Include or not to include conference abstracts in systematic reviews? Lessons learned from a large Cochrane network meta-analysis including 585 trials

Table 3 Sensitivity NMA analyses of NMA results primary and secondary outcomes including eligible abstracts

Outcome	Number of studies (new studies)§	Number of treatments (new treatments)	Number of pairwise comparisons (new comparisons)	Ranking^a treatment: new rank (+/−number of rank)	RR (95% CI)^b
Outcome	Number of studies (new studies)§	Number of treatments (new treatments)	Number of pairwise comparisons (new comparisons)	Ranking^a treatment: new rank (+/−number of rank)	Without abstracts	With abstracts
Vomiting 24 h	284 (+2) Ref. [5, 13]	65 (+0)	576 (+4)	NA	Palo RR 0.62 (0.48 to 0.80)	Palo RR 0.63 (0.50 to 0.79)
Headache	210 (+2) Ref. [5, 13]	53 (+0)	430 (+4)	Dexa-meto-onda: 16 (−8) Meto-onda: 30 (−7)	Meto-onda RR 0.77 (0.18 to 3.19) Palo RR 1.15 (0.89 to 1.49)	Meto-onda RR 0.94 (0.29 to 3.03) Palo RR 1.16 (0.91 to 1.49)
QT prolongation	19 (+1) Ref. [13]	13 (+0)	38 (+1)	NA	NA	NA
Vomiting ‘early’	265 (+2) Ref. [7, 12]	70 (+1)	575 (+11)	Perp: 31 (−12) Drop-meto: 42 (+10)	NA	NA
Vomiting ‘late’	207 (+1) Ref. [7]	52 (+1)	439 (+1)	NA	NA	NA
Nausea	327 (+3) Ref. [2, 5, 9]	65 (+0)	680 (+10)	Dime: 38 (+6)	NA	NA
Complete response	138 (+2) Ref. [9, 13]	38 (+0)	285 (+7)	NA	NA	NA

§References from Additional file 2, eligible abstracts. ^aRanking: we named substantial changes > 5 only. ^bChanges in the interpretation of effects according to the range of equivalence. For vomiting, nausea, and complete response, the range of equivalence is RR 0.8 to 1.25; for headache and QT prolongation, it is RR 0.9 to 1.11. CI confidence interval, NA not applicable, Ref. reference, RR risk ratio, dexa dexamethasone, dime dimenhydrinate, drop droperidol, meto metoclopramide, onda ondansetron, palo palonosetron, perp, perphenazine

ISSN: 2046-4053