Cardiac Surgery Risk Models: A Position Article

Artificial intelligence in cardiovascular procedures: a bibliometric and visual analysis study

Background

The integration of artificial intelligence (AI) into cardiovascular procedures has significantly advanced diagnostic accuracy, outcome prediction, and robotic-assisted surgeries. However, a comprehensive bibliometric analysis of AI's impact in this field is lacking. This study examines research trends, key contributors, and emerging themes in AI-driven cardiovascular interventions.

Methods

We retrieved relevant publications from the Web of Science Core Collection and analyzed them using VOSviewer, CiteSpace, and Biblioshiny to map research trends and collaborations.

Results

AI-related cardiovascular research has grown substantially from 1993 to 2024, with a sharp increase from 2020 to 2023, peaking at 93 publications in 2023. The USA (127 papers), China (79), and England (31) were the top contributors, with Harvard University leading institutional output (17 papers). Frontiers in Cardiovascular Medicine was the most prolific journal. Core research themes included "machine learning," "mortality," and "cardiac surgery," with emerging trends in "association," "implantation," and "aortic stenosis," underscoring AI's expanding role in predictive modeling and surgical outcomes.

Conclusion

AI demonstrates transformative potential in cardiovascular procedures, particularly in diagnostic imaging, predictive modeling, and patient management. This bibliometric analysis highlights the growing interest in AI applications and provides a framework for integrating AI into clinical workflows to enhance diagnostic accuracy, treatment strategies, and patient outcomes.

Survey of New Zealand Arthroplasty Surgeons on Surgeon-Level Outcome Reporting

BACKGROUND

Surgeon-specific outcome monitoring has become increasingly prevalent over the last 3 decades. The New Zealand Orthopaedic Association monitors individual surgeon performance through 2 mechanisms: arthroplasty revision rates derived from the New Zealand Joint Registry and a practice visit program. Despite remaining confidential, surgeon-level outcome reporting remains contentious. The purpose of this survey was to evaluate the opinions of hip and knee arthroplasty surgeons in New Zealand on the perceived importance of outcome monitoring, current methods used to evaluate surgeon-specific outcomes, and potential improvements identified through literature review and discussion with other registries.

METHODS

The survey consisted of 9 questions on surgeon-specific outcome reporting, using a five-point Likert scale, and 5 demographic questions. It was distributed to all current hip and knee arthroplasty surgeons. There were 151 hip and knee arthroplasty surgeons who completed the survey, a response rate of 50%.

RESULTS

Respondents agreed that monitoring arthroplasty outcomes is important and that revision rates are an acceptable measure of performance. Reporting risk-adjusted revision rates and more recent timeframes were supported, as was including patient-reported outcomes when monitoring performance. Surgeons did not support public reporting of surgeon-level or hospital-level outcomes.

CONCLUSION

The findings of this survey support the use of revision rates to confidentially monitor surgeon-level arthroplasty outcomes and suggest that concurrent use of patient-reported outcome measures would be acceptable.

Redefining "low risk": Outcomes of surgical aortic valve replacement in low-risk patients in the transcatheter aortic valve replacement era

OBJECTIVES

Guidelines suggest aortic valve replacement (AVR) for low-risk asymptomatic patients. Indications for transcatheter AVR now include low-risk patients, making it imperative to understand state-of-the-art surgical AVR (SAVR) in this population. Therefore, we compared SAVR outcomes in low-risk patients with those expected from Society of Thoracic Surgeons (STS) models and assessed their intermediate-term survival.

METHODS

From January 2005 to January 2017, 3493 isolated SAVRs were performed in 3474 patients with STS predicted risk of mortality <4%. Observed operative mortality and composite major morbidity or mortality were compared with STS-expected outcomes according to calendar year of surgery. Logistic regression analysis was used to identify risk factors for these outcomes. Patients were followed for time-related mortality.

RESULTS

With 15 observed operative deaths (0.43%) compared with 55 expected (1.6%), the observed:expected ratio was 0.27 for mortality (95% confidence interval [CI], 0.14-0.42), stroke 0.65 (95% CI, 0.41-0.89), and reoperation 0.50 (95% CI, 0.42-0.60). Major morbidity or mortality steadily declined, with probabilities of 8.6%, 6.7%, and 5.2% in 2006, 2011, and 2016, respectively, while STS-expected risk remained at approximately 12%. Mitral valve regurgitation, ventricular hypertrophy, pulmonary, renal, and hepatic failure, coronary artery disease, and earlier surgery date were residual risk factors. Survival was 98%, 91%, and 82% at 1, 5, and 9 years, respectively, superior to that predicted for the US age-race-sex-matched population.

CONCLUSIONS

STS risk models overestimate contemporary SAVR risk at a high-volume center, supporting efforts to create a more agile quality assessment program. SAVR in low-risk patients provides durable survival benefit, supporting early surgery and providing a benchmark for transcatheter AVR.

