2018-19 Section 7-Neoplastic and Inflammatory Diseases of the Head and Neck eBook

incision and drainage (I&D) using a prospective, multi- center outcomes database. The primary objective was to determine if delay in drainage of deep neck abscess in either population was associated with increased 30-day abscess-specific M&M.

between adults and pediatrics to account for differences in vital signs and other factors specific to either population. Laboratory values were examined using definitions for abnormal values and reported as high or low. 13 Indicator variables were used to analyze missing laboratory values. We report cells in tables with an n < 5 with an asterisk in order to protect patient privacy and prevent identification of specific patients, unless the cell is reporting a com- posite outcome. Outcomes Standard definitions for ACS–NSQIP and ACS–NSQIPP post- operative outcomes have been described previously. 11,12 We defined abscess-specific M&M as the occurrence of sepsis, septic shock, pneumonia, unplanned intubation, mechanical ventilation for greater than 48 hours, deep- or organ-space surgical site infection, and death because these complications best reflect potential compli- cations of deep neck abscess and are shared by both databases. Post- operative sepsis is defined in the same manner as preoperative sepsis when the signs and symptoms begin postoperatively or when a patient with preoperative sepsis experiences postoperative wor- sening. Additional complications that were examined include super- ficial surgical site infection, stroke, peripheral nerve injury, intraventricular hemorrhage, return to surgery, and length of stay. We additionally analyzed length of stay with prolonged length of stay defined as 7 days. Statistical Analysis We assessed continuous variables using analysis of variance. We assessed categorical variables using a chi-square for a linear trend test because our primary outcome of interest, time until sur- gical drainage, was linear. We substituted a Fisher’s exact test for the chi-square for linear trend when n < 5. Length of stay was determined to be nonparametric and was assessed using the Kruskal-Wallis test. The Kruskal-Wallis test is similar to analysis of variance analysis but is useful for nonparametric data, in this case the length of stay. P values < 0.05 were considered statistically significant, and all P values were 2-sided. We performed multivari- ate logistic regression using a significance test of the covariate vari- able selection strategy and including variables with a P value < 0.20 on bivariate analysis to determine the influence of individual patient variables on the main outcome variable. Variables found to be significant in either adults or pediatrics were included in the other corresponding model. Model performance was evaluated using the Hosmer-Lemeshow test and C statistic. The Hosmer- Lemeshow test is useful for determining goodness of fit for logistic regression models, that is, that predictions obtained from the model reliably reflect occurrence of events in the data. A significant result would indicate that the model is not well calibrated to the data set. The C statistic assesses how well the model discriminates subjects having an event from those not having that event. A value > 0.7 is generally considered acceptable discrimination. All laboratory val- ues were included in models because continuous variables to retain all information. We report odds ratios (OR) with 95% confidence intervals (CI) (95% CI) for unadjusted and adjusted analysis. Sta- tistics were done using SPSS 22 (SPSS Inc., Chicago, IL). RESULTS We identified 347 adult patients in ACS-NSQIP from 2005 to 2013, and 665 pediatric patients in ACS–NSQIPP from 2012 to 2013, with a deep neck abscess treated with I&D (Fig. 1). The frequency of surgical drainage by day- after-admission-groups is shown in Figure 2. Patient demo- graphics and comorbidities are summarized in Table I.

MATERIALS AND METHODS Data Source and Patient Selection

We conducted a prospective, multi-institutional, risk- adjusted cohort study using the American College of Surgeons National Surgical Quality Improvement Program (ACS–NSQIP) from 2005 to 2013 and its companion pediatric database, ACS–NSQIP Pediatric (ACS–NSQIPP), from 2012 to 2013. ACS–NSQIP and ACS–NSQIPP are designed to improve surgi- cal quality by collecting short-term perioperative outcomes on patients undergoing surgery at 435 and 50 participating sites, respectively, including community and academic hospitals. The details of sampling and methodology for these databases have been described previously. 7 Patients 0 to 18 years of age were included in ACS–NSQIPP, and patients > 18 years of age were included in ACS–NSQIP. This study was deemed exempt from review by the Northwestern University and the Ann & Robert H. Lurie Children’s Hospital of Chicago Internal Review Board. We included patients who underwent I&D in the operating room for a diagnosis of deep neck abscess by selecting Current Pro- cedure Terminology codes 21501 (I&D deep abscess or hematoma, soft tissues of neck or thorax), 42720 (I&D abscess; retropharyng- eal or parapharyngeal, intraoral approach), and 42725 (I&D abscess; retropharyngeal or parapharyngeal, external approach). We excluded I&D of hematoma or I&D of the thorax by selecting postoperative International Classification of Diseases, Ninth Revi- sion (ICD-9), codes 478.22 (parapharyngeal abscess), 478.24 (retro- pharyngeal abscess), and 682.1 (cellulitis and abscess of neck). We excluded patients who underwent I&D more than 7 days after admission to eliminate patients who potentially developed a deep neck abscess during admission. We defined our predictor groups as surgery on day 0, days 1 to 2, or days 3 to 7, which correspond to urgent drainage; drainage within 48 hours; and delayed drainage, respectively. We selected these time points based on clinically applicable intervals and review of the literature. 8–10 After complet- ing our analysis, we then performed a sensitivity analysis using surgical timing categories of day 0–1, days 2 to 3, and days 4 to 7 to assess if our results were influenced by our groups. Preoperative Variables Potential predictors for postoperative adverse events included patient demographics, comorbidities, intraoperative, and infectious variables. Demographic variables were age, gender, race/ethnicity, and body mass index. Comorbidity variables eval- uated in both groups were diabetes mellitus (DM), bleeding disor- der, and steroid use for a chronic condition. In adults, additional comorbidity variables included preoperative dyspnea, severe chronic obstructive pulmonary disease, hypertension requiring medication, dialysis, > 10% weight loss in last 6 months, conges- tive heart failure, ascites, and disseminated cancer. In children, comorbidity variables also included congenital malformation, con- genital heart disease, asthma, and preoperative pneumonia. Surgi- cal variables included American Society of Anesthesiologists (ASA) class and surgical approach (external or intraoral). Infectious variables included preoperative sepsis, which is categorized into systemic inflammatory response syndrome (SIRS); sepsis; or septic shock based on examination of preoperative vital signs and laboratory values, as described in each program’s respective user guide. 11,12 It is notable that the definition of SIRS varies slightly

Laryngoscope 126: August 2016

Cramer et al.: Delay in Drainage of Deep Neck Abscess

211