Performance Of Risk Models Research Articles

A novel risk score for predicting left atrial and left atrial appendage thrombogenic milieu in patients with non-valvular atrial fibrillation

BackgroundAlthough the CHADS2 and CHA2DS2-VASc scoring systems are commonly used as measures of thromboembolic risk in patients with nonvalvular atrial fibrillation (NVAF), data are inconsistent as to their value in predicting the presence of left atrial (LA) and/or left atrial appendage (LAA) thrombogenic milieu (TM). The present study aimed to establish a novel risk score to assess the risk of LA and/or LAATM in NVAF patients. MethodsThis is a retrospective case-control study that included 125 consecutive patients with NVAF plus TM, as evidenced by transesophageal echocardiography (TEE) during a period from1 January 2010 to 1 February 2017. The controls were 1098 NVAF patients without TM during the same period. Risk factors for LA and/or LAATM were identified using univariable analysis and multivariable logistic regression. The risk score model was developed based on 10-fold validation and multiple regression. Risk model performance was evaluated using receiver operating characteristic (ROC) curves. Net reclassification improvement (NRI) was used for the comparison of C-statistics. The AUCs were compared using the Z test. ResultsAmong all 1223 NVAF patients, 125 (10.22%) patients had LA and/or LAATM. A score system (0−12) was developed based on the following 6 independent variables identified by 10-fold validation with sequential methods. Different points were assigned for each variable, according to multivariable regression using relative coefficients (coefficient of the index risk factor divided by the lowest coefficient among the 6 risk factors; rounded to closest integer): 1 for blood type A and N-terminal B-type natriuretic peptide (NT-proBNP) ≥864.85 pg/mL; 2 for LAD ≥43.5 mm and age ≥ 73.5 years old; 3 for previous heart failure and previous stroke or TIA. The present risk score system had a sensitivity of 58.3%, specificity of 91.4 and accuracy of 81.6%. The area under the ROC curve (AUC) was 0.832, (95% CI: 0.784–0.881; P < 0.001). The negative predictive value (NPV) was 92% when we set the cut-off point at 4; when the cut-off point was set at 8, the positive predictive value (PPV) was 85.7%. Compared with CHADS2 and CHA2DS2-VASc score, the present novel risk score has better predictive power [net reclassification improvement (NRI) +96.3% and +66.2%, respectively; all P < 0.001]. ConclusionThis study developed a novel risk score to help predicting LA and/or LAATM in NVAF patients, which had higher accuracy than CHADS2 and CHA2DS2-VASc score system.

Aggressive lymphoma subtype is a risk factor for venous thrombosis. Development of lymphoma - specific venous thrombosis prediction models.

Venous thromboembolic events (VTE) are a frequent complication of lymphoma. We conducted a retrospective analysis to compare VTE risk in diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL). Subjects were randomly assigned to training and validation sets to identify risk factors of VTE and evaluate risk model performance, including the Khorana score. A group of 790 patients were diagnosed from 2002 to 2014 (DLBCL = 542, FL = 248). Median follow- up was 49 months. We observed 106 VTE, with higher incidence in DLBCL (5-year cumulative incidence = 16.3% vs 3.8% in FL patients). Five-year OS for patients with VTE was 51.4% vs 73.1% in patients without VTE (P < .001). Baseline VTE risk factors identified in the training cohort included lymphoma subtype, previous VTE, ECOG performance status ≥2, decreased albumin, increased calcium, elevated WBC, absolute lymphocyte count or monocyte count, and presence of bulky disease. Addition of new variables to the Khorana score improved its performance measured by Akaike information criterion and Concordance index. A new risk model including lymphoma subtype, albumin, WBC count, and bulky disease was validated in time-based ROC analyses. These findings were confirmed in the validation cohort. Lymphoma subtypes have different VTE risk. The effect of lymphoma subtype was independent from disease burden and the use of systemic therapy. The Khorana risk-score was validated in time to event analyses, and a more robust lymphoma-specific VTE risk score is proposed. These findings suggest lymphoma patients with highest VTE risk can be identified with baseline parameters.

