The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Mortality » Dropping of Observations (Neonatal Mortality Rate Estimations and Determinants)
Dropping of Observations [message #20099] Fri, 25 September 2020 07:58 Go to next message
shujaat.smc@gmail.com is currently offline  shujaat.smc@gmail.com
Messages: 75
Registered: July 2020
Senior Member
Dear DHS Representative,

I am using PDHS(Pakistan-DHS 2017-18) for assessing neonatal mortality rate and determinants using survival analysis methodology.

I am estimating mortality rate and its determinants in neonates with birth weight>= 2500gms.

In this case can I drop those observation (m19=birthweight) whose weight is less than 2500gms ???

If I drop these LBW observations what will be effect on survey weight V005.

If I cannot drops LBW then how I should declare my data set as survival analysis data so that i got the results only for neonates with birthweight above 2500gms when I am using the reference category of BW 4.5 and above.

Kindly Reply at your earliest.

Best Regards

Dr. Hussain
Re: Dropping of Observations [message #20101 is a reply to message #20099] Fri, 25 September 2020 08:58 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

There is very little you can do with birthweight (m19) in this survey. It is only provided for about 18% of births. The rest are "not weighed" or "don't know". The 18% that include a numerical value have a lot of heaping and questionable extreme values. About 15% of the numerical values are heaped right at 2500g. I would not have a lot of confidence in these numbers.

If you do proceed, I would recommend that initially you construct 5 categories, something like these: m19<2500; m19=2500; m19>2500 but <9996; m19=9996; and m19=9998. Then check for a statistical association between that categorical variable and neonatal or infant mortality.

The interpretation is probably affected by the fact that children for whom a birthweight is given are much more likely to have taken place in a facility, which gives a survival advantage, even if the child was LBW.

You should definitely not drop the LBW children and you do not need to modify the sampling weights.

Survival Data Setting in Stata for Neonatal Mortality Determinants [message #20121 is a reply to message #20101] Sun, 27 September 2020 08:31 Go to previous message
shujaat.smc@gmail.com is currently offline  shujaat.smc@gmail.com
Messages: 75
Registered: July 2020
Senior Member
Dear DHS Representative,

I am very much thankful to you for your quick responses.

I am following an article in which the author has mentioned that "The outcome was neonatal deaths recorded as a binary variable. The age of neonatal death was measured in days and for deaths within 24 hours value of 0.01 days was used".

For generating variables to specify my data set in stata as survival data how should I generate the time variable for neonates and specifying a value of 0.01 for death at the day of birth.

Please also comment on the outcome variable I have generated;

gen neonataldeath=1 if b6<=128
replace neonataldeath=0 if b6>=201

Best Regards

Dr. Hussain
Previous Topic: regression to explore the relationship between neonatal mortality & maternal health seeking behavioo
Next Topic: NMR, PMR, and IMR estimation with syncmrates
Goto Forum:
  


Current Time: Fri Mar 29 09:11:06 Coordinated Universal Time 2024