The DHS Program User Forum      
Discussions regarding The DHS Program data and results
Home » Countries » Tanzania » Computing the U5 mortality, infant and neonatal mortality
Computing the U5 mortality, infant and neonatal mortality [message #5632] Thu, 18 June 2015 05:56 Go to next message
Robert is currently offline  Robert
Messages: 18
Registered: April 2015
Location: Dar Es Salaam, Tanzania

Hi Everyone

I have generated the under five mortality, neonatal and infant mortality in stata using the Tanzania DHS for 2004. However, the confusion is the variables that i am using. In the DHS there is b5, b6 and b7. The b5 gives the variable when the child is alive or not and b6 gives the age at death of the child reported in the questionnaire months while b7 gives age at death of the child in completed months which is gives a calculated age at death from the reported information.

These are the commands that I have been able to generate. The question is did I do the right commands and the right variables? Could someone assist?

gen age at death=b7
gen u5m=ad
replace u5m=1 if ad<60
replace u5m=0 if ad>=60

ge neonatal_m=1 if ad<1
replace neonatal_m=0 if ad>=1
ge infant_m=1 if ad<=12
replace infant_m=0 if ad>12

Re: Computing the U5 mortality, infant and neonatal mortality [message #6688 is a reply to message #5632] Sat, 27 June 2015 17:09 Go to previous message
Messages: 290
Registered: March 2013
Senior Member

That looks OK, but one thing to note is that, for each calculation of mortality rate, you need to only include children who have reached the relevant age. By that I mean, to compute the U1 mortality rate, you have to drop all children who are not yet 1 - as I "like" to say (meaning hate to say), a 6 month old baby has 6 more months in which they COULD die, so we don't know if they are a 0 or a 1 until they reach 12 months.

You may also want to limit your analysis to children born in the last X-number of years (say 10 years old - 120 months - in my example below )

One other little thing - in your coding, this might be a little cleaner, especially if there are missing and/or weird values to your variables:

*finding the children who have died before age 12 months
gen u1 = 0
replace u1=1 if ad<12 & b5==1
*the b5 is just for confirmation - if "ad" exists, b5 should be 1.

*finding the right group of births to compute the mean proportion of deaths among
gen psuedo_age_at_survey = v008-b3
*you need to know the age of the living children, and how long it has been since the dead children were born (since they don't have an "age" in the survey, but you need to be able to exclude those who would not be 12 months yet, because you don't have a comparable group of not dead children there, as stated above).

*insert your "svyset" commands here to set strata, PSUs, and weights.

svy: mean u1 if pseudo_age_at_survey>12 & psuedo_age_at_survey<120

Previous Topic: Testing for the pool-ability of DHS data
Next Topic: Creating strata for Tanzania 2010 women's questionnaire
Goto Forum:

Current Time: Tue Apr 24 16:50:24 Eastern Daylight Time 2018