Table 4 Results of the two applications.

From: G-computation, propensity score-based methods, and targeted maximum likelihood estimator for causal inference with different covariates sets: a comparative simulation study

application

method

set

\({\hat{\pi }}_{0}\)

\({\hat{\pi }}_{1}\)

\(\hat{\theta }\)

SE

95% CI

A - Multiple sclerosis

GC

outcome

20.3

28.2

0.432

0.189

[0.062, 0.802]

treatment*

20.3

28.3

0.436

0.195

[0.054, 0.819]

common*

20.3

28.3

0.436

0.195

[0.054, 0.819]

entire

20.3

28.2

0.431

0.191

[0.056, 0.806]

IPTW

outcome

21.2

28.8

0.406

0.195

[0.023, 0.789]

treatment*

20.3

28.2

0.433

0.191

[0.059, 0.808]

common*

20.3

28.2

0.433

0.191

[0.059, 0.808]

entire

21.3

28.9

0.406

0.196

[0.022, 0.791]

TMLE

outcome

21.2

28.8

0.407

0.195

[0.025, 0.790]

treatment*

20.3

28.2

0.433

0.190

[0.061, 0.806]

common*

20.3

28.2

0.433

0.190

[0.061, 0.806]

entire

21.1

28.9

0.410

0.196

[0.026, 0.794]

FM

outcome

19.1

29.0

0.549

0.189

[0.178, 0.921]

treatment*

19.9

30.6

0.575

0.187

[0.210, 0.941]

common*

19.9

30.6

0.575

0.187

[0.210, 0.941]

entire

21.1

31.9

0.561

0.183

[0.201, 0.920]

B - ICU

GC

outcome

66.3

81.1

0.778

0.294

[0.201, 1.354]

treatment

65.3

81.1

0.824

0.298

[0.240, 1.407]

common

65.0

81.1

0.836

0.289

[0.270, 1.402]

entire

66.5

81.1

0.769

0.295

[0.191, 1.347]

IPTW

outcome

31.0

81.1

0.656

0.356

[−0.042, 1.354]

treatment

68.2

81.1

0.693

0.355

[−0.002, 1.388]

common

67.4

81.1

0.729

0.353

[0.038, 1.421]

entire

69.2

81.1

0.645

0.362

[−0.064, 1.354]

TMLE

outcome

66.2

79.6

0.692

0.293

[0.118, 1.266]

treatment

65.4

80.2

0.758

0.288

[0.194, 1.322]

common

64.8

79.9

0.769

0.298

[0.185, 1.354]

entire

66.4

79.4

0.668

0.285

[0.109, 1.228]

FM

outcome

73.8

81.1

0.419

0.342

[−0.252, 1.090]

treatment

67.2

81.1

0.739

0.337

[0.078, 1.399]

common

65.1

81.1

0.831

0.336

[0.173, 1.490]

entire

66.2

81.1

0.782

0.336

[0.123, 1.442]

  1. *Treatment and common sets contain same covariates. π0: Percentage of event in the Natalizumab (or control) group, π1: Percentage of event in the Fingolimod (or Barbiturates) group, SE: standard error.