We undertook a retrospective case-control study to assess the clinical variables associated with infections in open fractures. A total of 1492 open fractures were retrieved; these were Gustilo and Anderson grade I in 663 (44.4%), grade II in 370 (24.8%), grade III in 310 (20.8%) and unclassifiable in 149 (10.0%). The median duration of prophylaxis was three days (interquartile range (IQR) 1 to 3), and the median number of surgical interventions was two (1 to 9). We identified 54 infections (3.6%) occurring at a median of ten days (IQR 5 to 20) after trauma. Pathogens intrinsically resistant to the empirical antibiotic regimen used (enterococci, Enterobacter spp, Pseudomonas spp) were documented in 35 of 49 cases (71%). In multivariable regression analyses, grade III fractures and vascular injury or compartment syndrome were significantly associated with infection. Overall, compared with one day of antibiotic treatment, two to three days (odds ratio (OR) 0.6 (95% confidence interval (CI) 0.2 to 2.0)), four to five days (OR 1.2 (95% CI 0.3 to 4.9)), or > five days (OR 1.4 (95% CI 0.4 to 4.4)) did not show any significant differences in the infection risk. These results were similar when multivariable analysis was performed for grade III fractures only (OR 0.3 (95% CI 0.1 to 3.4); OR 0.6 (95% CI 0.2 to 2.1); and OR 1.7 (95% CI 0.5 to 6.2), respectively).

Infection in open fractures is related to the extent of tissue damage but not to the duration of prophylactic antibiotic therapy. Even for grade III fractures, a one-day course of prophylactic antibiotics might be as effective as prolonged prophylaxis.

Cite this article: Bone Joint J 2013;95-B:831–7.

Between June 1999 and May 2003 we undertook direct primary closure of the skin wounds of 173 patients with Gustilo and Anderson grade-IIIA and grade-IIIB open fractures. These patients were selected from a consecutive group of 557 with type-III injuries presenting during this time. Strict criteria for inclusion in the study included debridement within 12 hours of injury, no sewage or organic contamination, no skin loss either primarily or secondarily during debridement, a Ganga Hospital open injury skin score of 1 or 2 with a total score of ten or less, the presence of bleeding skin margins, the ability to approximate wound edges without tension and the absence of peripheral vascular disease. In addition, patients with polytrauma were excluded.

At a mean follow-up of 6.2 years (5 to 7), the outcome was excellent in 150 (86.7%), good in 11 (6.4%) and poor in 12 (6.9%). A total of 33 complications occurred in 23 patients including superficial infection in 11, deep infection in five and the requirement for a secondary skin flap in three. Six patients developed nonunion requiring further surgery, one of whom declined additional measures to treat an established infected nonunion.

Immediate skin closure when performed selectively with the above indications proved to be a safe procedure.