A Novel Method for Identifying a Parsimonious and Accurate Predictive Model for Multiple Clinical Outcomes

Mapping Intimacies ◽

10.21203/rs.2.20249/v1 ◽

2020 ◽

Author(s):

Ledif Grisell Diaz-Ramirez ◽

Sei J. Lee ◽

Alexander K. Smith ◽

Siqi Gan ◽

Walter John Boscardin

Keyword(s):

Clinical Outcomes ◽

Predictive Accuracy ◽

Information Criterion ◽

Prognostic Models ◽

Nursing Home Admission ◽

Multiple Outcomes ◽

C Statistic ◽

Union Model ◽

Comparable Accuracy ◽

Novel Method

Abstract Background: Most methods for developing clinical prognostic models focus on identifying parsimonious and accurate models to predict a single outcome; however, patients and providers often want to predict multiple outcomes simultaneously. For example, older adults are often interested in predicting nursing home admission as well as mortality. We propose and evaluate a novel predictor selection method for multiple outcomes.Methods: Our proposed method selected the best subset of common predictors based on the minimum average normalized Bayesian Information Criterion (BIC) across outcomes: the Best Average BIC (baBIC) model. We compared the predictive accuracy (Harrell’s C-statistic) and parsimony (number of predictors) of the baBIC model with a subset of common predictors obtained from the union of optimal models for each outcome (Union model). We used example data from the Health and Retirement Study (HRS) to demonstrate our method and conducted a simulation study to investigate performance considering correlated and uncorrelated outcomes.Results: In the example data, the average Harrell’s C-statistics across outcomes of the baBIC and Union models were comparable (0.657 vs. 0.662 respectively). Despite the similar discrimination, the baBIC model was more parsimonious than the Union model (15 vs. 23 predictors respectively). Likewise, in the simulations with correlated outcomes, the mean C-statistic across outcomes of the baBIC and Union models were the same after rounding: 0.650, and the baBIC model had an average number of predictors of 13.8 (95% CI: 13.7, 13.9) compared with 21.6 (95% CI: 21.5, 21.7) in the Union model. In the simulations, the baBIC method performed well by identifying on average the same predictors as in the example data 90.4% times for correlated outcomes.Conclusions: Our method identified a common subset of variables to predict multiple clinical outcomes with superior parsimony and comparable accuracy to current methods.

A Novel Method for Identifying a Parsimonious and Accurate Predictive Model for Multiple Clinical Outcomes

10.21203/rs.2.20249/v2 ◽

2020 ◽

Author(s):

Ledif Grisell Diaz-Ramirez ◽

Sei J. Lee ◽

Alexander K. Smith ◽

Siqi Gan ◽

Walter John Boscardin

Keyword(s):

Clinical Outcomes ◽

Predictive Accuracy ◽

Information Criterion ◽

Prognostic Models ◽

Nursing Home Admission ◽

Multiple Outcomes ◽

C Statistic ◽

Union Model ◽

Comparable Accuracy ◽

Novel Method

Abstract Background: Most methods for developing clinical prognostic models focus on identifying parsimonious and accurate models to predict a single outcome; however, patients and providers often want to predict multiple outcomes simultaneously. For example, older adults are often interested in predicting nursing home admission as well as mortality. We propose and evaluate a novel predictor selection method for multiple outcomes.Methods: Our proposed method selected the best subset of common predictors based on the minimum average normalized Bayesian Information Criterion (BIC) across outcomes: the Best Average BIC (baBIC) model. We compared the predictive accuracy (Harrell’s C-statistic) and parsimony (number of predictors) of the baBIC model with a subset of common predictors obtained from the union of optimal models for each outcome (Union model). We used example data from the Health and Retirement Study (HRS) to demonstrate our method and conducted a simulation study to investigate performance considering correlated and uncorrelated outcomes.Results: In the example data, the average Harrell’s C-statistics across outcomes of the baBIC and Union models were comparable (0.657 vs. 0.662 respectively). Despite the similar discrimination, the baBIC model was more parsimonious than the Union model (15 vs. 23 predictors respectively). Likewise, in two simulation scenarios with correlated and uncorrelated outcomes, the mean C-statistic across outcomes of the baBIC and Union models were very similar, and the baBIC model had on average fewer predictors. In the simulations, the baBIC method performed well by identifying the correct predictors most of the time and excluding the incorrect predictors in the majority of the simulations.Conclusions: Our method identified a common subset of variables to predict multiple clinical outcomes with superior parsimony and comparable accuracy to current methods.

