Experts vs. machine - comparison of machine learning to expert-informed prediction of outcome after major liver surgery.

BACKGROUND: Machine learning (ML) has been successfully implemented for classification tasks (e.g., cancer diagnosis). ML performance for more challenging predictions is largely unexplored. This study's objective was to compare machine learning vs. expert-informed predictions for surgical outcome in patients undergoing major liver surgery. METHODS: Single tertiary center data on preoperative parameters and postoperative complications for elective hepatic surgery patients were included (2008-2021). Expert-informed prediction models were established on 14 parameters identified by two expert liver surgeons to impact on postoperative outcome. ML models used all available preoperative patient variables (n = 62). Model performance was compared for predicting 3-month postoperative overall morbidity. Temporal validation and additional analysis in major liver resection patients were conducted. RESULTS: 889 patients included. Expert-informed models showed low average bias (2-5 CCI points) with high over/underprediction. ML models performed similarly: average prediction 5-10 points higher than observed CCI values with high variability (95% CI -30 to 50). No performance improvement for major liver surgery patients. CONCLUSION: No clinical relevance in the application of ML for predicting postoperative overall morbidity was found. Despite being a novel hype, ML has the potential for application in clinical practice. However, at this stage it does not replace established approaches of prediction modelling.

Surgical outcome improvement by shared decision-making: value of a preoperative multidisciplinary target clinic for the elderly in colorectal surgery.

PURPOSE: Frailty and comorbidities increase the risk of postoperative complications and raise treatment costs. Perioperative optimisation is shown to improve surgical outcomes for the elderly. The aim of this study was to assess the impact of introducing a multidisciplinary preoperative clinic for older patients (Colchester Older Persons' Evaluation for Surgery (COPES) clinic) undergoing major colorectal surgery. METHODS: This 5-year single centre study included patients >65 years with ≥3 comorbidities undergoing major colorectal surgery. From October 2018, patients with these characteristics were evaluated and optimised in the COPES clinic by a geriatrician and an anaesthetist. Outcomes were compared to high-risk patients operated on prior to COPES (pre-COPES group). The primary outcomes were postoperative morbidity at discharge and 6 months measured by the Comprehensive Complication Index. Patients were matched on age and number of comorbidities. RESULTS: A total of 54 patients were enrolled in the pre-COPES and 18 in the COPES group. After matching, the results were comparable for both groups. The length of stay was shorter in the COPES group and the recurrence rate was higher; however, it did not reach statistical significance in both findings. CONCLUSION: This clinic intends to improve treatment quality, placing emphasis on shared decision-making. More focus should be put on patient-reported outcomes and experiences. Especially for elderly patients, quality of life and maintaining independence are often their priority. To determine the true value of a preoperative multidisciplinary clinic targeting elderly comorbid patients, a prospective study with larger cohort is needed, focusing not only on objective outcomes but also on patient-reported outcomes.

Benchmarks in colorectal surgery: multinational study to define quality thresholds in high and low anterior resection.

BACKGROUND: Benchmark comparisons in surgery allow identification of gaps in the quality of care provided. The aim of this study was to determine quality thresholds for high (HAR) and low (LAR) anterior resections in colorectal cancer surgery by applying the concept of benchmarking. METHODS: This 5-year multinational retrospective study included patients who underwent anterior resection for cancer in 19 high-volume centres on five continents. Benchmarks were defined for 11 relevant postoperative variables at discharge, 3 months, and 6 months (for LAR). Benchmarks were calculated for two separate cohorts: patients without (ideal) and those with (non-ideal) outcome-relevant co-morbidities. Benchmark cut-offs were defined as the 75th percentile of each centre's median value. RESULTS: A total of 3903 patients who underwent HAR and 3726 who had LAR for cancer were analysed. After 3 months' follow-up, the mortality benchmark in HAR for ideal and non-ideal patients was 0.0 versus 3.0 per cent, and in LAR it was 0.0 versus 2.2 per cent. Benchmark results for anastomotic leakage were 5.0 versus 6.9 per cent for HAR, and 13.6 versus 11.8 per cent for LAR. The overall morbidity benchmark in HAR was a Comprehensive Complication Index (CCI®) score of 8.6 versus 14.7, and that for LAR was CCI® score 11.9 versus 18.3. CONCLUSION: Regular comparison of individual-surgeon or -unit outcome data against benchmark thresholds may identify gaps in care quality that can improve patient outcome.

