Feb 20, 2026 · 12 min read

My Doctors Almost Got a Heart Attack When They Saw My Cholesterol

Almost every person I know believes that LDL is “bad cholesterol.” It is one of the most successful pieces of medical PR in history, right up there with the food pyramid and the idea that dietary fat makes you fat. If you say “my cholesterol is high” at a kids party, people look at you like you’ve just confessed to smoking.

So here goes: My cholesterol is high.

Here’s why I don’t care.

Sirens

My medical aid tests total cholesterol annually. Not LDL. Not HDL. Not triglycerides. Total cholesterol — a number so crude it’s like judging a neighbourhood by counting the total number of vehicles without distinguishing ambulances from tanks. My total cholesterol is above their threshold, so they classify me as “high risk” and charge me accordingly. Animals.

Frustrated, because I have never felt healthier, I spent about some money getting a proper panel done. Probably cheaper than a year of statins, and considerably more informative. Here’s what came back:

Marker	Result	Notes
LDL Cholesterol	4.0 mmol/L	Flagged high
HDL Cholesterol	1.4 mmol/L	Healthy
Triglycerides	0.69 mmol/L	Excellent
ApoB	1.01 g/L	Near population average
Lp(a)	35.2 mg/dL	Low risk
hs-CRP	<1.0 mg/L	No inflammation
Fasting Insulin	3.7 mIU/L	Highly insulin sensitive
HbA1c	5.1%	Perfect blood sugar control

Every doctor I’ve spoken to has recommended statins based on the LDL number. Not one of them mentioned Lp(a). Most didn’t know what it was. I had to request the test myself.

The meme

“LDL = bad cholesterol” has become so deeply embedded that questioning it feels like questioning gravity. But the science behind it is considerably less settled than the PR suggests.

Here’s the short version of what I believe is going on.

LDL-C is a measure of the total amount of cholesterol being carried in your LDL particles. It tells you nothing about how many particles there are, how big they are, or whether they’re the kind that actually cause problems. Small, dense LDL particles are genuinely dangerous since they tend to penetrate arterial walls more easily and are more prone to oxidation.¹ Large, buoyant LDL particles are completely different. Same cholesterol measurement, very different risk.

My triglycerides are low. My insulin sensitivity is excellent. That combination is associated with larger, more buoyant LDL particles — the kind that are least likely to cause damage.² But the lab report doesn’t know or care about any of this. It sees a number above a threshold, prints it in red and says HIGH.

The base rate problem

Coincidentally, Nassim Taleb (I am a fan boy) published a piece in November 2025 about what he called a base rate fallacy in cardiovascular risk assessment.³

The argument: all the risk calculations that link LDL to heart disease are computed on a general population. That population includes people who are clearly not all the same. If you specifically look at Lp(a) — a genetically determined lipoprotein — around a third of people have higher levels than the rest of us. And higher levels of Lp(a) independently increases cardiovascular risk by 2-3x. You can’t change it with diet or exercise.

If you don’t separate these groups, you get an average risk number that looks worse for most people. That number overstates the risk for people with low Lp(a) and understates it for people with high Lp(a). Taleb estimates the overstatement at roughly 30%.⁴

My Lp(a) is 35.2 mg/dL — well below the 50 mg/dL risk threshold. Which means the standard risk tables that my doctors used to recommend statins are systematically overstating my cardiovascular risk. And they didn’t test for it because nobody told them to.

This is a conditional probability problem. The kind of thing you’d expect a first-year statistics student to flag. But the guideline says LDL above X equals statin, and following the guideline is safe. Questioning it is not.

What other factors are there out there, like Lp(a), that we haven’t even heard of?

What actually builds plaque

The simplified story: LDL enters your arterial wall, gets oxidised, triggers an immune response, builds up into plaque. This seems like it makes sense, but it skips the important question — what lets LDL into the arterial wall in the first place?

The answer is endothelial dysfunction. That is, damage or inflammation in the lining of your arteries, caused by things like hypertension, insulin resistance, chronic inflammation, smoking, or oxidative stress.⁵ If your endothelium is healthy, LDL particles — even lots of them — have a much harder time getting in.

