How to obtain confidence interval & p-value from stmh command result table

[Paul Somwe]

Good Morning,

I am a relatively new STATA15 user. I need some help obtaining the confidence interval and p-value output after executing the stmh command. e.g:

stmh age_at_start, c(2,1)


         failure _d:  ltfup_status
   analysis time _t:  date_of_nextapp/365.25
  enter on or after:  time art_start_date
                 id:  artnumber

Maximum likelihood estimate of the rate ratio
  comparing age_at_artstart2==2 vs. age_at_artstart2==1

RR estimate, and lower and upper 95% confidence limits

        ----------------------------------------------------------
             RR       chi2        P>chi2      [95% Conf. Interval]
        ----------------------------------------------------------
          2.041     143.33        0.0000         1.812      2.300
        ----------------------------------------------------------

I am only able to get the RR value using the return list command. Your help will be greatly appreciated.

Thanks,

Paul

[Rich Goldstein]

the p-value and the CI are saved in the returned results (as shown in the help file) so your question is unclear; you can, for example, -display- them:

di "p-value is " r(p)

etc.

but we can't give exact code unless we know what you want to do with these

[Paul Somwe]

Hi Rich,

I have tried display r(p), display display r(lb) and display r(ub), still not getting any results. I want to populate these values into an excel file using putexcel. Kindly help.

Thanks,

Paul

[Rich Goldstein]

as you are still not following the FAQ advice (show exact commands and sets of commands), I can only guess that you are doing something that "destroys" the returned values prior to your attempts; please read the FAQ and show us exactly what you are telling Stata to do and exactly how Stata is responding

[Paul Somwe]

Hi Rich, the exact commands I executed and associated results are:

. stmh viral_load_status, c(1,0)

         failure _d:  ltfup_status
   analysis time _t:  date_of_nextapp/365.25
  enter on or after:  time art_start_date
                 id:  artnumber

Maximum likelihood estimate of the rate ratio
  comparing viral_load_status==1 vs. viral_load_status==0

RR estimate, and lower and upper 95% confidence limits

        ----------------------------------------------------------
             RR       chi2        P>chi2      [95% Conf. Interval]
        ----------------------------------------------------------
          3.550     419.71        0.0000         3.119      4.041
        ----------------------------------------------------------

. return list

scalars:
                 r(RR) =  3.550137764026656

The return list command is only giving me the RR value. The p-value and confidence interval values are not given. These are the values I want so that I may popuate them into an excel file using putexcel.

Thanks,

Paul

[Joseph Coveney]

What do you get when you type

which stmh.ado

at the command line?

I don't know what Stata Release 15 had, but this is what I get with the next release.

*! version 6.1.1 09apr2020

Do you get something more like the following?

. type "http://www.stata.com/stb/stb41/ssa10_1/stmh.ado"
*! version 2.0 dgc/mh Sep 1997 STB-41 ssa10.1
*! this an st wrapper for dymh

[Paul Somwe]

Hi Joseph,

See below:

. which stmh.ado
C:\Program Files (x86)\ado\base\s\stmh.ado
*! version 6.1.0  16apr2007

is there something you are trying to ascertain?

Thanks,

paul

[Andrew Musau]

The return list command is only giving me the RR value. The p-value and confidence interval values are not given.

I do not think that this is a version issue. One workaround is to log the session when executing the command. Then you can read these numbers using infile. My example shows that these will be in line 14 and as the table is standard, I do not think that this will be variable.

webuse diet, clear
qui stset dox, origin(time dob) enter(time doe) id(id) scale(365.25) fail(fail==1 3 13)
preserve
qui log using firstfile, name(log1) text replace
stmh hienergy
log close log1
insheet using  "firstfile.log", clear
keep in 14
local chi2=  word(v1, 2)
local p= word(v1, 3)
local ci= "[`=word(v1, 4)' `=word(v1, 5)']"
restore

Res.:

. stmh hienergy

         failure _d:  fail == 1 3 13
   analysis time _t:  (dox-origin)/365.25
             origin:  time dob
  enter on or after:  time doe
                 id:  id

Maximum likelihood estimate of the rate ratio
  comparing hienergy==1 vs. hienergy==0

RR estimate, and lower and upper 95% confidence limits

        ----------------------------------------------------------
             RR       chi2        P>chi2      [95% Conf. Interval]
        ----------------------------------------------------------
          0.520       4.84        0.0277         0.288      0.941
        ----------------------------------------------------------


. di "`chi2'"
4.84

. di "`p'"
0.0277

. di "`ci'"
[0.288 0.941]

[Rich Goldstein]

Andrew Musau regardless of whether #8 is correct about the version not being the problem the OP should update their version of Stata (it is free to do)

[Andrew Musau]

Absolutely.

