| Previous adverse pregnancy outcome variable [message #11834] |
Wed, 15 February 2017 22:02  |
chr8850
Messages: 20
Registered: July 2016
|
Member |
|
|
Hi there,
I am working with the Afghanistan 2010 AMS and using the women's data have generated a pregnancy file using reshape that contains a list of the mother's pregnancies in the last five years. Using the pregnancy history data I generated a pregnancy outcome variable to determine if the outcome was a live birth, stillbirth or miscarriage. What i would like to do it be able to is determine if prior to the mothers' most recent pregnancy (identified as those cases where idx==q402) whether she had any other adverse pregnancy outcomes. At the moment the file is in such a way that the pregnancies are listed from 1 to X with 1 being her first pregnancy. I'm not sure how much information you need, but below I'm pasting the code used for generating the pregnancy outcome variable and a caseid for each woman as there was not one in the file.
For case ID:
egen caseid =concat (qhclust qhnumber qline),f(%03.0f)
For pregnancy outcome variable:
gen pregoutcomeb = .
replace pregoutcomeb = 1 if q314== 1 // q314- baby born alive born dead (1=born alive, 2=born dead)
replace pregoutcomeb = 2 if pregoutcomeb == . & q314 ==2 & q325 >=7 & q325 <98 // q325- how many months did pregnancy last?
replace pregoutcomeb = 3 if pregoutcomeb == . & (q325 <7 | q325 >= 98) //
lab def pregoutcomeb 1 "Live birth" 2 "Stillbirth" 3 "Miscarriage"
lab val pregoutcomeb pregoutcomeb
lab var pregoutcomeb "Pregnancy outcome"
I have noted that the women's data file includes a variable "psurv" that refers to survival of the preceding pregnancy before the most recent pregnancy. However some of the cases don't exactly match the pregnancy outcome variable that I generated (for example there is a case classified as a stillbirth but when i looked at the question where the mother was asked if the baby was born alive born dead, she said born alive, so i have called that a live birth rather than a stillbirth) and this variable only has two outcomes - live births and stillbirths with miscarriages being classified as stillbirths from what i can see.
It would also be good to know how I can count how many of each outcome the mother has as well i.e for each mother how many Live births, stillbirths and miscarriages in a lifetime.
Any help you can provide would be much appreciated
best,
Ali
|
|
|
|
|
|
|
|
|
|
| Re: Previous adverse pregnancy outcome variable [message #12113 is a reply to message #12091] |
Tue, 28 March 2017 21:49  |
chr8850
Messages: 20
Registered: July 2016
|
Member |
|
|
Hi Han, thanks for your reply.
I used the women.dta file from 2010 Afghanistan Mortality Survey. There is a variable called psurv_* for each birth in that file.
There is a variable q309 but that refers to the question on how many stillbirths a woman had in her lifetime. I used the pregnancy history to determine stillbirths, not q309. What i need to know is apart from the woman's most recent birth whether she had a stillbirth previous to that.
best
Ali
|
|
|
|
| Re: Previous adverse pregnancy outcome variable [message #12114 is a reply to message #12113] |
Wed, 29 March 2017 04:37 |
HanR
Messages: 19
Registered: March 2013
Location: Netherlands
|
Member |
|
|
Hi Ali,
Some thoughts:
From your message below I understand that you created a new file with the female as identifier (qhclust qhnumber qline). The unit of analysis in your file is one woman. This would mean that all occurrences of variables in the pregnancy history have their own unique name (as in for example q314_01 and q314_02 etc.). You would then be able to determine whether a previous pregnancy was a stillbirth or not by looking at the previous occurrences of q314 and the length of the pregnancy in q325 of the same occurrences (consider using arrays here).
The variable psurv_* does not distinguish between spontaneous/induced abortions and stillbirth. If you are specifically only looking at still births (born dead after 7+ months of pregnancy) then this variable will not help you.
If you want to know whether the respondent had any non-live birth before the last one then you could use q309 by simply looking at the outcome of the last birth and the total number of non-live births in q309.
Good luck.
Han.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
