With the approval of our institutional ethics committee and after having obtained written informed consent, patients were recruited for this prospective, randomized clinical study (German Clinical Trial Register number DRKS00000760). Inclusion criteria were: age >18 years and scheduled elective surgical intervention in the supine position with predicted anaesthesia duration between 60 and 180 minutes. Exclusion criteria were: BMI >35 kg/m
2, ASA status III or higher, known risk of aspiration, known or predicted difficult airway. The patients were assigned to their groups with a computer-generated randomisation list (
http://www.randomizer.org). The sealed envelope method was used for blinding.
Leak pressure (LP) was the primary endpoint. Secondary endpoints were speed of insertion, insertion success rates, fibreoptically assessed in situ position, dynamic airway compliance and signs of airway morbidity.
Airway management and ventilation
Two senior anaesthesia registrars (SC and TG) skilled in placing SGA performed all cases. Although the recommended insertion techniques of the three SGA are quite similar the investigators were required to perform a minimum of 15 insertions with all three devices before starting patient recruitment, particularly because they had previous experience with reusable laryngeal mask type devices (classical laryngeal mask airway and PLMA) but not i-gel™ and LMA-S. Depth of anaesthesia was assessed by performing a jaw thrust manoeuvre [
9,
10]. The size of the device was determined according to the manufacturers’ weight-based recommendations.
If two attempts to insert the initially randomized SGA failed, the study protocol prescribed a change to one of the other two devices selected randomly (coin toss). If two attempts with the second SGA device were also unsuccessful the trachea was intubated.
Failed insertion of the SGA was defined as the inability to position the device within 60 seconds, an air leak through the drainage channel during positive pressure ventilation despite corrective manoeuvres (e.g. deeper insertion or up-and-down-manoeuvre) [
11], inability to introduce a suction catheter (12Ch for the i-gel™, 16Ch for the LMA-S and LTS-D) beyond the tip of the device, inability to establish successful ventilation with a stable end-expiratory CO
2 signal with a targeted expiratory tidal volume of 7 ml/kg because of leakage or airway obstruction.
The time required for successful insertion was defined as the time from placing the SGA in the front of the patient’s mouth to its placement in the correct position. A non-blinded observer who was not involved in the study recorded the time needed for insertion.
After successful insertion, the cuff of the LMA-S was inflated to a pressure of 60 cmH2O. The cuff of the LTS-D was initially inflated with the volume indicated on the syringe provided by the manufacturer for emergency use. The cuff pressure was measured at this inflation volume and air was then withdrawn until cuff pressure was also 60 cmH2O.
Ventilation was pressure-controlled (PCV) with a positive end-expiratory pressure (PEEP) of 3 cmH2O, a respiratory rate between 14 and 16 and an inspiratory to expiratory ratio of 1:1.5. It was adapted to give an end-tidal CO2 of 35-40 mmHg.
Leak pressure was determined as a function of cuff pressure for the SGA with inflatable cuffs (LMA-S and the LTS-D). Cuff pressure was measured with a manometer (VBM Medical GmbH, Sulz, Germany; range from 0 to 110 cmH
2O) that was connected to the SGA through a three-way stopcock with an attached syringe. Air was injected until the pre-determined cuff pressure was obtained and the required inflation volume was recorded. Cuff pressures started at 0 cmH
20 (completely deflated and equilibrated to ambient pressure) and were increased in 10 cmH
20 increments to 60 cmH
20. It was further increased in the LTD-S in two 20 cmH
20 increments to a maximum of 100 cmH
2O. The pressure limit of the anaesthesia circuit was set to 35 cmH
2O and airway pressure was increased steadily with a continuous flow of oxygen (3 l/min). Leakage was defined as air escape audible with a stethoscope placed on the larynx, and leak pressure was defined as the airway pressure at which leakage was first detected [
12]. After these measurements the cuff was inflated to 60 cmH
2O and kept at that pressure for the remainder of the study duration.
Airway pressures (PAW) and tidal volumes (Vt) were recorded and averaged over one minute after insertion, as well as after equilibration of ventilator settings. Dynamic airway compliance (Cdyn) was calculated using the formula . The LP defined the maximum inspiratory airway pressure setting.
Fibreoptic evaluation
Fibreoptic (FO) evaluation of the SGA’s position was performed after successful insertion and determination of the airway pressures and tidal volumes. The position was assessed using a previously described four points score (1 = only vocal cords seen; 2 = cords and/or arytenoids seen; 3 = only epiglottis seen; 4 = other (e.g. cuff, pharynx, etc)) [
13].
If an unexpected reduction in tidal volume occurred during stable anaesthesia and unchanged ventilator settings, the airway was assessed with the fibrescope for dislocation of the device or obstruction due to glottic narrowing or laryngospasm. Glottic narrowing was differentiated from laryngospasm by a partial closure of the vocal cords that was not reversed by deepening anaesthesia or by the administration of a neuromuscular blocker.
Airway morbidity
All devices were evaluated for traces of blood on the mask bowl (LMA-S, i-gel™) or the airway apertures and cuff (LTS-D). The patients were questioned about sore throat, discomfort during swallowing and hoarseness at one hour and at 24 hours after anaesthesia. These complaints were classified as none, mild, or severe.
Statistical analysis
Published data on leak pressures were used to estimate the necessary sample size. Assuming a mean LP of 24 cmH
2O for the i-gel™ [
3,
14], 26 cmH
2O for the LMA-S [
6,
7,
15], and 28 cmH
2O for the LTS-D [
8,
16] and an assumed standard deviation of 5 cmH
2O for all devices, a total sample size of 83 was calculated to detect differences with 90% power and a significance level of 0.05 [
17]. To allow for potential dropouts, a sample size of 40 patients per group was chosen.
The data were documented in an Excel™ spreadsheet and analyzed using SPSS Statistics™ software (IBM SPSS Inc., Chicago, IL, USA). Depending on the level of measurement of the dependent variables, ANOVA, rank variance analysis (Kruskall-Wallis), or multinomial logistic regressions and chi-square tests were used.