Clinical Philosophy · Biochemical Individuality · Why You Are Not a Statistic
Roger Williams published Biochemical Individuality in 1956. His central argument — that human beings vary enormously in their nutritional needs, metabolic patterns, and physiological responses, and that medicine treating populations as if they were individuals was making a fundamental error — has been confirmed at the molecular level by every branch of biological science since. And yet most health advice is still written for Tom, Dick, and Harry simultaneously. This is why it so often doesn't work for you.
In the summer of 1999, I had two clients — women of almost identical age, similar weight, similar presenting symptoms of fatigue, poor sleep, and digestive discomfort. I put both of them on what was, at the time, my best clinical understanding of an appropriate nutritional protocol. One improved significantly within six weeks. The other felt worse.
This experience — which I have now seen repeated in more variations than I can count across 37 years of practice — is what biochemical individuality looks like in a clinical setting. Not as a theoretical concept. As a practical reality that has to be accounted for if you want to actually help the person in front of you rather than the average person you imagined when you designed the protocol.
Roger Williams, the biochemist who first systematically documented biochemical individuality, was writing in an era before genomics, before microbiome science, before epigenetics, before nutrigenomics. He was working from anatomy, chemistry, and careful observation. What he documented — that the range of human variation in nutritional requirements, enzymatic activity, organ size, metabolic rate, and physiological response is far wider than medicine acknowledges — has turned out to be not just correct but profoundly understated. The molecular biology of the past 30 years has filled in the mechanisms beneath observations that took Williams a career to establish.
This post is an attempt to explain why that matters — for your health specifically, right now, in ways that have direct clinical consequences for whether the health advice you're following is actually appropriate for you.
It means that your nutritional requirements, your drug responses, your disease risks, your detoxification capacity, your gut microbiome composition, your hormonal patterns, your metabolic rate, and your response to stress are all genuinely, measurably different from those of the person sitting next to you — and that those differences are large enough to matter clinically.
This is not a soft wellness concept. It is a hard biological fact, and the evidence for it spans multiple scientific disciplines.
Medicine is built on population-level evidence. The randomised controlled trial — the gold standard of medical evidence — takes a group of people, applies an intervention, and measures the average effect. The result tells you what happens to the average person. It cannot tell you what will happen to you — because you are not the average person. You are a specific person with a specific biochemistry, a specific history, a specific microbiome, and a specific pattern of genetic polymorphisms that the trial population almost certainly doesn't represent.
This is not a criticism of population medicine. It is the right tool for what it does — establishing that an intervention has an effect in a population, identifying risk factors at scale, informing public health decisions. The problem arises when population-level evidence is applied without modification to the individual clinical encounter — when the drug prescribed, the supplement recommended, the dietary guideline issued, or the reference range used is calibrated to the population average and applied without adjustment to a specific person whose biochemistry may sit far from that average.
Reference ranges are the most visible example. The "normal" range for a blood marker is typically defined as the range within which 95% of the reference population falls. That population is often not healthy — it is drawn from people who happened to present for blood tests, which skews toward those with health concerns. The range is statistical, not functional. A TSH of 3.8 mIU/L is within the normal range. It is also associated with reduced T4 to T3 conversion efficiency, increased cardiovascular risk, and symptoms of hypothyroidism in a meaningful proportion of people who carry it. "Normal" is not the same as optimal — and optimal varies between individuals in ways the reference range cannot capture.
The health advice written for everyone is, by definition, written for no one in particular. The closer it gets to the population average, the further it gets from you specifically.
There is a particular form of health-related social comparison that is both ubiquitous and clinically counterproductive. It goes: someone you know follows a specific diet and feels wonderful; therefore you should follow the same diet. Or: your colleague takes a specific supplement and their energy has transformed; therefore you should take the same supplement. Or — the modern version — a personalised algorithm on a health platform has identified certain patterns in your data that match patterns in other users, and recommends the interventions that worked for them.
Your neighbour thrives on a high-fat, moderate-protein diet. You feel worse on it. Both responses are correct — you are both responding accurately to your own biochemistry. Their Metabolic Nature may be Kinetic. Yours may be Grounded. The same macronutrient ratio that drives their energy impairs yours because your oxidative rate, your mitochondrial efficiency, and your hormonal response to dietary fat are genuinely different from theirs.
Your colleague's vitamin D supplement took their level from 38 to 85 nmol/L on 2,000 IU daily. Yours barely moved on the same dose. They have a responsive VDR variant. You may have a less responsive one, or a gut absorption issue, or a magnesium insufficiency impairing the conversion pathway. The same supplement, the same dose, meaningfully different outcomes — for biological reasons that are specific to each of you.
The personalised health app that compared your sleep data to 10,000 other users and recommended earlier bedtime and reduced caffeine after noon is not wrong — those are reasonable population-level recommendations. But it cannot see that your 3am waking is driven by low progesterone and elevated cortisol in the early morning hours, which requires a very different clinical response than generic sleep hygiene advice.
