Antibiotics and Dental Implant Surgery: What the Trials Say About Who Needs Them

Antibiotics at implant surgery are one of the few questions in this field on which I have changed my mind twice, and I think that is the correct response to the evidence. The argument for them is simple: an implant is a large foreign body placed into a wound in a mouth full of bacteria, infections around biomaterials are hard to treat, and an infected implant almost always has to come out. The argument against is equally simple: antibiotics have side effects ranging from stomach upset to life-threatening allergy, every course selects for resistant bacteria, and most implant surgery is short, clean and performed in healthy people. The 2013 Cochrane review by Esposito, Grusovin and Worthington laid both arguments out in its background section before it counted anything, and it is still the fairest summary of why reasonable surgeons disagree.

What the Cochrane review counted was six randomized trials with 1,162 participants, most of them comparing a single 2 g dose of amoxicillin an hour before surgery against placebo. Patients who received no antibiotic were significantly more likely to lose an implant: a risk ratio of 0.33 in favor of antibiotics, with a confidence interval of 0.16 to 0.67, and no heterogeneity between trials. Based on a 6% failure rate without antibiotics, the number needed to treat to prevent one patient losing an implant was 25. Prosthesis failure showed a borderline benefit; post-operative infection and adverse events showed no significant difference, and only two minor adverse events were recorded across all six trials. The authors concluded that a single 2 g or 3 g dose of amoxicillin an hour before surgery significantly reduces implant failure and that it might be sensible to suggest it — while noting that no trial had compared different antibiotics or doses and that the value of post-operative courses remained unknown.

That review moved a generation of surgeons, me included, to a single pre-operative dose as routine. The decade since has complicated the picture in a way that is instructive rather than contradictory, and it comes down to one word in the Cochrane conclusion: implants placed in ordinary conditions.

Lund's 2015 complex systematic review, which combined Cochrane-style methods with an assessment of the reviews themselves, found seven earlier systematic reviews of which five were at high risk of bias, and four primary trials sound enough to pool. The pooled result again favored antibiotics, with a risk ratio of 0.39 and a number needed to treat of 50 — a 2% absolute reduction in implant loss. But none of the four trials showed a significant benefit on its own, and the sub-analysis reached a conclusion that the pooled number hides: there appeared to be no benefit from prophylaxis in uncomplicated implant surgery in healthy patients. The benefit in the pooled data was being carried by the more complex cases.

Romandini's 2019 network meta-analysis of nine trials and 1,693 participants approached the same data from the prescriber's side, asking which protocol was most likely to be best. The answer was a single 3 g dose of amoxicillin an hour before surgery, with a 32.5% probability of being the best protocol, while the most widely used regimen — the single 2 g dose — had a probability of only 0.2%. The authors judged prophylaxis protective against early failure overall, found the evidence insufficient to recommend a specific dose with confidence, and concluded that post-operative courses were not justified by the available literature. That last point is the one I would underline for any patient who has been sent home with a week of antibiotics after a routine implant: the trials do not support it.

Then came the trial that the field had been missing. Momand and colleagues, in 2022, randomized 474 patients across seven Swedish clinics to 2 g of amoxicillin or an identical placebo before implant surgery, double-blinded, in healthy or relatively healthy patients. Implant failures occurred in six patients in the antibiotic group, 2.5%, and seven in the placebo group, 3.0%. Post-operative infections occurred in two and five patients respectively, 0.8% and 2.1%. No patient reported an adverse event. Neither difference was significant, and the authors concluded that prophylaxis is likely of small benefit and should be avoided in most cases, given the growth of antibiotic resistance. Their 2024 systematic review, restricted to trials in overall healthy patients and pooling seven of them with 1,859 patients and 3,014 implants, found no significant difference in early failure — a risk ratio of 0.66 with a confidence interval from 0.30 to 1.47 — an absolute risk difference of 0.7%, and a number needed to treat of 143. Their conclusion: the results do not support routine prophylaxis for dental implant surgery.

