The learning curve for intraoperative neuromonitoring of the recurrent laryngeal nerve in thyroid surgery

The introduction of neuromonitoring to thyroid surgery made it possible not only to easily identify the RLN but also to understand the physiology and mechanisms of RLN injury [1, 6, 13‐16]. Since neuromonitoring of the RLN and the external branch of the superior laryngeal nerve was standardized, the use of this technique has become increasingly widespread [1, 17]. In 2011, the first conference of the Polish Research Group for Neuromonitoring of the Polish Club of Endocrine Surgeons was held and the need to use IONM in at least some selected thyroid operations was agreed upon [18]. Since then, there has been a slow but steady growth of interest in this technique in Poland, and the number of centers performing operations with IONM continues to grow [11].

However, the introduction of IONM to thyroid surgery requires previous experience in thyroid surgery, as well as training, both theoretical and practical, in neuromonitoring [8, 9, 11, 19]. It also requires independent experience of a number of thyroid operations with IONM to improve one’s surgical technique and to reduce the number of postoperative complications [20, 21]. Pragacz et al. also reported how important this kind of training was before deploying IONM in everyday clinical practice [11, 19].

Despite solid preparation for the introduction of neuromonitoring in thyroid surgery, the authors did not entirely avoid problems during the initial stage of IONM use. These were primarily technical problems, which arose not from anything difficult about the use of the equipment but from a lack of experience with it. In the first year working with IONM (2012), problems occurred in 12.87 % of the thyroid operations, but in the next year (2013) there was already a sharp drop to 4.3 %, which was maintained at nearly the same level (4.6 %) in 2014 (p = 0.0219). The most common technical problem throughout the study period was rotation of the electrodes on the endotracheal tube in relation to the vocal cords. Especially during the initial phase of IONM use, many of the technical problems require knowledge of the algorithms for monitoring tube placement during intubation and problem solving in instances of signal loss. The International Nerve Monitoring Study Group recommendations for tube placement and for troubleshooting were adhered to [1]. To avoid inadequate endotracheal positioning, patients were intubated with the largest endotracheal tube possible to optimize electrode contact with the vocal cords—in women this was usually size 7.0, and in men size 7.5. After intubation and patient positioning, but before the start of surgery, preoperative tube position verification testing was carried out. We usually used the “tap test,” and in most patients this test was accurate. Whenever there were any problems or doubts, a repeat laryngoscopy was done. At the beginning of each operation, predissection vagal stimulation was the most important test to confirm overall system function. Moreover, knowledge of the standards for interpreting intraoperative loss of signal is essential when using IONM. In cases where EMG activity was not present or the amplitude was below 100 uV, the laryngeal twitch was assessed. If the laryngeal twitch was present, we suspected a malfunction on the recording side; usually, it was due to malpositioning of the endotracheal tube electrodes. Malpositioning usually involves tube rotation but may also entail inadequate tube depth. In all cases of malpositioning, a maneuver correcting the endotracheal tube placement was done by the surgeon or anesthesiologist. In two operations, we observed that the grounding electrode had slipped out of place due to perspiration. To prevent this problem, we always affixed this electrode very carefully at the beginning of surgery. We used contralateral vagal assessment in instances of signal loss, especially in bilateral surgery. If the contralateral EMG signal from the vagus nerve was absent, we checked for rotation of the tube. If the laryngeal twitch was absent or if we had the contralateral EMG signal, we suspected true loss of signal. We did not observe any intraoperative RLN stimulation errors.

A similar proportion of technical problems (10 %) was described by Dionigi et al. in those authors’ first 152 operations using IONM, showing a lower rate of problems as the number of procedures performed rose. They reported that the rates of problem-free use of IONM were 80 % in the first 50 operations, 92 % in the next 50, and 98 % in the last 50 (p < 0.05) [9]. Pragacz et al. reported an even higher rate of technical problems—24 %—in the initial period of IONM use (50 thyroid procedures); however, in the next 50 operations, the rate dropped to 8 % (p = 0.029) [11]. The results of the current study and those reported by other authors indicate that in order to master the basic techniques of neuromonitoring and minimize technical problems, 50 to 100 thyroid procedures with IONM have to be performed [9, 11].

Numerous publications about neuromonitoring in thyroid surgery show a high rate of RLN identification with IONM [1, 6, 13, 22‐24]. At our center, effective RLN identification with IONM in the first year was quite high: 92.11 %. In the initial phase of IONM use, the main problems causing difficulty in nerve visualization were inexperience in mapping on the one hand and technical problems on the other. In the later years of the study period, along with the decrease in technical problems, there was an increase in nerve identification, by 3 % in 2013 and by another 4 % in 2014. After 3 years of working with IONM (more than 200 thyroid operations), this increase was at the level of statistical significance (p = 0.022). None of the patients presented postoperative palsy. Even when there was difficulty finding the nerve (R1), we always checked the vagus nerve (V2) after a lobe resection. A positive signal from the vagus nerve was particularly useful before resecting the second lobe of the thyroid in bilateral thyroid operations. Pragacz et al. noted a statistically significant increase in nerve identification (p = 0.006) after only 50 monitored thyroid operations and that 84 % rate increased to 99 % after the next 50 procedures [11, 19].

