Acute kidney injury (AKI) after liver transplantation has been reported to be associated with increased mortality. Recently, machine learning approaches were reported to have better predictive ability than the classic statistical analysis. We compared the performance of machine learning approaches with that of logistic regression analysis to predict AKI after liver transplantation. We reviewed 1211 patients and preoperative and intraoperative anesthesia and surgery-related variables were obtained. The primary outcome was postoperative AKI defined by acute kidney injury network criteria. The following machine learning techniques were used: decision tree, random forest, gradient boosting machine, support vector machine, naïve Bayes, multilayer perceptron, and deep belief networks. These techniques were compared with logistic regression analysis regarding the area under the receiver-operating characteristic curve (AUROC). AKI developed in 365 patients (30.1%). The performance in terms of AUROC was best in gradient boosting machine among all analyses to predict AKI of all stages (0.90, 95% confidence interval [CI] 0.86–0.93) or stage 2 or 3 AKI. The AUROC of logistic regression analysis was 0.61 (95% CI 0.56–0.66). Decision tree and random forest techniques showed moderate performance (AUROC 0.86 and 0.85, respectively). The AUROC of support the vector machine, naïve Bayes, neural network, and deep belief network was smaller than that of the other models. In our comparison of seven machine learning approaches with logistic regression analysis, the gradient boosting machine showed the best performance with the highest AUROC. An internet-based risk estimator was developed based on our model of gradient boosting. However, prospective studies are required to validate our results.

Machine learning approaches were introduced for better or comparable predictive ability than statistical analysis to predict postoperative outcomes. We sought to compare the performance of machine learning approaches with that of logistic regression analysis to predict acute kidney injury after cardiac surgery. We retrospectively reviewed 2010 patients who underwent open heart surgery and thoracic aortic surgery. Baseline medical condition, intraoperative anesthesia, and surgery-related data were obtained. The primary outcome was postoperative acute kidney injury (AKI) defined according to the Kidney Disease Improving Global Outcomes criteria. The following machine learning techniques were used: decision tree, random forest, extreme gradient boosting, support vector machine, neural network classifier, and deep learning. The performance of these techniques was compared with that of logistic regression analysis regarding the area under the receiver-operating characteristic curve (AUC). During the first postoperative week, AKI occurred in 770 patients (38.3%). The best performance regarding AUC was achieved by the gradient boosting machine to predict the AKI of all stages (0.78, 95% confidence interval (CI) 0.75–0.80) or stage 2 or 3 AKI. The AUC of logistic regression analysis was 0.69 (95% CI 0.66–0.72). Decision tree, random forest, and support vector machine showed similar performance to logistic regression. In our comprehensive comparison of machine learning approaches with logistic regression analysis, gradient boosting technique showed the best performance with the highest AUC and lower error rate. We developed an Internet–based risk estimator which could be used for real-time processing of patient data to estimate the risk of AKI at the end of surgery.

This study was performed to assess whether intubation is more difficult in obese patients and to assess the ability of a new index: the ratio of the neck circumference to thyromental distance (NC/TM), to predict difficult intubation in obese patients. The incidence of difficult tracheal intubation in 123 obese (BMI≥27.5 kg m⁻²) and 125 non-obese patients was compared. Difficult intubation was determined using the intubation difficulty scale (IDS≥5). The NC/TM ratio was calculated and its ability to predict difficult intubation in obese patients was compared with that of established predictors including high BMI, the Mallampati score, the Wilson score, NC, width of mouth opening, sternomental distance, TM, and a previous history of difficult intubation. Difficult intubation was more frequent in obese patients than in non-obese patients (13.8% vs 4.8%; P=0.016). Multivariate analysis revealed that the Mallampati score, the Wilson score, and NC/TM independently predicted difficult intubation in obese patients. Among these three indices, NC/TM showed the highest sensitivity and a negative predictive value, and largest area under the curve on an ROC curve. Difficult intubation was more common in obese patients and the NC/TM was a better method for predicting difficult intubation than other established indices.

