The Dangerous Habit of Trusting Risk Without Proof
Hatched by Carlos Franco
May 05, 2026
9 min read
5 views
33%
When a risk label becomes a substitute for real evidence
What if the most dangerous thing in medicine is not uncertainty itself, but our tendency to confuse a plausible explanation with a proven outcome?
That question sits underneath two very different clinical realities. In one, dementia is often spoken about as if it were simply a normal extension of aging, even though many people live into their 90s and beyond without it. In the other, a treatment can be approved for a narrow, highly specific purpose based on a surrogate measure, even when there is no controlled evidence that it improves the outcome patients care about most.
These two facts point to the same deeper problem: humans are extremely good at managing risk narratives, and much worse at proving actual benefit. We like categories. We like warning signs. We like interventions that appear to move an important number in the right direction. But biology does not reward our desire for neatness. It rewards outcomes, and often in ways that are less convenient, less obvious, and harder to measure.
That is the hidden link between dementia and preterm birth prevention. Both force us to confront a difficult truth: a condition can be real without being inevitable, and a treatment can be rational without being truly effective.
The first confusion: treating age like destiny
Dementia is not a normal part of aging, even though it becomes more common with age. That distinction matters more than it first appears. When people blur the line between aging and disease, they stop looking for causes, interventions, and possibilities. They begin to treat cognitive decline the way people treat gray hair, as something unpleasant but simply expected.
That mindset is corrosive. It changes how families interpret memory lapses in a loved one. It changes how clinicians respond to early symptoms. It changes how society allocates urgency and research attention. If decline is framed as inevitable, then prevention becomes optional and diagnosis becomes sentimental rather than clinical.
But the fact that many people remain cognitively intact into advanced age reveals something important: aging increases vulnerability, but it does not erase individual variation. Some brains maintain resilience. Others become vulnerable to overlapping processes, many of which may be clinically labeled as Alzheimer’s disease, related disorders, or mixed forms of decline.
This is where the more interesting question emerges. We often ask, “What is the disease?” But the better question is, “What changed the balance between resilience and vulnerability?” In other words, the real problem is not simply the presence of damage. It is the failure of the brain’s protective systems to keep damage below a threshold that disrupts daily life.
The boundary between normal function and disease is often not a line, but a tipping point.
That insight matters far beyond neurology. It helps explain why two people of the same age can look very different clinically, and why a diagnosis often reflects accumulated interaction rather than a single cause. The diagnosis names the result. The struggle is in understanding the process.
The second confusion: mistaking a surrogate for the thing itself
Now consider a very different clinical story. A medication can be indicated to reduce the risk of preterm birth in a narrowly defined group: women with a singleton pregnancy and a history of spontaneous singleton preterm birth. Yet the evidence basis described is not direct improvement in neonatal mortality or morbidity. Instead, the measure is improvement in the proportion of women who delivered before 37 weeks.
That sounds close to the outcome people care about, but it is not the same outcome. Delivering before 37 weeks is a meaningful signal, but it is still a signal. It is a proxy, a surrogate. And surrogates can mislead.
A useful analogy is a car’s dashboard. If the fuel gauge moves upward, you know something has changed. But you do not yet know whether the engine runs better, whether the car is safer, or whether the destination is closer. A better number is not always a better life. Medicine is filled with interventions that improve laboratory values, imaging findings, or intermediate markers without reliably improving the outcomes patients actually experience.
That is why narrow indications matter. The more specific the evidence, the more carefully we should respect its boundaries. If a treatment was shown to work in one population, we cannot assume it works in all populations with similar risk factors. Risk factor overlap is not the same thing as evidence of benefit.
This is a subtle but profound lesson. A treatment can be biologically plausible, statistically suggestive, and still not be broadly validated. That does not make the treatment meaningless. It makes it conditional. It means the claim is precise rather than universal.
And precision is one of medicine’s least glamorous virtues. People prefer broad promises. But broad promises are exactly where evidence gets stretched, misapplied, or overgeneralized.
The shared mistake: replacing outcomes with stories
At first glance, dementia and preterm birth prevention seem unrelated. One concerns cognitive decline in older adults. The other concerns pregnancy management. Yet both expose a common failure in reasoning: we often let a compelling story stand in for a demonstrated outcome.
With dementia, the story is that aging naturally brings forgetfulness, so decline feels expected and therefore less urgent. With preterm birth prevention, the story is that reducing an early delivery marker must mean better neonatal health, so the proxy feels close enough to the real thing. In both cases, the story is emotionally satisfying. In both cases, it can also be wrong.
The deeper issue is not just medical. It is epistemic. Humans are drawn to explanations that feel coherent before they are proven effective. We like narratives because they reduce anxiety. But in medicine, a coherent narrative can quietly become a dangerous simplification.
