SAMURAI: Sensitivity analysis of a meta-analysis with unpublished but registered analytical investigations (software)

Table 4 Summary effects for the Hpylori dataset after all unpublished studies assigned the same default outlook

Scenario number	Outlook assigned to all 7 unpublished studies	Relative risk (RR) assigned to all unpublished studies	Summary effect (RR)	Lower 95% confidence limit	Upper 95% confidence limit	τ²**
1	very positive	0.33	0.518	0.432	0.621	<0.001
2	positive	0.5	0.561	0.471	0.669	<0.001
3	no effect	1	0.697	0.585	0.829	0.011
4	negative	2	0.801	0.609	1.052	0.317
4a*	negative	1.5	0.759	0.602	0.958	0.166
5	very negative	3	0.866	0.616	1.217	0.636
5a*	very negative	2.5	0.841	0.614	1.151	0.501
6	very positive CL	LCL _pub,1-α	0.546	0.457	0.652	<0.001
7	positive CL	$0.5 ({\hat{RR}}_{pub} + LC L_{pub, 1 - α})$	0.573	0.481	0.682	<0.001
8	current effect	${\hat{RR}}_{pub}$	0.591	0.497	0.703	<0.001
9	negative CL	$0.5 ({\hat{RR}}_{pub} + UC L_{pub, 1 - α})$	0.603	0.507	0.716	<0.001
10	very negative CL	UCL _pub,1-α	0.634	0.534	0.752	<0.001

(The random number seed used to generate this table was 106.)
*In these cases all unpublished studies were assigned an RR different from the default RR.
**A displayed value of ‘0’ for ‘tau2’ indicates a τ² value less than 0.001.

ISSN: 2046-4053