Key points
- Useful information misleads when it is asked to answer the wrong kind of question.
- A diagnosis can guide care while narrowing what others notice about the person.
- Population evidence offers probabilities, not a finished account of one individual.
- Scientific skepticism protects the boundaries of a claim rather than dismissing expertise.
I first read the story of Norman Cousins in college, and for decades, I remembered it with remarkable confidence
Cousins had terminal cancer. He left the hospital, watched comedy films, and somehow laughed himself into remission
It was a terrific story. Clean, hopeful, and easy to repeat. It may even have contributed to my decision to become a physician. It was also quite different from the story Cousins himself told
He had suffered from a severe, disabling illness that physicians believed might be irreversible, but not terminal cancer. His recovery occurred alongside medical care, large doses of vitamin C, changes in his surroundings, sleep, and comedy films that he believed gave him periods of pain relief. Cousins later acknowledged that one case carried little scientific weight. He had hesitated to tell it because he did not want to create false hope
My memory had quietly rewritten it anyway
The qualifications disappeared. So did the competing explanations. Laughter moved from one part of a complicated recovery to its apparent cause. What remained was the version most worth retelling: a dying man had laughed himself well
When the summary takes over
Human beings need summaries. Reality arrives with too much detail to reconsider from the beginning each time we make a decision
A year of medical care becomes a paragraph in the chart. Thousands of observations become an average. A scientific paper becomes a headline. A person’s complicated experience becomes a diagnosis
Usually, this helps. The trouble begins when we forget how much was removed. Apparently evolution neglected to give memory a footnoting system
Philosophers call one version of this mistake a category error. “What color is the number seven?” cannot be answered because it asks one kind of thing to provide another kind of answer
The real-world versions are harder to notice because they begin with useful information. A diagnosis describes part of a patient. A clinical trial tells us what happened among the people studied. A genetic association may affect the probability of disease
None of these facts has to be wrong in order to become misleading. We only have to give one of them more authority than it can carry
The diagnosis that changes what people see
Medicine cannot function without categories. Diagnoses guide testing and treatment. For patients who have spent years without an explanation, a diagnosis can bring relief and direction
Patients do not arrive pre-categorized. They arrive with winding histories, conflicting responsibilities, financial limits, medication sensitivities, prior treatment injuries, and bodies that rarely resemble the clean examples used for teaching
I have watched labels gradually take over a clinical story. A diagnosis first explains a cluster of symptoms. Later, it is used to explain why the patient is upset, why treatment failed, and what will probably happen next. Once the label seems to explain everything, fewer people keep looking
The effect is especially visible with words such as “noncompliant,” “behavioral,” or “drug seeking.” Each may refer to a legitimate concern. Once the label enters the chart, though, cost, side effects, fear, transportation, prior trauma, and the burden of the treatment plan may receive less attention
A patient who arrived in 4K becomes a lower-resolution version in the chart
Sometimes the important question is not whether the diagnosis is correct. It is what the diagnosis has caused everyone to stop noticing
What happens when research travels
A clinical trial may show that a treatment helped a proportion of patients under specified conditions. That information matters. Without population evidence, medicine drifts toward habit, anecdote, and personal confidence
Bias Essential Reads
Do Americans Believe Diversity Is the Enemy of Meritocracy?
How Islamophobia Harms Us All
The patient usually has a different question: Will it help me?
The study offers a starting point, not a personal forecast. The person in the room may differ from the study population in age, medication use, prior treatment history, physiology, or several other ways. Clinical judgment is the work of deciding how much those differences matter
As research moves beyond the paper, its boundaries tend to fall away. An association becomes a cause because it takes fewer words and sounds more decisive. A result in animals becomes a possible treatment for people. A group average becomes advice for everyone
The headline may remain close enough to the paper to withstand a factual objection. The understanding it leaves behind can still exceed the research
Cannabis makes this unusually visible. A regulator, a researcher, a worried family, and a patient may all be discussing the same substance while asking different questions. Each frame contains useful information. The clinical matter becomes distorted when one frame is allowed to settle the whole discussion
Where the claim stops
Scientific uncertainty makes some people uneasy. If research cannot provide a guarantee, they begin to suspect that expertise has little to offer and that every explanation deserves equal consideration
Good skepticism asks more of us. It asks what was measured, in whom, against what comparison, and over how much time. It asks who was excluded and whether a result about a group has quietly become a conclusion about one person
Careful scientific language often sounds qualified because the qualifications carry meaning. “In this population,” “under these conditions,” and “during the study period” tell us where the claim stops
For years, I carried an appealing version of Norman Cousins’s recovery. It had a dying man, an unlikely intervention, and an ending that seemed to prove something reassuring about the mind’s influence over the body
The account he published was less tidy. There was serious illness, medical care, several simultaneous interventions, pain relief, recovery, and genuine uncertainty about how those pieces fit together
I find the corrected version more useful now. It preserves the experience without asking it to become proof. It also reminds me that the easiest version to remember is often the one that discarded the most
The next time a diagnosis, statistic, or headline seems to explain everything, what might come back into view if you asked what had been left out?
Cousins N. Anatomy of an illness, as perceived by the patient. N Engl J Med. 1976;295(26):1458-1463. doi:10.1056/NEJM197612232952605
