Home » Topics » Nutrition and Anthropometry » Mozambique: Standard DHS, 2022-23, KR Dataset (nt_mdd, nt_mmf, nt_mad estimate does not match the report)
| Mozambique: Standard DHS, 2022-23, KR Dataset [message #29674] |
Fri, 19 July 2024 08:04  |
Bright Richfield
Messages: 3
Registered: January 2021
|
Member |
|
|
Dear DHS
Hope you are well.
I am having difficult time replicating nt_mdd, nt_mmf, nt_mad estimated in the https://dhsprogram.com/publications/publication-FR389-DHS-Fi nal-Reports.cfm
This is Mozambique: Standard DHS, 2022-23, KR Dataset.
More specifically, table 11.8 last three columns to the left of final column (Page 269).
Picture is uploaded. The denominator seems right.
Can you confirm whether DHS is able to replicate those values? Mine is slightly off.
If so, I must something wrong. Here are the steps I took.
After executing https://github.com/DHSProgram/DHS-Indicators-Stata/blob/mast er/Chap11_NT/DHS8/!NTmain.do,
I ran https://github.com/DHSProgram/DHS-Indicators-Stata/blob/mast er/Chap11_NT/DHS8/NT_IYCF.do.
The only change I made was to add replace nt_frtveg=1 if food1==1 in line 264 to add country specific food to the above do file.
This is based on line 213 which is based on variable v414wa. v414wa is amaranth leaves, cacana leaves, spinach, broccoli or moringa leaves
Even without this addition/change, nt_mdd, nt_mmf, nt_mad still does not match. I have attached the outputs.
Can you kindly advise?
Thanks so much.
-
Attachment: Pic 1.JPG
(Size: 140.59KB, Downloaded 30 times)
-
Attachment: nt_mdd.JPG
(Size: 80.55KB, Downloaded 36 times)
-
Attachment: nt_mmf.JPG
(Size: 82.26KB, Downloaded 29 times)
-
Attachment: nt_mad.JPG
(Size: 74.74KB, Downloaded 33 times)
|
|
|
|
|
|
| Re: Mozambique: Standard DHS, 2022-23, KR Dataset [message #29747 is a reply to message #29695] |
Tue, 30 July 2024 10:16  |
Bright Richfield
Messages: 3
Registered: January 2021
|
Member |
|
|
HI DHS
Thanks for taking the time to respond. Thanks for CSPro text file. Unfortunately, I was not able get what I needed, but I am making some progress to pull my own weight.
Ok. Back to Table 11.8 for Mozambique: Standard DHS, 2022-23, KR Dataset.
I was able to replicate nt_mdd (minimum dietary diversity) for (1) breast-fed, (2) non-breastfed and (3) both. Yay!
I was able to replicate nt_mmf (minimum meal frequency) for (1) breast-fed but not in the non-breastfed group.
https://dhsprogram.com/Data/Guide-to-DHS-Statistics/Minimum_ Dietary_Diversity_Minimum_Meal_Frequency_Minimum_Milk_Feedin g_Frequency_and_Minimum.htm
According to the new guideline noted above, I think there might be something I am not figuring out for non-breastfeeding group. Below is the rule:
2) A minimum meal frequency during the previous day or night of:
a) 2 or more solid, semi-solid or soft feeds for breastfeeding children age 6-8 months, or
b) 3 or more solid, semi-solid or soft feeds for breastfeeding children age 9-23 months; or
c) 4 or more solid, semi-solid or soft food or milk feeds for non-breastfeeding children age 6-23 months where at least one of the feeds must be a solid, semi-solid, or soft feed.
I think it is c) that may be the problem. As for a) and b), I was able to replicate that because nt_mmf is correct in table 11.8.
The word of c) or for non-breastfeeding children is bit tricky to interpret and I tried many combination.
Below is the code from the DHS Github site. I am only pasting variables and lines that are necessary.
https://github.com/DHSProgram/DHS-Indicators-Stata/blob/mast er/Chap11_NT/DHS8/NT_IYCF.do
//Fed milk or milk products
gen totmilkf = 0
replace totmilkf=totmilkf + v469e if v469e<8
replace totmilkf=totmilkf + v469f if v469f<8
replace totmilkf=totmilkf + v469x if v469x<8
gen nt_milkfeeds= ( totmilkf>=2 | m4==95) if inrange(age,6,23)
replace nt_milkfeeds=. if m4==95
label values nt_milkfeeds yesno
label var nt_milkfeeds "Non-breastfed child given at least two milk feeds in day/night before survey- youngest child age 6-23 months"
//Min meal frequency
gen feedings=totmilkf
replace feedings= feedings + m39 if m39>0 & m39<8
gen nt_mmf = (m4==95 & inrange(m39,2,7) & inrange(age,6,8)) | (m4==95 & inrange(m39,3,7) & inrange(age,9,23)) | (m4!=95 & feedings>=4 & inrange(age,6,23))
replace nt_mmf=. if age<6
label values nt_mmf yesno
label var nt_mmf "Child with minimum meal frequency- youngest child 6-23 months"
Here may be where the problem is: gen nt_mmf = (m4==95 & inrange(m39,2,7) & inrange(age,6,8)) | (m4==95 & inrange(m39,3,7) & inrange(age,9,23)) | (m4!=95 & feedings>=4 & inrange(age,6,23))
More specifically, the latter third: ...... | (m4!=95 & feedings>=4 & inrange(age,6,23))
. tab nt_mmf [iw=wt] if m4!=95
nt_mmf | Freq. Percent Cum.
