TABLE III.

Results from the Random-Effects Model: Hazard Ratios for Revision After Mobile Compared with Fixed-Bearing Posterior-Stabilized Knee Replacement^*

Mobile-Bearing PS, Relative to Fixed-Bearing PS	HR (95% CI)	P Value
Time in yr
0 to 1	1.76 (1.215-2.55)	0.003
1 to 2	1.062 (0.934-1.208)	0.357
2 to 3	1.017 (0.915-1.13)	0.759
3 to 4	1.037 (0.939-1.145)	0.475
4 to 5	1.012 (0.917-1.118)	0.809
5 to 6	1.012 (0.908-1.128)	0.831
6 to 7	1.023 (0.869-1.205)	0.785
7 to 8	1.06 (0.811-1.386)	0.667
8 to 9	1.059 (0.769-1.458)	0.726
Sex
Male	Ref.
Female	0.799 (0.747-0.854)	<0.001
Age in yr
>65	Ref.
≤65	0.556 (0.52-0.595)	<0.001
Resurfacing
No	Ref.
Yes	0.67 (0.62-0.724)	<0.001
Fixation
Cemented	Ref.
Cementless	1.491 (1.259-1.766)	<0.001
Hybrid	1.280 (1.01-1.491)	0.002
Fixed registry effects†	—	—
Random registry effects‡	—	—

^*Confidence intervals and p values are based on a t distribution. HR = hazard ratio, CI = confidence interval, and PS = posterior-stabilized knee prostheses.

^†Fixed registry effects were included in this model (five coefficients), but the results are omitted from this table because a precondition of data sharing was no reporting of comparisons among registries. The estimated intercept of the fixed registry effects was −5.522 (standard error, 0.134).

^‡The estimated intercept of the random registry effects was 0.020 (standard error, 0.020).