Don’t average human diversity
The worst thing that someone could tell me is that I am average, I believed as a kid. That would mean I am not interesting, or worthy of attention, as there is already a pre-existing drawer waiting to store people like me.
Many times since then I cursed my young self’s wish to be different, and realized how easy would it be if I am in fact average.
How great would it be if my body temp is not 35,5 °C, but “normal”, if my food allergies are not existent and ever present embarrassment to my social life? If the way I think is not querdenkerish (I got a Microsoft “querdenker” t-shirt as a proof), life would be easier. For me, and the ones that encounter me. It would be amazing.
But actually, I am average in many things: many, many of them. At least many of the ones I experienced and could quantify to a certain degree.
Like the average-me does, all of us are falling within some sort of average, and at the same time not. And we don’t know it until we investigate, and are familiar with our “normal/healthy” state. Perhaps we are Schroedinger cats of averageness?
MEDICAL AVERAGES
Average in medical terms (officially named “normal” coming from the Normal Distribution) is a value into which a 95% of populations values fall into, with 2,5% being higher and 2,5% being lower than the normal.
It becomes tougher to determine normal once we start digging into the complex, chronic conditions.
Indicators of health and disease are different and they are many, all connected into (more often than not) an invisible web, so when one of the indicators gets out of the balance, some of the others will lose their spot, and shift too. The problem is that shifts, although detected through technology, will not be large enough to be considered as symptoms of a condition. Not until one is out of 95% confidence rate. For that, it might take time.
There is a problem with such an approach, as it heavily depends where are you positioned in this 95% bell curve. Are you very left, and need to shift really a lot to get out to the right 2,5%? Or… are you very right, just at the border of not normal, and might cross that border just as easily as Gregory Peck did in The Scarlet and the Black. The shift is significantly different in the two cases above, but it will be read in the same way.
So, where lies the problem? In the averages. When talking about our changing health, we should rather be looking at the size of individual shifts in diverse indicators instead of trying to squeeze us all under the bell curve.
At the point when diseases become increasingly more complex, we are at a high risk of missing the appropriate diagnosis, thanks to two major things: (I) ignoring symptom diversity: being evaluated for an average on each of the symptoms that is currently medically accepted for that specific disease, while ignoring the set of symptoms we actually display, and (II) not having specific enough prediction charts for our demographics.
At the start of many complex diseases, symptoms are subtle, and that is the third problem when trying to have a good disease detection. What we have today is a low-fidelity diagnostics, and it is bound to miss many of the early signs of complex diseases.
Us, humans are diverse, and this should be taken more in consideration when diagnosing, however complex and painful and long and confusing it is. It should be done!
WHERE DO MEDICAL AVERAGES COME FROM?
From test subjects in clinical research and studies.
Until 20 years ago, about 90% of clinical research was done using middle aged white males as a test group. Even in the cases when other demographics were included in the research, their diversity contributing to the results was largely ignored.
So, quite some of the prediction charts in our doctors offices might still be coming from that very research. And that is a bad way to address human diversity: diversity in terms of health, symptoms of the disease, treatment sensitivity or side effects to the later one.
Understandably for the medicine 50 or more years ago, it was very hard to serve diverse populations and create multiple prediction charts, so it was the best to find some common denominator to start dealing with disease detection. In the case of biomedicine that common denominator just happened to be middle aged white males.
Why not including women in the studies?
There was different reasoning as why this was done. For one, they belong to a risky group, research claimed:
1. Menstrual-cycle related hormonal fluctuations can alter the symptoms.
2. There is a possibility of pregnancy, which translates to an immediate dropping out of the study.
It is costly to include, study and understand the differences.
This was unfortunately a recipe for many modern day healthcare disasters. The most notable one is heart related conditions that differ in symptom display between men and women, where there was a significant increase in woman mortality due to faulty disease prediction charts.
The truth is that even nowadays we are uncertain how will different individuals worldwide react to a certain treatment, drug, or how would a specific disease be displayed in them.
AVERAGING BEYOND MEDICINE
Historically, averaging did not only put human lives at risk in medical terms. Tod Rose’s “The End of Average” book beautifully depicts the flaw in non-medical, but mechanical human averaging.
One of the examples brought to light was many incidents and accidents occurring in the late 40s, where fighter jet pilots seemed unable to take control of their jets. The error stemmed from the late 20s, when the first-ever cockpit was designed. At that time engineers have measured many physical dimensions of hundreds of male pilots, and averaged them and used it to produce averaged seats, averaged height of the windshield, averaged shape and size of flight helmets, and averaged distances to reach pedals or a stick.
Thanks to a brilliant scientist Lt. Gilbert S. Daniels mystery was revealed. He asked himself how many pilots are really average? He calculated averages of the 10 physical dimensions considered the most important for the fighter jet cockpit design. He measured over 4000 people, and he compared their measurements to the average pilot one. Not a single one fit within the average range on all 10 dimensions!
Coming back to medicine and health, in today’s age of collecting a (virtually) worldwide pool of big data on health, we can correct many diagnostic wrongs that happened not so recently. I will write more about that next time.
References
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0094147