------------+-----------------------------------
0 | 493.75178 80.61 80.61
1 |118.7828269 19.39 100.00
------------+-----------------------------------
Total | 612.534607 100.00
. tab nt_mmf [iw=wt]
nt_mmf | Freq. Percent Cum.
------------+-----------------------------------
0 | 1,951.5073 72.90 72.90
1 | 725.393534 27.10 100.00
------------+-----------------------------------
Total | 2,676.9009 100.00
The nt_mmf overestimate by more than 2.1% for non-breastfed group and about 0.5% for overall.
I tried below to account for where at least one of the feeds must be a solid, semi-solid, or soft feed for non-breastfed group.
gen nt_mmf1 = (m4==95 & inrange(m39,2,7) & inrange(age,6,8)) | (m4==95 & inrange(m39,3,7) & inrange(age,9,23)) | (m4!=95 & feedings>=4 & inrange(age,6,23) & inrange(m39,1,7))
I made the changes on my own based on recommendation below which is also from DHS website.
2) A minimum meal frequency:
a) Defined for breastfed children as receiving solid or semi-solid foods
b) at least twice a day for infants 6-8 months (m4 = 95 & b19 in 6:8 & m39 in 2:7) or
c) at least three times a day for children 9-23 months (m4 = 95 & b19 in 9:23 & m39 in 3:7)
d) Defined for non-breastfed children age 6-23 months as receiving solid, semi-solid, or soft foods or milk feeds at least four times a day (m4 ≠ 95 & total milk feeds (sum of v469e (if in 1:7) plus v469f (if in 1:7) plus v469x (if in 1:7 and v413a = 1) >= 2) plus solid feeds (m39 if in 1:7) >= 4) where at least one of the feeds must be a solid, semi-solid, or soft food (m39 in 1:7)
. tab nt_mmf1 [iw=wt] if m4!=95
nt_mmf1 | Freq. Percent Cum.
------------+-----------------------------------
0 | 494.561481 80.74 80.74
1 | 117.973126 19.26 100.00
------------+-----------------------------------
Total | 612.534607 100.00
. tab nt_mmf1 [iw=wt]
nt_mmf1 | Freq. Percent Cum.
------------+-----------------------------------
0 |1,952.31705 72.93 72.93
1 | 724.583833 27.07 100.00
------------+-----------------------------------
Total | 2,676.9009 100.00
The % is slightly better than before.
So here is the request. Can you confirm below? More specifically, just the part where it begins from m4!=95 for non-breast fed group.
gen nt_mmf = (m4==95 & inrange(m39,2,7) & inrange(age,6,8)) | (m4==95 & inrange(m39,3,7) & inrange(age,9,23)) | (m4!=95 & feedings>=4 & inrange(age,6,23))
I think the wording of the new guideline may not have caught on with the community.
Thanks so much.
|
|
|
|
| Re: Mozambique: Standard DHS, 2022-23, KR Dataset [message #29767 is a reply to message #29747] |
Wed, 31 July 2024 11:49 |
Janet-DHS
Messages: 809
Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member, Tom Pullum:
I admire your attention to detail. You have obviously put a great deal of effort into understanding the indicator and the coding. I too am very detail-oriented. It can be very frustrating when an indicator is defined with multiple components that are essentially arbitrary. I have participated in many meetings where complex definitions of indicators are developed, and will just say that the people who come up with these complex definitions are NOT the people who have to write the computer code!
I believe you are questioning whether component c of the indicator, that is,
c) 4 or more solid, semi-solid or soft food or milk feeds for non-breastfeeding children age 6-23 months where at least one of the feeds must be a solid, semi-solid, or soft feed.
is correctly captured by the last part (in bold) of the "gen" command below:
gen nt_mmf = (m4==95 & inrange(m39,2,7) & inrange(age,6,8)) | (m4==95 & inrange(m39,3,7) & inrange(age,9,23)) | (m4!=95 & feedings>=4 & inrange(age,6,23))
My response would be that it is not correct, because the condition "at least one of the feeds must be a solid, semi-solid, or soft feed" is not included. You say that this condition would be represented by "(m39 in 1:7)" and I agree; "& inrange(m39,1,7)" should be included in the bolded section above. I will notify others on the DHS analysis team that the GitHub programs for table 11.8 needs to be corrected to include that condition.
In the CSPro program for table 11.8, which you have, there is the following around line 60: "if minfeedn then nbrfmin = 3;" . minfeed is defined earlier in the CSPro code. It is 1 if m39 is 1 to 7, and 0 otherwise.
To summarize, as I see it, Table 11.8 was correctly constructed; the CSPro code matches with the definition of the indicator. The only error is in the GitHub code, which does not include the m39 condition in the bolded segment. I will try to have that inconsistency fixed. Thank you for uncovering it.
|
|
|
|
Goto Forum:
Current Time: Fri Aug 30 07:11:47 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
