To determine the effect of reamed versus nonreamed intramedullary (IM) nailing of lower extremity long bone fractures on the rates of nonunion, implant failure, malunion, compartment syndrome, pulmonary embolus, and infection.

Quantitative systematic review of prospective, randomized controlled trials.

MEDLINE and SCISEARCH computer searches provided lists of published randomized clinical trials from 1969 to 1998. Extensive hand searches of major orthopaedic journals, bibliographies of major orthopaedic texts, and personal files identified additional studies.

Of 676 citations initially identified, sixty proved potentially eligible, of which four published and five unpublished randomized trials met all eligibility criteria. Each of three investigators assessed study quality and abstracted relevant data.

The pooled relative risk of reamed versus nonreamed nails (nine trials, n = 646 patients) was 0.33 [95% confidence interval (CI), 0.16 to 0.68; p = 0.004]. The absolute risk difference in nonunion rates with reamed IM nailing was 7.0 percent (95% CI, 1 to 11 percent). Thus, one nonunion could be prevented for every fourteen patients treated with reamed IM nailing [number needed to treat (NNT) = 14.28]. The risk ratios for secondary outcome measures were: implant failure, 0.30 (95% CI, 0.16 to 0.58; p < 0.001); malunion, 1.06 (95% CI, 0.32 to 3.57); pulmonary embolus, 1.10 (95% CI, 0.26 to 4.76); compartment syndrome, 0.45 (95% CI, 0.13 to 1.56); and infection, 0.98 (95% CI, 0.21 to 4.76). Sensitivity analyses suggested that reported rates of nonunion and implant failure were higher in studies of lower quality. The type of long bone fractured (tibia or femur), the degree of soft tissue injury (open or closed), study quality, and whether a study was published or unpublished did not significantly alter the relative risk of nonunion between reamed and nonreamed IM nailing.

There is evidence from a pooled analysis of randomized trials that reamed IM nailing of lower extremity long bone fractures significantly reduces rates of nonunion and implant failure in comparison with nonreamed nailing.