My hs-CRP is undetectable, which means there is no systemic inflammation. My fasting insulin is 3.7 which means I am highly insulin sensitive. My blood pressure is normal. I have never smoked. The conditions that enable LDL to cause damage are, as far as anyone can measure, absent.

There’s also emerging research on metabolically healthy people with very high LDL. The KETO-CTA study looked at 80 individuals with mean LDL of 7.0 mmol/L — almost double mine! — paired with high HDL and low triglycerides, and compared their coronary plaque against matched controls with an average LDL of 3.2 mmol/L. There’s no significant difference in plaque burden. And no correlation between LDL level and plaque.⁶ It’s one study with limitations, but it’s not nothing and challenges the dogma.

The cost of being wrong

If LDL really is driving my cardiovascular risk, I should do something about it. But statins aren’t risk-free. For primary prevention — people like me without existing heart disease — the number needed to treat is roughly 100-200 over five years.⁷ That means 99-199 people take a drug with real side effects so that one person benefits.

Those side effects include muscle weakness (directly undermining the thing that actually protects longevity⁸) and a 9-12% increase in Type 2 diabetes risk⁹ (treating a cardiovascular risk factor by creating a metabolic disease).

I believe there is an asymmetry here, but it runs in the opposite direction to what most people assume.

The real risk isn’t that I’m ignoring a dangerous number. It’s that the system is optimised for the median patient — overweight, sedentary, metabolically unwell — and is applying that same playbook to everyone regardless of context. For a lean, fit, 37-year-old with low inflammation, low Lp(a), excellent insulin sensitivity and low triglycerides, the expected value of a statin is very likely negative.

For someone with existing cardiovascular disease, or high Lp(a), or metabolic syndrome? Different calculation entirely. Context is everything, and the problem is that nobody is checking the context.

What I actually do

If you couldn’t guess, I don’t take statins.

What I do instead:

I avoid things I’m reasonably confident are bad for me. Sugary drinks, factory-processed food, seed oils,¹⁰ alcohol. I mostly resist the tyranny of the available.¹¹

I eat without anxiety. Good meat, fatty cuts, biltong, fish, eggs and lots of fruit. I don’t worry about dietary cholesterol because the evidence that it meaningfully drives serum cholesterol in most people is weak.¹²

I stay fit and relatively strong. Zone 2 running, BJJ, resistance training. This is probably the single most important thing on the list — and it’s the thing that gets undermined if you take a drug that causes muscle weakness. I might post about that in the future.

Lastly, I monitor the markers that actually matter: HDL, triglycerides, ApoB, Lp(a), hs-CRP, fasting insulin, Testosterone. The things that give you a real picture of metabolic health and cardiovascular risk, not a single number ripped from context.

The thing that bothers me

I’m not a doctor. I’m a random engineer who sometimes reads papers because I find this stuff interesting. The fact that I — a curious layperson who spent some money and a few weekends reading — have a more nuanced understanding of my own cardiovascular risk than the doctors who tried to prescribe me medication is not a flex. It’s a real problem.

These are smart, well-trained people. They’re not failing because they’re incompetent. They’re failing because the framing of LDL as “bad cholesterol” has been so effective that it has made curiosity unnecessary. The guideline is the guideline. Why look deeper when the number tells you what to do?

I wrote recently about the longevity industry’s relationship with credibility. Peter Attia — the man who built a $100,000/year concierge longevity practice — was one of the loudest voices insisting that LDL management was the cornerstone of cardiovascular health (his focus on VO2Max has also been curious). Anything you heard from him should be weighed against the fact that his judgment about who to spend time with was, charitably, catastrophic.¹³ That doesn’t make him wrong about LDL specifically, but it should make you wonder whether his longevity insights are as robust as his marketing.

The science on cholesterol and cardiovascular disease is not settled in the way that the medical establishment presents it. There are compelling alternative hypotheses about what drives plaque formation, serious questions about whether LDL-C is even the right thing to measure, and a growing body of evidence that metabolic context matters far more than any single number.

Next time I speak to a doctor, I might send them this way.

