidelines for antimicrobial prophylaxis in surgery. Am J Health Syst Pharm. 2013;70(3):195-283. DOI: 10.2146/ajhp120568. PMID: 23327981.

  4. [4]

    Wyles CC, Hevesi M, Osmon DR, Park MA, Habermann EB, Lewallen DG, Berry DJ, Sierra RJ. 2019 John Charnley Award: Increased risk of prosthetic joint infection following primary total knee and hip arthroplasty with the use of alternative antibiotics to cefazolin: the value of allergy testing for antibiotic prophylaxis. Bone Joint J. 2019;101-B(6 Supple B):9-15. DOI: 10.1302/0301-620X.101B6.BJJ-2018-1407.R1. PMID: 31146571.

  5. [5]

    Stanford Health Care. Surgical Antimicrobial Prophylaxis Guideline (institutional guideline, revised 2025-07-02; hospital-wide susceptibility data from 2024). Full text

  6. [6]

    Wainwright JD, Gay SS, Nguyen A, Weiss WM, Wenke JC. Corticosteroid injection up to 8 weeks before anterior cruciate ligament reconstruction doubles the incidence of postoperative infection. Arthroscopy. 2025;41:4021-4025.e1. DOI: 10.1016/j.arthro.2025.02.027. PMID: 40056948.

  7. [7]

    Naendrup JH, Marche B, de Sa D, Koenen P, Otchwemah R, Wafaisade A, Pfeiffer TR. Vancomycin-soaking of the graft reduces the incidence of septic arthritis following ACL reconstruction: results of a systematic review and meta-analysis. Knee Surg Sports Traumatol Arthrosc. 2020;28(4):1005-1013. DOI: 10.1007/s00167-019-05353-1. PMID: 30656372.

  8. [8]

    Figueroa F, Figueroa D, Calvo R, Vaisman A, Nuñez M, Putnis S. Vancomycin presoaking of the graft appears to prevent infection after anterior cruciate ligament reconstruction: a systematic review and meta-analysis. HSS J. 2022;18(1):138-144. DOI: 10.1177/15563316211011682. PMID: 35087344.

  9. [9]

    Hu M, Zhang Y, Shang G, Guo J, Xu H, Ma X, Yang X, Xiang S. Vancomycin presoak reduces infection in anterior cruciate ligament reconstruction: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2023;24(1):267. DOI: 10.1186/s12891-023-06331-y. PMID: 37020216. (Table 2 and the forest plot record Offerhaus soaked-group infections as 8/853, which does not match the original 0/853)

  10. [10]

    Kottek T, Kunovac B, Vrgoč G, Ivković A, Janković S. Vancomycin-soaking of the graft significantly reduces the incidence of septic arthritis following anterior cruciate ligament reconstruction: comparison of cohorts from the single-centre database. Int Orthop. 2025;49(9):2097-2102. DOI: 10.1007/s00264-025-06594-1. PMID: 40586927.

  11. [11]

    Bohu Y, Klouche S, Sezer HB, Herman S, Grimaud O, Gerometta A, Meyer A, Lefevre N. Vancomycin-soaked autografts during ACL reconstruction reduce the risk of post-operative infection without affecting return to sport or knee function. Knee Surg Sports Traumatol Arthrosc. 2020;28(8):2578-2585. DOI: 10.1007/s00167-020-05879-9. PMID: 32025764. (Correction: same journal 2020;28(8):2586. DOI: 10.1007/s00167-020-05923-8. PMID: 32140740)

  12. [12]

    Bansal A, Lamplot JD, VandenBerg J, Brophy RH. Meta-analysis of the risk of infections after anterior cruciate ligament reconstruction by graft type. Am J Sports Med. 2018;46(6):1500-1508. DOI: 10.1177/0363546517714450. PMID: 28737955.

  13. [13]

    Vertullo CJ, Quick M, Jones A, Grayson JE. A surgical technique using presoaked vancomycin hamstring grafts to decrease the risk of infection after anterior cruciate ligament reconstruction. Arthroscopy. 2012;28(3):337-342. DOI: 10.1016/j.arthro.2011.08.301. PMID: 22112612.

