Complications, Hospital Length of Stay, and Readmission after Total Laryngectomy

Abstract

Background

The purpose of this study was to describe the 30-day incidence of complications after total laryngectomy (TL) in a high-volume institution and their impact on hospital length of stay (LOS) and readmission rates.

Methods

A retrospective, cohort study of all patients who underwent TL at MD Anderson Cancer Center from January 1, 2010 through June 30, 2013 was conducted. Patient demographics, treatment history, LOS, and 30-day post-TL complications and readmissions were extracted from the medical record. Univariate associations were analyzed and stepwise backward selection methods were used to fit multivariate models.

Results

245 patients were included. Complications occurred in 83 patients (33.9%) including 3 (1.2%) deaths. Wound complications occurred in 53 patients (21.6%), of which 34 were pharyngocutaneous fistula (PCF) (13.9% overall). 34 patients (13.9%) were readmitted within 30 days. Multivariate analysis revealed the following: wound complications were associated with former (OR 5.1, p=0.03) and current (OR 5.8, p=0.02) smokers; PCF was associated with prior wide-field radiation (OR 3.1, p=0.01) but not prior narrow-field (larynx only) radiation (OR 1.4, p=0.61); LOS was associated with type of flap (p=0.002) and post-operative hematoma (p=0.05); and readmissions were associated with pre-operative hypoalbuminemia (p=0.003) and post-operative wound complications (p<0.001).

Conclusions

Complications occurred in nearly one third of TL patients, particularly among patients with poor wound healing risk factors such as prior smoking and radiation. Expectedly, LOS was longer among reconstructed patients. Readmission was associated with hypoalbuminemia and post-operative wound complications. These data inform quality improvement efforts and counseling of high-risk patients undergoing TL.

Introduction

Contemporary treatment of advanced laryngeal carcinoma has gradually evolved, in large part due to non-surgical treatment intensification. However, total laryngectomy (TL) still serves a critical role, whether in the advanced primary tumor, as salvage therapy, or for a dysfunctional larynx from prior treatment^1–3. The need for TL persists for the dysfunctional larynx despite continuing efforts to improve laryngeal cancer outcomes and preserve a functional organ through radiation dose optimization strategies^4–7. More commonly, TL remains a primary means of treatment for locally advanced disease and is often the only viable option for cure in the salvage setting. Yet, there remains a relative paucity of outcomes data from large samples of patients treated with TL in the modern era^8–10.

It is well established that the contemporary cohort of patients undergoing TL is different than many historical cohorts prior to the era of organ preservation therapy^{11, 12}. Moreover, TL is increasingly concentrated in higher-volume, academic centers due to a variety of influences including reimbursement, treatment complexity, and the necessity for multidisciplinary care to achieve optimal outcomes^{13, 14}. Lastly, as improved quality of care and resource utilization become clear mandates from public and private payers, institutions, and patients, the ability to anticipate initial treatment outcomes for high acuity cases such as TL based on pre-operative risk stratification is increasingly important.

This study was conceived to help address the need for resource optimization and quality improvement at our own institution, to provide insight into a group of patients amenable to clinical trial interventions aimed at outcome improvement, and to provide insight into the post-operative complications, LOS, and readmission rates in a contemporary cohort of TL patients at a high-volume institution.

Methods

Patient selection and clinical variables

This is an IRB-approved, retrospective cohort analysis that reviewed the medical records of all consecutive patients who underwent TL at MD Anderson Cancer Center between January 1, 2010 and June 30, 2013. A waiver of informed consent was obtained. Pre-operative clinical variables of interest were collected, including age, sex, cancer stage, primary tumor location and histology, smoking and alcohol history, pre-operative albumin and thyroid stimulating hormone (TSH) levels, past medical history including diabetes mellitus and immunosuppression, any prior interventions such as pre-operative tracheotomy, partial laryngeal surgery, therapeutic radiation, and/or systemic treatment, and reason for TL (i.e. primary tumor, salvage treatment, or non-functional larynx). Operative details including extent of laryngopharyngeal resection, completion of neck dissection(s), method of pharyngeal closure (vertical, horizontal, or “T”), use and type of regional or free tissue reconstruction, and placement of primary tracheoesophageal prosthesis (TEP) were tabulated. 30-day post-operative complications, LOS, and readmissions within 30 days were also reviewed.

