| Haemoglobin levels reported for DHS India 1998-1999 [message #24895] |
Sun, 31 July 2022 08:57  |
teeb
Messages: 11
Registered: July 2022
|
Member |
|
|
Hi I'm trying to calculate anemia levels from the DHS India 1998-1999 data for women using the IR file. While for the later rounds, anemia is classifed as mild, moderate, severe, and no anemia, the 1998-1999 reports hemoglobin levels in g/dl.
A summary of the variable s902 reveals values range from 0 to 992, with a mean of 117.21 These are imposssibly large values. Is it something to do with the unit of the variable?
Would appreciate any clarification.
|
|
|
|
| Re: Haemoglobin levels reported for DHS India 1998-1999 [message #24905 is a reply to message #24895] |
Mon, 01 August 2022 12:08  |
Bridgette-DHS
Messages: 3038
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
The hemoglobin concentration in the DHS files (hc56 for children in the PR files now, or hw56 in the KR files) includes a factor of 10 to get rid of a decimal point. You should divide the numbers in the data files by 10. For example, 100 is 10.0.
In many surveys there are impossibly large or impossibly small values. The anemia variable (now hc57 or hw57) has 4 categories. Both 1 (severe anemia) and 4 (not anemic) are open-ended. Because there don't seem to be standard upper or lower bounds for plausibility, the extreme values are typically just included in the 1 and 4. I suggest that you compare hc56 and hc57 in the most recent India survey to see whether limits of plausibility were used there, and if so, apply them to the 1998-99 survey.
Five years ago we did a methodological report on hemoglobin measurement in DHS surveys, and it may be helpful: https://www.dhsprogram.com/pubs/pdf/MR18/MR18.pdf .
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
