The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Mortality » MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) (When variables exist in multiples, which variable should be used for the analysis )
MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #29501] Fri, 28 June 2024 16:02 Go to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear DHS Team,

Good evening.

Please I have several questions.

I'm working on variable MM9, M13 and M14.

I understand that MM9 in the Nigeria Demographic Health Survey (NDHS) indicates if the respondent's sister was pregnant when she died, if she died during childbirth, within 6 weeks after the delivery or within 2 months after the delivery.

However, the SPSS dataset NGIR7BFL has some variables I am interested in and they are all in multiples.

MM9 is from MM9$01 - MM9$20 (maternal mortality)

M13 is from M13$1 - M13$6 (timing of first antenatal check)

M14 is from M14$1 - M14$6 (number of antenatal visits)

Please, how can I combine the variables mm9_01 to mm9_20 to a single variable?
Please, how can I combine the variables m13_01 to m13_6 to a single variable?
Please, how can I combine the variables m14_01 to m14_6 to a single variable?


My dillemma, is which of these variables in these ranges for MM9, M13 and M14 should be analysed?

My research is looking at the association between timing of first antenatal check and number of antenatal visits during pregnancy, and maternal mortality when controlling for highest educational level, and ethnicity among pregnant women in Nigeria.

Thank you

[Updated on: Sat, 29 June 2024 18:48]

Report message to a moderator

Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #29526 is a reply to message #29501] Tue, 02 July 2024 10:31 Go to previous messageGo to next message
Janet-DHS is currently offline  Janet-DHS
Messages: 888
Registered: April 2022
Senior Member
Following is a response from DHS staff member, Tom Pullum:
In the IR file, you will find a group of b variables that are indexed from 1 to 20. They refer to up to 20 births. They are reshaped in the BR file and KR files. The mm variables are analogous but they refer to up to 20 siblings. To analyze them you must reshape them into a file that is like the BR file but has siblings as cases. Beginning with DHS-8, DHS now produces an "SR" file that is a reshaping of the mm variables, but for this Nigeria survey you will have to do that yourself. You should be able to find the relevant SPSS code in the GitHub program for adult and maternal mortality, the generic chapter 16 (https:// github.com/DHSProgram/DHS-Indicators-SPSS/tree/master/Chap16 _AM).

M13 and m14 and other "m" variables are indexed from 1 to 6, for up to 6 children born in the past 5 years, and are reshaped in the KR file. You do not need to do the reshaping yourself. In the KR file, the children are indexed by bidx. Some of the m variables are only given for the most recent birth (bidx=1).
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30069 is a reply to message #29526] Thu, 19 September 2024 06:00 Go to previous messageGo to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear DHS program manager,

Good morning,

Thank you for your email a few months ago.

I am still struggling with this and do not have a solution yet. I am working with the Nigeria Demographic Health Survey (NDHS)

Please I need your guidance on what to do with the SPSS code in the GitHub program for adult and maternal mortality so that I am sure I am doing the correct thing. I do not understand how to go about using the code to generate one MM9 variable.

I ran an analysis but it returned 40,709 missing values. That was a red flag for me. Please kindly provide some guidance.

Thank you
kind regards

Kanayo Omonoju
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30110 is a reply to message #30069] Tue, 24 September 2024 20:53 Go to previous messageGo to next message
Janet-DHS is currently offline  Janet-DHS
Messages: 888
Registered: April 2022
Senior Member
Following is a response from DHS staff member, Tom Pullum:

I only use Stata and will respond with that, but you should be able to translate to SPSS. I think the problem is that when you reshape the data you are keeping the full allowance of up to 20 siblings. You need to delete the lines of data that go beyond the number of siblings that a woman actually had. Also you need to restrict to women who died at age 15-49.

I entered the following lines:
use "...NGIR7BFL.DTA", clear

* remove leading zeroes in the subscripts
rename *_0* *_*

* For this example just keep mm1 and mm7 and mm9
keep v001 v002 v003 mm1_* mm9_*

reshape long mm1_ mm9_, i(v001 v002 v003) j(index)

rename *_ *
tab mm1, m
* Drop the lines that go beyond the woman's actual number of siblings
drop if mm1==.

tab mm9 if mm1==2 & mm7>=15 & mm7<=49,m

I see 616,859 blank lines in the data, before dropping the lines with mm1=1:
. tab mm1, m

 

        mm1 |      Freq.     Percent        Cum.