Institutional Care Pathways

While care of individual patients may vary according to surgeon preference, some post-operative management is standardized by plan of care pathways for patients undergoing TL at MD Anderson. Patients undergoing primary TL without flap reconstruction receive peri-operative antibiotics for at least 24 hours post-operatively and often initiate an oral diet on post-operative day 7 with or without modified barium swallow (MBS). Patients requiring flap reconstruction are typically maintained on broad spectrum antibiotics for at least 7 days, and undergo MBS at 2 weeks or more, depending on the type of reconstruction and previous non-surgical therapy, prior to the initiation of an oral diet.

Classification of Variables

Several variables were subject to further grouping in order to assist with analysis and interpretation of data. Primary tumor sites were divided into laryngeal (comprised of supraglottic, glottic, and subglottic), pharyngeal (oro- and hypopharyngeal), and thyroid/other (tracheal, esophageal). Immunosuppression was defined as patients taking chronic glucocorticoids, biologic response modifiers, antimetabolites, or immunosuppressive transplant medications at the time of surgery. Patients with prior non-surgical therapy were divided into prior narrow-field radiation (treatment of larynx only), prior wide-field radiation alone (treatment of larynx/pharynx and draining lymphatics), and prior wide-field radiation with systemic therapy. Extent of surgical resection was classified as TL alone, TL with partial pharyngectomy and/or glossectomy, and total laryngopharyngectomy (TLP). Wound complications were defined as development of PCF or cellulitis, neck abscess, or neck incisional dehiscence in the absence of evidence indicating a salivary leak. Lastly, hypoalbuminemia and hypothyroidism were defined according to institutional clinical laboratory thresholds of albumin<3.5g/dL and TSH>4.2μIU/mL, respectively.

Statistical analysis

Patient, tumor, prior/current treatment characteristics, complications, LOS, and readmission variables were summarized by descriptive statistics. Covariates exhibiting significant association with complications, LOS, and readmission were identified initially using univariate analysis. Multivariate regression models were fit using backwards stepwise method for variable entry and removal¹⁵. Covariates that attained a p-value <0.20 were subsequently considered for multiple logistic and linear regression analysis for complications or readmission and LOS, respectively. A p-value <0.05 conferred statistical significance. Final results are reported using the set of covariates that attained statistical significance in multiple regression analysis. Statistical analyses were performed using STATA v14.0 (StataCorp LP, College Station, TX) data analysis software.

Results

Patients and Prior Treatment

Two hundred forty-five patients were included in the analysis. Table 1 lists patient demographics, tumor specifics, prior cancer treatment, and relevant past medical history. Mean age was 64.1 years, 79% were male, and 70% had primary tumors in the larynx. Pre-operatively, 20% of patients were hypothyroid, 32% had hypoalbuminemia, and 29% had diabetes mellitus and/or were immunosuppressed. 56% were former smokers and 30% were current smokers. 62% of patients had prior radiation, composed of 26% “narrow-field” and 74% “wide-field” patients with or without concurrent chemotherapy. Patients with prior radiation were an average of 4.3 (mean 2.1) years from completion of treatment at the time of surgery. 67% of patients had either regional or free tissue transfer reconstruction, 73% of which had prior radiation. Twenty patients were treated with neoadjuvant chemotherapy prior to primary TL. 85% of patients (79/93) undergoing primary TL had T4 tumors.

Table 1.