A Novel Method for Identifying a Parsimonious and Accurate Predictive Model for Multiple Clinical Outcomes

10.21203/rs.2.20249/v3 ◽

2021 ◽

Author(s):

Ledif Grisell Diaz-Ramirez ◽

Sei J. Lee ◽

Alexander K. Smith ◽

Siqi Gan ◽

Walter John Boscardin

Keyword(s):

Clinical Outcomes ◽

Predictive Accuracy ◽

Information Criterion ◽

Study Data ◽

Prognostic Models ◽

Nursing Home Admission ◽

Multiple Outcomes ◽

Intersection Method ◽

Case Study Data

Abstract Background and Objective: Most methods for developing clinical prognostic models focus on identifying parsimonious and accurate models to predict a single outcome; however, patients and providers often want to predict multiple outcomes simultaneously. As an example, for older adults one is often interested in predicting nursing home admission as well as mortality. We propose and evaluate a novel predictor-selection computing method for multiple outcomes and provide the code for its implementation.Methods: Our proposed algorithm selected the best subset of common predictors based on the minimum average normalized Bayesian Information Criterion (BIC) across outcomes: the Best Average BIC (baBIC) method. We compared the predictive accuracy (Harrell’s C-statistic) and parsimony (number of predictors) of the model obtained using the baBIC method with: 1) a subset of common predictors obtained from the union of optimal models for each outcome (Union method), 2) a subset obtained from the intersection of optimal models for each outcome (Intersection method), and 3) a model with no variable selection (Full method). We used a case-study data from the Health and Retirement Study (HRS) to demonstrate our method and conducted a simulation study to investigate performance.Results: In the case-study data and simulations, the average Harrell’s C-statistics across outcomes of the models obtained with the baBIC and Union methods were comparable. Despite the similar discrimination, the baBIC method produced more parsimonious models than the Union method. In contrast, the models selected with the Intersection method were the most parsimonious, but with worst predictive accuracy, and the opposite was true in the Full method. In the simulations, the baBIC method performed well by identifying many of the predictors selected in the baBIC model of the case-study data most of the time and excluding those not selected in the majority of the simulations.Conclusions: Our method identified a common subset of variables to predict multiple clinical outcomes with superior balance between parsimony and predictive accuracy to current methods.

A novel method of coronary stent sizing using intravascular ultrasound: Safety and clinical outcomes

International Journal of Angiology ◽

10.1055/s-0031-1278317 ◽

2009 ◽

Vol 18 (01) ◽

pp. 22-24

Author(s):

Christopher B Wong ◽

Norman D Hansen

Keyword(s):

Clinical Outcomes ◽

Intravascular Ultrasound ◽

Coronary Stent ◽

Novel Method

Artificial Intelligence Predicts and Explains West Nile Virus Risks Across Europe: Extraordinary Outbreaks Determined by Climate and Local Factors

10.1101/2020.07.24.20146829 ◽

2020 ◽

Author(s):

Albert A Gayle

Keyword(s):

West Nile Virus ◽

Large Scale ◽

Predictive Accuracy ◽

Local Effect ◽

Research Literature ◽

West Nile ◽

Priority Direction ◽

Minimal Data ◽

Novel Method ◽

Using Data

Year-to-year emergence of West Nile virus has been sporadic and notoriously hard to predict. In Europe, 2018 saw a dramatic increase in the number of cases and locations affected. In this work, we demonstrate a novel method for predicting outbreaks and understanding what drives them. This method creates a simple model for each region that directly explains how each variable affects risk. Behind the scenes, each local explanation model is produced by a state-of-the-art AI engine. This engine unpacks and restructures output from an XGBoost machine learning ensemble. XGBoost, well-known for its predictive accuracy, has always been considered a "black box" system. Not any more. With only minimal data curation and no "tuning", our model predicted where the 2018 outbreak would occur with an AUC of 97%. This model was trained using data from 2010-2016 that reflected many domains of knowledge. Climate, sociodemographic, economic, and biodiversity data were all included. Our model furthermore explained the specific drivers of the 2018 outbreak for each affected region. These effect predictions were found to be consistent with the research literature in terms of priority, direction, magnitude, and size of effect. Aggregation and statistical analysis of local effects revealed strong cross-scale interactions. From this, we concluded that the 2018 outbreak was driven by large-scale climatic anomalies enhancing the local effect of mosquito vectors. We also identified substantial areas across Europe at risk for sudden outbreak, similar to that experienced in 2018. Taken as a whole, these findings highlight the role of climate in the emergence and transmission of West Nile virus. Furthermore, they demonstrate the crucial role that the emerging "eXplainable AI" (XAI) paradigm will have in predicting and controlling disease.

