Poor Calibration Research Articles

Predictive Value of the Get With The Guidelines Heart Failure Risk Score in Unselected Cardiac Intensive Care Unit Patients.

BackgroundThe cardiac intensive care unit (CICU) population is no longer composed of only patients with acute coronary syndromes, and includes those with acute heart failure and multiple comorbidities. We hypothesized that the GWTG‐HF (Get With The Guidelines–Heart Failure) risk score that predicts inpatient mortality in hospitalized patients with heart failure would predict mortality in CICU patients.Methods and ResultsWe retrospectively analyzed CICU patients at a tertiary care hospital from 2007 to 2015. The GWTG‐HF risk score was calculated at CICU admission. As a secondary analysis, the EFFECT (Enhanced Feedback for Effective Cardiac Treatment), OPTIMIZE‐HF (Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients With Heart Failure), and ADHERE (Acute Decompensated Heart Failure National Registry) risk scores were calculated. Kaplan–Meier survival analysis and the area under the receiver operating characteristic curve value were determined for inpatient and 1‐year mortality. The GWTG‐HF risk score was calculated in 9532 (95%) patients, with a median value of 40 (interquartile range, 35–47). Inpatient mortality occurred in 824 (8.6%) patients, and 2075 (21.8%) patients died by 1 year. Patients who died in hospital had a significantly higher mean GWTG‐HF score (47.7 versus 40.2; P<0.001). Inpatient and 1‐year mortality increased in each GWTG‐HF risk score quartile (P<0.0001). Discrimination of the GWTG‐HF, EFFECT, OPTIMIZE‐HF, and ADHERE risk scores was assessed using area under the receiver operating characteristic curve values for hospital mortality, and were similar for all risk scores (0.72–0.74; P>0.05). The Hosmer–Lemeshow statistic suggested poor calibration for hospital mortality by the GWTG‐HF risk score (P<0.001).ConclusionsThe GWTG‐HF risk score and other heart failure prediction tools demonstrate good discrimination for inpatient and 1‐year mortality in a heterogeneous cohort of CICU patients. Our study emphasizes that prognostic variables overlap in cardiac patients, regardless of the admission diagnosis.

How Accurate Are the Surgical Risk Preoperative Assessment System (SURPAS) Universal Calculators in Total Joint Arthroplasty?

Surgical outcome prediction models are useful for many purposes, including informed consent, shared decision making, preoperative mitigation of modifiable risk, and risk-adjusted quality measures. The recently reported Surgical Risk Preoperative Assessment System (SURPAS) universal risk calculators were developed using 2005-2012 American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP), and they demonstrated excellent overall and specialty-specific performance. However, surgeons must assess whether universal calculators are accurate for the small subset of procedures they perform. To our knowledge, SURPAS has not been tested in a subset of patients undergoing lower-extremity total joint arthroplasty (TJA). How accurate are SURPAS models' predictions for patients undergoing TJA? We identified an internal subset of patients undergoing non-emergency THA or TKA from the 2012 ACS-NSQIP, the most recent year of the SURPAS development dataset. To assess the accuracy of SURPAS prediction models, 30-day postoperative outcomes were defined as in the original SURPAS study: mortality, overall morbidity, and six complication clusters-pulmonary, infectious, cardiac or transfusion, renal, venous thromboembolic, and neurologic. We calculated predicted outcome probabilities by applying coefficients from the published SURPAS logistic regression models to the TJA cohort. Discrimination was assessed with C-indexes, and calibration was assessed with Hosmer-Lemeshow 10-group chi-square tests and decile plots. The 30-day postoperative mortality rate for TJA was 0.1%, substantially lower than the 1% mortality rate in the SURPAS development dataset. The most common postoperative complications for TJA were intraoperative or postoperative transfusion (16%), urinary tract infection (5%), and vein thrombosis (3%). The C-indexes for joint arthroplasty ranged from 0.56 for venous thromboembolism (95% CI 0.53 to 0.59 versus SURPAS C-index 0.78) to 0.82 for mortality (95% CI 0.76 to 0.88 versus SURPAS C-index 0.94). All joint arthroplasty C-index estimates, including CIs, were lower than those reported in the original SURPAS development study. Decile plots and Hosmer-Lemeshow tests indicated poor calibration. Observed mortality rates were lower than expected for patients in all risk deciles (lowest decile: no observed deaths, 0.0% versus expected 0.1%; highest decile: observed mortality 0.7% versus expected 2%; p < 0.001). Conversely, observed morbidity rates were higher than expected across all risk deciles (lowest decile: observed 12% versus expected 8%; highest decile: observed morbidity 32% versus expected 25%; p < 0.001) CONCLUSIONS: The universal SURPAS risk models have lower accuracy for TJA procedures than they do for the wider range of procedures in which the SURPAS models were originally developed. These results suggest that SURPAS model estimates must be evaluated for individual surgical procedures or within restricted groups of related procedures such as joint arthroplasty. Given substantial variation in patient populations and outcomes across numerous surgical procedures, universal perioperative risk calculators may not produce accurate and reliable results for specific procedures. Surgeons and healthcare administrators should use risk calculators developed and validated for specific procedures most relevant to each decision. Continued work is needed to assess the accuracy of universal risk calculators in more narrow procedural categories based on similarity of outcome event rates and prevalence of predictive variables across procedures.

