Scenario: Need to take some tests for a patient to get them onto a drug trial for asthma meds. In order to get through they have to have a lung function that is not too good, and not too bad. How this is determined is by comparing their results to a chart based on their height, weight, age and weight as a % predicted value.
Problem: the software is American, and the race categories they use are Caucasian, African American, Hispanic or Latino, (East) Asian, Southeast Asian or Other. This is troublesome as this can skew the chart by a few %, introducing error I'd rather do without. White British & White Other have different groupings, and different from American Caucasian. Afro-Caribbean & African both tend to do better relative to African American. This can get even more skewed by self-selection in smaller American demographic groups, e.g. the caucasian chart tends to do better because it grabs more
Most asian patients we see are south asian, not east. One day south asian patient comes in with well-controlled asthma. There's no south asian option. As he's on the border of the "too healthy" line, depending on what chart we use for him his predicted % value could include or exclude him.
LW: ...We could put you under "other"
P: what about asian?
LW: their asian is east asian
P: I won't forgive you if you put me as east asian
LW: we could put you as caucasian. You'd definitely pass with caucasian
P: you've gone too far
Nurse: you want to try southeast asian?
P: do southeast asians have good lungs?
LW: we'll find out
P: I'll allow southeast asian. It's the closest one to me