Endotracheal intubation is a frequent procedure in the Neonatal Intensive Care Unit (NICU). Although neonatal awake intubation is not recommended, except for emergency situations, no consensus exists to date supporting the use of a drug or a drug combination in neonates 1 . Clinical studies addressing premedication before endotracheal intubation in neonates include randomised, controlled trials vs placebo 2 3 or comparison of different regimens 4 5 6 7 and observational studies 8 9 10 11 12 . Based on these, experts have recommended the use of a short onset opioid 1 13 14 . Fentanyl is the most studied synthetic opioid in this context 6 8 9 11 . In contrast with the scarce data regarding sufentanil use in neonates 15 and the latest expert recommendations 1 , French neonatologists use sufentanil more frequently than fentanyl for intubating neonates either in the NICU 16 or in the delivery room 17 . An increase in sufentanil use in German NICUs has also been recently reported 18 . Some trials have demonstrated the interest of associating a short onset opioid with a muscle relaxant 4 6 11 19 . However muscle relaxants are rarely used in France 16 17 20 . Randomised controlled trials (RCT) usually provide reliable evidence. However, real practice does not always correspond to bedside conditions 16 . As such, observational studies are frequently beneficial in providing data from actual clinical practice 21 22 . The latter studies are necessary to objectively evaluate the consequences of actual practices. In February 2007, we implemented and prospectively evaluated a protocol in our NICU for anaesthesia before endotracheal intubation in non-life threatening situations combining atropine, sufentanil and atracurium. This protocol was elaborated in collaboration with the paediatric anaesthesiologists from our institution and written after a review of the existing published evidence at that time. Our objectives were to describe the real conditions of the procedure, to document tolerance and to collect adverse events in a group of premature infants born under 32 weeks of gestational age (GA).

This observational study showed that in very premature infants the association of atropine, sufentanil and atracurium provided good intubation conditions as rated by the operator and a 74% success rate for first attempt. No significant changes in heart rate, mean arterial blood pressure or pulse oxymetry recorded at predefined time points were observed. Desaturations below 80% lasting for more than 60 consecutive seconds were however frequent (51%) and a significant increase in TcPCO₂ was observed within the 30 minutes following the first drug injection.

The use of atropine before endotracheal intubation is controversial 1 26 . In a recent observational study, 25% of neonates undergoing tracheal intubation in level 3 units experienced bradycardia between 100 and 60 bpm and 21% experienced bradycardia < 60 bpm 27 . In this study atropine use decreased the frequency of bradycardia between 100 and 60 bpm, but not the frequency of bradycradia < 60 bpm. In addition recent results are reassuring regarding atropine use in neonates 28 . Considering these observations and the significance of vagal tone in premature infants 29 we still include atropine as part of our premedication before endotracheal intubation although strong supporting evidence is still lacking.

Opioids have been the most studied drugs for neonatal intubation, either alone or combined to a muscle relaxant 2 4 5 6 8 9 11 19 . The use of morphine is arguable because of its delayed onset of action 1 and studies showing that it may be less effective than other drug regimens for intubating neonates 5 7 . Rapid onset, short acting opioids used as anaesthetics for tracheal intubation include alfentanil, fentanyl, sufentanil and remifentanil. Sufentanil has shown a faster elimination and a shorter site effect than fentanyl in some circumstances 30 . Thus, it could be the preferred drug in premature infants who usually have altered elimination capacities. Since some recent European studies have reported a frequent use of sufentanil in preterm neonates 16 17 or a trend to increasing its use 18 , we consider that data on tolerance and safety are necessary. Remifentanil also seems to be a promising opioid for tracheal intubation either alone 10 or in association with other drugs 5 . Nonetheless, one of its drawbacks is the difficulty to determine its appropriate dose 31 .

In 6 other studies associating a synthetic opioid with a muscle relaxant, the first attempt success rate was below 74% (range 35% to 67%) in 4 studies 4 9 11 19 and over 74% (77% and 90%) in 2 studies 2 8 . In a recently published observational study, the overall success rate for the first intubation attempt in neonates across 5 academic level III centres was 44% 32 . For residents, this rate fell to 20.3%. We consider that the rate we observed in our study is rather high, especially among junior operators. The observed success rate of first attempt in a population where 75% of infants had a weight below 1000 g at the time of intubation could be explained, at least partially, by the satisfactory conditions for upper airway visualization as assessed by the operator using the quality of sedation scale. Studies exploring the combination of a fast acting opioid and a muscle relaxant found that experienced personnel had a higher success rate than inexperienced personnel 8 11 . We did not find this difference. This could be explained by a lack of power so we cannot conclude on the efficacy of our regimen on this outcome.

