Eight RCTs (972 participants, range 40 to 252) were included. Randomisation and allocation concealment were considered adequate in four trials. All eight studies defined eligibility criteria and reported follow-up for at least one year. Half of the trials used intention-to-treat analyses. One study reported that surgeons were experienced in both operations and one study reported that outcome assessors were blinded. Two studies reported losses to follow-up of less than 5%.
There was an almost significant increased risk of dislocation within one year for total hip arthroplasty compared with hemiarthroplasty (RR 3.98, 95% CI 0.98 to 16.12; Ι²=46%, five RCTs); a statistically significant increased risk of dislocation for patients treated with total hip arthroplasty (RR 2.40, 95% CI 1.21 to 4.76; Ι²=13%, six RCTs) was seen for all follow-up periods up to 13 years.
There was a statistically non-significant 59% reduced risk of revision within one year for total hip arthroplasty compared with hemiarthroplasty (RR 0.41, 95% CI 0.16 to 1.03; Ι²=0%, four RCTs). A statistically significant 69% reduced risk of revision for patients treated with total hip arthroplasty (RR 0.31, 95% CI 0.17 to 0.59; Ι²=0%, six RCTs) was seen for all follow-up periods up to 13 years.
All eight RCTs reported differences in hip scores that favoured total hip arthroplasty (meta-analysis was not performed).
There were no statistically significant differences for risk of any surgery (reduction of dislocations, revisions and all other surgical interventions) within one year and for all follow-up periods and for risk of mortality within one year and for all follow-up periods. In subgroup analyses there were no statistically significant differences between risk ratios for confounding variables.