Question Can big-data analysis of clinical audits help to find new risk factors and predict adverse events associated with colorectal cancer surgery? Findings This cohort study found that machine... Show moreQuestion Can big-data analysis of clinical audits help to find new risk factors and predict adverse events associated with colorectal cancer surgery? Findings This cohort study found that machine learning applied to a clinical audit containing 62 501 records and 103 preoperative variables of surgically treated patients with colorectal cancer outperformed conventional scores in predicting 30-day postoperative mortality but with similar performance as a preexisting case-mix model. New risk factors for several other adverse events may be identified. Meaning This study suggests that machine learning methods may be of additional value in analyzing quality indicators in colorectal cancer surgery, thereby providing directions to optimize case-mix corrections for benchmarking in clinical auditing.Importance Quality improvement programs for colorectal cancer surgery have been introduced with benchmarking based on quality indicators, such as mortality. Detailed (pre)operative characteristics may offer relevant information for proper case-mix correction. Objective To investigate the added value of machine learning to predict quality indicators for colorectal cancer surgery and identify previously unrecognized predictors of 30-day mortality based on a large, nationwide colorectal cancer registry that collected extensive data on comorbidities. Design, Setting, and Participants All patients who underwent resection for primary colorectal cancer registered in the Dutch ColoRectal Audit between January 1, 2011, and December 31, 2016, were included. Multiple machine learning models (multivariable logistic regression, elastic net regression, support vector machine, random forest, and gradient boosting) were made to predict quality indicators. Model performance was compared with conventionally used scores. Risk factors were identified by logistic regression analyses and Shapley additive explanations (ie, SHAP values). Statistical analysis was performed between March 1 and September 30, 2020. Main Outcomes and Measures The primary outcome of this cohort study was 30-day mortality. Prediction models were trained on a training set by performing 5-fold cross-validation, and outcomes were measured by the area under the receiver operating characteristic curve on the test set. Machine learning was further used to identify risk factors, measured by odds ratios and SHAP values. Results This cohort study included 62 501 records, most patients were male (35 116 [56.2%]), were aged 61 to 80 years (41 560 [66.5%]), and had an American Society of Anesthesiology score of II (35 679 [57.1%]). A 30-day mortality rate of 2.7% (n = 1693) was found. The area under the curve of the best machine learning model for 30-day mortality (0.82; 95% CI, 0.79-0.85) was significantly higher than the American Society of Anesthesiology score (0.74; 95% CI, 0.71-0.77; P < .001), Charlson Comorbidity Index (0.66; 95% CI, 0.63-0.70; P < .001), and preoperative score to predict postoperative mortality (0.73; 95% CI, 0.70-0.77; P < .001). Hypertension, myocardial infarction, chronic obstructive pulmonary disease, and asthma were comorbidities with a high risk for increased mortality. Machine learning identified specific risk factors for a complicated course, intensive care unit admission, prolonged hospital stay, and readmission. Laparoscopic surgery was associated with a decreased risk for all adverse outcomes. Conclusions and Relevance This study found that machine learning methods outperformed conventional scores to predict 30-day mortality after colorectal cancer surgery, identified specific patient groups at risk for adverse outcomes, and provided directions to optimize benchmarking in clinical audits.This cohort study investigates the ability of machine learning to predict quality indicators for colorectal cancer surgery and identify previously unrecognized predictors of 30-day mortality based on a large nationwide colorectal cancer registry that collected extensive data on comorbidities. Show less
Purpose Interhospital referral is a consequence of centralization of complex oncological care but might negatively impact waiting time, a quality indicator in the Netherlands. This study aims to... Show morePurpose Interhospital referral is a consequence of centralization of complex oncological care but might negatively impact waiting time, a quality indicator in the Netherlands. This study aims to evaluate characteristics and waiting times of patients with primary colorectal cancer who are referred between hospitals. Methods Data were extracted from the Dutch ColoRectal Audit (2015-2019). Waiting time between first tumor-positive biopsy until first treatment was compared between subgroups stratified for referral status, disease stage, and type of hospital. Results In total, 46,561 patients were included. Patients treated for colon or rectal cancer in secondary care hospitals were referred in 12.2% and 14.7%, respectively. In tertiary care hospitals, corresponding referral rates were 43.8% and 66.4%. Referred patients in tertiary care hospitals were younger, but had a more advanced disease stage, and underwent more often multivisceral resection and simultaneous metastasectomy than non-referred patients in secondary care hospitals (p<0.001). Referred patients were more often treated within national quality standards for waiting time compared to non-referred patients (p<0.001). For referred patients, longer waiting times prior to MDT were observed compared to non-referred patients within each hospital type, although most time was spent post-MDT. Conclusion A large proportion of colorectal cancer patients that are treated in tertiary care hospitals are referred from another hospital but mostly treated within standards for waiting time. These patients are younger but often have a more advanced disease. This suggests that these patients are willing to travel more but also reflects successful centralization of complex oncological patients in the Netherlands. Show less
