The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » India » NFHS-5 (2019-21) (Maternal & other Variables Clarification)
NFHS-5 (2019-21) [message #29910] Wed, 21 August 2024 13:33 Go to next message
sak_bn is currently offline  sak_bn
Messages: 5
Registered: August 2024
Member
Hi;

I was reviewing some variables and looking for clarification regarding them. I am looking to do analysis based on the women who had given birth live/non-live or any adverse birth outcomes based on past 5 years. As such I plan to use variables: m_19 (birth weight; unsure if it should be m19_1, 2 , 3, 4 , 5 or 6 and whether its for the recorded births from past 5 years); v234 (unsure if it should v228 or v239 or s234 if considering past 5 years)

The variable V457 indicates anemia level. Is this something measured during pregnancy? if not what variable is for measured during pregnancy in past 5 years? Also is there a variable to recode if the birth was a pre-term birth in past 5 years.

Looking forward to response!

Thanks
Sak_bn
Re: NFHS-5 (2019-21) [message #29916 is a reply to message #29910] Thu, 22 August 2024 07:12 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

Following is a response from Senior DHS Staff member, Tom Pullum:

You are apparently using the IR file, which has women as units. Their births in the past five years are numbered with subscripts 1, 2, 3, etc., with 1 for the most recent birth. It will be much easier if you use the KR file, which has children as units. In that file, m19 has no subscripts.

As for choosing among "v234 (unsure if it should v228 or v239 or s234", I don't know what you are looking for. These variables are clearly labeled.

The anemia level of the mother, v457, is based on a blood test taken at the time of the survey. Values are given for pregnant women but we don't know the outcome of the current pregnancy. Hemoglobin analyses would normally not be used for women who are pregnant or within two months after a birth. You will find footnotes explaining that in the relevant tables in the final report. You cannot use current anemia level to explain previous pregnancy outcomes, if that is what you are trying to do.
Re: NFHS-5 (2019-21) [message #29917 is a reply to message #29916] Thu, 22 August 2024 07:25 Go to previous messageGo to next message
sak_bn is currently offline  sak_bn
Messages: 5
Registered: August 2024
Member
Thank-you for your prompt response. I am actually trying to evaluate how ANC service leads to adverse birth outcomes. If I use KR file, it would only give me data related to live births, however, my outcome is adverse birth outcomes which includes low birth weight, abortion, still births, preterm birth(unsure if there is a variable as I didn't found or may have missed).

As for anemia, I believe there won't be any records for anemia level during most recent pregnancy correct?

I am also looking for diseases factors during pregnancy such as hypertension, diabetes during pregnancy.

For variables, v234 (unsure if it should v228 or v239 or s234". These variables are related to stillbirth, abortions and miscarriages. I am specifically looking for one's in last give years only and if there is a breakdown and variables of how many stillbirths, miscarriages and stillbirths in last five years specifically

Thanks for all your help!
Re: NFHS-5 (2019-21) [message #29924 is a reply to message #29917] Thu, 22 August 2024 13:54 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

Following is a response from Senior DHS Staff member, Tom Pullum:

Thanks for your clarification. You may not be able to do quite what you want to do, because this survey did not include pregnancy histories. The calendar (vcal_1) includes pregnancy outcomes other than live births (with the letter T for termination) but it does not classify them with respect to whether they were a stillbirth, miscarriage, or abortion, EXCEPT for the most recent termination in the past 5 years, for which the breakdown is given with s234.

Also, in this survey, ANC information is provided only for live births. It is not asked for terminations.

You may have to reduce to the most recent pregnancy outcome in the past 5 years, because it is the only one you can classify as a birth / stillbirth / miscarriage / abortion, but even so, you will only have ANC information if it was a live birth.

Hemoglobin / anemia information is current status, that is, on the day of the survey. There is currently much discussion at CDC and WHO of the link between hemoglobin concentration and anemia. For many years, DHS has measured hemoglobin concentration and converted it to an anemia classification, but that may not continue. You may want to look into that link.

As for the connection between ANC and birth outcomes, such as stillbirths and early neonatal deaths, it has proven difficult to show that more ANC leads to better outcomes. Women who have a problematic pregnancy tend to have more ANC visits AND more risk of a poor outcome. This produces a spurious counter-intuitive relationship between the number of visits and the outcome.
Re: NFHS-5 (2019-21) [message #29925 is a reply to message #29924] Thu, 22 August 2024 14:48 Go to previous messageGo to next message
sak_bn is currently offline  sak_bn
Messages: 5
Registered: August 2024
Member
Thank you for your prompt response.

It seems that I might have to restructure my whole question and hypothesis in that case as there is no ANC information for women with stillbirths, miscarriages and abortions in last five years. Perhaps switch to other adverse outcomes such as low birth weight? Apart from this do we have variable which determines pre-term birth for the most recent birth or preeclampsia or other birth disorders for women giving live birth in past 5 years? For the most recent birth I might have to use 1 out of all 6 for the most recent birth.

I am also looking for diseases factors during pregnancy such as hypertension, diabetes during pregnancy and particularly looking for variables s728a,b, c,d,e,f,and g. I believe this is the status of the those disease/conditions during time of interview and not during pregnancy? I am unsure if I can link those conditions and use them for my analysis to evaluate if those conditions play any role during pregnancy.

Any help and guidance would really be appreciated.

