The Three Pillars Behind Almost Every Modern Illness

Only around 1 out of 8 American adults is metabolically healthy.

Araujo, Cai and Stevens took 8,721 adults from the national health survey covering 2009 to 2016 and found 12.2 percent (95% CI 10.9 to 13.6) met every criterion at once. That is about 1 in 8.2. (PMID 30484738)

The criteria are stricter than the phrase sounds. Five measures have to be optimal at once, with no related medication: waist under 102 cm for men and under 88 cm for women, fasting glucose under 100 mg/dL together with HbA1c under 5.7 percent, blood pressure under 120 over 80, triglycerides under 150 mg/dL, and HDL of at least 40 mg/dL for men or 50 mg/dL for women.

Read the medication clause again. A blood pressure of 118 over 76 achieved with a daily pill counts as not healthy. In the same data, 8.0 percent of overweight adults and 0.5 percent of adults with obesity qualified.

The figure has been updated, and the newer one is worse. O'Hearn and colleagues analysed 55,081 adults across 1999 to 2018 and found 6.8 percent (95% CI 5.4 to 8.1) with optimal cardiometabolic health in 2017 to 2018, roughly 1 in 15. (PMID 35798448) Not a like for like replacement, since that composite also requires the absence of diagnosed cardiovascular disease and uses adiposity rather than waist alone. Still a reason not to call 1 in 8 the current state of the evidence.

These conditions arrive as siloed problems. Failing heart, see a cardiologist. Blood sugar, an endocrinologist. Memory loss belongs to a neurologist. Underneath them, research keeps turning up the same three: chronic inflammation, insulin resistance, and dysfunctional mitochondria.

01Pillar one: chronic, low-level inflammation

Inflammation is how the body answers tissue damage, and healing needs it. Cut your hand on glass: the pain stops you using it, the swelling brings in the cells that repair it. Useful and necessary.

Now imagine it never going away. Trucks outside your house and sirens wailing for years, for decades. You would stop noticing the noise.

Franceschi and colleagues named that process inflammaging in 2000. (PMID 10911963)

What it is linked to, and how firmly. A 2018 review in Nature Reviews Cardiology names the diseases and flags the gap in the same breath. It lists inflammaging as a risk factor for chronic kidney disease, diabetes, cancer, depression, dementia and sarcopenia. Separately, for cardiovascular disease, it says clinical trials suggest the association is causal. (PMID 30065258)

Cardiovascular disease is where the trials are in. Whether damping inflammation helps with the rest is, in those authors' own word, controversial.

Diabetes is a strong prospective association and nothing more. Pradhan and colleagues ran a nested case-control study inside the Women's Health Study, 188 women who developed diabetes against 362 matched controls drawn from 27,628. Elevated CRP and IL-6 predicted incident type 2 diabetes, relative risks of 15.7 and 7.5 for the top quartile against the bottom. That design cannot make it causal. (PMID 11466099)

Cardiovascular disease is the one with randomised evidence. CANTOS gave canakinumab, an antibody against interleukin 1 beta, to over 10,000 patients who had already had a heart attack and had elevated hsCRP. The 150 mg dose cut recurrent major cardiovascular events without lowering LDL at all. (PMID 28845751) I have not printed an exact randomised total, because the registry record and the paper give different figures and I could not reach the paper's abstract to settle it. The same trial found fewer lung cancers on the drug, hazard ratios 0.61 at 150 mg and 0.33 at 300 mg, but that was an exploratory secondary analysis and the authors called their own result hypothesis-generating. (PMID 28855077)

Cognition comes from cohort data. ARIC measured inflammation in 12,336 people at midlife, average age 56.8, and followed cognition for 20 years. The top quartile of midlife inflammation declined 7.8 percent more steeply than the bottom. (PMID 30760633) A small effect.

02Pillar two: insulin resistance

Insulin is a messenger your pancreas releases after you eat. It knocks on the door of your cells and tells them to absorb blood sugar. Now imagine ringing that doorbell a couple hundred times a day, for years. The cells start to ignore it, and your body compensates by making more insulin to force a response.

