The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Tanzania » Maternal age at first birth
Maternal age at first birth Sat, 16 August 2014 04:17
 methodkazaura@yahoo.no Messages: 4Registered: July 2014 Location: Tanzania Member
I am trying to establish the time trends of child mortality (infant and child) among first born in Tanzania.

Why do I find differences in value between:

(a) A variable indicating 'maternal age at first birth' (directly from the data set: (V212) AND
(b) Year the first child were born - Year the mother was born) (B2\$01-V010) = Age of the mother at first birth

Then, what is the 'near' correct age at birth of the 'FIRST BORN'?

Also, aiming to use first born,I think BIDX\$01 refers to the FIRST birth and BORDX\$01 is similar to BIDX0\$? (in the reverse order).

So, looking for the first birth, I should concentrate preferably with BIDX\$01. Am I correct??

Method
Re: Maternal age at first birth [message #2730 is a reply to message #2728] Sun, 17 August 2014 23:28
 Trevor-DHS Messages: 789Registered: January 2013 Senior Member
Dear Method,

I see two problems with your calculating:
1) B2\$01 is the year of birth of the last birth, not the first. The birth history is in reverse chronological order.
2) The calculation of age at first birth in v212 is made in months and years, not just years.

I don't know the term 'near' correct age. Can you explain where you find this?

BIDX is the index of the birth history and BORD is birth order. As you not BORD is in reverse order to BIDX, but with the first entry in the birth history being the most recent birth, not the first.
Re: Maternal age at first birth [message #2732 is a reply to message #2730] Mon, 18 August 2014 00:45
 methodkazaura@yahoo.no Messages: 4Registered: July 2014 Location: Tanzania Member
Dear Trevor,

Thank you for the explanation. However, first, it is now more difficult because I though V212 is the age (years) of the mother at FIRST birth. This is indicated in the variable description as "Age of respondent at 1st birth"; values include, for example 9, 10, 11, ..., 41. Certainly, these can not be age in years and months, unless I am mistaken. Second, does your explanation mean BIDX\$01 refers to the first birth?

Method (Tanzania)
Re: Maternal age at first birth [message #2734 is a reply to message #2728] Mon, 18 August 2014 13:35
 Trevor-DHS Messages: 789Registered: January 2013 Senior Member
Dear Method

v212 is competed age in years, but calculation is not done just by subtracting year of birth or the mother from year of birth of the first child, but uses the months of birth of the mother and child too. For example, if a respondent was born in September 1990, and her child was born in February 2010, we don't just subtract 1990 from 2010 - we use the months too. The completed age of the respondent at first birth would be 19, not 20.

For the births, any of the birth history variables with \$01 on the end relate to the last birth, anything with \$02 relate to the birth before last, etc. Thus the variables containing the first birth will differ from case to case, depending on how many births the respondent had. However, the date of first birth in century month code (CMC) format is stored in v211. Similarly the date of birth of the respondent in CMC format is in v011. Thus you can calculate the age at first birth by subtracting v011 from v211 to get the age at first birth in months, and then divide by 12 to get age at first birth in years:

Age at first birth = int((v011 - v211) / 12). (Depending on your software you would use int or trunc to get the integer part or truncate to a while number).

Re: Maternal age at first birth [message #2736 is a reply to message #2734] Tue, 19 August 2014 02:03
 methodkazaura@yahoo.no Messages: 4Registered: July 2014 Location: Tanzania Member
Dear Trevor,

Thank you again for the quick reply. I certainly got you right concerning AGE AT FIRST BIRTH. Although you said age at first birth = int((v011 - v211) / 12), i came to realize it was int ((v211-v011)/12). That is done and it quite 100% corresponding to V212. Thank for that.

There are two variables (BIDX\$01 and BORD\$01) that link the child's birth. To me, the the former (BIDX\$01) refers to the 'most recent' birth and later (BORD\$01) refers to the 'older' child. The 'older' child is what i consider to be the first birth. I am asking these questions to assess the mortality level (and trend) for the FIRST BORN (who may be the 'oldest' children. Then, I will come to see their survival status (if dead, how old were they?).

Nevertheless, I thank you for your continued support.

Method
Re: Maternal age at first birth [message #2738 is a reply to message #2736] Tue, 19 August 2014 11:12
 Trevor-DHS Messages: 789Registered: January 2013 Senior Member
Hi Method,

you are right about the calculation - I wrote that backwards. It should be int((v211-v011)/12) as you have it.

BIDX\$01 and BORD\$01 are referring to the same child - all birth history variables with \$01 are the same child, and that is the last born child.
BIDX\$01 will be 1 - the index number for the birth history, going from oldest to youngest.
BORD\$01 is the birth order, and, for example, if the woman had 8 children BORD\$01 would be 8.

similarly, for the child before last the variables all end with \$02.
BIDX\$02 will be 2 - the index number in the birth history.
BORD\$02 is the birth order, and, continuing the example from above, for a woman with 8 children it would have the value 7.

If a woman gave birth to 8 children, you will find the information about the first born child in the variables with \$08 on the end.
Similarly if a woman gave birth to 6 children, you will find the information for the first born child in the variables ending with \$06.

I hope this helps to clarify the structure of the data.
Re: Maternal age at first birth [message #13065 is a reply to message #2738] Thu, 14 September 2017 10:27
 gafap Messages: 3Registered: April 2017 Member
Dear Trevor,

May I ask, how do you calculate, for example, age at second birth? (not second to last)

thank you!
best,
g
Re: Maternal age at first birth [message #13195 is a reply to message #13065] Mon, 02 October 2017 18:15
 Trevor-DHS Messages: 789Registered: January 2013 Senior Member
In Stata, you could do something like the following:
```use ZZIR62FL.DTA
rename b*_0* b*_*

gen age_2nd_birth = .
forvalues i = 1/20 {
replace age_2nd_birth = int((b3_`i' - v011)/12) if bord_`i' == 2
}
tab age_2nd_birth```
This code loops through all 20 entries in the birth history and finds the second birth (where bord = 2) and calculates the age at the second birth using the century month code of that birth.
 Previous Topic: Stratification for 1996, 1999, 2004-2005 dataset Next Topic: Difference between key indicator report and final report
Goto Forum:

Current Time: Tue Apr 23 18:49:09 Coordinated Universal Time 2024