Clinical phenotypes of cardiac sarcoidosis by latent class analysis

European Heart Journal ◽

10.1093/eurheartj/ehab724.2763 ◽

2021 ◽

Vol 42 (Supplement_1) ◽

Author(s):

Q Dai ◽

B Bose ◽

P Li ◽

B Liu ◽

L Jin ◽

...

Keyword(s):

Heart Failure ◽

Latent Class Analysis ◽

Clinical Outcomes ◽

Ventricular Arrhythmias ◽

Latent Class ◽

Cardiac Sarcoidosis ◽

Information Criterion ◽

Cardiometabolic Syndrome ◽

Study Population ◽

Myocardial Involvement

Abstract Background Sarcoidosis is a systemic granulomatous disease with cardiac involvement reported in 20–27% of patients [1]. Cardiac sarcoidosis (CS) can lead to atrial or ventricular arrhythmias, various conduction system disorders, heart failure or sudden cardiac death, depending on the location of myocardial involvement [2]. Previous studies have investigated the possible types of CS based on the distribution of myocardial involvement on imaging as well as the role of genetic factors [3,4]. However, there are no studies describing the clinical heterogeneity of CS patients. Purpose In order to determine if clinical clusters exist in CS, we carried out a latent class analysis (LCA) to explore potential phenotypes in a large sample of CS patients from the National Inpatient Sample (NIS). Methods We identified 848 patients with a diagnosis of CS from the NIS in 2016–2018. A LCA was performed based on comorbidities. Utilizing the Bayesian information criterion and Akaike's information criterion we divided our study population into 3 cohorts. We subsequently applied the LCA model for our study population to fit each patient into one of the 3 cohorts. Finally, we compared the clinical outcomes among the 3 groups. Results Following LCA, patients in cohort 3 were strongly associated with a cardiometabolic syndrome profile with the highest prevalence of congestive heart failure (CHF, 95.1%), chronic kidney disease (CKD, 69.7%), diabetes mellitus (68.9%), hyperlipidemia (52.5%) and obesity (45.1%). Patients in cohort 2 had an intermediate prevalence of cardiometabolic syndrome with a universal diagnosis of hypertension (100%) but with the lowest number of CHF (32.5%) patients and none with CKD. Finally, patients in cohort 1 had the least comorbidities in comparison to the other groups but there was a higher prevalence of CHF (71.7%). There was no significant difference in mortality among the 3 groups, but acute respiratory failure was the highest in cohort 3. However, ventricular arrhythmias were more prevalent in cohort 1 patients (Table). Conclusion We identified 3 different types of CS based on their clinical phenotype. The clinical outcomes varied among the cohorts with ventricular arrhythmias being the most prevalent in patients with the least cardiometabolic comorbidities. FUNDunding Acknowledgement Type of funding sources: None.

Novel Prediction Models of Nonrelapse Mortality in Patients Specific to Each Myeloablative Conditioning before Allogeneic Hematopoietic Stem Cell Transplantation: A Multicenter Analysis from the Kanto Study Group for Cell Therapy (KSGCT)

Blood ◽

10.1182/blood-2020-134539 ◽

2020 ◽

Vol 136 (Supplement 1) ◽

pp. 28-29

Author(s):

Arata Ishii ◽

Akira Yokota ◽

Emiko Sakaida ◽

Shokichi Tsukamoto ◽

Katsuhiro Shono ◽

...

Keyword(s):