How to Establish Benchmarks for Surgical Outcomes?: A Checklist Based on an International Expert Delphi Consensus.

OBJECTIVE: To define a standardized methodology for establishing benchmarks for relevant outcomes in surgery. SUMMARY BACKGROUND DATA: Benchmarking is an established tool to improve quality in industry and economics, and is emerging in assessing outcome values in surgery. Despite a recent 10-step approach to identify such benchmark values, a standardized and more widely agreed-on approach is still lacking. METHODS: A multinational web-based Delphi survey with a focus on methodological requirements for establishing benchmarks for surgical outcomes was performed. Participants were selected among internationally renowned specialists in abdominal, vascular, and thoracic surgery. Consensus was defined as ≥70% agreement and results were used to develop a checklist to establish benchmarks in surgery. RESULTS: Forty-one surgical opinion leaders from 19 countries and 5 continents were involved. Experts' response rates were 98% and 80% in rounds 1 and 2, respectively. Upon completion of the final Delphi round, consensus was successfully achieved for 26 of 36 items covering the following areas: center eligibility, validation of databases, patient cohort selection, procedure selection, duration of follow-up, statistical analysis, and publication requirements regarding center-specific outcomes. CONCLUSIONS: This multinational Delphi survey represents the first expert-led process for developing a standardized approach for establishing benchmarks for relevant outcome measures in surgery. The provided consensual checklist customizes the methodology of outcome reporting in surgery and thus improves reproducibility and comparability of data and should ultimately serve to improve quality of care.

Can Early Postoperative Complications Predict High Morbidity and Decrease Failure to Rescue Following Major Abdominal Surgery?

OBJECTIVE: To assess whether specific patterns of early postoperative complications may predict overall severe morbidity after major surgery, warranting early escalation of care and prevention of failure to rescue. SUMMARY OF BACKGROUND DATA: It is unclear whether early postoperative complications predict a poor outcome. Detailed knowledge of the chronology and type of early complications after major surgery may alert clinicians when to expect higher risk for subsequent major negative events. METHODS: All 90-day postoperative events following complex pancreas, liver, and rectal surgeries, and liver transplantation were analyzed over a 3-year period in a single tertiary center. Each complication was recorded regarding severity, type (cardiac, infectious, etc), etiology (surgical/medical), and timing of occurrence. The Comprehensive Complication Index (CCI), covering the first 7 postoperative days, was calculated as a measure for early cumulative postoperative morbidity. The statistical analysis (descriptive, sequence pattern analyses, and logistic regression analyses) aimed to detect any combinations of events predicting poor outcome as defined by a cumulative CCI ≥37.1 at 90-days. RESULTS: The occurrence of ≥2 complications, irrespective of severity, type or etiology, was strongly associated with a severe postoperative course (P < 0.001). Even 2 mild complications (≤ grade II) greatly increased the chance for high morbidity compared to patients with 0 or 1 complication within the first postoperative week (odds ratio 10.2, 95% confidence interval 5.82-17.98). The CCI at postoperative day 7 strongly predicted high 90-day morbidity (odds ratio 3.96 per 10 CCI points, P < 0.001). CONCLUSION: Multiple complications of any cause or severity within the first postoperative days represents a "warning-signal" for overall high morbidity by 90 days, which should be used to trigger an escalation of care to prevent failure to rescue and eventually poor outcome.

Defining Global Benchmarks in Bariatric Surgery: A Retrospective Multicenter Analysis of Minimally Invasive Roux-en-Y Gastric Bypass and Sleeve Gastrectomy.

