The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Child Health » Child mortality variables (Neonatal, infant and childmortality variables in the DHS 2019-21 India dataset )
Child mortality variables [message #24955] Tue, 09 August 2022 13:22 Go to next message
aishasarmad is currently offline  aishasarmad
Messages: 4
Registered: August 2022
Member
Hi! I am doing my master's dissertation on domestic violence and child mortality using the DHS 2019-21 dataset for India. I am struggling to generate a concise variable for childhood mortality (using neonatal, infant, and child mortality). Could someone help me with the code for this in Stata?
Re: Child mortality variables [message #24982 is a reply to message #24955] Fri, 12 August 2022 16:52 Go to previous messageGo to next message
Janet-DHS is currently offline  Janet-DHS
Messages: 779
Registered: April 2022
Senior Member
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

We may be able to help with this but I think we need more information about what are your units of analysis. The DV data is collected for women and is in the IR file. Child mortality is a characteristic of children and is usually analyzed with the KR or BR files. The woman's record in the IR file includes information about ALL of her births, with a subscript for each birth, so if you want to construct a measure of childhood mortality you will probably want to summarize her experience in some way. You will have to decide whether to summarize all her births, just the latest one, etc.

Alternatively, you may be working with child survival as an outcome, in the KR or BR file, which would be simpler.

Maybe you can clarify why you want to relate childhood mortality to the reported experience of domestic violence. If you believe they are related, is it because DV increases the risk of a child death? Or something else? And how is the age at death relevant?

Having said all this, the simple answer to your question is that there is a neonatal death if b6<28, an infant death if b7<12, and an under-five death is b7<60.
Re: Child mortality variables [message #24984 is a reply to message #24982] Fri, 12 August 2022 18:35 Go to previous messageGo to next message
aishasarmad is currently offline  aishasarmad
Messages: 4
Registered: August 2022
Member
Thank you for your reply, I appreciate it!

I want to generate a variable for individual child mortality. I want to examine if children of mothers who have experienced domestic violence are more likely to die within 30 days, a year, or 5 years after being born, than those who don't experience violence. I want to use a linear probability model where the dependent variable is a measure of child mortality for a child born to a mother who is a respondent to a DHS survey.

The IR file has information on fertility histories provided by women (v201), number of living children (v218), the month of interview (v006), year of interview (v007), year of birth of child (B2) and whether child was alive or dead at time of interview (B5) amongst others, but I am not sure how exactly I can use these (or other variables) to generate a variable or variables that capture neonatal mortality, infant mortality and under-5 mortality.I believe I might need to reshape or merge datasets first but I am not quite sure.

Further, since age heaping is observed in the data, I want to include the thirtieth day, twelfth and sixtieth months in each of these measures and also exclude from the analysis any children who are younger than these thresholds at the time of the survey.
I am not that proficient in Stata which is why I am a bit stuck at generating code for this. I will be grateful if you can help me with it as I am struggling quite a bit.
Re: Child mortality variables [message #25014 is a reply to message #24984] Fri, 19 August 2022 19:38 Go to previous messageGo to next message
aishasarmad is currently offline  aishasarmad
Messages: 4
Registered: August 2022
Member
I believe the first step would be to merge the BR file (base file) and IR file so the child becomes the unit of analysis. After this I would need to generate 3 variables for mortality (0/1 for whether the child died before the ages of one month, one year or 5 years) and 3 variables for mortality sample (0/1 for whether the child was born at least a month before date of interview, a year before date of interview or 5 years before date of interview). Then my regressions will have the outcome as the mortality measure and the sample I will use to estimate the regression will be of children who are exposed to the mortality risk.

Could you help me with the Stata code required to do this, please?

Re: Child mortality variables [message #25015 is a reply to message #24982] Fri, 19 August 2022 19:39 Go to previous messageGo to next message
aishasarmad is currently offline  aishasarmad
Messages: 4
Registered: August 2022
Member
I believe the first step would be to merge the BR file (base file) and IR file so the child becomes the unit of analysis. After this I would need to generate 3 variables for mortality (0/1 for whether the child died before the ages of one month, one year or 5 years) and 3 variables for mortality sample (0/1 for whether the child was born at least a month before date of interview, a year before date of interview or 5 years before date of interview). Then my regressions will have the outcome as the mortality measure and the sample I will use to estimate the regression will be of children who are exposed to the mortality risk.

Could you help me with the Stata code required to do this, please?
Re: Child mortality variables [message #25124 is a reply to message #25015] Thu, 01 September 2022 16:29 Go to previous message
Janet-DHS is currently offline  Janet-DHS
Messages: 779
Registered: April 2022
Senior Member
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

The BR file includes almost all of the variables for the mother that are in the IR file, so an IR/BR merge may not be needed.

The variables that give age at death are b6 and b7 (b7 is a recode of b6). If these variables are coded NA (".") then the child is still alive (and b5=1). The easiest constructions are that a death is neonatal if b6<128, infant if b7<12, under-5 if b7<60. Whether and how you deal with heaping is up to you. The age of the child in months is given by hw1 and age in days is given by b19. We use 365.25/12 as the number of days in a month.

We apologize for the delay in this response and hope it will still be useful.
Previous Topic: Penta or DPT
Next Topic: Difference between PCV and Pneumococcal vaccine?
Goto Forum:
  


Current Time: Wed Jul 24 18:18:15 Coordinated Universal Time 2024