Risk Factors ◽

Stem Cell ◽

Liver Function ◽

Cell Therapy ◽

Cell Transplantation ◽

Research Funding ◽

Prognostic Models ◽

Myeloablative Conditioning ◽

C Statistic ◽

Otsuka Pharmaceutical

Introduction Despite recent developments on various transplantation procedures and supportive therapy, nonrelapse mortality (NRM) after allogeneic stem cell transplantation (allo-SCT) remains an essential issue. In choosing the appropriate regimen for allo-SCT, decision-making information that considers the complexity of different risk factors is vital. The Hematopoietic Cell Transplantation-Comorbidity Index (HCT-CI), which was initially derived and validated by investigators at the Fred Hutchinson Cancer Research Center to predict NRM, has become a widely validated tool for predicting outcomes in many transplant settings (Sorror et al. Blood. 2005). It can also stratify patients for the risk of other outcomes, including overall survival and graft versus host disease. Patients with a high HCT-CI score tend to prefer allo-SCT with reduced-intensity conditioning. Conversely, for those who prefer allo-SCT with myeloablative conditioning (MAC) and has a low HCT-CI score, a prognostic indicator is unnecessary. Furthermore, the risk factors for NRM may differ among various conditioning regimens. Therefore, the current study aimed to establish a new prognostic model for patients specific to each MAC regimen before allo-SCT. Methods We performed a retrospective cohort study to develop prognostic models of NRM in patients conditioned with cyclophosphamide/total body irradiation (Cy/TBI) or busulfan/cyclophosphamide (Bu/Cy). We selected patients who had leukemia and lymphoma in remission or had untreated or stable myelodysplastic syndrome and experienced initial allo-SCT relapse between 2007 and 2017 in the Kanto Study of Group for Cell Therapy (KSGCT). The primary outcome measure was 2-year NRM. Furthermore, we evaluated variables such as patient age, albumin, liver function, renal function, respiratory function, ejection fraction (EF), C-reactive protein (CRP), stem cell source, donor type, antithymocyte globulin use, performance status, recipient/donor sexes, time interval from diagnosis to transplant, and HCT-CI score. To identify a set of variables for Cox proportional hazards, we used an Akaike Information Criterion (AIC)-based variable selection procedure. We assigned weights to individual parameters according to their prognostic significance in Cox proportional hazard models. The identified model's discriminative ability was assessed by Harrell's C-statistic calculated using the bootstrap method. Results Among the 555 patients analyzed, 338 received Cy/TBI, and 217 received Bu/Cy. In Cy/TBI and Bu/Cy, the median age was 39 (11-60) and 44 (18-62) years, the HCT-CI score ≤ 2 was observed in 82.1% and 87.6%, and 2-year NRM was found in 13.5% and 16.0% of the patients, respectively. Before transplantation, the most dominant parameters in Cy/TBI were abnormal liver function (AST/ALT or bilirubin >upper limit of normal) and albumin value < 4.5g/dL, whereas those in Bu/Cy were age >40 years, EF < 65 %, and CRP ≥ 0.2 mg/dL. Internal validation with bootstrap resampling showed good discrimination, with C-statistic values of 0.70 (95% CI: 0.69-0.71) in Cy/TBI and 0.68 (95% CI: 0.67-0.69) in Bu/Cy. Each of the abovementioned parameters, including age >40 years, was scored as 1 point. To evaluate the 2-year NRM, we divided the total scores into three risk groups. In the Cy/TBI group, the NRM was 6.9% in low (score 0-1, n = 186), 19.5% in intermediate (score 2, n = 127), and 35.3% in high (score 3, n = 25) scores. In the Bu/Cy group, the NRM was 8.3% in low (score 0-1, n = 93), 21.7% in intermediate (score 2, n = 98), and 29.8% in high (score 3, n = 26) scores (Figure). Higher scores were strongly associated with worse NRM and survival. Conclusions Our prognostic models for NRM estimation can distinguish patients with a high NRM risk. To our knowledge, these models are the first prognostic models used to estimate NRM for standard-risk patients specific to each MAC regimen. This new simple index may help predict NRM and choose an appropriate conditioning regimen before allo-SCT. Figure 1 Disclosures Nakasone: Takeda Pharmaceutical: Honoraria; Otsuka Pharmaceutical: Honoraria; Bristol-Myers Squibb: Honoraria; Celgene: Honoraria; Pfizer: Honoraria; Novartis: Honoraria; Janssen Pharmaceutical: Honoraria; Eisai: Honoraria; Chugai Pharmaceutical: Honoraria; Nippon Shinyaku: Honoraria. Fujisawa:Takeda Pharmaceutical Company Limited.: Speakers Bureau; Astellas Pharma Inc.: Research Funding, Speakers Bureau; Otsuka Pharmaceutical: Speakers Bureau; Pfizer Japan Inc.: Research Funding, Speakers Bureau; Bristol-Myers Squibb Company: Speakers Bureau; Novartis Pharma KK: Research Funding, Speakers Bureau; Celgene: Speakers Bureau; Janssen Pharmaceutical K.K: Speakers Bureau; NIPPON SHINYAKU CO.,LTD.: Research Funding. Nakaseko:Novartis Pharma KK: Speakers Bureau; Pfizer Japan Inc.: Speakers Bureau. Kanda:Novartis: Honoraria; Kyowa Kirin: Honoraria, Research Funding; Bristol-Myers Squibb: Honoraria; Takeda Pharmaceuticals: Honoraria; Alexion Pharmaceuticals: Honoraria; Shire: Honoraria; Daiichi Sankyo: Honoraria; Ono Pharmaceutical: Honoraria; Nippon Shinyaku: Honoraria, Research Funding; Mochida Pharmaceutical: Honoraria; Mundipharma: Honoraria; Sanofi: Honoraria, Research Funding; Meiji Seika Kaisha: Honoraria; Shionogi: Research Funding; Otsuka: Honoraria, Research Funding; Celgene: Honoraria; Chugai Pharma: Honoraria, Research Funding; Eisai: Honoraria, Research Funding; Janssen: Honoraria; Astellas Pharma: Honoraria, Research Funding; Sumitomo Dainippon Pharma: Honoraria; Pfizer: Honoraria, Research Funding; Merck Sharp & Dohme: Honoraria.

