Customize your JAMA Network experience by selecting one or more topics from the list below.
Sangal NR, Nishimori K, Zhao E, Siddiqui SH, Baredes S, Chan Woo Park R. Understanding Risk Factors Associated With Unplanned Reoperation in Major Head and Neck Surgery. JAMA Otolaryngol Head Neck Surg. 2018;144(11):1044–1051. doi:10.1001/jamaoto.2018.2049
Which risk factors are associated with unplanned reoperation following major head and neck surgery?
In this cohort study using the American College of Surgeons National Surgical Quality Improvement Program database, 14% of patients (275 of 1941) underwent a reoperation following glossectomy, mandibulectomy, laryngectomy, or pharyngectomy. Independent factors associated with reoperation included black race, disseminated cancer, total operation time, superficial and deep surgical site infections, wound dehiscence, and ventilator dependence.
Risk stratification may identify those patients more likely to need reoperation following major head and neck surgery and may assist in surgical planning.
Understanding the preoperative, intraoperative, and postoperative risk factors of reoperation is the optimal way to approach decreasing its incidence.
To identify risk factors of unplanned reoperation following major operations of the head and neck.
Design, Setting, and Participants
This retrospective cohort study queried the American College of Surgeons National Surgical Quality Improvement Program database and identified 2475 cases of major operations of the head and neck performed between 2005 and 2014. Specific operations analyzed were glossectomy, mandibulectomy, laryngectomy, and pharyngectomy. Univariate and multivariate analyses were performed to compare demographic and clinical characteristics of patients with or without unplanned reoperation. Data were analyzed between September and November 2017.
Main Outcomes and Measures
The primary outcome was incidence of unplanned reoperation in patients with major operations in the head and neck region. An additional aim was to assess the risk factors associated with an increased likelihood of reoperation.
In total, 1941 patients were included in this study (1298 [66.9%] males), with most patients (961 [49.5%]) between 61 and 80 years of age. The overall unplanned reoperation rate within 30 days after the principal operative procedure was 14.2% (275 patients). The operative procedure with the highest reoperation rate was pharyngectomy (8 of 46 [17.4%]), followed by glossectomy (95 of 632 [15.0%]), laryngectomy (53 of 399 [13.3%]), and mandibulectomy (25 of 240 [10.4%]). Among the unplanned reoperation patients, 516 patients (76.8%) underwent reoperation during their initial hospital admission and 156 patients (23.2%) after readmission. The mean (SD) number of days from the principal operative procedure to unplanned reoperation was 8.5 (3.6) days for initial-admission reoperations and 16.0 (4.8) days for readmission reoperations. The most common unplanned reoperation procedures overall included repair, surgical exploration, and revision procedures on arteries and veins (47 of 2475 [1.9%]), incision procedures on the soft tissue of the neck and thorax (37 of 1941 [1.9%]), and incision and drainage procedures on the skin, subcutaneous, and accessory structures (21 of 1941 [1.1%]). Multivariate analysis results indicated that the independent risk factors for unplanned reoperation following a major cancer operation of the head or neck included black race (odds ratio [OR], 1.72; 95% CI, 1.09-2.74), disseminated cancer (OR, 1.85; 95% CI, 1.14-3.00), greater total operation time (OR, 2.05; 95% CI, 1.49-2.82), superficial (OR, 2.56; 95% CI, 1.55-4.24) or deep (OR, 4.83; 95% CI, 2.60-8.95) surgical site infection, wound dehiscence (OR, 8.36; 95% CI, 5.10-13.69), and ventilator dependence up to 48 hours after surgery (OR, 2.95; 95% CI, 1.79-4.87).
Conclusions and Relevance
The identification of a significant association of black race, disseminated cancer, total operation time, surgical site infection in either the superficial or deep spaces, wound dehiscence, or ventilator dependence for more than 48 hours after surgery with increased risk of reoperation in major head and neck surgery may guide the modification and adaptation of these risk factors to decrease the burden that unplanned reoperation places on patients, surgeons, and the health care system.
