Computing ARI [message #23717] |
Tue, 16 November 2021 20:21 |
Michaelo
Messages: 15 Registered: October 2017
|
Member |
|
|
Dear DHS team
I am exploring symptoms of Acute respiratory infections for some countries in SSA. Many thanks for the queries shared by dhs users on this forum on the topic and also to the dhs team for sharing with us the links to the GIthub syntax and codes. They ve been helpful resources in creating this syntax for ARI symptoms in the 2 weeks before the survey. The results I had matched with the report in Ghana 2014(table 10.5) and Kenya) 2014.(table 10.4 However, I am unable match the results for Nigeria(table 10.5) and Angola.
I would therefore appreciate any help to rectify the issue. Thanks
Michaelo
Please kindly find the details
dataset used
1. GHKR72FL
2. NGKR7BFL
3. KEKR72FL
4. AOKR71FL.
Data was weighted using V005
From user forum responses and github I used
do if b5<>0.
compute ch_ari=0.
if h31b=1 & (h31=2) & (h31c=1 | h31c=3) ch_ari=1.
end if.
Results 1 using GH dataset Yes 193 (3.6 %)
Frequency Percent Valid Percent Cumulative Percent
No 5238 92.0 96.4 96.4
Yes 193 3.4 3.6 100.0
Total 5431 95.4 100.0
Missing System 264 4.6
Total 5695 100.0
2.Results 2 using NG dataset (Yes 2.3% but 2.6 in report)
ARI symptoms in the 2 weeks before the survey
Frequency Percent Valid Percent Cumulative Percent
Valid No 30159 88.2 97.7 97.7
Yes 722 2.1 2.3 100.0
Total 30881 90.3 100.0
Missing System 3311 9.7
Total 34193 100.0
3 Results using AO dataset
ARI symptoms in the 2 weeks before the survey
Frequency Percent Valid Percent Cumulative Percent
Valid No 12319 92.2 97.2 97.2
Yes 350 2.6 2.8 100.0
Total 12669 94.9 100.0
Missing System 688 5.1
Total 13356 100.0
The KE dataset
Yielded similar report in the DHS report indicating
1582(8.5%) yes
MO
[Updated on: Tue, 16 November 2021 20:25] Report message to a moderator
|
|
|
Re: Computing ARI [message #23725 is a reply to message #23717] |
Wed, 17 November 2021 13:20 |
Shireen-DHS
Messages: 140 Registered: August 2020 Location: USA
|
Senior Member |
|
|
Hello,
Thank you for your question.
I have used the code provided on GitHub in the CH_ARI_FV syntax file and was able to match the final report table 10.5 for the Nigeria and Angola surveys you mention. The code you have used is not exactly correct since you added (h31=2). As indicated in the GitHub code, the code to construct the ARI variable differs if the variable h31c is included or not in the survey. So you need to check for this. If h31c is present, you do not use h31. See: https://github.com/DHSProgram/DHS-Indicators-SPSS/blob/maste r/Chap10_CH/CH_ARI_FV.sps
* if h31c is present and not empty.
do if h31c_included = 1 and b5 <> 0.
compute ch_ari=0.
if h31b=1 & (h31c=1 | h31c=3) ch_ari=1 .
else if h31c_included <> 1 and b5 <> 0.
compute ch_ari=0.
if h31b=1 & (h31=2) ch_ari=1.
end if.
For both these surveys h31c is present so you should have used the first part of the code to construct the ARI variable, i.e.:
do if b5<>0.
compute ch_ari=0.
if h31b=1 & (h31c=1 | h31c=3) ch_ari=1
You also should check the main file for this chapter because you need to see if you should calculate the child's age based on b19 if it is present in the dataset.
Thank you.
Best,
Shireen Assaf
The DHS Program
|
|
|