What factors are associated with the development of postoperative delirium in patients undergoing head and neck free flap reconstruction?
In this cohort study of 515 patients undergoing free flap reconstruction, increased age, male sex, increased operative time, advanced nodal disease, and tobacco use were associated with an increased risk of developing postoperative delirium after head and neck free flap reconstruction. Preoperative alcohol abstinence was identified as a protective factor.
Risk stratification for postoperative delirium provides a tiered framework for quality improvement protocols to reduce the complications and costs associated with postoperative delirium, and preoperative withdrawal from alcohol use may prevent the development of postoperative delirium.
Postoperative delirium (POD) is associated with an increased rate of adverse events, higher health care costs, and longer hospital stays. At present, limited data are available regarding the risk factors for developing POD in patients undergoing head and neck free flap reconstruction. Identification of patients at high risk of developing POD will allow implementation of risk-mitigation strategies.
To determine the frequency of and risk factors associated with POD in patients undergoing free flap reconstruction secondary to head and neck disease.
Design, Setting, and Participants
This retrospective cohort study included 515 patients undergoing free flap reconstruction from January 1, 2006, through December 31, 2012, at the James Cancer Hospital and Solove Research Institute, The Ohio State University Comprehensive Care Center, a tertiary care cancer hospital. Preoperative, intraoperative, and postoperative data were collected retrospectively. Data from January 1, 2006, through December 31, 2012, were analyzed, and the final date of data analysis was January 8, 2018.
Head and neck free flap reconstruction.
Main Outcomes and Measures
The primary outcome was the development of POD as defined by the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition). Univariable and multivariable logistic regression were used to identify risk factors associated with POD.
Five hundred fifteen patients underwent free flap reconstruction during the study period (66.2% male; mean [SD] age, 60.1 [12.8] years). Of these, 56 patients (10.9%) developed POD. On multivariable analysis, risk factors associated with POD included increased age (odds ratio [OR], 1.06; 95% CI, 1.02-1.11), male sex (OR, 5.02; 95% CI, 1.47-17.20), increased operative time (OR for each 1-minute increase, 1.004 [95% CI, 1.001-1.006]; OR for each 1-hour increase, 1.26 [95% CI, 1.08-1.46]), advanced nodal disease (OR, 3.00; 95% CI, 1.39-6.46), and tobacco use (OR, 7.23; 95% CI, 1.43-36.60). Preoperative abstinence from alcohol was identified as a protective factor (OR, 0.24; 95% CI, 0.12-0.51).
Conclusions and Relevance
This study identified variables associated with a higher risk of developing POD. Although many of these risk factors are nonmodifiable, they provide a target population for quality improvement initiatives. Furthermore, preoperative alcohol abstinence may be useful in preventing POD.
Postoperative delirium (POD) is a cerebral disturbance characterized by fluctuating patterns of disorganized thinking, altered levels of consciousness, and varying degrees of inattention.1,2 Three forms exist, including hyperactive (agitation, aggressiveness, and hallucination), hypoactive (decreased attention, lethargy, and apathy), and mixed. The signs and symptoms typically develop within the first 72 hours after surgery and last for several days, with a minority of cases persisting as cognitive dysfunction.3 The development of POD is a well-recognized complication that occurs after major surgical procedures. The reported incidence of POD after head and neck surgery ranges from 11% to 26%.4-11 Postoperative delirium results in increased health care costs and is associated with a higher rate of adverse events and longer length of stay.12-14 The cornerstone for reducing the risk and severity of POD is early recognition of at-risk patient populations alongside active perioperative management.
Patients receiving free flap reconstruction after head and neck surgery are at an increased risk of developing POD owing to the length of anesthesia, high rates of malnutrition, and associated alcohol use disorders.15 Although many studies have focused on free flap failure as an important complication, few studies address the medical complications, such as POD, in this population. Previously, our group has reported 2 studies using this database and cohort describing (1) the general predictors of complications in head and neck free flap patients and (2) predictors of hospital quality metrics.16,17 The development of any medical or surgical complication was the most important predictor of increased hospital length of stay, readmission, or unplanned return to the operating room. Wound healing and infectious complications were the most common complication in the cohort and were predicted by longer operative time, advanced age, and comorbidities. A subset of the patients in these studies developed POD. The objective of this study is to determine the frequency of and risk factors associated with POD in these patients and to specifically evaluate the role of alcohol abuse as a contributing factor.
