Time From Injury to First Antibiotic Administration and Deep Infection Risk in Open Fractures: A Systematic Review and Meta-Analysis

Deep infection is a serious complication of open fractures, and prompt antibiotic prophylaxis is widely emphasized.However, evidence regarding the association between the timing of initial antibiotic administration and infection risk remains inconsistent, and commonly cited time thresholds are based largely on individual studies.This review makes two contributions.First, it separates studies by the clock used to measure delay, since time from injury and time from arrival begin at different zero points and are different exposures that should not be pooled.Second, it reframes the field's conflicting results as a measurement quality problem, because the disagreement tracks how infection was defined.PubMed, Embase, and Scopus were searched from inception to August 2026 (PRISMA 2020; PROSPERO CRD420261473036) for observational studies reporting time from injury or hospital arrival to the first prophylactic antibiotic and a deep or fracture-related infection outcome.Injury-anchored and arrivalanchored studies formed separate, prespecified strata that were never pooled in the primary analysis; ORs used random-effects models, a post hoc subgroup compared studies by infection-definition validity, and bias and certainty were assessed using Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I), Methodological Index for Non-Randomized Studies (MINORS), and Grading of Recommendations Assessment, Development and Evaluation (GRADE).Twenty-four studies were included, and 16 were poolable (5,283 patients, 508 infections).Delay was not significantly associated with infection in either stratum (injury-anchored OR 1.62, 95% CI 0.87-3.00,k = 11; arrival-anchored OR 1.34, 95% CI 0.86-2.07,k = 5), and no time threshold was estimable.Forcing both anchors into one pool produced a borderline OR 1.53 (95% CI 1.00-2.32),but this is a pooling artifact: a borderline value assembled by adding precision across two individually nonsignificant strata that measure different exposures (anchor moderator p = 0.811).The disagreement instead tracked infection definition: the four studies using a validated definition reported OR 4.34 (95% CI 2.45-7.69,k = 4), roughly four-fold higher odds with delay, while the seven using an operative proxy reported OR 0.86 (95% CI 0.63-1.17,k = 7), with a significant subgroup difference (p = 0.0015).This definition-based signal is post hoc and confounded, since it cannot be separated from whether a study used an adjusted analysis, and overall certainty was very low; it is a hypothesis for prospective testing, not a reliable estimate of the effect of delay.Testing it will require prospective studies that measure timing from injury, define infection by validated criteria, and report adjusted estimates, ideally pooled as individual participant data.Two points follow for practice.A nonsignificant pooled result under very low-certainty evidence is absence of evidence, not license to relax timing.Moreover, the numeric thresholds built into current audit metrics and protocols are not evidence-based, even though giving antibiotics as early as feasible remains sound on biological and guideline grounds.