Search tips
Search criteria

Results 1-25 (762736)

Clipboard (0)

Related Articles

1.  Comparison of the performance of the CRUSADE, ACUITY-HORIZONS, and ACTION bleeding risk scores in STEMI undergoing primary PCI: insights from a cohort of 1391 patients 
To compare the performance of the CRUSADE, ACUITY-HORIZONS, and ACTION risk models in the ST-segment elevation myocardial infarction (STEMI) undergoing primary percutaneous coronary intervention (PPCI).
We studied all consecutive patients with STEMI who underwent PPCI at our institution between 2006 and 2010 (n=1391). The CRUSADE, ACUITY-HORIZONS, and ACTION risk scores were calculated based on the patients’ clinical characteristics. The occurrence of in-hospital major bleeding (defined as the composite of intracranial or intraocular bleeding, access site haemorrhage requiring intervention, reduction in haemoglobin ≥4 g/dl without or ≥3g/dl with overt bleeding source, reoperation for bleeding, or blood transfusion) reached 9.8%. Calibration and discrimination of the three risk models were evaluated by the Hosmer−Lemeshow test and the C-statistic, respectively. We compared the predictive accuracy of the risk scores by the DeLong non-parametric test.
Calibration of the three risk scores was adequate, given the non-significant results of Hosmer−Lemeshow test for the three risk models. Discrimination of CRUSADE, ACUITY-HORIZONS, and ACTION models was good (C-statistic 0.77, 0.70, and 0.78, respectively). The CRUSADE and ACTION risk scores had a greater predictive accuracy than the ACUITY-HORIZONS risk model (z=3.89, p-value=0.0001 and z=3.51, p-value=0.0004, respectively). There was no significant difference between the CRUSADE and ACTION models (z=0.63, p=0.531).
The CRUSADE, ACUITY-HORIZONS, and ACTION scores are useful tools for the risk stratification of bleeding in STEMI treated by PPCI. Our findings favour the CRUSADE and ACTION risk models over the ACUITY-HORIZONS risk score.
PMCID: PMC3760576  PMID: 24062930
Bleeding; ischaemic heart disease; myocardial infarction; primary percutaneous coronary intervention; risk score
2.  Prediction of the First Variceal Haemorrhage 
HPB Surgery  1997;10(4):255-258.
We followed 87 cirrhotic patients with esophageal varices and without previous hemorrhage for a mean period of 24 mo to prospectively evaluate the occurance of variceal bleeding within (early) or after (late) 6 mo from entry and the contribution of portal Doppler ultrasound parameters to the prediction of early and late hemorrhage. Clinical, biochemical, endoscopic and portal Doppler ultrasound parameters were recorded at entry. Variceal bleeding occurred in 22 patients (25.3%). Nine (40.9%) bled within the first 6 mo. Cox regression analysis identified variceal size, cherry-red spots, serum bilirubin and congestion index of the portal vein (the ratio of portal vein [cross-sectional area] and portal blood flow velocity) as the only independent predictors of first variceal hemorrhage. Discriminant analysis was used to find the prognostic index cut off points to identify patients who bled within 6 mo (prognostic group 1) or after 6 mo (prognostic group 2) or remained free of bleeding (prognostic group 3). The cumulative proportion of patients correctly classified was 73% in prognostic group 1, 47% in prognostic group 2 and more than 80% in prognostic group 3. The addition of Doppler ultrasound flowmetry to clinical, biochemical and endoscopic parameter only improved the classification of patients with early bleeding.
PMCID: PMC2423869  PMID: 9184882
3.  Length of intensive care unit stay following cardiac surgery: is it impossible to find a universal prediction model? 
Accurate models for prediction of a prolonged intensive care unit (ICU) stay following cardiac surgery may be developed using Cox proportional hazards regression. Our aims were to develop a preoperative and intraoperative model to predict the length of the ICU stay and to compare our models with published risk models, including the EuroSCORE II.
Models were developed using data from all patients undergoing cardiac surgery at St. Olavs Hospital, Trondheim, Norway from 2000–2007 (n = 4994). Internal validation and calibration were performed by bootstrapping. Discrimination was assessed by areas under the receiver operating characteristics curves and calibration for the published logistic regression models with the Hosmer-Lemeshow test.
Despite a diverse risk profile, 93.7% of the patients had an ICU stay <2 days, in keeping with our fast-track regimen. Our models showed good calibration and excellent discrimination for prediction of a prolonged stay of more than 2, 5 or 7 days. Discrimination by the EuroSCORE II and other published models was good, but calibration was poor (Hosmer-Lemeshow test: P < 0.0001), probably due to the short ICU stays of almost all our patients. None of the models were useful for prediction of ICU stay in individual patients because most patients in all risk categories of all models had short ICU stays (75th percentiles: 1 day).
A universal model for prediction of ICU stay may be difficult to develop, as the distribution of length of stay may depend on both medical factors and institutional policies governing ICU discharge.
PMCID: PMC3480596  PMID: 22833511
Risk prediction; Length of ICU stay; Cardiac surgery
4.  Development and validation of a risk score for hospitalization for heart failure in patients with Type 2 Diabetes Mellitus 
There are no risk scores available for predicting heart failure in Type 2 diabetes mellitus (T2DM). Based on the Hong Kong Diabetes Registry, this study aimed to develop and validate a risk score for predicting heart failure that needs hospitalisation in T2DM.