OBJECTIVE: To define "best possible" outcomes for bariatric surgery (BS)(Roux-en-Y gastric bypass [RYGB] and sleeve gastrectomy [SG]). BACKGROUND: Reference values for optimal surgical outcomes in well-defined low-risk bariatric patients have not been established so far. Consequently, outcome comparison across centers and over time is impeded by heterogeneity in case-mix. METHODS: Out of 39,424 elective BS performed in 19 high-volume academic centers from 3 continents between June 2012 and May 2017, we identified 4120 RYGB and 1457 SG low-risk cases defined by absence of previous abdominal surgery, concomitant procedures, diabetes mellitus, sleep apnea, cardiopathy, renal insufficiency, inflammatory bowel disease, immunosuppression, anticoagulation, BMI>50 kg/m and age>65 years. We chose clinically relevant endpoints covering the intra- and postoperative course. Complications were graded by severity using the comprehensive complication index. Benchmark values were defined as the 75th percentile of the participating centers' median values for respective quality indicators. RESULTS: Patients were mainly females (78%), aged 38±11 years, with a baseline BMI 40.8 ±â€Š5.8 kg/m. Over 90 days, 7.2% of RYGB and 6.2% of SG patients presented at least 1 complication and no patients died (mortality in nonbenchmark cases: 0.06%). The most frequent reasons for readmission after 90-days following both procedures were symptomatic cholelithiasis and abdominal pain of unknown origin. Benchmark values for both RYGB and SG at 90-days postoperatively were 5.5% Clavien-Dindo grade ≥IIIa complication rate, 5.5% readmission rate, and comprehensive complication index ≤33.73 in the subgroup of patients presenting at least 1 grade ≥II complication. CONCLUSION: Benchmark cutoffs targeting perioperative outcomes in BS offer a new tool in surgical quality-metrics and may be implemented in quality-improvement cycle.ClinicalTrials.gov Identifier NCT03440138.

Procedural Surgical RCTs in Daily Practice: Do Surgeons Adopt Or Is It Just a Waste of Time?

OBJECTIVE: To assess the adoption of recommendation from randomized clinical trials (RCTs) and investigate factors favoring or preventing adoption. BACKGROUND: RCT are considered to be the cornerstone of evidence-based medicine by representing the highest level of evidence. As such, we expect RCT's recommendations to be followed rigorously in daily surgical practice. METHODS: We performed a structured search for RCTs published in the medical and surgical literature from 2009 to 2013, allowing a minimum of 5-year follow-up to convincingly test implementation. We focused on comparative technical or procedural RCTs trials addressing the domains of general, colorectal, hepatobiliary, upper gastrointestinal and vascular surgery. In a second step we composed a survey of 29 questions among ESA members as well as collaborators from their institutions to investigate the adoption of surgical RCTs recommendation. RESULTS: The survey based on 36 RCTs (median 5-yr citation index 85 (24-474), from 21 different countries, published in 15 high-ranked journals with a median impact factor of 3.3 (1.23-7.9) at the time of publication. Overall, less than half of the respondents (47%) appeared to adhere to the recommendations of a specific RCT within their field of expertise, even when included in formal guidelines. Adoption of a new surgical practice was favored by watching videos (46%) as well as assisting live operations (18%), while skepticism regarding the methodology of a surgical RCT (40%) appears to be the major reason to resist adoption. CONCLUSION: In conclusion, surgical RCTs appear to have moderate impact on daily surgical practice. While RCTs are still accepted to provide the highest level of evidence, alternative methods of evaluating surgical innovations should also be explored.

Benchmark analyses in minimally invasive esophagectomy-impact on surgical quality improvement.

Over the last decades, benchmarking has become an established management tool to improve quality in commercial economics. It is a rather new concept in the healthcare industry, and a confusingly wide range of approaches referring to "benchmarking" have been employed in the field of minimally invasive esophageal cancer surgery. It is our conviction that benchmarking will be an essential element of surgical research in the future. Therefore, defining and implementing standards is not only a desirable, but a vital step. Recently, we have introduced a standardized method of establishing valid benchmarks for surgical quality improvement including ideal outcome thresholds for total minimally invasive transthoracic esophagectomy (ttMIE). The present article aims at discussing the actual literature on benchmarking in minimally invasive esophagectomy (MIE) and at fueling the debate on how to further improve the current practice of surgical outcome research.

The Comprehensive Complication Index (CCI®) is a Novel Cost Assessment Tool for Surgical Procedures.

