Machine learning methods applied to audit of surgical margins after curative surgery for head and neck cancer.

Most surgical specialties have attempted to address concerns about the unfair comparison of outcomes by 'risk-adjusting' data to benchmark specialty-specific outcomes that are indicative of quality of care. We explore the ability to predict for positive margin status so that effective benchmarking that will account for complexity of case mix is possible. A dataset of care episodes recorded as a clinical audit of margin status after surgery for head and neck squamous cell carcinoma (n=1316) was analysed within the Waikato Environment for Knowledge Analyisis (WEKA) machine learning programme. The outcome was a classification model that can predict for positivity of tumour margins (defined as less than 1mm) using data on preoperative demographics, operations, functional status, and tumour stage. Positive resection margins of less than 1mm were common, and varied considerably between treatment units (19%-29%). Four algorithms were compared to attempt to risk-adjust for case complexity. The 'champion' model was a Naïve Bayes classifier (AUROC 0.72) that suggested acceptable discrimination. Calibration was good (Hosmer-Lemershow goodness-of-fit test p=0.9). Adjusted positive margin rates are presented on a funnel plot. Subspecialty groups within oral and maxillofacial surgery are seeking metrics that will allow for meaningful comparison of the quality of care delivered by surgical units in the UK. To enable metrics to be effective, we argue that they can be modelled so that meaningful benchmarking, which takes account of variation in complexity of patient need or care, is possible.

BAOMS QOMS: findings from the pilot phase and lessons learned in the feasibility evaluation of a national quality improvement initiative.

The BAOMS QOMS pilot was developed and run in six England OMFS units between December 2019 - April 2020. The aims of this pilot project were: to evaluate feasibility of the questionnaires developed for the audit and how effective they were with regards to quality improvement, to test the processes associated with the data collection system and finally, to provide baseline data to support patient data collection without the requirement of prospective consent. The pilot included a series of six audits (oral and dentoalveolar [ODA], oncology, orthognathic, reconstruction, trauma, and skin). Data entry was clinician-led in five OMFS units and in one unit (EKHU), it was additionally supported by members of the clinical coding team. One hundred and twenty-eight REDCap account user details were issued and of these, 45 (35%) completed registration and 22 (17%) were active users who participated in the pilot data entry. Disproportionate focus on individual audits within QOMS was seen, though not all units offered the full range of service audited. Users suggest the skin and ODA audits were sufficiently clear, but improvement is required in the oncology and reconstruction questionnaire particularly. The pilot was successful in aiding the project team identify areas of weaknesses and strength in the design of the REDCap registry and implementation of the next phase of the initiative. The information and experience gained has to date enabled a successful application for section 251 approval from the HRA and progress for the next phase of national data collection.

Machine learning methods applied to audit of surgical outcomes after treatment for cancer of the head and neck.

Most surgical specialties have attempted to address concerns about unfair comparison of outcomes by "risk-adjusting" data to benchmark specialty-specific outcomes that are indicative of the quality of care. We are building on previous work in head and neck surgery to address the current need for a robust validated means of risk adjustment. A dataset of care episodes, which were recorded as a clinical audit of complications after operations for squamous cell carcinoma (SCC) of the head and neck (n=1254), was analysed with the Waikarto Environment for Knowledge Analysis (WEKA) machine learning tool. This produced 4 classification models that could predict complications using data on the preoperative demographics of the patients, operation, functional status, and tumour stage. Three of them performed acceptably: one that predicted "any complication" within 30 days (area under the receiver operating characteristic curve (AUROC) 0.72), one that predicted severe complications (Clavien-Dindo grade 3 or above) within 30 days (AUROC 0.70), and one that predicted a prolonged duration of hospital stay of more than 15 days, (AUROC 0.81). The final model, which was developed on a subgroup of patients who had free tissue transfer (n=443), performed poorly (AUROC 0.59). Subspecialty groups within oral and maxillofacial surgery are seeking metrics that will allow a meaningful comparison of the quality of care delivered by surgical units in the UK. For these metrics to be effective they must show variation between units and be amendable to change by service personnel. Published baseline data must also be available. They should be modelled effectively so that meaningful comparison, which takes account of variations in the complexity of the patients' needs or care, is possible.