------------+-----------------------------------

       male |    112,971       13.51       13.51

     female |    106,590       12.74       26.25

          . |    616,859       73.75      100.00

------------+-----------------------------------

      Total |    836,420      100.00

The last line in the program produces this:
. tab mm9 if mm1==2 & mm7>=15 & mm7<=49,m

 

                    mm9 |      Freq.     Percent        Cum.

------------------------+-----------------------------------

      death not related |      2,484       67.52       67.52

    died while pregnant |        415       11.28       78.80

   died during delivery |        535       14.54       93.34

6 weeks after delivery |        202        5.49       98.83

2 months after delivery |         43        1.17      100.00

------------------------+-----------------------------------

                  Total |      3,679      100.00

There are no "missing" cases if you restrict to women age 15-49 at death. These are the only siblings to whom mm9 applies.

Note that there is an updated version of the files, with "C" in place of "B".

Please let us know if you are still seeing blanks or NA cases
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30138 is a reply to message #30110] Wed, 02 October 2024 09:47 Go to previous messageGo to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Good afternoon Janet,

Thank you very much for your response.

Please I have several questions:

1. From this statement below, please how do I know the number of siblings a woman had? To help me know which data to delete correctly.

"You need to delete the lines of data that go beyond the number of siblings that a woman actually had."


2. The NGIR7BFL individual recode data set has 41,821 cases of women 15-49 years who were interviewed.

Please can you kindly explain what the 836,420 stands for, if 41,821 is the number of cases?

3. In the attached, I did a manual computation on an Excel sheet to better understand the data. I have tried to combine the 20 MM9 variables in the NGIR7BFL data but I am still getting missing cases.

I did a quick descriptive analysis of the 20 MM9 variables in the dataset (MM9$01 to MM9 $20)and got the below:

Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
MM9$01- Sibling's death and pregnancy 1112 1 6 1.59 1.146
MM9$02 -Sibling's death and pregnancy 834 1 6 1.56 1.108
MM9$03-Sibling's death and pregnancy 707 1 6 1.63 1.121
MM9$04-Sibling's death and pregnancy 553 1 6 1.65 1.166
MM9$05-Sibling's death and pregnancy 367 1 5 1.66 1.110
MM9$06-Sibling's death and pregnancy 264 1 6 1.69 1.297
MM9$07-Sibling's death and pregnancy 174 1 5 1.53 .995
MM9$08-Sibling's death and pregnancy 96 1 6 1.59 1.091
MM9$09-Sibling's death and pregnancy 46 1 5 1.50 .913
MM9$10-Sibling's death and pregnancy 25 1 5 1.84 1.405
MM9$11-Sibling's death and pregnancy 10 1 3 1.30 .675
MM9$12- Sibling's death and pregnancy 1 1 1 1.00 .
MM9$13- Sibling's death and pregnancy 0
MM9$14- Sibling's death and pregnancy 1 1 1 1.00 .
MM9$15-Sibling's death and pregnancy 0
MM9$16- Sibling's death and pregnancy 0
MM9$17- Sibling's death and pregnancy 0
MM9$18- Sibling's death and pregnancy 0
MM9$19- Sibling's death and pregnancy 0
MM9$20- Sibling's death and pregnancy 0
Valid N (listwise) 0


Please I want to confirm what the N represents.

What does the 1112 in MM9$01 imply?

4. I am unable to translate the stata code to SPSS, please can I kindly get some guidance

Thank you

kind regards
Kanayo Omonoju



Thank you
Kanayo
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30143 is a reply to message #30110] Thu, 03 October 2024 10:38 Go to previous messageGo to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear Janet,

Good afternoon,

Please I am still having missing cases.
Please kindly provide some guidance
Please kindly see attached.