This retrospective review included all cases requiring head and neck free flap reconstruction (including patients requiring secondary reconstruction for nonmalignant conditions) from January 1, 2006, through December 31, 2012. All surgical procedures were performed by 4 head and neck reconstructive surgeons (E.O., A.A., T.N.T., and M.O.). We used electronic health records, the primary charting system at our institution, and archived paper records to identify the most relevant data points. Data retrieval was performed by 2 reviewers (B.T. and J.S.) independent of the attending surgeons, and surgeons were blinded to the individual patient outcomes. The study protocol was granted approval with a waiver of informed consent by the Research Ethics Board at the James Cancer Hospital and Solove Research Institute of The Ohio State University Comprehensive Care Center, Columbus.
Patients treated before February 17, 2009, were admitted directly to the intensive care unit after surgery. Patients treated after this date were admitted directly to a head and neck surgical–specific floor18 unless they received a craniotomy or a thoracotomy. The nursing staff on this floor had a ratio of nurse to patients of 1:2. The free flap protocol included flap checks every hour for the first 24 hours, followed by every 2 hours for the next 24 hours, then every 4 hours for the remainder of the hospitalization. Tracheostomy tube change and downsize, if needed, was performed on postoperative day 5 with subsequent capping and decannulation as appropriate. For patients with mucosal reconstruction, enteral feedings were administered via a nasogastric tube or a gastrostomy tube. If clinically appropriate, speech language pathology assessment was performed on postoperative day 6 to evaluate for potential oral intake.
A preoperative medical evaluation was performed for most of the patients, including collection of medical history, physical examination findings, and laboratory data. All patients had relevant intraoperative and postoperative data collected during hospitalization and at outpatient clinic visits. Variables were divided into preoperative, intraoperative, and postoperative groups. Preoperative variables included age, sex, tumor site, Charlson Comorbidity Index (scores range from 0-33, with higher scores indicating more comorbidities), smoking history, history of an alcohol use disorder, presence of an active alcohol use disorder, induction chemotherapy, history of radiotherapy, history of head and neck cancer, and specific assessment of comorbidities (type 2 diabetes, coronary artery disease, chronic obstructive pulmonary disease, chronic kidney disease, hypothyroidism, and malnutrition). Intraoperative variables included duration of the operation, type of free flap, estimated blood loss, flap ischemia time, volume of intravenous fluid (crystalloid and colloid) infused, and volume of blood products (packed red blood cells and fresh frozen plasma) transfused. Postoperative variables included length of time spent in the intensive care unit, postoperative complications, and final cancer staging.
Delirium was identified by documentation that corresponded to the diagnostic criteria as stated in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition), the current edition at the time of the study.19 Symptoms of POD were characterized by (1) disturbance of consciousness and a change in cognitive capacity, (2) symptom development during a short period (hours to days), and (3) fluctuating mental status changes. During hospitalization, patients displaying symptoms of delirium were evaluated separately by at least 2 head and neck surgeons, and if all physicians agreed the symptoms were consistent with POD, a diagnosis was made. Distinctions were not made among the hyperactive, hypoactive, and mixed subtypes of delirium.
Data from January 1, 2006, through December 31, 2012, were analyzed, and the final date of data analysis was January 8, 2018. We compared results for each risk factor using a univariable logistic regression model and considered these findings for the multivariable analysis using a backward selection algorithm if univariable P ≤ .10 (2 sided). Collinearity of variables was assessed using variance inflation factors before simultaneously entering the values into the multivariable analysis to prevent the formation of an unstable regression model. All regression model results were reported as an odds ratio (OR) with an associated 95% CI. All analyses were performed using SAS software (version 9.2; SAS Institute, Inc).
A total of 515 patients undergoing head and neck free flap reconstruction were included in the study. The mean (SD) age of the patients was 60.1 (12.8) years, with 341 male (66.2%) and 174 female (33.8%). At the time of surgery, 202 patients (39.2%) smoked tobacco and 194 (37.7%) had an alcohol use disorder. The most common disease location was the oral cavity (338 [65.6%]) followed by the oropharynx, hypopharynx, or larynx (101 [19.6%]). The mean (SD) operative time was 544.5 (130.2) minutes, with 173 operations (33.6%) completed in no more than 8 hours. The mean (SD) blood loss was 457.5 (394.0) mL, with 380 patients (73.8%) losing no more than 500 mL. Blood transfusions were required in 160 patients (31.1%). The mean (SD) volume of crystalloid infusion was 5508.4 (1962.8) mL. Postoperative stays in the intensive care unit were required for 146 patients (28.3%). All demographic and clinical characteristics of the study patients are provided in Table 1.