Case-mix adjustment in audit of length of hospital stay in patients operated on for cancer of the head and neck.

Patients treated surgically for squamous cell carcinoma (SCC) of the head and neck form a heterogeneous group, and it is difficult to take this variation into account when measuring the quality of care. We have tested the feasibility of mathematical models that allow for the adjustment for case mix when auditing the length of hospital stay as a proxy indicator of the quality of care. We completed a case-note audit of 733 surgical episodes of care for SCC of the head and neck in five cancer networks, and used logistic regression and decision tree analysis to adjust for case mix using pertinent preoperative variables. Risk adjustment models of length of stay included age, alcohol, T classification, performance status, tracheostomy, high-risk status, and complexity of operation. The risk-adjusted length of stay differed significantly between the cancer networks studied (p<0.001). The models performed acceptably for the purpose of audit when this was under 15 days. Length of stay is a measurable outcome that can be used as a benchmark of surgical care. Audits of this after operations for cancer of the head and neck, if reported in national clinical audits, should take case mix into account.

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.

Subjective perception of paraesthesia of the lower lip after bilateral sagittal split osteotomy at a district general hospital.

Our aim was to assess patients' perception of paraesthesia of the lower lip after bilateral sagittal split osteotomy (BSSO) at a district general hospital. Patients who had BSSO between August 2013 and August 2014 (n=46) were asked to score their perception of numbness between 0-10 (0=normal sensation. 10=complete loss of sensation/total numbness) one day postoperatively and then weekly for seven weeks, and at three months, 6 months, and one year. Data was collected on score sheets and by regular contact by telephone. Of the 46 operated on, 31 were female and 15 male. Data were available one year postoperatively for 43 patients. Ten of the 92 sides were reported as feeling normal on day 1 postoperatively, three-quarters as feeling normal at six months, and 79 at one year. On multivariate analysis there was no significant difference in postoperative sensation at one year between sides operated on by the registrar (left) and consultant (right) operated (p=0.76). Our results compared favourably with the limited data available in similarly designed studies.

Validating a benchmarking tool for audit of early outcomes after operations for head and neck cancer.

INTRODUCTION In 2013 all UK surgical specialties, with the exception of head and neck surgery, published outcome data adjusted for case mix for indicator operations. This paper reports a pilot study to validate a previously published risk adjustment score on patients from separate UK cancer centres. METHODS A case note audit was performed of 1,075 patients undergoing 1,218 operations for head and neck squamous cell carcinoma under general anaesthesia in 4 surgical centres. A logistic regression equation predicting for all complications, previously validated internally at sites A-C, was tested on a fourth external validation sample (site D, 172 operations) using receiver operating characteristic curves, Hosmer-Lemeshow goodness of fit analysis and Brier scores. RESULTS Thirty-day complication rates varied widely (34-51%) between the centres. The predictive score allowed imperfect risk adjustment (area under the curve: 0.70), with Hosmer-Lemeshow analysis suggesting good calibration. The Brier score changed from 0.19 for sites A-C to 0.23 when site D was also included, suggesting poor accuracy overall. CONCLUSIONS Marked differences in operative risk and patient case mix captured by the risk adjustment score do not explain all the differences in observed outcomes. Further investigation with different methods is recommended to improve modelling of risk. Morbidity is common, and usually has a major impact on patient recovery, ward occupancy, hospital finances and patient perception of quality of care. We hope comparative audit will highlight good performance and challenge underperformance where it exists.

Comparison of proximal isovelocity surface area method with pressure half-time and planimetry in evaluation of mitral stenosis.