Carbonate ecosystem and environmental evolution during the Valanginian–Barremian: Discussion on possible controlling factors in the Apennine Carbonate Platform record (Italy)

Shallow-water carbonate records are often characterised by low stratigraphic resolution and poor chronostratigraphic calibration despite abundant and diversified fossil content. We present here a stratigraphic study of a Valanginian–Barremian carbonate platform section, exposed in the southern Apennines (Italy). In the San Lorenzello section, we produced a high-resolution dataset integrating sedimentology, biostratigraphy, cyclostratigraphy and sequence stratigraphy to investigate how neritic biofacies responded to palaeonvironmental and palaeoclimatic changes in the time interval after the Weissert event. Fossil assemblages and lithofacies characteristics allowed the reconstruction of the paleoenvironmental history of the section, which is part of a shallow carbonate platform influenced by waves activity, storm events and sea-level changes. The latter induced cyclic emersions of this flat-topped subtropical platform, witnessed by karstification and pedogenesis. Investigations lead to a refinement of the previous biostratigraphy, also producing a close agreement between the biostratigraphic and C-isotope position of the Valanginian–Hauterivian boundary. Numerous green algae and benthic foraminifera bioevents have been identified and six biozones have been defined and anchored to the southern Tethys biozonal schemes.We used biostratigraphy and previous C-isotope stratigraphy to correlate local to global stratigraphic events and to discuss controlling factors influenced this carbonate platform system. An eustatic control on the hierarchical organization of the superbundles, also interpreted in terms of depositional sequences, is evidenced. Most Trangressive/Regressive Facies Trends (T/RFTs) boundaries have been correlated with the “global” Sequence Boundaries and to the Cretaceous eustatic curve. A correspondence between calcareous algae and benthic foraminiferal turnover phases and long-term eustatic events is also shown.

Wild goose chase – no predictable patient subgroups benefit from meniscal surgery: patient-reported outcomes of 641 patients 1 year after surgery

BackgroundDespite absence of evidence of a clinical benefit of arthroscopic partial meniscectomy (APM), many surgeons claim that subgroups of patients benefit from APM.ObjectiveWe developed a prognostic model predicting change in...