Risk prediction models in patients undergoing percutaneous coronary intervention: A collaborative analysis from a Japanese administrative dataset and nationwide academic procedure registry

BACKGROUND

Contemporary guidelines emphasize the importance of risk stratification in improving the quality of care for patients undergoing percutaneous coronary intervention (PCI). We aimed to investigate whether adding information from a procedure-based academic registry to administrative claims data would improve the performance of risk prediction model.

METHODS

We combined two nationally representative administrative and clinical databases. The study cohort comprised 43,095 patients; 18,719 and 23, 525 with acute [ACS] and chronic [CCS] coronary syndrome, respectively. Each population was randomly divided into the logistic regression model (derivation cohort, 80%) and model validation (validation cohort, 20%) groups. The performances of the following models were compared using C-statistics: (1) variables restricted to baseline claims data (model #1), (2) clinical registry data (model #2), and (3) expanded to both claims and clinical registry data (model #3). The primary outcomes were in-hospital mortality and bleeding.

RESULTS

The primary outcomes occurred in 3.7% (in-hospital mortality)/5.0% (bleeding) of patients with ACS and 0.21%/0.95% of CCS patients. For each event, the model performance was 0.65 (95% confidence interval [CI], 0.60-0.69) /0.67 (0.63-0.71) in ACS and 0.52 (0.35-0.76) /0.62 (0.54-0.70) for CCS patients in model #1, 0.83 (0.80-0.87) /0.77 (0.74-0.81) in ACS and 0.76 (0.60-0.92) /0.67 (0.59-0.75) in CCS for model #2, and 0.83 (0.79-0.86) /0.78 (0.75-0.81) in ACS and 0.76 (0.61-0.92) /0.67 (0.58-0.74) in CCS for model #3.

CONCLUSIONS

Combining clinical information from the academic registry with claims databases improved its performance in predicting adverse events.

Observed versus predicted mortality after isolated tricuspid valve surgery

BACKGROUND

Aim of this study is to analyse the performances of Clinical Risk Score (CRS) and European System for Cardiac Operative Risk Evaluation (EuroSCORE)-II in isolated tricuspid surgery.

METHODS

Three hundred and eighty-three patients (54 ± 16 year; 54% female) were enrolled. Receiver operating characteristic analysis was performed to evaluate the relationship between the true positive fraction of test results and the false-positive fraction for a procedure.

RESULTS

Considering the 30-day mortality the area under the curve was 0.6 (95% confidence interval [CI] 0.50-0.72) for EuroSCORE II and 0.7 (95% CI 0.56-0.84) for CRS-score. The ratio of expected/observed mortality showed underestimation when considering EuroSCORE-II (min. 0.46-max. 0.6). At multivariate analysis, the CRS score (p = .005) was predictor of late cardiac death.

CONCLUSION

We suggest using both scores to obtain a range of expected mortality. CRS to speculate on late survival.

Development of machine learning models for mortality risk prediction after cardiac surgery

BACKGROUND

We developed machine learning models that combine preoperative and intraoperative risk factors to predict mortality after cardiac surgery.

METHODS

Machine learning involving random forest, neural network, support vector machine, and gradient boosting machine was developed and compared with the risk scores of EuroSCORE I and II, Society of Thoracic Surgeons (STS), as well as a logistic regression model. Clinical data were collected from patients undergoing adult cardiac surgery at the First Medical Centre of Chinese PLA General Hospital between December 2008 and December 2017. The primary outcome was post-operative mortality. Model performance was estimated using several metrics, including sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC). The visualization algorithm was implemented using Shapley's additive explanations.

RESULTS

A total of 5,443 patients were enrolled during the study period. The mean EuroSCORE II score was 3.7%, and the actual in-hospital mortality rate was 2.7%. For predicting operative mortality after cardiac surgery, the AUC scores were 0.87, 0.79, 0.81, and 0.82 for random forest, neural network, support vector machine, and gradient boosting machine, compared with 0.70, 0.73, 0.71, and 0.74 for EuroSCORE I and II, STS, and logistic regression model. Shapley's additive explanations analysis of random forest yielded the top-20 predictors and individual-level explanations for each prediction.

CONCLUSIONS

Machine learning models based on available clinical data may be superior to clinical scoring tools in predicting postoperative mortality in patients following cardiac surgery. Explanatory models show the potential to provide personalized risk profiles for individuals by accounting for the contribution of influencing factors. Additional prospective multicenter studies are warranted to confirm the clinical benefit of these machine learning-driven models.

Calibration Factors for STS Risk Model Predictions: Why, How and When They Are Used

The Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database is the world's premier adult cardiac surgery outcomes registry. This tutorial explains: how STS updates the risk models that are used to calculate the predicted risks of adverse events in the registry; why STS quarterly adjusts or "calibrates" the observed to expected ratios to equal one (O/E=1), effectively making the annual number of adverse events predicted by the model match the annual number of adverse events observed in the entire registry; the differences between the calibrated and uncalibrated O/E ratios; and how and when to use each.

Risk-adjusted rates between hospitals for adverse outcomes of very-low-birth-weight infants

BACKGROUND/PURPOSE

To analyze the amount of variation in these risk-adjusted adverse outcomes corresponding to the care of premature births. In addition, hospitals were ranked according to their unadjusted and adjusted rates, and we assessed the degree of concordance between these rankings. Finally, the correlations of hospital-adjusted adverse outcomes were also tested.

