Statins: What "Experts" Get Wrong

Are you ready? I'm about to take you inside a study that broke media headlines. Millions more people should be taking life-saving statins. Most feared statin side effects are unfounded, landmark study finds. And every time a study like this drops and sounds rigorous, headlines and influencers run rampant, but without first digesting the data. And as an MD-PhD researcher obsessed with the nuances of heart health science, I I find this so frustrating. And the more I looked into this paper, this specific paper, the more I got obsessed that all of these reporters and doctors and so-called expert bodies, they're getting it totally wrong. The headlines scream "Safety with Certainty," but the data hint at new risks. Risks no one is talking about. I mean, hold up, could statins actually make you fat? Trust me, you're gonna wanna sit through for this one, because I'm not only gonna rewrite what you know about statins, but rewire how you think about certainty in science. I know that's a tall order, but I will deliver. Watch me. Now let's go. It is a communication failure, the p-value propaganda that erects brick walls in front of informed consent. People have questioned whether the display is meant to be intentionally confusing, a statistical lie, incompetence, intent to mislead, or maybe just laziness. People still died. Human people still died. Okay, now before we begin, hang in there for just a little bit longer than usual for a really important preamble. I want to offer you both a challenge and a promise. The challenge is this: listen, really listen to what I am saying, and more importantly, to what I am not saying, because there's going to be a theme that will run through this video. It's something you need to understand deeply, that science and medicine don't operate in binaries— blacks and whites, ones and zeros. So to say The fact that we are not certain of something is not to say we are certain of the opposite. I know that sounds vague right now, but it'll make a lot more sense as we go through this paper together. Now for the promise. I didn't want to make this video. I generally don't like making reaction content. I don't really enjoy it, and this falls into that category. I'd much rather spend my time reading new Cell and Nature papers, getting genuinely excited about the newness of discovery and bringing those scientific frontiers to you rather than intellectually sword fighting with what I see to be distortions of the evidence. But where there is real opportunity to deliver genuine clarification and educational value, I do what I think is necessary. So I'm not doing this for easy clicks, to stir outrage, or to feed conspiracy theory thinking. Yes, there's a hook, there's a lead. The headlines scream "Safety with Certainty, but the data hint at new risks." That's to get you right here where I want you, but everything that I'm going to say is grounded in logic, verbally precise, and honest. And sometimes the truth is uncomfortable. I think we'll find that out in the comments later. But that aside, and as always, what I give you here is the concise version. If you want extra nuance, more depth, and more references, you can find that at the written newsletter linked in the video notes. Now, here's the roadmap. First, I'm gonna walk you through this big new study, what they did, and what made it such massive news. Then I'm gonna explain in chapter 2, the data. And how they actually should be interpreted. Third, I'm going to show you the big blind spot and why the omission matters so much. Then fourth, I'll reveal other limitations, big ones, and why they need to be part of the conversation. Then finally, in part 5, I'll respond to pushback. So with that, let's begin. Part 1: The Study. This was a massive labor, it truly was. A meta-analysis of large double-blind statin randomized controlled trials to assess for the negative side effects caused by statins. That means the research group pooled data from studies where patients were randomized to a statin or a placebo, and both the patient and the provider were blinded to the assignment. In the end, they ended up including 19 total trials with a median follow-up period of 4.5 years and a total of 123,940 individuals. Using the side effects listed on statin product labels, they assessed for 66 additional undesirable outcomes beyond effects on muscle and diabetes, which they present previously reported. We just kind of accept those risks. Their headline conclusion, and now I'm going to quote right from the Lancet Group's professional LinkedIn account, "New: Almost all side effects on statin labels not caused by statins." Now bank that claim in your brain because I'm going to prove that it is unequivocally inaccurate and effectively a statistical lie based on their own data. That brings us to part 2, the data. Now, this is where we get to have a little bit of fun together. I want to show you how to understand their main graph. Now, fair warning, looks confusing as heck, right? In fact, people have questioned whether the display is meant to be intentionally confusing since this sort of graph called the forest plot isn't usually presented this way. It's usually presented in a clearer vertical form. So to help you understand it, indulge me as I walk you through a simplified vertical version, and then we're going to translate it into the graph at hand, and you are going to understand what's going on. So here's what you need to know. The horizontal axis represents what's known as a relative risk, or risk ratio, abbreviated RR. That's the probability of an event occurring in the treatment group—so say the statin group—versus the control. So a relative risk of 1, a risk ratio of 1, means the risk of an event occurring in the intervention group divided by that in the control group is 1. In other words, no increase in risk. Values greater than 1, so to the right of the line, are associated with an increased risk of the outcome given the intervention. So in this case, statins will be increasing risk. And if it's less than 1, the opposite—statins reduce risk. Now the risks for various specific outcomes are plotted vertically, and the circles represent the average risk ratio, and the horizontal line represents the confidence interval. The confidence interval is critically important. It tells us how noisy the data are. And, very importantly, if the confidence interval crosses the risk ratio line of 1, we say the result is non-significant. Now pause. Stop. I have another problem for you. This word "non-significant"—what