[Paul Somwe]

Thank you all, will try your suggestions!

[Paul Somwe]

I do not think that this is a version issue. One workaround is to log the session when executing the command. Then you can read these numbers using infile. My example shows that these will be in line 14 and as the table is standard, I do not think that this will be variable.

webuse diet, clear
qui stset dox, origin(time dob) enter(time doe) id(id) scale(365.25) fail(fail==1 3 13)
preserve
qui log using firstfile, name(log1) text replace
stmh hienergy
log close log1
insheet using "firstfile.log", clear
keep in 14
local chi2= word(v1, 2)
local p= word(v1, 3)
local ci= "[`=word(v1, 4)' `=word(v1, 5)']"
restore

Res.:

. stmh hienergy

failure _d: fail == 1 3 13
analysis time _t: (dox-origin)/365.25
origin: time dob
enter on or after: time doe
id: id

Maximum likelihood estimate of the rate ratio
comparing hienergy==1 vs. hienergy==0

RR estimate, and lower and upper 95% confidence limits

----------------------------------------------------------
RR chi2 P>chi2 [95% Conf. Interval]
----------------------------------------------------------
0.520 4.84 0.0277 0.288 0.941
----------------------------------------------------------


. di "`chi2'"
4.84

. di "`p'"
0.0277

. di "`ci'"
[0.288 0.941]

Hi Andrew,

I have tried your example and still the chi2, p, and ci interval values are not being displayed. I am honestly stuck now because am not quite sure what could be the issue.

Thanks,

Paul

[Andrew Musau]

Can you show your exact commands and the resulting Stata output? Place these within CODE delimiters.

[Paul Somwe]

Can you show your exact commands and the resulting Stata output? Place these within CODE delimiters.

Hi Andrew Musau ,

I ran the exact code you shared i.e.

. webuse diet, clear
(Diet data with dates)

. qui stset dox, origin(time dob) enter(time doe) id(id) scale(365.25) fail(fail==1 3 13)

. preserve

. qui log using firstfile, name(log1) text replace

. stmh hienergy

         failure _d:  fail == 1 3 13
   analysis time _t:  (dox-origin)/365.25
             origin:  time dob
  enter on or after:  time doe
                 id:  id

Maximum likelihood estimate of the rate ratio
  comparing hienergy==1 vs. hienergy==0

RR estimate, and lower and upper 95% confidence limits

        ----------------------------------------------------------
             RR       chi2        P>chi2      [95% Conf. Interval]
        ----------------------------------------------------------
          0.520       4.84        0.0277         0.288      0.941
        ----------------------------------------------------------

. log close log1
      name:  log1
       log:  C:\Users\Paul.Somwe\Documents\Tx abstract\firstfile.log
  log type:  text
 closed on:   7 Jul 2020, 13:04:21
--------------------------------------------------------------------------------------------------------------------------

. insheet using "firstfile.log", clear
(1 var, 20 obs)

. keep in 14
(19 observations deleted)

. local chi2= word(v1, 2)

. local p= word(v1, 3)

. local ci= "[`=word(v1, 4)' `=word(v1, 5)']"

. restore

. 
. stmh hienergy

         failure _d:  fail == 1 3 13
   analysis time _t:  (dox-origin)/365.25
             origin:  time dob
  enter on or after:  time doe
                 id:  id

Maximum likelihood estimate of the rate ratio
  comparing hienergy==1 vs. hienergy==0

RR estimate, and lower and upper 95% confidence limits

        ----------------------------------------------------------
             RR       chi2        P>chi2      [95% Conf. Interval]
        ----------------------------------------------------------
          0.520       4.84        0.0277         0.288      0.941
        ----------------------------------------------------------

. di "`chi2'"


. di "`p'"


. di "`ci'"
[ ]

Notice that the chi2, p and ci values are not being displayed?

Thanks,

Paul

[Andrew Musau]

Thanks for the example. What is biting is the observation number. It's better to automate the search rather than relying on a manual approach.

webuse diet, clear
qui stset dox, origin(time dob) enter(time doe) id(id) scale(365.25) fail(fail==1 3 13)
preserve
qui log using firstfile, name(log1) text replace
stmh hienergy
log close log1
insheet using  "firstfile.log", clear
gen obs=_n
qui sum obs if regexm(v1, "P>chi2")
keep in `=r(min)+2'
local chi2=  word(v1, 2)
local p= word(v1, 3)
local ci= "[`=word(v1, 4)' `=word(v1, 5)']"
restore
di "`chi2'"
di "`p'"
di "`ci'"