  14. [14]

    Mirahmadi A, Hosseini-Monfared P, Farrokhi M, Minaie R, Bateni A, Kazemi SM. Extended oral antibiotics fail to reduce surgical site infection in orthopedic surgeries: a comparative study. PLoS One. 2025;20(9):e0330685. DOI: 10.1371/journal.pone.0330685. PMID: 40911548. (Procedure-stratified data in Supplementary Table S1)

  15. [15]

    Siddiqi A, Forte SA, Docter S, Bryant D, Sheth NP, Chen AF. Perioperative antibiotic prophylaxis in total joint arthroplasty: a systematic review and meta-analysis. J Bone Joint Surg Am. 2019;101(9):828-842. DOI: 10.2106/JBJS.18.00990. PMID: 31045673.

  16. [16]

    Inabathula A, Dilley JE, Ziemba-Davis M, Warth LC, Azzam KA, Ireland PH, Meneghini RM. Extended oral antibiotic prophylaxis in high-risk patients substantially reduces primary total hip and knee arthroplasty 90-day infection rate. J Bone Joint Surg Am. 2018;100(24):2103-2109. DOI: 10.2106/JBJS.17.01485. PMID: 30562290.

  17. [17]

    Kheir MM, Dilley JE, Ziemba-Davis M, Meneghini RM. The AAHKS Clinical Research Award: extended oral antibiotics prevent periprosthetic joint infection in high-risk cases: 3855 patients with 1-year follow-up. J Arthroplasty. 2021;36(7 Suppl):S18-S25. DOI: 10.1016/j.arth.2021.01.051. PMID: 33589279.

  18. [18]

    Carender CN, DeMik DE, Glass NA, Noiseux NO, Brown TS, Bedard NA. Do extended oral postoperative antibiotics prevent early periprosthetic joint infection in morbidly obese patients undergoing primary total joint arthroplasty? J Arthroplasty. 2021;36(8):2716-2721. DOI: 10.1016/j.arth.2021.03.018. PMID: 33781639.

  19. [19]

    Flynn JB, Yokhana SS, Wilson JM, Schultz JD, Hymel AM, Martin JR. Not so fast: extended oral antibiotic prophylaxis does not reduce 90-day infection rate following joint arthroplasty. J Arthroplasty. 2024;39(9 Suppl 2):S122-S128. DOI: 10.1016/j.arth.2024.04.064. PMID: 38685337.

  20. [20]

    Fuqua AA, Worden JA, Ayeni AM, Bundschuh KE, Wilson JM, Premkumar A. Extended oral antibiotic prophylaxis and periprosthetic joint infection-free survivorship after primary total hip arthroplasty. Arthroplast Today. 2025;33:101694. DOI: 10.1016/j.artd.2025.101694. PMID: 40585394.

  21. [21]

    Raju A, Jahagirdar OB, Eslam Pour A, Golden M, Molloy I, Grauer JN, Wiznia DH. No improvement in infection or complication rate with extended oral antibiotic prophylaxis after primary total joint arthroplasty. J Arthroplasty. 2026 (online ahead of print). DOI: 10.1016/j.arth.2026.05.044. PMID: 42173207.

  22. [22]

    Haziza S, Greenberg M, Spagnuola J, Duchnycz R, Harbaugh T, Abdo ZE. Association of extended oral antibiotic prophylaxis with 90-day complication rates after primary total knee arthroplasty: a propensity-matched retrospective cohort study. J Arthroplasty. 2026 (online ahead of print). DOI: 10.1016/j.arth.2026.08.020. PMID: 42600869.

  23. [23]

    Dasari SP, Kanumuri SD, Yang J, Manner PA, Fernando ND, Hernandez NM. Extended prophylactic antibiotics for primary and aseptic revision total joint arthroplasty: a meta-analysis. J Arthroplasty. 2024;39(9S2):S476-S487. DOI: 10.1016/j.arth.2024.01.014. PMID: 38237874.

  24. [24]

    Cooper AM, Munhoz Lima AL, Luo TD, Arshi A, Spangehl MJ, et al. 2025 ICM: Antibiotic prophylaxis in primary joint arthroplasty. J Arthroplasty. 2025;41:S362-S375. DOI: 10.1016/j.arth.2025.08.061. PMID: 41161514.

  25. [25]

    General Office of the National Health and Family Planning Commission, Office of the State Administration of Traditional Chinese Medicine, Bureau of Pharmaceuticals and Equipment of the Health Department of the General Logistics Department. Guiding Principles for Clinical Use of Antibacterials (2015 edition). Document Guo Wei Ban Yi Fa [2015] No. 43.

