Length of Stay Prediction Models for Oral Cancer Surgery: Machine Learning, Statistical and ACS-NSQIP

OBJECTIVE

Accurate prediction of hospital length of stay (LOS) following surgical management of oral cavity cancer (OCC) may be associated with improved patient counseling, hospital resource utilization and cost. The objective of this study was to compare the performance of statistical models, a machine learning (ML) model, and The American College of Surgeons National Surgical Quality Improvement Program's (ACS-NSQIP) calculator in predicting LOS following surgery for OCC.

MATERIALS AND METHODS

A retrospective multicenter database study was performed at two major academic head and neck cancer centers. Patients with OCC who underwent major free flap reconstructive surgery between January 2008 and June 2019 surgery were selected. Data were pooled and split into training and validation datasets. Statistical and ML models were developed, and performance was evaluated by comparing predicted and actual LOS using correlation coefficient values and percent accuracy.

RESULTS

Totally 837 patients were selected with mean patient age being 62.5 ± 11.7 [SD] years and 67% being male. The ML model demonstrated the best accuracy (validation correlation 0.48, 4-day accuracy 70%), compared with the statistical models: multivariate analysis (0.45, 67%) and least absolute shrinkage and selection operator (0.42, 70%). All were superior to the ACS-NSQIP calculator's performance (0.23, 59%).

CONCLUSION

We developed statistical and ML models that predicted LOS following major free flap reconstructive surgery for OCC. Our models demonstrated superior predictive performance to the ACS-NSQIP calculator. The ML model identified several novel predictors of LOS. These models must be validated in other institutions before being used in clinical practice.

LEVEL OF EVIDENCE

Level 3 Laryngoscope, 2024.

The Utility of Comorbidity Indices in Assessing Head and Neck Surgery Outcomes: A Systematic Review

OBJECTIVE

To evaluate the utility of comorbidity index (CI) scores in predicting outcomes in head and neck surgery (HNS). The CIs evaluated were the Charlson Comorbidity Index (CCI), Elixhauser Comorbidity Index (ECI), Kaplan-Feinstein Index (KFI), American Society of Anesthesiologists Physical Status (ASA-PS), Adult Comorbidity Evaluation-27 (ACE-27), National Cancer Institute Comorbidity Index (NCI-CI), and the Washington University Head and Neck Comorbidity Index (WUHNCI).

METHODS

We report a systematic review according to the 2009 Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Electronic databases (PubMed, Cochrane, and Embase) and manual search of bibliographies identified manuscripts addressing how CI scores related to HNS outcomes.

RESULTS

A total of 116 studies associated CI scores with HNS outcomes. CIs were represented in the literature as follows: ASA-PS (70/116), CCI (39/116), ACE-27 (24/116), KFI (7/116), NCI-CI (3/116), ECI (2/116), and WUHNCI (1/116). The most frequently cited justification for calculating each CI (if provided) was: CCI for its validation in other studies, ACE-27 for its utility in cancer patients, and ECI for its comprehensive design. In general, the CCI and ACE-27 were predictive of mortality in HNS. The ECI was most consistent in predicting >1-year mortality. The ACE-27 and KFI were most consistent in predicting medical complications.

CONCLUSION

Despite inconsistencies in the literature, CIs provide insights into the impact of comorbidities on outcomes in HNS. These scores should be employed as an adjunct in the preoperative assessment of HNS patients. Comparative studies are needed to identify indices that are most reliable in predicting HNS outcomes.

LEVEL OF EVIDENCE

NA Laryngoscope, 2021.

Identification of patients for a delayed extubation strategy versus elective tracheostomy for postoperative airway management in major oral cancer surgery: A prospective observational study in seven hundred and twenty patients

OBJECTIVES

Tracheostomy (TT) and delayed extubation (DE) are two approaches to postoperative airway management in patients after major oral cancer surgery. We planned a study to determine the safety of overnight intubation followed by extubation the next morning (DE) compared to elective TT and to identify factors that were associated with a safe DE (maintenance of a patent airway).

MATERIAL AND METHODS

We conducted a prospective observational study in a tertiary referral cancer care center. We included adult patients undergoing elective major oral cancer surgery under general anesthesia with tracheal intubation. The decision regarding postoperative airway management using either TT or DE was made according to the usual practice at our center.

