Following institutional ethics committee approval, American Society of Anesthesiologists physical status classification 1-3 adult patients scheduled to undergo surgery under general anesthesia were recruited. Patient characteristics such as age, gender, weight, and height were noted. Before surgery, all patients received a full-scale airway evaluation. Patients were included in this study if a staff anesthesiologist not involving in the study determined that they would require an AOTI based on history of prior difficult intubation or the presence of clinical predictors for the difficult airways. These predictors included previous history of multiple or failed laryngoscopy, Mallampati class 3 or 4 with a history of severe snoring and observed pauses in breathing during sleeping in the supine position, thyromental distance < 60 mm, limited mouth opening with interincisor distance < 30 mm, and head and neck movement < 80o (13). The head and neck movement was measured as described by Wilson and colleagues (14) . The exclusion criteria were inability to cooperate with adequate airway assessment, respiratory tract pathology (e.g., intrinsic laryngeal abnormalities), history of cardiovascular, hepatic, renal and coagulation diseases, pregnancy, inadequate transillumination of the anterior neck (e.g., grossly obese patients or patients with neck scar), long-term use of opioids or sedatives and risk of regurgitation–aspiration.
A total of 85 patients that met the inclusion criteria were prospectively enrolled into the study. During the preoperative visit, the details of the lightwand and fiberoptic techniques for airway topicalization and AOTI were explained for each recruited patient. Patients were also informed that they had the right to decline from participation. Because 5 patients did not agree to participate after interview, 80 cases were eventually included in the study and informed patient consent was obtained. Patients did not receive any premedication, were fasting for at least 6 hours and were restricted from oral intake of clear fluid for 4 hours. Perioperative monitoring included a 3-lead electrocardiogram (ECG), pulse oximetry, non-invasive blood pressure (BP) and capnography. In the preoperative holding area, a 20-gauge intravenous cannula was inserted and intravenous atropine 10 μg/kg was administered for its antisialogogue effect (1). A 18-gauge indwelling intravenous cannula was also inserted into the antebrachial or antecubital vein on the contralateral arm for serial plasma lidocaine level sampling. In all patients, the posterior pharynx was anesthetized with five intra-oral sprays using 10% lidocaine (Xylocaine® 10% oral spray, Astra® Pharmaceutical Products, Inc, Westborough, MA); each depression of the release button delivered 0.1 ml (10 mg). In the operating room, patients received fentanyl 1.5 μg/kg and midazolam intravenously to achieve anxiolysis as defined by an Observer's Assessment of Alertness/Sedation Scale (OAA/S) of 14-16 (15). If the OAA/S was less than 14, or the patient was not cooperative due to excess sedation, the patient was excluded from this study.
Once the desired level of sedation was achieved, patients were randomly assigned into the lightwand (LW) group and fiberoptic bronchoscope (FOB) group (N=40 per group). Randomization was performed using computer generated random numbers, enclosed in sealed envelopes. In the LW group, a middle size wand of the Trachlight™ (Laerdal Medical Corporation, New York, USA) and a MADgic® atomizer (Wolfe Tory Medical Inc, Salt Lake City, UT) were assembled together as the combined unit using the method previously described (11). In the FOB group, a 1.1 mm single-orifice end hole epidural catheter was threaded through the suction channel of a FOB with an outer diameter of 3.1 mm (Olympus LF-DP, Tokyo, Japan) (3). Both the airway topicalization and AOTI with the lightwand and fiberoptic techniques were accomplished by the experienced anesthetists using the two methods. They were trained in the same postgraduate education programme of difficult airway management, had been engaging in clinical anesthesia for at least 10 years, and had performed the awake intubation in more than 50 patients with known difficult airways using the lightwand and FOB, respectively, before the start of the study.
Patient was positioned supine with the head and neck in a neutral position. The jaw was lifted upward to elevate the epiglottis and enlarge the pharyngeal cavity. In the LW group, the combined unit was passed in the midline until a bright well circumscribed circle of light was seen at the level of the hyoid, which indicated that its tip was located in the epiglottic vallecula. At this time, 1 ml of 2% lidocaine was sprayed using the MADgic® atomizer. Then the fine left or right rotation of the combined unit was done to obtain a bright glow in the lateral aspect of the larynx, which indicated that its tip was placed in the pyriform recess, and 2 ml of 2% lidocaine was sprayed in two aliquots onto the bilateral pyriform recess. This procedure was repeated after 5 minutes. Five minutes after the second supraglottic spray, the combined unit was again inserted until a central, clear and bright transillumination on the cricothyroid membrane, which suggested a correct positioning of its tip in the laryngeal aperture, was observed. At this time, 3 ml of 2% lidocaine was sprayed during inspiration to anesthetize laryngeal and tracheal areas (11) .
In the FOB group, the FOB was inserted through a suitable size Berman intubating airway (Vital Signs, Inc., Totowa, NJ, USA) into the hypopharynx (2) and its tip was first positioned at the epiglottic vallecula and then in vicinity of the piriform recess. Three ml of 2% lidocaine was slowly sprayed in three aliquots of 1-ml onto these supraglottic areas. After 5 minutes, this procedure was repeated. Following another 5-min waiting period, the FOB was reinserted to expose the glottis and 0.5 ml of 2% lidocaine was sprayed into the laryngeal area. This procedure was repeated at 3-min intervals until adequate topical anesthesia of the larynx, as evidenced by cessation of the laryngeal response to further lidocaine administration (16,17). The FOB was then advanced into the trachea and its tip was positioned 2 cm below the glottis. During inspiration, 3 ml of 2% lidocaine was slowly sprayed into the trachea.