Both figures are correct for the population they describe, and reading them together is the whole lesson. The 2013 Cochrane pool included trials from an era of machined surfaces, longer surgeries and less selected patients, with a baseline failure rate of 6% in the placebo groups. The 2024 pool included only healthy patients treated with modern implants in routine surgery, where the baseline failure rate was around 3% and the room for an antibiotic to improve it was correspondingly small. The absolute benefit of prophylaxis shrinks as the surgery gets cleaner, the implant gets better and the patient gets healthier, until in the population most implant patients belong to, it is too small to detect in a trial of 474 people.

The older narrative review by Ahmad and Saad in 2012 makes the same point without a trial. Across 11,406 implants from controlled studies, success was 92% without antibiotics, 96% with a pre-operative dose alone, 97% with a post-operative course alone and 96% with both — differences the authors read as showing no benefit in low- and moderate-risk patients once the confounders were considered. Whether one accepts that reading or the Cochrane one, the direction of the evidence over twelve years is consistent: the routine, healthy, single-implant case does not need an antibiotic, and the benefit that does exist concentrates in the cases that are not routine.

That leaves the question of which cases are not routine, and here the trials are less help than judgment, because they excluded exactly the patients in whom prophylaxis is most plausible. Lang's 2012 review of implants placed into fresh extraction sockets found that a course of post-operative antibiotics was the only one of five factors examined that significantly affected survival — the one setting in which the literature still favors a course rather than a dose, though it is observational evidence. Naujokat's diabetes review notes that supportive antibiotics appear to improve outcomes in diabetic patients. Extensive grafting, sinus elevation, long surgeries with many implants, and immunocompromised patients are the situations the Cochrane background listed as classical indications for surgical prophylaxis, and nothing since has removed them.

For a healthy patient having one or two implants placed into healed bone in a short procedure, I no longer prescribe an antibiotic by default, and I explain why: the best trial evidence in that exact population shows no measurable benefit, the risk of a reaction is real if small, and every unnecessary course is a contribution to resistance that the next infected patient pays for. That is a change from what I did ten years ago, and it is a change the evidence forced. If a patient in that group has a strong preference for a single pre-operative dose, the Cochrane data give it a defensible basis and I do not refuse it; what I will not do is send anyone home with a week of tablets for a routine implant, because no trial supports it.

For the cases outside that group I do prescribe, and I prescribe on the pattern the trials examined: a single dose an hour before surgery, amoxicillin unless the patient is allergic, at 2 g or 3 g. Those cases are immediate implants into extraction sockets, particularly where the tooth was infected; sinus lifts and substantial grafts, where a large volume of foreign material is placed; full-arch surgery with many implants and a long operating time; patients with diabetes that is not well controlled, or another condition that blunts the immune response; and patients whose history includes a previous implant infection. Post-operative courses are reserved for the socket cases where Lang's data apply and for surgery into a site that was actively infected, and they are kept short.

One category is separate from all of this and often confused with it. Some patients need antibiotics before any dental procedure for a reason that has nothing to do with the implant: to protect a heart valve or a joint replacement. The American Heart Association's 2021 statement kept its 2007 recommendations unchanged for the specific cardiac conditions in which endocarditis prophylaxis is advised, and the American Dental Association's guidance sets out those categories together with its position that prophylaxis is generally not recommended for people with prosthetic joints. The American Academy of Orthopaedic Surgeons' 2025 guideline summary is the current orthopedic reference. That decision is made with the physician who manages the valve or the joint, and it is recorded on the antibiotic premedication acknowledgment before treatment; the medical-clearance page explains how it is coordinated.

If you are prescribed an antibiotic for implant surgery, three questions will tell you whether the prescription is evidence-based. Is it a single dose before surgery, or a course afterwards? The trials support the first and not, in ordinary cases, the second. What about my case makes it necessary? A good answer names something specific — an extraction socket, a graft, a medical condition — rather than habit. And is there an alternative if I am allergic to penicillin? There is, and the trials have not compared the alternatives, so the choice rests on the surgeon's judgment and your history. The same three questions apply to the blood-thinner and bone-medication conversations, which have their own pages.

None of this makes antibiotics unimportant. It makes them specific, which is what a drug with side effects and a societal cost should be. An implant that becomes infected is a serious problem, and where the evidence says a dose lowers that risk, the dose is given. Where the evidence says it does not, the patient is spared it. That is the whole of the principle, and it is the same principle behind every other page in this research series.