Different Nonlinear Regression Techniques and Sensitivity Analysis as Tools to Optimize Oil Viscosity Modeling

Resources ◽

10.3390/resources10100099 ◽

2021 ◽

Vol 10 (10) ◽

pp. 99

Author(s):

Dicho Stratiev ◽

Svetoslav Nenov ◽

Dimitar Nedanovski ◽

Ivelina Shishkova ◽

Rosen Dinkov ◽

...

Keyword(s):

Sensitivity Analysis ◽

Nonlinear Regression ◽

Information Criterion ◽

Least Square ◽

Gas Oil ◽

Oil Viscosity ◽

Prediction Ability ◽

Regression Techniques ◽

Comparable Accuracy ◽

Relative Errors

Four nonlinear regression techniques were explored to model gas oil viscosity on the base of Walther’s empirical equation. With the initial database of 41 primary and secondary vacuum gas oils, four models were developed with a comparable accuracy of viscosity calculation. The Akaike information criterion and Bayesian information criterion selected the least square relative errors (LSRE) model as the best one. The sensitivity analysis with respect to the given data also revealed that the LSRE model is the most stable one with the lowest values of standard deviations of derivatives. Verification of the gas oil viscosity prediction ability was carried out with another set of 43 gas oils showing remarkably better accuracy with the LSRE model. The LSRE was also found to predict better viscosity for the 43 test gas oils relative to the Aboul Seoud and Moharam model and the Kotzakoulakis and George.

Relevance of Brain Regions’ Eloquence Assessment in Patients With a Large Ischemic Core Treated With Mechanical Thrombectomy

Neurology ◽

10.1212/wnl.0000000000012863 ◽

2021 ◽

pp. 10.1212/WNL.0000000000012863

Author(s):

Basile Kerleroux ◽

Joseph Benzakoun ◽

Kévin Janot ◽

Cyril Dargazanli ◽

Dimitri Daly Eraya ◽

...

Keyword(s):

Logistic Regression ◽

Random Forest ◽

Regression Model ◽

Logistic Regression Model ◽

Mechanical Thrombectomy ◽

Predictive Accuracy ◽

Unmet Need ◽

C Statistic ◽

Ischemic Core ◽

Functional Relevance

ObjectiveIndividualized patient selection for mechanical thrombectomy (MT) in patients with acute ischemic stroke (AIS) and large ischemic core (LIC) at baseline is an unmet need.We tested the hypothesis, that assessing the functional relevance of both the infarcted and hypo-perfused brain tissue, would improve the selection framework of patients with LIC for MT.MethodsMulticenter, retrospective, study of adult with LIC (ischemic core volume > 70ml on MR-DWI), with MRI perfusion, treated with MT or best medical management (BMM).Primary outcome was 3-month modified-Rankin-Scale (mRS), favourable if 0-3. Global and regional-eloquence-based core-perfusion mismatch ratios were derived. The predictive accuracy for clinical outcome of eloquent regions involvement was compared in multivariable and bootstrap-random-forest models.ResultsA total of 138 patients with baseline LIC were included (MT n=96 or BMM n=42; mean age±SD, 72.4±14.4years; 34.1% females; mRS=0-3: 45.1%). Mean core and critically-hypo-perfused volume were 100.4ml±36.3ml and 157.6±56.2ml respectively and did not differ between groups. Models considering the functional relevance of the infarct location showed a better accuracy for the prediction of mRS=0-3 with a c-Statistic of 0.76 and 0.83 for logistic regression model and bootstrap-random-forest testing sets respectively. In these models, the interaction between treatment effect of MT and the mismatch was significant (p=0.04). In comparison in the logistic regression model disregarding functional eloquence the c-Statistic was 0.67 and the interaction between MT and the mismatch was insignificant.ConclusionConsidering functional eloquence of hypo-perfused tissue in patients with a large infarct core at baseline allows for a more precise estimation of treatment expected benefit.

