Estimates for exclusive BF and early initiation of BF not matching with DHS report [message #14018] |
Sat, 03 February 2018 07:13 |
dgodha
Messages: 44 Registered: November 2016 Location: India
|
Member |
|
|
Hello,
My exclusive breastfeeding (EBF) estimates are not matching those in the DHS reports for multiple countries (except for Bangladesh)- varying from discrepancies in the proportion to that in the sample size. I have used the following Stata code:*use AFKR70FL.DTA/ NPKR7HFL.DTA/ PKKR61FL.DTA/ IAKR71FL.DTA
*Keep live children under 2
gen age=v008-b3
keep if age<24 & b5==1
* Child lives with respondent, not elsewhere
keep if b9==0
bysort v001 v002 v003: egen minbidx=min(bidx)
* Need to drop those that are bidx==2 and minbidx==1
drop if bidx>minbidx
** Exclusive breast feeding
gen water=0
gen liquids=0
gen milk=0
gen solids=0
gen breast=0
gen bottle=0
*TO DETERMINE IF CHILD IS GIVEN WATER, SUGAR WATER, JUICE, TEA OR OTHER.
replace water=1 if (v409>=1 & v409<=7)
* IF GIVEN OTHER LIQUIDS
foreach xvar of varlist v409a v410 v410* v413* {
replace liquids=1 if `xvar'>=1 & `xvar'<=7
}
replace liquids=1 if v412c>=1 & v412c<=7
* IF GIVEN POWDER/TINNED milk, FORMULA OR FRESH milk
foreach xvar of varlist v411 v411a v412 v414p {
replace milk=1 if `xvar'>=1 & `xvar'<=7
}
* IF STILL BREASTFEEDING
replace breast=1 if m4==95
* IF WAS EVER BOTTLE FED
replace bottle=1 if m38==1
*IF GIVEN ANY SOLID FOOD
foreach xvar of varlist v414* {
replace solids=1 if `xvar'>=1 & `xvar'<=7
}
replace solids=1 if v412a==1 | v412b==1
gen diet=7
replace diet=0 if water==0 & liquids==0 & milk==0 & solids==0
replace diet=1 if water==1 & liquids==0 & milk==0 & solids==0
replace diet=2 if liquids==1 & milk==0 & solids==0
replace diet=3 if milk==1 & solids==0
replace diet=4 if milk==0 & solids==1
replace diet=5 if milk==1 & solids==1
replace diet=6 if breast==0
gen ebf=0
replace ebf=1 if diet==0
lab de A 0 "No" 1 "Yes"
la val ebf A
lab var ebf "Exclusive BF"
The estimates I am getting for different countries is as follows:** AFGHANISTAN - Matching Table 11.2, page 192 for EBF in children 0-5 months **
ta ebf if age<6 [iw=v005/1000000]
/*43.36% & 3068 vs. 43.3% & 3095 in DHS report*/
** NEPAL - Matching Table 11.3, page 242 for EBF in children 0-5 months **
ta ebf if age<6 [iw=v005/1000000]
/*68.35% & 421 vs. 66.1% & 445 in DHS report*/
** PAKISTAN - Matching Table 11.3 for EBF in children 0-5 months **
ta ebf if age<6 [iw=v005/1000000]
/*37.76% & 1151 vs. 37.7% & 1164 in DHS report*/
** MADHYA PRADESH, INDIA - For Matching Table for EBF in children 0-5 months **
ta ebf if age<6 & v024==19 [iw=sv005/1000000]
/*58.27% & 2044 vs. 58.4% & 2047 in DHS report*/
ta ebf if age<4 & v024==19 [iw=sv005/1000000]
/*67.13% & 1245 vs. 67.3% & 1255 in DHS report*/
ta ebf if age>=6 & age<=9 & v024==19 [iw=sv005/1000000]
/* 18.38 % & 1669 vs. 18.6% & 1696 in DHS report*/
ta ebf if age>=12 & age<=15 & v024==19 [iw=sv005/1000000]
/*5.38% & 1522 vs. 5.4% & 1556 in DHS report*/
ta ebf if age>=12 & age<=23 & v024==19 [iw=sv005/1000000]
/*3.55% & 4226 vs. 3.6% & 4531 in DHS report*/
In case of breastfeeding within one hour, my estimates do not match for Nepal though they match for the other countries. I have used the following code:]*Finding the youngest child for each mother
bysort v001 v002 v003: egen minbidx=min(bidx)
*Need to drop those that are bidx==2 and minbidx==1
drop if bidx>minbidx
*Child's age
gen age=v008-b3
** Matching Table 11.1, page 241 for breast feeding within 1 hour of birth in children under 2**
gen ei=0 if age<24
replace ei=1 if m34<=100 & age<24
svy:tab ei v025, col
Estimates:
/*55.07% & 1930 vs. 54.9% & 1978 in DHS report*/
/*Urban: 57.07% & 1034 vs. 57% & 1062 in DHS report*/
/*Rural: 52.76% & 897 vs. 52.5% & 916 in DHS report*/
I have already looked at the forum threads and the DHS guide multiple times but have not been able to find the problem. I will appreciate help on this.
Thank you
Deepali
|
|
|
|
Re: Estimates for exclusive BF and early initiation of BF not matching with DHS report [message #15133 is a reply to message #15131] |
Thu, 07 June 2018 06:45 |
dgodha
Messages: 44 Registered: November 2016 Location: India
|
Member |
|
|
Yes please, I have been eagerly waiting for your response. I have been in touch with Bridgette and she had assured me of a response.
I also decided to try other ways to check my estimates. I think the problem is in the weighting. I say this because I tried to calculate simple estimates on background characteristics like age group, marital status as shown in Table 3.1 on Page 51 of the report. My proportions and unweighted counts were an exact match but not the weighted counts- they were way off.
Thank you
Deepali
|
|
|
|
|