Area under the curve for repeated measures anova

[Maria Majeed]

Hello Everyone,

I have been stuck at plotting area under the curve for a panel dataset (id) with six repeated measurements at three different conditions (just a difference in a month) for all same nine patients.
I carried out the analysis by using repeated measures of ANOVA. However, my manager would like to see AUC for all three months.

I used Pkexamin command but received an error "time variable takes on repeated values" (indeed it involves repeated measurements of time, with six readings)
When I used pksumm command, it would not give me p-value (just .) but AUC of each category

Secondly, in case I have AUC values for all three months, how would I be able to compare three AUCs (months) with a p-value

Thanks

[Joseph Coveney]

Try something like the following. Begin at the "Begin here" comment; the first part of the do-file just creates a fictitious dataset that has the same essential structure as yours for illustration.

.ÿ
.ÿversionÿ16.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿsetÿseedÿ`=strreverse("1551445")'

.ÿquietlyÿsetÿobsÿ9

.ÿ
.ÿgenerateÿbyteÿpidÿ=ÿ_n

.ÿgenerateÿdoubleÿpid_uÿ=ÿrnormal()

.ÿ
.ÿquietlyÿexpandÿ3ÿÿÿÿÿÿÿÿ

.ÿbysortÿpid:ÿgenerateÿbyteÿcdnÿ=ÿ_n

.ÿ
.ÿquietlyÿexpandÿ6

.ÿbysortÿpidÿcdn:ÿgenerateÿbyteÿtimÿ=ÿ_n

.ÿ
.ÿgenerateÿdoubleÿoutÿ=ÿpid_uÿ+ÿ10ÿ-ÿ(timÿ-ÿ3.5)^2ÿ+ÿrnormal()

.ÿassertÿoutÿ>ÿ0

.ÿ
.ÿ*
.ÿ*ÿBeginÿhere
.ÿ*
.ÿegenÿlongÿpid_cdnÿ=ÿgroup(pidÿcdn)

.ÿ
.ÿsetÿtypeÿdouble

.ÿpkcollapseÿtimÿout,ÿid(pid_cdn)ÿstat(auc)ÿtrapezoidÿkeep(pidÿcdn)ÿnodots

.ÿ
.ÿxtregÿauc_outÿi.cdn,ÿi(pid)ÿfe

Fixed-effectsÿ(within)ÿregressionÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ27
Groupÿvariable:ÿpidÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿgroupsÿÿ=ÿÿÿÿÿÿÿÿÿÿ9

R-sq:ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿObsÿperÿgroup:
ÿÿÿÿÿwithinÿÿ=ÿ0.1373ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿminÿ=ÿÿÿÿÿÿÿÿÿÿ3
ÿÿÿÿÿbetweenÿ=ÿÿÿÿÿÿ.ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿavgÿ=ÿÿÿÿÿÿÿÿ3.0
ÿÿÿÿÿoverallÿ=ÿ0.0188ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿmaxÿ=ÿÿÿÿÿÿÿÿÿÿ3

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿF(2,16)ÿÿÿÿÿÿÿÿÿÿÿ=ÿÿÿÿÿÿÿ1.27
corr(u_i,ÿXb)ÿÿ=ÿ0.0000ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿFÿÿÿÿÿÿÿÿÿÿ=ÿÿÿÿÿ0.3067

------------------------------------------------------------------------------
ÿÿÿÿÿauc_outÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿtÿÿÿÿP>|t|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿcdnÿ|
ÿÿÿÿÿÿÿÿÿÿ2ÿÿ|ÿÿÿ.9303635ÿÿÿ.9332006ÿÿÿÿÿ1.00ÿÿÿ0.334ÿÿÿÿ-1.047933ÿÿÿÿÿ2.90866
ÿÿÿÿÿÿÿÿÿÿ3ÿÿ|ÿÿÿ1.472431ÿÿÿ.9332006ÿÿÿÿÿ1.58ÿÿÿ0.134ÿÿÿÿ-.5058659ÿÿÿÿ3.450728
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿ_consÿ|ÿÿÿ38.49384ÿÿÿ.6598725ÿÿÿÿ58.34ÿÿÿ0.000ÿÿÿÿÿ37.09497ÿÿÿÿ39.89271
-------------+----------------------------------------------------------------
ÿÿÿÿÿsigma_uÿ|ÿÿ4.3739292
ÿÿÿÿÿsigma_eÿ|ÿÿ1.9796174
ÿÿÿÿÿÿÿÿÿrhoÿ|ÿÿ.82998434ÿÿÿ(fractionÿofÿvarianceÿdueÿtoÿu_i)
------------------------------------------------------------------------------
Fÿtestÿthatÿallÿu_i=0:ÿF(8,ÿ16)ÿ=ÿ14.65ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿFÿ=ÿ0.0000

.ÿtestparmÿi.cdn

ÿ(ÿ1)ÿÿ2.cdnÿ=ÿ0
ÿ(ÿ2)ÿÿ3.cdnÿ=ÿ0

ÿÿÿÿÿÿÿF(ÿÿ2,ÿÿÿÿ16)ÿ=ÿÿÿÿ1.27
ÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿFÿ=ÿÿÿÿ0.3067

.ÿ
.ÿ//ÿor
.ÿ
.ÿanovaÿauc_outÿcdnÿpid

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿ=ÿÿÿÿÿÿÿÿÿ27ÿÿÿÿR-squaredÿÿÿÿÿ=ÿÿ0.8821
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿRootÿMSEÿÿÿÿÿÿ=ÿÿÿÿ1.97962ÿÿÿÿAdjÿR-squaredÿ=ÿÿ0.8084

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿPartialÿSSÿÿÿÿÿÿÿÿÿdfÿÿÿÿÿÿÿÿÿMSÿÿÿÿÿÿÿÿFÿÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿ-----------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿModelÿ|ÿÿ469.13255ÿÿÿÿÿÿÿÿÿ10ÿÿÿ46.913255ÿÿÿÿÿ11.97ÿÿ0.0000
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿcdnÿ|ÿÿ9.9823981ÿÿÿÿÿÿÿÿÿÿ2ÿÿÿ4.9911991ÿÿÿÿÿÿ1.27ÿÿ0.3067
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿpidÿ|ÿÿ459.15016ÿÿÿÿÿÿÿÿÿÿ8ÿÿÿ57.393769ÿÿÿÿÿ14.65ÿÿ0.0000
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿ62.702158ÿÿÿÿÿÿÿÿÿ16ÿÿÿ3.9188849ÿÿ
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿ-----------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿTotalÿ|ÿÿ531.83471ÿÿÿÿÿÿÿÿÿ26ÿÿÿ20.455181ÿÿ

.ÿ
.ÿexit

endÿofÿdo-file


.

[Maria Majeed]

Hi Many many thanks for your response,

Can you please indicate what these variables cdn, tim, out, pid refer to in your example. I started using stata just six months ago so I may have missed out in your post.

Thanks

[Maria Majeed]

Hello Mr Joseph,

Thank you so much for your response, I figured out the comparison in your variables and how it applies to mine.

Would you be able to tell me if I want to compare the section of AUC (1-30MIN) to AUC (30-60 MIN) within and between different categories of grouping variable?

Thanks once again
stay blessed

[Joseph Coveney]

Would you be able to tell me if I want to compare the section of AUC (1-30MIN) to AUC (30-60 MIN) within and between different categories of grouping variable?

It's certainly doable, but it sounds like your manager is fishing. I would limit ad hoc analysis to graphical exploration and would restrict hypothesis testing to that which is driven by the research question in advance.