The Performance of Risk Prediction Models

Gerds, Thomas Alexander; Cai, Tianxi; Schumacher, Martin

doi:10.1002/bimj.200810443

Cited by 249 publications

(185 citation statements)

References 82 publications

Supporting

Mentioning

180

Contrasting

Unclassified

Order By: Relevance

“…Cox analysis was used to evaluate the ability of the SCS to predict survival in the validation cohort. The BS [12,13] was used to assess the accuracy of the SCS, French registry and PHC registry equations in predicting 1-and 2-yr survival in the validation cohort allowing for censoring. It measures the mean squared deviation of predicted probability from the actual outcome.…”

Section: Discussionmentioning

confidence: 99%

Predicting survival in pulmonary arterial hypertension in the UK

et al. 2012

View full text Add to dashboard Cite

Contemporary prognostic equations in pulmonary arterial hypertension (PAH) derived from US and French cohorts may not perform as well in the UK as a locally derived scoring scheme. The aim of the study was to develop and validate a UK risk score to predict prognosis in PAH.Baseline mortality predictors identified by multivariate Cox analysis in 182 incident PAH patients were used to derive the Scottish composite score (SCS). Its prognostic performance in an independent UK cohort was compared with the French registry and Pulmonary Hypertension Connection (PHC) registry equations using Brier scores (BS).The SCS based on age, sex, aetiology, right atrial pressure, cardiac output and 6-min walk distance predicted survival in the validation cohort (hazard ratio (HR) 1.7 per point increase; p,0.001) and provided further prognostic stratification in World Health Organization (WHO) functional class III patients (HR 1.8 per point increase; p,0.001). It was more accurate than the French registry equation in predicting 1-yr survival (BS: 0.092 versus 0.146; p50.001) and 2-yr survival (0.131 versus 0.255; p,0.001). There was no significant difference in BS between the SCS and PHC registry equation.The SCS predicts survival and can be used to supplement WHO functional class in prognostication.

show abstract

Section: Discussionmentioning

confidence: 99%

Predicting survival in pulmonary arterial hypertension in the UK

et al. 2012

View full text Add to dashboard Cite

show abstract

“…Brier score values range between 0 and 1, the smaller values indicating good performance of the risk prediction model at given time points (a useful risk prediction model should not have a value >0.25). Three prediction errors were compared, as suggested by Gerds et al 21 : (1) 0.632+ prediction error estimate, a weighted combination of the apparent error on the full data set and the bootstrap cross-validation error estimate with K=10 parts and B=100 bootstrap samples; (2) null model prediction error, an estimation of fit without the prognostic variables by using the Kaplan-Meier estimate; and (3) no-information error of the full data set, an evaluation of the prognostic variables in artificially permuted data where the recurrence response is independent of the predictors. Two-sided values of P<0.05 were considered significant.…”

Section: Statistical Analysesmentioning

confidence: 99%

Predictors of Long-Term Recurrent Vascular Events After Ischemic Stroke at Young Age

et al. 2014

View full text Add to dashboard Cite

Background-Data on long-term risk and predictors of recurrent thrombotic events after ischemic stroke at a young age are limited. Methods and Results-We followed 1867 patients with first-ever ischemic stroke who were 18 to 45 years of age (mean age, 36.8±7.1 years; women, 49.0%), as part of the Italian Project on Stroke in Young Adults (IPSYS). Median follow-up was 40 months (25th to 75th percentile, 53). The primary end point was a composite of ischemic stroke, transient ischemic attack, myocardial infarction, or other arterial events. One hundred sixty-three patients had recurrent thrombotic events (average rate, 2.26 per 100 person-years at risk). At 10 years, cumulative risk was 14.7% (95% confidence interval, 12.2%-17.9%) for primary end point, 14.0% (95% confidence interval, 11.4%-17.1%) for brain ischemia, and 0.7% (95% confidence interval, 0.4%-1.3%) for myocardial infarction or other arterial events. Familial history of stroke, migraine with aura, circulating antiphospholipid antibodies, discontinuation of antiplatelet and antihypertensive medications, and any increase of 1 traditional vascular risk factor were independent predictors of the composite end point in multivariable Cox proportional hazards analysis. A point-scoring system for each variable was generated by their β-coefficients, and a predictive score (IPSYS score) was calculated as the sum of the weighted scores. The area under the receiver operating characteristic curve of the 0-to 5-year score was 0.66 (95% confidence interval, 0.61-0.71; mean, 10-fold internally cross-validated area under the receiver operating characteristic curve, 0.65).© 2014 American Heart Association, Inc. 1 Although it is well documented that such a risk is much lower in young patients with stroke than in elderly patients, information on what specific factors may predict recurrent events in younger age groups are limited. Most data derive from single-center studies enrolling several hundred patients or less, 2 using different thresholds of age to define young, and sometimes being biased by the inadequate capture of cases, the inclusion of different ethnic groups, and the high number of patients lost to follow-up.3 This makes such studies somewhat heterogeneous and their findings poorly comparable. In addition, the influential effect of some specific factors is missing in most previous studies. This is the case, for example, of patients' adherence to secondary prevention therapies, which is likely to impact the recurrence of potentially avoidable vascular events. The Italian Project on Stroke in Young Adults (IPSYS) provides the opportunity to investigate these issues owing to its large sample size, the homogeneous demographic characteristics and clinical phenotype of the subjects included, and the standard diagnostic workup. Therefore, in the present study we aimed at (1) elucidating the predictors of long-term recurrent vascular events after first-ever IS, and the extent to which these factors can be modified, which implicates the potential of reducing this risk,...

show abstract

“…15 The most important property of a predictive model is discrimination, the ability of a model to distinguish patients who are likely to have clinical HIT from those who are not. 16 Model discrimination was assessed with the concordance probability, or c-index. 17 The c-index is equivalent to the area under the receiver operating characteristic curve 18 and is related to Somers' Dyx rank correlation [Dyx ϭ (c Ϫ 0.5)/0.5].…”

Section: Assessment Of Model Performancementioning

confidence: 99%