This is the sentence I got wrong. The first version of this piece said your body can hold normal blood sugar for 10 to 15 years on average, hiding the dysfunction the whole time. That figure has no primary source. It comes from a textbook chapter (PMID 29939616), and the references that chapter cites for it do not include a study that measured such an interval. The chapter also says insulin resistance precedes the development of type 2 diabetes. I turned that into insulin resistance hiding behind normal blood sugar, a different and stronger claim.

What the longitudinal data show. Whitehall II modelled the run up to diagnosis in 6,538 civil servants, 505 of whom developed diabetes, over a median 9.7 years of follow up. Trajectories were modelled from up to 13 years out, and that 13 years is the lookback window rather than a finding. The finding is that changes appear as much as 3 to 6 years before diagnosis. Fasting glucose rose linearly across the whole period and then climbed steeply, from 5.79 to 7.40 mmol/L, starting three years out. Insulin sensitivity fell steeply only in the final five. (PMID 19515410)

So it is not a dam that holds and then bursts. It is a floor that has been settling a fraction of an inch a year the entire time, then drops several inches in the last three.

The closest published number to the one I printed. Harris and colleagues in 1992 calculated that retinopathy onset had occurred roughly 4 to 7 years before diabetes was diagnosed, and since other data suggested diabetes may be present five years before retinopathy appears, they wrote that onset may occur 9 to 12 years before clinical diagnosis. (PMID 1516497) Their own title and headline conclusion is the shorter figure. The longer one is a hedge on a 1992 extrapolation, and it measures how long diabetes precedes diagnosis, not how long insulin resistance precedes abnormal glucose. Different claim from mine.

On type 3 diabetes. Steen, Wands, de la Monte and colleagues found insulin and IGF signalling markedly reduced in Alzheimer's brains in 2005. The question is posed in their title and the term is formally proposed in the closing sentence. (PMID 15750215) The paper that made it famous is the 2008 review titled "Alzheimer's disease is type 3 diabetes-evidence reviewed", and that flat declarative form is exactly why the label spread. Crediting the 2008 paper with the coinage is a common error.

The label is a hypothesis, not a diagnosis, and the two largest attempts to treat Alzheimer's by going at insulin signalling both failed. SNIFF gave 40 IU of intranasal insulin daily for 12 months in mild cognitive impairment and mild Alzheimer's, with no cognitive or functional benefit. (PMID 32568367) 289 people were randomised and 240 formed the primary analysis population, because the device used for the first 49 was unreliable and had to be replaced, which makes it a less clean test than it looks. Then evoke and evoke plus tested oral semaglutide in 3,808 patients with amyloid confirmed early Alzheimer's, across 566 sites in 40 countries, and missed the primary endpoint in both. (PMID 41865758)

For the case in favour, a 2024 review in Ageing Research Reviews surveys the shared mechanisms without endorsing the label. Its title opens "True or false?", which tells you where it lands. (PMID 38955264)

03Pillar three: mitochondrial dysfunction

Mitochondria turn the food you eat and the oxygen you breathe into energy. When they stop making as much as they used to, the parts of your body that need the most suffer first. You may find yourself lethargic in a way that sleeping more does not fix.

Researchers believe this may be a core hallmark of aging. That is the one claim here still carrying no citation, and I am leaving it visible.

04The hidden connection

The three pillars form a loop. Inflammation drives cellular insulin resistance, which wreaks havoc on your mitochondria, which produces more inflammation. What you get is a collection of conditions rather than one disease, and that combination is what we call aging.

I still find the model useful. In the diabetes analysis of CANTOS, blocking interleukin 1 beta produced large reductions in hsCRP and IL-6, and over a median 3.7 years it did not reduce the incidence of new onset diabetes at all. The hazard ratio was 1.02 (95% CI 0.87 to 1.19). HbA1c fell for six to nine months and then the benefit disappeared. (PMID 29544870)

If inflammation were the load-bearing wall in this house, taking it out should have brought the other rooms down with it. They took it out, in ten thousand people, under placebo control. The heart attacks fell. The diabetes did not move.

That is the cleanest published dissociation between the first pillar and the second, and it argues these three are correlated rather than causally stacked. The loop may still be real.

