unknown egen function zanthro()

[Alexa Ayarza]

Hi! I'm using zanthro to get zscores, but the values look between [-3,3]
I used this code to get the height-for-age:

egen haz = zanthro(talla,ha,US), xvar(edad) gender(sex) gencode(male=1, female=2) ageunit(year) nocutoff

This is how it looks:

edad    talla    sex    haz
17    1.76    Hombre    -13.84882
17    1.67    Hombre    -13.84922
13    1.44    Mujer    -18.02584

[Nick Cox]

Yes; it’s community-contributed, so not downloaded when you install Stata. You need to install explicitly.

search zanthro

should identify locations. You may have to rummage around to identify the latest version.

[Sa Li]

Hi I also use the zanthro command to calculate height-for-age z score, and the values between [-4.94,4.93],but i am not very sure the the reference to confirm whether this children is stunted, is HAZ score<-2，then the children can be seen being stunted. Is that correct? also for the referecne of whz and bmiz. Thank you in advance, I would be very appreciate for your reply.

[Nick Cox]

The thread title is unfortunate because posters all know about this function (not a command) and are using it. #3 raises specialist questions. If you don't get a good answer from users of the software, I suggest you email the first author.

[Sa Li]

Thank you, may be i made an inappropriate explanation. Because i am not a native-speaker of English. Thanks for your reply!

[Nick Cox]

Sa Li You didn't choose the thread title, but no real problem either way. Someone else may join in. My guess is that results from this code are extraordinary sensitive to what is regarded as the reference population, but I am not a medical statistician, epidemiologist or medic.

[Sa Li]

Thank you for your reply. Really appreciate that. i will find the thread title. Yes you are right, my reference population is children, the age group is from 0-5. But i was trying to fing the related reference, seems they are quite different in the articles or the WHO reference. so i am very confusing now.

[Clyde Schechter]

I am an epidemiologist. To my knowledge, there is no universally agreed upon definition of stunting. While most people define it as Z < -2, there are various reference distributions that can be used to calculate the z-score. -zanthro()- offers, for babies born at term, the US, UK, and WHO standards. If you are studying only children in economically advanced countries, using the US or UK distribution would be reasonable. But it will identify as stunted many children from poorer countries who are developing normally given the limited access to food in their environment and the endemicity of certain diseases. The WHO standard is more appropriate for those environments, unless you want your definition of stunted to include children who are small primarily because they are being raised in a poor country. Put more categorically, it depends on whether you want to include stunting due to broad environmental effects in your classification, or just biological and localized environmental problems.

[Sa Li]

Hi, really grateful for your reply. i was trying to use the WHO growth standards. The below are the codes to calculate the z scores.

egen zhawho= zanthro(HEIGHT_C,ha,WHO), xvar(age) gender(GENDER) gencode(male=1, female=2)
egen zwhwho = zanthro(WEIGHT_C,wh,WHO), xvar(HEIGHT_C) gender(GENDER) gencode(male=1, female=2)
egen zbawho = zanthro(bmi_C,ba,WHO), xvar(age) gender(GENDER) gencode(male=1, female=2)

now i am trying to define "stunting" "wasting" or "overweight" and calculate the prevalence of child stunting, child wasting, child overweight, so i need to be sure about the reference. Also my age groups are 0-5 years.

[Clyde Schechter]

So, having settled on the WHO reference population, you need only be clear about the cutoffs to define. Stunting, as already mentioned is generally taken to be Z < -2 in height for age. Wasting is generally defined as Z < -2 in weight for height. And overweight is generally defined as Z > 1 in BMI for age.

You cannot use an age group as your age variable with zanthro*. You must specify an actual age: it can be measured in years, days, months, or weeks. If the measurement unit is other than years, you must specify that with the -ageunit()- option.

*I am unaware of any software for these statistics that allows calculation from 5 year age groups. Nor do I see how it would even be remotely possible. The distributions of height, age, and weight change enormously over a five-year period in childhood. I don't see any way to get appropriate results based only on age groups.

[Sa Li]

Thank you so much. Now i totally understand. i am sorry i made a misunderstanding explanation. My age unit is in years like 0, 1, 2, 3, 4, 5 years.

sum age

Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
age | 4,771 2.87424 1.665191 0 5

Thanks a lot!