OBJECTIVES: This study sought to 1) compare the accuracy of the proximal isovelocity surface area (PISA) and Doppler pressure half-time methods and planimetry for echocardiographic estimation of mitral valve area; 2) evaluate the effect of atrial fibrillation on the accuracy of the PISA method; and 3) assess factors used to correct PISA area estimates for leaflet angulation. BACKGROUND: Despite recognized limitations of traditional echocardiographic methods for estimating mitral valve area, there has been no systematic comparison with the PISA method in a single cohort. METHODS: Area estimates were obtained in patients with mitral stenosis by the Gorlin hydraulic formula, PISA and pressure half-time method in 48 patients and by planimetry in 36. Two different factors were used to correct PISA estimates for leaflet angle (theta): 1) plane-angle factor (theta/180 [theta in degrees]); and 2) solid-angle factor [1-cos(theta/2)]. RESULTS: After exclusion of patients with significant mitral regurgitation, the correlation between Gorlin and PISA areas (0.88) was significantly greater (p < 0.04) than that between Gorlin and pressure half-time (0.78) or Gorlin and planimetry (0.72). The correlation between Gorlin and PISA area estimates was lower in atrial fibrillation than sinus rhythm (0.69 vs. 0.93), but the standard error of the estimate was only slightly greater (0.24 vs. 0.19 cm2). The average ratio of the solid- to the plane-angle correction factors was approximately equal to previously reported values of the orifice contraction coefficient for tapering stenosis. CONCLUSIONS: 1) The accuracy of PISA area estimates in mitral stenosis is at least comparable to those of planimetry and pressure half-time. 2) Reasonable accuracy of the PISA method is possible in irregular rhythms. 3) A simple leaflet angle correction factor, theta/180 (theta in degrees), yields the physical orifice area because it overestimates the vena contracta area by a factor approximately equal to the contraction coefficient for a tapering stenosis.

CT-angiography for the detection of a lower gastrointestinal bleeding source.

The evaluation of lower gastrointestinal bleeding (LGIB) often involves the collaborative efforts of the gastroenterologist, radiologist, and surgeon. Efforts to localize the acute LGIB have traditionally involved colonoscopy, technetium-labeled red blood cell (RBC) scintigraphy, angiography, or a combination of these modalities. The sensitivity of each method of diagnosis is limited, with the most common cause of a negative study the spontaneous cessation of hemorrhage. Other technical factors include vasospasm, lack of adequate contrast volume or exposure time, a venous bleeding source, and a large surface bleeding area. We report the use of multidetector computed tomography (MDCT), or CT-angiography (CT-A), in the initial evaluation of LGIB, and speculate on the incorporation of this technique into a diagnostic algorithm to treat LGIB. MDCT may offer a very sensitive means to evaluate the source of acute LGIB, while avoiding some of the morbidity and intense resource use of contrast angiography, and may provide unique morphologic information regarding the type of pathology. Screening with the more rapid and available MDCT, followed by either directed therapeutic angiography or surgical management, may represent a reasonable algorithm for the early evaluation and management of acute LGIB in which an active bleeding source is strongly suspected.

Sudden cardiac death in postmyocardial-infarction patients.

Patients who survive a myocardial infarction are at increased risk for sudden death, owing largely to ventricular arrhythmia. In this article, we will review the epidemiology of sudden cardiac death in postmyocardial-infarction patients, arrhythmia mechanisms and substrate leading to cardiac arrest, identifying possible risk factors for sudden cardiac death (SCD) in high risk population and apply risk stratification strategies for prevention of SCD. We will also review relevant major trials and evidence-based therapy currently used, in addition to the indications and role of implantable cardioverter-defibrillators in this population. We will end this review with a summary of the current guidelines recommendations and a look into the future of this domain.

Diagnostic investigation of parotid neoplasms: a 16-year experience of freehand fine needle aspiration cytology and ultrasound-guided core needle biopsy.

This study aimed to examine the diagnostic yield of fine needle aspiration cytology (FNAC) and ultrasound-guided core needle biopsy (USCB) in the diagnosis of parotid neoplasia. A 16-year retrospective analysis was performed of patients entered into our pathology database with a final diagnosis of parotid neoplasia. FNAC and USCB data were compared to surgical excision where available. One hundred and twenty FNAC, 313 USCB, and 259 surgical specimens were analyzed from 397 patients. Fifty-six percent of FNAC and 4% of USCB were non-diagnostic. One hundred and thirty-two (33%) patients had a final diagnosis made by USCB and did not undergo surgery. Surgery was performed in 257 (65%) patients, 226 (88%) of whom had a preoperative biopsy. Most lesions were benign, but there were 62 parotid and 13 haematological malignancies diagnosed; false-negative results were obtained in three FNAC and two USCB samples. The sensitivity and specificity of FNAC were 70% and 89%, respectively, and for USCB were 93% and 100%, respectively. This study represents the largest series of patients with a parotid neoplasm undergoing USCB for diagnosis. USCB is highly accurate with a low non-diagnostic rate and should be considered an integral part of parotid assessment.