Failure of airway spray attempt was defined as withdrawal of the combined unit or FOB from the patient's mouth because of inability to be directed to the targeted area. A maximum of three attempts was permitted. Both the number of attempts required for each airway spray and causes of failed attempts were noted. The time for each airway spray in different targeted areas, namely the period from initial insertion of the combined unit or FOB to its withdrawal from the patient's mouth after completion of airway topicalization (including the time required for repeating attempts), was recorded using a digital stopwatch. The total time for airway sprays were measured as the time from first insertion of the combined unit or FOB to its withdrawal from the patient's mouth after the last airway spray (including the awaiting time between repeated airway sprays). During each airway spray, an independent investigator assessed and scored a patient's discomfort using a 4-point scale: no response=1; slight gagging=2; moderate gagging=3; severe gagging or patient's inability to tolerate=4. Gagging was considered slight if only one episode of gagging occurred, moderate if 2-3 gagging episodes occurred, and severe if more than 3 episodes occurred (16). After completion of airway sprays, the operator was asked to grade his/her subjective opinion about the difficulty of the procedure by a visual analogue scales (VAS) ruler from 0 to 10 cm, where 0 was described as very easy and 10 as impossible. The total doses of lidocaine used for airway sprays (including 50 mg used for intra-oral sprays) were noted. Patients were also asked whether they had experienced any local anesthetic side-effects, such as dysphoria, dizziness, nausea, and shivering, visual and auditory disturbances, involuntary movements, etc. (17) .
Five minutes after completion of airway topicalization, AOTI was performed using the lightwand and the FOB in the LW group and FOB group, respectively. After insertion of the endotracheal tube (ETT) into the trachea, an independent investigator scored patient's reaction using a modified 6-point scale (no reaction=1; no change or a single change in the facial expression: slight reaction=2; grimacing facial expressions: moderate reaction=3; severe facial grimace but retained ability to follow verbal command and a reflex with no head movements: severe reaction=4; severe facial grimace associated with discomforting head movements, but still ability to obey verbal command: very severe reaction=5; and severe facial grimace associated with protective head and limb movements hindering the procedure, and inability to obey any verbal command: noncooperation=6)(16). Cough severity was rated on a 4-point scale (no cough=1; slight coughing=2; moderate coughing=3; severe coughing=4). Coughing was considered slight if no more than 2 coughs in sequence occurred, moderate if 3-5 coughs in sequence occurred, and severe if more than 5 coughs in sequence occurred (16). Intubating conditions were assessed using a 3-point scale (excellent=no response and cough; adequate=both patient's coughing and reaction scores were ≤ 3; unacceptable=both patient's coughing and reaction scores were ≥ 4)(12) . The intubation time, defined as the period from initial insertion of the lightwand or FOB to start of ventilation through the ETT, was measured with a stopwatch. The number of intubation attempts was also noted. The operator was asked to grade his/her subjective opinion about difficulty of intubation by a VAS ruler from 0 to 10 cm (12). The independent investigator also recorded BP and HR before airway sprays with the lightwand or FOB (baseline), immediately after completion of supraglottic and first laryngeal spray, at intubation, and 1 minute after intubation. Following confirmation of correct ETT placement by chest auscultation and capnography, general anesthesia was induced with intravenous propofol and maintained with intravenous fentanyl, 1-2% end-tidal isoflurane plus 60% nitrous oxide in oxygen, and vecuronium for muscle relaxation.
A baseline blood sample (5 ml) was obtained from the indwelling intravenous cannula before any lidocaine was administered (T0). Further blood samples were obtained immediately following the second supraglottic spray and the first laryngeal spray, and at 10-min intervals until 60 minutes had elapsed from the final lidocaine spray. The plasma lidocaine concentrations were assayed using high-performance liquid chromatography with ultraviolet detection accurate to 0.02 μg/ml.
At 24 hours following surgery, an independent investigator unaware of the patient's group assignment, used VAS rulers marked 0-100 cm to determine the patients' recall, and levels of discomfort and pain during the periods of airway spray and AOTI. The patients were also asked whether they had sore throat or other uncomfortable sensation of the throat. The VAS rulers from 0 to 100 cm described "no recall" to "perfect recall", "no discomfort" to "extreme discomfort" and "no pain to intolerable pain" (15) .
Sample size selections were based on the results of primary observed variables of our preliminary trial including 20 patients in each group. Power calculation indicated that 39 patients in each group would at least be required to detect a difference of 20% between the groups with respect to success rate of the first laryngeal spray by one attempt for a type I error of 0.05 and a power of 0.8 for a two-tailed 2×2 chi-squared test. A total of 40 patients per group was studied to account for methodological difficulties that could have led to exclusion from the study. Unless otherwise stated, the quantitative data were expressed as mean±SD with a range and the non-quantitative data were expressed as median with a range. The statistical analysis of data was performed with SPSS (Version 11.5, SPSS Inc., Chicago, IL). Comparison of two means between groups was performed using the Student t-test, and comparison of two medians between groups was performed using the Mann-Whitney U test. Comparison of percentages between groups was performed using a chi square test, a Fisher exact test and a McNemar's test as appropriate. The intragroup comparisons of hemodynamic data at different observed points were done using repeated measures analysis of variance. All comparisons were two-sided and P＜0.05 was considered statistically significant.