Patient Characteristics (n = 245), Univariate Analysis Results^*

	# Patients (%)	Any Complication	Wound Complication	Fistula	Readmission
Age (mean)	64.1	0.99 (0.97-1.02)	1.01 (0.98-1.04)	1.02 (0.98-1.06)	1.01 (0.97-1.04)
Sex		1.43 (0.76-2.69)	1.28 (0.62-2.61)	1.41 (0.61-3.23)	1.17 (0.49-2.76)
Male	193 (78.8)
Female	52 (21.2)
Primary Tumor Site
Larynx	172 (69.8)	-	-	-	-
Pharynx	57 (23.3)	1.69 (0.92-3.13)	1.63 (0.82-3.26)	2.36 (1.08-5.16)	1.44 (0.64-3.26)
Thyroid/Other	17 (6.9)	0.67 (0.21-2.14)	1.29 (0.39-4.20)	1.07 (0.23-5.03)	0.90 (0.19-4.21)
Prior Radiation
None	93 (38.0)	-	-	-	-
Narrow-field	39 (15.9)	0.57 (0.23-1.39)	0.66 (0.22-1.93)	1.40 (0.39-5.10)	0.56 (0.15-2.12)
Wide-field	113 (46.1)	1.52 (0.85-2.70)	1.69 (0.87-3.30)	3.14 (1.28-7.69)	1.36 (0.62-2.98)
Prior partial laryngeal surgery	30 (12.2)	0.26 (0.09-0.79)	0.37 (0.11-1.26)	0.66 (0.19-2.31)	0.95 (0.31-2.91)
Smoking history
Never	34 (13.9)	-	-	-	-
Former	138 (56.3)	2.06 (0.83-5.07)	4.83 (1.10-21.3)	6.6 (0.86-50.7)	2.87 (0.64-12.9)
Current	73 (29.8)	2.4 (0.92-6.24)	5.63 (1.23-25.8)	5.24 (0.64-42.7)	2.84 (0.59-13.6)
Alcohol history >2 drinks/day
Never	118 (48.2)	-	-	-	-
Former	59 (24.1)	1.95 (1.01-3.73)	1.49 (0.70-3.14)	1.65 (0.68-4.02)	1.15 (0.47-2.78)
Current	68 (27.7)	1.26 (0.67-2.40)	1.34 (0.65-2.78)	1.56 (0.66-3.70)	0.97 (0.40-2.34)
Hypoalbuminemia (<3.5g/dL)	79 (32.2)	1.31 (0.75-2.28)	1.10 (0.58-2.10)	0.86 (0.39-1.89)	2.77 (1.32-5.78)
Hypothyroid (TSH>4.2μIU/mL)	47 (19.7)	1.12 (0.58-2.19)	0.79 (0.36-1.76)	1.06 (0.43-2.62)	1.11 (0.45-2.74)
Diabetes mellitus	40 (16.3)	0.51 (0.23-1.13)	0.89 (0.38-2.06)	1.12 (0.43-2.90)	0.65 (0.22-1.95)
Immunosuppression	36 (14.7)	0.72 (0.33-1.57)	0.41 (0.14-1.21)	0.75 (0.25-2.26)	0.52 (0.15-1.81)

^*Odds Ratio (95% Confidence Interval). Abbreviations: TSH, thyroid stimulating hormone

Operative Details

Ninety-three patients (38%) underwent TL for primary cancer treatment, 137 (56%) were salvage procedures, and 15 (6.1%) laryngectomies were performed for chronic laryngeal dysfunction secondary to prior organ preservation therapy. 43.7% had TL alone without partial or total pharyngectomy. 85.7% had concurrent neck dissection. Among all patients, 67% underwent reconstruction with regional or free tissue transfer, including 20 pectoralis major (17 were onlay), 124 anterolateral thigh, 10 radial forearm, and 10 other flaps (including three jejunal flaps). Five patients had more than one free flap for reconstruction of complex defects. Among patients with prior radiation, 54% of narrow-field and 88% of wide-field patients underwent flap placement at time of initial pharyngeal closure or reconstruction. Primary TEP was performed on 80 patients (33%), including 21 that were performed at the same time as flap reconstruction (or 12.7% of flap cases). See Table 2 for a summary of operative characteristics.

Table 2.

