Skip to main content

Table 2 Comparison of Bayesian models; posterior distributions of odds ratios and 95% credible intervals (CrI)

From: Comparison of anti-malarial drug efficacy in the treatment of uncomplicated malaria in African children and adults using network meta-analysis

 

Basic parameters

Model 2

Model 3

Consistency

Inconsistency

Odds ratios

95% CrI

Odds ratios

95% CrI

AQSP

\( d_{12} \)

0.87

0.50–1.51

0.96

0.45–2.05

ASAQ

\( d_{13} \)

1.12

0.82–1.54

1.09

0.76–1.57

ASAQCPH

\( d_{14} \)

1.55

0.27–8.76

1.91

0.24–14.95

ASATPG*

\( d_{35} \)

3.82

0.49–29.9

3.39

0.41–28.13

ASCD

\( d_{16} \)

0.59

0.26–1.32

0.76

0.26–2.18

ASMQ

\( d_{17} \)

1.20

0.52–2.80

1.26

0.47–3.36

ASNAPH

\( d_{18} \)

0.85

0.06–11.78

0.82

0.06–11.77

ASPY

\( d_{19} \)

1.41

0.42–4.75

1.41

0.38–5.23

ASSMP

\( d_{1,10} \)

1.15

0.46–2.90

1.65

0.48–5.68

ASSP

\( d_{1,11} \)

1.36

0.73–2.55

0.99

0.25–3.89

DHAP**

\( d_{1, 12} \)

1.92

1.30–2.82

1.88

1.20–2.95

DHAPT

\( \begin{array}{*{20}c} { } \\ { d} \\ \end{array}_{1,13} \)

1.48

0.22–10.09

1.47

0.19–11.03

\( \tau^{2} \) (SD)

 

1.45 (0.30)

 

1.26 (0.27)

 

\( \sqrt {\tau^{2} } \) (SD)

 

0.84 (0.08)

 

0.90 (0.097)

 

\( \hat{D} \)

 

732

 

728

 

\( p_{D} \)

 

145

 

148

 

DIC

 

1024.07

 

1025.04

 
  1. Results are extracted from the random effects model under consistency and inconsistency assumptions. \(d_{1k}\) is the relative effect of treatment k compared to treatment 1, here AL
  2. SD estimate of the standard deviation of the parameter, σ estimate of the variability in treatment effect between trials within pairwise contrasts, τ estimate of the between study variation, \(\hat{D}\): deviance estimating likelihood, \(p_{D}\) number of parameters in the model, DIC deviance information criterion (when it is small, it gives the best model)
  3. aDHAP was superior to AL in all fitted models
  4. bASATPG was the only ACT compared directly to ASAQ; the difference of efficacy of these ACT medicines was not statistically significant