| Definition of ARI symptoms [message #14864] |
Wed, 09 May 2018 14:40  |
Ezusha
Messages: 1
Registered: May 2018
Location: Ethiopia
|
Member |
|
|
am planning to study the prevalence of symptoms of Acute Respiratory (ARI) from 2016 DHS data for Ethiopia. Unfortunately, I got a problem while the prevalence from the Children's Record (KD) obtained accessed on April 24, 2018. This is because there are different definitions of symptoms of ARI and none of them produced similar results to the prevalence reported in the final report, which was 691 (6.6%).
Annotation of the final report defines Symptoms of ARI include cough accompanied by short, rapid breathing that is chest-related and/or by difficult breathing that is chest-related. Using this definition, though the exact definition could not be computed using the available variables, I used the code: h31 ==2 & (h31b==1 | h31c==1 | h31c==3) and the result was 796. This is higher than the number reported on the final report (Page 177).
On Guide to DHS statistics (Sep 2006, Page 112), the numerator is described as Number of children ill with a cough accompanied by short, rapid breathing at any time during the two weeks preceding the interview. I used this code 'ari=1 if h31 == 2 & h31b==1' and the results were identical with the above code.
When I search on the DHS forum for further explanations I found out that the outcome variable is (Symptoms of ARI) is computed using the following code: ari=1 if h31b==1&( h31c==1 | h31c==3). Nevertheless, the result was 473, which is much less than the reported on the final report. I have recorded 'ari' into missing for b5=0 (not alive), which only affected the denominator. So would anyone tell me which variables and coding should I use and what can I do if the results are still inconsistent with the final report.
|
|
|
|
| Re: Definition of ARI symptoms [message #15153 is a reply to message #14864] |
Fri, 08 June 2018 19:51  |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
A response from Senior Data Processing expert, Ladys Ortiz,
Quote:
Dear user
Your numbers are not matching with the report because you are not weighting the cases. After running some frequencies, I got 691 vs 473 (weighted and unweighted).
FYI: to apply the weight factors you should use variable V005 divided by 1,000,000.
Thanks
Ladys Ortiz
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
