The Headline-Grabbing Statin Study Everyone Is Oversimplifying
A new Lancet study says statin side effects are overblown. Headlines are running that millions more people should be taking statins. Let's see truth through nuance...
Two days ago, a massive paper dropped in one of the world’s top medical journals, The Lancet.
It was a massive labor: a meta-analysis of data from large double-blind randomized controlled trials (RCTs) to assess the negative side effects caused by statins.
And even within the first day of publication, it landed on legacy and social media with the force of an academic sledgehammer—BOOOM!
Headlines are outright advising that millions more people should be taking statins.
The thrust of the paper is simple: the side effects of statin therapy are overblown.
The evidence base for that claim comes from the “best possible” type of data: high-quality RCTs that can separate causal effects of statins from confounding and placebo/nocebo effects.
And, as you might expect if you’ve been following StayCurious Metabolism for a while… the headlines (and most doctors) missed something big—literally. (That will make sense later.)
We’re going to get into the nuances. And I promise to show, not just tell. But first, let me describe the study.
Study Design & Methods
A meta-analysis of double-blinded statin trials means they pooled data from studies where:
Patients were randomized to statins or placebo
Both the patient and the provider were blinded to assignment
These studies are designed to help determine whether effects are causal and to reduce confounding.
The authors also applied additional inclusion criteria to increase rigor:
At least 1,000 participants per trial
A scheduled treatment period of at least 2 years
In the end, they included 19 total trials with a median follow-up of 4.5 years and a total of 123,940 individuals.
Using adverse effects listed on statin product labels, they assessed 66 “additional undesirable outcomes,” beyond effects on muscles and diabetes risk (which they present as “previously reported”).
Their headline conclusion: the data do not support a causal relationship between statin therapy and most of the conditions assessed.
So… case closed? Statins are safe. Let’s put them in the drinking water and make cardiovascular disease an orphan disease.
Or maybe we should take a moment to more rigorously assess the data…
How to Understand Figure 1
Now is where we get to have some fun together, my StayCurious nuance nerds!
Let’s begin with Figure 1 (below). I know it looks messy, but here’s the way to read it:
A risk ratio (RR) (red arrow) represents the relative risk of an outcome in the statin group versus placebo.
RR > 1 = increased risk with statins
RR < 1 = decreased risk with statins
The circles represent the average estimate, and the line through them is the confidence interval—how much uncertainty surrounds that estimate.
Now, listen carefully because the next part is critical.
If that line crosses RR = 1, the effect is deemed “non-significant.”
So, even low-intensity statin therapy cause a “significant” increase in new-onset diabetes (pink circle)
But statins the effect of statins on gynecomastia (cyan circle) is “non-significant.”
The effect size on gynecomastia—male breast enlargement—is actually larger. But the confidence interval is wider, and it crosses that magic probability 1.0 line. So, it’s “non-significant.”
Let me double click on this because it’s important.
“Significant” has nothing to do with an effect size or biological truth.
“Significant” is about the probability the effect is truly caused by the intervention given a relatively arbitrary probability threshold that we’ve chosen.
In general, that threshold is 5%, as in we need to be “95% certain” that a negative effect is caused by statins in order to call it “significant.”
Thus, if we are—statistically speaking—94% certain a negative effect is caused by statins the effect is “non-significant.”
In the form of the graph, that is represented as the black lines (confidence intervals) running through the grey circles.
And if the black line crosses the 1.0 mark, even slightly, the effect is “non-significant.”
With me so far? I hope so.
But let’s hammer the point home with an example because I really need you to understand this concept.
*Nuance note: For those who are more into statistics, I discuss the paper’s methods regarding on ‘false discovery rate’ (FDR) in the post-script
Thought Puzzle: 5X More Death but “Non-Significant”
Imagine I had a treatment “Z” and conducted a RCT with the primary outcome being death.
If the RR of death for “Z” is 5 (meaning a fivefold increased risk of death), but the confidence interval crossed 1.0—say the confidence interval was 0.99 to 9.01—we’d still call that “non-significant.”
There’s a clear trend to a horrible outcome. A fivefold increased risk of death. The biological reality hasn’t changed. We’re just not quite “certain” enough by a hair.
But, if the study had just a little bit more power the confidence interval would shrink, and we might end up with a situation in which intervention “Z” has a RR of 5 for death (same as before) with a confidence interval of 1.01 – 8.99.
This time, it doesn’t cross that magical 1.0 line.
Thus, we’d conclude, shockingly, “Oh my goodness! Z causes a 5X in death!”
That’s why “non-significant” can be a misleading phrase in real-world interpretation.
Now, if you’re with me so far, I invite you to be a medical detective:
What Do You See in Figure 1?
Look above… and look carefully…
Statins Cause a Trend to Increase in Weight Gain
There are a lot of trending negative effects. But one I find particularly interesting is weight increase.
Look at the gray circle: there’s clearly at least a moderate trend toward increased risk for weight gain with statin therapy.
But the confidence interval just barely crosses the 1.0 line (yellow line segment).
Meaning?
In the most rigorous trials, statin therapy is trending toward a causal effect on weight gain—just shy of the arbitrary “statistical significance” cutoff.
Funny… that didn’t make it into any of the major news articles on this study. Go figure.
So, to be clear, if the data imply we’re “85–90% certain” statins cause weight gain, that still gets labeled “non-significant.”
And THAT is what most people miss. And if you’re following along, you’re in the top 1% of critical thinkers. Congratulate yourself! WAHOO!
(Or go back, re-read, and then congratulate yourself!)
Now take a moment to pause and explore Figure 1 above.
If you look across the graph, you see multiple outcomes where the trend is toward harm—even if they don’t quite hit that magic line.
And we do see clear (significant) signals in places:
Increased risk of new-onset diabetes
Increases in liver damage markers
Edema/swelling.
So, this is already more complex than “statins are safe, more people should take them.”
But this complexity is obscured in headlines that read “Millions More People Should be Taking Life-Saving Statins.”
Otherwise put, there are concerns here.
But if they don’t cross the statistical threshold, we’re going to talk like they basically don’t exist—because we can use the word “non-significant.”
That’s not evidence-based medicine. That’s deception or—if I’m being generous—lazy thinking.
But what comes next—I love even more…
This where you put your critical thinking hat on tight and see what most doctors and scientists don’t see. What follows takes us into true StayCurious sleuth territory! We’ll unpack:
The dose–response clue hiding in plain sight
Limitations of these data
“Beyond the Study Duration” Excuse
The Issue of Bio-individuality
Excluded Biochemical Analyses
Excluded Observational Data (and risk of premature death)
The sleight of hand around “burden of proof”
Why an “innocent until proven guilty” is a recipe for poor outcomes
You’ll also get access to the full StayCurious Heart Health archives!
If you want to be part of a truly special community—one united by bona fide curiosity and intellectual integrity—there’s no better time to join than now.
