Surgical Antibiotic Prophylaxis: 6 Latest Clinical Insights
Where the compendium moved the goalposts on duration, colorectal preparation and vancomycin — and what the trials underneath it actually show.
One-Minute Takeaway
- Surgical antibiotic prophylaxis is anchored by the 2022 SHEA/IDSA/APIC compendium update, which grades the core recommendation as high-quality evidence and directs that prophylaxis stop at incisional closure.
- Across 52 randomised trials and 19,273 participants, continuing prophylaxis after closure did not reduce surgical site infection (RR 0.89, 95% CI 0.79–1.00).
- Oral plus intravenous antibiotics before elective colorectal resection reduced infection by 44% versus mechanical preparation alone (RR 0.56, 95% CI 0.42–0.74).
- A reported β-lactam allergy was associated with higher infection risk across 460,284 patients (RR 1.55, 95% CI 1.24–1.94), driven by the substitute agents it prompts.
- Agent choice in surgical antibiotic prophylaxis is now procedure-specific: piperacillin-tazobactam reduced infection versus cefoxitin after open pancreatoduodenectomy (19.8% vs 32.8%).
Why Surgical Antibiotic Prophylaxis Matters
Surgical site infection is the complication a trainee is most likely to influence and least likely to notice at the time. It follows 1–3% of inpatient operations: the National Healthcare Safety Network recorded 21,186 infections in 2021, across 2,759,027 procedures.
The consequences are out of proportion to that rate. Patients with an infection carry a two- to eleven-fold higher risk of death than those without one. Among infected patients who die, 77% of deaths are directly attributable to it.
Up to 60% are preventable using evidence-based guidelines. Surgical antibiotic prophylaxis has the shortest path from one decision to a measurable effect, and is the task most often delegated to the most junior person in the room.
Four surgical antibiotic prophylaxis decisions carry most of that effect: agent, dose, timing and stopping point. All four have moved since the 2013 clinical practice guideline that still governs agent selection — mostly not in the direction residents were taught.
What the Guideline Recommends
Two documents govern surgical antibiotic prophylaxis in North American practice, and they do different jobs. The 2013 ASHP/IDSA/SIS/SHEA clinical practice guideline remains the reference for agent and dose by procedure. The 2022 SHEA/IDSA/APIC compendium update, published in 2023, sets the strategy and grades it.
It classifies surgical antibiotic prophylaxis delivered to evidence-based standards as an essential practice at high quality of evidence. Essential is a specific category: every acute-care hospital should adopt it, not reach for it once the basics are covered.
- TimingBegin within 1 hour before incision; 2 hours for vancomycin and fluoroquinolones, which take longer to infuse. For caesarean delivery, give before skin incision rather than after cord clamping.
- DoseCefazolin 2 g up to 120 kg, 3 g above it, 30–40 mg/kg in children. Vancomycin 15 mg/kg. Gentamicin 5 mg/kg adult, 2.5 mg/kg paediatric.
- RedosingRedose at two half-lives from the preoperative dose — four hours for cefazolin — and after blood loss over 1,500 mL.
- DurationDiscontinue after incisional closure in the operating room, even when drains are inserted.
- ColorectalCombine parenteral and oral prophylaxis before elective colorectal surgery, at high quality of evidence. Extend the logic wherever colonic entry is possible.
- AllergyTake a thorough allergy history. Self-reported β-lactam allergy has been linked to higher infection risk through the alternatives it prompts; many such patients can safely receive a β-lactam.
- VancomycinNot routine. Consider it in patients known to be colonised with methicillin-resistant Staphylococcus aureus, particularly where prosthetic material is implanted.
Three entries changed position rather than wording, and those are where surgical antibiotic prophylaxis actually moved. Where earlier documents tolerated continuation up to 24 hours, the update directs discontinuation at incisional closure — drains or no drains. Combined parenteral and oral prophylaxis was a 2014 sub-bullet elevated to a standalone essential practice.