Keeping up with the Joneses in health terms is not just unhelpful — it can actively misdirect you. The intervention that worked for someone else worked because it addressed something specific in their biochemistry. If that thing is not present in yours, the intervention will either do nothing or, in some cases, make things worse. The person who tried the same probiotic their friend swore by and felt significantly worse is not having a nocebo response. They may have SIBO, or a histamine intolerance, or a dysbiosis pattern that high-dose Lactobacillus supplementation actually worsens. Their friend's microbiome needed what that probiotic provides. Theirs didn't.
Conventional medicine asks: what disease do you have, and what is the established treatment for that disease?
Functional investigation asks: what is actually happening in your specific biochemistry right now, and what does your specific body need?
These are different questions and they produce different answers — not because functional medicine rejects the evidence base of conventional medicine, but because it applies that evidence base at a different level of resolution. The level of the individual rather than the population. The level of the mechanism rather than the diagnosis. The level of what is happening now rather than what has already caused enough damage to be named.
The distinction between functional and medical is also the distinction between current and historic. A diagnosis names something that has already happened — a pattern of damage or dysfunction that has reached the threshold for classification. Functional investigation reveals what is building before it reaches that threshold — the cortisol dysregulation before the burnout, the gut permeability before the autoimmune diagnosis, the ferritin decline before the anaemia, the insulin resistance before the type 2 diabetes. The gap between "building" and "named" is where the most clinically important and most modifiable work happens.
The digital age has produced something that looks like personalisation but isn't quite. Search algorithms personalise based on your search history, your location, your device, and the patterns of people who searched similar things. Health information platforms personalise based on what users with similar demographic characteristics found helpful. AI systems trained on population data give you the best average answer to your question.
None of these are personalised to your biochemistry. They are personalised to your behaviour — which is a very different thing. Your search history tells the algorithm what you're interested in. Your blood chemistry, your gut microbiome, your genetic polymorphisms, your hormonal pattern, and your toxic burden tell a clinician what is actually happening inside you. The first produces a customised information feed. The second produces a clinical picture.
The difference matters most when what you need is not more information but more clarity about your specific situation. The person who has read everything about thyroid health, tried multiple interventions, and still feels unwell is not suffering from a lack of information. They may be suffering from a lack of investigation — a proper map of what is actually happening in their specific biochemistry that tells them not what works for thyroid patients in general, but what their thyroid situation specifically requires.
Information personalised to your interests is everywhere. Investigation personalised to your biochemistry is rare. That gap is the most important gap in modern health.
There is another dimension to biochemical individuality that Williams identified and that modern science has confirmed: it is not static. You are not the same biochemically as you were five years ago, or as you will be five years from now. Your microbiome changes in response to diet, antibiotics, stress, and environment. Your methylation patterns shift in response to nutrient availability and toxic exposure. Your hormonal landscape changes with age, stress, illness, and pregnancy. Your mitochondrial function responds to exercise, to sleep, to nutritional status.
This means that even a genuinely personalised investigation is a snapshot — the most useful snapshot available, but a snapshot nonetheless. The protocol that was exactly right for you 18 months ago may need revision because you are not the same person metabolically. This is not a failure of the approach. It is the biological reality that personalised medicine has to account for and population medicine cannot.
The value of periodic functional investigation is not just the initial picture. It is the ability to track change over time — to see whether the interventions are working, whether new stressors have shifted the picture, whether markers that were marginal are improving or worsening. This is health management at the level of resolution that actually reflects how biology works.
There is no one-size-fits-all health plan. There is no supplement stack that works for everyone. There is no diet that is optimal for every human body. There is only what is appropriate for a specific person given their specific biochemistry — and finding that out requires investigation, not assumption.
It means that before you follow the dietary protocol your friend swears by, before you take the supplement that worked for your colleague, before you accept that your blood results are normal because they fall within a population reference range — it is worth asking a more fundamental question.
Is this calibrated to me?
Not to the average person. Not to the population. Not to the person next to you whose health looks similar from the outside but whose biochemistry may be profoundly different. To you. Your gut bacteria. Your genetic polymorphisms. Your hormonal pattern. Your mitochondrial function. Your toxic burden. Your specific nutritional gaps and excesses.
This is what functional investigation does — and why it produces results that generic health advice, however well-intentioned and however evidence-based at the population level, cannot reliably replicate. Not because the generic advice is wrong. Because it is right for someone else.
Roger Williams understood this in 1956 from anatomy and chemistry. The molecular biology of the past seven decades has confirmed it in extraordinary detail. The clinical practice that takes it seriously — that begins every case with the question "what is actually happening in this specific person?" rather than "what is the standard protocol for this presentation?" — produces outcomes that reflect what is possible when you treat the individual rather than the population.
That is what this practice is built on. It always has been.
Not the population. Not the average. Not what worked for someone else. Your specific biochemistry, investigated properly, interpreted clinically, acted on specifically.