Summary The programmed intermittent epidural bolus technique has shown superiority to continuous epidural infusion techniques, with or without patient‐controlled epidural analgesia for pain relief, reduced motor block and patient satisfaction. Many institutions still use patient‐controlled epidural analgesia without a background infusion, and a comparative study between programmed intermittent epidural bolus and patient‐controlled epidural analgesia without a background infusion has not yet been performed. We performed a randomised, two‐centre, double‐blind, controlled trial of these two techniques. The primary outcome was the incidence of breakthrough pain requiring a top‐up dose by an anaesthetist. Secondary outcomes included: motor block; pain scores; patient satisfaction; local anaesthetic consumption; and obstetric and neonatal outcomes. We recruited 130 nulliparous women who received initial spinal analgesia, and then epidural analgesia was initiated and maintained with either programmed intermittent epidural bolus or patient‐controlled epidural analgesia using ropivacaine 0.12% with sufentanil 0.75 µg·ml‐1. The programmed intermittent epidural bolus group had a programmed bolus of 10 ml every hour, with on‐demand patient‐controlled epidural analgesia boluses of 5 ml with a 20 min lockout, and the patient‐controlled epidural analgesia group had a 5 ml bolus with a 12 min lockout interval; the potential maximum volume per hour was the same in both groups. The patients in the programmed intermittent epidural bolus group had less frequent breakthrough pain compared with the patient‐controlled epidural analgesia group, 7 (10.9%) vs. 38 (62.3%; p < 0.0001), respectively. There was a significant difference in motor block (modified Bromage score ≤ 4) frequency between groups, programmed intermittent epidural bolus group 1 (1.6%) vs. patient‐controlled epidural analgesia group 8 (13.1%); p = 0.015. The programmed intermittent epidural bolus group had greater local anaesthetic consumption with fewer patient‐controlled epidural analgesia boluses. Patient satisfaction scores and obstetric or neonatal outcomes were not different between groups. In conclusion, we found that a programmed intermittent epidural bolus technique using 10 ml programmed boluses and 5 ml patient‐controlled epidural analgesia boluses was superior to a patient‐controlled epidural analgesia technique using 5 ml boluses and no background infusion.

process. This is a good example where it is not just about "statistics" but the clinical significance of approaching the study objective, which altered the final study conclusion. And even after the peer-review driven modifications, and adjusting for many potential confounders, the study results seemed to be at least counterintuitive to the study aim.Other consideration in "big data" research is to take in account inherent flaws related to retrospective observational studies, mainly related with selection bias. In this particular, the confounding by indication may not be accounted through the source of information (birth certificates), and does not allow to show whether NA was indicated before or after the indication of the operative delivery.As clinicians seeking evidence for best practice, we should value different types of research designs while acknowledging their strengths and limitations.