7067 Hong Kong Chinese diabetes patients without history of heart failure, and without history and clinical evidence of coronary heart disease at baseline were analyzed. The subjects have been followed up for a median period of 5.5 years. Data were randomly and evenly assigned to a training dataset and a test dataset. Sex-stratified Cox proportional hazard regression was used to obtain predictors of HF-related hospitalization in the training dataset. Calibration was assessed using Hosmer-Lemeshow test and discrimination was examined using the area under receiver's operating characteristic curve (aROC) in the test dataset.
During the follow-up, 274 patients developed heart failure event/s that needed hospitalisation. Age, body mass index (BMI), spot urinary albumin to creatinine ratio (ACR), HbA1c, blood haemoglobin (Hb) at baseline and coronary heart disease during follow-up were predictors of HF-related hospitalization in the training dataset. HF-related hospitalization risk score = 0.0709 × age (year) + 0.0627 × BMI (kg/m2) + 0.1363 × HbA1c(%) + 0.9915 × Log10(1+ACR) (mg/mmol) - 0.3606 × Blood Hb(g/dL) + 0.8161 × CHD during follow-up (1 if yes). The 5-year probability of heart failure = 1-S0(5)EXP{0.9744 × (Risk Score - 2.3961)}. Where S0(5) = 0.9888 if male and 0.9809 if female. The predicted and observed 5-year probabilities of HF-related hospitalization were similar (p > 0.20) and the adjusted aROC was 0.920 for 5 years of follow-up.
The risk score had adequate performance. Further validations in other cohorts of patients with T2DM are needed before clinical use.
PMCID: PMC2377240  PMID: 18430204
5.  Calibration Belt for Quality-of-Care Assessment Based on Dichotomous Outcomes 
PLoS ONE  2011;6(2):e16110.
Prognostic models applied in medicine must be validated on independent samples, before their use can be recommended. The assessment of calibration, i.e., the model's ability to provide reliable predictions, is crucial in external validation studies. Besides having several shortcomings, statistical techniques such as the computation of the standardized mortality ratio (SMR) and its confidence intervals, the Hosmer–Lemeshow statistics, and the Cox calibration test, are all non-informative with respect to calibration across risk classes. Accordingly, calibration plots reporting expected versus observed outcomes across risk subsets have been used for many years. Erroneously, the points in the plot (frequently representing deciles of risk) have been connected with lines, generating false calibration curves. Here we propose a methodology to create a confidence band for the calibration curve based on a function that relates expected to observed probabilities across classes of risk. The calibration belt allows the ranges of risk to be spotted where there is a significant deviation from the ideal calibration, and the direction of the deviation to be indicated. This method thus offers a more analytical view in the assessment of quality of care, compared to other approaches.
PMCID: PMC3043050  PMID: 21373178
6.  ICU Acuity: Real-time Models versus Daily Models 
To explore the feasibility of real-time mortality risk assessment for ICU patients.
This study used retrospective analysis of mixed medical/surgical intensive care patients in a university hospital. Logistic regression was applied to 7048 development patients with several hundred candidate variables. Final models were selected by backward elimination on top cross-validated variables and validated on 3018 separate patients.
The real-time model demonstrated strong discrimination ability (Day 3 AUC=0.878). All models had circumstances where calibration was poor (Hosmer-Lemeshow goodness of fit test p < 0.1). The final models included variables known to be associated with mortality, but also more computationally intensive variables absent in other severity scores.
Real-time mortality prediction offers similar discrimination ability to daily models. Moreover, the discrimination of our real-time model performed favorably to a customized SAPS II (Day 3 AUC=0.878 vs AUC=0.849, p < 0.05) but generally had worse calibration.
PMCID: PMC2815497  PMID: 20351861
7.  A new model for end-stage liver disease improves prognostic capability after transjugular intrahepatic portosystemic shunt 
Background & Aims
Cirrhotics undergoing transjugular intrahepatic portosystemic shunt (TIPS) for refractory ascites or recurrent variceal bleeding are at risk for decompensation and death. This study examined whether a new model for end-stage liver disease (MELDNa), which incorporates serum sodium, is a better predictor of death or transplant after TIPS than the original MELD.
148 consecutive patients undergoing non-emergent TIPS for refractory ascites or recurrent variceal bleeding from 1997 to 2006 at a single center were evaluated retrospectively. Cox model analysis was performed with death or transplant within 6 months as the end point. The models were compared using the Harrell’s C index. Recursive partitioning determined the optimal MELDNa cut-off to maximize the risk-benefit ratio of TIPS.
The predictive ability of MELDNa was superior to MELD, particularly in patients with low MELD scores. The C indices (95% CI) for MELDNa and MELD were 0.65 (0.55, 0.71) and 0.58 (0.51, 0.67) using a cut-off score of 18, and 0.72 (0.60, 0.85) and 0.62 (0.49, 0.74) using a cut-off score of 15. Using a MELDNa > 15, 22% of patients were reclassified to a higher risk with an event rate of 44% compared to 10% when the score was ≤ 15.
MELDNa performed better than MELD in predicting death or transplant after TIPS, especially in patients with low MELD scores. In cirrhotics undergoing non-emergent TIPS, a MELD score ≤ 18 can provide a false positive prognosis; a MELDNa score ≤ 15 provides a more accurate risk prediction.
