Among the five RCT meeting the inclusion criteria [
8‐
12], the study by Holbrook and co-workers did not describe the fracture subtypes considered; Tornetta et al. treated only Gustilo type IIIB fractures; Tu et al. and Mohseni et al. treated both IIIA and IIIB fractures; and Inan et al. considered only IIIA fractures. No study seemed to include IIIC fractures. We divided patients into two approach-based groups: IMN and EF. Of the 57 patients described by Holbrook et al. [
8], 29 were treated by IMN and 28 by EF; 2 of the former patients had infection, 7 had fracture healing problems and 12 had “other complications”; of their EF patients 11 had infection, 9 had fracture healing problems and 11 had “other complications”. Tornetta et al. [
9] included 29 patients, 14 with IMN and 15 with EF; of those treated with IMN 3 had infection, none had fracture healing problems and 5 had “other complications”; among those managed by EF, there were 6 cases of infection, 2 fracture healing problems and 8 “other complications”. Tu et al. [
10] described 36 patients, 18 treated with IMN and 18 with EF; those receiving IMN had 4 infections, 3 fracture healing problems and 2 “other complications”, whereas patients treated with EF had 2 infections, 8 fracture healing problems and 3 “other complications”. Inan et al. [
11] treated 61 patients, 29 with IMN and 32 with EF; of the IMN patients, 3 had infection, 4 had fracture healing problems and 7 had “other complications”; of the EF patients, 2 had infection, 4 had fracture healing problems and 3 had “other complications”. Mohseni et al. [
12] considered 50 patients, 25 treated by IMN and 25 by EF; of those managed by IMN, 4 developed infection, 1 had fracture healing problems and 2 had “other complications”, whereas of those receiving EF 8 had infection, 8 had fracture healing problems and 3 had “other complications”.
The data from the 2 × 2 tables were used to obtain the forest plot for each outcome measure considered. The plot for the incidence of infection (Table
1) favours IMN due to a lower incidence of infection (OR = 0.48); the same applies to fracture healing problems (Table
2) (OR = 0.41). The results for “other complications” (Table
3) are not significant (OR = 1.14), providing no clear indication.
Table 1
Forest plot 1: infections
Table 2
Forest plot 2: fracture healing problems
Table 3
Forest plot 3: other complications