EDITOR’S PERSPECTIVE: What We Know about This Topic: IV anesthesia may impair anticancer immunity less than volatile anesthesia and therefore reduce recurrence risk What This Article Tells Us That Is New: In a large propensity-matched retrospective cohort analysis, the authors compared total IV and volatile anesthesia for breast cancer surgery Recurrence hazard was similar with each approach Selection of IV or volatile anesthesia should be based on factors other than cancer recurrence Background: The association between type of anesthesia used and recurrence of cancer remains controversial. This retrospective cohort study compared the influence of total IV anesthesia and inhalation anesthesia on the primary outcome of recurrence-free survival after breast cancer surgery. Methods: The authors reviewed the electronic medical records of patients who had breast cancer surgery at a tertiary care teaching hospital between January 2005 and December 2013. The patients were grouped according to whether IV or inhalation anesthesia was used for surgery. Propensity score matching was used to account for differences in baseline characteristics. Kaplan–Meier survival curves were constructed to evaluate the influence of type of anesthesia on recurrence-free survival and overall survival. The risks of cancer recurrence and all-cause mortality were compared between each type of anesthesia. Results: Of 7,678 patients who had breast cancer surgery during the study period, data for 5,331 patients were available for analysis (IV group, n = 3,085; inhalation group, n = 2,246). After propensity score matching, 1,766 patients remained in each group. Kaplan–Meier survival curves showed that there was no significant difference in recurrence-free survival or overall survival between the two groups, with 5-yr recurrence-free survival rates of 93.2% (95% CI, 91.9 to 94.5) in the IV group and 93.8% (95% CI, 92.6 to 95.1) in the inhalation group. Inhalation anesthesia had no significant impact on recurrence-free survival (hazard ratio, 0.96; 95% CI, 0.69 to 1.32; P = 0.782) or overall survival (hazard ratio, 0.96; 95% CI, 0.69 to 1.33, P = 0.805) when compared with total IV anesthesia. Conclusions: The authors found no association between type of anesthesia used and the long-term prognosis of breast cancer. The results of this retrospective cohort study do not suggest specific selection of IV or inhalation anesthesia for breast cancer surgery.

Acute kidney injury (AKI) after pediatric cardiac surgery is associated with high morbidity and mortality. Modifiable risk factors for postoperative AKI including perioperative anesthesia-related parameters were assessed. The authors conducted a single-center, retrospective cohort study of 220 patients (aged 10 days to 19 years) who underwent congenital cardiac surgery between January and December 2012. The incidence of AKI within 7 days postoperatively was determined using the Kidney Disease: Improving Global Outcomes (KDIGO) criteria. Ninety-two patients (41.8%) developed AKI and 18 (8.2%) required renal replacement therapy within the first postoperative week. Among patients who developed AKI, 57 patients (25.9%) were KDIGO stage 1, 27 patients (12.3%) were KDIGO stage 2, and eight patients (3.6%) were KDIGO stage 3. RACHS-1 (Risk-Adjusted classification for Congenital Heart Surgery) category, perioperative transfusion and fluid administration as well as fluid overload were compared between patients with and without AKI. Multivariable logistic regression analyses determined the risk factors for AKI. AKI was associated with longer hospital stay or ICU stay, and frequent sternal wound infections. Younger age (3 g/dl) from preoperative level on POD1 was entered into the multivariable analysis, it was independently associated with postoperative AKI (OR, 6.51; 95% CI, 2.23–19.03 compared with no increase). This association was significant after adjustment with patient demographics, medication history and RACHS-1 category (hemoglobin increase >3g/dl vs. no increase: adjusted OR, 6.94; 95% CI, 2.33–20.69), regardless of different age groups and cyanotic or non-cyanotic heart disease. Prospective trials are required to evaluate whether correction of preoperative anemia and prevention of hemoconcentration may ameliorate postoperative AKI in patients who underwent congenital cardiac surgery.

BACKGROUND: Supraclavicular (SC) and infraclavicular (IC) brachial plexus block (BPB) are commonly used for upper extremity surgery. Recent clinical studies have compared the effect of SC- and IC-BPB, but there have been controversies over spread of sensory blockade in each of the 4 peripheral nerve branches of brachial plexus. METHODS: This study included a systemic review, using the Medline and EMBASE database from their inceptions through March 2016. Randomized controlled trials (RCTs) comparing SC- and IC-BPB were included. The prespecified primary outcome was the incidences of incomplete sensory blockade in each of the 4 terminal nerve branches of brachial plexus. Secondary outcome included the incidence of successful blockade, performance time, onset of sensory block, duration of analgesia, and complication rates. RESULTS: Ten RCTs involving 676 patients were included. Pooled analyses showed the incidence of incomplete block at 30 minutes in radial nerve territory was significantly higher in IC-BPB, favoring SC-BPB (risk ratio 0.39; 95% confidence interval [0.17–0.88], P = .02, I ² = 0%). However, subgroup analysis according to the number of injections of IC-BPB showed that double or triple injections IC-BPB yielded no difference in the incomplete radial block. Furthermore, the incidence of incomplete ulnar block at 30 minutes was significantly lower in IC-BPB when using double or triple injection IC-BPB. There was no difference in the secondary outcomes between SC- and IC-BPB groups, with the exception of complication rates. The incidence of paresthesia/pain on local anesthetic injection, phrenic nerve palsy, and Horner syndrome was significantly higher in the SC group, favoring IC-BPB. CONCLUSIONS: This meta-analysis demonstrated that IC-BPB showed a significantly high incidence of incomplete radial nerve sensory block at 30 minutes, which may be avoided by double or triple injection. Furthermore, IC-BPB with multiple injection technique showed significantly lower incidence of incomplete ulnar block than SC-BPB. There were no differences in the incidence of successful blockade, block onset, and duration of analgesia between SC- and IC-BPB. Procedure-related paresthesia/pain and adjacent nerve-related complications were more frequent in SC-BPB. However, because of the small sample size, publication bias remains a concern when interpreting our results. Further studies with sufficient sample size and reporting large number of outcomes are required.