Twenty studies (4,508 patients, range 31 to 683) were included. Follow-up ranged from one to 13 years. Evidence quality was generally considered moderate except for three studies classed as low quality and two rated as high quality. Six trials reported a clearly adequate method of concealment and four used inadequate randomisation methods. Four studies included information about outcome assessor blinding. Loss to follow-up ranged from zero to 17.9% for the 13 trials that reported it.
Compared with internal fixation, arthroplasty reduced the risk of the major complications at one year (RR 0.33, 95% CI 0.21 to 0.54; 16 RCTs; Ι²=83%) and at five years (RR 0.22, 95% CI 0.16 to 0.31; three RCTs; Ι²=35%) following surgery. Incidence of reoperation was lower in the intervention group at one year (RR 0.22, 95% CI 0.15 to 0.34; 17 RCTs; Ι²=79%) and at five years (RR 0.13, 95% CI 0.08 to 0.24; three RCTs; Ι²=0%). Arthroplasty led to lower postoperative pain (RR 0.50, 95% CI 0.34 to 0.72; nine RCTs; Ι²=85%) and superior function (RR 0.59, 95% CI 0.44 to 0.79; nine RCTs; Ι²=80%) compared with internal fixation.
There was no statistical difference in mortality at one to three years after surgery between the two types of interventions, although at one year there was a trend in favour of internal fixation (RR 1.09, 95% CI 0.96 to 1.23; 19 RCTs; Ι²=13%). Results of sensitivity analyses were reported. There was no evidence of publication bias.