The Public Health Mistake: Treating Risk as an Individual Decision

What if the most dangerous question in public health is not, “Is this good or bad for one person?” but, “What happens when millions of different people encounter it at different moments in their lives?”

That shift sounds small. It is not. It separates a serious assessment of risk from a persuasive sales pitch, a clinical judgment from a stereotype, and prevention from the habit of explaining harm only after it becomes visible.

Tobacco regulation and dementia appear to belong to different intellectual worlds. One concerns whether a new product should be allowed onto the market. The other concerns the loss of memory, reasoning, and everyday independence. Yet they converge on a difficult question: How should we make decisions when the consequences are distributed across different people, unfold over time, and cannot be captured by a single average?

The answer offers a general framework for thinking about public health, technology, aging, and personal choice. It also exposes a common error: mistaking the absence of obvious harm for evidence of safety.

The Average Person Does Not Exist

A product can benefit one group while harming another. A change that helps current tobacco users stop may also make tobacco use more attractive to people who have never used it. A person who has no cognitive symptoms at age seventy may later develop serious impairment, while another person reaches their nineties without dementia. In both cases, the population is not a single body moving in one direction. It is a collection of trajectories.

This is why a meaningful public health judgment must ask more than whether an intervention produces benefits in some subjects. It must examine who benefits, who bears the risk, and who is newly exposed because the intervention exists.

Consider a hypothetical nicotine product designed to deliver fewer harmful substances than a conventional cigarette. An individual smoker might view it as a useful step toward quitting. But that is only one pathway. Another smoker might use both products rather than replacing one with the other. A young nonuser might find the new product appealing and begin using tobacco. A former user might relapse because the product appears safer. The same object has different consequences depending on the person encountering it and the behavioral path that follows.

A population wide decision therefore cannot be based on the most favorable use case. It must weigh the whole distribution of likely outcomes. This includes current users, former users, nonusers, and people who may be affected indirectly by changing norms, access, or exposure.

Dementia reveals the same principle from a different angle. Cognitive decline is often treated as if it were simply an inevitable consequence of getting older. But age is a risk context, not a diagnosis. Many people live into their nineties without dementia, while others experience a level of cognitive loss that interferes with daily life. The relevant question is not, “Is this person old?” It is, “Has cognitive functioning declined enough to alter the person’s ability to think, remember, reason, solve problems, and manage ordinary life?”

The distinction matters because a broad category can conceal radically different individual trajectories. “Older adults” is not a diagnosis. “A new tobacco product” is not a complete risk assessment. Labels become dangerous when they replace investigation.

Averages describe populations. Decisions affect trajectories.

Visibility Is Not the Same as Importance

Public attention is drawn to events that are immediate, vivid, and easy to attribute. A person uses a product, then feels an effect. A person forgets a familiar route, then a family notices a pattern. But many public health consequences are slower, indirect, or distributed across groups. They can be real without being immediately visible.

This creates a timing problem. When a new product is introduced, the most visible evidence may concern short term experience: taste, sensation, satisfaction, or whether an existing user finds it easier to switch. The less visible questions concern initiation, continued use, dual use, relapse, and the effects of manufacturing quality over time. A responsible decision must look beyond what can be noticed first.

The same temporal problem appears in cognitive health. Dementia is defined by interference with daily life, but the underlying process may be recognized only after years of subtle change. A missed appointment is easy to dismiss. Repeated difficulty with familiar tasks is harder to ignore. The final impairment is often more visible than the earlier trajectory that led to it.

This suggests a useful model: public health decisions should track transitions, not just states.

A state is a condition at a particular moment: a person currently uses tobacco, does not use tobacco, or has cognitive impairment. A transition is movement from one condition to another: nonuse to use, use to cessation, normal functioning to impairment, or mild difficulty to loss of independence.

Transitions are where policy has leverage. They are also where risk is frequently underestimated. A product may appear acceptable when evaluated only among existing users, yet produce a harmful population effect if it increases the number of people entering tobacco use. Similarly, a cognitive change may seem like normal aging when viewed as an isolated incident, yet signal a consequential transition when observed as a pattern.

A practical way to think about this is to ask four questions whenever a risk is being evaluated:

1. What is the current state? Who is already exposed, impaired, or vulnerable?
2. What transitions are likely? Who may start, stop, worsen, recover, or switch behaviors?
3. How quickly do consequences appear? Are we measuring immediate experience or long term function?
4. What evidence would reveal harm before it becomes severe?

These questions turn passive observation into active surveillance. They also resist the comforting but flawed idea that a problem matters only once it is dramatic.

The Hidden Cost of “Safer” Options

One of the most difficult ideas in risk assessment is that reducing risk for an existing user can still increase total harm if it expands the exposed population.

Imagine a town with one hundred current smokers. A new product helps twenty of them stop smoking entirely, but its appealing design leads ten nonusers to begin using tobacco. The arithmetic is not enough to settle the question, because the health consequences are not interchangeable. The twenty who quit may gain substantial benefits, while the ten new users may face years of exposure and an elevated chance of progressing to more harmful products. The point is not that the new product must be rejected. The point is that the favorable outcome for one group cannot be treated as proof of a favorable outcome for everyone.

This is the exposure expansion problem: an intervention can lower risk per user while increasing the number of users. The two variables must be considered together.

