Are you more trusting of people who tend to question health claims—what might be called “health skeptics”—than the people who make those claims in the first place?
Should skeptical claims get the same scrutiny?
I recently wrote an article on why people should be just as cautious about accepting the views of health skeptics. A few days later, Dr. Adam Cifu, co-founder of the Sensible Medicine Substack, invited me to write a follow-up piece: “Why you should be skeptical of the next skeptical take you read on Sensible Medicine.”
I said yes, though with mixed feelings. I subscribe to Sensible Medicine because I love its nuance, its community’s critical thinking, and how I learn something every time I read it. Sensible Medicine wasn’t top-of-mind when writing my article.
However, the site does have a focus on scrutinizing health claims and research suggesting benefits, occasionally in ways that make me skeptical of its skepticism.
The hype / hyper-skepticism cycle
As I argued in the original article, there’s a predictable news cycle in which exaggerated health claims produce an equal and opposite contrarian overreach. As part of this cycle, it’s common for skeptics to overstate the certainty with which we can debunk the original health claim, typically serviced by a headline stating the claim is definitively false.
Uncertainty in scientific research doesn’t make a hypothesis implausible. When it comes to observational studies, associations don’t prove causation, but it’s also true that absence of evidence isn’t evidence of absence. In science, evidence rarely reveals a binary yes or no.
People who brand themselves as skeptics of health claims often try to align their approach with the scientific method, noting that skepticism is essential in science. But scientific skepticism is not about reflexively rejecting health claims. It’s about being skeptical of your own hypothesis, whether the hypothesis is that something is beneficial, harmful, or does nothing at all.
Properly understood, skepticism refers to a method, not a position itself. When skepticism turns into an identity, it risks becoming pseudoskepticism.
These points might sound obvious. There’s nothing new or groundbreaking about them. Michael Shermer, founder of Skeptic magazine, has argued that skepticism isn’t a position you take in advance but a methodology for evaluating claims. However, as a journalist who watches how findings get translated into headlines and public advice, I think these ideas have gotten muddled in the hype / hyper-skepticism media cycle.
How this cycle may sometimes apply to Sensible Medicine: a coffee example
As much as Sensible Medicine helps me think about health research and reporting, I see occasional examples of what I described in my article.
One is a piece about coffee from earlier this year. The headline states: “Coffee is great, but it does not prevent dementia.” The article shows how observational data on the benefits of coffee, here from a 2026 JAMA study, are confounded, and there are excellent reasons to worry about confounding. We can’t conclude causality from an observational result, as media outlets sometimes imply. But I’d suggest readers of Sensible Medicine think carefully about the certainty of this article’s headline and whether its conclusions go beyond what we actually know.
The article argues scientists should “stop doing non-randomized observational studies on things like coffee, quinoa, chocolate, saunas, and blueberries” because they’re “worthless.” That seems like an extreme position — the type of view we might expect the hype / hyper-skepticism cycle to generate. An observational study can’t prove a hypothesis, but it can provide a signal that supports a hypothesis — which is consistent with other Sensible Medicine articles. This past July, Adam wrote that observational research gets the wrong conclusion “at least 20% of the time,” suggesting these studies can be informative, albeit with major limitations.
Additional studies that find the same association, if they don’t all have the same bias, can increase confidence. In the case of coffee, a 2024 meta-analysis of 33 studies reported lower cognitive-disorder risk among coffee drinkers. There’s also contradictory evidence; some research reviews have found no association. Another 2024 meta-analysis of cohort studies found no significant overall association when comparing the highest with the lowest coffee consumption, while still finding a nonlinear association suggesting lower dementia risk for people having 1 to 3 cups daily. It rated the evidence low-certainty.
None of this means coffee “does not prevent dementia”—or, more precisely, that coffee doesn’t reduce dementia risk. The more defensible take-home at this point is that we need more evidence to know whether it does or doesn’t.
As Meir Stampfer, an epidemiologist at Harvard’s T.H. Chan School of Public Health and coauthor of the JAMA study, told me: “In science we make a distinction between data that are worthless from data that are worth less” than stronger evidence.
There’s another reason for skepticism about the article’s skepticism: it doesn’t discuss other types of research pointing to the benefits of regular coffee drinking, or at least its biological plausibility. It’s well established from lab research that coffee contains biologically active compounds with plausible effects on neuroinflammation, oxidative stress, amyloid and tau pathways, and related biology. These compounds include caffeine, chlorogenic acids, caffeic acid, and trigonelline.
Convergence across study types, or triangulation, increases the likelihood of an association being causal. Turning the weaknesses of observational studies into the whole story seems to skew toward hyper-skepticism; it can obscure what we’ve learned from other methods, the mosaic of evidence needed to address questions about health.
Another example: exercise
Another Sensible Medicine article argues that many observational studies on exercise are “wasteful” and, more broadly, that the optimal amount and type of exercise is essentially unknowable. Again, there’s a lot of truth to the point that observational research involves important shortcomings. I learned plenty from this article’s analysis. But I think we’re entering hyper-skepticism territory with such a definitive statement that it’s a waste of time to study different types of exercise.
To the article’s credit, it doesn’t claim that all observational work is worthless. Instead, the article says it focuses only on observational studies of exercise similar to the examples discussed. However, some of the reasons given for dismissing these exercise studies—like the impossibility of completely removing confounders—aren’t unique to them; they’re structural features of observational epidemiology.
As with coffee, I think readers should consider that although observational research can’t prove causation, other research methods add credibility to the hypothesis that different forms of exercise affect the body and health outcomes differently. Experimenters can manipulate the intensity and type and observe the resulting physiological adaptations. We have RCTs comparing HIIT, sprint intervals, and moderate continuous exercise; training studies measuring VO₂ max; vascular-function experiments; metabolic studies; muscle biopsies; and mechanistic work. Distinct responses include mitochondrial signaling, muscle-fiber recruitment, vascular function, glucose metabolism, and cardiovascular adaptation.
These other forms of research don’t prove differences in lifespan, but they help us understand physical changes that affect health, which ultimately may contribute to how long people live. So why does the article conclude the whole endeavor of studying amounts, types, and lengths of exercise is “as silly as it is unknowable”?
Skeptical perspectives offer great value. It’s important for Sensible Medicine to keep pointing out confounding, weak study designs, and exaggerated causal claims. But we shouldn’t exempt these takes from scrutiny just because their rhetorical posture sounds more scientific. We should apply skepticism just as rigorously to the skeptical conclusion itself.
Matt Fuchs is a health and science journalist from Maryland. He is currently a Knight-Wallace Fellow at the University of Michigan.
Photo by Jamie Street on Unsplash



The article could’ve been much shorter. “Study A shows that condition B correlates with result C. Just because this isn’t evidence of causation doesn’t mean it’s not true.” No, It just means that it’s irrelevant.
Sorry to be harsh, but journalists thrive on observational studies because they give them something to write about. They are guaranteed click bait that produce ad profits. They also do a great disservice to the public.