Planned subgroup analysis
We conducted an additional analysis according to ten deciles of annual volume in order to assess an existence of a threshold. The number of vaginal births attended by a single CNM in the 10 deciles was: ≤83 (G1), 84-106 (G2), 107-123 (G3), 124-134 (G4), 135-150 (G5), 151-163 (G6), 164-193 (G7), 185-207 (G8), 208-245 (G9), 246-390 (G10).
Maternal outcomes by the different deciles are presented inFigure 2 . Neonatal outcomes by the different percentile are presented in Figure 3 . Overall, we could not detect a difference of a lower rate of maternal or neonatal complications associated with the higher CNM workload.
Primiparity analysis :
Maternal and neonatal outcomes in primiaparous women according to the CNM annual volume revealed a similar trend as was apparent in the entire study population (Table S1 and S2, respectively). Briefly women delivered by CNM with lower annual delivery volume had statistically significant (p<0.01) higher rates of any degree of spontaneous perineal lacerations (72.3% vs. 69.8%), 2nd degree perineal tear (20.6% vs. 19.2%), and lower rates of episiotomy (56.1% vs. 58.8%), and higher rates of uterine manual exploration for suspected retained placental products (2.1% vs 1.7%). There was no statistically significant difference in any of the neonatal outcomes aside from the composite adverse neonatal outcome that occurred in 12.3% of the neonates delivered by the low annual delivery volume CNM as opposed to 11.3% of those delivered by high annual delivery volume CNM (p=0.01).