Unplanned reoperations (URs) are a major stressor on the health care system of the United States.1,2 Unplanned reoperations increase patient morbidity and mortality, amplify health care use, and decrease access to care. By definition, URs increase length of hospital stays and hospital costs, with no improvement in outcomes. Surgical complications, including URs, have been shown to decrease satisfaction with care and are a source of significant concern for surgeons and patients alike.3 In addition, rates of URs are considered a hospital outcome measure, a tool to determine a hospital’s ability to provide safe and efficient care, making it a vital metric in the changing landscape of hospital reimbursement.4,5
Unplanned reoperations have been extensively studied in many surgical fields. However, in otolaryngology, there are few large studies assessing the incidence, impact, and risk factors associated with URs. Gaining a better understanding of why URs occur and how they can be prevented is essential for improving surgical services and by extension health care use.6 The contribution of URs to the health care system cannot be understated despite such reoperations being a relatively rare occurrence. Estimated incidence rates in most surgical fields fall between 2% and 10% of all surgical procedures, making URs difficult to study on an institution by institution basis.7-10 Database studies, which provide larger numbers of patients with more patient and institutional diversity, are well suited to study infrequent outcomes, such as URs. Here, we used the American College of Surgeon’s National Surgical Quality Improvement Program (ACS-NSQIP) database to identify the factors associated with URs among patients who had undergone head and neck surgical procedures. We aimed to provide insight into how preoperative risk stratification can decrease the incidence of URs in these complex operations.
A retrospective cohort analysis was conducted using the ACS-NSQIP database from 2005 to 2014 to identify potential risk factors associated with URs. The ACS-NSQIP database is a deidentified sampling of surgical cases from across the United States. The database contains information on more than 150 variables, including preoperative and intraoperative factors as well as 30-day postoperative outcomes. As of 2014, there were 517 participating medical sites. Data abstraction, including classification of race, for the database is performed by certified surgical clinical reviewers, and interrater reliability audits are completed regularly to ensure accuracy of collection and quality of data. The combined results of audits completed prior to and during 2014 revealed an overall disagreement rate of 2%. Owing to the deidentified nature of the database, the Rutgers New Jersey Medical School (Newark, New Jersey) Institutional Review Board deemed this study exempt from requiring their approval and waived the requirement of obtaining informed patient consent.
The 2005-2014 ACS-NSQIP database was queried for cases with a major surgery of the head and neck, including glossectomy, mandibulectomy, laryngectomy, and pharyngectomy, as the principal or concurrent procedure using the Current Procedural Terminology codes listed in Table 1. Only cases with available data for URs within a 30-day postoperative period were included in the analysis. Cases of UR not associated with the principal or concurrent operative procedure were excluded from the present analysis.
The primary outcome of interest was UR within 30 days following a major surgery of the head and neck. Demographic variables, such as age, sex, and race, were examined. Race was analyzed in this study to assess its variability between cohorts and to evaluate its role as a risk factor for reoperation. Preoperative clinical variables examined included body mass index calculated as weight in kilograms divided by height in meters squared, American Society of Anesthesiologists physical status classification, past medical conditions, previous treatments, and ventilator dependence 48 hours preceding surgery. Postoperative clinical outcomes within 30 days following surgery examined included superficial surgical site infection (SSI), organ or space SSI, wound dehiscence, systemic infection, ventilator dependence for more than 48 hours, and acute renal failure. Total primary operation time (minutes) and total length of hospital stay (days) were compared among patients who underwent a UR vs those who did not. Total operation time and length of hospital stay were categorized as binary variables, comparing operations greater than the 50th percentile with those less than this value.
Statistical analyses progressed from univariate to bivariate then multivariate analyses. Multivariable logistic regression modeling was used to assess the risk of a UR within 30 days following a major surgery of the head and neck. Pearson χ2 tests and independent samples t tests were used for categorical and continuous variables, respectively. Fisher exact tests and Mann-Whitney tests were used as indicated. For ordinal variables in χ2 analyses, gamma, a proportional reduction in error measure, was used as a symmetric measure of effect size. Significant variables in bivariate analysis were added to the multivariable regression model in a stepwise, progressive manner. Bootstrapping was performed to achieve precise estimates of confidence intervals. Akaike information criterion and Nagelkerke R2 were used to assess each model. Additional analyses were conducted to assess risk factors for URs by surgery type. A 2-sided P < .05 was considered statistically significant for all tests. All statistical analyses were performed using SPSS, version 23 (IBM) from September to November 2017.