Prognostic performance of the Society of Thoracic Surgeons risk score in patients with left main coronary artery disease undergoing revascularisation: a post hoc analysis of the EXCEL trial.

Accurate risk prediction in patients undergoing revascularisation is essential. We aimed to assess the predictive performance of Society of Thoracic Surgeons (STS) risk models in patients with left main coronary artery disease (LMCAD) undergoing coronary artery bypass grafting (CABG) or percutaneous coronary intervention with everolimus-eluting stents (PCI-EES). The predictive performance of STS risk models for perioperative mortality, stroke and renal failure was evaluated for their discriminative ability (C statistic) and calibration (Hosmer-Lemeshow goodness-of-fit-test; χ2 and p-values) among patients with LMCAD undergoing PCI-EES (n=935) and CABG (n=923) from the randomised EXCEL trial. STS risk scores, in CABG patients, showed good discrimination for 30-day mortality and average discrimination for stroke (C statistic 0.730 and 0.629, respectively) with average calibration. For PCI, STS risk scores had no discrimination for mortality (C statistic 0.507), yet good discrimination (C statistic 0.751) and calibration for stroke. The predictive performance for renal failure was good for CABG (C statistic 0.82), yet poor for PCI (C statistic 0.59). In selected patients with LMCAD from the EXCEL trial, STS risk models showed good predictive performance for CABG yet lacked predictive performance for PCI for perioperative mortality and renal failure. The STS stroke risk model was surprisingly more discriminating in PCI compared to CABG. Improved and procedure-specific risk prediction instruments are needed to accurately estimate adverse events after LMCAD revascularisation by CABG and PCI. ClinicalTrials.gov Identifier: NCT01205776.

Performance of risk models predicting cardiac complications in pregnant women with congenital heart disease: a meta-analysis.

Accurate risk stratification is important in pregnant women including the growing population with congenital heart disease (CHD). We meta-analysed the performance of risk models for CHD women. Six studies with 3426 pregnancies were studied, with cardiac complication rates of 6.7-20.6%. Pooled c-statistics (95% confidence interval) for scores were mWHO 0.71 (0.65-0.76), CARPREG 0.66 (0.61-0.71) and ZAHARA 0.71 (0.65-0.76). Current risk models had at best moderate discrimination for CHD women with significant room for improvement.

Jump Driven Risk Model Performance in Cryptocurrency Market

This paper aims at identifying a validated risk model for the cryptocurrency market. We propose a stochastic volatility model with co-jumps in return and volatility (SVCJ) to highlight the role of jumps in returns and volatility in affecting Value-at-Risk (VaR) and Expected Shortfall (ES) in cryptocurrency market. Validation results based on backtesting show that SVCJ model is superior in terms of statistical accuracy of VaR and ES estimates, compared to alternative models such as TGARCH (Threshold GARCH) volatility and RiskMetrics models. The results imply that for the cryptocurrency market, the best performing model is a stochastic process that accounts for both jumps in returns and volatility.

Metabolic Syndrome Risk Models and Scores: A Systematic Review

Globally the prevalence of Metabolic Syndrome (MetS) is high and may reach a frightening level in the nearest future given the current trends of rising prevalence of hypertension and obesity. The presence of metabolic syndrome is linked with increased risk of cardiovascular disease, diabetes and all-cause mortality. Despite the recent proliferation of models and scores for assessing the risk of developing MetS, there is hardly any used in practical setting. Hence, we conducted a systematic review to determine the performance of risk models and scores for predicting metabolic syndrome.

The Effect of Outcome Selection on the Performance of Prediction Models in Patients at Risk for Sepsis.

