Table 3.

Results of the Bayes factor analysis for the primary composite score (moderate to severe adverse neuropsychiatric treatment‐emergent events).

	Risk difference from published study	SE of risk difference from published study	P value from published study	Bayes factor	Interpretation of Bayes factor	Risk difference robustness region a
Primary composite endpoint
Overall
Varenicline versus placebo	0.0029	0.0060	0.626	0.07	Strong evidence for the null	[0.015, ∞]
Bupropion versus placebo	0.0081	0.0062	0.193	0.23	Moderate evidence for the null	[0.032, ∞]
Nicotine patch versus placebo	0.0018	0.0060	0.760	0.06	Strong evidence for the null	[0.013, ∞]
Non‐psychiatric cohort
Varenicline versus placebo	−0.0109	0.0060	0.072	0.08	Strong evidence for the null	[0.004, ∞]
Bupropion versus placebo	−0.0018	0.0067	0.792	0.08	Strong evidence for the null	[0.010, ∞]
Nicotine patch versus placebo	0.0008	0.0069	0.905	0.11	Strong evidence for the null	[0.013, ∞]
Psychiatric cohort
Varenicline versus placebo	0.0160	0.0103	0.119	0.52	Insensitive/anecdotal evidence for the null	[0.00, 0.067]
Bupropion versus placebo	0.0176	0.0104	0.090	0.71	Insensitive/anecdotal evidence for the null	[0.00, 0.078]
Nicotine patch versus placebo	0.0029	0.0097	0.766	0.07	Strong evidence for the null	[0.021, ∞]

^a Indicates the range of expected effect sizes that lead to the same qualitative conclusion (i.e. evidence for the null [Bayes factor <1/3] or data insensitivity [Bayes factor between 1/3 and 3]).