How to obtain bootstrap ROC after logistic regression

[Al Bothwell]

I have a binary outcome (positive blood culture, coded 0/1) and a continuous predictor (risk score, where higher number indicates greater risk). I run the following code:

logistic positivebloodculture riskscore, vce(bootstrap, reps(1000) seed(102703) dots(1))

Now I need a bootstrapped ROC curve from the logistic model. Any suggestions on how I can do this. I use Stata 16.

Thank you,
Al Bothwell

[Joseph Coveney]

To bootstrap the area under the receiver operating characteristic curve, you can try something like the following.

.ÿ
.ÿversionÿ16.0

.ÿ
.ÿclearÿ*

.ÿ
.ÿsetÿseedÿ`=strreverse("1529392")'

.ÿ
.ÿwebuseÿhanley

.ÿquietlyÿexpandÿpop

.ÿ
.ÿprogramÿdefineÿbootem
ÿÿ1.ÿÿÿÿÿÿÿÿÿversionÿ16.0
ÿÿ2.ÿÿÿÿÿÿÿÿÿsyntax
ÿÿ3.ÿ
.ÿÿÿÿÿÿÿÿÿlogitÿdiseaseÿc.rating
ÿÿ4.ÿÿÿÿÿÿÿÿÿlrocÿ,ÿnograph
ÿÿ5.ÿend

.ÿ
.ÿbootstrapÿrocÿ=ÿr(area),ÿnodots:ÿbootem

BootstrapÿresultsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿ2,513
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿReplicationsÿÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ50

ÿÿÿÿÿÿcommand:ÿÿbootem
ÿÿÿÿÿÿÿÿÿÿroc:ÿÿr(area)

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿObservedÿÿÿBootstrapÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNormal-based
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿzÿÿÿÿP>|z|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿrocÿ|ÿÿÿ.9886281ÿÿÿ.0018736ÿÿÿ527.65ÿÿÿ0.000ÿÿÿÿÿ.9849558ÿÿÿÿ.9923004
------------------------------------------------------------------------------

.ÿestatÿbootstrapÿ,ÿall

BootstrapÿresultsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿ2,513
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿReplicationsÿÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ50

ÿÿÿÿÿÿcommand:ÿÿbootem
ÿÿÿÿÿÿÿÿÿÿroc:ÿÿr(area)

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿObservedÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿBootstrap
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿCoef.ÿÿÿÿÿÿÿBiasÿÿÿÿStd.ÿErr.ÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿrocÿ|ÿÿÿ.98862809ÿÿ-.0001258ÿÿÿ.00187365ÿÿÿÿ.9849558ÿÿÿ.9923004ÿÿÿ(N)
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ.9834656ÿÿÿ.9917346ÿÿÿ(P)
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ.9850107ÿÿÿ.9917502ÿÿ(BC)
------------------------------------------------------------------------------
(N)ÿÿÿÿnormalÿconfidenceÿinterval
(P)ÿÿÿÿpercentileÿconfidenceÿinterval
(BC)ÿÿÿbias-correctedÿconfidenceÿinterval

.ÿ
.ÿbootstrapÿrocÿ=ÿr(area),ÿnodotsÿnowarn:ÿroctabÿdiseaseÿrating

BootstrapÿresultsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿ2,513
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿReplicationsÿÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ50

ÿÿÿÿÿÿcommand:ÿÿroctabÿdiseaseÿrating
ÿÿÿÿÿÿÿÿÿÿroc:ÿÿr(area)

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿObservedÿÿÿBootstrapÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNormal-based
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿzÿÿÿÿP>|z|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿrocÿ|ÿÿÿ.9886281ÿÿÿ.0018738ÿÿÿ527.60ÿÿÿ0.000ÿÿÿÿÿ.9849555ÿÿÿÿ.9923007
------------------------------------------------------------------------------

.ÿestatÿbootstrapÿ,ÿall

BootstrapÿresultsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿ2,513
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿReplicationsÿÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ50

ÿÿÿÿÿÿcommand:ÿÿroctabÿdiseaseÿrating
ÿÿÿÿÿÿÿÿÿÿroc:ÿÿr(area)

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿObservedÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿBootstrap
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿCoef.ÿÿÿÿÿÿÿBiasÿÿÿÿStd.ÿErr.ÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿrocÿ|ÿÿÿ.98862809ÿÿ-.0003885ÿÿÿ.00187381ÿÿÿÿ.9849555ÿÿÿ.9923007ÿÿÿ(N)
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ.9846106ÿÿÿÿ.991626ÿÿÿ(P)
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ.9850507ÿÿÿ.9919115ÿÿ(BC)
------------------------------------------------------------------------------
(N)ÿÿÿÿnormalÿconfidenceÿinterval
(P)ÿÿÿÿpercentileÿconfidenceÿinterval
(BC)ÿÿÿbias-correctedÿconfidenceÿinterval

.ÿ
.ÿexit

endÿofÿdo-file


.

[Al Bothwell]

Joseph, thank you for your help.


Al Bothwell.