does it mean? Does it mean that the intervention—the statin—does not cause an outcome? No. Heck no. By golly, no. No! God, please, no! What it means is the data aren't high resolution enough to claim with enough certainty that statins do cause the outcome. So, if we are 90% certain that statins cause, say, weight gain, and hold that thought, but it doesn't reach statistical significance because the confidence interval crosses that magical line of 1, we get to slap the term "non-significant" on it. Not Slytherin. Not Slytherin. So do you see the problem? Do you see the nucleus of confusion? Remember, in this hypothetical, where we are 90% sure that statins make you fatter, but because we want to have a very high bar for claiming certainty—X causes Y for sure—we don't give it the term significant. Instead, we give it the term non-significant. That is not a comment on biological reality. It is statistical semantics. Or if you want another example, we could have a drug that causes a 5-fold increased risk of death within a year, But if there's enough variation in a study such that it's, say, underpowered and the confidence interval just crosses that risk ratio line of 1, it's non-significant. There's still probably a 400% increased risk of outright death in a year in this example, but if it falls shy of that arbitrary threshold, we call it non-significant and it can get brushed aside. Now let's look at their graph and see what we find. So that brings us to part 3, the blind spot. Their graph is just what I explained, except they twisted it into a circle for who knows what reason. What is clear as mud though is that there's a ton of trends suggesting cause-effect relationships between statins and negative outcomes, including— wait, what's that? Weight gain. We also see clear causal trends for male breast enlargement, gynecomastia, skin conditions, and significant effects on liver function tests, nuanced at diabetes and edema, which is swelling. So do you see? Do you see now why the headlines are not supported by the data? No, the study did not show millions more people should be taking life-saving statins. It did not show that concerns are unfounded. No, it did not show that almost all statin side effects are not caused by statins, which is from the Lancet group themselves. I mean, if you're trying to present yourself as a serious scientific group, that This form of verbal imprecision reveals either incompetence, intent to mislead, or maybe just laziness. I'm not sure, but really none of the interpretations are good. But we're not done. So don't you dare move— unless you like to do exercises while watching YouTube, in which case good on you. Anyway, let's look at Figure 2. It looks similar to Figure 1, which we just saw, but it's asking a slightly different question. Instead of statin versus placebo control, it compares high-intensity statin versus low-intensity statin, looking for a dose-response. Now look, again, a trending increase in weight gain that just misses statistical significance. So hold up, pause, think. I believe you're smart, so let's logic this through together. There's a trend towards weight gain when you compare statin to placebo, Figure 1, that just misses significant—a causal connection. There's also a trend towards weight gain when you compare high-intensity versus low-intensity statin treatments. Again, both trends are hovering near statistical significant, that arbitrary threshold. So, what if you isolated people on high-intensity statin therapy and compared them to those not on statin therapy? What do you think? My prediction? You'd see a statistically significant weight gain. And then you get that label— statistically significant. Now, to be clear, they didn't do that test in this study, so I'm speculating. I'd try to do the analysis myself with colleagues if I had access to the individual participant-level data that this group did. Unfortunately, the raw data are guarded, and what tests the group chooses to run and not run is largely up to them. And maybe it's not my place to speculate on whether conflicts of interest influence what tests they'd run and didn't run, but it is interesting to note they had conflicts of interest with many pharmaceutical companies, including Merck, AstraZeneca, Eli Lilly, Novartis, Amgen, Pfizer, and so on. Maybe this influenced what tests they decided to run and how they reported the data, but I guess just take it for what it's worth. There's some transparency for you. That brings us to part 4. Other limitations. There are a ton, but let me highlight 4. The first is beyond the study duration. The paper actually clarifies that these findings can't exclude the possibility of adverse events associated with statin use beyond the study duration. So, the researchers report that statins don't increase dementia risk, but if the median follow-up period was 4.5 years, do we really have data to rule out that statins could, in some people, cause dementia over longer time horizons. No, we can't rule that out. And again, the claim is not that statins cause dementia. It's a claim about the limitations on the data to rule out possible adverse effects. Now, two, on bio-individuality, let me double-click on "some people," that phrase I just used, to really emphasize the point that my goal here is not to label statins or any medication as universally good or bad. Actually, a lot of my content, or at least some of it, focuses on underappreciated appreciated benefits of certain medications. You can see this letter on ezetimibe—super cool data on how it might improve dementia risk. So my goal here truly is to emphasize bio-individuality and the necessity of understanding how drugs function in the context of a specific person. There are so many variables that shape how someone responds to statins and whether the net effect is benefit or harm. You could ask questions like: what's the person's baseline metabolic health, genetics, their Lp, their microbiome? I have deep dives onto each of these topics that I'm going to link below, but the specifics are less important than the point, centrally, which is biology? It's deeply complex. Those complexities should be humbling, and they should be treated as such, rather than flattened into blanket statements delivered with alienating certainty. But moving on. 3. Excluding observational side effects. Let me double-click on the idea that not all adverse effects will be identified in randomized controlled trials, Now let me use a real-world example. Cerivastatin, one of the earlier statins. After its release to the global