Diversity when interpreting evidence in network meta-analyses (NMAs) on similar topics: an example case of NMAs on diabetic macular oedema

Table 3 Summary of three key outcomes on aflibercept versus ranibizumab in included NMAs

	Régnier 2014 [14]	Korobelnik 2015 [13]	Zhang 2016 [15]	Muston 2018 [16]	Virgili 2018 [17]
Gain ≥ 10 ETDRS letters at 12 months (three NMAs^a)
OR [95% CrI]	0.63 [0.19 to 1.63]	1.59 [0.75 to 3.35]	NR	1.79 [0.63 to 4.06]	NR
Trials reporting these data^b in each NMA
Elman 2010 [21] [DRCR.net Protocol I]	Included	Included	NR	Included	NR
Mitchell 2011 [22] [RESTORE]	Included	Included	NR	Included	NR
Korobelnik 2014 [23] [VIVID; VISTA]	Included	Included	NR	Included	NR
Massin 2010 [19] [RESOLVE]	Included	Not included^c	NR	Not included^c	NR
Googe 2011 [24] [DRCR.net Protocol J]	Not included^d	Included	NR	Included	NR
Do DV 2012 [25] [Da VINCI]	Included	Not included^d	NR	Not included^d	NR
Ishibashi 2015 [18] [REVEAL]	Not included focus on Asian population^e	Included^e	NR	Included	NR
RESPOND [26] [NCT01135914]	Included	Not included^f	NR	Included	NR
Nguyen 2009 [20] [READ-2]	Included^e	Not included^g	NR	Not included^g	NR
Average change in BCVA^h at 12 months MD [95% CrI] (five NMAs)
	*4.5* [1.5 to 7]ⁱ	4.67 [2.45 to 6.87]	2.07 [− 0.97 to 5.33]	5.20 [1.90 to 8.52]	4 [2.5 to 5.5]
Gain ETDRS letters^h at 12 months OR [95% CrI] (five NMAs)
≥ 10	*0.63* [0.19–1.63]	1.59 [0.75–3.35]	NR	1.79 [0.63–4.06]	NR
≥ 15	NR	NR	NR	2.30 [1.12–4.20]	1.33 [1.06–1.67]^j

CrI, credible interval; BCVA, best-corrected visual acuity; ETDRS, Early Treatment Diabetic Retinopathy Study; NMA, network meta-analysis; NR, not reported; OR, odds ratio
^aZhang (2016) and Virgili et al. (2018) did not report this outcome
^bThe additional reasons presented for ‘included’ or ‘not included’ were identified by the author team of this review, not identified in the original texts of NMAs
^cData unavailable on ranibizumab 0.5 mg
^dUnclear reason for exclusion
^eIncluded in the sensitivity analysis
^fUnpublished when NMA conducted
^gData only reported at 6 months
^hHigher values represent better visual acuity measured using ETDRS letters
ⁱData were analysed by the author team of this review (Bayesian network model/random effects, using the ADDIS software), not reported in the original texts of Re´gnier (2014)
^jData were risk ratio (RR) and its 95% CrI

ISSN: 2046-4053