Thank you

kind regards
Kanayo
  • Attachment: 03102024.spv
    (Size: 8.71KB, Downloaded 19 times)
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30148 is a reply to message #30138] Thu, 03 October 2024 14:18 Go to previous messageGo to next message
Janet-DHS is currently offline  Janet-DHS
Messages: 888
Registered: April 2022
Senior Member
Following is a response from DHS staff member, Tom Pullum:

You can calculate the number of siblings with these lines (in Stata), before reshaping the IR file:
use "C:\Users\26216\ICF\Analysis - Shared Resources\Data\DHSdata\NGIR7BFL.DTA", clear

 

rename *_0* *_*

 

* Calculate the number of siblings each woman has

gen nsibs=0

forvalues lsibs=1/20 {

replace nsibs=nsibs+1 if mm1_`lsibs'<.

}

 

tab nsibs

tab nsibs [iweight=v005/1000000]

summarize nsibs [iweight=v005/1000000]

The weighted mean is 5.23 siblings. Note that you have to allow for the possibility of 0 siblings. This can be modified to give the number of siblings who are males, females, still alive, etc. I am repeating the Stata lines I gave earlier, because what I pasted the first time omitted m7. The full program is below--but it does not include the lines to calculate the adult and maternal mortality rates. Those lines are on our GitHub site.

use "C:\Users\26216\ICF\Analysis - Shared Resources\Data\DHSdata\NGIR7BFL.DTA", clear

 

rename *_0* *_*

 

* Calculate the number of siblings each woman has

gen nsibs=0

forvalues lsibs=1/20 {

replace nsibs=nsibs+1 if mm1_`lsibs'<.

}

 

tab nsibs

tab nsibs [iweight=v005/1000000]

summarize nsibs [iweight=v005/1000000]

 


* For this example just keep mm1 and mm7 and mm9

keep v001 v002 v003 mm1_* mm7_* mm9_*


reshape long mm1_ mm7_ mm9_, i(v001 v002 v003) j(index)


rename *_ *

tab mm1, m

* Drop the lines that go beyond the woman's actual number of siblings

drop if mm1==.


tab mm9 if mm1==2 & mm7>=15 & mm7<=49,m
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30180 is a reply to message #30110] Wed, 09 October 2024 08:48 Go to previous messageGo to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear Janet,

Good afternoon

Please I need a clarification:


2. The NGIR7BFL individual recode data set has 41,821 cases of women 15-49 years who were interviewed.

Please can you kindly explain what the 836,420 stands for, if 41,821 is the number of cases?

What do the missing systems imply? Are they missing cases?

Thank you
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30181 is a reply to message #30180] Wed, 09 October 2024 08:48 Go to previous messageGo to next message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear Janet,

Good afternoon

Please I need a clarification:


2. The NGIR7BFL individual recode data set has 41,821 cases of women 15-49 years who were interviewed.

Please can you kindly explain what the 836,420 stands for, if 41,821 is the number of cases?

What do the missing systems imply? Are they missing cases?

Thank you
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30195 is a reply to message #30181] Fri, 11 October 2024 16:13 Go to previous messageGo to next message
Janet-DHS is currently offline  Janet-DHS
Messages: 888
Registered: April 2022
Senior Member
Following is a response from DHS staff member, Tom Pullum:
The questionnaires have lines for up to 20 siblings (siblings are defined to be other children of the same mother). I don't know whether any respondent in any DHS survey has ever needed all 20 lines. The extra lines are blank, which in Stata is a dot for Not Applicable or NA. On the respondent's record the siblings' variables have subscripts _01 through _20.

In the reshape long step, a file of siblings is constructed, with 20 lines for each respondent, one line for each sibling, whether or not the sibling ever existed. Most of the lines are just blank. We drop those lines.

That's all there is to it. This is just a data processing step to go from a "wide" file to a "long" file and drop blank lines. Nothing is missing.
Re: MM9 in the NGIR7BFL dataset of NDHS is from MM9$01 - MM9$20 (maternal mortality) [message #30196 is a reply to message #30195] Fri, 11 October 2024 18:48 Go to previous message
prekan2000 is currently offline  prekan2000
Messages: 7
Registered: June 2024
Member
Dear Janet,

Thank you.
Previous Topic: Case-crossover analysis with DHS
Next Topic: Stillbirth
Goto Forum:
  


Current Time: Fri Nov 22 23:34:43 Coordinated Universal Time 2024