market, cerivastatin was linked to dozens of deaths, ultimately leading to its withdrawal worldwide. Those harms were never proven in randomized controlled trials, so if we just looked at those specific RCT data, we'd say the results were not statistically significant. But guess what? People still died. Human people still died. And there was enough certainty that cerivastatin caused those deaths that it was withdrawn from the global market. So although this meta-analysis of double-blinded placebo-controlled trials sounds rigorous, it also excludes a lot of data that provide important, maybe even life-or-death insights. And 4, biochemical analyses were excluded as well. This is another key limitation. All outcomes reported were derived solely from adverse event reports, clinical reports, They did not incorporate systematic biochemical analyses, and that matters a lot. So let's relate this to something else we've talked about: the demonstrated effects of statins on slashing GLP-1 levels to a substantial degree. Those are biochemical data, but they wouldn't show up in an adverse event report like something like muscle pain might. Yet, this can plausibly help explain the trending signal around weight gain. I'm gonna do a whole nother letter on the mechanisms behind statins and potential weight gain, including GLP-1s, but also the fact they cut leptin levels in some trials by as much as 67%. You can check that below, I'll drop the link when it's available. So what do we do with that? Do we say just because we haven't collected the highest level of definitive causal proof for a clinical effect that we should just assume there's nothing going on here? Innocent until proven guilty? I hope you don't need me to answer this question for you. Because obviously not. And for more on those data, you can see these videos. But now let me get to the final chapter, 5: The pushback. Broadly, the pushback falls into 3 categories. The first is what I call fearmongering accusations. Kind of what it sounds like. The argument goes, "Nick, you're scaring people away from potentially life-saving medications. They might act against medical advice, be non-compliant, and suffer harm, even death." My response to this is quite simple. People will always bring their own confirmation bias to a situation. I can't control that. So seriously, what would you have me do? Put yourself in my shoes. Would you stay silent simply because some people are going to misrepresent your position, intentionally or out of laziness? Here's my real frustration that most people aren't hearing. It has nothing to do with value judgments around LDL statins or individualized medical care. The problem is the distortion of truth, the p-value propaganda that erects brick walls in front of informed consent. And I suspect, if I know my audience well, that's what frustrates many people, you as well. People have the right to understand what they're doing to their bodies, full stop. And closely related pushback 2: "But statins save lives!" Let's be clear, this is a dodge, although not a good one. Charge! Fire in the hole! In metabolically compromised individuals and in secondary prevention, statins may indeed confer a net benefit. Fine, every tool has its use. But that doesn't mean every tool is appropriate for every case. Hammers are great at driving nails, but improper for opening jars of pickles. Ignoring the risks simply because a benefit exists for someone misses the entire point of medicine. And third, and I find this category kind of amusing, attempting to lump me in with the woo-woo wellness influencers. To paint me as some renegade rejecting evidence-based medicine to build some cultish following. The irony honestly makes me smile. I was valedictorian in undergrad studying cell bio and biochemistry, earned a PhD from Oxford, an MD from Harvard, grew up with two parents who had four doctorates between them, and have two siblings who are also in medicine. My family is literally batting 5 for 5 MDs, with most of us having PhDs too. So I'm not exactly your stereotypical outsider. What these shallow attacks reveal, frankly, is embarrassing hypocrisy. These are the people who claim to value nuance and yet go out of their way to distort my position, or at least who are otherwise not taking time to digest what I'm actually saying, the true nuances. So you can choose to believe me or not, but I am here trying my best to present information to help people. To empower you with information to make the best choice for yourself. Whether that's speaking to the newly discovered potential benefits of pharmacotherapy, like here and here, or their underdiscussed harms, or sometimes just revealing how evidence-based medicine is less than intellectually rigorous, but mostly ego-defensive virtue signaling, because you deserve to know that too. So call me whatever names you'd like. I understand how this game is played, and that may be precisely why this message lands. All I have to do is lay out the incentive structures, the logic, and the evidence, and then let you decide for yourself. That's the beauty of it. That's the goal here. So here's the bottom line. This new study, this landmark study, does not show what the headlines or self-important experts as institutions claim it shows. This does not prove that statin side effects are unfounded, nor does it justify this sweeping certainty about safety. What this shows is something far more subtle and far more important: that signals of harm can be erased with rhetoric. That erasure is not science, it is a communication failure. Because medicine is not about binaries. A probability threshold doesn't dictate biological reality—how a medicine is affecting a person. And the real ethical obligation here isn't to protect a narrative, a drug, or an institution. It's to protect informed consent. And that is why I'm here. Not to be your guru, and certainly not to be your doctor, nor pretend I'm omniscient because some fancy schools gave me fancy pieces of paper. Yes, it's useful for optics, and I'll fight fire with fire, but I'm here to give you the respect you deserve by diving into the data. The good, the bad, the ugly, especially when it's confusing. And if you still don't believe me, at least give me and this community a chance to show you how science and medicine can evolve when we stay authentically curious. So stay curious, come with me on this journey, because even though I express frustration in this video, I am very optimistic about the future because of people like you who are willing to sit through videos like this.