Vancomycin moved the other way. The compendium expanded its discussion against routine vancomycin surgical antibiotic prophylaxis, narrowing it toward documented colonisation rather than healthcare-associated risk factors.
Evidence Reviewed
The surgical antibiotic prophylaxis evidence base is mixed: a few adequately powered trials, reviews pooling them, and large cohorts filling the gaps. The table separates them by design, because design decides how far a result travels. Sample sizes are analysed populations.
| Study / source | Design | N | Population | Comparison | Key result |
|---|---|---|---|---|---|
| Willis 2023 (Cochrane) | SR/MA, 21 RCTs | 5,264 | Elective colorectal resection, all on IV prophylaxis | Mechanical prep + oral antibiotics vs mechanical prep alone | SSI RR 0.56 (95% CI 0.42–0.74); leak RR 0.60 (0.36–0.99); moderate certainty |
| de Jonge 2020 (Lancet Infect Dis) | SR/MA, 52 RCTs | 19,273 | Mixed specialties, prophylaxis indicated | Postoperative continuation vs stopping at closure | SSI RR 0.89 (95% CI 0.79–1.00); no benefit where best practice was followed |
| Hanai 2025 (Infect Control Hosp Epidemiol) | SR/MA, 25 cohorts | 460,284 | Adults, mixed procedures | Reported β-lactam allergy vs none | SSI RR 1.55 (95% CI 1.24–1.94); β-lactam vs alternative RR 0.63 (0.42–0.94) |
| Huisman 2026 (BJS Open) | SR/MA, 3 RCTs + 30 cohorts | 99,211 | Surgical patients with obesity | Weight-based vs standard cefazolin dose | RCTs: RD +2.02% (95% CI −3.15 to 7.19); cohorts: RD −1.93% (−2.84 to −1.02); very low certainty |
| Hanai 2025 (Ann Gastroenterol Surg) | SR/MA, 7 cohorts | 4,671 | Operations lasting 3 hours or more | Intraoperative redosing vs no redosing | SSI OR 0.65 (95% CI 0.45–0.94); funnel plot asymmetry |
| de Jonge 2017 (Medicine) | SR/MA, 14 studies | 54,552 | Mixed specialties | Timing intervals before incision | No difference between 120–60 min and 60–0 min pre-incision |
| D’Angelica 2023 (JAMA) | Pragmatic phase 3 RCT, open-label | 778 | Open pancreatoduodenectomy, 26 hospitals | Piperacillin-tazobactam vs cefoxitin | 30-day SSI 19.8% vs 32.8%; difference −13.0% (95% CI −19.1 to −6.9) |
| MOBILE2 / Koskenvuo 2024 (JAMA Surg) | RCT, double-blind, placebo-controlled | 565 | Elective rectal resection, primary anastomosis | Mechanical prep + oral antibiotics vs prep + placebo | SSI 8.3% vs 16.7%, OR 0.45 (95% CI 0.27–0.77); dehiscence OR 0.39 (0.21–0.72) |
| Weber 2017 (Lancet Infect Dis) | Phase 3 RCT, superiority | 5,175 analysed | General surgery inpatients, Switzerland | Cefuroxime early (42 min) vs late (16 min) pre-incision | SSI OR 0.93 (95% CI 0.72–1.21); no advantage to early dosing |
| Sommerstein 2023 (JAMA Netw Open) | Cohort, 158 hospitals | 222,439 | 11 major procedures, cefuroxime | 0–30 min vs 61–120 min before incision | SSI aOR 0.85 (95% CI 0.78–0.93); lowest risk at 10–25 min |
| Sommerstein 2021 (JAMA Netw Open) | Cohort, 142 hospitals | 37,640 | Patients ≥80 kg, 9 major procedures | Cefuroxime 3.0 g vs 1.5 g | SSI aOR 0.89 (95% CI 0.78–1.02); not significant |
| Blumenthal 2018 (Clin Infect Dis) | Cohort, single centre | 8,385 | Arthroplasty, hysterectomy, colon, CABG | Reported penicillin allergy vs none | SSI aOR 1.51 (95% CI 1.02–2.22), via non-β-lactam agents |
SR/MA = systematic review and meta-analysis. SSI = surgical site infection. RD = risk difference. aOR = adjusted odds ratio.