PMCID: PMC2783337  PMID: 19560557
8.  Evaluation of accuracy of Euroscore risk model in prediction of perioperative mortality after coronary bypass graft surgery in Isfahan* 
This study aimed to evaluate the accuracy of Euroscore (European System for Cardiac Operative Risk Evaluation) in predicting perioperative mortality after cardiac surgery in Iranian patient population.
Data on 1362 patients undergoing coronary bypass graft surgery (CABG) from 2007 to 2009 were collected. Calibration was assessed by Hosmer-Lemeshow goodness-of-fit. Area under the curve (AUC) was used to assess score validity. Odds ratios were measured to evaluate the predictive value of each risk factor on mortality rate.
The overall perioperative in hospital mortality was 3.6% whereas the Euroscore predicted a mortality of 3.96%. Euroscore model fitted well in the validation databases. The mean AUC was 66%. Mean length of intensive care unit (ICU) stay was 2.5 ± 2.5 days. Among risk factors, only left ventricular dysfunction, age and neurologic dysfunction were found to be related to mortality rate.
Euroscore did not have acceptable discriminatory ability in perioperative in hospital mortality in Iranian patients. It seems that development of a local mortality risk scores corresponding to our patients epidemiologic characteristics may improve prediction of outcome.
PMCID: PMC3214397  PMID: 22091308
Euroscore model; hospital; mortality; preoperative; cardiac surgery
9.  Variceal pressure is a strong predictor of variceal haemorrhage in patients with cirrhosis as well as in patients with non-cirrhotic portal hypertension 
Gut  1999;45(4):618-621.
BACKGROUND—Variceal pressure is a strong predictor for a first variceal bleed in patients with cirrhosis.
AIMS—To evaluate whether variceal pressure is also a determinant of the risk of a first variceal bleed in patients with non-cirrhotic portal hypertension.
METHODS—Variceal pressure was measured non-invasively in 25 patients with non-cirrhotic portal hypertension and large varices while receiving a stable therapeutic regimen. Factors predictive of bleeding were compared with those observed in 87 cirrhotics.
RESULTS—The one year incidence of variceal bleeding was 32% (n=28) for the cirrhotic and 20% (n=5) for the non-cirrhotic patients. There was no difference in factors predicting the risk of bleeding between the groups, except for variceal pressure. For the same level of variceal pressure, the risk of variceal bleeding was lower in patients with non-cirrhotic portal hypertension. Multiple logistic regression analysis revealed the following variables as having a significant predictive power: variceal pressure (p=0.0001), red spots (p=0.004), and the time interval between the first observation of the varices and the moment of variceal pressure measurement (p=0.0046). For the non-cirrhotics the risk of bleeding increased with higher Child-Pugh score (p=0.0024); this was not the case for the cirrhotic patients (p=0.9521).
CONCLUSION—Variceal pressure is a major predictor of variceal bleeding in patients with cirrhosis as well as in patients with non-cirrhotic portal hypertension. The risk of bleeding in non-cirrhotics is less than in cirrhotics for the same level of variceal pressure. In patients with non-cirrhotic portal hypertension the risk of variceal bleeding increases more with advancing disease.

Keywords: variceal haemorrhage; variceal pressure; non-cirrhotic portal hypertension
PMCID: PMC1727662  PMID: 10486375
10.  Multivariate prediction of major adverse cardiac events after 9914 percutaneous coronary interventions in the north west of England 
Heart  2005;92(5):658-663.
To develop a multivariate prediction model for major adverse cardiac events (MACE) after percutaneous coronary interventions (PCIs) by using the North West Quality Improvement Programme in Cardiac Interventions (NWQIP) PCI Registry.
All NHS centres undertaking adult PCIs in north west England.
Retrospective analysis of prospectively collected data on 9914 consecutive patients undergoing adult PCI between 1 August 2001 and 31 December 2003. A multivariate logistic regression analysis was undertaken, with the forward stepwise technique, to identify independent risk factors for MACE. The area under the receiver operating characteristic (ROC) curve and the Hosmer‐Lemeshow goodness of fit statistic were calculated to assess the performance and calibration of the model, respectively. The statistical model was internally validated by using the technique of bootstrap resampling.
Main outcome measures
MACE, which were in‐hospital mortality, Q wave myocardial infarction, emergency coronary artery bypass graft surgery, and cerebrovascular accidents.
Independent variables identified with an increased risk of developing MACE were advanced age, female sex, cerebrovascular disease, cardiogenic shock, priority, and treatment of the left main stem or graft lesions during PCI. The ROC curve for the predicted probability of MACE was 0.76, indicating a good discrimination power. The prediction equation was well calibrated, predicting well at all levels of risk. Bootstrapping showed that estimates were stable.
A contemporaneous multivariate prediction model for MACE after PCI was developed. The NWQIP tool allows calculation of the risk of MACE permitting meaningful risk adjusted comparisons of performance between hospitals and operators.
PMCID: PMC1860907  PMID: 16159983
major adverse cardiac events; percutaneous coronary interventions; risk prediction
11.  Validation of four different risk stratification systems in patients undergoing off-pump coronary artery bypass surgery: a UK multicentre analysis of 2223 patients 
Heart  2003;89(4):432-435.
