After ethics committee approval (PR AG 129/2014), a prospective observational study was carried out at Vall d'Hebron University Hospital, a tertiary hospital in Barcelona, Spain, in accordance with the Standards for Reporting of Diagnostic Accuracy (STARD), including all consecutive adult patients undergoing total thyroidectomy during a 12-month period. All patients understood the various aspects of the study and gave written informed consent. Patient characteristics recorded were age, sex, body mass index, ASA grade, and reason for surgery, among others.
Intraoperative neuromonitoring: Neural monitoring was performed once the carotid neurovascular bundle had been identified and throughout the surgical dissection process to ensure functional integrity of the nerve. We used Avalanche XT® (Dr Langer Medical GmbH, Waldkirch, Germany) equipment fitted with paired electrodes on the tracheal tube, in contact with the vocal cords (7-10 mm above the upper edge of the tube). Intermittent neuromonitoring was performed in each patient according to the international standard guidelines for electrophysiological monitoring20. The VN and RLN were stimulated following the four-step technique (V1: initial vagal stimulation, prior identification and dissection of the VN, R1: initial stimulation of the RLN, R2: post-dissection RLN stimulation, V2: post-dissection vagal stimulation). Vagal stimulation confirmed proper functioning of the entire neuronal circuit, while also avoiding false negatives through stimulation of an injured distal RLN at the lesion site. All signals were recorded on a monitor and collected for evaluation. Vagal and RLN stimulation was programmed at 3 mA and 1 mA, respectively. The event threshold was set to 70-100 uV, and increased to 200 uV, to avoid false-positive events appearing on the monitor due to spontaneous low-level respiratory waves \< 100 uV. Electromyographic signals (latency and amplitude) as well as biphasic and triphasic waves were displayed on the monitoring screen. Lack of signal occurred when the monitor showed low-amplitude waves, disturbance of the isoelectric line, multiple waves of variable and scant amplitude, interference or substantial reduction in electromyography response (\< 100 uV), and a high-pitched warning sound.
Vocal cords were assessed using VS (Laryngeal Strobe 9400, Pentax Medical, USA) by a single experienced phonologist blinded to the IONM results. The VS exam was performed preoperatively and repeated one week after surgery. The diagnosis of palsy (absence of motion) was made on the basis of an observation of asymmetrical laryngeal motion. Patients with postoperative VCP began voice therapy and were monitored periodically to assess their progress.
Statistics:
The main aim of the present study was to identify independent risk factors for VCP, including those related with pre-dissection IONM values. Continuous variables are expressed as the mean (standard deviation) or the median and interquartile range (i.q.r.) prior to verification of normality by the Kolmogorov-Smirnov test, and categorical variables are expressed as the absolute value (percentage).
The IONM results (response amplitude in uV and latency in ms) were compared with VS findings and classified as true positives (TP) when IONM values were abnormal and VCP was confirmed by VS, as true negatives (TN) when IONM values were normal and vocal cord mobility was normal on VS, as false positives (FP) when IONM values were obviously abnormal but vocal cord mobility was normal on VS, and as false negatives (FN) when IONM values were normal but VS revealed VCP. These IONM results were used to calculate sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), Youden incidence, with the respective 95% confidence interval (CI).
Perioperative variables included sex, age, BMI, ASA grade, neoplasm diagnosis, lymphadenectomy, and surgery. The bivariate relationship between each factor and the VCP results was analysed using an independent sample t test or the Mann-Whitney U-test, as appropriate, for continuous variables, and using the χ2 or Fisher exact test for dichotomous categorical variables. Receiver operating characteristic (ROC) curves were also plotted to determine the best cut-off points for pre-dissection latency and response amplitude for both VN and RLN to predict postoperative palsy.
These cut-off points were then included in the forward stepwise multivariate logistic regression to identify independent risk factors for VCP. Candidate variables for the logistic regression included sex, age, BMI, neoplasm diagnosis, lymphadenectomy, surgery duration, American Society of Anesthesiologists (ASA) classification, and IONM-related predictive factors such as pre-dissection VN latency, pre-dissection VN response amplitude, pre-dissection RLN latency, and pre-dissection RLN response amplitude.
The odds ratio (OR) and 95% CI were also calculated for all exposure factors. In all cases, the level of significance was set at less than 0.05. The statistical analysis was performed using the software package SPSS version 23.0 (IBM, Armonk, New York, USA).