OBJECTIVE: The aim of this study was to identify a readily available, reproducible, and internationally applicable cost assessment tool for surgical procedures. SUMMARY OF BACKGROUND DATA: Strong economic pressure exists worldwide to slow down the rising of health care costs. Postoperative morbidity significantly impacts on cost in surgical patients. The comprehensive complication index (CCI), reflecting overall postoperative morbidity, may therefore serve as a new marker for cost. METHODS: Postoperative complications and total costs from a single tertiary center were prospectively collected (2014 to 2016) up to 3 months after surgery for a variety of abdominal procedures (n = 1388). CCI was used to quantify overall postoperative morbidity. Pearson correlation coefficient (rpears) was calculated for cost and CCI. For cost prediction, a linear regression model based on CCI, age, and type of surgery was developed and validated in an international cohort of patients. RESULTS: We found a high correlation between CCI and overall cost (rpears = 0.75) with the strongest correlation for more complex procedures. The prediction model performed very well (R = 0.82); each 10-point increase in CCI corresponded to a 14% increase to the baseline cost. Additional 12% of baseline cost must be added for patients older than 50 years, or 24% for those over 70 years. The validation cohorts showed a good match of predicted and observed cost. CONCLUSION: Overall postoperative morbidity correlates highly with cost. The CCI together with the type of surgery and patient age is a novel and reliable predictor of expenses in surgical patients. This finding may enable objective cost comparisons among centers, procedures, or over time obviating the need to look at complex country-specific cost calculations (www.assessurgery.com).

Defining Benchmarks in Liver Transplantation: A Multicenter Outcome Analysis Determining Best Achievable Results.

: This multicentric study of 17 high-volume centers presents 12 benchmark values for liver transplantation. Those values, mostly targeting markers of morbidity, were gathered from 2024 "low risk" cases, and may serve as reference to assess outcome of single or any groups of patients. OBJECTIVE: To propose benchmark outcome values in liver transplantation, serving as reference for assessing individual patients or any other patient groups. BACKGROUND: Best achievable results in liver transplantation, that is, benchmarks, are unknown. Consequently, outcome comparisons within or across centers over time remain speculative. METHODS: Out of 7492 liver transplantation performed in 17 international centers from 3 continents, we identified 2024 low risk adult cases with a laboratory model for end-stage liver disease score ≤20 points, a balance of risk score ≤9, and receiving a primary graft by donation after brain death. We chose clinically relevant endpoints covering intra- and postoperative course, with a focus on complications graded by severity including the complication comprehensive index (CCI). Respective benchmarks were derived from the median value in each center, and the 75 percentile was considered the benchmark cutoff. RESULTS: Benchmark cases represented 8% to 49% of cases per center. One-year patient-survival was 91.6% with 3.5% retransplantations. Eighty-two percent of patients developed at least 1 complication during 1-year follow-up. Biliary complications occurred in one-fifth of the patients up to 6 months after surgery. Benchmark cutoffs were ≤4 days for ICU stay, ≤18 days for hospital stay, ≤59% for patients with severe complications (≥ Grade III) and ≤42.1 for 1-year CCI. Comparisons with the next higher risk group (model for end stage liver disease 21-30) disclosed an increase in morbidity but within benchmark cutoffs for most, but not all indicators, while in patients receiving a second graft from 1 center (n = 50) outcome values were all outside of benchmark values. CONCLUSIONS: Despite excellent 1-year survival, morbidity in benchmark cases remains high with half of patients developing severe complications during 1-year follow-up. Benchmark cutoffs targeting morbidity parameters offer a valid tool to assess higher risk groups.

Spiradenocarcinoma: A Comprehensive Data Review.