The question therefore arises whether we are able to significantly reduce the frequency of RLN paralysis as we learn the method of IONM. In the current study, within 3 years, there was a decline in transient paralysis from 2.63 % of the RLNs exposed to the risk of damage in 2012 to 0.83 % in 2014; this decrease, however, was not at the level of statistical significance (p = 0.275). Permanent paralysis occurred only in two cases (0.83 % of the RLNs exposed to the risk of damage) during the first 101 thyroid operations performed during the initial stage of IONM implementation; it did not occur at all during the remaining study period (p = 0.505). There was no bilateral paralysis after procedures with neuromonitoring throughout the observation period. Jonas et al. reported virtually identical results in terms of the RLN injury rate during the IONM learning period. In those authors’ observations, transient paralysis occurred in 2.3 % and permanent paralysis in 0.8 % of the RLNs at risk; as in the present study, those figures concerned only the first of 3 years working with IONM [10]. A similar proportion of transient paralysis (2.6 %) and lack of permanent injury was reported by Dionigi et al. in the first 152 thyroid surgeries using IONM [9]. On the other hand, Pragacz et al. reported a statistically significant decrease in transient RLN paralysis between the initial learning period and the team’s later operations with IONM. In those authors’ study, in the introductory period transient paralysis occurred in 6.5 % of the RLNs at risk; this was drastically reduced to 1 % in the next 50 thyroid operations [11, 19]. While it is fairly easy to learn the techniques of IONM and RLN identification, it takes a long period of observation or a very large number of thyroid procedures to observe a statistically significant drop in the RLN paralysis rate, hence the tremendous value of multicenter studies [25, 26].

A study by Dralle et al. discussing the direction of changes in thyroid surgery is worth mentioning here. The authors noted that, over the years, there have been changes in resectional strategy, with an increase in the number of total thyroidectomies and a decrease in subtotal resection, but the rates for postoperative hypoparathyroidism and vocal cord palsy have decreased [26]. Despite the relatively small number of thyroid procedures performed at our center, we also observed a statistically significant increase in the ratio of radical thyroid procedures (p = 0.004). There was a slight decrease in the rate of postoperative palsy, but it was not statistically significant. These results at a single small department are similar to those observed by Dralle et al. in their multicenter study. There are many different factors influencing thyroid operation strategy, but we must emphasize that at our center these results were achieved after the introduction of IONM.

Regarding the value of IONM in decreasing the number of cases of postoperative RLN paralysis, a study by Alesina et al. is worth quoting. They compared thyroidectomies performed by inexperienced surgeons under the supervision of a consultant surgeon without IONM with thyroidectomies performed without experienced oversight but with the use of IONM. The study revealed that the total number of injuries in the two groups was similar—2.6 and 2.7 %, respectively. This proved how good IONM is, offering a high degree of safety during thyroid surgery even in the hands of surgeons with little prior experience [27].

An objective assessment of the value of IONM is an analysis of its sensitivity, specificity, positive and negative predictive value, and accuracy. In the present study, the sensitivity of IONM was 71.4 % in 2012; this increased to 100 % (p > 0.05) after the first 101 thyroid procedures. The positive predictive value (PPV) was 62.5 % in 2012 and 75 % in the 2013–2014 period (p < 0.05). Despite the statistically significant reduction in the number of technical problems (p = 0.0185), the increases in sensitivity and PPV were not statistically significant (p = 0.05). These PPV values did not give us grounds to perform “stage thyroidectomies” in either period working with IONM. Pragacz et al. reported an even lower PPV in the first period of IONM use: 55.6 % after the first 50 thyroid surgeries performed [11, 19]. In the literature, positive predictive value ranges from 9.2 to 92.1 %; it was only with the standardization of neuromonitoring that we find PPV above 50 % at most centers during the introductory phase of IONM use and increasing with experience [1, 28].

In contrast, in most reports, negative predictive value (NPV) is quite high from the beginning of IONM use, ranging from 92 to 100 % [1, 8]. In the present study, NPV in the initial period was 98 %, rising to 100 % in the second stage of IONM use. However, this increase was not statistically significant (p = 0.05). The increase in accuracy was the only one that was statistically significant after the first 101 thyroid procedures with IONM, increasing from 97.4 % in 2012 to 99.6 % in the later phase of the study period (p = 0.049).

In the context of learning curves, the duration of thyroid surgery is often given, with the expectation that as the surgical team gains experience with IONM the duration of the procedures will decrease. Such a dependency was reported by Dionigi et al. and Pragacz et al. [9, 11]. The present study did not confirm a correlation between the duration of surgery and the length of time working with IONM. Rather, the operating time was different each year we worked with IONM: It was 105 min in 2012, 115 min in 2013, and 110 min in 2014 (p < 0.05). Every thyroid operation is different, and the procedure duration is determined by the preoperative diagnosis, the type of operation performed (primary or reoperation), and the experience and individual skills of the surgeon. Interestingly, in our analysis it emerged that the only factor that significantly influenced the time of monitored thyroid surgery was the patients’ body weight (p = 0.004). As the BMI increased, so did the duration of surgery, which can be explained by more difficult access to both the vagus nerve and the RLN in overweight patients.

IONM seems to be an easy technique to introduce into everyday use in clinical practice, but full exploitation of this method—among other things, the decision to undertake “stage thyroidectomy”—requires good preparation before implementation, followed by the execution of at least 50–100 thyroid procedures with IONM. Duclos et al. discussed the individuality of the learning curve, specifying how very different it was among three surgeons performing monitored thyroid surgery for the first time [20].

Finally, it must be emphasized that the surgeons’ rapid learning process with IONM would not have been possible without good cooperation with the anesthesiology team, which has also been noted by other authors [1, 8, 9].