The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » General » Change in data values for "Treatment of Diarrhea" indicator?
Change in data values for "Treatment of Diarrhea" indicator? [message #20253] Thu, 15 October 2020 11:36 Go to next message
sahsan is currently offline  sahsan
Messages: 1
Registered: October 2020
Location: Ottawa

I had looked up and downloaded data on "Treatment of Diarrhea: Taken to a Health Facility: Three years preceding the survey" indicator for all countries available in June 2020. However, when I downloaded the same data this week, I found that individual values for some countries have changed even though the source (for example, Bangladesh's DHS 2014) remained the same. As a point of reference, Bangladesh and India's values were 38.7 and 69.3 respectively in the older dataset I downloaded whereas the newer datasets had 77.3 and 78.0 respectively now. I just wanted to verify whether there has been any change in how this indicator is being computed or is it just a mistake from my end (I am fairly confident I looked up and downloaded the data earlier exactly the way I am doing now). Any help will be much appreciated.

Re: Change in data values for "Treatment of Diarrhea" indicator? [message #20296 is a reply to message #20253] Tue, 20 October 2020 15:39 Go to previous message
Shireen-DHS is currently offline  Shireen-DHS
Messages: 45
Registered: August 2020
Location: USA
Hello Sameen,

I think you mean you looked up this indicator on STATcompiler. The current definition for "treatment of diarrhea: taken to health facility" includes all types of treatment except being taken to a traditional healer or traditional practitioner. Older surveys also excluded pharmacy and shops so the Final reports reflect that exclusion and would not match STATcompiler. However, STATcompiler uses the current definition. I computed the indicator using our standardized code from out code library on GitHub. The indicator you are interested in is found in Chapter 10 and specifically in the Stata do file : er/Chap10_CH

When I run the code for the Bangladesh surveys I match STATcompiler so it appears to be correct. You can try this yourself but please read the notes in the main file and the file before running any code. You also need to read our readme file here:

I see there is a large jump in Bangladesh from 34.2% in 2004 to 71.2% in 2007 but it appears to be calculated correctly.

Thank you.

Shireen Assaf
The DHS Program

Previous Topic: Question of STATA
Next Topic: Total number households reported vs v002
Goto Forum:

Current Time: Thu Dec 3 08:52:10 Coordinated Universal Time 2020