Quantitative bias analysis in practice: review of software for regression with unmeasured confounding

Kawabata, Emily; Tilling, Kate; Groenwold, Rolf  H. H.; Hughes, Rachael A.

doi:10.1186/s12874-023-01906-8

BMC Medical Research Methodology

Table 1 Software programs implementing a quantitative bias analysis to unmeasured confounding published 2011 to 2021

From: Quantitative bias analysis in practice: review of software for regression with unmeasured confounding

		Analysis of interest				Bias analysis
Name (Year created)	Environment	Type of analysis	Outcome	Exposure	Mediator	No. bias parameters	Benchmarking	Graphical plot	Tipping point
isa[37, 38] (2011)	Stata	simple^a	con^b	bin^c	−	2	yes	line	null effect, stat. sig.^d
gsa[11, 39] (2012)	Stata	simple	bin, con	bin, con, cat^e	−	2	yes	scatter	null effect, stat. sig.
SensitivityCase- Control[40, 41] (2012)	R	matched	bin	bin	−	1	no	no	none
causalsens[42, 43] (2013)	R	simple	con	bin	−	1	yes	line with CI^f	null effect, stat. sig.
mbsens[44] (2014)	Stata	matched	bin	bin	−	1	no	no	none
sensitivitymw[45,46,47] (2014)	R	matched	con, integer	bin	−	1	no	no	none
treatSens[48,49,50] (2014)	R	simple	con	bin, con	−	2 [\(+1\) for p(U)]^g	yes	contour	null effect, stat. sig.
sensitivitymv[45, 46, 51] (2015)	R	matched	con, integer	bin	−	1	no	no	none
EValue^h (2017) [52,53,54,55,56]	R, Stata, Web tool	simple, meta-analysis	bin, con, TTEⁱ	bin, con	−	2	no	line	null effect, stat. sig.
rmpw[57, 58] (2017)	R	mediation	bin, con, cat	bin	bin, con, cat	2 or 4	yes	contour	stat. sig.
sensitivityfull[45, 59, 60] (2017)	R	matched	con, integer	bin	−	1	no	no	no
submax[61] (2017)	R	matched	con, integer	bin	−	1	no	no	stat. sig.
Umediation[62] (2017)	R	mediation	bin, con	bin, con	bin, con	3 [\(+1\) or 2 for p(U)]	no	line	stat. sig.
konfound[63, 64] (2018)	R, Stata, Web tool	simple	bin, con	bin, con	−	2	yes	bar, causal diagram	stat. sig.
sensmediation[65, 66] (2018)	R	mediation	bin, con	bin, con	bin, con	1	no	line with CI	null effect, stat. sig.
sensitivityCalibration[32, 67] (2018)	R	matched	con	bin	−	3	yes	line	stat. sig.
sensemakr[68,69,70] (2019)	R, Stata, Web tool	simple	con	bin	−	2	yes	contour	null effect, stat. sig.
ui[71] (2019)	R	simple	bin	bin	−	2	no	line with CI and UI	null effect, stat. sig.
mediationsens[17, 72] (2020)	R	mediation	con, bin	bin	con, bin	2 or 3 [\(+1\) for p(U)]	yes	contour	null effect, stat. sig.
survsens[73, 74] (2020)	R	simple	TTE	bin	−	2 [\(+1\) for p(U)]	no	contour	null effect, stat. sig.
MediationSensitivity- Analysis[75] (2021)	Web tool	mediation	con	bin, con, cat	con	2	no	contour	null effect, stat. sig.

^a estimation of total effect of exposure in a sample of unmatched, independent observations from a single study; ^bcontinuous variable; ^c binary variable; ^d statistical significance; ^e categorical variable; ^f CI: confidence interval; ^g option to set the parameter of the marginal distribution of U; ^h R package EValue, Stata command evalue and web tool E-value calculator; ⁱ time to event variable

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com