Overall, there were 2475 major surgical cases of the head and neck between 2005 and 2014. Among these, 534 cases (21.6%) were excluded from the present analysis: 519 cases (21.0%) were excluded for lack of available data for URs, and 15 cases (0.6%) were excluded for URs unassociated with the principal or concurrent procedure. Patients who were excluded did not significantly differ from patients included in the analysis by age, sex, race, and American Society of Anesthesiologists scores. However, some preoperative clinical variables differed between patients excluded and patients included in the analysis, such as hypertension with medication, distant cancer, and some postoperative clinical outcomes, including superficial SSI, deep SSI, organ space SSI, wound dehiscence, pneumonia, reintubation, ventilator dependence for more than 48 hours, bleeding, and sepsis. Although these differences reached statistical significance, there was no clinical significance associated with the exclusion of these patients.
Of the 1941 patients included in the analysis, 275 patients (14.2%) underwent a UR within 30 days following surgery. The likelihood of a UR did not significantly differ by type of surgery (Table 2). Table 2 gives a summary of the demographic and clinical variables examined. Overall, more male patients (1298, 66.9%) than female patients underwent major head and neck surgery; however, the distribution of sex among patients without a UR compared with those with a UR was not significant (Kruskal gamma coefficient [γ], 0.03; 95% CI, −0.11 to 0.17). The patients who underwent a UR did not significantly differ from patients without reoperation by age (γ, 0.05; 95% CI, −0.06 to 0.16), race (γ, 0.14; 95% CI, −0.07 to 0.35), or body mass index (median [interquartile range], 26.22 [7.30] vs 26.36 [7.34]) (Table 2). Among clinical variables, patients who underwent a UR were more likely to have hypertension (γ, 0.17; 95% CI, 0.04-0.30), hemiplegia (γ, 0.72; 95% CI, 0.24-1.20), quadriplegia (γ, 1.00; 95% CI, 1.00-1.00), distant cancer (γ, 0.35; 95% CI, 0.16-0.53), wound infection (γ, 0.34; 95% CI, 0.09-0.58), recent weight loss (γ, 0.22; 95% CI, 0.04-0.40), bleeding disorder (γ, 0.35; 95% CI, 0.22-0.47), and radiotherapy for malignant neoplasm (γ, 0.60; 95% CI, 0.23-0.97) than patients without reoperation. Patients who underwent a UR were more likely to have postoperative complications of superficial SSI (γ, 0.48; 95% CI, 0.31-0.64), deep space SSI (γ, 0.77; 95% CI, 0.67-0.87), wound dehiscence (γ, 0.84; 95% CI, 0.77-0.90) pneumonia (γ, 0.35; 95% CI, 0.15-0.54), reintubation (γ, 0.52; 95% CI, 0.29-0.75), acute renal failure (γ, 0.72; 95% CI, 0.24-1.19), urinary tract infection (γ, 0.44; 95% CI, 0.11-0.78), postoperative bleeding (γ, 0.40; 95% CI, 0.28-0.51), graft failure (γ, 1.00; 95% CI, 0.99-1.00), deep vein thrombosis (γ, 0.77; 95% CI, 0.50-1.04), or sepsis (γ, 0.64; 95% CI, 0.49-0.79), and to be ventilator dependent for more than 48 hours (γ, 0.61; 95% CI, 0.48-0.74) (Table 3). The most common URs overall included repair, surgical exploration, and revision procedures on arteries and veins (47 of 2475 [1.9%]), incision procedures on the soft tissue of the neck and thorax (37 of 1941 [1.9%]), and incision and drainage procedures on the skin, subcutaneous, and accessory structures (21 of 1941 [1.1%]). Furthermore, patients with a UR had significantly longer primary total operation time (γ, 0.41; 95% CI, 0.29-0.52) and total length of hospital stay (γ, 0.59; 95% CI, 0.50-0.68) (Table 2). No significant differences were found for the other clinical variables examined.