Validation of a Risk Equation Predicting Hemodialysis Arteriovenous Fistula Primary Failure in Elderly

Background: Choice of vascular access for older hemodialysis patients presents a special challenge since the rate of arteriovenous fistula (AVF) primary failure is high. The Lok’s risk equation predicting AVF primary failure has achieved good prediction accuracy and holds great potential for clinical use, but it has not been validated in the United States older hemodialysis patients. Methods: We assembled a validation data set of 14,892 patients aged 67 years and older who initiated hemodialysis with a central venous catheter between July 1, 2010, and June 30, 2012, and had a subsequent, incident AVF placement from the United States Renal Data System. We examined the external validity of Lok’s model by applying it to this validation data set. The discriminatory accuracy and calibration were evaluated by the concordance index (C-statistics) and calibration plot, respectively. Results: The observed frequency of AVF primary failure varied from 0.45 to 0.53 in hemodialysis patients in the validation data set. The predicted probabilities of AVF primary failure calculated by using the Lok’s risk equation ranged from 0.08 to 0.61, and 77.8, 40.5, and 51.7% of patients were categorized as having high, intermediate, and low risk of AVF primary failure, respectively. The C-statistics of the Lok’s risk equation in the validation data set was 0.53 (95% CI 0.52–0.54). The predicted probabilities of AVF primary failure corresponded poorly with the observed proportions in the calibration plot. Conclusions: When externally applied to a cohort of U.S. older hemodialysis patients, the Lok’s risk equation exhibited poor discrimination and calibration accuracy. It is invalid to use it to predict AVF primary failure. A more complex model with strong predictors is expected to better serve clinical determination for AVF placement in this population.

SuperNNova: an open-source framework for Bayesian, neural network-based supernova classification

ABSTRACTWe introduce SuperNNova, an open-source supernova photometric classification framework that leverages recent advances in deep neural networks. Our core algorithm is a recurrent neural network (RNN) that is trained to classify light curves using only photometric information. Additional information such as host-galaxy redshift can be incorporated to improve performance. We evaluate our framework using realistic supernova simulations that include survey detection. We show that our method, for the type Ia versus non-Ia supernova classification problem, reaches accuracies greater than 96.92 ± 0.09 without any redshift information and up to 99.55 ± 0.06 when redshift, either photometric or spectroscopic, is available. Further, we show that our method attains unprecedented performance for the classification of incomplete light curves, reaching accuracies &amp;gt;86.4 ± 0.1 (&amp;gt;93.5 ± 0.8) without host-galaxy redshift (with redshift information) 2 d before maximum light. In contrast with previous methods, there is no need for time-consuming feature engineering and we show that our method scales to very large data sets with a modest computing budget. In addition, we investigate often neglected pitfalls of machine learning algorithms. We show that commonly used algorithms suffer from poor calibration and overconfidence on out-of-distribution samples when applied to supernova data. We devise extensive tests to estimate the robustness of classifiers and cast the learning procedure under a Bayesian light, demonstrating a much better handling of uncertainties. We study the benefits of Bayesian RNNs for SN Ia cosmology. Our code is open sourced and available on github1.

Can risk modelling improve treatment decisions in asymptomatic carotid stenosis?

BackgroundCarotid endarterectomy (CEA) is routinely performed for asymptomatic carotid stenosis, yet its average net benefit is small. Risk stratification may identify high risk patients that would clearly benefit from treatment.MethodsRetrospective cohort study using data from the Asymptomatic Carotid Atherosclerosis Study (ACAS). Risk factors for poor outcomes were included in backward and forward selection procedures to develop baseline risk models estimating the risk of non-perioperative ipsilateral stroke/TIA. Baseline risk was estimated for all ACAS participants and externally validated using data from the Atherosclerosis Risk in Communities (ARIC) study. Baseline risk was then included in a treatment risk model that explored the interaction of baseline risk and treatment status (CEA vs. medical management) on the patient-centered outcome of any stroke or death, including peri-operative events.ResultsThree baseline risk factors (BMI, creatinine and degree of contralateral stenosis) were selected into our baseline risk model (c-statistic 0.59 [95% CI 0.54–0.65]). The model stratified absolute risk between the lowest and highest risk quintiles (5.1% vs. 12.5%). External validation in ARIC found similar predictiveness (c-statistic 0.58 [0.49–0.67]), but poor calibration across the risk spectrum. In the treatment risk model, CEA was superior to medical management across the spectrum of baseline risk and the magnitude of the treatment effect varied widely between the lowest and highest absolute risk quintiles (3.2% vs. 10.7%).ConclusionEven modestly predictive risk stratification tools have the potential to meaningfully influence clinical decision making in asymptomatic carotid disease. However, our ACAS model requires target population recalibration prior to clinical application.