METHODS

The study utilized the 5-year Taiwan Premature Infant Follow-up Network (TPFN) database in Taiwan from 2014 to 2018, and the sample size was 6482. We calculated the "observed over expected" (OE) ratio every year to form the risk-adjusted adverse outcome rate for each hospital.

RESULTS

There was a larger variation in the risk-adjusted rate for NEC and the second-largest variation for IVH. Regarding the concordances between the unadjusted and adjusted ranks, the ranks for mortality had the lowest concordance (coefficient of concordance 0.64), and only a few of the risk-adjusted rates between outcomes were significantly correlated.

CONCLUSION

The results of the TPFN show that there is room to improve performance in terms of large variations in NEC and IVH. Furthermore, risk adjustment is important, especially for mortality, since the ranks for mortality have the lowest concordance. Finally, we cannot generate a conclusion regarding whether a hospital is high in quality if we only take 1 or 2 adverse outcomes as profiling measures because only a few of the risk-adjusted rates between outcomes were significantly correlated.

Risk Stratification for Transcatheter Aortic Valve Replacement

Risk assessment models developed from administrative and clinical databases are used for clinical decision making. Since these models are derived from a database, they have an inherent limitation of being as good as the data they are derived from. Many of these models under or overestimate certain clinical outcomes particularly mortality in certain group of patients. Undeniably, there is significant variability in all these models on account of patient population studied, the statistical analysis used to develop the model and the period during which these models were developed. This review aims to shed light on development and application of risk assessment models for cardiac surgery with special emphasis on risk stratification in severe aortic stenosis to select patients for transcatheter aortic valve replacement.

B-Type Natriuretic Peptide as a Predictor of Short-Term Mortality in On-Pump Coronary Artery Bypass Grafting

Objective

The present study refers to a determination of the preoperative B-type natriuretic peptide is a predictor of short-term all-cause mortality in patients undergoing on-pump coronary artery bypass graft surgeries.

Methods

Two hundred and twenty-one patients undergoing on-pump coronary artery bypass graft surgeries were evaluated prospectively during a 30-day postoperative follow-up period. Serum B-type natriuretic peptide concentration was measured without a 24-hour period prior to the surgical procedure and the value obtained was correlated with a short-term all-cause mortality.

Results

Data analysis showed that all-cause mortality rates were equal to 9.5% in 30 days. Accuracy analysis by the receiver operating characteristic curve found an ideal cut-off value of B-type natriuretic peptide equal to 150 pg/mL in relation to mortality (AUC=0.82, 95% CI=0.71-0.87, P<0.001). Multivariate analysis showed that B-type natriuretic peptide value greater than or equal to 150 pg/mL (P=0.030, HR=3.99, 95% CI=1.14-13.98) was an independent predictor of all-cause mortality in a 30-day follow-up period.

Conclusion

Preoperative serum B-type natriuretic peptide concentration is an independent predictor of short-term all-cause mortality in patients undergoing coronary artery bypass grafting with cardiopulmonary bypass.

Development and performance of a diagnostic/prognostic scoring system for breakthrough pain

OBJECTIVES

Variable prevalence and treatment of breakthrough pain (BTP) in different clinical contexts are partially due to the lack of reliable/validated diagnostic tools with prognostic capability. We report the statistical basis and performance analysis of a novel BTP scoring system based on the naïve Bayes classifier (NBC) approach and an 11-item IQ-BTP validated questionnaire. This system aims at classifying potential BTP presence in three likelihood classes: "High," "Intermediate," and "Low."

METHODS

Out of a training set of n=120 mixed chronic pain patients, predictors associated with the BTP likelihood variables (Pearson's χ² and/or Fisher's exact test) were employed for the NBC planning. Adjusting the binary classification to a three-likelihood classes case enabled the building of a scoring algorithm and to retrieve the score of each predictor's answer options and the Patient's Global Score (PGS). The latter medians were used to establish the NBC thresholds, needed to evaluate the scoring system performance (leave-one-out cross-validation).

RESULTS

Medians of PGS in the "High," "Intermediate," and "Low" likelihood classes were 3.44, 1.53, and -2.84, respectively. Leading predictors for the model (based on score differences) were flair frequency (ΔS=1.31), duration (ΔS=5.25), and predictability (ΔS=1.17). Percentages of correct classification were 63.6% for the "High" and of 100.0% for either the "Intermediate" and "Low" likelihood classes; overall accuracy of the scoring system was 90.9%.

CONCLUSION

The NBC-based BTP scoring system showed satisfactory performance in classifying potential BTP in three likelihood classes. The reliability, flexibility, and simplicity of this statistical approach may have significant relevance for BTP epidemiology and management. These results need further impact studies to generalize our findings.

Comparison of four risk scores for in-hospital mortality in patients undergoing heart valve surgery: A multicenter study in a Chinese population

BACKGROUND AND AIM OF THE STUDY

To compare four risk scores with regard to their validity to predict in-hospital mortality after heart valve surgery in a multicenter patient population of China.

