Trial-level characteristics associate with treatment effect estimates: a systematic review of meta-epidemiological studies

Table 2 Characteristics of 63 meta-epidemiological (ME) studies based on collection of meta-analyses^a

Characteristics	No. of ME study^b	% (95% CI)
Data sources^c
Cochrane review only	28/58	48 (35 to 62)
Non-Cochrane review only	3/58	5 (0 to 11)
Non-Cochrane review published in “high impact factor” journals	1/58	2 (0 to 5)
Both Cochrane and non-Cochrane review	27/58	47 (33 to 60)
Non-Cochrane review published in “high impact factor” journals and Cochrane	8/58	14 (5 to 23)
Type of meta-analyses
Aggregated data only	58	92 (85 to 99)
Aggregated and individual participant data	3	5 (0 to 10)
Individual participant data only	1	2 (0 to 5)
Network of aggregated data only	1	2 (0 to 5)
Management of overlapping meta-analyses	35	56 (43 to 68)
Reported minimum No. of trials included in each meta-analysis	37	59 (46 to 71)
Minimum No. of trials included in meta-analysis, median (Q1, Q3)^d, range^e	3 (2,4)	1 to 10
Criteria of selecting one meta-analysis within each systematic review^f
Primary outcome	10/44	23 (10 to 36)
Largest number of studies	4/44	9 (0 to 18)
Objective	3/44	7 (0 to 15)
First outcome	2/44	4 (0 to 11)
Others^g	5/44	11 (2 to 21)
More than one method	20/44	45 (30 to 61)
Data extraction sources^c
From individual trial	19/60	32 (20 to 44)
From meta-analysis	8/60	13 (4 to 22)
Both	33/60	55 (42 to 68)

^aValues are numbers of ME studies, percentage (95% confidence interval) unless stated otherwise. ^bDenominators are 63, unless stated otherwise. ^cDenominators are not equal to 63 as some ME studies did not report relate information. ^dQ1, Quartile 1. ^eAmong 63 ME studies, 26 ME studies did not report minimum number of trials included in meta-analysis. ^fDenominator is 44, as four ME studies included all meta-analysis within each systematic review while 15 ME studies did not report related information. ^gOthers includes selection methods of first outcome statistically significant (1, 2.1%), mortality (1, 2.1%), most clinically relevant (1, 2.1%), most homogeneous (1, 2.1%) and at random (1, 2.1%)

ISSN: 1471-2288