False-positive reversible perfusion defect during dobutamine-thallium imaging in left bundle branch block.

In the presence of pre-existing left bundle branch block (LBBB) exercise stress thallium scans have been associated with false-positive septal and apical perfusion abnormalities. Recent reports have documented a lower incidence of false-positive septal perfusion defects when pharmacologic agents such as dipyridamole or adenosine are utilized in patients with LBBB. Dobutamine, a synthetic catecholamine, is being used with increasing frequency in combination with perfusion agents for the diagnosis of coronary artery disease in patients unable to achieve an adequate exercise workload. Because the positive inotropic and chronotropic actions of doubtamine are similar to the physiologic effects of treadmill exercise, it is conceivable that false-positive perfusion abnormalities will be observed in patients with pre-existing LBBB undergoing dobutamine perfusion imaging. We describe a patient with underlying LBBB who underwent dobutamine thallium imaging which revealed septal and periapical defects. Subsequent coronary angiography showed these abnormalities to be false-positive. It is concluded that septal and periapical perfusion abnormalities during dobutamine thallium imaging may be false-positive and should be interpreted cautiously.

A phospholipase inhibitor modifies the pulmonary damage associated with peritonitis in rabbits.

Peritonitis has been produced in rabbits by the spreading of 5 ml of caecal contents throughout the peritoneal cavity. After sacrifice at 5 h, electron microscopy revealed a increase of 600% in PMN and 1200% in lymphocytes in the pulmonary capillaries of the test animals when compared to the sham operated group. The pulmonary capillaries in the test group showed a 27% reduction in their luminal area, which was associated with a 73% reduction in their patency due to occlusion by WBC and cell debris. In addition there was endothelial and epithelial disruption with basement membrane exposure. The PMN were degranulated and adherent to the pulmonary endothelium. Pretreatment with the phospholipase inhibitor, mepacrine, significantly attenuated these responses, so that there was only an increase of 200% in PMN and 450% in lymphocytes in the lungs with no evidence of degranulation or adhesion to the pulmonary endothelium. Furthermore there was no change in capillary luminal area when compared to the sham operated group. In addition there was only slight damage to the endothelium and epithelium with no exposure of the basement membrane. Peripheral WBC in both the test and mepacrine groups showed a similar 62% and 75% reduction after 5 h when compared to baseline values. The sham group did not show this change. These results suggest that phospholipase inhibition plays an important role in attenuating the pulmonary response to faecal peritonitis and may be of potential benefit in treating clinical septicaemia.

The effects of dobutamine, dopexamine and fluid on hepatic histological responses to porcine faecal peritonitis.

We studied the effects of two catecholamines with differing receptor profiles on hepatic blood flow and hepatic structure in a porcine model of faecal peritonitis. We treated animals with dopexamine (group Dp) or dobutamine (group Db) and fluid, or fluid alone as a control, to achieve a 25% increase in Qt from baseline values. After the induction of faecal peritonitis the increased Qt was maintained throughout the 8 h study period by adjustment of the fluid infusion rate. The dose of catecholamines remained constant. Hepatic blood flow was correspondingly maintained at above baseline values throughout the study. Post-mortem liver biopsy specimens were analysed from experimental animals and 5 sham animals who had not been instrumented or infected. In experimental animals there was a reduction in sinusoidal patency between sham and group Dp (76% of total sinusoid vs 51%, p less than 0.05) and group Dp and control (51% vs 33%, p less than 0.05) or groups Dp and Db (51% vs 34%, p less than 0.05) animals. This was accounted for by an increase in sinusoidal leukocytes and endothelial swelling. In addition to the changes noted above there was marked hepatocellular destruction in group Db. We conclude that maintenance of organ blood flow does not guarantee structural integrity in the sepsis syndrome and hepatocellular damage was greater in group Db than group Dp or control.

A narrow range, medium molecular weight pentastarch reduces structural organ damage in a hyperdynamic porcine model of sepsis.