Background: Various risk stratification systems have been developed in coronary artery bypass graft surgery (CABG), based mainly on patients undergoing procedures with cardiopulmonary bypass.
Objective: To assess the validity and applicability of the Parsonnet score, the EuroSCORE, the American College of Cardiology/American Heart Association (ACC/AHA) system, and the UK CABG Bayes model in patients undergoing off-pump coronary artery bypass surgery (OPCAB) in the UK.
Methods: Data on 2223 patients who underwent OPCAB in eight cardiac surgical centres were collected. Predicted mortality risk scores were calculated using the four systems and compared with observed mortality. Calibration was assessed by the Hosmer–Lemeshow (HL) test. Discrimination was assessed using the receiver operating characteristic (ROC) curve area.
Results: 30 of 2223 patients (1.3%) died in hospital. For the Parsonnet score the HL test was significant (p < 0.001) and the receiver operating characteristic curve (ROC) area was 0.74. For the EuroSCORE the HL test was also significant (p = 0.008) and the ROC area was 0.75. For the ACC/AHA system the HL test was non-significant (p = 0.7) and the ROC area was 0.75. For the UK CABG Bayes model the HL test was also non-significant (p = 0.3) and the ROC area was 0.81.
Conclusions: The UK CABG Bayes model is reasonably well calibrated and provides good discrimination when applied to OPCAB patients in the UK. Among the other three systems, the ACC/AHA system is well calibrated but its discrimination power was less than for the UK CABG Bayes model. These data suggest that the UK CABG Bayes model could be an appropriate risk stratification system to use for patients undergoing OPCAB in the UK.
PMCID: PMC1769277  PMID: 12639875
risk stratification; coronary artery bypass graft surgery
12.  Prevalence and predictors of over-the-counter medication use among pregnant women: a cross-sectional study in the Netherlands 
BMC Public Health  2013;13:185.
Over-the-counter-medication (OTC-medication) use during pregnancy can be potentially harmful for the fetus. To successfully counsel the patient it is important to know if the patient is at risk. In this study possible predictors for OTC-medication use were identified and a model was designed to predict OTC-medication use during pregnancy.
We performed a post-hoc analysis on data collected for a clustered clinical trial to study a screening strategy for Query fever. Pregnant women under supervision of a midwife were eligible for inclusion. These women filled out questionnaires during their pregnancy and post-partum. These questionnaires were used to determine the prevalence and to select possible predictors for OTC-medication use. These predictors were included in a prediction model using multivariate analysis. The discrimination and calibration of the model were assessed with Receiver Operating Characteristic analysis and the Hosmer and Lemeshow test.
Of the 1348 women enrolling in the clustered clinical trial, we included 1246 women in this analysis. The prevalence of OTC-medication use was 12.5%. The predictors for OTC-medication use in our cohort were: nulliparity, use of prescription medication, the presence of a comorbidity, Body Mass Index between 26 and 30 kg/m2 and General Practitioner visits. These predictors were used to design a prediction model for OTC-medication use. The area under the Receiver Operating Characteristic-curve of the prediction model was 0.667 (95% CI 0.620-0.714 P<0.001) and the predictive probabilities ranged from 6.6% to 57.4%. The Hosmer and Lemeshow goodness-of-fit test indicated good calibration of the model (P = 0.640).
It is possible to indicate women at risk for OTC-medication use during pregnancy, using five maternal characteristics that independently contribute to the prediction model. The predictors are easy to estimate and the model is easy to implement in daily practice.
PMCID: PMC3662163  PMID: 23452432
OTC-medication; Pregnancy; The Netherlands; Epidemiology; Prediction model
13.  Prediction models for risk of developing type 2 diabetes: systematic literature search and independent external validation study 
Objective To identify existing prediction models for the risk of development of type 2 diabetes and to externally validate them in a large independent cohort.
Data sources Systematic search of English, German, and Dutch literature in PubMed until February 2011 to identify prediction models for diabetes.
Design Performance of the models was assessed in terms of discrimination (C statistic) and calibration (calibration plots and Hosmer-Lemeshow test).The validation study was a prospective cohort study, with a case cohort study in a random subcohort.
Setting Models were applied to the Dutch cohort of the European Prospective Investigation into Cancer and Nutrition cohort study (EPIC-NL).
Participants 38 379 people aged 20-70 with no diabetes at baseline, 2506 of whom made up the random subcohort.
Outcome measure Incident type 2 diabetes.
Results The review identified 16 studies containing 25 prediction models. We considered 12 models as basic because they were based on variables that can be assessed non-invasively and 13 models as extended because they additionally included conventional biomarkers such as glucose concentration. During a median follow-up of 10.2 years there were 924 cases in the full EPIC-NL cohort and 79 in the random subcohort. The C statistic for the basic models ranged from 0.74 (95% confidence interval 0.73 to 0.75) to 0.84 (0.82 to 0.85) for risk at 7.5 years. For prediction models including biomarkers the C statistic ranged from 0.81 (0.80 to 0.83) to 0.93 (0.92 to 0.94). Most prediction models overestimated the observed risk of diabetes, particularly at higher observed risks. After adjustment for differences in incidence of diabetes, calibration improved considerably.
