The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » HIV » Replicating HIV testing coverage rates
Replicating HIV testing coverage rates [message #838] Mon, 14 October 2013 16:35 Go to next message
kh8v07 is currently offline  kh8v07
Messages: 2
Registered: October 2013
Location: Southampton
I am interested in the levels of HIV testing response/coverage for men and women at the regional level within countries in sub-Saharan Africa. However so far I have had problems replicating the results displayed in the country reports. I am using SPSS.

The DHS recode manual suggests that the variables I need for this when looking at individual women are v473a or v473b. I also believe that the data should not be weighted, either with the HIV sample weight 'HIV05' or the individual sample weight 'v005'. I am not sure where I am going wrong. Do I need to recode the HIV related variables in some way? Or is it because incomplete interviews have been dropped from the file? The recode manual says that for the result of the individual interview (v015) in flat format data files cases with a result code different than 1 (i.e. incomplete interviews) are dropped from the file. If they have been dropped, am I still able to compute response/coverage rates?

Thanks in advance for any help.

Re: Replicating HIV testing coverage rates [message #893 is a reply to message #838] Wed, 30 October 2013 16:16 Go to previous messageGo to next message
Messages: 1516
Registered: February 2013
Senior Member
Dear User,
I am not a programmer but was looking at the standard tables and it looks as though the variables used are:
Row Variables
HV025 Type of place of residence
HV024 Region
A Total variable for the 15-49 age group (if you're looking at women)
Column Variables
HIVRES1 recodes HA63
HIVRes2 recodes HA65
sex variable
HA63 is the result of measurement
HA65 is result of woman interview

Here is some of the CSPro code which may help you.

They are used as follows:
{ tabulate women for tables 14.1 & 14.2 }
{ as they are based on all eligible women }
isex = 1;
tot1549 = 0;
for i in RECH5_EDT do
box HA63 => hivres1;
1 => 1; { blood taken }
3 => 2; { refused }
2 => 3; { not present }
=> 4; { 4-not tested/5-inconclusive/6-other/missing }

{ if blood taken check if test was successful }
samptake = 0;
if HA63 = 1 then
{ load test data for women and modify hivres1 accordingly }
xline = edit("ZZ9",HA0);
samptake = loadcase( HIVTESTDAT, HV001, HV002, xline );
if samptake then
if !HIV03 in 0:7 then
hivres1 = 4; { test was performed but protocol didn't reach the end }
samptake = 0; { not enough blood will not be included in the denominator }
hivres1 = 4; { blood taken but sample not on data file }

if HV103(HA0) = 1 then { de-facto population }
box HA65 => hivres2; { result of individual interview }
1 => 1; { complete }
=> 2; { other }
v106w = HA66; { education from individual interview }
if HA66 in 8,missing then v106w = 9 endif;
hivage = int( HA1/5 ) - 2; { age in five years }

xtab( t1401 );
Re: Replicating HIV testing coverage rates [message #945 is a reply to message #893] Tue, 26 November 2013 12:54 Go to previous message
kh8v07 is currently offline  kh8v07
Messages: 2
Registered: October 2013
Location: Southampton
Ah, thank you! I was looking at the male and female recode files (merged with the HIV file) separately. This is quite confusing as the HIV variables appear to give the required information for each file. However the variables you have mentioned (HA63 and HA65) are obviously household variables, so I will use these instead! Thanks for your help.

Previous Topic: Svyset in Stata for HIV results and ART use data
Next Topic: Cameroon: Sample domain missing for weights
Goto Forum:

Current Time: Mon Jun 27 13:48:19 Coordinated Universal Time 2022