Our results suggest that the use of LMA compared to ETT is independently associated with a reduced need for postoperative invasive ventilation in VLBW infants with ROP undergoing laser photocoagulation. Our data show that the odds of postoperative mechanical ventilatory support are at least three times lower (based on propensity matching) with LMA than with ETT.

With positive pressure ventilation, supraglottic devices will always have a potential risk of aspiration of gastric content; thus, ETT is a safer option for securing the airway. However, there are several papers proving that LMA is a safe alternative in several clinical scenarios, and also reduces perioperative complications. In a former study comparing LMA, cuffed ETT and uncuffed ETT, it was reported that the use of LMA reduced the incidence of postoperative laryngospasm and coughing in infants aged between 1 to 24 months undergoing lower abdominal surgery.10 Drake-Brockman and colleagues found in a randomised control trial that the incidence of perioperative respiratory adverse events were three times higher in patients receiving ETT than in patients receiving LMA in an infant population under 1 year of age who underwent elective general anaesthesia.8 Another study comparing LMA and ETT in 70 neonates demonstrated that extubation was longer in patients who received ETT and had more postoperative respiratory adverse events.11 Previously, a small case series study suggested that LMA airway management was feasible in patients who underwent ROP treatment.12 Besides our results being consistent with the above-mentioned findings on LMA, our study also suggests that LMA may be a safe alternative to ETT in VLBW infants, providing safer care options for an especially vulnerable patient population. It is conceivable that LMA is less airway irritative and less likely to reactivate BPD, a common comorbidity in VLBW infants, thus reducing the risk for postoperative mechanical ventilatory support.

Previous studies have described the patterns of reintubation and its relationship to BPD in VLBW cohorts. In a cohort of infants with birth weight ≤1250 g, 47% were reintubated during hospitalisation, one-third of which for non-respiratory reasons, 5% for elective procedures.16 The same group has reported that time to reintubation independently modulated the odds of the combined outcome of death/BPD.17 Furthermore, others have found that infants reintubated within 7 days after extubation were significantly more likely to die or require prolonged respiratory support and hospitalisation.18 All these findings emphasize the notion that it is best to avoid invasive mechanical ventilation in VLBW infants who were once successfully extubated before. A ROP intervention related postoperative mechanical ventilation is not only a risk factor for further complications in an already compromised patient population, but also a financial burden for healthcare institutions as it does not only require a higher level of care, but also prolongs the length of stay of these infants in the NICU.

Further physiological considerations also support the use of LMA. Disma and colleagues stated that difficult intubation occurred in 5.8% of neonatal or infant anaesthesia, which was accompanied with a longer desaturation or bradycardia in a large amount of the cases.19 Moreover, a review suggested that hypoxia, hypotension and anaemia were the dreaded triad that increased morbidity and mortality in neonates undergoing anaesthesia.20 A recent review study pointed out that lately there is a shift from general anaesthesia towards other types of anaesthesia in preterm infants undergoing laser photocoagulation.21 The main reason for this phenomenon was to decrease adverse events associated with extubation postoperatively. These findings suggest that the use of LMA instead of ETT can increase patient safety by reducing the chance of possible adverse events associated with ETT and eliminating the use of muscle relaxants for ETT insertion in VLBW infants.

Although our data show advantages of LMAs, the device has some limitations and disadvantages.

Firstly, according to the manufacturer, the lowest limit of use of LMA is a body weight of 2000 g. In our experience, LMA can be used successfully even in patients weighing less than 1500 g. However, if the patient was too small, an anatomical mismatch might cause leakage, and the LMA might not seal properly, although we followed the instructions of the manufacturers, and LMA cuff pressure was checked using manometry. In these cases, when a leakage occurred and could not be solved in any other way, it led to conversion to ETT. Another possible issue is the problem of dead space due to the discrepancy between body weight of the infant and the size of the LMA, which can potentially lead to hypercapnia. Secondly, respiratory adverse events such as desaturation or hypercapnia are common in neonates.22 LMA, being a less invasive airway device, can be dislocated easily from its correct position, even with a slight touch by the ophthalmologist, resulting in respiratory compromise. While the anaesthesiologist is performing an intervention, ophthalmologic treatment has to be stopped due to the small body size of the patients and the fact that both the ophthalmologic intervention and the airway management take place at the head. Thus, any LMA dislocation can lengthen the entire procedure. It is important to point out that LMA airway management is not suitable for standard thoracic or abdominal surgeries in premature neonates; however, it may be considered in other minor surgical interventions besides laser photocoagulation, for example in inguinal hernia repair.

Our study has several strengths and limitations. According to our current understanding, this has been the largest study investigating the effect of LMA during general anaesthesia in VLBW infants. During the entire 7.5-year study period laser photocoagulations were performed by a single paediatric ophthalmologist, and anaesthesia was provided by the same team consisting of senior paediatric anaesthesiologists using uniform equipment and protocol for general anaesthesia. Due to the long study period, we could enrol a significant number of patients, which granted us the possibility to perform different and complex statistical analyses, all of which yielded very similar results.

Naturally, there are limitations to this study that should be noted. As an observational retrospective study, randomisation of patients was impossible. Airway management as a key determining factor on the outcome was entirely left to the discretion of the attending anaesthesiologist. Therefore, the choice of airway management was determined by whether the infant had been previously intubated and was dependent on mechanical ventilation, whether the infant had received any type of respiratory support (HFNC, NIV) or the body weight of the infant at the time of the laser photocoagulation. This can result in a selection bias, which we attempted to overcome by using complex statistical methods, multiple regression analysis and propensity score matching, and both methods yielded similar results. A further limitation may be that we did not collect and compared data other than respiratory ones, however the immediate benefit and safety of use of LMA instead of ETT expected in respiratory outcomes. Of note, we could not control for BPD as a comorbidity, as the majority of our patient were less than 36 weeks old at the time of laser photocoagulation, hence the BPD diagnosis could not be established at that point. However, we used the cumulative days of mechanical ventilation as a proxy measure, which predicts BPD and home oxygen need.23 Finally, we acknowledge that generalisability can possibly be a limitation of our study, as in some centres infants are not admitted to the operating theatre for laser photocoagulation treatment under general anaesthesia. However, our results may be relevant for the anaesthesia of other minor procedures in preterm infants.