Causal Arrow

In the 1990s, researchers accidentally discovered that asking ten questions about a patient's childhood could predict disease better than anything else medicine routinely measures. It costs next to nothing to administer, takes about three minutes, and patients would only need to take it one time ever, as their answers, once they are over 18, would never change. This simple assessment predicts lifelong disease probability as well as or better than all of our standard current screening tools combined: blood pressure, cholesterol, smoking history, and family history of disease.

The ACEs survey consists of ten questions about your childhood before the age of 18, and you answer yes that happened to me, or no it didn’t happen to me. That’s it. So if you answered “yes that happened to me” to 6 or more of the questions, your life expectancy is on average 20 years less than your peers who answered no to all of the questions.

The original study that figured out how predictive this was tracked 17,421 people over decades of their life. For contrast, most clinical trials that dictate our care across the entire medical model of healthcare today enroll an average of 65 patients and last an average of 6-12 weeks, with larger phase 3 trials capped at around 3,000. The original study focussed on a low-risk population of mostly white, middle class, college educated, insured people. They asked the participants these ten simple questions about their childhood, and then they tracked their health outcomes for decades. They found that about 60% of these people reported at least one yes on their assessment, and one out of eight people reported 4 or more yeses. And there was a direct dosage response, meaning for each additional number you answered yes to, your health outcomes were predictably worse. There have since been hundreds of studies in the 30 years since this original study, and through that expanded research those same results have been replicated across many different populations.

Now let’s look at a parallel set of research that was being done over this same 30 year timespan that shows actual pathostatic chemistry showing up in the body and how that correlates with disease.

Allostatic load

The term ‘allostasis’ was coined in 1988 by Sterling and Eyer. They saw something that challenged the framework of the time, noticing that black Americans had far higher rates of hypertension than genetically similar populations in West Africa. They had basically the same genetics, but existed in different environments, and they were showing vastly different disease rates. The belief into the 1980s was that your genes and constitution (namely if you were weak or hardy) determined whether you got disease and what diseases you got. This allostasis work proved that our physiological responses weren't fixed, but were actually adaptive to context, which was a revolutionary finding at the time. They were seeing pathostasis in action - the same genetics producing disease or health depending on pathostatic load: how much pathostatic chemistry you are carrying and for how long.

Then in 1993 researchers Bruce McEwen and Eliot Stellar took this concept further. They measured the same chemicals we've been looking at: cortisol, catecholamines, inflammatory markers, metabolic markers, and tracked what happened to people over time. They called the cumulative burden of these chemicals 'allostatic load.' And what they found was that these chemicals become dysregulated and then cascade through connected systems exactly the way our tree predicts.

In one study they tracked 738 adults over 5 years, and tested 12 allostatic biomarkers. In this study they found that for every single additional biomarker that was out of range at baseline, people had 35% higher odds of developing type 2 diabetes, 21% higher odds of cardiovascular disease, and 15-24% higher odds of physical impairment - even though they seemed healthy by regular medical measures when originally tested.

Once they had this data they divided participants into three groups: those with 0-3 biomarkers out of range (healthy), 4-5 out of range (moderate load), and 6 or more out of range (high load). Compared to the healthy group, people with just 4-5 dysregulated markers had nearly triple the odds of developing diabetes over the next 5 years (2.78 times higher). Those with 6 or more dysregulated markers had more than double the odds of cardiovascular disease (2.32 times), and double to triple the odds of physical impairment. These were people who seemed fine by medical standards, and this model could accurately forecast their disease trajectory years before symptoms appeared. And these findings are consistent across populations. Many studies have been conducted globally showing the same patterns. The relationship holds across age groups and cultural contexts. Higher allostatic load means worse health outcomes.

To date, thousands of articles, 2,465 as of the time of this publication, and over 30 years of data, predicts and explains disease far better than anything we are currently pointing to as a reason or a cause and yet this whole body of research is mostly not taught in medical school. And that is likely in large part due to the fact that the researchers framed the solution as ‘reduce stress and improve coping’, which we all know doesn’t actually do a whole lot. They see this as something that amplifies disease, not as something that causes it. But what this research does very well is show that tracking pathostasic chemicals and markers BEFORE disease arrives predicts disease better than anything else we’re measuring. Just like ACES do.

