| Re: Full Birth History for all birth in Preceding 5 years But ANC history for Most Recent Birth [message #20244 is a reply to message #20231] |
Wed, 14 October 2020 18:56  |
Bridgette-DHS
Messages: 3215
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
Yes, v008=b3<60 identifies births in the past five years, but if you use the KR file you will have that limitation. That file is limited to births in the past five years. Yes, in this survey the information about antenatal care is limited to the most recent birth in the past five years. The most recent birth has bidx=1. If you want to relate neonatal mortality to antenatal care, you can just limit to children in the KR file with bidx=1.
You can define a binary outcome that is 0 for survived and 1 for neonatal death, and then do a logit regression of that outcome on neonatal care, etc. However, in many surveys this relationship is weak or is in what seems to be the wrong direction. A difficult pregnancy may lead to more neonatal care but also a higher probability of a death, and therefore a negative relationship between care and survival.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