Postoperative delirium occurred in 56 patients (10.9%). Univariable and multivariable analyses revealed many significant risk factors for the development of POD (Table 2). All variance inflation factors were less than 10. On univariable analysis, we identified higher CCI (OR, 1.23; 95% CI, 1.12-1.36), specific comorbidities such as cardiac disease (OR, 2.12; 95% CI, 1.17-3.81) and pulmonary disease (OR, 3.03; 95% CI, 1.62-5.68), blood loss (OR, 2.36; 95% CI, 1.34-4.17), and infusion of crystalloid (OR, 1.0002; 95% CI, 1.0000-1.0003) or blood (OR, 1.77; 95% CI, 1.00-3.12) products as significant risk factors. These factors did not reach statistical significance on multivariable analysis. Increased age (OR, 1.03; 95% CI, 1.01-1.06), male sex (OR, 10.48; 95% CI, 3.23-33.99), longer operative time (OR, 1.004; 95% CI, 1.002-1.006), regional nodal metastases (OR for N1-N2 disease, 2.46 [95% CI, 1.21-4.98]; OR for N2b-N3 disease, 3.19 [95% CI, 1.64-6.21]), and active tobacco use (OR, 11.82; 95% CI, 2.77-50.42) were all identified as significant risk factors on univariable analysis, and these findings were sustained on multivariable analysis. The absence of an alcohol abuse disorder at the time of surgery was identified as a protective risk factor on univariable and multivariable analysis (OR, 0.20; 95% CI, 0.11-0.37). Tumor size, tumor location, tumor recurrence, type of free flap, and preoperative radiotherapy were not significant risk factors for the development of POD.
The mean (SD) age for patients developing POD was significantly higher than for patients who did not develop POD (64.7 [12.4] vs 59.5 [12.8] years), and most of the patients who developed POD were male (53 [94.6%]). The mean (SD) operative time was 65 minutes longer for patients developing POD (602.4 [107.0] vs 537.3 [131.1] minutes). Regional nodal disease conferred an increased risk with 16 of 107 patients with N1 and N2a disease (15.0%) and 20 of 108 with N2b and N3 disease (18.5%) developing POD compared with 20 of 300 with N0 status (6.7%), independently of operative time. Tobacco use was associated with a higher rate of POD development, with an elevated risk among active (35 of 200 [17.5%]) and former (19 of 37 [51.4%]) smokers compared with nonsmokers (2 of 113 [1.8%]). A higher mean (SD) Charlson Comorbidity Index was found to be a risk factor for the development of POD on univariable analysis (6.09 [2.56] vs 4.81 [2.22]) but did not reach statistical significance with multivariable analysis. Examination of comorbidities by organ system revealed the presence of cardiac and pulmonary disease to be significant risk factors.
Each year of increased age elevates the risk of developing POD by a factor of 1.06 (95% CI, 1.02-1.11). Male sex (OR, 5.02; 95% CI, 1.47-17.20) correlated with a higher rate of POD development. Active tobacco use was a significant risk factor (OR, 7.23; 95% CI, 1.43-36.60), whereas abstinence from alcohol was associated with prevention of POD (OR, 0.24; 95% CI, 0.12-0.51). For every 1-minute and 1-hour increase in operative time, the odds of developing POD increased multiplicatively by 1.004 (95% CI, 1.001-1.006) and 1.26 (95% CI, 1.08-1.46), respectively. Extensive regional nodal disease was a significant risk factor, with N2b and N3 disease (OR, 3.00; 95% CI, 1.39-6.46) having a higher association with the development of POD than N1 and N2a disease (OR, 2.61; 95% CI, 1.17-5.81).
The overall incidence of POD in our study was 10.9%, which is comparable to the range of 11% to 26% reported in previous studies.4-11 Multivariable analysis suggests that increasing age, male sex, longer operative time, regional nodal metastases, and active tobacco use are significant risk factors for the development of POD. In addition, preoperative abstinence from alcohol was identified as a negative risk factor for developing POD.