Conclusions Most basic prediction models can identify people at high risk of developing diabetes in a time frame of five to 10 years. Models including biomarkers classified cases slightly better than basic ones. Most models overestimated the actual risk of diabetes. Existing prediction models therefore perform well to identify those at high risk, but cannot sufficiently quantify actual risk of future diabetes.
PMCID: PMC3445426  PMID: 22990994
14.  Single portal pressure measurement predicts survival in cirrhotic patients with recent bleeding 
Gut  1999;44(2):264-269.
Background—Height of portal pressure correlates with severity of alcoholic cirrhosis. Portal pressure indices are not however used routinely as predictors of survival. 
Aims—To examine the clinical value of a single portal pressure measurement in predicting outcome in cirrhotic patients who have bled. 
Methods—A series of 105 cirrhotic patients who consecutively underwent hepatic venous pressure measurement were investigated. The main cause of cirrhosis was alcoholic (64.8%) and prior to admission all patients had bled from varices. 
Results—During the follow up period (median 566 days, range 10-2555), 33 patients died, and 54 developed variceal haemorrhage. Applying Cox regression analysis, hepatic venous pressure gradient, bilirubin, prothrombin time, ascites, and previous long term endoscopic treatment were the only statistically independent predictors of survival, irrespective of cirrhotic aetiology. The predictive value of the pressure gradient was much higher if the measurement was taken within the first or the second week from the bleeding and there was no association after 15 days. A hepatic venous pressure gradient of at least 16 mm Hg appeared to identify patients with a greatly increased risk of dying. 
Conclusions—Indirectly measured portal pressure is an independent predictor of survival in patients with both alcoholic and non-alcoholic cirrhosis. In patients with a previous variceal bleeding episode this predictive value seems to be better if the measurement is taken within the first two weeks from the bleeding episode. A greater use of this technique is recommended for the prognostic assessment and management of patients with chronic liver disease. 

Keywords: chronic liver disease; alcoholic cirrhosis; portal pressure
PMCID: PMC1727391  PMID: 9895388
15.  Validation of Pediatric Index of Mortality-2 Scoring System in a Single Pediatric Intensive Care Unit in Iran 
Iranian Journal of Pediatrics  2012;22(4):481-486.
A study to validate and calibrate Pediatric Index of Mortality-2 (PIM2) in children admitted to our pediatric intensive care unit (PICU).
This is a prospective cohort study performed in Bahrami Children's Hospital affiliated to Tehran University of Medical Sciences. We studied the patients admitted to PICU from May 2007 to November 2008. Clinical measures were identified upon arrival in PICU. We used PIM2 score and logistic regression analysis to compare expected mortality risk with observed mortality rate. Receiver operating characteristics (ROC) curve analysis was done and standardized mortality ratio was calculated. PIM2 Index assessment was performed by use of Hosmer and Lemeshow goodness-of-fit test.
240 patients were included in this study. The model fit was achieved adequately (P value=0.741). The area under the ROC curve was 0.795 (0.715-0.875 for 95% confidence interval) and standardized mortality ratio was 1.8 (1.28-2.465 for 95% confidence interval) High-risk group diagnosis with adjusted odds ratio (AOR)=14.75, pupil reaction to light (AOR=0.13) and duration of stay in PICU (AOR=1.03) had significant statistical association to pediatric mortality.
PIM2 is a good index for prediction of mortality in our pediatric intensive care unit. This study revealed that there is significant statistical association between the children mortality and the length of hspita;ization, pupillary light reflex and the risk level category on admission.
PMCID: PMC3533148  PMID: 23429969
Mortality; Pediatric intensive care unit; Pediatric Index of Mortality-2 (PIM2); Iran
16.  A comparison of Child-Pugh, APACHE II and APACHE III scoring systems in predicting hospital mortality of patients with liver cirrhosis 
The aim of this study was to assess the prognostic accuracy of Child-Pugh and APACHE II and III scoring systems in predicting short-term, hospital mortality of patients with liver cirrhosis.
200 admissions of 147 cirrhotic patients (44% viral-associated liver cirrhosis, 33% alcoholic, 18.5% cryptogenic, 4.5% both viral and alcoholic) were studied prospectively. Clinical and laboratory data conforming to the Child-Pugh, APACHE II and III scores were recorded on day 1 for all patients. Discrimination was evaluated using receiver operating characteristic (ROC) curves and area under a ROC curve (AUC). Calibration was estimated using the Hosmer-Lemeshow goodness-of-fit test.
Overall mortality was 11.5%. The mean Child-Pugh, APACHE II and III scores for survivors were found to be significantly lower than those of nonsurvivors. Discrimination was excellent for Child-Pugh (ROC AUC: 0.859) and APACHE III (ROC AUC: 0.816) scores, and acceptable for APACHE II score (ROC AUC: 0.759). Although the Hosmer-Lemeshow statistic revealed adequate goodness-of-fit for Child-Pugh score (P = 0.192), this was not the case for APACHE II and III scores (P = 0.004 and 0.003 respectively)
Our results indicate that, of the three models, Child-Pugh score had the least statistically significant discrepancy between predicted and observed mortality across the strata of increasing predicting mortality. This supports the hypothesis that APACHE scores do not work accurately outside ICU settings.
PMCID: PMC156886  PMID: 12735793
17.  Diagnosis based injury severity scaling: investigation of a method using Australian and New Zealand hospitalisations 
Injury Prevention  2004;10(6):379-383.