Correlation vs causation

Other than obesity one of the main theories we have about where diseases come from is genetics, so let’s look at that next.

In 1993 they discovered that Huntington's disease is caused by a specific genetic mutation - a CAG repeat expansion in the huntingtin gene. Medicine identified this by studying families where Huntington's ran across generations, found the mutation in affected individuals, and concluded the mutation causes the disease with near-complete penetrance, meaning they were saying that pretty much everyone with the gene will get the disease. But the thing is, they only tested people who already had Huntington's disease, or who had family members with Huntington's disease. They found the mutation in people with the disease, confirmed it was inherited, and declared it fully penetrant. Which is the definition of selection bias. But because of this seemingly full penetrance, Huntington’s disease became the poster child for genetic causation.

But when they finally did screen the general population 23 years later in a 2016 study of over 7,000 people, they found that approximately 1 in 400 individuals carry the expanded CAG repeat associated with Huntington's disease. That's 250 per 100,000 people. But Huntington's prevalence is only about 5-13 per 100,000. Which means roughly 95-98% of people with 'the Huntington's gene' never develop Huntington's disease. For context, Huntington's expression depends on how many times the CAG sequence repeats - anywhere from 36 to 41+ repeats is considered the "disease range," with higher numbers associated with more severe symptoms and earlier onset. When they finally looked at the general population instead of just symptomatic families, they found that having the mutation doesn't actually mean you'll develop the disease. Up to 86% of people with 36 repeats - the low end of the range - never got the disease in their lifetime. They found 10 people aged 67-95 with 36-39 repeats who showed no signs of Huntington's. Even at 40-41 repeats, traditionally called the "full penetrance" range where the disease should be inevitable, they documented asymptomatic carriers. The mutation is far more common than the disease. Most carriers never get sick.

So as rigorous research standards make very clear, if you only test people who have the disease or are in families where it appears, and you find a genetic mutation in those people, you can't conclude that everyone with that mutation gets the disease. That is correlation, not causation, which means you can only conclude that everyone with the disease has the mutation - which is a very different claim. But despite this, the belief that genetics are determinant is still pretty strongly believed, even though we know it’s not true. And this same flaw runs through genetic research on most chronic diseases. They study sick people, find genetic variants that are more common in sick populations than healthy ones, and declare those variants cause disease. But they're not screening everyone with those variants to see how many never get sick. They're not investigating what determines expression versus non-expression. They're looking only where they expect to find disease, then concluding what they find is universal.

For example even for the full penetrance range, 40+ CAG repeats is considered a golden ticket to disease for Huntington’s, we have only ever looked at sick people and their families. Because those are less common, the study above of over 7000 people in the general population only found 3 people in the full penetrance range, which means we have no idea whether there are people who don’t have family members with Huntington’s disease who are walking around with 40+ CAG repeats and no disease. We would have to do a much bigger population survey to make that assertion.

And then there is something called the Huntington’s phenocopy. Which is people who present as having Huntington’s disease who don’t have the gene at all. The response has been to look for other genes that might explain it, because by this point genetics is the explanation, so when the gene isn’t there the only reasonable explanation seems to be, it must be another gene we haven’t found yet. The current estimate is that about 20% of suspected Huntington’s cases are phenocopies, which is one in five people. Which means that having the gene doesn’t guarantee the disease, and you can actually get the disease without the gene present at all.

If even the most "genetic" disease in medicine depends on something beyond the gene to express, and the gene doesn’t always cause it, then genetics isn't the cause, it’s a vulnerability. It's where the chain breaks.

· · ·

So now we’ve walked through a lot of medical diagnosis, but what about psychiatric ones? How did disorders of the brain end up being somehow separate from disorders of the body anyway, being that they all stem from the same animal?

It turns out the answer isn't medical or scientific, it's historical and institutional. So let’s look at how this divide happened, and why it persists despite all evidence to the contrary.