MATERIALS AND METHODS

From January 2009 to December 2012, data from 12,412 consecutive patients older than 16 years who underwent heart valve surgery at four cardiac surgical centers were collected and scored according to the EuroSCORE II, Ambler risk score, NYC risk score, and STS risk score. The patients were divided into two subgroups according to the types of valve procedures, and the performance of the four risk scores for each group was assessed. Calibration was assessed by the Hosmer-Lemeshow (H-L) test. Discrimination was tested by calculating the area under the receiver operating characteristic (ROC) curve.

RESULTS

Observed mortality was 2.09% overall. The EuroSCORE II, Ambler score, and NYC score overpredicted observed mortality (Hosmer-Lemeshow: P = 0.002, P < 0.0001, and P < 0.0001, respectively) and the STS score underpredicted observed mortality (Hosmer-Lemeshow: P = 0.001). The discriminative power in the entire cohort for in-hospital mortality was highest for the STS score (0.735), followed by the EuroSCORE II score (0.704), NYC score (0.693), and Ambler score (0.674). Meanwhile, the STS score and EuroSCORE II give an accurate prediction in patients undergoing single valve surgery compared with the Ambler score and NYC score. However, all four risk scores give an imprecise prediction in patients undergoing multiple valve surgery.

CONCLUSIONS

Both the STS score and Euroscore II, especially the STS score, were suitable for individual operative risk in Chinese patients undergoing single valve surgery compared with the Ambler score and NYC score, however, all four risk scores were not suitable for prediction in Chinese patients undergoing multiple valve surgery. Therefore, the creation of a new model which accurately predicts outcomes in patients undergoing multiple valve surgery is possibly required in China.

The Impact of Aneurysm Morphology and Anatomic Characteristics on Long-Term Survival after Endovascular Abdominal Aortic Aneurysm Repair

BACKGROUND

Hostile anatomic characteristics in patients undergoing endovascular abdominal aortic aneurysm repair (EVAR) and the placement of endografts not in concordance with the specific device anatomic guidelines (or instructions for use [IFU]) have shown decreased technical success of the procedure. But these factors have never been evaluated in regard to patient postoperative survival. We sought to assess the association between survival and (1) aneurysm anatomy and characteristics and (2) implantation in compliance with manufacturer's anatomic IFU guidelines in patients undergoing endovascular aortic aneurysm repair.

METHODS

The cohort included 273 consecutive patients who underwent EVAR at Baylor Heart and Vascular Hospital between January 1, 2002 and December 31, 2009 and had their preoperative computed tomography (CT) scan digitally retrievable. The CT scans and operative notes were then reviewed, and the anatomic severity grading (ASG) score, maximum aneurysm diameter, thrombus width, patency of aortic side branch vessels, and implantation in compliance with IFU guidelines were assessed. The unadjusted association between survival (assessed until November 1, 2011) and these variables was assessed with the Kaplan-Meier method. Moreover, propensity-adjusted (for a comprehensive array of clinical and nonclinical risk factors) proportional hazard models were developed to assess the adjusted associations.

RESULTS

Seven (2.56%) patients died within 30 days from EVAR, and 88 (30.04%) patients died during the study follow-up. Patient mean survival was 6.3 years. The unadjusted analysis showed a statistically significant association between survival and thrombus width (P = 0.007), ASG score (P = 0.004), and implantation in compliance with IFU guidelines (P = 0.007). However, the adjusted analysis revealed that none of the anatomic and compliance factors were significantly associated with long-term survival (ASG, P = 0.149; diameter, P = 0.836; thrombus, P = 0.639; patency, P = 0.219; and implantation compliance, P = 0.219).

CONCLUSIONS

Unfavorable aneurysm morphologic characteristics and endograft implantation not in compliance with IFU guidelines did not adversely affect patient survival after EVAR in this group of patients. This implies that unfavorable anatomy, even that which would necessitate implantation of the EVAR device outside of the IFU guidelines, should not necessarily contraindicate EVAR.

Enhancing the Value of Population-Based Risk Scores for Institutional-Level Use

BACKGROUND

We hypothesized that factors associated with an institution's residual risk unaccounted for by population-based models may be identifiable and used to enhance the value of population-based risk scores for quality improvement.

METHODS

From January 2000 to January 2010, 4,971 patients underwent aortic valve replacement (AVR), either isolated (n = 2,660) or with concomitant coronary artery bypass grafting (AVR+CABG; n = 2,311). Operative mortality and major morbidity and mortality predicted by The Society of Thoracic Surgeons (STS) risk models were compared with observed values. After adjusting for patients' STS score, additional and refined risk factors were sought to explain residual risk. Differences between STS model coefficients (risk-factor strength) and those specific to our institution were calculated.