RESULTS

We screened a total of 4477 patients, 720 patients were included. DE was performed in 417 patients (58.4%) and TT in 303 patients (42.4%). On multivariable analysis, T1-T2 tumor stage, absence of extensive resection, primary closure or reconstruction using fasciocutaneous flap, absence of preoperative radiation, no neck dissection or unilateral neck dissection and shorter duration of anesthesia were independent predictors for a safe DE. Overall complications (4.3% versus 22.5%, p = 0.00) and airway complications (1.7% versus 8.7%, p = 0.00) were lower in the DE compared to the TT group respectively. DE was associated with a shorter hospital stay (7.2 ± 3.7 versus 11.5 ± 7.2 days, p = 0.00), time to oral intake and speech compared to TT.

CONCLUSIONS

A DE strategy after major oral cancer surgery is a safe alternative to TT in a select group of patients.

Outcomes for head and neck cancer patients admitted to intensive care in Australia and New Zealand between 2000 and 2016

OBJECTIVE

To report intensive care unit admission outcomes for head and neck cancer patients.

METHODS

A retrospective, observational cohort analysis of all Australian and New Zealander head and neck cancer patient intensive care unit admissions from January 2000 to June 2016, including data from 192 intensive care units.

RESULTS

There were 10 721 head and neck cancer patients, with a median age of 64 years (71.6 per cent male). Of admissions, 76.4 per cent were in public hospitals, 96.9 per cent were post-operative and 43.6 per cent required mechanical ventilation. Annual head and neck cancer admissions increased from 2000 to 2015 (from 348 to 1132 patients), but the overall proportion of intensive care unit admissions remained constant. In-hospital mortality was 2.7 per cent, and intensive care unit mortality was 0.7 per cent. The in-hospital mortality risk decreased three-fold (p < 0.001).

CONCLUSION

Head and neck cancer patients had low mortality in the intensive care unit and in hospital. Risk of dying decreased despite more intensive care unit admissions. This is the first large-scale cohort study quantifying intensive care unit utilisation by head and neck cancer patients. It informs future work investigating alternatives to the intensive care unit for these patients.

National Surgical Quality Improvement Program risk calculator validity in South Australian laryngectomy patients

BACKGROUND

Assessing an individual patient's post-operative risk profile prior to laryngectomy for cancer is difficult. The American College of Surgeons' National Surgical Quality Improvement Program (ACS-NSQIP) risk calculator was developed to better inform preoperative decision-making. The calculator uses patient-specific characteristics to estimate the risk of experiencing post-operative complications within 30 days of surgery. We investigated the ACS-NSQIP risk calculator's performance for Australian laryngectomy patients.

METHODS

The ACS-NSQIP risk calculator was used to retrospectively calculate the 30-day post-operative predicted outcomes in patients who underwent laryngectomy for laryngeal, hypopharyngeal and thyroid cancers (with laryngeal involvement) in two institutions in South Australia. These data were compared against the actual mortality, morbidity, complications and length of stay (LOS) collected from a retrospective chart review.

RESULTS

A total of 144 patients underwent surgical intervention for malignancies with laryngeal involvement. The median LOS was 25 days (range 13-197) compared to the predicted LOS of 6.5 days (range 3.5-12.5). Overall mortality was 2.78% with post-operative complications occurring in 63% of patients. The most common complication was wound infection, occurring in 33% of patients. Hosmer-Lemeshow plots demonstrated good agreement between predicted and observed rates for complications.

CONCLUSION

The ACS-NSQIP risk calculator effectively predicted post-operative complication rates in South Australian laryngeal cancer patients undergoing laryngectomy. However, differences in predicted and actual LOS may limit the usefulness of the calculator's LOS predictions for Australian patients.

Head and Neck Reconstructive Surgery: Characterization of the One-Team and Two-Team Approaches

PURPOSE

To the best of our knowledge, no studies have compared the patient profiles for 1- versus 2-team surgery within head and neck oncosurgery.