Risk prediction models for patients with suspected sepsis have been derived on and applied to various outcomes, including readily available outcomes such as hospital mortality and ICU admission as well as longer-term mortality outcomes that may be more important to patients. It is unknown how selecting different outcomes influences model performance in patients at risk for sepsis. Evaluate the impact of outcome selection on risk model performance and weighting of individual predictor variables. We retrospectively analyzed adults hospitalized with suspected infection from January 2014 to September 2017 at 12 hospitals. We used routinely collected clinical data to derive logistic regression models for four outcomes: hospital mortality, composite ICU length of stay greater than 72 hours or hospital mortality, 30-day mortality, and 90-day mortality. We compared the performance of the models using area under the receiver operating characteristic curve and calibration plots. Among 52,184 admissions, 2,030 (4%) experienced hospital mortality, 6,659 (13%) experienced the composite of hospital mortality or ICU length of stay greater than 72 hours, 3,417 (7%) experienced 30-day mortality, and 5,655 (11%) experienced 90-day mortality. Area under the receiver operating characteristic curves decreased when hospital-based models were applied to predict 30-day (hospital mortality = 0.88-0.85; -0.03, composite ICU length of stay greater than 72 hours or hospital mortality = 0.90-0.81; -0.09) and 90-day mortality (hospital mortality = 0.88-0.81; -0.07, composite ICU length of stay greater than 72 hours or hospital mortality = 0.90-0.76; -0.14; all p < 0.01). Models were well calibrated for derived (root-mean-square error = 5-15) but not alternate outcomes (root-mean-square error = 8-35). Risk models trained to predict readily available hospital-based outcomes in suspected sepsis show poorer discrimination and calibration when applied to 30- and 90-day mortality. Interpretation and application of risk models for patients at risk of sepsis should consider these findings.

Application of 17 Contrast-Induced Acute Kidney Injury Risk Prediction Models

Introduction: Contrast-induced acute kidney injury (CI-AKI) is a frequent complication of percutaneous coronary interventions (PCI). Various groups have developed and validated risk scores for CI-AKI. Although the majority of these risk scores achieve an adequate accuracy, their usability in clinical practice is limited and greatly debated. Objective: With the present study, we aimed to prospectively assess the diagnostic performance of recently published CI-AKI risk scores (up to 2018) in a cohort of patients undergoing PCI. Methods: We enrolled 1,247 consecutive patients (80% men, mean age 62 ± 10 years) treated with elective or urgent PCI. For each patient, we calculated the individual CI-AKI risk score based on 17 different risk models. CI-AKI was defined as an increase of ≥25% (liberal) or ≥0.5 mg/dL (strict) in pre-PCI serum creatinine 48 h after PCI. Results: CI-AKI definition and, therefore, CI-AKI incidence have a significant impact on risk model performance (median negative predictive value increased from 85 to 99%; median c-statistic increased from 0.516 to 0.603 using more strict definition criteria). All of the 17 published models were characterized by a weak-to-moderate discriminating ability mainly based on the identification of “true-negative” cases (median positive predictive value 19% with liberal criterion and 3% with strict criterion). In none of the models, c-statistic was >0.800 with either CI-AKI definition. Novel, different combinations of the >35 independent variables used in the published models either by down- or by up-scaling did not result in significant improvement in predictive performance. Conclusions: The predictive ability of all models was similar and only modest, derived mainly by identifying true-negative cases. A new approach is probably needed by adding novel markers or periprocedural characteristics.

Deploying Machine Learning to Validate Clinical Phenotypes and Factors Associated with Mortality Risk in 2,022 Critically Ill Patients with COVID-19 in Spain