Individual participant data validation of the PICNICC prediction model for febrile neutropenia

BackgroundRisk-stratified approaches to managing cancer therapies and their consequent complications rely on accurate predictions to work effectively. The risk-stratified management of fever with neutropenia is one such very common area...

Preoperative risk score for prediction of long-term outcomes after hepatectomy for intrahepatic cholangiocarcinoma: Report of a collaborative, international-based, external validation study

PurposeA preoperative risk score (PRS) to predict outcome of patients with intrahepatic cholangiocarcinoma treated by liver surgery could be clinically relevant.To assess accuracy for broadly adoption, external validation of predictive models on independent datasets is crucial.The objective of this study was to externally validate the score for prediction of long-term outcomes after liver surgery for intrahepatic cholangiocarcinoma proposed by Sasaki et al. and based on preoperative albumin, neutrophil-to-lymphocytes-ratio, CA19-9 and tumor size. MethodsPatients treated by liver surgery for intrahepatic cholangiocarcinoma at 11 international HPB centers from 2001 to 2018 were included in the external validation cohort. Harrell's c-index and Hosmer-Lemeshow analyses were used to test PRS discrimination and calibration. Kaplan–Meier curve for risk groups as described in the original study were displayed. ResultsA total of 355 patients with 174 deaths during the follow-up period (median = 41.7 months, IQR 32.8–50.6) were included. The median PRS value was 14.7 (IQR 10.7–20.6), with normal distribution across the cohort. A Cox regression on PRS covariates found coefficients similar to those of the derivation cohort, except for tumor size. Measures of discrimination estimated by Harrell's c-index was 0.61(95%CI:0.56–0.67) and Hosmer-Lemeshow p = 0.175. The Kaplan-Meyer estimation showed reasonable discrimination across risk groups, with 5years survival rate ranging from 20.1% to 0%. ConclusionIn this external validation cohort, the PRS had mild discrimination and poor calibration performance, similarly to the original publication. Nevertheless, its ability to identify different classes of risk is clinically useful, for a better tailoring of a therapeutic strategy.

2319. Clinical Predictors of Influenza and Hospitalization of Children with Influenza in an Emergent Care Setting

BackgroundObjective measures utilizing early vital sign data show promise in predicting more severe outcomes among adults with influenza, but data are sparse in children. The objectives of this study were to determine the value of vital signs in predicting influenza infection or hospitalization due to influenza infection among children evaluated in an emergency department (ED) or urgent care (UC) setting in Colorado.MethodsWe evaluated vital signs obtained from a prospective cohort study of children aged 6 months to 8 years of age with influenza like illness evaluated at an ED/UC site in Aurora, CO from 2016–2018, and who underwent influenza testing by PCR. We collected the first set of vital signs, peak heart rate and temperature, and converted heart rate (HR) and respiratory rate (RR) to z-scores by age. HR z scores were further adjusted for temperature. Bivariable analyses for each vital sign as a predictor of influenza-related hospitalization and influenza infection as main outcomes were conducted. Predictors with P < 0.2 were entered into a multivariable logistic regression model to determine odds ratios (OR) and 95% CI; model performance was assessed using the Brier score and discriminative ability with the C statistic.ResultsAmong 1478 children, 411 were positive for influenza, of which 28 were hospitalized. In multivariable analyses, among children with influenza infection, lower initial oxygen saturation (OR 0.87, 95% CI 0.78–0.98, P = 0.026) and higher adjusted respiratory rate (OR 2.09, 95% CI 1.21–3.61, P = 0.0085) were significant predictors of hospitalization (Figure 1). Among children with ILI, higher peak temperature (OR 1.46, 95% CI 1.30–1.63, P < 0.0001), lower adjusted peak heart rate (OR 0.79, 95% CI 0.69–0.90, P = 0.0005), higher initial oxygen saturation (OR 1.07, 95% CI 1.03–1.12 P = 0.002) and lower adjusted respiratory rate (OR 0.74, 95% CI 0.64–0.87, P = 0.0002) were significant predictors for having PCR-confirmed influenza. However, this model had poor calibration and discriminatory ability.ConclusionHigher respiratory rate adjusted for age and lower initial oxygen saturation were significant predictors of hospitalization among young children with PCR-confirmed influenza, but were not reliable discriminators of having influenza infection. Disclosures All authors: No reported disclosures.

