Create a continuous variable for blood pressure using two variables

[Sigrid Vikjord]

Dear List,

I wonder if any of you know how to make a continuous blood pressure variable from two variables? I have continuous variables for systolic and diastolic pressure: systolic_bp and diastolic_bp. These are expressed by the following ratio: systolic over diastolic pressure (for instance 145/83). I want to create a continuous variable, bp, that looks like this (example):

systolic_bp	diastolic_bp	bp
180	95	180/95
140	90	140/90
137	75	137/75
etc...

Any suggestions?

Best regards,
Sigrid

[Carlo Lazzaro]

Sigrid:
if you want to have the -bp- variable in numeric format, this is, I fear, unfeasible, unless you were comfortable with something like:

gen bp_ratio=systolic_bp/diastolic_bp

I think that the only chance to accomplish what you're after is something along the line of the following toy-example:

. set obs 1
number of observations (_N) was 0, now 1

. g sbp=120

. g dbp=80

. egen sbp_dbp_ratio=concat( sbp dbp), p(/)

. list

     +----------------------+
     | sbp   dbp   sbp_db~o |
     |----------------------|
  1. | 120    80     120/80 |
     +----------------------+

.

Obviously, the derivated variable does not support any calculation.

[Sigrid Vikjord]

Thank you for your reply, Carlos. That's what I was afraid of, I will have to use the two variables separately then.

[Clyde Schechter]

FWIW, I've been doing epidemiology since 1985, and blood pressure plays a role in a large proportion of the projects I work on. Blood pressure is, in my experience, almost always represented as separate numeric variables for systolic and diastolic. It is never used as a string variable containing a slash character: that's clinical notation that's fast to write by hand. But it's useless for analysis. The only circumstances I can recall where a single numeric variable for blood pressure is used is when they calculate (or approximate) a time weighted average of systolilc and diastolic blood pressure, sometimes called mean blood pressure. I've seen that in a few physiology oriented studies, never in clinically based research.

[Sigrid Vikjord]

Dear Clyde,
Thank you for valuable input. I would completely agree with you. I'm doing some cluster analysis where I need two versions of the blood pressure variable, one dichotomous (normal/high) and one continuous. I'm replicating results from another study, and cannot wrap my head around how they've created their variables. I will take your feedback with me when working with the analyses.