RESULTS

Observed operative mortality was less than predicted for AVR (1.6% [42 of 2,660] vs 2.8%, p < 0.0001) and AVR+CABG (2.6% [59 of 2,311] vs 4.9%, p < 0.0001). Observed major morbidity and mortality was also lower than predicted for isolated AVR (14.6% [389 of 2,660] vs 17.5%, p < 0.0001) and AVR+CABG (20.0% [462 of 2,311] vs 25.8%, p < 0.0001). Shorter height, higher bilirubin, and lower albumin were identified as additional institution-specific risk factors, and body surface area, creatinine, glomerular filtration rate, blood urea nitrogen, and heart failure across all levels of functional class were identified as refined risk-factor variables associated with residual risk. In many instances, risk-factor strength differed substantially from that of STS models.

CONCLUSIONS

Scores derived from population-based models can be enhanced for institutional level use by adjusting for institution-specific additional and refined risk factors. Identifying these and measuring differences in institution-specific versus population-based risk-factor strength can identify areas to target for quality improvement initiatives.

The Benefits and Pitfalls of the Use of Risk Stratification Tools in Cardiac Surgery

Risk assessment tools are increasingly used in surgery. In cardiac surgery, risk models are used for patient counselling, surgical decision-making, performance benchmarking, clinical research, evaluation of new therapies and quality assurance, among others. However, they have numerous disadvantages which need to be considered. This article evaluates the utility of risk assessment tools in cardiac surgery including a discussion of their advantages and disadvantages.

On the practice of ignoring center-patient interactions in evaluating hospital performance

We evaluate the performance of medical centers based on a continuous or binary patient outcome (e.g., 30‐day mortality). Common practice adjusts for differences in patient mix through outcome regression models, which include patient‐specific baseline covariates (e.g., age and disease stage) besides center effects. Because a large number of centers may need to be evaluated, the typical model postulates that the effect of a center on outcome is constant over patient characteristics. This may be violated, for example, when some centers are specialized in children or geriatric patients. Including interactions between certain patient characteristics and the many fixed center effects in the model increases the risk for overfitting, however, and could imply a loss of power for detecting centers with deviating mortality. Therefore, we assess how the common practice of ignoring such interactions impacts the bias and precision of directly and indirectly standardized risks. The reassuring conclusion is that the common practice of working with the main effects of a center has minor impact on hospital evaluation, unless some centers actually perform substantially better on a specific group of patients and there is strong confounding through the corresponding patient characteristic. The bias is then driven by an interplay of the relative center size, the overlap between covariate distributions, and the magnitude of the interaction effect. Interestingly, the bias on indirectly standardized risks is smaller than on directly standardized risks. We illustrate our findings by simulation and in an analysis of 30‐day mortality on Riksstroke. © 2015 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.

Predicting the risk of death following coronary artery bypass graft made simple: a retrospective study using the American College of Surgeons National Surgical Quality Improvement Program database

INTRODUCTION

Risk models to predict 30-day mortality following isolated coronary artery bypass graft is an active area of research. Simple risk predictors are particularly important for cardiothoracic surgeons who are coming under increased scrutiny since these physicians typically care for higher risk patients and thus expect worse outcomes. The objective of this study was to develop a 30-day postoperative mortality risk model for patients undergoing CABG using the American College of Surgeons National Surgical Quality Improvement Program database.

MATERIAL AND METHODS

Data was extracted and analyzed from the American College of Surgeons National Surgical Quality Improvement Program Participant Use Files (2005-2010). Patients that had ischemic heart disease (ICD9 410-414) undergoing one to four vessel CABG (CPT 33533-33536) were selected. To select for acquired heart disease, only patients age 40 and older were included. Multivariate logistic regression analysis was used to create a risk model. The C-statistic and the Hosmer-Lemeshow goodness-of-fit test were used to evaluate the model. Bootstrap-validated C-statistic was calculated.

RESULTS

A total of 2254 cases met selection criteria. Forty-nine patients (2.2%) died within 30 days. Six independent risk factors predictive of short-term mortality were identified including age, preoperative sodium, preoperative blood urea nitrogen, previous percutaneous coronary intervention, dyspnea at rest, and history of prior myocardial infarction. The C-statistic for this model was 0.773 while the bootstrap-validated C-statistic was 0.750. The Hosmer-Lemeshow test had a p-value of 0.675, suggesting the model does not overfit the data.

CONCLUSIONS

The American College of Surgeons National Surgical Quality Improvement Program risk model has good discrimination for 30-day mortality following coronary artery bypass graft surgery. The model employs six independent variables, making it easy to use in the clinical setting.

Relation between topical application of platelet-rich plasma and vancomycin and severe deep sternal wound infections after a first median sternotomy

Deep sternal wound infections (DSWIs) are serious complications of sternotomy, leading to increased mortality and costs of care. Topical applications of autologous platelet concentrate and vancomycin have both shown promise in preventing DSWIs. From January 1, 1998, to November 30, 2010, 1,866 patients without previous sternotomy underwent cardiac surgery at the Baylor University Medical Center, Dallas, by a single surgeon who systematically adopted application of a paste containing vancomycin, calcium-thrombin, and platelet-rich plasma (PRP paste) to the edges of sternal wounds before closure in December 2005. A propensity-adjusted logistic regression model employing Firth's penalized maximum likelihood method was used to assess the association between the use of the PRP paste (intervention) and the incidence of severe DSWI. Eleven patients (0.59%) developed severe DSWIs. All were among the 1,318 patients in the control group (0.83%); no severe DSWIs developed in the 548 patients in the intervention group. Both the unadjusted and adjusted associations between the study intervention and DSWI were statistically significant (unadjusted p value=0.021; adjusted p value=0.005; adjusted odds ratio=0.05, 95% confidence interval 0.01, 0.50). In conclusion, the PRP paste appears to prevent severe DSWIs.

