The salon off Route 9 in Framingham has two chairs and a window unit Marisol Vega runs into October. I was there for my hands, which after nineteen years of adjusting claims in other people's fields have not held up well. She turned the left one over, then the right, and took longer about it than the work required.
"The hands tell on you before you do," she said.
She has been at that address twenty-two years. She was careful, twice, to say she is not diagnosing anybody. She also said she has sent four clients to a doctor in that time and two of them thanked her afterward.
This one is a wrong arrow. It is also, more than most rules in this column, substantially correct about what it sees.
What would have to be true
For the rule to work as Marisol's clients hear it, the hand would have to carry information about what is coming.
Start with what it does carry, because the list is long and it is clinical. Clubbing of the fingertips shows up with pulmonary malignancy, COPD, cirrhosis, inflammatory bowel disease and congenital heart disease. Koilonychia, the spoon-shaped nail, tracks iron deficiency anemia and hemochromatosis. Splinter hemorrhages under the nail are a classic finding in endocarditis. Terry's nails, where the plate goes white and the tip keeps a narrow dark band, were first described in severe liver disease, and about 80 percent of cirrhotic patients have them. Beau's lines, the transverse grooves, come from any illness severe enough to stop the nail growing for a while.1
That is a real differential built out of one hand. And Marisol sees something almost no physician sees, which is a large number of adult nails with the polish off, repeatedly, over years. A nail salon is one of the few places where a hand gets looked at carefully by someone who has a baseline for that particular hand.
So the steelman is not weak. The observation is good. Hold onto that, because everything after this is about direction and none of it is about whether she can see.
What the hand is actually measuring
Here is the sentence.
The hand is two recording instruments running at different speeds. The palm creases are a fixed record of about three weeks in the first trimester. The nail is a rolling tape roughly six months long. Neither one is pointed forward.
Take the creases first. Palmar and digital flexion creases appear between weeks 10 and 13 of gestation. The thenar crease shows up near week 10 with the opposition of the thumb, and the proximal and distal palmar creases become distinct over the next two weeks.2 After that they are set. The lines on the palm of a sixty-year-old are the lines that were laid down before that person had a functioning diaphragm.
I believed for a long time that the creases were worn in by the fetus closing its hand, the way a work glove takes a fold where it bends. That is wrong. The creases appear before the digital movement phase begins, and the formation looks independent of the flexion movements themselves. I had it from a midwife and I repeated it for years without checking it.
Now the nail. A fingernail grows about 3 mm a month, which is roughly 1 mm every six to 10 days, and the whole plate turns over in three to six months. That growth rate is the useful part. A Beau's line can be dated by measuring from the proximal nail fold to the groove, which gives you the approximate week the growth arrest happened.1
Read that again with the arrow in mind. The nail is not telling you that an illness is coming. It is telling you, to within about a week, when one already happened.
Against what
The creases fail on base rate, and they fail badly.
A single transverse palmar crease, the one people call the simian crease, occurs in roughly one person in 30, and males are about twice as likely to have it as females. It appears more often in Down syndrome, fetal alcohol syndrome, Trisomy 13 and several other conditions. It is also, on its own, frequently just normal.3
Work that through. If a feature shows up in about three percent of everybody and you use it as a signal, almost every hand you flag belongs to someone with nothing wrong. The crease is real, its association is real, and as a test applied to one person in front of you it does not survive the arithmetic. It earns its keep inside a full clinical picture, alongside other findings, which is exactly how the clinical sources use it and exactly not how it gets used on a palm.
The nail does better because it is not being asked to predict anything. A dated Beau's line scored against the patient's own history usually finds the illness, because the illness is already in the history. That is a strong instrument aimed at the past.
Marisol's version of the rule is closer to the nail than to the creases. She is reading a tape of the last six months and calling it a warning. It is a warning, in the sense that a thing already underway can still be caught. It is not a forecast, and nothing on the palm is.
The same arrow, better funded
The live version of this error is the polygenic risk score.
A polygenic risk score adds up the small effects of many common variants into one number for a disease. The variants are fixed at conception. Like the palm creases, the record was written early and does not change, and like the palm creases it gets read as news about what is coming.
It has been scored against a baseline, and the scoring is published. Hingorani and colleagues went through the Polygenic Score Catalog and reported a median area under the curve of 0.65 across the scores they examined, and a median detection rate of 11 percent at a 5 percent false positive rate. For coronary artery disease the detection rate was 12 percent; for breast cancer, 10 percent. Their conclusion was that the scores "performed poorly in population screening, individual risk prediction, and population risk stratification," and that strong claims about their effect on healthcare "seem to be disproportionate to their performance."4
Eleven percent detection at a 5 percent false positive rate is a number Marisol would recognize. It is the simian crease with a better funding round.
The structure is identical in both halves. A genuine biological record, honestly observed, laid down at a fixed point and stable afterward. The information in it is real. The direction it points is backward, toward the conditions that wrote it, and the value gets claimed forward, toward a person's odds.
What the hand knows, it knew already. The calendar on it runs one way.
Editor's note. Category: wrong arrow. Filed to the Street desk. Marisol Vega is a composite and the salon is not a specific business; the account of how the rule was collected is written as such. The clinical findings, the crease developmental timeline, the nail growth rate and dating method, the single-crease prevalence and the polygenic risk score performance figures are real and sourced below. No source cited here measured whether nail salon visits function as cancer screening, and this piece does not claim they do.