Background: The identification of factors associated with Intensive Care Unit (ICU) mortality and derived clinical phenotypes in COVID-19 patients could help for a more tailored approach to clinical decision-making that improves prognostic outcomes. The objective was to analyze patient’s factors associated with mortality risk and utilize a Machine Learning(ML) to derive clinical COVID-19 phenotypes.Methods: Prospective, multicenter, observational study of critically ill patients with confirmed COVID-19 disease and acute respiratory failure admitted from 63 Intensive Care Units(ICU) in Spain. Patient features including demographics and clinical data at ICU admission were analyzed. Generalized linear models were used to determine ICU morality risk factors. An unsupervised clustering analysis was applied to determine presence of phenotypes. The prognostic models were validated and their performance was measured using accuracy test, sensitivity, specificity and ROC curves.Findings: The database included a total of 2,022 patients (mean age 64[IQR5-71] years, 1423(70·4%) male, median APACHE II score (13[IQR10-17]) and SOFA score (5[IQR3-7]) points. The ICU mortality rate was 32·6%. Of the 3 derived phenotypes, the C(severe) phenotype was the most common (857;42·5%) and was characterized by the interplay of older age (>65 years), high severity of illness and a higher likelihood of development shock. The A(mild) phenotype (537;26·7%) included older age (>65 years), fewer abnormal laboratory values and less development of complications and B (moderate) phenotype (623,30·8%) had similar characteristics of A phenotype but were more likely to present shock. Crude ICU mortality was 45·4%, 25·0% and 20·3% for the C, B and A phenotype respectively. The ICU mortality risk factors and model performance differed between whole population and phenotype classifications.Interpretation: The presented ML model identified three clinical phenotypes that significantly correlated with host-response patterns and ICU mortality. Different risk factors across the whole population and clinical phenotypes were observed which may limit the application of a “one-size-fits-all” model in practice.Funding Statement: This study was supported by the Spanish Intensive Care Society(SEMICYUC) and Ricardo Barri Casanovas Foundation.Declaration of Interests: All authors declare that they have no conflicts of interest.Ethics Approval Statement: The study was approved by the reference institutional review board at Joan XXIII University Hospital (IRB# CEIM/066/2020) and each participating site with a waiver of informed consent. All data values were anonymized prior to the phenotyping which consisted of clustering clinical variables on their association with COVID-19 mortality.

A Breakthrough Idea in Risk Measure Validation – Is the Way Paved for an Effective Expected Shortfall Backtest?

This research note is a kind of “call for attention” to recent developments in backtesting financial risk measures. This topic is relevant in relation to the regulatory monitoring of the performance of internal risk models used by banks in determining the minimum capital requirements for trading book portfolios. Backtesting is a process for checking the validity of risk estimation models. In his seminal work, Gneiting (2011) has proven that a prominent risk measure, Expected Shortfall (ES), lacks a property called elicitability. This finding has triggered a huge controversy on the issue of whether ES is backtestable at all. Due to the significant contribution of Acerbi and Szekely (2017, 2019) among others, the above-mentioned debate can be adequately and convincingly closed because there is a (re)solution. In particular, one can arrive at the conclusion that, building on its joint elicitability with Value-at- Risk (VaR), it is possible to introduce a so-called ridge backtest for ES. In fact, there is still an open question as to when and how the regulatory authorities will (re)act.

Mammographic Breast Density Assessed with Fully Automated Method and its Risk for Breast Cancer.

Objectives:We evaluated the association between breast cancer and breast density (BD) measured using fully automated software. We also evaluated the performance of cancer risk models such as only clinical risk factors, density related measures, and both clinical risk factors and density-related measures for determining cancer risk.Materials and Methods:This is a retrospective case–control study. The data were collected from August 2015 to December 2018. Two hundred fifty women with breast cancer and 400 control subjects were included in this study. We evaluated the BD qualitatively using breast imaging-reporting and data system density and quantitatively using 3D slicer. We also collected clinical factors such as age, familial history of breast cancer, menopausal status, number of births, body mass index, and hormonal replacement therapy use. We calculated the odds ratio (OR) for BD to determine the risk of breast cancer. We performed receiver operating characteristic (ROC) curve to assess the performance of cancer risk models.Results:The OR for the percentage BD for second, third, and fourth quartiles was 1.632 (95% confidence intervals [CI]: 1.102–2.416), 2.756 (95% CI: 1.704–4.458), and 3.163 (95% CI: 1.356–5.61). The area under ROC curve for clinical risk factors only, mammographic density measures, combined mammographic, and clinical risk factors was 0.578 (95% CI: 0.45, 0.64), 0.684 (95% CI: 0.58, 0.75), and 0.724 (95% CI: 0.64, 0.80), respectively.Conclusion:Mammographic BD was found to be positively associated with breast cancer. The density related measures combined clinical risk factors, and density model had good discriminatory power in identifying the cancer risk.