  26. [26]

    AlBuhairan B, Hind D, Hutchinson A. Antibiotic prophylaxis for wound infections in total joint arthroplasty: a systematic review. J Bone Joint Surg Br. 2008;90(7):915-919. DOI: 10.1302/0301-620X.90B7.20498. PMID: 18591602.

  27. [27]

    ClinicalTrials.gov. Parenteral versus combined parenteral with vancomycin-soaked graft in ACL reconstruction. NCT06166381 (Duhok University; planned enrollment 288; started 2024-01-25; snapshot retrieved 2026-09-08, status "Recruiting," last submitted update 2024-12-26, posted update 2024-12-30).

  28. [28]

    Khan DA, Banerji A, Blumenthal KG, et al. Drug allergy: a 2022 practice parameter update. J Allergy Clin Immunol. 2022;150(6):1333-1393. DOI: 10.1016/j.jaci.2022.08.028. PMID: 36122788.

  29. [29]

    Pérez-Prieto D, Portillo ME, Torres-Claramunt R, Pelfort X, Hinarejos P, Monllau JC. Contamination occurs during ACL graft harvesting and manipulation, but it can be easily eradicated. Knee Surg Sports Traumatol Arthrosc. 2018;26(2):558-562. DOI: 10.1007/s00167-017-4735-z. PMID: 28988312.

  30. [30]

    Thiagarajah SV, Henry J, Sivaprakasam V, Sutton PM, Charalambous CP. Evaluation of autograft contamination in anterior cruciate ligament reconstruction and its clinical impact: a systematic review and meta-analysis. JBJS Rev. 2025;13(8):e25.00108. DOI: 10.2106/JBJS.RVW.25.00108. PMID: 40845180. (Erratum: same journal 2025;13(9):eER.25.00108. DOI: 10.2106/JBJS.RVW.ER.25.00108. PMID: 40991740; corrects author affiliations only, no data change)

  31. [31]

    Scheu M, Bosselin R, Araos R, Ugalde J, Tuca MJ, Espinoza G. Bacterial analysis of hamstring autograft with vancomycin presoaking in primary anterior cruciate ligament reconstruction shows high incidence of colonization without early clinical infection. Arthrosc Sports Med Rehabil. 2025;7:101231. DOI: 10.1016/j.asmr.2025.101231. PMID: 41583803.

  32. [32]

    Carrozzo A, Saithna A, Ferreira A, Guy S, Chadli L, Monaco E, Pérez-Prieto D, Bohu Y, Vieira TD, Sonnery-Cottet B. Presoaking ACL grafts in vancomycin decreases the frequency of postoperative septic arthritis: a cohort study of 29,659 patients, systematic review, and meta-analysis from the SANTI Study Group. Orthop J Sports Med. 2022;10(2):23259671211073928. DOI: 10.1177/23259671211073928. PMID: 35155711.

  33. [33]

    Mousavi SH, Radi M, Kazemi D, Izadpanah F, Shahrouzian A. Evaluation the results of anterior cruciate ligament reconstruction with or without using antibiotic solution during graft preparation. Eur J Orthop Surg Traumatol. 2023;33(7):3033-3036. DOI: 10.1007/s00590-023-03522-0. PMID: 36988698.

  34. [34]

    Gourab K, Rani S, Gupta GK, Guria A, Kullu S, Pandey N, Ain Z. Assessing the outcomes of anterior cruciate ligament reconstruction with or without the use of antibiotic solution in graft preparation: a randomized control trial. Ann Afr Med. 2026 (online ahead of print; volume, issue, and pages pending). DOI: 10.4103/aam.aam_270_25. PMID: 41769928.

  35. [35]

    Pfeiffer TR, Althoff A, Krombholz S, Dautert M, Naendrup JH, Guenther D, Bouillon B, Thevis M. Vancomycin concentrations in synovial fluid do not reach chondrotoxic thresholds after anterior cruciate ligament reconstruction with vancomycin-soaked autologous soft tissue grafts: an in vivo prospective observational study in humans. Am J Sports Med. 2023;51(7):1715-1720. DOI: 10.1177/03635465231169040. PMID: 37161935.

