Regular readers know that among the many demigods in my polytheistic pantheon is the subaltern who stands guard to keep the faculties of Critical Thinking from being waylaid by cant or derailed by sensationalism.
This brings us to our two seemingly far afield and disconnected-from-each-other texts for today:
The flu shot piece (NYT, yesterday) reports that
As soon as swine flu vaccinations start next month, some people getting them will drop dead of heart attacks or strokes, some children will have seizures and some pregnant women will miscarry. […]
Every year, there are 1.1 million heart attacks in the United States, 795,000 strokes and 876,000 miscarriages, and 200,000 Americans have their first seizure. Inevitably, officials say, some of these will happen within hours or days of a flu shot. […]
“There are about 2,400 miscarriages a day in the U.S.,” said Dr. Jay C. Butler, chief of the swine flu vaccine task force at the federal Centers for Disease Control and Prevention. “You’ll see things that would have happened anyway. But the vaccine doesn’t cause miscarriages. It also doesn’t cause auto accidents, but they happen.”
To counter what the Centers for Disease Control and other federal agencies fear may be an all-but-inevitable surge in hysteria and alarmism about “side effects” of the vaccine, they have set up a war room, as well as a constantly updated Facebook page and Twitter feed. We should all, at the very least, wish them luck.
Otherwise we may find ourselves subject to the satanic twin of my Critical Thinking subaltern, “regulation by anecdote.” Think this doesn’t happen?
As Exhibit A, I would give you Sarbanes Oxley, a stupendous reaction to a handful of amazing outlier events. While Enron and Worldcom may have “shocked the conscience” (the strenuous and increasingly convoluted machinations of their perpetrators certainly shocked my conscience at the time), they lay so far outside the norm of corporate behavior that to base the most far-reaching reforms of our securities laws in several generations upon them seemed akin to legislating “knife ownership control” statutes based on the conduct of Jack the Ripper.
The larger moral from the swine flu story?
Human beings–by and large to our enormous credit–excel at finding patterns, teasing out cause and effect, and deducing what will be the probable after from the current before. But we are also more likely to exaggerate the actual probability of “A causes B” the more dreadful B is. Conversely, the more benign B is, the less likely we are to imagine causation. The benefit or peril from B, of course, has precisely nothing to do with the likelihood A was involved.
En garde.
The melanoma piece reports:
In recent years there has been a sharp rise in reported cases of malignant melanoma, the deadliest form of skin cancer [a 48% increase in just 13 years]. But a British study has found evidence that the epidemic may be due at least in part to “diagnostic drift,” a growing tendency to identify and treat benign lesions as malignant cancers. […]
The British researchers [also] found something odd in the data: almost all of the increase was in diagnoses of the earliest stage of the disease, where it is difficult, and sometimes impossible, to tell a malignant lesion from a melanocytic nevus, a type of benign mole. There was no change in the combined incidence of the later stages of the disease, and mortality increased only slightly.
First of all, I really wish I could take credit for coining the phrase “diagnostic drift”–a marvelously pithy piece of shorthand for a complex phenomenon. Alas, ’tis not to be.
The real interest of the piece comes in how it exposes the thought processes of dermatologists and epidemiologists discussing whether the increase is real, or drift.
The first clue that it really is drift and not real comes from the observation that the increase is disproportionately resident in very early stage diagnoses:
Melanoma is commonly diagnosed at various stages of severity, and if its incidence were truly increasing, the authors write, there would be increases found in all stages of the illness, not just the earliest. “We think that in borderline cases dermatologists and pathologists are erring on the side of caution,” said the lead author, Dr. Nick J. Levell.
Arguing that it’s real and not drift is Dr. Darrell S. Rigel, a professor of dermatology at New York University, who says:
“Every study over the past 10 years has shown that the absolute number of melanomas is rising,” […] The death rate from melanoma has also been going up, Dr. Rigel said, but so has the survival rate. In other words, while more people are getting the disease and dying from it, early identification and treatment has simultaneously allowed more people to survive.
Dr. Levell, our “drift” proponent, rebuts:
But if this were so “there would have to have been a coincidence of a large increase of just Stage 1 melanoma which had been almost exactly matched in both time and magnitude by a large improvement in therapeutic effect.” He finds this “improbable.”
Another expert weighs in on the side of “real” as follows:
Dr. Julide Tok Celebi, an associate professor of dermatology at Columbia, strongly disagreed with the study’s conclusions. The increase in melanoma is real, she said, and “the only logical explanation is environmental exposure.” She added that these days people were being exposed to “significantly greater” amounts of ultraviolet radiation.
Oh, yeah?
Levell (drift) again:
“Squamous and basal cell carcinomas are no doubt caused by sunlight,” he said, “and those increases are concentrated on the face and neck.” But the diagnoses of melanoma in the registry were mostly on the back, trunk and limbs, areas not consistently exposed to the sun. This means that exposure to sunlight cannot explain the increased number of lesions reported as malignant.
Ladies & Gentlemen of the jury, what say you? Drift or real?
The point, of course, is not which view is right, the point is to watch Critical Thinking in action.
And yes, you should always try this at home.
