Skip to main content

Table 4 Stopping rules based on the O’Brien-Fleming alpha spending method

From: The NICU Antibiotics and Outcomes (NANO) trial: a randomized multicenter clinical trial assessing empiric antibiotics and clinical outcomes in newborn preterm infants

Interim Decision action Z test statistic boundary valuea P value
First § Stop the trial due to higher incidence of composite adverse events in EA Z > 3.66474 P < 0.00025
§ Stop the trial due to higher incidence of composite adverse events in placebo Z < −3.66474 P < 0.00025
§ Continue the trial Z in [−3.66474, 3.66474] P3 0.00025
Second § Stop the trial due to higher incidence of composite adverse events in EA Z > 2.50210 P < 0.01235
§ Stop the trial due to higher incidence of composite adverse events in placebo Z < −2.50210 P < 0.01235
§ Continue the trial Z in [−2.50210, 2.50210] P3 0.01235
Final § Incidence of composite adverse events is significantly higher in EA Z > 1.96 P < 0.05
§ Incidence of composite adverse events is significantly higher in placebo Z < −1.96 P < 0.05
§ Incidence of composite adverse events is not significantly different between EA and placebo Z in [−1.96, 1.96] P3 0.05
  1. aZ test statistic = [(incidence of NEC/LOS/death in EA) − (incidence of NEC/LOS/death in placebo)]/SE