OBJECTIVE: to compare diafiltered 6% pentastarch (Pentafraction--PDP, MWn 120,000 and MWw 280,000) and native pentastarch (Pentaspan--PSP, MWn 63,000 and MWw 264,000 dalton) in a porcine model of faecal peritonitis. DESIGN: Randomised prospective study in 12 adolescent pigs. INTERVENTIONS: Prior to infection the study solution was infused to increase Qt by 25%. Thereafter adjustments in infusion rate were made (up to 1 l/h) in an attempt to maintain Qt at 25% above baseline values. MEASUREMENTS AND RESULTS: Animals were sacrificed at 8 h. Tissue was excised from the right lobe of liver and from the right lung and fixed for later electron microscopy and digital morphometric analysis. Patent sinusoidal lumen was significantly greater in group PDP compared to PSP (11.3% +/- 2.3% of liver tissue versus 4.8% +/- 1.1%, p < 0.05) and this was accounted for by a significantly lower proportion of sinusoidal lumen occluded with white cells (2.1% +/- 0.6% versus 6.6% +/- 1.9%, p < 0.05). Similarly, patent capillary represented a significantly higher proportion of lung tissue for group PDP versus PSP (26.2% +/- 1.9% versus 18.5% +/- 2.7%, p < 0.05). The arithmetic mean alveolar capillary barrier thickness was significantly greater in group PSP than in group PDP (4.3 +/- 0.3 microns versus 2.5 +/- 0.3 microns, p < 0.01). CONCLUSIONS: The molecular weight profile of Pentafraction was associated with less structural organ damage including less tissue oedema and less white cell occlusion.

Risk of dysphagia after transesophageal echocardiography during cardiac operations.

BACKGROUND: Dysphagia can be a significant complication following cardiac operations. This study evaluates its incidence and relationship to intraoperative transesophageal echocardiography (TEE) for specific indications versus known factors such as stroke or prolonged intubation. METHODS: Records of 838 consecutive cardiac surgical patients were reviewed, and categorized into those who received TEE for specific indications versus those who did not (nonTEE). Dysphagia was recorded when symptoms were confirmed by barium cineradiography. Multiple logistic regression identified significant factors causing dysphagia. RESULTS: TEE was significantly related to the development of postoperative dysphagia by multiple logistic regression (p < 0.001). After controlling for other significant factors (stroke, left ventricular ejection fraction, intubation time, duration of operation), the odds of dysphagia for TEE patients was 7.8 times greater than for nonTEE patients. CONCLUSIONS: TEE may be an independent risk factor for dysphagia following cardiac operations.

Echocardiography allows safer venous cannulation during excision of large right atrial masses.

BACKGROUND: Excision of large right atrial masses requires bicaval cannulation and cardiopulmonary bypass. Safe venous cannulation can be accomplished only by knowing the exact intracavitary location and extension of the mass to avoid fragmentation. Transthoracic echocardiography and intraoperative transesophageal echocardiography, although helpful, cannot always define the exact intracavitary relationships of the tumor. METHODS: We have used both intraoperative transesophageal and epicardial echocardiography to guide venous cannulation in 4 patients with large right atrial masses. Both echo images are used by the surgeon to select the exact site and method of cannulation to avoid fragmentation of the mass. Epicardial echocardiography complemented the images obtained by transesophageal echocardiography. RESULTS: The technique of combined transesophageal and epicardial echocardiography allowed safe venous cannulation in all 4 patients. Each of the right atrial masses was safely excised using case-specific cannulation techniques guided by the echocardiographic images. CONCLUSIONS: We propose the routine use of both intraoperative transesophageal and epicardial echocardiography in guiding venous cannulation for safe excision of large right atrial masses.

Superior vena caval syndrome: color flow Doppler detection of collateral venous channels.

Thrombosis of the superior vena cava (SVC) is an important clinical problem that requires prompt diagnosis. Confirmation of suspected SVC syndrome requires the use of an imaging study to document the obstruction and presence of collateral venous channels. Echocardiography is often used to evaluate suspected SVC obstruction; however, previous methods have allowed only indirect detection of collateral flow. We describe 3 patients with SVC obstruction in whom color flow Doppler allowed direct demonstration of collateral venous channels.