Skip to main content

Table 3 Comparison between estimates of σsjt (psjt for binary outcomes) obtained with the B-spline, P-spline and Fractional Polynomial (FP) network meta-analysis models. We report median, minimum and maximum evaluated over the 50 replicates of the average σsjt (respectively psjt) for each scenario

From: Bayesian splines versus fractional polynomials in network meta-analysis

Scenario

B-spline model

FP model

 

Median

Min

Max

Median

Min

Max

Linear

0.1007

0.1003

0.1010

0.1423

0.1172

0.1867

Logarithmic

0.1016

0.1014

0.1018

0.1439

0.1258

0.1761

Piecewise linear monotonic

0.1300

0.1190

0.1468

0.5701

0.5552

0.5875

Mixed

0.1321

0.1200

0.1497

0.6794

0.6666

0.6850

Non-monotonic

0.1399

0.1240

0.1760

0.9321

0.9321

0.9322

MTC

0.0829

0.0746

0.0939

0.1831

0.1236

0.3685

BEST-ITP

0.1352

0.1262

0.1426

0.2507

0.1712

0.4110

Piecewise linear monotonic (binary)

0.6128

0.5923

0.6246

0.6989

0.6830

0.7037

Non-monotonic (binary)

0.5940

0.5763

0.6011

0.7732

0.7372

0.8189

Piecewise linear (non-closed network)

0.1483

0.1203

0.1749

1.0382

1.0365

1.0408