Saturation values ≤ 80% were frequently observed in other studies although usually of shorter duration than in ours 9 11 27 . It should be noted that 99% of infants in Venkatesh et al.’s study 27 and all infants in Lemyre et al.’s study 11 were preoxygenated with 100% FiO₂. Dempsey et al. 9 recommended to obtain a saturation of at least 95% before starting the procedure. In our study the median FiO₂ and the median SpO₂ values one minute before the first drug injection were respectively 37% and 94%. The absence of systematic preoxygenation, of target SpO₂ recommendation and of positive end-expiratory pressure on our ventilation bags may explain the high incidence of desaturations in our patients. This advocates for proper positive end expiratory pressure use and preoxygenation in very low birth weight infants who are known to have reduced residual functional capacity as compared to older infants 23 . This is probably even more critical when a muscle relaxant is used resulting in additional lung derecruitment. On the other end, oxygen toxicity for the developing eye, brain and lung has been documented 33 and the risk/benefit ratio of preoxygenation on long-term outcome is unknown. Based on our experience, we believe that preoxygenation performed through a face-mask connected to the ventilator circuit should be used so that an effective positive end-expiratory pressure can be provided and peak inspiratory pressure can be controlled; we also believe that a target SpO₂ value of 95% should be aimed for.

The initial (M5) increase in TcPCO2 we observed is probably due to the laryngoscopy during which no respiratory movement exists due to muscle relaxant use. However we were surprised to observe the persistence of hypercapnia 10, 15 and 30 minutes after the first drug injection. Possible explanations for this may include a major derecruitment at the time of laryngoscopy resulting in collapsed compliance once mechanical ventilation is started or resumed, inappropriate ventilator settings after intubation and/or excessive permissive hypercapnia as illustrated by the similar values observed at the different time points for delta pressure and for respiratory rate. Another possible explanation is a decrease in peripheral perfusion (especially cutaneous) in spite of maintained central perfusion as illustrated by stable blood pressures, resulting in an unreliable transcutaneous PCO₂ measurement. Unfortunately no arterial blood gas was sampled at that time to confirm or not this hypothesis. Since no other study explored variations in TcPCO₂ or PCO₂ during very preterm infants’ intubation we might just have observed a phenomenon that was so far ignored. Future studies should explore gas exchange in the immediate period following intubation.

Thoracic rigidity could possibly be reduced by the injection of the muscle relaxant before the opioid. This sequence can be controversial since paralysis prior to anaesthesia may not be appropriate 34 . However, several publications have reported the injection of a muscle relaxant before an opioid 5 9 based on the hypothesis that obtaining adequate paralysis before the peak opioid effect appears might prevent thoracic rigidity. Therefore, we currently inject atracurium before sufentanil in our unit.

Sedatives such as midazolam or propofol are frequently used as premedications for endotracheal intubation 16 20 21 22 . Future studies should aim at comparing sedatives with the combination of an opioid and a muscle relaxant. The monitoring of vital signs including desaturations and PCO₂ should be included in future research.

Our study has several limitations. It is observational and has no comparative group. The collection of data was stopped after 7 months because caregivers were satisfied with this drug combination, this resulted in a limited number of studied cases. We did not record the total number of intubations performed during the study period. Therefore we don’t know if the studied population is representative of all very premature infants intubated in our NICU. Especially we ignore if similar observations would be obtained in emergency intubations. The median postnatal age in our study was 10 days, which is older than the age at intubation observed in other studies 5 10 21 22 . In our unit premature infants born below 28 weeks GA receive prophylactic surfactant in the delivery room. Thus, most intubations performed in very premature infants in our NICU are re-intubations. This explains the age at intubation in our population. Our observations might then not be extrapolated to recently born premature infants. The use of a muscle relaxant precluded the use of any pain scale since body movements or facial expression cannot be evaluated during paralysis. Skin conductance might have provided information on analgesic efficacy of our regimen 35 . However, at the time of our study this technique was not widely in use and the presence of atropine, an anticholinergic drug, in our regimen precluded the use of skin conductance. We did not record time to spontaneous limb and chest movements’ recovery. This information would have helped in the comparison with other opioid plus muscle relaxant regimens. We did not either record the time from decision to intubate to tube fixation which is of interest in an emergency situation. Our definition of the duration of intubation procedure does not provide the time of glottis exposure as others have reported 22 . Interpretation of this parameter is therefore difficult.

Although the combination of atropine, sufentanil and atracurium for intubating very premature infants offers good intubation conditions without any significant change in patients’ heart rate and blood pressure, proper preoxygenation and alternative drugs to muscle relaxant avoiding paralysis and subsequent lung derecruitment need to be further explored in order to improve tolerance and conditions of neonatal intubation.

ETT: Endotracheal tube; GA: Gestational age; IQR: Interquartile range; MAP: Mean arterial blood pressure; NICU: Neonatal intensive care unit; PCO2: Partial pressure of carbon dioxide; RCT: Randomised controlled trial; SpO2: Pulse oxymetry.

All authors have no conflict of interest concerning this study

XD designed the study, performed statistical analyses, wrote the first draft, reviewed and revised the manuscript. SD participated to data collection, reviewed and revised the manuscript. PD contributed to statistical analyses, reviewed and revised the manuscript. SB contributed to the study design, reviewed and revised the manuscript. GD contributed to the study design, reviewed and revised the manuscript. LC reviewed and revised the manuscript. RC contributed to study design and statistical analyses, reviewed and revised the manuscript. All authors approved the final manuscript as submitted.