Think of it this way: a forecast is not the weather. A risk score is not a fate. A surrogate marker is not a patient’s life. These substitutions are tempting because they make complexity manageable. But when the substitution becomes invisible, the map starts to replace the terrain.
The real danger is not that we lack information. It is that we mistake the most legible information for the most important information.
This is why evidence-based medicine is not just about data. It is about resisting the human urge to overtrust what is easiest to measure. In dementia, the thing easiest to dismiss is early cognitive change. In obstetrics, the thing easiest to overtrust is the endpoint that can be counted before the baby is born. Different contexts, same cognitive trap.
A better framework: the three thresholds of medical truth
To make sense of these cases, it helps to use a simple framework: medical truth passes through three thresholds.
1. The threshold of plausibility
This is where an idea makes biological sense. A treatment might reduce preterm birth because of a hormonal mechanism. A memory problem might suggest a neurodegenerative process because the symptoms fit a known pattern. Plausibility is useful, but it is only the beginning.
2. The threshold of measurable change
This is where an intervention changes a marker, score, or intermediate outcome. A medication may shift the rate of delivery before 37 weeks. A cognitive test may detect early impairment. These changes matter, but they do not yet prove that the person’s life is better.
3. The threshold of lived benefit
This is the hardest threshold. It asks whether people live longer, function better, suffer less, or maintain independence. It is the level at which medicine stops being an elegant theory and becomes an improvement in human experience.
Most errors happen when we stop at the second threshold and speak as if we reached the third.
That framework also clarifies why some conditions demand humility. Dementia is not one disease but often a cluster of overlapping brain disorders. Preterm birth risk is not one mechanism but a web of interacting factors, many of which may not respond to a single intervention. Complexity resists simple solutions, and when we ignore that, we end up worshipping proxies.
A mature medical culture would ask three questions before celebrating any intervention:
- Does it make biological sense?
- Does it change a measurable marker?
- Does it improve what matters to people?
If the answer to the third question is absent, the result should be considered provisional, not triumphant.
What this means for patients, families, and clinicians
This is not an argument against treatment. It is an argument against confusing risk management with outcome certainty.
For families facing dementia, the lesson is to reject fatalism. Decline is not just age doing what age does. It is a signal that deserves attention, diagnosis, and planning. The earlier the distinction is made between normal aging and cognitive disorder, the more room there is for support, safety, and quality of life.
For pregnant patients and clinicians, the lesson is different but related. A treatment aimed at preventing preterm birth should be judged within the population where evidence exists, and with an awareness that reducing a surrogate outcome is not identical to preventing harm to the baby. That does not mean ignoring interventions. It means choosing them with clear eyes.
There is also a broader lesson about how to live with uncertainty. Good medicine does not require pretending that risk can be eliminated. It requires knowing which risks are meaningful, which claims are validated, and which metrics are merely convenient stand-ins.
In practical terms, that means asking sharper questions:
- Is this problem being treated as inevitable when it is not?
- Is this intervention proven for this exact situation, or just for something that looks similar?
- Am I being shown an outcome, or a proxy for an outcome?
- If the marker improves, do we actually know that people do better?
These questions do not make medicine slower for the sake of being difficult. They make it more honest. And honesty is often the difference between care that feels reassuring and care that is truly effective.
Key Takeaways
- Do not confuse common with normal. A condition can become more frequent with age or risk without being an inevitable part of life.
- Do not confuse a surrogate with a real outcome. A better marker is not automatically better health.
- Respect the boundaries of evidence. A treatment shown to work in one narrow group should not be assumed effective in broader groups.
- Ask the three threshold questions. Is it plausible, measurable, and actually beneficial in lived experience?
- Watch for storytelling bias. When a medical explanation feels neat, pause and ask whether it is proven or merely satisfying.
The real lesson: evidence is a discipline of restraint
The deepest connection between dementia and preterm birth prevention is not that both involve medical risk. It is that both expose how easily humans slide from pattern recognition into overconfidence.
We want aging to explain decline. We want a changed marker to explain a successful treatment. But biology is not obligated to honor our desire for simplicity. It often gives us partial truths, conditional truths, and truths that must be held with restraint.
That is why the most important habit in medicine may be the hardest one to practice: the willingness to stop at the point where evidence ends, even when the story feels complete.
Because once we begin confusing plausibility with proof, we risk two failures at once. We may dismiss a serious disease as just aging. Or we may elevate a narrow intervention into a universal solution.
The deeper discipline is to keep asking, with both humility and precision: not just, “Does this make sense?” but, “Does this actually change what matters?” That question is where better medicine begins, and where better thinking begins too.
Sources
Hatch New Ideas with Glasp AI 🐣
Glasp AI allows you to hatch new ideas based on your curated content. Let's curate and create with Glasp AI :)
Start Hatching 🐣