Probabilistic recalibration of forecasts

We present a scheme by which a probabilistic forecasting system whose predictions have a poor probabilistic calibration may be recalibrated through the incorporation of past performance information in order to produce a new forecasting system that is demonstrably superior to the original, inasmuch as one may use it to win wagers consistently against someone who is using the original system. The scheme utilizes Gaussian process (GP) modeling to estimate a probability distribution over the probability integral transform (PIT) of a scalar predictand. The GP density estimate gives closed-form access to information entropy measures that are associated with the estimated distribution, which allows the prediction of winnings in wagers against the base forecasting system. A separate consequence of the procedure is that the recalibrated forecast has a uniform expected PIT distribution. One distinguishing feature of the procedure is that it is appropriate even if the PIT values are not i.i.d. The recalibration scheme is formulated in a framework that exploits the deep connections among information theory, forecasting, and betting. We demonstrate the effectiveness of the scheme in two case studies: a laboratory experiment with a nonlinear circuit and seasonal forecasts of the intensity of the El Niño-Southern Oscillation phenomenon.

P658CHA2DS2-VASc score calibration in anticoagulated vs non-anticoagulated patients in a healthcare area

Abstract Introduction Atrial fibrillation (AF) is a highly prevalent heart disease, affecting a significant proportion of patients over 65 years old. The CHA2DS2-VASc score predicts 1 year risk of a thromboembolic (TE) event and is well validated against several populations. However, calibration may vary if there is subgroup heterogeneity. Purpose To compare the CHA2DS2-VASc score calibration, in patients with or without anticoagulation (AC) in a real population of AF patients in our healthcare area. Methods Patients with an episode with atrial fibrillation/flutter were selected from a general population in a healthcare area (383,000 subjects), with 21/12/2013 as a cut-off date. Patients with valve disease, anticoagulation or antiplatelet therapy were identified. The CHA2DS2-VASc score was calculated as stipulated in the European Society of Cardiology guidelines. A CHA2DS2-VASc score of 0 is considered to be low risk for TE events (0% at 1 year), score of 1 intermediate risk (0.6% rate at 1 year), and greater than 1 high risk (3% rate at 1 year). Quantitative variables are presented as mean and standard deviation (SD). Categorical variables were presented as frequencies and percentages. A logistic regression was fitted to predict 1-year risk TE outcomes with CHA2DS2-VASc as the only covariate. Model calibration was assessed using the predicted versus actual probabilities of TE events. All analyses were performed using R v.3.4 (R Core Team, Vienna, Austria) with the packages rms and ggplot2. Results CHA2DS2-VASc was calculated in 7990 patients with AF. A total of 1824 patients were excluded either due to valvular disease (846) or due to previous antiplatelet treatment (1047). From them, 143 patients were excluded for an incomplete follow-up time (&lt;1 year). As of December 31, 2015, 67 stroke cases had been notified from 6023 patients (1.1%) (Table 1). Mortality rate was 181 (3%) at 1 year. Patients presented overall low risks of stroke with a poor score calibration. Higher scores presented risks that were lower than predicted by CHA2DS2-VASc. Event rate at 1 year was similar regardless of the AC regime at the initial date, and also similar to a previous cohort of anticoagulated patients (Lip et al.). This similarity may indicate confounding by later AC therapy initiation, before the final assessment date. Table 1. Comparison of thromboembolic event rates in several studies % (No-AC) % (AC) % Lip 2010 % Poli 2011 % Friberg 2012 % Okumura 2014 0.01 0.01 0.02 4.5 4.5 1 AC: anticoagulation. CHA2DS2VASc score calibration Conclusion Higher CHA2DS2-VASc scores are not associated to higher risks of stroke in our healthcare area, in patients with non-valvular AF and without antiplatelet therapy.