Table 4 gives the final iteration of a 4-step multivariable regression model. Model 1 includes demographic variables and total operation time. Model 2 includes all model 1 variables and preoperative clinical variables significant in bivariate analysis. Model 3 includes all model 2 variables and postoperative clinical outcomes significant in bivariate analysis. The final regression, model 4, includes demographic variables, total operation time, and clinical variables influential in the other 3 models. Unplanned reoperation was associated with black race (odds ratio [OR], 1.72; 95% CI, 1.09-2.74), disseminated cancer (OR, 1.85; 95% CI, 1.14-3.00), greater total operation time (OR, 2.05; 95% CI, 1.49-2.82), superficial (OR, 2.56; 95% CI, 1.55-4.24) and deep (OR, 4.83; 95% CI, 2.60-8.95) surgical site infections, wound dehiscence (OR, 8.36; 95% CI, 5.10-13.69), and ventilator dependence up to 48 hours after surgery (OR, 2.95; 95% CI, 1.79-4.87).
Multivariable regression models identified previously were stratified by surgery type. For glossectomy, after adjusting for other variables, a UR was associated with other race (OR, 0.24; 95% CI, 0.67-0.87), deep SSI (OR, 4.76; 95% CI, 1.81-12.52), wound dehiscence (OR, 6.47; 95% CI, 3.32-12.62), and ventilator dependence for more than 48 hours (OR, 2.67; 95% CI, 1.38-5.15). For mandibulectomy, after adjusting for other variables, a UR was associated with superficial SSI (OR, 3.98; 95% CI, 1.34-11.83) and wound dehiscence (OR, 33.64; 95% CI, 7.71-146.70). For laryngectomy, after adjusting for other variables, a UR was associated with male sex (OR, 2.62; 95% CI, 1.15-5.99), preoperative hypertension (OR, 1.94; 95% CI, 1.00-3.73), deep SSI (OR, 12.00; 95% CI, 3.63-39.65), and wound dehiscence (OR, 7.73; 95% CI, 2.86-20.90). For pharyngectomy, after adjusting for other variables, a UR was associated with black race (OR, 19.70; 95% CI, 2.50-155.51), deep SSI (OR, 96.01; 95% CI, 3.90-2388.10), wound dehiscence (OR, 81.44; 95% CI 3.07-2157.13), and ventilator dependence for more than 48 hours (OR, 121.16; 95% CI, 10.63-1380.43).
Unplanned reoperations are a major burden on the health care system and are associated with significant morbidity and mortality.11,12 In the current health care delivery system, any inefficiency is a target for reimbursement modification, making the control of UR rates a major goal for all institutions. The Hospital Readmissions Reduction Program of the Affordable Care Act enacted a penalty against hospitals with significant readmission burden in 2012. This burden is strongly associated with URs among surgical patients, making it a reduction priority for hospital systems.13-16 It is important for clinicians to understand preoperative, perioperative, and postoperative risk factors. To our knowledge, the present study provides the first multi-institutional analysis of risk factors associated with URs occurring within 30 days following major head and neck surgical procedures.
We used the ACS-NSQIP database to identify cases of major surgical procedures complicated by a UR. The NSQIP is a nationally validated, risk-adjusted, outcomes-based program, designed to measure and improve the quality of surgical care. Using the ACS-NSQIP database, we identified 1941 major surgical procedures of the head and neck between 2005 and 2014. The rate of URs in these operations was 14.2%. This figure is significantly higher than that reported in other studies in different fields. The lowest risk of reoperation was found in cases of outpatient plastic surgery, for which Lim et al6 identified a UR rate of 0.7% in 11 468 outpatient surgical procedures. For inpatient spinal surgery, Kim et al17 found UR rates to be between 4% and 8%. In a study of 852 patients identified from the NSQIP database who underwent laryngectomy for laryngeal cancer, Helman et al13 reported that 12.4% were brought back to the operating room. The type of operation conducted is essential in the consideration of acceptable rates of URs because case by case variation is significant. Even still, the burden of URs is significant in major surgical procedures of the head and neck, likely due to anatomic considerations and baseline patient risk factors, making risk stratification important.