Objective: To assess the performance of the International Classification of Diseases (ICD) based injury severity score, ICISS, when applied to two versions of the 10th edition of ICD, ICD-10 and ICD-10-AM.
Design: ICISS was assessed on its ability to predict threat to life using logistic regression modelling. Models used ICISS and age as predictors and survival as the outcome.
Setting: Australia and New Zealand.
Patients or subjects: Hospitalisations with an ICD-10-AM principal diagnosis in the range S00–T89 from 1 July 1999 to 30 June 2001 (Australia) or 1 July 1999 to 31 December 2001 (New Zealand).
Interventions: None.
Main outcome measures: The models were assessed in terms of their discrimination, measured by the concordance score, and calibration, measured using calibration curves and the Hosmer-Lemeshow statistic.
Results: 523 633 Australian and 124 767 New Zealand hospitalisations were selected, including 7230 and 1565 deaths respectively. Discrimination was high in all the fitted models with concordance scores of 0.885 to 0.910. Calibration results were also promising with all calibration curves being close to linear, though ICISS appeared to underestimate mortality somewhat for cases with an ICISS score less than 0.6. Overall ICISS performed better when applied to the Australian than the New Zealand hospitalisations. Australian and New Zealand hospitalisations were very similar. ICISS was also only a little more successful when ICD-10-AM rather than mapped ICD-10 was used.
Conclusions: ICISS appears to be a reasonable way to estimate severity for databases using ICD-10 or ICD-10-AM. It is also likely to work well for other clinical variants of ICD-10.
PMCID: PMC1730146  PMID: 15583261
18.  Prediction of large esophageal varices in cirrhotic patients using classification and regression tree analysis 
Clinics  2011;66(1):119-124.
Recent guidelines recommend that all cirrhotic patients should undergo endoscopic screening for esophageal varices. That identifying cirrhotic patients with esophageal varices by noninvasive predictors would allow for the restriction of the performance of endoscopy to patients with a high risk of having varices. This study aimed to develop a decision model based on classification and regression tree analysis for the prediction of large esophageal varices in cirrhotic patients.
309 cirrhotic patients (training sample, 187 patients; test sample 122 patients) were included. Within the training sample, the classification and regression tree analysis was used to identify predictors and prediction model of large esophageal varices. The prediction model was then further evaluated in the test sample and different Child‐Pugh classes.
The prevalence of large esophageal varices in cirrhotic patients was 50.8%. A tree model that was consisted of spleen width, portal vein diameter and prothrombin time was developed by classification and regression tree analysis achieved a diagnostic accuracy of 84% for prediction of large esophageal varices. When reconstructed into two groups, the rate of varices was 83.2% for high‐risk group and 15.2% for low‐risk group. Accuracy of the tree model was maintained in the test sample and different Child‐Pugh classes.
A decision tree model that consists of spleen width, portal vein diameter and prothrombin time may be useful for prediction of large esophageal varices in cirrhotic patients.
PMCID: PMC3044565  PMID: 21437447
Classification and regression tree analysis; Tree model; Esophageal varices; Predictor; Cirrhosis
19.  Validation of the chronic disease score-infectious disease (CDS-ID) for the prediction of hospital-associated clostridium difficile infection (CDI) within a retrospective cohort 
BMC Infectious Diseases  2013;13:150.
Aggregate comorbidity scores are useful for summarizing risk and confounder control in studies of hospital-associated infections. The Chronic Disease Score – Infectious Diseases (CDS-ID) was developed for this purpose, but it has not been validated for use in studies of Clostridium difficile Infection (CDI). The aim of this study was to assess the discrimination, calibration and potential for confounder control of CDS-ID compared to age alone or individual comorbid conditions.
Secondary analysis of a retrospective cohort study of adult inpatients with 2 or more days of antibiotic exposure at a tertiary care facility during 2005. Logistic regression models were used to predict the development of CDI up to 60 days post-discharge. Model discrimination and calibration were assessed using the c-statistic and Hosmer-Lemeshow (HL) tests, respectively. C-statistics were compared using chi-square tests.
CDI developed in 185 out of 7,792 patients. The CDS-ID was a better standalone predictor of CDI than age (c-statistic 0.653 vs 0.609, P=0.04). The best discrimination was observed when CDS-ID and age were both used to predict CDI (c-statistic 0.680). All models had acceptable calibration (P>0.05).
The CDS-ID is a valid tool for summarizing risk of CDI associated with comorbid conditions.
PMCID: PMC3614868  PMID: 23530876
Comorbidity score; Clostridium difficile infection; Validation; Chronic disease score; Infectious disease (CDS-ID)
20.  Comparing the predictive validity of three contemporary bleeding risk scores in acute coronary syndrome 
Haemorrhagic complications are strongly linked with adverse outcomes in acute coronary syndrome (ACS) patients. Various risk scores (RS) are available to predict bleeding risk in these patients. We compared the performance of three contemporary bleeding RS in ACS.
We studied 4500 consecutive patients with ACS. We calculated the ACTION, CRUSADE, and Mehran et al. (2010) bleeding RS, and evaluated their performance for predicting their own major bleeding events and TIMI serious (major or minor) bleeding episodes, in patients with either non-ST-elevation ACS (NSTEACS) or ST-elevation myocardial infarction (STEMI). Calibration (Hosmer–Lemeshow test, HL) and discrimination (c-statistic) for the three RS were computed and compared.
