Mar 072014

CC BY-NC-SA 2.0 - flickr user tyfn - Thanks for the image and hope you're recovered!This study started with good intentions: To try to clarify “normal” laboratory results for various blood tests for transgender people. Definitely need data on that. The current common wisdom, as I understand it, is to use the reference ranges for physiology. So according to that thinking, for a trans man you use “female” ranges before testosterone and “male” ranges after testosterone. But I haven’t seen much research verifying that this is the best way to go.

Sometimes articles come to my attention from unexpected places. This one is not from any of my common sources, like the Journal of Sexual Medicine or the Archives of Sexual Behavior. This is from the American Journal of Medicine, a journal I had no experience with before now. And it’s going to be a difficult one to break down without me sounding highly critical. So let’s just get this out of the way before I go further: I think this was a badly done study, to the point where its results and conclusions are unreliable. There. Now that that’s out of the way, let’s look at what they did…

These researchers pulled clinical data from 55 trans women in two clinics – one in California, the other in Georgia. These women had been on “hormone therapy” for at least 6 months, up to 10 years. Laboratory results which were commonly checked included:

  • liver function tests (ALP, ALT, AST)
  • renal function tests (creatinine, BUN)
  • hemoglobin, hematocrit
  • lipids (total cholesterol, triglycerides, LDL, HDL)
  • sodium and potassium

But these two clinics used two different labs. So instead of using the standardized reference ranges, they compared the transgender participant values to… the blood values of 20 “healthy” cis men and 20 “healthy” cis women.

Here’s what they found:

  • No statistical difference between trans women, cis women, and cis men on these tests: ALT, AST, BUN, HDL, total cholesterol, sodium
  • Trans women’s values were closer to cis female than cis male ranges on these tests: hematocrit, hemoglobin, potassium
  • Trans women’s values were closer to cis male than cis female ranges for these tests: ALP, creatinine
  • Trans women’s values were higher than either cis male or cis female ranges for triglycerides

BUT I checked the reference ranges for these tests. Not all of them are actually different between cis men and cis women! The two I know are are hematocrit and hemoglobin. This is because testosterone stimulates production of red blood cells! It’s why trans men should get tested regularly – too many red blood cells can lead to clotting. There’s a little evidence that ideal HDL levels may vary between cis men and cis women, but I only found one source for that.

Even more important – all of these laboratory results were within “normal” reference ranges. I checked. Reference ranges do vary from lab to lab, but there are standards. And the researchers never correlated any of these results with clinical results. Did that very very slight difference in creatinine levels actually affect anyone’s health? Dunno. But probably not, since it’s still within “normal”!

This leads to other problems I have with the methodology of this study….

  • “Hormone therapy” was never defined. These women could have been on spironolactone or not, on any form of estrogen (or not)… both factors do affect blood levels! Spironolactone has the potential to drastically affect potassium.
  • Whether the trans women participants had their gonads or not was not addressed. This is also crucially important for blood levels
  • Estrogen and testosterone levels were not reported. Same objection as above.
  • Other confounding variable, such as drug use, BMI, and activity level were not reported
  • “Healthy” was never defined for the controls
  • Whether the trans women in this study were “healthy” was never declared

Many of these variables are known to be more common in trans women than in the cis population (e.g., high BMI, drug use, low exercise levels). Given a sample size of only 55, I’m highly suspicious that the observed differences in this study were either noise or due to those other variables. I am left staring at the paper in baffled bewilderment, struggling to understand how this got published. Good intent, but badly executed.