Operative Details (n = 245), Univariate Analysis Results^*

	# Patients (%)	Any Complication	Wound Complication	Fistula	Readmission
Reason for Laryngectomy
Primary surgery	93 (38.0)	-	-	-	-
Salvage	137 (55.9)	1.22 (0.70-2.15)	1.48 (0.77-2.84)	3.02 (1.25-7.25)	1.22 (0.57-2.62)
Dysfunctional larynx	15 (6.1)	1.10 (0.35-3.52)	0.69 (0.14-3.34)	***	1.22 (0.57-2.62)
Resection Type
TL	107 (43.7)	-	-	-	-
TL + PP/Glossectomy	84 (34.3)	1.49 (0.81-2.77)	1.45 (0.70-2.97)	1.46 (0.59-3.63)	1.21 (0.52-2.80)
TLP	54 (22.0)	2.15 (1.08-4.27)	2.08 (0.96-4.51)	3.08 (1.25-7.58)	1.45 (0.58-3.63)
Neck Dissection
No	35 (14.3)	-	-	-	-
Yes	210 (85.7)	1.54 (0.69-3.46)	1.12 (0.46-2.73)	1.29 (0.43-3.92)	1.85 (0.53-6.40)
Pharyngeal closure (no flap)
Unknown	40 (50.0)	-	-	-	-
Vertical	27 (33.8)	0.62 (0.19-2.04)	0.39 (0.7-20.3)	0.49 (0.05-4.94)	0.28 (0.03-2.51)
Horizontal	9 (11.3)	0.34 (0.04-3.05)	**	***	****
T-type	4 (5.0)	0.91 (0.09-9.69)	1.62 (0.15-17.9)	4.22 (0.33-54.1)	2.40 (0.21-27.8)
Flap Reconstruction
No	81 (33.1)	-	-	-	-
Yes	164 (66.9)	1.55 (0.86-2.77)	2.19 (1.06-4.53)	2.57 (1.02-6.49)	1.21 (0.55-2.68)
Type of Flap Reconstruction
Pectoralis	20 (12.2)	-	-	-	-
Anterolateral thigh	124 (75.6)	0.92 (0.35 -2.40)	1.09 (0.37-3.23)	0.86 (0.26-2.83)	0.44 (0.14-1.39)
Radial forearm	10 (6.1)	0.64 (0.13-3.25)	**	***	0.33 (0.03-3.33)
Other	10 (6.1)	1.5 (0.36-6.92)	2.0 (0.40-10.1)	1.00 (0.15-6.67)	1.29 (0.24-6.96)
Primary TEP (all)	80 (32.7)	0.71 (0.40-1.26)	0.53 (0.26-1.08)	0.49 (0.20-1.18)	0.98 (0.45-2.13)
without flap	59 (73.8)	-	-	-	-
with flap	21 (26.3)	0.76 (0.28-2.05)	0.58 (0.16-2.05)	0.63 (0.14-2.84)	1.04 (0.29-3.73)

^*Odds Ratio (95% Confidence Interval);

^**No patients had wound complications;

^***No patients had fistula;

^****No patients were readmitted. Abbreviations: TL, total laryngectomy; PP, partial pharyngectomy; TLP, total laryngopharyngectomy; TEP, tracheoesophageal prosthesis

Overall Complications and Treatment Required

Overall, 144 complications occurred in 83 patients (34%) including 3 (1.2%) deaths (Table 3). PCF, non-fistulous wound complications, and hematoma were most common. Eleven patients were treated for pneumonia during their post-operative hospitalization and four developed a chyle leak. Nine patients developed less common complications including myocardial infarction (1), syndrome of inappropriate antidiuretic hormone secretion (2), and severe postoperative hypoparathyroidism requiring temporary intravenous calcium infusion (1). Among treatments for overall complications, 58 patients received antibiotics, 51 required additional surgical procedures, and 14 received additional wound care without return to the operating room. The majority of patients (38/51) who required a return to the operating room did so for a washout and/or closure revision. The remaining 13 patients required additional reconstruction (10 pectoralis major flaps and 3 free flaps), of which 5 occurred in patients who were primarily closed initially. Four of these patients were salvage cases. Univariate analysis results for overall complications are listed in Tables 1 and 2. Multivariate analysis determined that prior laryngeal surgery was associated with a decreased likelihood for overall complications (OR 0.27, 95% CI 0.09-0.82) while patients undergoing more extensive resections, namely TLP, had a statistically significant increased likelihood for overall complications (OR 1.44, 95% CI 1.02-2.03).