The NSQIP collects demographic characteristics, preoperative medical diagnoses, intraoperative measures, and postoperative complications. We assessed the contributions of these features to risk of reoperation. Using χ2 analyses of all major head and neck surgical procedures, we found no significant differences in rates of reoperation by demographic characteristics such as age cohort, sex, race, or body mass index. Many preoperative medical diagnoses were found to be significantly different between reoperations and successful primary operations. There was also a significant increase in the total primary operation time, by greater than 100 minutes among patients who underwent reoperations. A majority of postoperative complications were found to have increased rates among reoperations.
Our multivariate regression analyses of the factors identified to be significantly different among reoperations found that for black race the OR was 1.72, a similar rate as has been found in other surgical fields.10,18 A recent retrospective cohort analysis of septorhinoplasties conducted between 2005 and 2009 in the Healthcare Cost and Utilization inpatient database found revisit rates in black patients to be significantly increased.19 This increased risk is certainly multifactorial, and a full understanding of this phenomenon is beyond the scope of the present study. However, a study by Dimick et al20 using national Medicare data for patients undergoing high-risk surgical procedures showed that black patients are up to 96% more likely to undergo surgery at lower-quality hospitals in segregated regions. Disseminated cancer is defined by NSQIP as a patient who has a primary cancer that has disseminated to a major organ, and the patient and the presence of multiple metastases indicate that the cancer is widespread, fulminant, or near terminal. Patients with this preoperative diagnosis were found to have a 1.85 times increased risk of reoperation in the present study. Wallis et al21 similarly describe disseminated cancer as a predisposing factor for developing major complications after surgery among patients undergoing urologic surgery, many of whom require reoperation. Preoperative radiotherapy for malignant neoplasm, although associated with URs in our bivariate analyses, did not have a significant association in our multivariate regression analysis. However, other studies have shown that preoperative radiation therapy increases the risk of complications, including URs.22,23 A possible explanation for this discrepancy is the small sample size; less than 25% of our total data included data on radiotherapy. Future analyses using more complete data would provide a better understanding of the association between radiotherapy and reoperation. In addition, having greater than the median time (415 minutes) in the operating room or greater than the median time to discharge (8 days) was strongly associated with increased risk of reoperation in multivariate regression analysis.
Postoperative complications were most strongly associated with risk of reoperation. Operations complicated by superficial or deep surgical site infection had increased risk of reoperation. Surgical site infection in head and neck operations has been shown to be associated with contaminated field, advanced cancer stage, flap reconstruction, or long operative time.24 In colorectal surgery, a graded association between increasing operation duration and incidence of deep and superficial SSI has been identified.25,26 Both factors were shown to be independent significant risk factors in our analysis. However, superficial SSI was not significantly associated with increased risk. Postoperative wound dehiscence as a surgical complication was also found to be associated with URs, as expected given that every dehiscence requires reoperation. Previous studies of the head and neck have shown that these failures of closure occur at the highest rates in free flaps, a modality that is commonly used in the major surgical procedures described in the present study.27 In addition, postoperative ventilation for more than 48 hours was independently associated with URs. This is consistent with previous studies that have found prolonged ventilation to be a risk factor for reoperation and readmission.28,29 Prolonged ventilation itself may be a sequela of unfavorable preoperative and intraoperative factors. Patients with certain preexisting comorbid conditions may already be prone to ventilator dependence as well as other postoperative complications. Thus, prolonged ventilation may simply serve as a surrogate marker of a UR. Furthermore, patients who require prolonged ventilation, and thus prolonged bedridden hospital stay, are likely to acquire more hospital-associated complications that can subsequently contribute to the need for reoperation.