Key Changes in Surgical Antibiotic Prophylaxis and Their Basis
1. Stop at skin closure, not at midnight
The most consequential change in surgical antibiotic prophylaxis is subtractive. The pooled analysis behind it included 52 of 83 randomised trials and 19,273 participants, across gastrointestinal, cardiac, thoracic, gynaecological, orthopaedic and maxillofacial surgery.
Continuing prophylaxis after the operation, rather than stopping at the end of it, produced a relative risk of infection of 0.89 (95% CI 0.79–1.00), with low heterogeneity. The interval touches unity. That is not a signal on which to keep dosing.
The prespecified subgroup is the part worth carrying: where trials met best-practice standards for surgical antibiotic prophylaxis, continuation yielded no benefit. It compensates for a first dose given badly, not one given correctly. The harms — resistance, Clostridioides difficile, acute kidney injury — sit on the other side of the ledger.
2. Oral antibiotics before colorectal resection are now essential, not optional
The Cochrane review on this surgical antibiotic prophylaxis question analysed 21 randomised trials and 5,264 participants, all on intravenous prophylaxis. Seventeen compared mechanical preparation plus oral antibiotics against mechanical preparation alone.
Adding oral antibiotics reduced surgical site infection by 44% (RR 0.56, 95% CI 0.42–0.74; 16 trials) and anastomotic leakage by 40% (RR 0.60, 95% CI 0.36–0.99). Both were graded moderate certainty. Mortality, ileus and length of stay were unchanged.
MOBILE2 then tested the same question in rectal resection, double-blind and placebo-controlled. Among 565 patients, infection occurred in 8.3% of the oral-antibiotic group versus 16.7% of placebo (OR 0.45, 95% CI 0.27–0.77), with dehiscence in parallel (OR 0.39, 95% CI 0.21–0.72). What this does not settle is whether the mechanical component adds anything: three trials found no difference against oral antibiotics alone, at very low certainty.
3. The allergy label is a modifiable risk factor
Roughly one surgical patient in ten carries a penicillin allergy label, and most are not allergic. The largest synthesis pooled 25 observational studies and 460,284 patients. A reported β-lactam allergy was associated with 55% higher risk of surgical site infection (RR 1.55, 95% CI 1.24–1.94).
The mechanism is not immunological. Within the same synthesis, patients who received a β-lactam had lower infection risk than those given a non-β-lactam alternative (RR 0.63, 95% CI 0.42–0.94).
The cohort that first framed the problem agrees. Across 8,385 patients undergoing 9,004 procedures, a reported penicillin allergy was associated with 51% higher adjusted odds of infection (aOR 1.51, 95% CI 1.02–2.22). Its mediation analysis put the direct effect of the label at aOR 1.00 (95% CI 0.36–2.78): null.
Both estimates are observational, vulnerable to confounding. The compendium’s response is a clinical instruction, not a research one: take the history, because many patients labelled allergic can receive cefazolin safely. The allergy field is part of surgical antibiotic prophylaxis now, not a footnote.
4. Timing narrowed — but the trial and the registry disagree
This is the cleanest live disagreement in surgical antibiotic prophylaxis, and worth knowing both sides. The meta-analysis that informed the WHO guideline pooled 14 studies and 54,552 patients, finding no difference between dosing 120–60 minutes before incision and 60–0 minutes.
The only adequately powered trial pushed inside that window, assigning 5,580 general surgery patients to cefuroxime early or late — medians of 42 and 16 minutes before incision. Among 5,175 analysed, early administration did not reduce infection (OR 0.93, 95% CI 0.72–1.21).
The Swiss national surveillance cohort disagrees, at scale. Across 222,439 patients receiving cefuroxime for 11 major procedures, dosing 0–30 minutes before incision was associated with lower infection odds than 61–120 minutes (aOR 0.85, 95% CI 0.78–0.93). The lowest risk sat in the 10–25 minute window.
