Best Practice in Surgical Treatment of Malignant Head and Neck Tumors

Purpose of review: Defining the best practice of surgical care for patients affected by malignant head and neck tumors is of great importance. In this review we aim to describe the evolution of “best practice” guidelines in the context of quality-of-care measures and discuss current evidence on “best practice” for the surgical treatment of cancers of the sino-nasal tract, skull base, aero-digestive tract, and the neck.

Recent findings: Current evidence based on certain structure and outcome indicators, but mostly based on process indicators already helps defining the framework of “Best practice” for head and neck cancer surgery. However, many aspects of surgical treatment still require in-depth research.

Summary: While a framework of “Best practice” strategies already exists for the conduction of the surgical treatment of head and neck cancers, many questions still require additional research in particular in case of rare histologies in the head and neck region.

Validating a risk stratification tool for audit of early outcome after operations for squamous cell carcinoma of the head and neck

The aim of this study was to validate a case-mix adjustment tool (neural network) for the audit of postoperative outcomes. We tested its calibration and discrimination on two unseen groups of patients being treated for squamous cell carcinoma (SCC) of the head and neck and compared observed complication rates with predicted rates. A total of 196 patients who were treated at two UK NHS institutions between 2016 and 2018 were audited. Preoperative data pertaining to risk (T classification, complexity of operation, and "high-risk" status) were collected, together with data on postoperative complications. Diagnostic test statistics and receiver operating curves (ROC) were used to test the performance of the tool. The score was well calibrated (predicted and observed complication rates both 43%), but discrimination suggested only fair accuracy (ROC 0.66 - 0.68). Adjustment of case mix for the audit of postoperative complications is difficult, although our model suggests that departmental audit is possible, and its accuracy is equivalent to that of other national audits. Further work may elucidate key variables that have not yet been assessed.

A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models

OBJECTIVES

The objective of this study was to compare performance of logistic regression (LR) with machine learning (ML) for clinical prediction modeling in the literature.

STUDY DESIGN AND SETTING

We conducted a Medline literature search (1/2016 to 8/2017) and extracted comparisons between LR and ML models for binary outcomes.

RESULTS

We included 71 of 927 studies. The median sample size was 1,250 (range 72-3,994,872), with 19 predictors considered (range 5-563) and eight events per predictor (range 0.3-6,697). The most common ML methods were classification trees, random forests, artificial neural networks, and support vector machines. In 48 (68%) studies, we observed potential bias in the validation procedures. Sixty-four (90%) studies used the area under the receiver operating characteristic curve (AUC) to assess discrimination. Calibration was not addressed in 56 (79%) studies. We identified 282 comparisons between an LR and ML model (AUC range, 0.52-0.99). For 145 comparisons at low risk of bias, the difference in logit(AUC) between LR and ML was 0.00 (95% confidence interval, -0.18 to 0.18). For 137 comparisons at high risk of bias, logit(AUC) was 0.34 (0.20-0.47) higher for ML.

CONCLUSION

We found no evidence of superior performance of ML over LR. Improvements in methodology and reporting are needed for studies that compare modeling algorithms.