Abstract 62: Refining Angiographic Biomarkers of Reperfusion: Modified TICI Is Superior to TIMI for Predicting Clinical Outcomes After Intra-arterial Therapy

Stroke ◽

10.1161/str.44.suppl_1.a62 ◽

2013 ◽

Vol 44 (suppl_1) ◽

Author(s):

Albert J Yoo ◽

Claus Z Simonsen ◽

Shyam Prabhakaran ◽

Zeshan A Chaudhry ◽

Mohammad Issa ◽

...

Keyword(s):

Acute Stroke ◽

Predictive Accuracy ◽

Stroke Patients ◽

Critical Determinant ◽

Term Outcome ◽

Long Term Outcome ◽

Good Outcome ◽

C Statistic ◽

Standard Tool ◽

Angiographic Data

Background: Tissue reperfusion is a critical determinant of outcomes after intra-arterial therapy (IAT). However, there is no standardized method for grading angiographic reperfusion. We sought to compare the two most commonly used reperfusion scales, TIMI and modified TICI (m-TICI), for predicting good long-term outcome following IAT. Methods: From a multicenter database, we evaluated acute stroke patients presenting with middle cerebral artery (MCA) M1 occlusions who underwent IAT. Clinical and angiographic data were collected. Reperfusion for each case was graded using the TIMI and m-TICI scales. The primary distinction between these scales is that partial reperfusion (TIMI 2) is sub-divided into minor (m-TICI 2A: <50%) and major (m-TICI 2B: 50-99%) grades in the m-TICI system. The performance of these scales for predicting a good 90-day outcome (mRS 0-2) was evaluated using the c-statistic. Results: There were 313 acute stroke patients with MCA M1 segment occlusions who underwent IAT at 6 academic centers; 171 (54.6%) were female. Mean age was 65.3±16.5 years. There were 157 (50.2%) right-sided strokes. Median baseline NIHSS score was 17 (IQR 15-20). Good outcome at 90 days was achieved in 32.5%. For predicting good outcome, the c-statistic was significantly higher for m-TICI (0.74 vs. 0.68; p<0.0001). The threshold that maximized predictive accuracy was m-TICI ≥2B (sensitivity 78%, specificity 65%; figure). Conclusions: The modified TICI scale is superior to the TIMI scale for predicting clinical outcome after IAT, and should be the standard tool for grading angiographic reperfusion. An m-TICI score ≥2B (≥50% reperfusion) is the optimal biomarker for successful reperfusion.

Assessing the Veterans Affairs Surgical Quality Improvement Program Risk Calculator in Cholecystectomy

The American Surgeon ◽

10.1177/000313481808400662 ◽

2018 ◽

Vol 84 (6) ◽

pp. 1039-1042 ◽

Cited By ~ 2

Author(s):

Jonathan B. Imran ◽

Oswaldo Renteria ◽

Maria Ruiz ◽

Thai H. Pham ◽

Ali A. Mokdad ◽

...

Keyword(s):

Quality Improvement ◽

Predictive Accuracy ◽

Weighted Least Squares ◽

Veterans Affairs ◽

Quality Improvement Program ◽

Least Squares Regression ◽

Risk Calculator ◽

Improvement Program ◽

Surgical Quality ◽

C Statistic

The Veterans Affairs Surgical Quality Improvement Program (VASQIP) risk calculator has been validated for several operations but has not been assessed specifically for cholecystectomy. Our aim was to externally validate the VASQIP calculator's accuracy in predicting 30-day morbidity and mortality (M&M) for patients undergoing cholecystectomy. A retrospective review of patients undergoing cholecystectomy at the North Texas Veterans Affairs hospital was performed. The VASQIP risk calculator was used to determine predicted 30-day M&M, which was compared with actual M&M. The predictive accuracy of the Veterans Affairs risk calculator was assessed using the C-statistic and a graphical assessment of a locally weighted least squares regression smoother. Overall, 848 patients were included in the study. Actual M&M were 6.3 and 0.94 per cent, respectively, whereas predicted M&M were 6.0 and 0.54 per cent. The C-statistic was 0.75 for morbidity and 0.78 for mortality. In our analysis, the VASQIP risk calculator reasonably predicted 30-day M&M.