Abstract
AIM: In critically ill mechanically ventilated children, midazolam is used first line for sedation, however its exact sedative effects have been difficult to quantify. In this analysis, we use parametric time-to-event (PTTE) analysis to quantify the effects of midazolam in critically ill children. METHODS: In the PTTE analysis, data was analyzed from a published study in mechanically ventilated children in which blinded midazolam or placebo infusions were administered during a sedation interruption phase until, based on COMFORT-B and NISS scores, patients became undersedated and unblinded midazolam was restarted. Using NONMEM® v.7.4.3., restart of unblinded midazolam was analysed as event. Patients in the trial were divided into internal and external validation cohorts prior to analysis. RESULTS: Data contained 138 events from 79 individuals (37 blinded midazolam; 42 blinded placebo). In the PTTE model, the baseline hazard was best described by a constant function. Midazolam reduced the hazard for restart of unblinded midazolam due to undersedation by 51%. In the blinded midazolam group, time to midazolam restart was 26 h versus 58 h in patients with low versus high disease severity upon admission (PRISM II < 10 versus > 21), respectively. For blinded placebo, these times were 14 h and 33 h, respectively. The model performed well in an external validation with 42 individuals. CONCLUSION: The PTTE analysis effectively quantified the effect of midazolam in prolonging sedation and also the influence of disease severity on sedation in mechanically ventilated critically ill children, and provides a valuable tool to quantify the effect of sedatives. Clinical trial number and registry URL: Netherlands Trial Register, Trial NL1913 (NTR2030), date registered 28 September 2009 https://www.trialregister.nl/trial/1913 .