Validation of the REVEAL Prognostic Equation and Risk Score Calculator in Incident Systemic Sclerosis-Associated Pulmonary Arterial Hypertension.

A prognostic equation and risk score calculator derived from the Registry to Evaluate Early and Long-term Pulmonary Arterial Hypertension Disease Management (REVEAL) are being used to predict 1-year survival in patients with pulmonary arterial hypertension (PAH), but little is known about the performance of these REVEAL survival prediction tools in systemic sclerosis (SSc)-associated PAH (SSc-PAH). Prospectively gathered data from the Johns Hopkins Pulmonary Hypertension Program and Pulmonary Hypertension Assessment and Recognition of Outcome in Scleroderma Registries were used to evaluate the predictive accuracy of the REVEAL models for predicting the probability of 1-year survival in patients with SSc-PAH. Model discrimination was assessed by comparison of the Harrell's C-index, model fit was assessed using multivariable regression techniques, and model calibration was assessed by comparing predicted to observed survival estimates. The validation cohort consisted of 292 SSc-PAH patients with a 1-year survival rate of 87.4%. The C-index for predictive accuracy of the REVEAL prognostic equation (0.734, 95% confidence interval [95% CI] 0.652-0.816) and for the risk score (0.743, 95% CI 0.663-0.823) demonstrated good discrimination, comparable to that in the model development cohort. The calibration slope was 0.707 (95% CI 0.400-1.014), indicating that the overall model fit was marginal (P = 0.06). The magnitude of risk assigned to low distance on the 6-minute walk test (6MWD) and elevated serum levels of brain natriuretic peptide (BNP) was lower in the validation cohort than was originally seen in the REVEAL derivation cohort. Model calibration was poor, particularly for the highest risk groups. In predicting 1-year survival in patients newly diagnosed as having SSc-PAH, the REVEAL prognostic equation and risk score provide very good discrimination but poor calibration. REVEAL prediction scores should be interpreted with caution in newly diagnosed SSc-PAH patients, particularly those with higher predicted risk of poor 1-year survival resulting from a low 6MWD or a high BNP serum level.

National risk prediction model for perioperative mortality in non-cardiac surgery.

Many multivariable models to calculate mortality risk after surgery are limited by insufficient sample size at development or by application to cohorts distinct from derivation populations. The aims of this study were to validate the Surgical Outcome Risk Tool (SORT) for a New Zealand population and to develop an extended NZRISK model to calculate 1-month, 1-year and 2-year mortality after non-cardiac surgery. Data from the New Zealand National Minimum Data Set for patients having surgery between January 2013 and December 2014 were used to validate SORT. A random 75 per cent split of the data was used to develop the NZRISK model, which was validated in the other 25 per cent of the data set. External validation of SORT in the 360 140 patients who underwent surgery in the study period showed good discrimination (area under the receiver operating characteristic curve (AUROC) value of 0·906) but poor calibration (McFadden's pseudo-R2 0·137, calibration slope 5·32), indicating it was invalid in this national surgical population. Internal validation of the NZRISK model, which incorporates sex and ethnicity in addition to the variables used in SORT for 1-month, 1-year and 2-year outcomes, demonstrated excellent discrimination with AUROC values of 0·921, 0·904 and 0·895 respectively, and excellent calibration (McFadden's pseudo-R2 0·275, 0·308 and 0·312 respectively). Calibration slopes were 1·12, 1·02 and 1·02 respectively. The SORT performed poorly in this national population. However, inclusion of sex and ethnicity in the NZRISK model improved performance. Calculation of mortality risk beyond 30 days after surgery adds to the utility of this tool for shared decision-making.

