Background: Few studies address the dynamic effect of opioids on respiration. Models with intact feedback control of carbon dioxide on ventilation (non-steady-state models) that correctly... Show moreBackground: Few studies address the dynamic effect of opioids on respiration. Models with intact feedback control of carbon dioxide on ventilation (non-steady-state models) that correctly incorporate the complex interaction among drug concentration, end-tidal partial pressure of carbon dioxide concentration, and ventilation yield reliable descriptions and predictions of the behavior of opioids. The authors measured the effect of remifentanil on respiration and developed a model of remifentanil-induced respiratory depression. Methods: Ten male healthy volunteers received remifentanil infusions with different infusion speeds (target concentrations: 4-9 ng/ml; at infusion rates: 0.17-9 ng . ml(-1) . min(-1)) while awake and at the background of low-dose propofol. The data were analyzed with a nonlinear model consisting of two additive linear parts, one describing the depressant effect of remifentanil and the other describing the stimulatory effect of carbon dioxide on ventilation. Results: The model adequately described the data including the occurrence of apnea. Most important model parameters were as follows: C-50 for respiratory depression 1.6 +/- 0.03 ng/ml, gain of the respiratory controller (G) 0.42 - 0.1 l.min(-1) . Torr(-1), and remifentanil blood effect site equilibration half-life (t1/2k(e0)) 0.53 +/- 0.2 min. Propofol caused a 20-50% reduction of C50 and G but had no effect on t1/2k(e0). Apnea occurred during propofol infusion only. A simulation study revealed an increase in apnea duration at infusion speeds of 2.5-0.5 ng.ml(-1).min(-1) followed by a reduction. At an infusion speed of <= 0.31 ng.ml(-1).min(-1), no apnea was seen. Conclusions: The effect of varying remifentanil infusions with and without a background of low-dose propofol on ventilation and end-tidal partial pressure of carbon dioxide concentration was described successfully using a non-steady-state model of the ventilatory control system. The model allows meaningful simulations and predictions. Show less