Introduction: The revised Dutch colorectal cancer guideline (2014), led to an overall decrease in preoperative radiotherapy (RT) use. This study evaluates hospital variation in RT use for... Show moreIntroduction: The revised Dutch colorectal cancer guideline (2014), led to an overall decrease in preoperative radiotherapy (RT) use. This study evaluates hospital variation in RT use for resectable rectal cancer and the influence of guideline revision, including the nationwide impact of changing RT application on short term outcomes.Methods: Data of surgically resected rectal cancer patients registered in the Dutch ColoRectal Audit were extracted between 2011 and 2017. Patients were divided into groups based on time of guideline revision (<2014 and >= 2014). Primary outcome was guideline adherence at hospital level regarding RT application, stratified for three stage groups. Secondary outcomes included positive circumferential resection (CRM+) and 30-day complicated postoperative course.Results: The groups consisted of 7364 and 12,057 patients, respectively. In total, 6772 patients did not receive RT (17.6% (<2014) vs. 45.7% (>= 2014), p < 0.001). The largest increase of surgery alone was observed for cT1-2N0 stage rectal cancer (35.1% vs. 91.8%, p < 0.001), with a substantial decrease in hospital variation (IQR 22.2-50.0% vs. IQR 87.6-98.0%). For cT1-3N1MRF-stage rectal cancer, a substantial amount of hospital variation in short course RT remained after guideline revision (IQR 26.8-54.1% vs. IQR 26.2-50.0%). A significant decrease in CRMthorn (5.8% vs. 4.2%, p < 0.001) and complicated course (22.5% vs. 18.5%, p < 0.001) was observed.Conclusions: Radiotherapy for early-stage rectal cancer was uniformly abandoned after guideline revision, while substantial hospital variation remained for intermediate risk resectable rectal cancer in the Netherlands. The substantial nationwide decrease in the use of RT for rectal cancer treatment did not negatively impact CRM involvement. (C) 2020 Elsevier Ltd, BASO similar to The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved. Show less
Aim This study aimed to determine predictive factors for the circumferential resection margin (CRM) within two northern European countries with supposed similarity in providing rectal cancer care.... Show moreAim This study aimed to determine predictive factors for the circumferential resection margin (CRM) within two northern European countries with supposed similarity in providing rectal cancer care. Method Data for all patients undergoing rectal resection for clinical tumour node metastasis (TNM) stage I-III rectal cancer were extracted from the Swedish ColoRectal Cancer Registry and the Dutch ColoRectal Audit (2011-2015). Separate analyses were performed for cT1-3 and cT4 stage. Predictive factors for the CRM were determined using univariable and multivariable logistic regression analyses. Results A total of 6444 Swedish and 12 089 Dutch patients were analysed. Over time the number of hospitals treating rectal cancer decreased from 52 to 42 in Sweden, and 82 to 79 in the Netherlands. In the Swedish population, proportions of cT4 stage (17% vs 8%), multivisceral resection (14% vs 7%) and abdominoperineal excision (APR) (37% vs 31%) were higher. The overall proportion of patients with a positive CRM (CRM+) was 7.8% in Sweden and 5.4% in the Netherlands. In both populations with cT1-3 stage disease, common independent risk factors for CRM+ were cT3, APR and multivisceral resection. No common risk factors for CRM+ in cT4 stage disease were found. An independent impact of hospital volume on CRM+ could be demonstrated for the cT1-3 Dutch population. Conclusion Within two northern European countries with implemented clinical auditing, rectal cancer care might potentially be improved by further optimizing the treatment of distal and locally advanced rectal cancer. Show less
Background A multicentre cohort study was performed to analyse the motivations for surgical referral of patients with benign colorectal lesions, and to evaluate the endoscopic and pathological... Show moreBackground A multicentre cohort study was performed to analyse the motivations for surgical referral of patients with benign colorectal lesions, and to evaluate the endoscopic and pathological characteristics of these lesions as well as short-term surgical outcomes. Methods Patients who underwent surgery for a benign colorectal lesion in 15 Dutch hospitals between January 2014 and December 2017 were selected from the pathology registry. Lesions were defined as complex when at least one of the following features was present: size at least 40 mm, difficult location according to the endoscopist, previous failed attempt at resection, or non-lifting sign. Results A total of 358 patients were included (322 colonic and 36 rectal lesions). The main reasons for surgical referral of lesions in the colon and rectum were large size (33 center dot 5 and 47 per cent respectively) and suspicion of invasive growth (31 center dot 1 and 58 per cent). Benign lesions could be categorized as complex in 80 center dot 6 per cent for colonic and 80 per cent for rectal locations. Surgery consisted of local excision in 5 center dot 9 and 64 per cent of colonic and rectal lesions respectively, and complicated postoperative course rates were noted in 11 center dot 2 and 3 per cent. In the majority of patients, no attempt was made to resect the lesion endoscopically (77 center dot 0 per cent of colonic and 83 per cent of rectal lesions). Conclusion The vast majority of the benign lesions referred for surgical resection could be classified as complex. Considering the substantial morbidity of surgery for benign colonic lesions, reassessment for endoscopic resection by another advanced endoscopy centre seems to be underused and should be encouraged. Show less