Table 3.

Complications and Readmission (n = 245)

	# Patients (%)
Any complication	83 (33.9)
Wound complications	53 (21.6)
Pharyngocutaneous Fistula	34 (13.9)
Non-Fistula
Cellulitis	10 (4.1)
Neck abscess	4 (1.6)
Incisional dehiscence	13 (5.3)
Hematoma	22 (9.0)
Chyle leak	4 (1.6)
Pneumonia	11 (4.5)
Flap failure	0 (0)
Other	9 (3.7)
Death	3 (1.2)
Readmission (within 30 days)	34 (13.9)

Wound complications

Wound complications (PCF or non-fistulous complications of cellulitis, neck abscess, or neck incisional dehiscence) occurred in 53 patients. Ten patients developed cellulitis, 4 had neck abscesses, and 13 experienced dehiscence of their neck incision in the absence of a recognized salivary fistula. Considering wound complications on the whole, univariate analysis revealed that extent of surgical resection, need for reconstruction, smoking history, and prior radiation therapy were associated with increased odds of these outcomes. Pre-operative tracheotomy and primary TEP placement were negatively associated with wound complications. On multivariate analysis, only prior smoking history, including both former and current smokers, maintained statistical association, conferring a respective 5.1 (95% CI 1.1–22.4) and 5.8 (95% CI 1.3–26.5) times increased likelihood for wound complication (Figure 1).

Figure 1.

Multivariate analysis results for wound complications, pharyngocutaneous fistula, and hospital readmission.

Pharyngocutaneous Fistula

PCF occurred in 34 (13.9%) patients, 97% of which were managed with antibiotics, 56% local wound care in addition to 23 patients (68%) who returned at least once to the operating room. Though return to operating room was common overall, it was not more likely in flap patients. Increased risk for PCF on univariate analysis was found among former and current smokers, patients with prior radiation therapy, those with primary tumors in the pharynx as well as those undergoing TLP, and patients requiring flap reconstruction. There were no significant differences in fistula between no reconstruction and flap reconstruction in salvage cases. Again, pre-operative tracheotomy and primary TEP were associated with decreased odds of salivary fistula. Multivariate analysis demonstrated that patients with prior wide-field but not narrow-field radiation were 3.1 times more likely (95% CI 1.3-7.7) to develop PCF than those without prior radiation (Figure 1). Prior treatment with concurrent chemotherapy did not increase the risk of fistula compared to wide-field radiation alone.

Length of Stay

Average hospitalization was 8.1 days (median 7, range 3-27). Patients undergoing TL for laryngeal dysfunction had the shortest mean stay at 7.1 days, followed by primary TL at 8.0 days, and salvage TL at 8.3 days though median LOS was 7 days for all. Considering flap-reconstructed patients, mean LOS was 8.8 days (median 8) compared with a mean of 6.6 days (median 6) in those not requiring flaps. Univariate analysis found extent of surgical resection and type of flap to be associated with longer LOS. Wound complications, fistula in particular, and hematoma were also associated with extended hospitalization. Multivariate analysis including all significant covariates revealed that type of flap (p=0.002) and hematoma (p=0.05) predicted longer LOS. Examining flap patients more closely, pectoralis major reconstructions had the shortest LOS, followed by anterolateral thigh, radial forearm, and, lastly, other flaps (Figure 2, p=0.001). Patients with post-operative hematoma stayed in the hospital almost 3 days longer on average than those not developing a hematoma (10.7 vs 7.8 days, respectively) (p=0.0004). Given the impact of developing a hematoma on LOS, additional univariate analysis revealed that salvage surgery (and prior radiation therapy) as well as pharyngeal tumor site and need for flap reconstruction were statistically significant though none of these associations maintained statistically significance in a multivariate model.