Thanks
Re: NFHS-5 (2019-21) [message #29926 is a reply to message #29925] Fri, 23 August 2024 11:31 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

Following is a response from Senior DHS Staff member, Tom Pullum:

A cross-sectional survey such as a DHS (or NFHS) is a poor substitute for the kind of longitudinal clinical study that you would like to have.

s728 covers 7 chronic diseases but all of them are self-reports. The respondents may report that they have "sought treatment" but we don't know whether they are currently receiving treatment or medication. We do know more about hypertension, because blood pressure readings were included in the NFHS-5 (enter "lookfor systolic" to find the variable names).

Even if the woman self-reports a chronic disease at the time of the survey, we don't know whether she had it at the time of any pregnancy.

I looked at the frequencies. For example, of the 724,115 women in the IR file in the NFHS-5, 13,589 (weighted frequency) say they currently have diabetes. 1,576 of the women with diabetes had a birth in the past 5 years. Of them, 87 had a child in the past 5 years who died as an infant. I think it would be very difficult to find a relationship between currently having diabetes and having had an infant death, especially if you take potential confounders into account. Because these are self-reports and the temporal sequence is reversed, any reviewers would be skeptical.

The issues of sample size, self-reports, and reverse timing would be present for other outcomes such as low birthweight. Note that if eclampsia develops, there is a high risk that the mother dies and the case is lost entirely.

Your research topic is certainly important. The problem is just that you may not be able to get very far with these data. I can't come up with any serious suggestions. Maybe other users can help out.

Re: NFHS-5 (2019-21) [message #29931 is a reply to message #29926] Sun, 25 August 2024 10:55 Go to previous messageGo to next message
sak_bn is currently offline  sak_bn
Messages: 5
Registered: August 2024
Member
Thank you for your response.

I agree and the results would be skeptical. I was trying to enter command in Stata to see the weighted frequency results that you obtained but unable to get the similar results. Would you be able to confirm what command you used to achieve 13,589 and then 1576 with 87 had a child in past 5 years who died of diabetes?

Also, the variable m42 indicates only if blood pressure taken in terms of yes or no but not exactly how much was the blood pressure. As per what I have seen there is any variable which gives exact blood pressure measurement or glucose level measurement during pregnancy. Is this correct?

Thanks
Re: NFHS-5 (2019-21) [message #29936 is a reply to message #29931] Mon, 26 August 2024 07:46 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

Following is a response from Senior DHS Staff member, Tom Pullum:

Thanks for giving the specific numbers, but where did you see them? If they are in a table in the final report, please give me the table number. If they are on STATcompiler, please tell me exactly what indicator. Or were these numbers given on the forum?


Re: NFHS-5 (2019-21) [message #29951 is a reply to message #29936] Wed, 28 August 2024 13:06 Go to previous messageGo to next message
sak_bn is currently offline  sak_bn
Messages: 5
Registered: August 2024
Member
Hi

These were the numbers referring from the response by your on August 23rd stating: "I looked at the frequencies. For example, of the 724,115 women in the IR file in the NFHS-5, 13,589 (weighted frequency) say they currently have diabetes. 1,576 of the women with diabetes had a birth in the past 5 years. Of them, 87 had a child in the past 5 years who died as an infant. I think it would be very difficult to find a relationship between currently having diabetes and having had an infant death, especially if you take potential confounders into account. Because these are self-reports and the temporal sequence is reversed, any reviewers would be skeptical."

I was trying to enter command in Stata to see the weighted frequency results that you obtained but unable to get the similar results. Would you be able to confirm what command you used to achieve 13,589 and then 1576 with 87 had a child in past 5 years who died of diabetes?

Also, just to confirm there is no variable reflecting Hypertension, anemia or blood glucose levels/measurements during the most recent pregnancy/birth in DHS India 2019-21 survey? I was looking for the ones from 1998, and there was one for anemia during pregnancy so wanted to confirm.

Thanks
Re: NFHS-5 (2019-21) [message #29955 is a reply to message #29951] Thu, 29 August 2024 07:53 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member
Following is a response from Senior DHS staff member, Tom Pullum:

Embarrassing! When forum questions come to me they often do not have the history attached. Either that happened, or I just didn't see it. I will paste the Stata lines below, along with two tables, and will highlight the relevant numbers in those tables.

I included possible births and deaths in the month of interview. Sometimes we don't do that. Also children born in the past year did not have full exposure to the risk of an infant death. This was intended just to give a sense of how many cases there are, and other approaches could give slightly different numbers.

use "...IAIR7EFL.DTA", clear

lookfor diabetes

*remove leading zeroes from indices

rename *_0* *_*

*up to 6 births in the past five years

gen births=0
gen deaths=0

forvalues lb=1/6 {
replace births=1 if v008-b3_`lb'<=60
replace deaths=1 if v008-b3_`lb'<=60 & b7_`lb'<=11
}

label variable births "Had 1+ births in the past 5 years"
label variable deaths "Had 1+ infant deaths to those births"

tab s728a births [iweight=v005/1000000],m 
tab s728a deaths [iweight=v005/1000000],m


. tab s728a births [iweight=v005/1000000],m 

/index.php?t=getfile&id=2393&private=0
  • Attachment: tab s728a.png
    (Size: 20.87KB, Downloaded 185 times)
Previous Topic: INDIA NFHS 5
Next Topic: Regarding the Blood Glucose level SB74 in IR file
Goto Forum:
  


Current Time: Wed Nov 27 11:24:46 Coordinated Universal Time 2024