Table 3. Restricted mean survival time estimates from the stratified Cox Model.

Time point	RMST HA group [95% CI]		RMST N-HA group [95% CI]		RMST difference (±SD)	Zucker test (p-value)
365 days (1 year)	333.18	[332.91, 333.46]	282.35	[280.40, 284.30]	+50.83 (±1.10)	< 0.0001
1095 days (3 years)	681.56	[677.93, 685.18]	507.95	[497.87, 518.02]	+173.61 (±5.16)	< 0.0001
1825 days (5 years)	807.37	[799.84, 814.90]	592.42	[576.36, 608.49]	+214.94 (±7.95)	< 0.0001
2794 days (7.5 years)	842.18	[832.13, 852.23]	625.41	[605.23, 645.60]	+216.77 (±9.70)	< 0.0001

The use of RMST to analyze time-to TKR of patients with knee OA allows testing the difference between treatment groups at different time points from diagnosis. For instance, the difference of RMST at 365 days is about 51 days in favor of HA group, we might say that HA might increase time-to TKR of knee OA patients by 51 days during the first year following diagnosis.RMST: restricted mean survival time, CI: confidence interval, HA group: patients who received at least one IA HA injection during their follow-up period, N-HA group: patients who received only CS as IA injections, TKR: total knee replacement, OA: osteoarthritis, IA: intra-articular, HA: hyaluronic acid, CS: corticosteroids.