Calculating*Pediatric Blood Pressure Z-scores

[Patrick Abi Nader]

Hello All,

I am looking for a Stata user written command that I can use to calculate blood pressure z-scores for adolescents based on the following NIH report. I found an R macro and the following 2012 Stata thread with no answers. Has anyone developed a Stata package that I can use?

Thank you for your time,
Patrick

[Phil Bromiley]

I don't know of one, but it is easy to do.

egen sd=sd(x)
g z=x/sd

[Clyde Schechter]

Phil Bromiley I think you misinterpreted the question. He is not looking to standardize the BP variable in his own data. Rather, the NIH reference he links to as tables of a reference population distribution for pediatric blood pressures and I believe he wants to convert his sample of blood pressure measurements to z-scores relative to that distribution. Unfortunately, the distribution is presented there as a series of age- and sex-specific percentiles, and there are no means or standard deviations provided.

Actually, in a medical or epidemiological context, when people refer to z-scores they usually mean scores expressed as z-scores in a standard population distribution that has been widely accepted as a reference. The use of z-scores calculated by standardizing in your own sample is actually pretty uncommon in our discipline (except in psychiatry/psychology!)

For BMI there is a nifty user-written -egen- function, -zanthro()- which does this, and actually gives a choice among several generally accepted reference populations. But I'm unaware of anything in Stata that does this for blood pressure. (Nor for bone density scores, another variable commonly used this way.)

[Patrick Abi Nader]

Hello Phil and Clyde,

Clyde thank you for clarifying my question. You understood me correctly. Do you think I can take the NIH age and sex specific percentiles and match them to my data set? My end goal is to classify individuals into prehypertensive or hypertensive based on specific perentile cut offs 90th to 95th (prehypertensive) and >95th hypertensive. In my data set have diastolic blood pressure, systolic blood pressure, sex, and age.

Thank you
Patrick

[Clyde Schechter]

Yes, I would sit down with the table in the report and manually create a new Stata data set with four variables: age, sex, pctile90, and pctile95 (the latter being the 90th and 95th percentiles of sex-specific BP for age). Then I would -merge- that data set with your original data, so now each observation in your data is associated with its corresponding 90th and 95th percentiles. Then from that point doing the classification is easy enough.