At one month, pain was reported by 137 (48) patients in the injection group versus 164 (58) in the control group (relative risk; RR: ; 95 CI: -; p = ). After three months, these values were 58 (21) and 63 (24), respectively (RR: ; 95 CI: -; p = ), and at 6 months: 39 (15) and 44 (17) (RR: ; 95 CI: -; p = ). No subgroups were detectable in which the relative risk for pain at one month after inclusion substantially differed from the overall estimate. At one month, the median severity of pain in the injection group was 2 (on a 100-points scale) versus 6 in the control group (p = ). At later follow-up, there was no longer any statistically significant difference in the severity of pain between the two groups. No patient had major adverse events related to the epidural injection.