For RS-specific major bleeding, ACTION and CRUSADE showed the best prognostic discrimination in STEMI (c=0.734 and 0.791, respectively; p=0.04), and in NSTEACS (c=0.791 and 0.810; p=0.4); being CRUSADE significantly superior to Mehran et al. in both ACS types (p<0.05). All RS performed well in patients undergoing coronary arteriography using either a radial or femoral approach (all c≥0.718); however, their discriminative capacity was modest in patients not undergoing coronary arteriography and in those previously on oral anticoagulant (all c<0.70). For TIMI serious bleeding, ACTION and CRUSADE displayed the highest c-index values in both STEMI (0.724 and 0.703, respectively; p=0.3) and NSTEACS (c=0.733 and 0.744, respectively; p=0.6); however, calibration of ACTION was poor in both ACS types (HL p<0.05).
Of contemporary bleeding RS, the CRUSADE score was found to be the most accurate quantitative tool for NSTEACS and STEMI patients undergoing coronary arteriography.
PMCID: PMC3760544  PMID: 24062910
Acute coronary syndrome; haemorrhage; risk score
21.  Calibrating predictive model estimates to support personalized medicine 
Predictive models that generate individualized estimates for medically relevant outcomes are playing increasing roles in clinical care and translational research. However, current methods for calibrating these estimates lose valuable information. Our goal is to develop a new calibration method to conserve as much information as possible, and would compare favorably to existing methods in terms of important performance measures: discrimination and calibration.
Material and methods
We propose an adaptive technique that utilizes individualized confidence intervals (CIs) to calibrate predictions. We evaluate this new method, adaptive calibration of predictions (ACP), in artificial and real-world medical classification problems, in terms of areas under the ROC curves, the Hosmer-Lemeshow goodness-of-fit test, mean squared error, and computational complexity.
ACP compared favorably to other calibration methods such as binning, Platt scaling, and isotonic regression. In several experiments, binning, isotonic regression, and Platt scaling failed to improve the calibration of a logistic regression model, whereas ACP consistently improved the calibration while maintaining the same discrimination or even improving it in some experiments. In addition, the ACP algorithm is not computationally expensive.
The calculation of CIs for individual predictions may be cumbersome for certain predictive models. ACP is not completely parameter-free: the length of the CI employed may affect its results.
ACP can generate estimates that may be more suitable for individualized predictions than estimates that are calibrated using existing methods. Further studies are necessary to explore the limitations of ACP.
PMCID: PMC3277613  PMID: 21984587
22.  Balloon-occluded Retrograde Transvenous Obliteration (BRTO): Preprocedural Evaluation and Imaging 
Patients undergoing balloon retrograde transvenous obliteration (BRTO) are mostly decompensated cirrhotic with either bleeding gastric varices (GV) or hepatic encephalopathy. It is crucial that clinicians are up-to-date with the assessments needed prior to BRTO to anticipate and prevent complications, and to deliver critical quality care. These patients will require preprocedural assessments and management, including endoscopic, clinical, laboratory, and imaging evaluation. Endoscopic evaluation is mandatory prior to BRTO, and it is highly recommended that it be performed at the same institution where BRTO will be performed. It is essential that clinicians are aware of the potential benefits and complications that may result from BRTO. These complications should be anticipated and prevented when possible. For GV bleeders, there should be consideration of a transvenous intrahepatic portosystemic shunt (TIPS) during or before BRTO in patients with refractory ascites or pleural effusion, as well as endoscopic banding or a TIPS in patients with high-risk esophageal varices. Patients undergoing BRTO are usually complicated and require a team approach. In this article, the authors address these assessment and preparatory management and planning procedures prior to the BRTO procedure as well as expected outcomes and potential complications.
PMCID: PMC3312157  PMID: 22942546
Gastric varices; portal hypertension; liver cirrhosis; BRTO; TIPS; splenorenal shunt
23.  Clinical outcome and predictors of survival after TIPS insertion in patients with liver cirrhosis 
AIM: To determine the clinical outcome and predictors of survival after transjugular intrahepatic portosystemic stent shunt (TIPS) implantation in cirrhotic patients.
METHODS: Eighty-one patients with liver cirrhosis and consequential portal hypertension had TIPS implantation (bare metal) for either refractory ascites (RA) (n = 27) or variceal bleeding (VB) (n = 54). Endpoints for the study were: technical success, stent occlusion and stent stenosis, rebleeding, RA and mortality. Clinical records of patients were collected and analysed. Baseline characteristics [e.g., age, sex, CHILD score and the model for end-stage liver disease score (MELD score), underlying disease] were retrieved. The Kaplan-Meier method was employed to calculate survival from the time of TIPS implantation and comparisons were made by log rank test. A multivariate analysis of factors influencing survival was carried out using the Cox proportional hazards regression model. Results were expressed as medians and ranges. Comparisons between groups were performed by using the Mann-Whitney U-test and the χ2 test as appropriate.
