2017-18 HSC Section 3 Green Book

Original Investigation Research

Complications After Obstructive Sleep Apnea Upper Airway Surgery

23 patients were excluded because the information compiled in the database was incomplete, such as absent results of pre- operative sleep study and incomplete documentation of the intraoperative and postoperative findings. The characteris- tics of the 95 patients are presented in Table 1 . Within this co- hort, the mean (SD) age was 37.7 (1.6) years, which was a rela- tively younger populationundergoingOSA surgery; 87 patients (92%) were male; and 83 (87%) were Malay. The young age of the patients reflects the more aggressive approach in treating younger patients with severe OSA in our population. 20 Pa- tients with more severe OSA were observed to have greater bodymass index (BMI) (Cohen d, 0.27; 95%CI, −0.28 to 0.82), longer surgical time (Cohen d, 1.57; 95%CI, 0.95-2.15), andolder age (Cohen d, 3.06; 95% CI, 2.29-3.77). Thirty-eight patients (40%) were prescribed CPAP before surgery, of whom8 patients had central apnea components in their preoperative polysomnographic results. In these pa- tients, no correlation was seen between the amount of posi- tive pressure they required preoperatively or the central ap- nea components in their polysomnographic results with operative complications. Forty-eight of 95 patients (51%) de- veloped at least 1 operative complication. The complication rates are presented according to the time of occurrence (ie, in- traoperative, immediately postoperative [from the recovery roomperiod until within 24 hours of surgery], andwithin 2-30 postoperative days). For analysis purposes, an overall dichoto- mous variable was created for patients who experienced any operative complication, and a dichotomous variable was cre- ated for each complication category to identify patients with more than 1 complication in each category ( Table 2 ). Because some patients experienced multiple complications, the over- all rate of complications is less than the sumof the individual complication subcategory. The most common complications were respiratory, andmost complications occurred during the intraoperative period. Overall, respiratory and cardiovascu- lar complications occurred early, whereas surgical complica- tions happened after day 1 of surgery. The patients underwent the following procedures. Five pa- tients (5%) underwent nasal surgery only, 33 (35%) under- went palatopharyngeal surgery only, 4 (4%) underwent tongue base surgery only, and 53 (56%) underwent a combination of at least 2 operations simultaneously. We categorized patients into complication and noncomplication groups and com- pared these groups todetermine factors that contributed toop- erative complications ( Table 3 ). Complication is affectedby age (Cohen d, 2.78; 95%CI, 2.19-3.32) and BMI (Cohen d, 3.2; 95% CI, 2.57-3.78). The presence of comorbid disease generally in- creases the rate of complication, but this finding was not sta- tistically significant ( P = .07). Within the polysomnographic factors, lowest oxygen desaturation (Lsat) (Cohen d, 5.5; 95% CI, 4.58-6.33) and longest apnea duration (Cohen d, 4.0; 95% CI, 3.27-4.66) were significantly associated with complica- tions but not AHI. Complication rates were not affected by the sites of surgery and whether a single operation or concurrent multilevel operations were performed. Logistic regressionwasused toanalyze themultivariableas- sociation of OSA variables, age, BMI, smoking, and type of sur- gerywithpostoperativecomplications.Analyseswereperformed

desaturation of 85%or less in the postoperative period, bron- chospasm, and respiratory arrest. Cardiovascular complica- tions includedmyocardial infarction, thromboembolic events, arrhythmias, persistent hypertension requiringmedical inter- vention, and stroke. Surgical complications within the first 30 days of operationwerebleeding, surgical site infections,wound breakdown, readmission to the operating theater for further surgical procedure, and hospital readmission for any reasons related to the surgery after initial discharge from the ward. Comorbidconditions includedhistoryof diagnosis andcur- rent treatment for hypertension, type 1 diabetes, cardiovas- cular disease (coronary artery disease, heart failure, atrial fi- brillation, ventricular ectopy), pulmonary disease (asthma, chronic obstructive pulmonary disease, bronchitis, or pulmo- naryhypertension), hypothyroidism, andkidneydisease (blood creatinine concentrations above the age-adjusted norm). After postanesthesia recovery in the operating theater, pa- tientsweretransferredtotheintensivecareunitwithstrict1-on-1 nursing care or to anacute care sectionof a generalward,where theywereunder 24-hour continuous oximetrymonitoring. The decision for the transfer was made by the attending anesthesi- ologist (A.H.M.). It is a standard practice in our center to apply CPAP or BiPAP postoperatively in the postanesthesia care unit in patients who were using these devices preoperatively. Statistical Analysis Analysiswas performedusing SPSS statistical software forWin- dows, version 19 (IBM Inc). Using the estimated incidence of overall complications of surgery in patients with OSA at 30% with a 10% margin of error, 3,4,7,17,18 the sample size was cal- culated using a single proportion formula in Power Analysis and Sample Size, 11th edition (NCSS Statistical Software), soft- ware. The estimated minimum sample size was 92. The severityofOSAwasquantified intomild,moderate, and severeusingAHI according to theAASMcriteria. 15,19 Patient and procedural characteristicswere presented according to theOSA severity. Datawere calculated inmean (SD) for continuous vari- ables and frequency (percentage) for categorical variables. Data characteristics were compared across the OSA severity using analysisofvarianceortheKruskal-Wallistestforcontinuousvari- ables and the χ 2 or Fisher exact test for categorical variables. Theprevalence and types of complicationsweredescribed. Patientswere then categorized into complication andnoncom- plicationgroups andcompared for eachpotential risk factor. The unpaired, 1-tailed t test or Mann-Whitney test and χ 2 or Fisher exact test were performed to analyze data with respect to the presence or absence of a specific risk factor for complications. Logistic regression was used to analyze the multivariable association of potential confounding factors with operative complications. In all cases, P < .05 was considered statisti- cally significant. Results Within the study period, 135 patients underwent upper air- way surgery as a treatment for OSA. Seventeen patients were lost to follow-up and were excluded from the study. Another

(Reprinted) JAMA Otolaryngology–Head & Neck Surgery March 2017 Volume 143, Number 3

jamaotolaryngology.com

Copyright 2017 American Medical Association. All rights reserved.

137