These are not the same claim, which is why both can hold. The trial randomised within roughly 75 minutes and found no advantage to being early; the cohort used a 61–120 minute reference and found an advantage to being late. The surgical antibiotic prophylaxis rule survives either way: give it inside the hour, and being close to incision is not a fault to correct.
5. Agent selection has become procedure-specific
Cefazolin remains the surgical antibiotic prophylaxis default for most operations. What changed is that one procedure now has randomised evidence against the standard-care cephalosporin — not a claim available a decade ago.
A pragmatic phase 3 trial randomised 778 adults undergoing open pancreatoduodenectomy at 26 North American hospitals to piperacillin-tazobactam or cefoxitin, stopping early at a predefined interim analysis. Thirty-day surgical site infection occurred in 19.8% versus 32.8% (difference −13.0%, 95% CI −19.1 to −6.9).
The secondary outcomes moved with the primary. Clinically relevant pancreatic fistula fell from 19.0% to 12.7% (difference −6.3%, 95% CI −11.4 to −1.2), and sepsis from 7.5% to 4.2%. The result should not be generalised: it applies where biliary instrumentation changes the flora, and is not an argument for broadening surgical antibiotic prophylaxis elsewhere.
6. Weight-based dosing rests on weaker ground than the rule implies
The surgical antibiotic prophylaxis instruction is simple: cefazolin 2 g up to 120 kg, 3 g above it. The evidence behind the threshold is not, and the number is softer than it sounds.
A 2026 meta-analysis assembled 33 studies and 99,211 patients with obesity. Its three randomised trials contributed one infection among 103 patients between them, and showed no reduction with weight-based dosing (risk difference +2.02%, 95% CI −3.15 to 7.19). Six observational studies, covering 45,554 patients, pointed the other way (risk difference −1.93%, 95% CI −2.84 to −1.02) at very low GRADE certainty.
A Swiss study of 37,640 patients over 80 kg found no overall link between double-dose cefuroxime and infection (aOR 0.89, 95% CI 0.78–1.02). But among the 2,594 patients above 120 kg it favoured the higher dose, short of significance (aOR 0.65, 95% CI 0.42–1.04).
A 2025 pharmacokinetic consensus review is cited as finding no basis to exceed 2 g of cefazolin for operations under four hours. The compendium concedes the conflict, noting that a single 3 g dose carries few adverse events. Follow the guideline; hold the certainty loosely — it is the softest number in surgical antibiotic prophylaxis.
Clinical Pearl
Redosing is the surgical antibiotic prophylaxis decision most often missed and the easiest to defend. Across seven cohorts and 4,671 patients undergoing operations of three hours or more, it was associated with lower infection odds (OR 0.65, 95% CI 0.45–0.94). For cefazolin the interval is four hours from the preoperative dose — or immediately, after blood loss above 1,500 mL; set the timer at induction.
Quality and Consistency of the Evidence
Strength varies sharply across the six surgical antibiotic prophylaxis changes above, rated by outcome rather than recommendation:
- Stopping surgical antibiotic prophylaxis at incisional closure High
52 randomised trials, low heterogeneity, consistent across six specialties, concordant with the compendium’s high grade. - Oral plus intravenous antibiotics before elective colorectal resection High
A moderate-certainty Cochrane estimate reinforced by a later double-blind trial that reproduced both the size and direction of the effect. - Agent selection in open pancreatoduodenectomy Moderate
One large pragmatic trial: open-label, stopped early, one procedure. Strong internally, narrow externally. - Timing within 60 minutes of incision Moderate
Concordant on the boundary, discordant inside it. One trial found no difference within the window; a large cohort favoured the final half hour. - β-lactam allergy label as a risk factor Moderate
25 observational studies, directionally consistent, with a coherent mechanism. Confounding by indication cannot be excluded; no randomised data exist. - Weight-based cefazolin dosing above 120 kg Low
Randomised data are nearly absent. Observational and pharmacokinetic evidence conflict; GRADE certainty was rated very low. - Intraoperative redosing Low
Observational only, with funnel plot asymmetry suggesting publication bias. A randomised trial may no longer be ethical, which locks the evidence here.
