Gender

"There are more intersex people than there are people over a hundred years old."

This sentence was uttered during a discussion about a possible web app that could help to diagnose a disease that increases in prevalence with the patient's age and is more prevalent among females than males, when medical experts and app designers considered the question of how many categories should be used to elicit patient age and gender. The diagnosticians' instinct was to create categories like the 8 decade-long bins (shown at right), but to brush aside questions about whether subtleties about gender should likewise be elicited.

In designing the app, the solution for capturing the variation in age is of course to just ask the patient's age or year of birth, but this approach is not available for eliciting gender for intersex or gender-nonconforming patients. As a matter of basic human respect, it seems that an app ought to allow for answers other than simply "M" or "F". The need for such flexibility rises still more in cases where gender identity makes a quantitative difference in a medical diagnosis. In the case of this particular disease, it is not clear whether such flexibility would help with the diagnosis. This is a result of the larger practical problem which is that the epidemiological and etiological characteristics for people whose gender does not fit neatly in the two typical categories have usually not been studied. This general problem is not restricted to gender minorities. Racial and ethnic minorities, children and even women (who are in the majority) have been traditionally underrepresented in medical and public health studies. This is a serious problem because, as researchers have recognized at least since criticism of the Framingham Study, evidence-based medicine depends critically on the demographic and medical similarity between the research population and the patients at hand. This is, of course, an instance of the general problem in inferential statistics known as the reference class problem.

So is it true that there are more intersex people than there are people over a hundred years old? It seems the answer seems to be yes. Crude calculations (below in green) based on prevalence estimates available on the web suggest there are between twice and ten times as many.

Prevalence of people living to be over 100 years of age

0.00028 = 28/100000 // United States (https://en.wikipedia.org/wiki/Centenarian)

0.00021 = 21.5/100000 // United Kingdom (https://en.wikipedia.org/wiki/Centenarian)

0.00029 = 97000/333000000 // United States (https://www.weforum.org)

[ 0.00008, 0.00037] = [8,37]/100000 // Wales has the highest prevalence in the UK (https://www.ons.gov.uk)

Prevalence of people not conforming to typical gender dimorphism

[ 0.0022, 0.0036] = 1/[2000,5000]+1/[500,1000]+1/1000+1/[18000,50000] // karyotypes: X, XXY, XYY, XXXY (https://www.mvorganizing.org)

[ 0.0005, 0.00067] = 1 / [1500,2000] // intersex noticed at birth (https://en.wikipedia.org/wiki/Intersex)

0.017 = 1.7% // atypical chromosomal, gonadal, genital, sexual development (https://en.wikipedia.org/wiki/Intersex)

Other references

Colyvan, M. and Regan, H.M. 2007. Legal decisions and the reference-class problem. International Journal of Evidence and Proof 11(4): 274–285. http://www.colyvan.com/papers/legaldec.pdf

Colyvan, M., Regan, H.M., and Ferson, S. 2001. Is it a crime to belong to a reference class? Journal of Political Philosophy 9(2):168–181. Reprinted in H.E. Kyburg, and M. Thalos, (eds.). 2003. Probability is the Very Guide of Life, Chicago: Open Court, pages 331347. http://www.colyvan.com/papers/shonubi.pdf

David Gray. 2009. Risk assessment gone mad: the rise of risk evaluation and mass public deception. British Journal of Cardiology 16: 117–18. https://bjcardio.co.uk/2009/05/risk-assessment-gone-mad-the-rise-of-risk-evaluation-and-mass-public-deception/