Skip to main content

Table 3 Measures of goodness of fit of fixed and random effects models for each of the five network meta-analyses

From: Patient characteristics as effect modifiers for psoriasis biologic treatment response: an assessment using network meta-analysis subgroups

Measure of goodness of fitRandom effects (uniform prior)Random effects (log-normal prior)Fixed effects
Licensed doses network
Residual deviancea162.78177.54209.77
pD117.98106.2991.61
Deviance information criterion (DIC)280.76283.83301.38
Between-study standard deviation, posterior median (95% credible interval)0.31 (0.17–0.45)0.19 (0.12–0.28)
Network of patients with no previous biologic use (< 25% had previous use)
Residual devianceb82.1082.8888.85
pD59.1256.352.45
Deviance information criterion (DIC)141.22139.20141.30
Between-study standard deviation, posterior median (95% credible interval)0.19 (0.01–0.41)0.14 (0.09–0.23)
Network of patients with PASI score ≤ 25
Residual deviancec143.89152.67173.06
pD99.1690.5879.61
Deviance information criterion (DIC)243.05243.26252.67
Between-study standard deviation, posterior median (95% credible interval)0.2574 (0.114–0.408)0.16 (0.10–0.24)
Network of patients with weight ≤ 90 kg
Residual devianced66.4074.1780.02
pD51.5944.7842.14
Deviance information criterion (DIC)117.99118.95122.16
Between-study standard deviation, posterior median (95% credible interval)0.40 (0.08–0.76)0.15 (0.09–0.24)
Network of ≥ 90% white patients
Residual deviancee100.57112.47126.65
pD78.5771.6263.83
Deviance information criterion (DIC)179.14184.09190.48
Between-study standard deviation, posterior median (95% credible interval)0.311 (0.13–0.51)0.17 (0.10–0.25)
  1. a165 unconstrained data points, pD number of parameters for licensed doses network
  2. b80 unconstrained data points, pD number of parameters
  3. c143 unconstrained data points, pD number of parameters
  4. d65 unconstrained data points, pD number of parameters
  5. e103 unconstrained data points, pD number of parameters