The DHS Program User Forum      
Discussions regarding The DHS Program data and results
Home » Topics » Malaria » Reproduce household ownership of ITN's using household member recode files
Reproduce household ownership of ITN's using household member recode files [message #11308] Thu, 01 December 2016 14:06 Go to next message
galka is currently offline  galka
Messages: 1
Registered: December 2016
Location: Basel, Switzerland
Member
Hello,

I would like to calculate a range of indicators related to coverage and use of malaria interventions. Some of these are customarily reported at household level, others for children under 5 etc. To avoid downloading multiple datasets I was hoping to be able to reproduce the statistics using just Household Member Recode files. This, however, proved tricky.

I am able to replicate "Proportion of households with at least one mosquito net: Any mosquito net" in both HR and PR files matching DHS final report. When it comes to "Proportion of households with at least one mosquito net: ITN" I can only replicate in HR files, estimates in PR are off.

This seems to be the case in most surveys, the difference in magnitude, however, varies by country.
Here, for example, I tabulate data from Ghana 2014 DHS survey:

/* ======= using HR file */
use ghhr71fl.dta, clear;

gen anynet_hh= (hml1>0);
egen nmbitn_hh= rowtotal(hml10_*);
gen itnnet_hh= (nmbitn_hh>0);

sum anynet_hh itnnet_hh nmbitn_hh [w= hv005];

Variable | Obs Weight Mean Std. Dev. Min Max
-------------+---------------------------------------------- -------------------
anynet_hh | 11,835 1.1835e+10 .6958722 .4600565 0 1
itnnet_hh | 11,835 1.1835e+10 .6834536 .4651484 0 1
nmbitn_hh | 11,835 1.1835e+10 1.332503 1.25887 0 7

/* ======= using PR file */
use ghpr71fl.dta, clear;
gen anynet= (hml1>0) if hml1<.;
by hv001 hv002, sort: egen nmbitn= total(hml10) ;
gen itnnet= (nmbitn>0);
sum anynet itnnet nmbitn [w= hv005] if hvidx==1;

Variable | Obs Weight Mean Std. Dev. Min Max
-------------+---------------------------------------------- -------------------
anynet | 11,835 1.1835e+10 .6958722 .4600565 0 1
itnnet | 11,835 1.1835e+10 .4066891 .4912367 0 1
nmbitn | 11,835 1.1835e+10 1.223365 1.914324 0 17

I also looked within the two files and found that there is disagreement in between the 2 files with respect to ITN ownership.

/* ======= using HR file */
hv001 hv002 hml10_1 hml10_2 hml10_3 hml10_4 hml10_5 hml10_6 hml10_7 nmbitn_hh
1 2 yes 1
1 6 yes 1
1 11 yes yes 2
1 14 yes yes 2
1 19 yes yes 2



/* ======= using PR file */
hv001 hv002 hvidx hml10 nmbitn
1 2 1 0
1 2 2 0
1 2 3 0
1 2 4 0
1 2 5 0
1 6 1 0
1 6 2 0
1 6 3 0
1 11 1 0
1 14 1 0
1 14 2 0
1 14 3 0
1 19 1 0
1 19 2 0
1 19 3 0
1 19 4 0
1 19 5 0
1 19 6 0
1 19 7 0
1 19 8 0
1 19 9 0
1 19 10 0


Please, advise whether and if so - how can household level statistics be replicated in PR files.

Many thanks,
Katya

[Updated on: Thu, 01 December 2016 14:07]

Report message to a moderator

Re: Reproduce household ownership of ITN's using household member recode files [message #11373 is a reply to message #11308] Fri, 09 December 2016 16:23 Go to previous message
Liz-DHS
Messages: 1264
Registered: February 2013
Senior Member
Dear User,
A reply from malaria expert, Dr. Lia Florey:
Quote:

Dear Katya,

Thank you for your question. If you look in the PR file and compare hml10 and hml12 you will see that hml10 is actually representing whether or not an individual slept under an ITN the night before the survey and not representing ownership of an ITN. This is likely the reason for the discrepancies that you observed. I would recommend calculating the household-level variables you need in the HR file. If needed you can then merge those household-level variables into the PR file by hv001 and hv002.

As a warning, you will also not be able to calculate the indicators on IPTp or case management of fever using the PR file. If you have datasets with anemia and parasitemia data those indicators can be calculated using the PR file (and using this file is necessary if you are trying to replicate the indicators from the report).

I hope this is helpful.

Best,
Lia

Previous Topic: Barriers to the utilization of malaria preventive measures in Nigeria among pregnant women - Variabl
Next Topic: How to make and interpreting Bivariate statistics analysis?
Goto Forum:
  


Current Time: Thu Jul 19 13:19:45 Eastern Daylight Time 2018