Skip to main content

Table 3 Monte Carlo Results: Estimation of recurrence of adenomas at three years (true recurrence rate: 0.495) under a situation of independent censoring with five auxiliary variables.

From: Estimation of colorectal adenoma recurrence with dependent censoring

Method

Esta

Bias

SDb

SEc

CRd

λ e= 0.12; = 0.30; early censoring rateg= 23.0%; |r h| = 0.35

NPMLE

0.501

0.006

0.0405

0.0404

93.8

KM

0.485

-0.010

0.0371

0.0381

94.2

Logit

0.417

-0.078

0.0339

0.0348

36.6

WLogit

0.490

-0.005

0.0375

0.0391

96.2

WKMr+c(4,1)

0.482

-0.013

0.0369

0.0375

94.4

WKMr(4)

0.480

-0.015

0.0366

0.0372

94.4

WKMc(4)

0.484

-0.011

0.0368

0.0380

94.6

λ = 0.17; p ec = 0.40; early censoring rate = 30.5%; |r| = 0.36

NPMLE

0.501

0.006

0.0429

0.0431

94.0

KM

0.481

-0.014

0.0383

0.0394

93.6

Logit

0.390

-0.105

0.0337

0.0344

13.0

WLogit

0.489

-0.006

0.0392

0.0407

94.0

WKMr+c(4, 1)

0.477

-0.018

0.0380

0.0385

92.6

WKMr(4)

0.474

-0.021

0.0377

0.0380

91.8

WKMc(4)

0.480

-0.015

0.0376

0.0392

94.0

λ = 0.23; p ec = 0.50; early censoring rate = 38.3%; |r| = 0.37

NPMLE

0.506

0.011

0.0460

0.0465

94.8

KM

0.480

-0.015

0.0403

0.0409

94.6

Logit

0.363

-0.132

0.0340

0.0339

2.4

WLogit

0.491

-0.004

0.0414

0.0426

95.6

WKMr+c(4,1)

0.474

-0.021

0.0395

0.0396

90.6

WKMr(4)

0.471

-0.024

0.0387

0.0391

89.8

WKMc(4)

0.478

-0.017

0.0392

0.0406

94.0

  1. aaverage of 500 estimated recurrence rates.
  2. bempirical standard deviation of 500 point estimates.
  3. caverage of 500 estimated standard errors.
  4. dfraction of 95% confidence intervals which contain the true value.
  5. ehazard rate for time to colonoscopy.
  6. fproportion of participants with early colonoscopy.
  7. gright censoring occurs before the end of the trial (3 years).
  8. hcorrelation coefficient between the two risk scores.