Considerations When Using Breast Cancer Risk Models for Women with Negative BRCA1/BRCA2 Mutation Results.

The performance of breast cancer risk models for women with a family history but negative BRCA1 and/or BRCA2 mutation test results is uncertain. We calculated the cumulative 10-year invasive breast cancer risk at cohort entry for 14 657 unaffected women (96.1% had an affected relative) not known to carry BRCA1 or BRCA2 mutations at baseline using three pedigree-based models (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm, BRCAPRO, and International Breast Cancer Intervention Study). During follow-up, 482 women were diagnosed with invasive breast cancer. Mutation testing was conducted independent of incident cancers. All models underpredicted risk by 26.3%-56.7% for women who tested negative but whose relatives had not been tested (n = 1363; 63 breast cancers). Although replication studies with larger sample sizes are needed, until these models are recalibrated for women who test negative and have no relatives tested, caution should be used when considering changing the breast cancer risk management intensity of such women based on risk estimates from these models.

P1853Comparative evaluation of risk model performance for prediction of 30-day mortality in transcatheter aortic valve replacement

Abstract Background and purpose EuroSCORE (ES) and Society of Thoracic Surgeons (STS) risk prediction models are routinely used to guide decision-making for transcatheter aortic valve replacement (TAVR), however their accuracy remains limited, especially in very old and high-risk patients. New and updated scoring models have thus been developed to improve risk stratification. We performed a comparative evaluation of classical and new risk scoring models for prediction of 30d mortality in transcatheter aortic valve interventions. Methods and results A total of 1,569 patients undergoing transfemoral (TF, n=1.235) or transapical (TA, n=334) TAVR from 2009 to 2018 were included in a single-center all-comer analysis. Six risk scoring models (logES_I, ES_II, STS-PROM, FRANCE-2, OBSERVANT, GAV-2) were calculated for all patients and evaluated for prediction of 30d mortality in their model discrimination (c-indices with 95% confidence intervals (CI)) and calibration (graphical evaluation). Mean classical risk scores confirmed an intermediate-to-high-risk patient collective (logES_I 27.0±16.9%; STS-PROM 7.0±6.4%), mean 30-day mortality was 3.4% (TF 2.3%; TA 7.8%). Overall discrimination performance was best in FRANCE-2 (c-index 0.73, 95% CI 0.67–0.80), followed by STS-PROM (c-index 0.68, 95% CI 0.62–0.75), OBSERVANT (c-index 0.68, 95% CI 0.61–0.76), ES_II (c-index 0.64) and logES_I and GAV-2 (both c-indices 0.63). FRANCE-2 discriminated best in TF TAVR (c-index 0.72; range of c-indices 0.63 to 0.72), while OBSERVANT performed best in TA TAVR (c-index 0.70; range of c-indices 0.61 to 0.70). All risk scoring models – with the exception of lowest-risk deciles of STS-PROM and ES_II – showed an overestimation of mortality probability in all risk strata. Conclusion FRANCE-2 and OBSERVANT risk models showed superior discrimination performance to classical risk scoring models in TF and TA TAVR, however all models tended to overestimate mortality probability.

Prediction of acute care service utilization for patients with hematologic malignancies.

