Twenty studies comprising 24 comparisons were included in the review (n at least 112,094; 2 studies did not report the sample size) . Three studies were RCTs and 16 were cohort studies, of which five were multi-centre, seven were single-centre, two were matched pair and two were bilaterally controlled. The design of one study was not reported.
There was no statistically significant difference in the survivorship proportion between cemented and uncemented fixation groups in the main analysis of all 24 comparisons: the pooled difference was -0.005 (95% confidence interval, CI: -0.031, 0.022). Subgroup analysis of the RCTs (4 comparisons) also showed no significant difference between the groups (difference 0.016, 95% CI: -0.054, 0.086), and the results of randomised and non-randomised studies did not differ significantly (p=0.30). None of the other reported a priori subgroup analyses showed a significant difference between the groups. However, the analysis of specific components showed that comparisons using a threaded or macrogrowth implant favoured cemented fixation (difference in survivorship -0.054, 95% CI: -0.090, -0.018; 2 comparisons), whilst those using a micro-ingrowth or ongrowth cup showed no significant difference (0.031, 95% CI: -0.056, 0.12; 2 comparisons); the difference between the subgroups was significant (p=0.02). A similar effect was found when stainless steel or cobalt chrome cemented stems (difference -0.051, 95% CI: -0.098, -0.004 favouring cemented fixation; 5 comparisons) were contrasted with titanium cemented stems (difference 0.12, 95% CI: 0.051, 0.19 favouring uncemented designs; 2 comparisons); the difference between the subgroups was significant (p<0.001).
Meta-regression found that year of publication was associated with improvements in the survival of uncemented implants compared with cemented implants.
No evidence of publication bias was detected (p=0.50).