One pattern is worth naming. Where surgical antibiotic prophylaxis has randomised evidence, the answer has usually been less: shorter duration, no advantage to earlier dosing, no benefit from a larger dose. Where it is observational, the answer has more often been more — and trials stopped early tend to overstate effect size.
What the Evidence Does Not Show
- No randomised trial has tested the 120 kg cefazolin threshold against a clinical endpoint. The three available trials contributed one infection between them.
- No randomised trial has compared intraoperative redosing with none. That surgical antibiotic prophylaxis recommendation rests on seven cohorts showing publication bias.
- Whether mechanical preparation adds anything to oral antibiotics remains unresolved: three trials, 960 participants, very low certainty.
- Piperacillin-tazobactam is untested in laparoscopic or robotic pancreatoduodenectomy; the trial enrolled open procedures only.
- No trial has randomised patients carrying a β-lactam allergy label to cefazolin versus an alternative. Every estimate is observational.
- The compendium’s surgical antibiotic prophylaxis search closed in August 2021. Both MOBILE2 and the pancreatoduodenectomy trial post-date the guidance now used to support them.
That last gap is the one to hold onto. Guidance and trial evidence are not synchronised, and the interval between them is where a resident’s reading earns its keep.
Practical Implications
The evidence supports a short list of surgical antibiotic prophylaxis defaults, small enough to hold between timeout and closure.
- Cefazolin 2 g intravenously, inside 60 minutes of incision. Above 120 kg the guideline says 3 g; give it, knowing the threshold rests on very low-certainty evidence. Vancomycin and fluoroquinolones need the 120-minute window.
- Redose cefazolin at four hours, or after blood loss above 1,500 mL. The decision most often lost to a long case.
- Stop at skin closure. Not at 24 hours, not when the drain comes out. The trials that appear to support continuation are those where the first dose was given wrong.
- Add oral antibiotics before elective colorectal and rectal resection. The randomised evidence here is stronger than for most of what surrounds it.
- Interrogate the allergy label before substituting. A childhood rash is not an indication for clindamycin, and the substitution is where the risk sits.
- For open pancreatoduodenectomy, use piperacillin-tazobactam. That evidence does not travel to other operations.
None of this is a directive; it is what the evidence supports, mapped onto the decisions that fall to the person holding the chart. Where local surgical antibiotic prophylaxis protocol conflicts, the protocol governs. Bring the evidence to the stewardship committee, not the anaesthetic room.
Evidence Grade and Bottom Line
Surgical antibiotic prophylaxis by component: high for duration and colorectal oral antibiotics; moderate for timing, agent selection and the allergy label; low for weight-based dosing and redosing.
- Strong evidence supports stopping surgical antibiotic prophylaxis at incisional closure, and adding oral antibiotics before elective colorectal resection.
- Evidence suggests the 60-minute pre-incision window is the operative rule, with no penalty for dosing near the incision and no benefit from dosing early.
- Evidence suggests a reported β-lactam allergy is associated with higher infection risk through the agents it prompts, and most such labels do not preclude cefazolin.
- Preliminary evidence suggests redosing helps in long operations, entirely from observational data.
- Insufficient evidence to determine whether the 3 g cefazolin threshold changes clinical outcomes.
Surgical antibiotic prophylaxis has narrowed rather than expanded. The defensible position is a correct single dose, redosed when the operation runs long, oral antibiotics where the colon is opened, and nothing after the skin is closed. Everything else — the extra day of cefazolin, the reflexive clindamycin, the third gram — is unsupported or carried by evidence too weak for the confidence it is given.
