Reading the Forecast Backward
Why "most treated people get no benefit" misunderstands prevention
Preventive care rests on risk prediction and risk reduction. Those are two separate things, and confusing them produces most of the bad arguments against prevention.
The risk prediction part is nearly impossible at the level that matters to a patient, which is whether a particular person will have an event. The best we can do is sort people broadly into loose buckets of cardiovascular disease (CVD) risk, and even those buckets are wide and uncertain. We cannot tell an individual that they will have a heart attack, only that people who resemble them have CVD events at some rate over some time period.
The risk reduction part is where the misleading numbers get quoted. An intervention lowers average risk by some relative proportion, usually around 25% to 33% for many preventive therapies. That is a relative reduction, and keeping the difference between relative and absolute risk reduction in mind is where the whole discussion can succeed or fail. If one applies a 30% relative risk reduction to a person with a genuinely high predicted risk, say a 20% to 30% chance of an adverse event over ten years, then the absolute benefit over that ten-year window comes out to 6% to 9%. And even that figure understates the value of treatment, because risk reduction keeps accruing in the years beyond the window during which it was measured.
Sadly, many people respond to the arithmetic by declaring that over 90% of those treated get no benefit, and they mean it as an indictment of preventive care. But that claim is a retrospective count dressed up as a prospective fact. After the fact, sure, you can identify who had events and who did not. But the decision to treat is made a priori, when everyone carries the same lowered probability and no one’s outcome is known. The probability shift is a property of each person’s situation at the moment of treatment, not a lottery ticket that only the eventual event-havers cash. Counting the untouched afterward and calling the treatment wasted confuses the two frames.
It is the same error as reading a weather forecast backward. A 30% chance of rain was not wrong because the day stayed dry. The probability was a real statement about a distribution of possible days, made before you knew which one you would get. The therapy lowered everyone’s probability at the moment of the decision, and that reduction was real for each person even though only some were ever going to have the event it prevents.
This holds across preventive care, not only for statins, but also for blood pressure lowering and anticoagulation in atrial fibrillation, where the same objection surfaces in the same form and fails for the same reason. In each case, the intervention shifts a probability for everyone treated, and in each case only a minority were ever destined to have the event that the shift prevents.
None of this makes the decision automatic. The benefit is a probability shift, but so are the harms, such as bleeding on an anticoagulant, or the side effects, cost, and daily burden of any of these therapies. When the absolute benefit is modest, whether it justifies those harms is a genuine question, and the answer depends on what the individual values. That is what makes these decisions preference-sensitive rather than automatic. And preferences in this setting are more than tastes, like “I prefer vanilla over chocolate.” They are real risk assessment inputs, because the probabilities of benefit and harm are close enough that individual values legitimately tip the balance.
But the critical point is that the reduction in risk is real for every treated person at the moment of the decision, whatever the eventual outcome, because that is when the probability is shifted and that is when the choice is made. Treating that shift as nothing, on the grounds that most treated people would have done fine anyway, reads the forecast backward.
The age old problem in estimating benefit-risk in preventative therapies (including vaccines). You always know who had an adverse event but never know who did not have en event due to the treatment so the patient benefit risk ratio is always skewed emotionally. Good graphics on this by David Spiegelhalter: "Patient reactions to a web-based cardiovascular risk calculator in type 2 diabetes: a qualitative study in primary care" https://pubmed.ncbi.nlm.nih.gov/25733436/
One of the hardest shifts is helping people think in probabilities instead of certainties. We tend to judge decisions by outcomes, but in prevention, the quality of a decision is determined by the information available at the time, not by whether the unfavorable event ultimately happened.