Skip to main content


Fig. 4 | BMC Medicine

Fig. 4

From: Risk of serious infections with immunosuppressive drugs and glucocorticoids for lupus nephritis: a systematic review and network meta-analysis

Fig. 4

The figure shows the deviances from consistency and the inconsistency models. Consistency model assumes that the evidence derived from direct and indirect estimates should be in agreement. The inconsistency model does not make this assumption. Ideally, all deviances should be 2 or below. There were 71 unconstrained data points. Three data points show deviances of > 2 (extreme right side in the figure). Residual deviances from consistency versus inconsistency models were 72.49 versus 67.73 for 69 data points each. Deviance information criteria was 325.65 versus 325.96, respectively, indicating that the inconsistency model is a slightly better fit for the data than the consistency model. Thus, the model showed some evidence of inconsistency

Back to article page