Skip to main content

Table 2 Predictive validity for indirect standardization, logistic regression and multilevel random intercept approach respectively using the mean Brier score and AUC for assessing the predictive abilities of the respective model using the validation set

From: Comparison of three different methods for risk adjustment in neonatal medicine

 

Indirect Standardization

Logistic regression

Multilevel

Variable

Brier Score

AUC

Brier Score

AUC

Brier Score

AUC

Mortality

0.070

0.896

0.113

0.935

0.109

0.936

In hosp. mortality

0.060

0.831

0.095

0.892

0.093

0.893

Late onset sepsis

0.068

0.788

0.067

0.804

0.067

0.803

NECa

0.023

0.716

0.024

0.694

0.024

0.694

sIVHb

0.050

0.765

0.049

0.786

0.049

0.786

BPDc

0.084

0.807

0.195

0.843

0.195

0.843

sROPd

0.015

0.765

0.016

0.773

0.016

0.773

  1. a NEC Necrotizing enterocolitis. Values based on imputed data
  2. b sIVH intra−/periventricular haemorrhage grade 3–4
  3. c BPD bronchopulmonary dysplasia
  4. d sROP retinopathy of prematurity grade 3 and above. Values based on imputed data