Spatial inventory of selected atmospheric emissions from oil industry in Ecuadorian Amazon: Insights from comparisons among satellite and institutional datasets

Atmospheric emissions from oil activities impact human health, socioeconomic status and exacerbate global warming. This study was conducted in the North-eastern Ecuadorian Amazon, a rich biodiverse and cultural area. This study aimed to show the benefits of public institutional data to advance hazard mapping knowledge for comprehensible risk evaluation. A spatial inventory was built from publicly disclosed reports spanning ten years (2003–2012). Emissions were estimated for gas flaring, associated black carbon (BC) and greenhouse gases (i.e., CO2 and CH4). To assess the quality of publicly available data, the calculated emissions were compared with satellite observations and historical energy statistics from the United Nations (UN). Results indicate total gas flared for this period of 7.6 Gm3, corresponding to 782 Mm3 yr−1, and equivalent to a 3.7–4.5 kt yr−1 of BC. These values were in agreement with the UN estimates, suggesting that publicly available data are of acceptable quality. In contrast, the results from energy censuses diverged from satellite observation data, which might be explained by a poor calibration of satellite sensors. Study results enabled emissions mapping at a higher spatial scale than previous studies. Black carbon presented the highest results with 29.4–148.0 kg m−2 yr−1 in the cities of Shushufindi and Joya de Los Sachas. Greenhouse gases were up to twenty-fold higher than previous estimates. Publicly disclosed data estimates were discussed in terms of their potential on evaluations for climate, local health and economic impacts, to raise environmental monitoring and accountability in governmental institutions.

Validation of the preoperative score to predict postoperative mortality (POSPOM) in patients undergoing major emergency abdominal surgery.

Patients undergoing major emergency abdominal surgery have a high mortality rate. Preoperative risk prediction tools of in-hospital mortality could assist clinical identification of patients at increased risk and thereby aid clinical decision-making and postoperative pathways. The aim of this study was to validate the preoperative score to predict mortality (POSPOM) in a population of patients undergoing major emergency abdominal surgery. POSPOM was investigated in a retrospectively collected cohort of patients undergoing major emergency abdominal surgery at a Danish University Hospital from 2010 to 2016. Predicted in-hospital mortality by POSPOM was compared to observed in-hospital mortality. Calibration was assessed by Hosmer-Lemeshow goodness-of-fit and calibration plot. Discrimination was assessed by area under the receiver operating characteristic curve and accuracy was assessed with Brier score. The study included 979 patients (513 females) with a median age of 64 (IQR 55-77) years. The majority of patients underwent open surgery (94.5%). The observed in-hospital mortality rate was 10.9%. The estimated mean in-hospital mortality rate by POSPOM was 6.7%. POSPOM showed a good discrimination [AUC 0.82 (95% CI 0.78-0.85)] and an excellent accuracy [Brier score 0.09 (95% CI 0.07-0.10)]. However, a poor calibration was found (p < 0.01) as POSPOM underestimated in-hospital mortality. POSPOM is not an ideal prediction model for in-hospital mortality in patients undergoing major emergency abdominal surgery due a poor calibration.

External validation of the International Risk Prediction Algorithm for the onset of generalized anxiety and/or panic syndromes (The Predict A) in the US general population

IntroductionMultivariable risk prediction algorithms are useful for making clinical decisions and health planning. While prediction algorithms for new onset of anxiety disorders in Europe and elsewhere have been developed, the performance of these algorithms in the Americas is not known. The objective of this study was to validate the PredictA algorithm for new onset of anxiety and/or panic disorders in the US general population. MethodsLongitudinal study design was conducted with approximate 2-year follow-up data from a total of 24 626 individuals who participated in Wave 1 and 2 of the US National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) and who did not have generalized anxiety disorder (GAD) and panic disorder in the past year at Wave 1. The PredictA algorithm was directly applied to the selected participants. ResultsAmong the participants, 5.4% developed GAD and/or panic disorder over two years. The PredictA algorithm had a discriminative power (C-statistics = 0.62, 95%CI: 0.61; 0.64), but poor calibration (p < 0.001) with the NESARC data. The observed and the mean predicted risk of GAD and/or panic disorders in the NESARC were 5.3% and 3.6%, respectively. Particularly, the observed and predicted risks of GAD and/or panic disorders in the highest decile of risk score in the NESARC participants were 13.3% and 10.4%, respectively. ConclusionThe PredictA algorithm has acceptable discrimination, but the calibration with the NESARC data was poor. The PredictA algorithm is likely to underestimate the risk of GAD/panic disorders in the US population. Therefore, the use of PredictA in the US general population for predicting individual risk of GAD and/or panic disorders is not encouraged.