I am not a doctor. I am a bad engineer. This is not medical advice. Use your own common sense please.

Small, dense LDL (sdLDL) particles have reduced LDL receptor affinity (longer plasma retention), easier transendothelial transport, and lower antioxidant content making them more susceptible to oxidation. See “Atherosclerosis Development and Progression: The Role of Atherogenic Small, Dense LDL,” PMC, 2022. ↩
Low triglycerides with high insulin sensitivity are characteristic of the “Pattern A” lipid phenotype — larger, more buoyant LDL particles. High triglycerides with insulin resistance drive “Pattern B” — smaller, denser, more atherogenic particles. Same LDL-C number, different story. ↩
Taleb, N.N. “Medical Mistakes with Probability, 2: Optimal LDL Levels.” Substack, November 19, 2025. ↩
The maths: Lp(a) and LDL-C risks are additive and independent (Bhatia et al., “Independence of Lipoprotein(a) and Low-Density Lipoprotein Cholesterol–Mediated Cardiovascular Risk,” Circulation, 2025). Since high Lp(a) increases risk 2-3x and affects 15-28% of the population, the unconditional risk estimate for any LDL level includes this subpopulation’s contribution. Remove it, and the risk for low-Lp(a) individuals drops by ~30%. Taleb also flags a possible J-curve for LDL in primary prevention from Chang et al. (2024), JAMA Network Open — a ~400,000-person Chinese dataset where optimal LDL was higher than current guidelines suggest. ↩
“Mechanisms of Oxidized LDL-Mediated Endothelial Dysfunction and Its Consequences for the Development of Atherosclerosis,” Frontiers in Cardiovascular Medicine, 2022. ↩
Feldman et al., “Carbohydrate Restriction-Induced Elevations in LDL-Cholesterol and Atherosclerosis: The KETO Trial,” JACC: Advances, 2024. The one-year follow-up (Soto-Mota, Norwitz, Feldman et al., JACC: Advances, July 2025) found baseline plaque predicted progression, but ApoB did not. Results were mixed — some participants showed rapid progression, others minimal change. The study had no control group, which limits conclusions. Suggestive, not conclusive. ↩
Primary prevention NNT varies by baseline risk. For the overall primary prevention population, the Cochrane Review (2013) found ~18 events avoided per 1000 treated over 5 years (NNT ~56). For low-risk subgroups — people like me without additional risk factors — persistence-adjusted NNT rises to ~117-146 over 5 years (post-hoc analysis, Family Practice, 2018). Secondary prevention NNT is substantially better (~30-50), which is why statins clearly make sense for people with existing cardiovascular disease. ↩
Sarcopenia — age-related muscle loss — is one of the strongest predictors of mortality in older adults. A drug that causes muscle weakness as a side effect in 5-15% of users is directly working against the thing that keeps you alive longer. Sheesh. ↩
Sattar et al., “Statins and risk of incident diabetes: a collaborative meta-analysis of randomised statin trials,” The Lancet, 2010. ↩
The seed oil debate is tribal and the evidence is mixed. The mechanistic argument — that linoleic acid in seed oils increases LDL oxidation and promotes endothelial damage — is plausible but not conclusively demonstrated in vivo. I avoid them not because I’m certain they’re harmful, but because the downside of being wrong is zero and I don’t trust “big canola”! I just use butter and olive oil instead. It’s not exactly a sacrifice. ↩
The availability heuristic applied to diet: what’s in your house is what you’ll eat when willpower is depleted. Engineer the environment, don’t rely on discipline. ↩
Dietary cholesterol has a modest effect on blood cholesterol for most people. The 2015 USDA Dietary Guidelines Advisory Committee noted that “cholesterol is not a nutrient of concern for overconsumption” — though the 2020-2025 guidelines walked this back slightly. Your body produces the vast majority of its own cholesterol and adjusts production based on intake. ↩
Attia’s name appeared over 1,700 times in the Epstein files released January 2026. He has since lost a number of roles and is suffering from a small credibility crisis. None of this means all his medical views are wrong, though I suspect there is an inverse correlation between the hubris in his emails with old Jeff and the level of rigour in his work. ↩