The idea has applications far beyond tobacco. A medication with fewer side effects may be prescribed to many more people, including those with little need for it. A digital tool that makes life easier for most users may encourage dependency in people who are already vulnerable. A screening test may identify some serious conditions while also producing anxiety and unnecessary interventions in healthy people. Safety is never only a property of an object. It is a property of an object placed inside a social system.

That is why manufacturing, processing, packing, and controls matter. A product is not merely its advertised chemical profile or intended use. It is also the result of a production system, a distribution system, a marketing environment, and a set of real world patterns that may diverge from the instructions.

Dementia offers a parallel lesson about complexity. Many people with dementia have Alzheimer’s disease alongside one or more related disorders that share clinical or brain scanning features. The observable condition may be real even when the underlying causes are mixed. A single label can describe the lived problem without fully explaining its biological composition.

This is a broader principle of diagnosis: the visible outcome and the causal structure are not always the same thing. If several pathways converge on the same impairment, then a solution aimed at only one pathway may be incomplete. Likewise, if different behaviors produce different population outcomes, then a product cannot be evaluated solely by its intended purpose.

From Product Approval to Cognitive Care: A Shared Logic

The strongest connection between these subjects is not that one causes the other. The available ideas do not establish such a claim, and it would be irresponsible to invent one. Their deeper connection is methodological.

Both require a form of reasoning that is population aware, trajectory sensitive, and skeptical of simple categories.

A useful framework is the three lens test.

Lens one: The individual lens

What happens to a particular person? Does an existing user become more likely to stop? Does a person with cognitive difficulties lose the ability to manage daily activities? Individual experience is concrete and morally important. It is where abstract policy becomes lived reality.

Lens two: The transition lens

Who is likely to enter or leave a state because of the intervention or condition? Does a nonuser begin using tobacco? Does a person with minor cognitive changes progress toward dependence? Does a current user move toward cessation or merely add another product? This lens captures direction and momentum.

Lens three: The system lens

What happens when the choice is repeated across a population and shaped by institutions? How is the product manufactured and controlled? How does availability alter behavior? How do families, clinicians, and public agencies respond when impairment becomes common? This lens reveals effects that no individual case can show.

Weak reasoning usually stops at the first lens. It asks whether one person can benefit, or whether one person appears fine. Strong reasoning moves through all three.

This framework also clarifies why early action is difficult. At the individual level, evidence may seem ambiguous. At the system level, waiting can allow a small transition rate to accumulate into a large burden. If one million people are exposed, even a rare adverse pathway may affect many thousands. Conversely, a modest benefit repeated across a large group may matter enormously.

The proper response is not panic or certainty. It is calibrated humility: make decisions using the best available evidence, identify the groups at risk, monitor transitions, and revise policy when real world outcomes diverge from predictions.

A Better Definition of Protection

“Protection” is often imagined as the elimination of risk. That standard is impossible for most products and conditions. A more useful definition is the disciplined management of tradeoffs across time and across groups.

For a new tobacco product, protection requires asking whether the total balance of benefits and risks is favorable for the population, not merely whether the product is preferable to one existing product for some current users. It requires attention to nonusers, initiation, cessation, and the reliability of the production process.

For cognitive health, protection requires refusing two opposite errors. The first is fatalism: assuming that dementia is simply normal aging and therefore nothing can be learned from early changes. The second is overdiagnosis: treating every lapse as evidence of disease. Protection means distinguishing ordinary variation from meaningful interference with daily life, while remaining alert to patterns and transitions.

These are not separate habits of mind. They are versions of the same discipline: do not confuse a category with a conclusion.

“Reduced exposure” is not automatically “safe.” “Older” is not automatically “demented.” “A benefit for current users” is not automatically “a benefit for the population.” Each statement may contain a useful observation, but none is sufficient without context.

Key Takeaways

• Evaluate who changes, not only who benefits. Whenever you consider a health intervention, ask whether it could attract new users, create new dependencies, or shift risk toward people who were previously unexposed.

• Track transitions early. Keep records of repeated changes in behavior, function, or symptoms. A pattern often contains more information than an isolated event.

• Separate age and circumstance from diagnosis. Aging can increase vulnerability, but it does not by itself establish cognitive disease. Focus on changes that interfere with daily life.

• Inspect the system around the product. Intended use is only one part of risk. Manufacturing quality, controls, marketing, availability, and actual behavior can change the outcome.

• Use calibrated humility. Make the best decision available now, state what remains uncertain, and create a way to detect whether the prediction was wrong.

The deepest lesson is that public health is not a contest between optimism and alarm. It is an exercise in seeing what ordinary attention misses: the nonuser who becomes a user, the current user who does not actually quit, the early cognitive change dismissed as age, the mixed cause hidden behind a single label, and the manufacturing failure concealed by a promising design.

A society protects people not by pretending risk can be abolished, but by refusing to evaluate it in isolation. The right unit of analysis is not the product alone, the symptom alone, or the average person alone. It is the changing relationship among individuals, trajectories, and systems.

Once we learn to see health decisions this way, a new product is no longer merely something someone might choose, and a cognitive symptom is no longer merely something someone might forget. Both become signals in a larger question: What future becomes more likely when this possibility enters the world?