Table 2.

NMA results including certainty assessments

Comparison	Number of trials with direct comparisons	Direct estimate (95% CI)	Indirect estimate (95% CrI)	NMA estimate (95% CrI) (higher of direct or indirect confidence)
Light starch vs saline	4	1.07 (0.89, 1.29) M1	0.59 (0.25, 1.35) VL1,2,3	1.04 (0.87, 1.25) M
Heavy starch vs saline	3	0.64 (0.30, 1.37) M1	1.13 (0.71, 1.80) VL1,2	0.95 (0.64, 1.41) M
Albumin vs saline	2	0.81 (0.64, 1.03) M1	0.96 (0.14, 6.31) VL2,4	0.82 (0.65, 1.04) M
Balanced crystalloid vs saline	0	–	0.78 (0.58, 1.05) L1,2	0.78 (0.58, 1.05) L
Gelatin vs saline	0	–	1.04 (0.46, 2.32) VL1,2	1.04 (0.46, 2.32) VL
Heavy starch vs light starch	0	–	0.91 (0.63, 1.33) L1,2	0.91 (0.63, 1.33) L
Albumin vs light starch	0	–	0.79 (0.59, 1.06) L1,2	0.79 (0.59, 1.06) L
Balanced crystalloid vs light starch	2	0.80 (0.61, 1.04) M3	0.44 (0.19, 0.97) M2	0.75 (0.58, 0.97) M
Gelatin vs light starch	0	–	1.00 (0.44, 2.21) VL1,2	1.00 (0.44, 2.21) VL
Albumin vs heavy starch	2	1.40 (0.35, 5.56) L4	0.83 (0.52, 1.33) L1,2	0.87 (0.55, 1.36) L
Balanced crystalloid vs heavy starch	1	0.74 (0.52, 1.05) M1	1.35 (0.63, 2.92) VL2,4	0.82 (0.60, 1.13) M
Gelatin vs heavy starch	1	1.09 (0.55, 2.19) L4	–	1.10 (0.54, 2.21) L
Balanced crystalloid vs albumin	0	–	0.95 (0.65, 1.38) VL1,2	0.95 (0.65, 1.38) VL
Gelatin vs albumin	0	–	1.26 (0.55, 2.90) VL2,4	1.26 (0.55, 2.90) VL
Gelatin vs balanced crystalloid	0	–	1.34 (0.61, 2.89) VL2,4	1.34 (0.61, 2.89) VL

“From Annals of Internal Medicine, Rochwerg B et al, Fluid Resuscitation in Sepsis: A systematic review and network meta-analysis, 161, 5, 347-55.”

CI confidence interval, CrI credible interval; QoE: H high, M moderate, L low, VL very low

¹—rated down for imprecision, ²—rated down for indirectness, ³—rated down for inconsistency (I ² = 80%, p = 0.03 for heterogeneity), ⁴—rated down 2 levels for imprecision