Hwang SM, et al. J Korean Med Sci. 2014 Feb;29(2):287-291. http://dx.doi.org/10.3346/jkms.2014.29.2.287

In the emergency department (ED), extracorporeal membrane oxygenation (ECMO) can be used as a rescue treatment modality for patients with refractory circulatory and/or respiratory failure. Serious consideration must be given to the indication, and the PRESERVE and RESP scores for mortality have been investigated. However these scores were validated to predict survival in patients who received mainly veno-venous (VV) ECMO in the intensive care unit. The aim of the present study was to investigate the factors that predicted the outcomes for patients who received mixed mode (veno-arterial [VA] and VV) ECMO support in the ED. This single center retrospective study included 65 patients who received ECMO support at the ED for circulatory or respiratory failure between January 2009 and December 2013. Pre-ECMO SAPS II and other variables were evaluated and compared for predicting mortality. Fifty-four percent of patients received ECMO-cardiopulmonary resuscitation (E-CPR), 31 % received VA and V-AV ECMO, and 15 % received VV ECMO. The 28-day and 60-month mortality rates were 52 % and 63 %. In the multivariate analysis, only the pre-ECMO Simplified Acute Physiology Score II (SAPS II) (odd ratio: 1.189, 95 % confidence interval: 1.032–1.370, p = 0.016) could predict the 28-day mortality. The area under the receiver operating characteristic curve and the optimal cutoff value for pre-ECMO SAPS II in predicting 28-day mortality was 0.852 (95 % CI: 0.753–0.951, p < 0.001) and 80 (sensitivity of 97.1 % and specificity of 71.0 %), respectively. Validation of the 80 cutoff value revealed a statistically significant difference for the 28-day and 60-month mortality rates in the overall, E-CPR, and VA groups (28-day: p < 0.001, p = 0.004, p = 0.005; 60-month: p < 0.001, p = 0.004, p = 0.020). In the Kaplan-Meier analysis, the 28-day and 60-month survival rates were lower among the patients with a pre-ECMO SAPS II of ≤80, compared to those with a score of >80 (both, p < 0.001). The pre-ECMO SAPS II could be helpful for identifying patients with refractory acute circulatory and/or respiratory failure who will respond to ECMO support in the ED.

Kwon Y, et al. Korean J Anesthesiol. 2015 Feb;68(1):32-36. http://dx.doi.org/10.4097/kjae.2015.68.1.32

Background: We evaluated the endotracheal tube cuff pressure (P _cuff ) changes during pneumoperitoneum for laparoscopic cholecystectomy and the correlations between body mass index (BMI), pneumoperitoneum time, and P _cuff changes. Methods: Total 60 patients undergoing laparoscopic cholecystectomy were allocated to either a study group (BMI ≥ 25 kg/m ² ) or a control group (BMI < 25 kg/m ² ). The endotracheal intubation was performed with a high-volume low-pressure cuffed oral endotracheal tube. A manometer was connected to the pilot balloon using a 3-way stopcock and the cuff was inflated. The change in P _cuff was defined as the difference between the pressure just before intra-abdominal CO ₂ insufflation and the pressure before CO ₂ desufflation. Results: P _cuff increased to 5.3 ± 3.6 cmH ₂ O in the study group and 5.7 ± 5.4 cmH ₂ O in the control group. There was no significant difference between two groups. While BMI was not correlated with change in P _cuff (r = 0.022, p = 0.867), there was a significant correlation between change in P _cuff and the duration of pneumoperitoneum (r = 0.309, p = 0.016). Conclusion: The change in P _cuff was not affected by BMI and was significantly correlated with pneumoperitoneum time. We recommend regular measurement and adjustment of P _cuff during laparoscopic surgery.

Postoperative nausea and vomiting (PONV) are common complications after surgery. This study aimed to present the utilization of machine learning for predicting PONV and provide insights based on a large amount of data. This retrospective study included data on perioperative features of patients, such as patient characteristics and perioperative factors, from two hospitals. Logistic regression algorithms, random forest, light-gradient boosting machines, and multilayer perceptrons were used as machine learning algorithms to develop the models. The dataset of this study included 106,860 adult patients, with an overall incidence rate of 14.4% for PONV. The area under the receiver operating characteristic curve (AUROC) of the models was 0.60–0.67. In the prediction models that included only the known risk and mitigating factors of PONV, the AUROC of the models was 0.54–0.69. Some features were found to be associated with patient-controlled analgesia, with opioids being the most important feature in almost all models. In conclusion, machine learning provides valuable insights into PONV prediction, the selection of significant features for prediction, and feature engineering.