Calculating the Index of Volatility in Inhomogeneous Levy Models

The problem of calculating an analog of volatility index (VIX) in exponential Levy models is considered. To obtain the relation for the original index, an assumption is made about the market diffusion model. Unlike Levy models, diffusion models are not able to describe sharp changes of asset prices and offer a poorer calibration flexibility. Relations for calculating an analog of VIX for the exponential Levy model are therefore used, including one with a determinate time change. An explicit form of the relation for the index computation is obtained for the special case of the gamma dispersion model.

Evaluation of quality of care in trauma patients using international scoring systems.

To evaluate the ability of the TRISS and PS14 models to predict mortality rates in our medical system and population. A retrospective observational study was carried out over a 66-month period. The study was conducted in the Trauma Intensive Care Unit (ICU) of a third level hospital. All severe trauma patients (Injury Severity Score≥16 and/or Revised Trauma Score <12) aged> 14 years were included. Medical care data were prospectively recorded. The "W" statistic (difference between expected and observed mortality for every 100 patients) and its significance were calculated for each model. Discrimination and calibration were evaluated by means of receiver operating characteristic (ROC) curves, and the Hosmer-Lemeshow test and GiViTI calibration belt, respectively. A total of 1240 patients were included. Survival at hospital discharge was 81.9%. The "W" scores for the TRISS, TRISS 2010 and PS14 models were+6.72 (P<.01), +1.48 (P=.08) and +2.74 (P<.01) respectively. Subgroup analysis revealed significant favorable results for some populations. The areas under the ROC curve for the TRISS, TRISS 2010 and PS14 models were 0.915, 0.919 and 0.914, respectively. There were no significant differences among them (P>.05). Both the Hosmer-Lemeshow test and GiViTI calibration belt demonstrated poor calibration for the three models. These models are suitable tools for assessing quality of care in a Trauma ICU, affording excellent discrimination but poor calibration. In our institution, survival rates higher than expected were observed.

Performance of atherosclerotic cardiovascular risk prediction models in a rural Northern Chinese population: Results from the Fangshan Cohort Study

Performance of Pooled Cohort Equations (PCEs) for atherosclerotic cardiovascular disease (ASCVD) risks varied across populations. Whether the recently developed Prediction for ASCVD Risk in China (China-PAR) model could accurately predict cardiovascular risks in real practice remains unclear. A population-based cohort study in rural Beijing in the "stroke belt" in North China was used to externally validate PCE and China-PAR models for 5-year ASCVD risk prediction. Expected 5-year prediction risk using China-PAR model was compared with PCE (white). The models were assessed for calibration, discrimination, and reclassification. Among 11,169 adults aged 40 to 79 years over a median 6.44 years of follow-up, 1,921 participants developed a first ASCVD event during total 70,951 person-years. China-PAR model fairly predicted ASCVD risk in men but overestimated by 29.4% risk in women (calibration χ2 = 81.4, P < .001). Underestimations were shown by PCE as 76.2% in men and 88.2% in women with poor calibration (both P < .001). However, discrimination was similar in both models: C-statistics in men were 0.685 (95% CI 0.660-0.710) for China-PAR and 0.675 (95% CI 0.649-0.701) for PCE; C-statistics in women were 0.711 (95% CI 0.694-0.728) for China-PAR and 0.714 (95% CI 0.697-0.731) for PCE. Moreover, China-PAR did not substantially improve accuracy of reclassification compared with PCE. China-PAR outperformed PCE in 5-year ASCVD risk prediction in this rural Northern Chinese population at average population risk level, fairly predicted risk in men, but overestimated risk in women; however, China-PAR did not meaningfully improve the accuracy of discrimination and reclassification at individual risk level.