Owing to this significant variance of UR rates by procedure, we stratified head and neck cases by major surgery type. We found that the operative procedure with the highest UR rate among the procedures examined was pharyngectomy (8 of 46 [17.4%]), followed by glossectomy (95 of 632 [15.0%]), laryngectomy (53 of 399 [13.3%]), and mandibulectomy (25 of 240 [10.4%]). Using the model developed in the present study for all major head and neck surgical procedures, we identified differences in risk factors. For glossectomy, after adjusting for other variables, a UR was associated with other race (defined as Asian, Pacific Islander, or Hispanic individuals), organ or space SSI, wound dehiscence, and ventilator dependence for more than 48 hours. Most reoperations after glossectomy are associated with failure of oral floor reconstruction and free flaps after total glossectomy.30 For mandibulectomy, a UR was associated with superficial SSI and wound dehiscence as well as with risk of failure of free flaps and decrease in wound healing secondary to bisphosphonate use.31 For laryngectomy, a UR was associated with male sex, preoperative hypertension, organ or space SSI, and wound dehiscence. A study by Megwalu and Sikora32 of patients with late-stage laryngeal carcinoma in the Survival, Epidemiology, and End Results (SEER) database found an outcome disparity in laryngectomy by sex, which we also found in risk of reoperation. Although the sample size of patients identified with pharyngectomy in the present study was too small to draw significant conclusions from, our preliminary data suggested that the risk factors for reoperation following this procedure were similar to those of other major head and neck surgical procedures.
To our knowledge, the present study is the first to assess the factors that predispose patients to URs in major head and neck surgical procedures. Here, we found that factors with the strongest risk of reoperation were postoperative complications, such as infections of the surgical site and wound dehiscence. This information will help surgeons to counsel patients on reoperation risk reduction in the perioperative period. Given the study design and multicenter nature, the analysis has its limitations. The ACS-NSQIP database provides the most comprehensive and diverse survey of surgical operative characteristics. However, this strength relies on the integrity of the data input by the many contributing health care systems. In addition, reoperations are tracked in the database up to 30 days postoperatively; thus, the present study may have missed some late sequelae of the surgical procedures that may lead to reoperation, leading to underestimation of true rates. Our analysis was also confined by the variables collected by the program, there are certainly other important factors that we were unable to assess, such as the experience of the surgeon and the teaching status of the hospital. In addition, given that most reoperations were drainage of fluid collections, it would be important to understand the role of deep vein thrombosis prophylaxis and preoperative blood thinning medications regarding the risk of a UR. These variables are not collected by NSQIP but would be important to assess in future study. Despite these limitations, the findings of the present study should help guide the modification and adaptation of the identified risk factors to decrease the burden that URs place on patients, surgeons, and the health care system.
Unplanned reoperation is an important indicator of surgical quality in an increasingly outcome-driven health care system. We provided, to our knowledge, the first analysis of risk factors associated with URs in major head and neck surgery, a field with the highest burden of URs among surgical specialties. Identification of these preoperative, intraoperative, and postoperative factors are essential for the understanding and implementation of surgical risk stratification and modification. We found black race, disseminated cancer, prolonged operation time and length of hospital stay, SSI in the superficial and deep spaces, wound dehiscence, and ventilator dependence of more than 48 hours after surgery to be significantly associated with increased risk of reoperation. We believe that these findings can be used to better understand surgical risk of reoperation in head and neck surgery.
Accepted for Publication: July 5, 2018.
Corresponding Author: Richard Chan Woo Park, MD, Department of Otolaryngology–Head and Neck Surgery, Rutgers New Jersey Medical School, 90 Bergen St, Ste 8100, Newark, NJ 07103 (firstname.lastname@example.org).
Published Online: September 27, 2018. doi:10.1001/jamaoto.2018.2049
Author Contributions: Dr Chan Woo Park had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Sangal, Nishimori, Zhao, Siddiqui, Chan Woo Park.
Acquisition, analysis, or interpretation of data: Sangal, Nishimori, Siddiqui, Baredes, Chan Woo Park.
Drafting of the manuscript: Sangal, Nishimori, Chan Woo Park.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sangal, Nishimori, Zhao, Siddiqui.
Administrative, technical, or material support: Siddiqui, Baredes, Chan Woo Park.
Supervision: Sangal, Siddiqui, Chan Woo Park.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Meeting Presentation: This study was presented at the AHNS 2018 Annual Meeting; April 19, 2018; National Harbor, Maryland.