Prevalence of and risk factors of renal artery stenosis in patients with resistant hypertension

Renal artery stenosis (RAS) is a common cause of secondary hypertension. Renal artery angiography is the gold standard for diagnosing RAS. The aim of this study is to report (1) the prevalence of RAS in patients with resistant hypertension and (2) the association of RAS with peripheral vascular disease (PVD) and diabetes mellitus (DM). We studied 285 consecutive patients (mean age: 72.5 years) with resistant hypertension (systolic blood pressure >140 mm Hg despite administration of at least 3 antihypertensive drugs) who underwent renal artery angiography at Baylor Heart and Vascular Hospital from January 2006 to December 2010. Sixty-nine cases of RAS were identified (incidence: 24.2%). The propensity-adjusted analysis (controlling for clinical and nonclinical risk factors) showed a strong and significant association between RAS and PVD (odds ratio 5.15, 95% confidence interval 2.68 to 9.89, p <0.0001). However, the association between RAS and DM, a previously defined risk factor for RAS, was not significant in this cohort (odds ratio 0.63, 95% confidence interval 0.34 to 1.19, p = 0.16). In conclusion, results from this study define the prevalence of RAS in patients with resistant hypertension. Patients with PVD were found to be 5 times more likely to experience RAS than patients without PVD, whereas DM did not confer any increased risk.

Data Quality Improvement in Clinical Databases Using Statistical Quality Control: Review and Case Study

Ensuring the quality of data being collected in clinical and medical contexts is a concern for data managers and users. Quality assurance frameworks, systematic audits, and correction procedures have been proposed to enhance the accuracy and completeness of databases. Following an overview of the undertaken approaches, particularly statistical methods, the authors promote acceptance sampling plans (ASPs) and statistical process control (SPC) tools, including control charts and root cause analysis, as the technical core of the data quality improvement mechanism. They review ASP and SPC techniques and discuss their implementation in data quality evaluation and improvement. Two case studies are presented in which the authors apply some of the techniques to databases maintained by a local hospital. Finally, guidelines are proposed for which techniques are appropriate with regard to dataflow and database specifications.

The European Association for Cardio-Thoracic Surgery (EACTS) database: an introduction

OBJECTIVES

Continuous monitoring of surgical outcomes through benchmarking and the identification of best practices has become increasingly important. A structured approach to data collection, coupled with validation, analysis and reporting, is a powerful tool in these endeavours. However, inconsistencies in standards and practices have made comparisons within and between European countries cumbersome. The European Association for Cardio-Thoracic Surgery (EACTS) has established a large international database with the goals of (i) working with other organizations towards universal data collection and creating a European-wide repository of information on the practice of cardio-thoracic surgery, and (ii) disseminating that information in scientific, peer-reviewed articles. We report on the process of data collection, as well as on an overview of the data in the database.

METHODS

The EACTS Database Committee met for the first time in Monaco, September 2002, to establish the ground rules for the process of setting up the database. Subsequently, data have been collected and merged by Dendrite Clinical Systems Ltd.

RESULTS

As of December 2008, the database included 1,074,168 patient records from 366 hospitals located in 29 countries. The latest submission from the years 2006-08 included 404,721 records. The largest contributors were the UK (32.0%), Germany (20.9%) and Belgium (7.3%). Isolated coronary bypass surgery was the most frequently performed operation; the proportion of surgical workload that comprised isolated coronary artery bypass grafting varied from country to country: 30% in Spain and almost 70% in Denmark. Isolated valve procedures constituted 12% of all procedures in Norway and 32% in Spain. Baseline demographics showed an increase in the mean age and the percentage of patients that were female over time. Remarkably, the mortality rates for all procedures declined over the period analysed, to 2.2% (95% confidence interval [CI] 2.2-2.3%) for isolated coronary bypass, 3.4% (95% CI 3.3-3.5%) for isolated valve and 6.2% (95% CI 6.0-6.5%) for bypass + valve procedures.

CONCLUSION

The EACTS database has proven to be an important step forward in providing opportunities for monitoring cardiac surgical care across Europe. As the database continues to expand, it will facilitate research projects, establish benchmarking standards and identify potential areas for quality improvements.

Using administrative data for mortality risk adjustment in pediatric congenital cardiac surgery

OBJECTIVE

To evaluate the performance of risk-adjustment models from the University HealthSystem Consortium and the Agency for Healthcare Research Quality on an administrative dataset for children undergoing congenital cardiac surgery.

DESIGN

Retrospective cross-sectional cohort analysis.