Despite the existence of numerous models for stratifying patients at risk of developing POD, no single model is ready for clinical implementation.11,20 This situation exists largely because the individual risk factors for developing POD are not well defined. In the several reported studies of patients with head and neck cancer, the results are significantly heterogeneous, with various independent risk factors being identified through multivariable analysis (increased age, male sex, increased operative time, increased blood loss, active alcohol use, pain control, presence of psychiatric disorder, and marital status).4-11,21 In addition, most of the patients in those studies did not undergo free flap reconstruction. A recently performed meta-analysis regarding risk factors for POD after major head and neck surgery identified 10 statistically significant risk factors for developing POD, including increased age, being older than 70 years, male sex, duration of surgery, history of hypertension, blood transfusion, tracheotomy, American Society of Anesthesiologists classification of 3 or greater, neck dissection, and free flap reconstruction.22
Since Marcantonio et al15 reported the predictive model of POD, the binary value of 70 years or older has been used as a variable to report POD. The studies in head and neck surgery confirm this variable as reliable. Booka et al4 identified being older than 70 years as the only risk factor associated with developing POD on multivariable analysis. Hasegawa et al10 showed that increasing age in their entire cohort was not a risk factor for developing POD (mean [SD] age of 49.0 [8.9] years with delirium vs 67.4 [12.9] years without delirium), but subgroup multivariable analysis identified patients 75 years or older as having an increased risk of developing POD (OR, 6.83; 95% CI, 2.47-18.90). In addition, Wang et al7 stratified the age of patients into decades of life and found that the incidence of POD after major head and neck surgery linearly increased with each decade of life (R = 0.97). Our results agree with those of most previous studies. We identified increasing age as a significant risk factor for developing POD, with each year of increased age elevating the risk multiplicatively by a factor of 1.06 (95% CI, 1.02-1.11).
Patients undergoing head and neck cancer resection with free flap reconstruction are invariably subjected to lengthy operative times. Previous studies evaluating operative time as a risk factor for POD have reported diverse results. Shah et al5 and Yamagata et al9 identified surgical times longer than 6 and 10 hours, respectively, to be significant risk factors for developing POD on multivariable analyses, whereas Hasegawa et al10 reported increased operative time as a risk factor on univariable analysis. However, increased length of operative time was not found to be a risk factor in the studies by other investigators.4,6,11 In our study, increased operative time was a significant risk factor for developing POD on multivariable analysis, with the mean surgical time for patients developing POD being 602.4 minutes compared with 537.4 minutes in those not developing POD. For every 1-minute and 1-hour increase in operative time, the odds of developing POD increased multiplicatively by 1.004 (95% CI, 1.001-1.006) and 1.26 (95% CI, 1.08-1.46), respectively. Therefore, surgery performed at a tertiary academic institution where 2-surgeon teams are used may reduce the incidence of POD by decreasing the operative time.
Alcohol and tobacco abuse are the main risk factors for developing head and neck squamous cell carcinoma. In keeping with previous studies,5,7,8,23 our results indicate a higher rate of POD in patients with an active alcohol use disorder. Our study is unique by demonstrating that, whether or not a patient has a history of an alcohol use disorder, abstinence around the time of surgery significantly reduces the risk of developing POD. In addition, a recent case-control study23 conducted at our institution on the use of a preoperative alcohol-abstinence contract reduced the rate of POD from 73% in the noncontracted group to 0 in the contracted group. In that study,23 15 high-risk alcohol misusers undergoing free flap reconstruction for head and neck cancer entered into an abstinence contract compared with 30 high-risk alcohol users who were treated with a therapeutic withdrawal protocol based on the Clinical Institute Withdrawal Assessment for Alcohol. With a mean of 14.8 days of abstinence before surgery, none of the contracted group developed POD compared with 22 (73.3%) in the noncontracted group. Together these results strongly advocate for the implementation of preoperative alcohol abstinence strategies to prevent the occurrence of POD.
Most individual studies on POD after head and neck surgery do not identify male sex as a risk factor. However, a meta-analysis of these studies suggests that male sex is a risk factor for developing POD (pooled OR, 1.92; 62% heterogeneity).22 Within other surgical disciplines, the results are similarly heterogeneous. A meta-analysis of patients undergoing vascular surgery24 identified male sex as a risk factor for developing POD, whereas a meta-analysis of patients undergoing gastrointestinal surgery25 found sex to have no influence. Our data indicate that male sex is a significant risk factor (OR, 5.02; 95% CI, 1.47-17.20) and closely correlates to the report by Shiiba et al6 (OR, 5.49; 95% CI, 1.58-19.01). The discrepancy among studies is potentially related to a higher incidence of confounding risk factors present in the male population that were not always controlled for with study design and statistical analysis.