05The boring stuff that actually works

I am not a fan of miracle cures. The things with the most robust evidence are rarely interesting. Three of them I can now put numbers behind.

Sleep. A meta-analysis of 72 studies covering more than 50,000 participants, including both prospective cohorts and experimental sleep deprivation, found sleep disturbance associated with CRP (effect size 0.12) and IL-6 (0.20). (PMID 26140821) Those effects are small, and the nulls matter. TNF-alpha showed no association with sleep disturbance, and short sleep duration showed no association with IL-6. "Poor sleep drives inflammation" is more than this source carries.

Muscle. In NHANES III, across 13,644 people, each 10 percent increase in skeletal muscle index was associated with an 11 percent relative reduction in HOMA-IR (95% CI 6 to 15 percent) and a 12 percent relative reduction in prediabetes and diabetes prevalence (95% CI 1 to 21 percent). (PMID 21778224) That is cross-sectional, so it is association and not protection, and the authors themselves call for intervention research. The mechanistic case for why muscle sits at the centre of this pillar is DeFronzo's 2009 Banting Lecture. (PMID 19336687)

Diet. This is the strong one, and it is causal. DiRECT cluster-randomised primary care practices to an intensive weight management programme. At 12 months, 46 percent of the intervention group were in remission from type 2 diabetes and off all antidiabetic drugs, against 4 percent of controls, with a clean dose response by weight lost: zero percent among those who gained weight, 34 percent at 5 to 10 kg, and 86 percent among those who lost 15 kg or more. (PMID 29221645)

One sourcing note. If you see PREDIMED cited for diet and cardiovascular outcomes, the 2013 paper was retracted and the 2018 republication is the one to use. (PMID 29897866) The reason was randomisation irregularities at one of the study sites and in household co-enrollment, with results materially unchanged on reanalysis. That breakdown is reported from industry summaries rather than the primary record, since I could not retrieve the retraction notice itself.

06My take

What I keep returning to is how the 10 to 15 year figure got into my draft at all. I did not invent it. I read it somewhere reputable sounding, it matched the story I was already telling, and I typed it. A textbook says it, an encyclopedia repeats the textbook, health writers repeat the encyclopedia, and by the time it reaches you it has the texture of something measured.

There was no silent window. Glucose was drifting upward the entire time, in data anyone could have drawn, and the drama sits in the last three years.

The same discipline applies to the framing this newsletter is built on. Three pillars is a useful map. The one time somebody tested a link on that map under randomisation, the link did not hold. I am keeping the map and telling you where it failed.

Which number about your own health, the one you repeat to friends or to your doctor, have you never traced back to a study? Send it to me and I will go find where it came from.

Frequently asked

What are the three pillars behind most age related disease?+

Chronic low grade inflammation, insulin resistance, and mitochondrial dysfunction. Not diseases in themselves, but problems sitting underneath heart disease, type 2 diabetes and dementia.

Does chronic inflammation cause type 2 diabetes?+

The association is strong and the one randomised test of it did not hold. In the CANTOS trial, blocking an inflammatory signal cut further heart attacks and did nothing to new cases of diabetes.

How long does insulin resistance build before diabetes is diagnosed?+

The figure that circulates is 10 to 15 years, and I could not find a study behind it. In 6,538 British civil servants tracked before diagnosis, glucose was drifting upward the whole way and climbed steeply only in the final three years.

Is Alzheimer's disease really type 3 diabetes?+

It is a hypothesis, not a diagnosis. The two largest attempts to treat Alzheimer's by going at insulin signalling both missed their primary endpoint.

What actually works on all three pillars?+

The dull things, and only one has causal evidence here. In DiRECT, 46 percent of the weight management group were in remission from type 2 diabetes at 12 months against 4 percent of controls.

---

This content is for educational and informational purposes only and is not medical advice. Peptides discussed are research compounds and may not be approved for human use. Nothing here should be used to diagnose, treat, cure, or prevent any disease. Always consult a qualified healthcare professional before starting any peptide, supplement, or protocol. Individual responses vary. Do not self-administer compounds without proper medical supervision.