RESULTS: No difference could be seen in terms of age, sex, underlying disease or degree of portal pressure gradient (PPG) reduction between the ascites and the bleeding group. The PPG significantly decreased from 23.4 ± 5.3 mmHg (VB) vs 22.1 ± 5.5 mmHg (RA) before TIPS to 11.8 ± 4.0 vs 11.7 ± 4.2 after TIPS implantation (P = 0.001 within each group). There was a tendency towards more patients with stage CHILD A in the bleeding group compared to the ascites group (24 vs 6, P = 0.052). The median survival for the ascites group was 29 mo compared to > 60 mo for the bleeding group (P = 0.009). The number of radiological controls for stent patency was 6.3 for bleeders and 3.8 for ascites patients (P = 0.029). Kaplan-Meier calculation indicated that stent occlusion at first control (P = 0.027), ascites prior to TIPS implantation (P = 0.009), CHILD stage (P = 0.013), MELD score (P = 0.001) and those patients not having undergone liver transplantation (P = 0.024) were significant predictors of survival. In the Cox regression model, stent occlusion (P = 0.022), RA (P = 0.043), CHILD stage (P = 0.015) and MELD score (P = 0.004) turned out to be independent prognostic factors of survival. The anticoagulation management (P = 0.097), the porto-systemic pressure gradient (P = 0.460) and rebleeding episodes (P = 0.765) had no significant effect on the overall survival.
CONCLUSION: RA, stent occlusion, initial CHILD stage and MELD score are independent predictors of survival in patients with TIPS, speaking for a close follow-up in these circumstances.
PMCID: PMC3468853  PMID: 23066315
Transjugular intrahepatic portosystemic stent shunt; Liver cirrhosis; Ascites; Gastrointestinal hemorrhage; Treatment outcome
24.  Liver Stiffness Measurement by Fibroscan Predicts the Presence and Size of Esophageal Varices in Egyptian Patients with HCV Related Liver Cirrhosis 
Background and Aim: Liver stiffness measured by transient elastgraphy correlates with Hepatic vein pressure gradient, liver Stiffness value of 21 kpa predicts significant portal hypertension. Aim is to predict esophageal varices presence by fibroscan and possible grading by degree of liver stiffness in HCV related cirrhotic patients.
Material and Methods: Thirty two HCV related cirrhotic patients were recruited, age > 18 years, BMI< 35, no history of: upper GI bleeding, hepatocellular carcinoma, abdominal collaterals, ascites. Patients underwent clinical examination, laboratory investigations, abdominal ultrasonography, upper endoscopy and fibroscan. They divided into (Group I= no varices, Group II =small varices (Grade 1 & 2), Group III = large varices (Grade 3 & 4).
Results: Age is higher in Group III than I & II (55+6.6 vs 49.5+4.7 & 48.9+4.7, p-value 0.04) respectively, Groups were gender & BMI matched, fibroscan values in Group I vs II & III were 27 Vs 49.4, p value 0.01, cutoff 29.7 Kpa (sensitivity 95% & specificity 67%) while its value in Group II vs III were 38.4 vs 60.4, p value 0.002, cutoff 38.2 Kpa (sensitivity 100% & specificity77.3%). Platelet count, splenic size, platelet count/splenic size in Group I vs II & III were 107.166 vs 72.900, 13.8 vs 15.4, 803.6 vs 478, p value 0.01, 0.008, 0.005, cutoff 80.000, 14.5, 545, sensitivity & specificity (85%&75%, 75%&75%, 85%&84%) respectively. On multivariate analysis fibroscan (OR 1.113; p=0.005) & platelet count/splenic size (OR 0.995; p=0.012) were positive predictors of esophageal varices presence.
Conclusion: Fibroscan is a good non-invasive method to predict esophageal varices presence & possible grading with high sensitivity.
PMCID: PMC3843460  PMID: 24298490
Fibroscan; Esophageal varices; Grading; Non-invasive methods
25.  Mortality from isolated coronary bypass surgery: a comparison of the Society of Thoracic Surgeons and the EuroSCORE risk prediction algorithms 
We compared the performances of the additive European System for Cardiac Operative Risk Evaluation, EuroSCORE (AES) and logistic EuroSCORE (LES) with the Society of Thoracic Surgeons' risk prediction algorithm in terms of discrimination and calibration in predicting mortality in patients undergoing isolated coronary artery bypass grafting (CABG) at a single institution in Pakistan. Both models were applied to 380 patients, operated upon at the Aga Khan University Hospital from August 2009 to July 2010. The actual mortality was 2.89%. The mean AES of all patients was 4.36 ± 3.58%, the mean LES was 5.96 ± 9.18% and the mean Society of Thoracic Surgeons' (STS) score was 2.30 ± 4.16%. The Hosmer–Lemeshow goodness-of-fit test gave a P-value of 0.801 for AES, 0.699 for LES and 0.981 for STS. The area under the receiver operating characteristic curve was 0.866 for AES, 0.842 for LES and 0.899 for STS. STS outperformed AES and LES both in terms of calibration and discrimination. STS, however, underestimated mortality in the top 20% of patients having an STS score >2.88, thus overall STS estimates were lower than actual mortality. We conclude that STS is a more accurate model for risk assessment as compared to additive and logistic EuroSCORE models in the Pakistani population.
PMCID: PMC3290369  PMID: 22184465
European System for Cardiac Operative Risk Evaluation; Society of Thoracic Surgeons; Coronary artery bypass grafting; Risk stratification; Pakistan

Results 1-25 (762736)