Classifying information-sharing methods

Table 1 A categorisation of papers that shared information according to the ‘core’ relationship that they used and the PICOS dimension that direct and indirect evidence differ in

PICOS dimension of indirectness	‘Core‘ relationship
	Functional	Exchangeability based	Prior-based	Multivariate
Intervention	Lumping: [30, 32–43]	RE: [34, 38–42]	SIP: No refs	B: [36]
	C: [35, 39, 40, 44]	RW: [35]	MixP: No refs	W: No refs
	L: [30, 35–38, 42, 45–48]	MLM: [32, 33, 35, 44, 46, 49, 50]	PP: No refs	BW: No refs
	N-L: [35, 51–53]			S: No refs
Population	Lumping: [14, 33]	RE: [34, 54]	SIP: [55]	B: No refs
	C: No refs	RW: No refs	MixP: [56]	W: No refs
	L: [33]	MLM: [55]	PP: [55]	BW: No refs
	N-L: No refs			S: No refs
Outcomes	Lumping: [32]	RE: No refs	SIP: No refs	B: [30, 31, 57–60]
	C: [57]	RW: [45, 61]	MixP: No refs	W: [62]
	L: [31]	MLM: [30]	PP: No refs	BW: [26, 63–78]
	N-L: [61, 79]			S: [80–84]
Designs	Lumping: No refs	RE: No refs	SIP: [85–90]	B: No refs
	C: No refs	RW: No refs	MixP: No refs	W: No refs
	L: [42, 50, 90–99]	MLM: [85, 86, 88, 100]	PP: [101]	BW: No refs
	N-L: No refs			S: No refs

The ‘PICOS dimension of indirectness’ denotes the PICOS part (i.e. Population, Intervention etc.) on which the direct evidence differ from the indirect in terms of the research question they address.
C: Constraint, L: Linear relationship (e.g. meta-regression), N-L: Non-linear, RE: Random-Effect, RW: Random-Walk, MLM: Multi-level model, SIP: Standard Informative Prior, MixP: Mixture prior, PP: Power-prior, B: Only between-studies correlation modelled, W: Only within-study correlation modelled, B&W: Both within-study and between-studies correlations modelled separately, S: Within-study and between-studies correlations modelled simultaneously as one parameter

ISSN: 1471-2288