Figure 2.

Length of stay stratified by type of flap reconstruction.

Readmission

Thirty-four patients (13.9%) were readmitted to the hospital within 30 days of discharge. Univariate analysis demonstrated that current and former smokers, patients with hypoalbuminemia, and those who developed wound complications, fistula in particular, or pneumonia were more likely to require readmission. Patients who had undergone pre-operative tracheotomy and those with diabetes or immunosuppression were negatively associated with readmission. Multivariate analysis revealed patients with wound complications (p<0.001) to be highly associated with readmission as well as those with low albumin pre-operatively (p=0.003) (Figure 1). More specifically, 22% of patients with pre-operative hypoalbuminemia and 51% of patients who developed post-operative wound complications required readmission within 30 days. Readmitted patients had an average pre-operative albumin level of 3.4 (median 3.3) compared to an average of 3.7 (median 3.9) in those who remained outpatients (p=0.01).

Discussion

Patients undergoing TL at a single, high-volume institution during a relatively short interval represent a contemporary cohort that had either prior non-surgical therapy with persistent/recurrent cancer or chronic laryngeal dysfunction or a large and/or functionally-limiting primary tumor at presentation. Almost two-thirds of patients had prior radiation therapy and the vast majority without prior radiation (85%) had stage T4 primary tumors. Complications occurred in roughly one third of patients, including a 21.6% incidence of wound complications and a 13.8% incidence of PCF. Length of stay was highly associated with development of post-operative hematoma and with flap type for patients who required additional reconstruction in excess of primary pharyngeal closure. Lastly, readmission was significantly associated with low albumin levels and development of wound complications.

Wound complications, specifically PCF, and hematoma are well-recognized as the most common complications in primary and salvage TL^16–18. This study found a strong association between wound complications and both current and former smokers, supporting prior evidence linking smoking to overall complications and PCF in TL¹⁹. Despite the clinically apparent negative impact of smoking on wound healing, a recent systematic analysis of smoking and peri-operative complications in head and neck cancer found mixed results, largely owing to the heterogeneity of studies in terms of methodology, smoking classification, and outcomes of interest. Nevertheless, this association is supported by strong physiologic plausibility and evidence of relative tissue hypoxemia in current and/or recent smokers as well as knowledge of common co-morbidities among smokers that negatively impact wound healing^{20, 21}. In the absence of biochemical confirmation of smoking cessation, it’s true that these groups may be subject to misclassification. Given that both former and current smokers were at increased risk, this further bolsters efforts to support smoking cessation prior to both surgical and non-surgical treatment, since smoking in the latter group not only influences treatment toxicity and efficacy but also because they may eventually require a salvage procedure.

Many studies have documented an association between PCF and prior radiation^22–24. In an era when the type and location of prior radiation therapy is of paramount importance to treatment decision-making, few studies examining post-TL complications make the distinction between the narrow radiation fields used in early glottic cancers and the wider fields used for more advanced glottic tumors or those in other laryngeal and pharyngeal locations²⁵. Though population-level studies and meta-analyses may also be limited in their ability to derive this granular information, publications reporting an association between PCF and prior chemoradiation or extra-laryngeal primary tumor sites can likely be assumed to have been patients with prior “wide-field” treatment as was found in this study^{17, 26, 27}. Indeed, this finding strongly reinforces the argument that specifics of non-surgical treatment, particularly prior radiation therapy in excess of a “narrow-field” approach, must be considered for surgical planning and counseling in patients requiring a salvage TL. Time since completion of radiation was not significantly associated with the outcomes of interest. Nevertheless, these radiation data suggest the need to distinguish narrow-field radiotherapy failures as more favorable when attempting to risk stratify for PCF since this group did not have significantly elevated risk for PCF despite approximately 35% less use of flap reconstruction. Alternatively, it may be argued that this lack of difference in fistula rates is influenced by the selection bias of the surgeons’ impression for the initial need of flap reconstruction and cannot be accounted for through a retrospective study.