INTRODUCTION: Spiradenocarcinomas (SCs) are rare and potentially aggressive skin adnexal tumors. Optimal treatment has not yet been established. Experiences with this carcinoma are mostly presented in case reports and few case series. OBJECTIVE: To generate to a synopsis of published data on SC with regard to diagnostic procedures, treatment, and outcome. RESULTS: Median patient age was 60 years and sex distribution was balanced. Tumor manifestations were evenly distributed within the sweat gland carrying skin. The most commonly reported symptom was accelerated growth of a longstanding indolent lesion, typically present for more than 2 years. Metastatic spread to the lung, bone, lymph nodes, liver, kidney, and breast has been documented. For staging computed tomography (CT) and positron emission tomography-CT are recommended, especially for detection of hematogenic metastases and lymph node involvement. Clear resection margins and tumor free regional lymph nodes reduce recurrence and carcinoma related death. Although low-grade SCs were reported over 3 times more often, high-grade carcinomas show a greater likelihood for recurrence and lethal outcome. CONCLUSION: Suspicion of an SC should lead to performance of a magnetic resonance imaging for defining tumor extent, and a fludeoxyglucose positron emission tomography-CT for detection of metastases. Radical tumor excision and resection of tumor involved regional lymph nodes are essential for a curative approach. Histopathological evaluation should involve determination of tumor differentiation grade, because high-grade carcinomas seem to have a much more aggressive behavior. Excision of distant metastases has no therapeutic value. Follow-up needs to be carried out in short intervals with frequent imaging.

The Comprehensive Complication Index (CCI®): Added Value and Clinical Perspectives 3 Years "Down the Line".

OBJECTIVE: To explore the added value of the comprehensive complication index (CCI) to standard assessment of postoperative morbidity, and to clarify potential controversies for its application. BACKGROUND: The CCI was introduced about 3 years ago as a novel metric of postoperative morbidity, integrating in a single formula all complications by severity, ranging from 0 (uneventful course) to 100 (death). It remains unclear, how often the CCI adds to standard reporting of complications and how to apply it in complex postoperative courses. METHODS: CCI data were prospectively collected over a 1-year period at our institution. The proportion of patients with more than 1 complication and the severity of those complications were assessed to determine the additional value of the CCI compared to the Clavien-Dindo classification. Complex and controversial cases were presented to 90 surgeons worldwide to achieve consensus in weighing each postoperative event. Descriptive statistics were used to evaluate agreement among surgeons and to suggest solutions for consistent use of the CCI. RESULTS: Complications were identified in 24% (290/1212) of the general surgical population. Of those, 44% (127/290) developed more than 1 complication by the time of discharge, and thereby CCI added information to the standard grading system of complications. Information gained by the CCI increased with the complexity of surgery and observation time. CONCLUSIONS: The CCI adds information on postoperative morbidity in almost half of the patients developing complications, with particular value following extensive surgery and longer postoperative observation up to 3 months. Each single complication, independently of their inter-connection, should be included in the CCI calculation to best mirror the patients' postoperative morbidity.

Prognostic factors for impaired plasma sodium homeostasis after transsphenoidal surgery.

OBJECTIVE: Surgical manipulation of the pituitary stalk, neurohypophysis or the hypothalamus may disturb control of the plasma sodium level. The factors that might predict the risk of postoperative sodium imbalance are not clear, and were investigated in this study. METHODS: A retrospective survey of 129 surgical records for the occurrence of plasma sodium levels outside the normal range, following transsphenoidal procedures. Median patient age was 49 (range 20-78) years, 65 female. 73 of the operated lesions were non-functioning pituitary adenomas. Patients were considered to have impaired plasma sodium balance if the range of 135-145 mmol/L was not maintained. RESULTS: Of all 129 surgical cases, 68 (53%) experienced an imbalance in sodium levels. Severe sodium imbalance (≥ 149 or ≤ 131 mmol/L) was observed in 28 patients (22%). 13 showed hypernatraemia (median day 1), and 15 hyponatraemia (median day 6). Tumour size was associated with an increased incidence of sodium imbalance, particularly in patients younger than 49 years; surgery resulted in sodium imbalance in 38% of young patients operated on for tumours < 22 mm and in 76% of young patients, operated on for tumours ≥ 22 mm. Overall, tumour size increased with patients' age, and tumour size was less predictive for sodium disturbances in elderly patients. Median time in hospital was 5 days for patients without sodium imbalance, 6 days for patients with hypernatraemia and 11 days for patients with hyponatraemia. CONCLUSIONS: Following pituitary surgery, patients with large tumours, in particular those of young age, are at higher risk for losing control of their plasma sodium level. Increased ADH secretion (hyponatraemia), but not transient diabetes insipidus was associated with a prolonged hospital stay. Postoperative follow-up of patients with sellar tumours should include careful monitoring of plasma sodium levels within the first two postoperative weeks and clear patients' instructions.