SETTING

Multi-institutional database of administrative data provided by the University HealthSystem Consortium.

PATIENTS

Children whose discharge diagnosis had an associated cardiac surgical procedure.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

The performance of two risk-adjustment modeling schemata was measured in terms of discrimination and calibration, and receiver operating characteristic curves were compared. Model calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test. A total of 19,436 patients were included in the analysis with 816 deaths and an unadjusted overall mortality rate of 4.2%. The University HealthSystem Consortium models applied to the entire population resulted in an area under the curve = 0.73, and by comparison, the Agency for Healthcare Research Quality risk-adjustment model revealed area under the curve = 0.86. The risk-adjustment model of the University HealthSystem Consortium subgroup of Circulatory System Major Diagnostic Category 5 showed better performance with area under the curve = 0.81. Calibration using the Hosmer-Lemeshow test failed to show good agreement between the predicted and actual outcomes across the University HealthSystem Consortium mortality risk groups with an overall standardized mortality ratio of 1.2 (95% CI, 1.1-1.3; p < 0.0001) and poor predictive ability for the highest risk group, with a nearly 1.5-fold overprediction of death. The Agency for Healthcare Research Quality model shared similar calibration results with an overall standardized mortality ratio of 1.6 (95% CI, 1.5-1.7; p < 0.0001) and a nearly two-fold underprediction of death in the highest risk group.

CONCLUSIONS

Administrative data can be used to create risk-adjustment models in the congenital cardiac surgery population. Risk-adjustment models generated from administrative data may represent an attractive addition to clinically derived models in pediatric congenital cardiac surgery patients and should be considered for use either alone or in combination with clinical data in future analyses where mortality is a measure of performance and quality.

The society of thoracic surgeons national database

AIMS

The Society of Thoracic Surgeons (STS) National Database collects detailed clinical information on patients undergoing adult cardiac, paediatric and congenital cardiac, and general thoracic surgical operations. These data are used to support risk-adjusted, nationally benchmarked performance assessment and feedback; voluntary public reporting; quality improvement initiatives; guideline development; appropriateness determination; shared decision making; research using cross-sectional and longitudinal registry linkages; comparative effectiveness studies; government collaborations including postmarket surveillance; regulatory compliance and reimbursement strategies.

INTERVENTIONS

All database participants receive feedback reports which they may voluntarily share with their hospitals or payers, or publicly report. STS analyses are regularly used as the basis for local, regional and national quality improvement efforts.

POPULATION

More than 90% of adult cardiac programmes in the USA participate, as do the majority of paediatric cardiac programmes, and general thoracic participation continues to increase. Since the inception of the Database in 1989, more than 5 million patient records have been submitted.

BASELINE DATA

Each of the three subspecialty databases includes several hundred variables that characterise patient demographics, diagnosis, medical history, clinical risk factors and urgency of presentation, operative details and postoperative course including adverse outcomes.

DATA CAPTURE

Data are entered by trained data abstractors and by the care team, using detailed data specifications for each element.

DATA QUALITY

Quality and consistency checks assure accurate and complete data, missing data are rare, and audits are performed annually of selected participant sites.

ENDPOINTS

All major outcomes are reported including complications, status at discharge and mortality.

DATA ACCESS

Applications for STS Database participants to use aggregate national data for research are available at http://www.sts.org/quality-research-patient-safety/research/publications-and-research/access-data-sts-national-database.

Initial application in the EACTS and STS Congenital Heart Surgery Databases of an empirically derived methodology of complexity adjustment to evaluate surgical case mix and results

OBJECTIVES

Outcomes evaluation is enhanced by assignment of operative procedures to appropriate categories based upon relative average risk. Formal risk modelling is challenging when a large number of operation types exist, including relatively rare procedures. Complexity stratification provides an alternative methodology. We report the initial application in the Congenital Heart Surgery Databases of the Society of Thoracic Surgeons (STS) and the European Association for Cardio-thoracic Surgery (EACTS) of an empirically derived system of complexity adjustment to evaluate surgical case mix and results.

METHODS

Complexity stratification is a method of analysis in which the data are divided into relatively homogeneous groups (called strata). A complexity stratification tool named the STS-EACTS Congenital Heart Surgery Mortality Categories (STAT Mortality Categories) was previously developed based on the analysis of 77,294 operations entered in the Congenital Heart Surgery Databases of EACTS (33,360 operations) and STS (43,934 patients). Procedure-specific mortality rate estimates were calculated using a Bayesian model that adjusted for small denominators. Operations were sorted by increasing risk and grouped into five categories (the STAT Mortality Categories) that were designed to minimize within-category variation and maximize between-category variation. We report here the initial application of this methodology in the EACTS Congenital Heart Surgery Database (47,187 operations performed over 4 years: 2006-09) and the STS Congenital Heart Surgery Database (64,307 operations performed over 4 years: 2006-09).