Given the high morbidity of PCF, salvage TL represents one of the most common indications within head and neck surgery for regional or free tissue reconstruction^{14, 28–30}. The use of vascularized tissue reconstruction of the neopharynx, even in an onlay fashion, has greatly reduced the incidence of PCF and wound complications^{18, 31–33}. Consistent with this trend, over two-thirds of patients in this study were reconstructed with regional or free tissue transfer. Considering complication outcomes by type of reconstruction, we found no statistically significant differences but this must be interpreted recognizing that the vast majority of reconstruction is through anterolateral thigh flaps and pectoralis flaps are mainly reserved for two scenarios, namely to bolster closure in the salvage setting where pharyngeal mucosa is entirely spared and/or because free tissue transfer was deemed too risky from an operative risk standpoint. Analysis of patient hospitalization was associated with incremental increases in LOS by flap type, a finding which is supported by other publications and clinical practice³⁴. LOS must be considered in the context of departmental and institutional characteristics such as case volume and plan of care pathways. Such pathways have been shown to be beneficial from a LOS, complication, and readmission standpoint in several groups of post-surgical head and neck cancer patients including TL^{13, 35–38}.

Readmission rates were predictably higher among patients who developed wound complications, occurring in patients who developed wound complications during their initial hospitalization, continued management after discharge, and subsequently required readmission as well as through the new development of such complications after an initially uncomplicated post-operative hospitalization. Prior population level and single institution studies have documented similar findings^{39, 40}. Unfortunately, a retrospective study is unable to provide insight into the nuances of hospitalizations and possible decision-making process that might explain readmissions (or premature discharge) for individual or groups of patients. Our study did reveal that pre-operative hypoalbuminemia was associated with 2.8 greater odds of readmission. Prior studies have found an association between low pre-operative albumin (or pre-operative dietary counseling) and wound complications, which aligns well with our findings and provides a measureable, actionable item to optimize outcomes and reduce readmission rates^{23, 41, 42}. Because pre-operative nutritional status represents a potentially modifiable risk factor, this study argues that formal involvement of a clinical dietitian for nutritional optimization should also be a common part of the multidisciplinary care of TL patients preoperatively rather than performed in an ad hoc fashion for those who have the most obvious malnourishment.

The primary limitation of this study rests in its retrospective design, which predisposes data to possible differential misclassification, and may bias results either towards or away from the null. Also due to the nature of data collection and specifics of the electronic medical record, details on duration and type of antibiotic prophylaxis or treatment, severity of complications such as PCF, and unrecognized contributing factors that may have led to prolonged hospitalization or readmission in individual patients could not be taken into account. Given that our institution has a large referral base from both surgical and non-surgical providers, details on prior treatments including exact radiation dose, timing, and type of systemic therapy were not always available, meaning that analysis could also not be conducted to this degree of specificity. Nevertheless, given the patient volume at a single institution, and relatively short interval in comparison to many other publications, this study was able to minimize surgical heterogeneity, improve generalizability for a modern cohort of TL patients, and was adequately powered for multivariate analysis to adjust for potential confounding factors that influence post-operative complications, LOS, and readmission.

These data constitute valuable evidence on the use of TL in the era of organ preservation, enhance surgical quality improvement efforts through outcome optimization, and inform the need for evidence-based risk stratification as increasing emphasis is placed on shared decision-making and resource utilization.

Condensed Abstract.

In patients undergoing primary and salvage total laryngectomy (TL), wound complications including pharyngocutaneous fistula occurred in nearly one-third, particularly among smokers and those with prior wide-field radiation, length of stay was expectedly longer among reconstructed patients, and readmission was associated with pre-operative hypoalbuminemia and post-operative wound complications. These data are intended to inform quality improvement efforts, assist with risk stratification, and aid in counseling of patients undergoing TL.