Strengths and Limitations
The present study is the largest single study, to our knowledge, to analyze the incidence and risk factors of POD in patients undergoing head and neck free flap reconstruction. The strengths of our study include the size of our database with granular data points and identifiable associations between modifiable risk factors and the development of POD. This study has several limitations, including but not limited to its retrospective nature. We did not use a mental health professional to diagnose POD in our patients, so delirium in many patients may not have been diagnosed, especially the hypoactive subtype. Also, potential unmeasured confounding remains from variables not measured in our study (eg, preoperative pain control, sleep disturbance, comorbid psychiatric history, marital status, sedative use). In particular, the presence of preoperative cognitive impairment, a commonly identified risk factor for POD, was not objectively assessed. Future studies should account for these variables to provide further insight into risk factors for developing POD.
Our findings suggest that many identifiable risk factors are associated with developing POD after free flap reconstruction for head and neck disease. Nonmodifiable risk factors include increased age, male sex, and increased regional disease. Modifiable risk factors include operative time and active alcohol and tobacco use disorders. Prospective studies that implement preoperative predictive nomograms with risk reduction strategies are needed to realize the clinical meaning of these findings. Multidisciplinary collaboration with anesthesiologists, geriatricians, and internists to create a standardized risk reduction protocol is warranted. Further analysis on the cost-effectiveness of an active risk reduction protocol would be of benefit before expanding to an institutional level. Head and neck surgeons should stratify patients with the above characteristics into a high-risk category and closely monitor them during the perioperative period. Preoperative alcohol abstinence contracts and 2-surgeon teams to reduce the operative time may reduce the incidence of POD.
Accepted for Publication: November 8, 2018.
Published Online: January 3, 2019. doi:10.1001/jamaoto.2018.3820
Correction: This article was corrected on February 28, 2019, to correct an omission of 2 reference citations and to add discussion of these related studies to the article introduction.
Corresponding Author: Matthew Old, MD, Department of Otolaryngology–Head & Neck Surgery, Division of Head & Neck Oncology, The Ohio State University, James Cancer Hospital and Solove Research Institute, 456 W 10th Ave, Room 4a, Columbus, OH 43210 (email@example.com).
Author Contributions: Dr Old had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Tweel, Sitapara, Ozer.
Study concept and design: Old.
Acquisition, analysis, or interpretation of data: Densky, Eskander, Kang, Chan, Tweel, Sitapara, Agrawal, Carrau, Rocco, Old.
Drafting of the manuscript: Densky, Eskander, Old.
Critical revision of the manuscript for important intellectual content: Eskander, Kang, Chan, Tweel, Sitapara, Ozer, Agrawal, Carrau, Rocco, Old.
Statistical analysis: Densky, Tweel, Sitapara, Old.
Administrative, technical, or material support: Densky, Kang, Tweel, Sitapara, Carrau, Rocco, Old.
Study supervision: Eskander, Ozer, Carrau, Old.
Conflict of Interest Disclosures: None reported.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Washington, DC: American Psychiatric Publishing; 2013.
LS. Postoperative delirium and postoperative cognitive dysfunction in the elderly: what are the differences? Minerva Anestesiol
. 2011;77(7):742-749.PubMedGoogle Scholar
MK. Development of a nomogram for predicting the probability of postoperative delirium in patients undergoing free flap reconstruction for head and neck cancer. Eur J Surg Oncol
. 2017;43(4):683-688. doi:10.1016/j.ejso.2016.09.018PubMedGoogle ScholarCrossref
YR. Health economic implications of perioperative delirium in older patients after surgery for a fragility hip fracture. J Bone Joint Surg Am
. 2015;97(10):829-836. doi:10.2106/JBJS.N.00724PubMedGoogle ScholarCrossref
et al. Intensive care unit versus non-intensive care unit postoperative management of head and neck free flaps: comparative effectiveness and cost comparisons. Head Neck
. 2014;36(4):536-539. doi:10.1002/hed.23325PubMedGoogle ScholarCrossref
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed, text revision. Washington, DC: American Psychiatric Association; 2000.
LCC, van Meenen
DMP, de Rooij
SE, ter Riet
G. Risk prediction models for postoperative delirium: a systematic review and meta-analysis. J Am Geriatr Soc
. 2014;62(12):2383-2390. doi:10.1111/jgs.13138PubMedGoogle ScholarCrossref
et al. Comparison of clinical outcomes following head and neck surgery among patients who contract to abstain from alcohol vs patients who abuse alcohol. JAMA Otolaryngol Head Neck Surg
. 2017;143(12):1181-1186. doi:10.1001/jamaoto.2017.0553PubMedGoogle ScholarCrossref
J. Systematic review and meta-analysis of risk factors for postoperative delirium among older patients undergoing gastrointestinal surgery. Br J Surg
. 2016;103(2):e21-e28. doi:10.1002/bjs.10062PubMedGoogle ScholarCrossref