Skip to main content

Table 6 Estimates of regression parameters and their standard error obtained from MLE and Firth’s procedure by fitting Weibull AFT model for prostate cancer data under separation and near-to-separation

From: On estimation for accelerated failure time models with small or rare event survival data

 

Separation

Near-to-Separation

 

MLE

Firth

MLE

Firth

Predictors

Estimates

SE

Estimates

SE

Estimates

SE

Estimates

SE

Treatment

11.842

9,561.118

1.129

0.409

1.146

0.600

0.846

0.271

Age

11.475

0.00

1.013

0.428

0.380

0.804

0.132

0.565

WT

0.309

0.502

− 0.112

0.199

− 1.316

0.597

− 0.598

0.213

PF

-0.895

0.888

-0.981

0.388

-1.138

1.417

-0.854

0.720

HX

0.534

0.615

0.503

0.251

1.547

1.057

0.905

0.616

HG

-1.141

0.731

-1.063

0.313

13.101

7,825.785

1.587

0.684

SZ

-0.537

0.778

-0.054

0.259

0.506

0.918

0.389

0.591

SG

-1.966

0.872

-0.321

0.092

-2.071

0.743

-0.345

0.080

Intercept

5.218

0.728

7.671

0.181

5.924

1.006

8.171

0.252

scale (b)

0.518

0.182

0.398

0.096

0.694

0.219

0.608

0.145

  1. WT = weight index, PF= performance rating, HX= history of cardiovascular disease, HG= serum haemoglobin, SZ= size of primary lesion, SG= Gleason stage/grade category