312 Background: Improving the ability to predict which patients are at increased risk for acute care services subsequent to hospitalization can lead to more effective interventions and quality care. This study evaluated the performance of risk models to predict 30-day acute service utilization among adult inpatients with newly diagnosed hematologic malignancies. The study explored the impact of medical complications on potentially preventable service utilization, aligned with specifications for proposed measure OP-35 from the CMS Hospital Outpatient Quality Reporting Program. Methods: The study included 933 unique adult patients admitted to four acute care inpatient facilities within a non-profit community based healthcare system during calendar years 2012 to 2017. Risk models integrated a comprehensive set of data elements using clinical information from electronic medical records and facility oncology registries. Predictive models were constructed using a multivariable logistic regression, with calculation of standardized coefficients to rank order variables with the greatest impact on unplanned readmissions. Exploratory data mining techniques were conducted to augment classification of at-risk patients. Results: The mean age of study participants was 65 years and 55.1% were male. In order of prevalence, the types of hematologic malignancy were lymphoma (48.7%), leukemia (35.2%), myeloma (14.0%), and mixed types (2.1%). Approximately one fifth of patients received unplanned acute care services within 30 days after the index hospitalization (lymphoma = 23.1%, myeloma = 22.1%, leukemia = 18.6%). Among the risk models, the most influential predictors included initial length of stay, severity of illness, and consult for social work. Overall, 56.6% of patients who were readmitted within 30 days of hospitalization presented with one or more symptoms associated with the CMS medical complication measure, including cytopenia, infection, negative hydration, and/or pain. Conclusions: Optimizing quality care in a value-based health care environment requires proactive approaches such as risk prediction models for identifying patients who can benefit most from care transition interventions.

Risk models and scores for metabolic syndrome: systematic review protocol

IntroductionMetabolic syndrome ‘a clustering of risk factors which includes hypertension central obesity, impaired glucose metabolism with insulin resistance and dyslipidaemia’ affects approximately 20%–25% of the global adult population. Individuals with...

Can Classification and Regression Tree Analysis Help Identify Clinically Meaningful Risk Groups for Hip Fracture Prediction in Older American Men (The MrOS Cohort Study)?

ABSTRACTAlthough the WHO fracture risk algorithm (FRAX) is used to predict fracture, the utility of some simple machine‐learning methods, such as classification and regression trees (CARTs) should be evaluated to determine their efficacy in fracture prediction. Follow‐up time for the hip fracture analyses of 5977 community‐dwelling American men aged ≥65 years old was truncated to 10 years. There were 172 (2.9%) men who had an incident nontraumatic hip fracture. The CARTs were developed using hip BMD and common clinical risk factors as follows: model 1 = using classification with continuous variables of age, total hip BMD, and femoral neck BMD, or together with common clinical risk factors; and model 2 = using classification with continuous variables of age, total hip BMD, femoral neck BMD, FRAX score, osteoporosis by T‐score at the hip, and common clinical risk factors. The predictive performance of risk models derived from CARTs was compared with the basic classification of FRAX at 3% (basic model). From model 1, discriminators selected by CART were total hip BMD, age, and femoral neck BMD; no other clinical risk factors were selected. From model 2, discriminators selected by CART were FRAX score, femoral neck BMD, and age. Compared with the basic model using only a high‐risk group by FRAX ≥3%, no significantly improved predictive performance was demonstrated by model 1 or model 2 as identified by CART with the area under the receiver‐operating characteristic curve for each model of 0.714 (95% CI, 0.676 to 0.751) or 0.726 (95% CI, 0.690 to 0.762) versus 0.703 (95% CI, 0.667 to 0.740), respectively. The improved overall net reclassification improvement index was 0.02 (95% CI, –0.04 to 0.08) and 0.05 (95% CI, –0.01 to 0.10), respectively. Although a FRAX category is a good clinical indicator for hip fracture risk, a simple classification by age and BMD may provide an alternative way to estimate a clinical risk level of 3.0%. © 2019 The Authors. JBMR Plus is published by Wiley Periodicals, Inc. on behalf of the American Society for Bone and Mineral Research.