  36. [36]

    Parinyakhup W, Chavalparit P, Saeloh Sotthibandhu D, Boonriong T, Maliwankul K, Klabklay P, Chuaychoosakoon C. Optimizing vancomycin soaking protocols for anterior cruciate ligament reconstruction. Antibiotics (Basel). 2025;14(4):332. DOI: 10.3390/antibiotics14040332. PMID: 40298474. (Original gauze concentrations conflict between Abstract/Results 2.5/5/10 and Methods 5/10/20 mg/mL)

  37. [37]

    Weninger P, Weninger V, Steffel C, Herbst A. Vancomycin-soaked quadriceps tendon autografts but not hamstring tendon autografts are associated with early graft failure (graft resorption) after primary anterior cruciate ligament reconstruction: a single-center experience. Indian J Orthop. 2026;60:672-677. DOI: 10.1007/s43465-025-01632-x. PMID: 41908996.

  38. [38]

    Branch-Elliman W, O'Brien W, Strymish J, Itani K, Wyatt C, Gupta K. Association of duration and type of surgical prophylaxis with antimicrobial-associated adverse events. JAMA Surg. 2019;154(7):590-598. DOI: 10.1001/jamasurg.2019.0569. PMID: 31017647.

  39. [39]

    Zhai Q, Li X. Letter to the editor: Vancomycin-soaking of the graft significantly reduces the incidence of septic arthritis following anterior cruciate ligament reconstruction: comparison of cohorts from the single-centre database. Int Orthop. 2025. DOI: 10.1007/s00264-025-06647-5. PMID: 40888911.

  40. [40]

    Fuqua AA, et al. Extended oral antibiotic prophylaxis and PJI-free survivorship after primary total knee arthroplasty. Knee. 2025;56:1-10. DOI: 10.1016/j.knee.2025.04.005. PMID: 40383007.

  41. [41]

    Carney J, Heckmann N, Mayer EN, Alluri RK, Vangsness CT Jr, Hatch GF 3rd, Weber AE. Should antibiotics be administered before arthroscopic knee surgery? A systematic review of the literature. World J Orthop. 2018;9(11):262-270. DOI: 10.5312/wjo.v9.i11.262. PMID: 30479973.

  42. [42]

    Offerhaus C, Balke M, Hente J, Gehling M, Blendl S, Höher J. Vancomycin pre-soaking of the graft reduces postoperative infection rate without increasing risk of graft failure and arthrofibrosis in ACL reconstruction. Knee Surg Sports Traumatol Arthrosc. 2019;27(9):3014-3021. DOI: 10.1007/s00167-018-5323-6. PMID: 30666370.

  43. [43]

    Armstrong RW, Bolding F, Joseph R. Septic arthritis following arthroscopy: clinical syndromes and analysis of risk factors. Arthroscopy. 1992;8(2):213-223. DOI: 10.1016/0749-8063(92)90039-E. PMID: 1637435. (Abstract writes 4,256 cases; Methods and Results write 4,264; the original is internally inconsistent)

  44. [44]

    Armstrong RW, Bolding F. Septic arthritis after arthroscopy: the contributing roles of intraarticular steroids and environmental factors. Am J Infect Control. 1994;22(1):16-18. DOI: 10.1016/0196-6553(94)90086-8. PMID: 8172371.

  45. [45]

    Bundschuh KE, Muffly BT, Ayeni AM, Heo KY, Khawaja SR, Tocio AJ, Karzon AL, Premkumar A, Guild GN 3rd. Should all patients receive extended oral antibiotic prophylaxis? Defining its role in patients undergoing primary and aseptic revision total joint arthroplasty. J Arthroplasty. 2024;39(9 Suppl 2):S117-S121.e4. DOI: 10.1016/j.arth.2024.01.012. PMID: 38218558.

  46. [46]

    Eranki V, Wallis A, Smith S. Haematogenous spread of Staphylococcus aureus from an iliacus abscess to an ACL reconstructed knee. Case Rep Orthop. 2013;2013:914329. DOI: 10.1155/2013/914329. PMID: 23762701.

  47. [47]

    Neri T, et al. Salmonella infection after anterior cruciate ligament reconstruction: a case report. World J Orthop. 2021;12(4):246-253. DOI: 10.5312/wjo.v12.i4.246. PMID: 33959488.