Table 2.
Outcomes of the stepped wedge cluster-randomised trial
| Control period | Intervention period | Unadjusted model | Mixed effects model (adjusted for study design) | |||
|---|---|---|---|---|---|---|
| n (per 10 000 deliveries) | n (per 10 000 deliveries) | IRR (95% CI)a | P-valuea | IRR (95% CI)b | P-valueb | |
| Total, n | 28 183 | 31 928 | ||||
| Primary outcome | ||||||
| Maternal death due to PPH or invasive procedures for PPHc | 19 (6.7) | 37 (11.6) | 1.72 (0.99–2.99) | 0.06 | 4.08 (1.07–15.58) | 0.04 |
| Secondary outcomes | ||||||
| Maternal death due to PPH | 10 (3.5) | 15 (4.7) | 1.32 (0.59–2.95) | 0.49 | 2.23 (0.35–14.07) | 0.39 |
| Hysterectomy due to PPH | 7 (2.5) | 13 (4.1) | 1.64 (0.65–4.11) | 0.29 | 4.38 (0.47–41.09) | 0.20 |
| Conservative surgery for PPHd | 5 (1.8) | 16 (5.1) | 2.82 (1.03–7.71) | 0.04 | Cannot estimatee | |
| Blood transfusion for PPH | 282 (100.1) | 311 (97.4) | 0.97 (0.83–1.14) | 0.74 | 1.24 (0.86–1.80) | 0.25 |
| Transfer to higher level care after PPH diagnosis | 21 (7.5) | 16 (5.0) | 0.67 (0.35–1.29) | 0.23 | 3.05 (0.79–11.70 | 0.10 |
CI, confidence interval; IRR, incident rate ratio; PPH, postpartum haemorrhage.
Derived from simple Poisson regression models.
Derived from mixed effects models include study site (cluster) as a random effect and study time period as a fixed effect.
Invasive procedures defined as hysterectomy or conservative surgical procedures (includes arterial ligation, B Lynch/compression sutures, repairof ruptured uterus).
Surgical intervention for PPH that requires laparotomy but excludes hysterectomy (includes arterial ligation, B Lynch/compression sutures, repair of ruptured uterus).
Mixed effects Poisson regression model could not generate an estimate as there were no conservative surgical procedures observed among sites that did not start UBT in the first intervention phase.