PATIENTS AND METHODS

A retrospective study of the data from 2968 patients who had undergone concurrent head and neck extirpative and reconstructive surgery in the National Surgical Quality Improvement Program (2010 to 2017) was conducted. Patients were stratified into 1- and 2-team surgery groups, and the demographic data were compared. Univariate analyses of the outcomes before and after propensity score matching were conducted.

RESULTS

Most ablative and reconstructive head and neck procedures (68.5%) were performed using a 1-team approach. The patients who had undergone 2-team surgery were more likely to have a higher American Society of Anesthesiologists classification (P < .001), to require mandibulectomy (P < .001) or glossectomy (P < .001), and to receive a microvascular free flap (P < .001) but were less likely to require parotidectomy (P < .001) or to receive a rotational flap (P < .001). Before propensity score matching, the patients undergoing 2-team surgery had longer operative times (P < .001), longer postoperative stays (P < .001), greater rates of a return to the operating room (P = .001), and an increased rate of complications (P < .001). After propensity score matching, the 2-team approach continued to have longer operative times (P < .001) and an increased incidence of complications (P < .001) but no significant differences in the length of stay or rate of return to the operating room after Bonferroni's correction.

CONCLUSIONS

Nationally, most head and neck ablative and reconstructive surgeries were completed by 1 team. More complicated reconstructive procedures involving microvascular free flaps have been more commonly performed by 2 teams, resulting in slightly longer operative times and greater associated complication rates.

Discharge Delay in Head and Neck Free Flap Surgery: Risk Factors and Strategies to Minimize Hospital Days

Objective

Length of stay (LOS) includes time medically necessary in the hospital and time waiting for discharge (DC) afterward. This DC delay is determined in head and neck free flap patients. Reasons for and factors leading to DC delay, as well as associated adverse outcomes, are elucidated.

Methods

Retrospective chart review was performed for all head and neck free flap surgeries from 2012 to 2017. Data including demographics, comorbidities, and perioperative factors were collected. Regression analyses were performed to identify factors associated with DC delay.

Results

In total, 264 patients were included. Mean total LOS was 13.1 days. DC delay occurred in 65% of patients with a mean of 4.8 days. Factors associated with DC delay on univariate analysis included Medicaid/self-pay insurance, DC to a facility, and not having children ( P < .05). Multivariate analysis showed prolonged medically necessary LOS and surgery on a Monday/Friday ( P < .05) were associated with DC delay. Top reasons for DC delay included case management shortages, rejection by facility, and awaiting supplies. Eleven percent experienced complications during the DC delay.

Discussion

DC delay can add days and complications to the LOS. Prevention begins preoperatively with DC planning involving the patient’s closest family. Understanding limitations of the patient’s insurance may help plan DC destination. Optimizing hospital resources when available should be a focus.

Implications for Practice

Head and neck free flap patients require a team of teams unified in optimizing quality of care. DC delay is a novel quality metric reflecting the team’s overall performance. Through strategic DC planning and capitalizing on available resources, DC delay can be minimized.

Insurance Status, an Important Predictor of Oral Cancer Surgery Outcomes

PURPOSE

Patients without insurance, or using Medicaid, generally have a lower socioeconomic status. They have less access to screening and regular medical care, resulting in later diagnosis of oral cancer. This study examined the association between insurance status and the likelihood of complications after head and neck cancer surgery.

MATERIALS AND METHODS

A retrospective cross-sectional study was implemented to determine whether patients' insurance status is associated with increased complications and length of stay after oral cancer surgery. Patients were grouped into 4 cohorts: 1) private insurance, 2) Medicare, 3) Medicaid, and 4) uninsured. Patients were stratified further to consider age, gender, initial staging, pre-existing comorbidities, and social history. Data were analyzed with χ(2) test, 1-way analysis of variance, odds ratios, and binary logistic regression.

RESULTS

This study consisted of 89 surgically treated patients. The uninsured and Medicaid groups had the highest incidence of postoperative complications. Uninsured patients, followed by the Medicare cohort, were the most likely to have an extended length of stay.

CONCLUSION

Uninsured and Medicaid patients are at increased probability for major and minor complications after head and neck cancer surgery. Uninsured patients also showed the greatest tendency for a prolonged length of hospital stay. This could reflect their lack of preventive care, increased use of tobacco and alcohol, presentation with more advanced disease, and delays in initiating treatment.