Short-term forecasting of bark beetle outbreaks on two economically important conifer tree species

In recent years bark beetles have been shown to be an important risk factor in European forests. An early warning system is needed to mitigate bark beetle damage, and short-term forecasting models that assist efforts to identify attacked trees comprise an important part of such a system. The aim of this study was to develop short-term forecasting models of the probability of bark beetle outbreaks on two important conifer tree species: Norway spruce (Picea abies) and silver fir (Abies alba). For the development of the models, we used a time series of 20 years of sanitary felling because of bark beetles and relief data (altitude, slope and exposition), several soil variables, climate data (temperature and SPI), sanitary felling because of bark beetles, sanitary felling due to harmful abiotic factors, and amount of weakened trees due to bark beetles. The forecasting variable was sanitary felling because of bark beetles in the current year. The models were developed with a general linear model with binomial error distribution. For the probability of bark beetle outbreaks on silver fir, the amount of fir, soil base saturation percentage, sanitary felling of attacked fir, weakened fir, and sanitary felling because of abiotic factors increased the probability of sanitary felling because of fir bark beetles. Altitude, exposition, slope, phosphorus, soil depth, soil cation exchange capacity, SPI and temperature decreased the probability of sanitary felling because of fir bark beetles. For Norway spruce, the amount of Norway spruce, soil base saturation percentage, SPI, temperature, amount of sanitary felling in the previous year, amount of weakened trees in the previous year, and amount of sanitary felling because of abiotic factors in the previous year increased the probability of sanitary felling of Norway spruce because of bark beetles in the current year. Slope, soil cation exchange capacity, and precipitation decreased the probability of sanitary felling because of bark beetles in the current year. The performance of the bark beetle risk model for Norway spruce was very good. The performance of the model for silver fir was also good, but not on par with that for Norway spruce. Therefore, additional research on fir bark beetles is needed to further improve the risk model for bark beetle attacks on silver fir.

A clinical model to predict the risk of synchronous bone metastasis in newly diagnosed colorectal cancer: a population-based study

BackgroundThe early detection of synchronous bone metastasis (BM) in newly diagnosed colorectal cancer (CRC) affects its initial management and prognosis. A clinical model to individually predict the risk of developing BM would be attractive in current clinical practice.MethodsA total of 55,869 CRC patients were identified from Surveillance, Epidemiology, and End Results (SEER) database, of whom 317 patients were diagnosed with synchronous BM. Risk factors for BM in CRC patients was identified using multivariable logistic regression. A weighted scoring system was built with beta-coefficients (P < 0.05). A random sample of 75% of the CRC patients was used to establish the risk model, and the remaining 25% was used to validate its accuracy of this model. The performance of risk model was estimated by receiver operating curve (ROC) analysis.ResultsThe risk model consisted of 8 risk factors including rectal cancer, poorly-undifferentiation, signet-ring cell carcinoma, CEA positive, lymph node metastasis, brain metastasis, liver metastasis and lung metastasis. The areas under the receiver operating curve (AUROC) were 0.903 and 0.889 in the development and validation cohort. Patients with scores from 0 to 4 points had about 0.1% risk of developing BM, and the risk increased to about 30% in patients with scores ≥15 points.ConclusionsThis clinical risk model is accurate enough to identify the CRC patients with high risk of synchronous BM and to further provide more individualized clinical decision.

Validation of National Cardiovascular Data Registry risk models for mortality, bleeding and acute kidney injury in interventional cardiology at a German Heart Center.