RESULTS

In the STS Congenital Heart Surgery Database, operations classified as STAT Mortality Categories 1-5 were (1): 17332, (2): 20114, (3): 9494, (4): 14525 and (5): 2842. Discharge mortality was (1): 0.54%, (2): 1.6%, (3): 2.4%, (4): 7.5% and (5): 17.8%. In the EACTS Congenital Heart Surgery Database, operations classified as STAT Mortality Categories 1-5 were (1): 19874, (2): 12196, (3): 5614, (4): 8287 and (5): 1216. Discharge mortality was (1): 0.99%, (2): 2.9%, (3): 5.0%, (4): 10.3% and (5): 25.0%.

CONCLUSIONS

The STAT Mortality Categories facilitate analysis of outcomes across the wide spectrum of distinct congenital heart surgery operations including infrequently performed procedures.

Artificial neural networks versus proportional hazards Cox models to predict 45-year all-cause mortality in the Italian Rural Areas of the Seven Countries Study

BACKGROUND

Projection pursuit regression, multilayer feed-forward networks, multivariate adaptive regression splines and trees (including survival trees) have challenged classic multivariable models such as the multiple logistic function, the proportional hazards life table Cox model (Cox), the Poisson's model, and the Weibull's life table model to perform multivariable predictions. However, only artificial neural networks (NN) have become popular in medical applications.

RESULTS

We compared several Cox versus NN models in predicting 45-year all-cause mortality (45-ACM) by 18 risk factors selected a priori: age; father life status; mother life status; family history of cardiovascular diseases; job-related physical activity; cigarette smoking; body mass index (linear and quadratic terms); arm circumference; mean blood pressure; heart rate; forced expiratory volume; serum cholesterol; corneal arcus; diagnoses of cardiovascular diseases, cancer and diabetes; minor ECG abnormalities at rest. Two Italian rural cohorts of the Seven Countries Study, made up of men aged 40 to 59 years, enrolled and first examined in 1960 in Italy. Cox models were estimated by: a) forcing all factors; b) a forward-; and c) a backward-stepwise procedure. Observed cases of deaths and of survivors were computed in decile classes of estimated risk. Forced and stepwise NN were run and compared by C-statistics (ROC analysis) with the Cox models. Out of 1591 men, 1447 died. Model global accuracies were extremely high by all methods (ROCs > 0.810) but there was no clear-cut superiority of any model to predict 45-ACM. The highest ROCs (> 0.838) were observed by NN. There were inter-model variations to select predictive covariates: whereas all models concurred to define the role of 10 covariates (mainly cardiovascular risk factors), family history, heart rate and minor ECG abnormalities were not contributors by Cox models but were so by forced NN. Forced expiratory volume and arm circumference (two protectors), were not selected by stepwise NN but were so by the Cox models.

CONCLUSIONS

There were similar global accuracies of NN versus Cox models to predict 45-ACM. NN detected specific predictive covariates having a common thread with physical fitness as related to job physical activity such as arm circumference and forced expiratory volume. Future attention should be concentrated on why NN versus Cox models detect different predictors.

Preoperative hyponatremia predicts outcomes after cardiac surgery

OBJECTIVE

To study the effect of preoperative hyponatremia (Na <135 mEq/L) on outcomes after cardiac surgery.

METHODS

From 2002 to 2008, 4370 patients had cardiac surgery at our institution (CABG in 2238, valve in 597, CABG valve in 537, other in 998). The institution electronic medical records, STS database, and Social Security death index data were analyzed. The association of hyponatremia with mortality, hospital length of stay (LOS), and complications was analyzed using regression analysis.

RESULTS

Prevalence of hyponatremia was 21%. Patients with preoperative hyponatremia had lower left ventricular ejection fraction (39% ± 17% versus 46% ± 14%, P < 0.001) and glomerular filtration rate (69 ± 32 mg/min/1.73 m(2)versus 74 ± 27 mg/min/1.73 m(2), P < 0.001) and higher median EuroSCORE (19% versus 9%, P < 0.001), NYHA class 3-4 (77% versus 65%, P < 0.001), prevalence of chronic obstructive pulmonary disease (25% versus 18%, P < 0.001), and arteriopathy (20% versus 13%, P < 0.001). Hyponatremia was associated with increased early mortality (9% versus 4%, P < 0.001), late mortality (24% versus 16%, P < 0.001), and LOS (13 versus 8 d, P < 0.001). Mortality increased with the severity of hyponatremia. After adjusting for baseline and operative variables, hyponatremia was associated with increased hazard of mortality (hazard ratio [HR] 1.31, 95% confidence interval [CI] 1.14-1.52, P < 0.001), risk of early mortality (odds ratio [OR] 1.52, 95% CI 1.09-2.12, P < 0.001), late mortality (HR 1.37, 95% CI 1.16-1.62, P < 0.001), LOS (multiplier 1.26, 95% CI 1.15-1.39, P < 0.001), operative complications (OR 1.30, 95% CI 1.00-1.69, P = 0.051), and dialysis (OR 1.64, 95% CI 1.11-2.44, P = 0.013).

CONCLUSIONS

Preoperative hyponatremia is common, especially in high-risk patients. It is an independent risk factor for mortality, prolonged hospitalization, and complications after cardiac surgery.