Seven studies (n=700 knees, range 46 to 230 knees) were included in the meta-analysis. Follow-up ranged from 0.5 years to 15 years. Four studies reported adequate allocation concealment. Two studies reported adequate randomisation. One study reported adequate blinding.
Based on five studies where follow-up was five years or less (610 knees), the authors estimated there was a statistically significantly increased risk of revision in the total knee arthroplasty compared with the unicompartmental knee arthroplasty groups (RR 3.47, 95% CI 1.23 to 9.77, I2=0%, fixed-effect model). In the two studies (148 knees) with follow-up of more than five years, there was no statistically significant difference in the risk of revision between groups.
Based on five studies (546 knees), risk of complications was lower in unicompartmental than in total knee arthroplasty groups (RR 0.20, 95% CI 0.08 to 0.52, I2=0%, fixed-effect model).
Based on five studies (n=586 knees), total knee arthroplasty groups had a statistically significantly increased range of motion compared with unicompartmental knee arthroplasty groups (WMD 6.43, 95% CI 3.46 to 9.40, I2=70%, random-effects model).
Based on four studies (452 knees) there were no statistically significant differences in Knee Society Scores in the two groups (I2=65%, fixed-effect and random-effects models).