Article Information and References
- Willis MA, Toews I, Soltau SLV, et al. Preoperative combined mechanical and oral antibiotic bowel preparation for preventing complications in elective colorectal surgery. Cochrane Database Syst Rev. 2023;2(2):CD014909. DOI: https://doi.org/10.1002/14651858.CD014909.pub2
- de Jonge SW, Boldingh QJJ, Solomkin JS, et al. Effect of postoperative continuation of antibiotic prophylaxis on the incidence of surgical site infection: a systematic review and meta-analysis. Lancet Infect Dis. 2020;20(10):1182–1192. DOI: https://doi.org/10.1016/S1473-3099(20)30084-0
- Hanai Y, Hirai J, Matsuo K, et al. Consequences of reported β-lactam allergy on perioperative outcomes: a systematic review and meta-analysis of surgical site infection risk. Infect Control Hosp Epidemiol. 2025;46(11):1140–1150. DOI: https://doi.org/10.1017/ice.2025.10294
- Huisman H, Huinink K, Bontekoning N, et al. Weight-based dosing of surgical antibiotic prophylaxis in patients with obesity: meta-analysis. BJS Open. 2026;10(2):zrag015. DOI: https://doi.org/10.1093/bjsopen/zrag015
- Hanai Y, Hirai J, Kobayashi M, et al. Intraoperative redosing of antibiotics for prevention of surgical site infections: a systematic review and meta-analysis. Ann Gastroenterol Surg. 2025;9(2):369–378. DOI: https://doi.org/10.1002/ags3.12866
- de Jonge SW, Gans SL, Atema JJ, et al. Timing of preoperative antibiotic prophylaxis in 54,552 patients and the risk of surgical site infection: a systematic review and meta-analysis. Medicine (Baltimore). 2017;96(29):e6903. DOI: https://doi.org/10.1097/MD.0000000000006903
- Märtson AG, Barber KE, Crass RL, et al. The pharmacokinetics of antibiotics in patients with obesity: a systematic review and consensus guidelines for dose adjustments. Lancet Infect Dis. 2025;25(9):e504–e515. DOI: https://doi.org/10.1016/S1473-3099(25)00155-0
- Calderwood MS, Anderson DJ, Bratzler DW, et al. Strategies to prevent surgical site infections in acute-care hospitals: 2022 update. Infect Control Hosp Epidemiol. 2023;44(5):695–720. DOI: https://doi.org/10.1017/ice.2023.67
- 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: https://doi.org/10.2146/ajhp120568
- D’Angelica MI, Ellis RJ, Liu JB, et al. Piperacillin-tazobactam compared with cefoxitin as antimicrobial prophylaxis for pancreatoduodenectomy: a randomized clinical trial. JAMA. 2023;329(18):1579–1588. DOI: https://doi.org/10.1001/jama.2023.5728
- Koskenvuo L, Lunkka P, Varpe P, et al. Morbidity after mechanical bowel preparation and oral antibiotics prior to rectal resection: the MOBILE2 randomized clinical trial. JAMA Surg. 2024;159(6):606–614. DOI: https://doi.org/10.1001/jamasurg.2024.0184
- Weber WP, Mujagic E, Zwahlen M, et al. Timing of surgical antimicrobial prophylaxis: a phase 3 randomised controlled trial. Lancet Infect Dis. 2017;17(6):605–614. DOI: https://doi.org/10.1016/S1473-3099(17)30176-7
- Sommerstein R, Troillet N, Harbarth S, et al. Timing of cefuroxime surgical antimicrobial prophylaxis and its association with surgical site infections. JAMA Netw Open. 2023;6(6):e2317370. DOI: https://doi.org/10.1001/jamanetworkopen.2023.17370
- Sommerstein R, Atkinson A, Kuster SP, et al. Association between antimicrobial prophylaxis with double-dose cefuroxime and surgical site infections in patients weighing 80 kg or more. JAMA Netw Open. 2021;4(12):e2138926. DOI: https://doi.org/10.1001/jamanetworkopen.2021.38926
- Blumenthal KG, Ryan EE, Li Y, et al. The impact of a reported penicillin allergy on surgical site infection risk. Clin Infect Dis. 2018;66(3):329–336. DOI: https://doi.org/10.1093/cid/cix794