Previously published drug interaction models predict loss of response for transoesophageal echocardiography sedation well but not response to oesophageal instrumentation.

Response surface models (RSMs) were used to predict effects of multiple drugs interactions. Our study was aimed to validate accuracy of the previous published volunteer models during transoesophageal echocardiography (TEE). This is a cross-sectional study with 20 patients scheduled for transesophageal echocardiography in Taipei Veterans General Hospital, Taiwan. Effect-site concentration pairs of alfentanil and propofol were recorded and converted to equivalent remifentanil and propofol effect-site concentrations. Observer's Assessment of Alertness/Sedation (OAA/S) scores were assessed every 2 minutes. Using these data, previous published models of loss of response (LOR), intolerable ventilatory depression (IVD), and loss of response to esophageal instrumentation (LREI) were then estimated. Accuracy of prediction is assessed by calculating the difference between the true response and the model-predicted probability. Clinical events such as interruption of TEE were recorded. The average procedure time was 11 minutes. Accuracy for prediction of LOR and LREI is 63.6% and 38.5%, respectively. There were four patients experienced desaturation for less than 1 minute, which were not predicted by IVD model, and one interruption of TEE due to involuntary movement. The previous published drug-interaction RSMs predict LOR well but not LREI for TEE sedation. Further studies using response surface methodology are needed to improve quality for TEE sedation and clinical implementation.