The National Cardiovascular Data Registry (NCDR) risk scores for mortality, bleeding and acute kidney injury (AKI) are accurate outcome predictors of coronary catheterization procedures in North American populations. However, their application in German clinical practice remained elusive and we thus aimed to verify their use. NCDR scores for mortality, bleeding and AKI and corresponding clinical outcomes were retrospectively assessed in patients undergoing catheterization for ST-segment elevation myocardial infarction (STEMI), non-ST-segment elevation myocardial infarction (NSTEMI) or for elective coronary procedures at a German Heart Center from 2014 to 2017. Risk model performance was assessed using receiver-operating-characteristic curves (discrimination) and graphical analysis/logistic regression (calibration). A total of 1637 patients were included, procedures were performed for STEMI (565 patients, 34.5%), NSTEMI (572 patients, 34.9%) and elective purposes (500 patients, 30.5%); 6% (13% of STEMI and 5% of NSTEMI patients) presented in cardiogenic shock and 3% with resuscitated cardiac arrest. Radial access was used in 38% of procedures and cross-over was necessary in 5%; PCI was performed in 60% of procedures. In-hospital mortality was 6.3% (STEMI 14.5%; NSTEMI 3.7%; elective 0%) and major bleedings occurred in 5.6% (STEMI 10.6%; NSTEMI 5.4%; elective 0.2%); AKI was detected in 18.1% of patients (STEMI 23.7%; NSTEMI 27.3%; elective 1.4%), amounting to KDIGO stage I/II/III in 11.5%/3.5%/3.2%. NCDR risk models discriminated very well for mortality [AUC 0.93 with 95% confidence interval (CI) 0.91-0.95] and well for major bleeding (AUC 0.82, CI 0.78-0.86) and any AKI (AUC 0.83, CI 0.81-0.86). Discrimination in the subgroup of patients with PCI was comparable (mortality: AUC 0.90; major bleeding: AUC 0.78; any AKI: AUC 0.79). However, calibration showed considerable underestimation of mortality and AKI in high-risk patients, while major bleeding was consistently overestimated (Hosmer-Lemeshow p < 0.02 for all outcomes). The NCDR risk models showed excellent performance in discriminating high-risk from low-risk patients in contemporary German interventional cardiology. Model calibration for adverse event probability prediction, however, is limited and demands recalibration, especially in high-risk patients.

Performance of Prognostic HeartFailure Models in Patients With Nonischemic Cardiomyopathy Undergoing VentricularTachycardia Ablation.

This study sought to assess the performance of established risk models in predicting outcomes after catheter ablation (CA) in patients with nonischemic dilated cardiomyopathy (NIDCM) and ventricular tachycardia (VT). A correct pre-procedural risk stratification of patients with NIDCM and VT undergoing CA is crucial. The performance of different pre-procedural risk stratification approaches to predict outcomes of CA of VT in patients with NIDCM is unknown. The study compared the performance of 8 prognostic scores (SHFM [Seattle HeartFailure Model], MAGGIC [Meta-analysis Global Group in Chronic HeartFailure], ADHERE [Acute Decompensated HeartFailure National Registry], EFFECT [Enhanced Feedback for Effective Cardiac Treatment-Heart Failure], OPTIMIZE-HF [Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with HeartFailure], CHARM [Candesartan in Heart Failure-Assessment of Reduction in Mortality], EuroSCORE [European System for Cardiac Operative Risk Evaluation], and PAINESD [Chronic Obstructive Pulmonary Disease, Age > 60 Years, Ischemic Cardiomyopathy, New York Heart Association Functional Class III or IV, Ejection Fraction<25%, Presentation With VT Storm, Diabetes Mellitus]) for the endpoints of death/cardiac transplantation and VT recurrence in 282 consecutive patients (age 59 ± 15 years, left ventricular ejection fraction: 36 ± 13%) with NIDCM undergoing CA of VT. Discrimination and calibration of each model were evaluated through area under the curve (AUC) of receiver-operating characteristic curve and goodness-of-fit test. After a median follow-up of 48 (interquartile range: 19-67) months, 43 patients (15%) died, 24 (9%) underwent heart transplantation, and 58 (21%) experienced VT recurrence. The prognostic accuracy of SHFM (AUC= 0.89; goodness-of-fit p= 0.68 for death/transplant and AUC= 0.77; goodness-of-fit p= 0.16 for VT recurrence) and PAINESD (AUC= 0.83; goodness-of-fit p= 0.24 for death/transplant and AUC= 0.68; goodness-of-fit p= 0.58 for VT recurrence) were significantly superior to that of other scores. In patients with NIDCM and VT undergoing CA, the SHFM and PAINESD risk scores are powerful predictors of recurrent VT and death/transplant during follow-up, with similar performance and significantly superior to other scores. A pre-procedural calculation of the SHFM and PAINESD can be useful to predict outcomes.