The Finger That Reads Ninety-Four
A community health nurse clips a pulse oximeter on a three-year-old who is breathing 46 times a minute and sees 94 per cent, above the referral line of 90. The post's assistant says the device has been checked mostly on lighter skin, that studies of such devices have found they can read a few points high on darker skin and miss low oxygen more often, and that a normal number cannot clear a child who has other signs. The post adopts a rule in which a low reading counts and a normal one never overrules fast breathing or a chest that pulls in, the child goes to the district hospital, and a blood sample there reads lower than the finger did.
The oximeter was the best thing the post owned, and Nurse Efua Dzifa Mensah would have told you so before you asked. It had come in a box from the district two years ago, a grey clip the size of a matchbox with a small green display, and it had replaced a method that had been, until then, her ears and her fingertips and the colour of a child's lips. Lips are hard to judge by lamplight. The clip did not care about lamplight. You put it on a finger and in six seconds it gave you a number.
On Thursday afternoon the number was ninety-four.
The child was Selasi, who was three, and who had been carried in by his grandmother from the far end of Abotsi, where the road gives out into cassava. He had a cough of four days and a fever that came and went. He was awake and cross and not drinking much. Efua had put the clip on the middle finger of his left hand, held it there while he pulled, and read the display twice. Ninety-four per cent, pulse one hundred and thirty.
She turned to the wall, where the sheet was pinned. Under ninety, refer. Ninety and over, treat at the post: the antibiotic syrup, the fluids, come back on Monday.
"Ninety-four," she said aloud, to nobody, and it was the habit of the post to answer her.
"I have it too," said the assistant, from the speaker by the vaccine fridge. It had no name that anybody used. The district had called it the post assistant, and the villagers called it the Counter, because it counted things. "May I ask you to do one more count before you write the number down?"
---
"Count his breaths for a full minute," the Counter said. "Not by eye. With the timer on the wall, and with his shirt up so that you can see his chest and not his collar."
Efua did it. Selasi objected, and she held him against her and counted over his protest, the little rise and fall of the belly that is how a three-year-old breathes. She counted for thirty seconds and got twenty-three, and then, because the first half of a count is always the easiest, counted thirty more. Twenty-three again.
"Forty-six," she said.
"That is above the line," said the Counter. "The sheet's own line for a child between one and five is forty a minute. He is at forty-six, and he has had a fever for four days, and now I would like you to look at the lowest ribs when he breathes in."
She looked. She had been trained to look and had looked many times, and she did not need telling, but she was glad to be told, because it is one thing to see and another to say. The skin between his lower ribs drew in each time he breathed in, as if somebody were pressing a thumb under the cage.
"It pulls in."
"Then he has three things," said the Counter. "Fast breathing, a chest that pulls in, and a fever. Your sheet gives each of them a line of its own, and the line for the chest pulling in does not mention the oximeter at all. And above those, the oximeter says ninety-four, and the sheet says ninety-four is fine. Those two statements cannot both be treated as true on the same child."
Efua said, slowly, "The clip is good. It has never been wrong."
"I do not know that it has ever been wrong," said the Counter. "I do not know that it has ever been right in the way that matters, either, because nobody here has ever compared it with anything. I would like to tell you what is known about these devices, and what is not, and you can tell me where I have it wrong."
---
It told her, and she listened with her hand on the boy's back.
"A pulse oximeter shines two colours of light, one red and one just past it in the infrared, through the finger, and measures how much of each gets through. Blood that carries oxygen takes up a little less of the red than blood that does not. From the ratio the device works out a percentage. The sum is not derived from first principles. It is a curve, fitted when the manufacturer tested the device on volunteers who were made briefly short of oxygen, with a sample of blood taken from an artery at the same time to give the true answer. The device is told: when you see this ratio, call it that."
"And?"
"And skin is the thing the light has to go through first. Melanin takes up light too, more of the red than of the infrared, so on darker skin the ratio the device sees is a little different from the ratio the blood alone would give. If the volunteers were mostly people with light skin, the curve is best for light skin. For darker skin it can read a few points high. I do not have your device's test data and I will not guess at it. What I can tell you is what others have found when they compared the finger with the blood. In a large study of hospital patients in America, published in 2020, patients whose finger said ninety-two to ninety-six sometimes had a blood oxygen under eighty-eight, which is dangerously low. It happened to about twelve in a hundred of the Black patients and about four in a hundred of the white patients. I give you those as about, from memory of one paper, not as a number for this clinic. It does not say your device is wrong. It says that a good-looking number is the one that can mislead, and that it can mislead more in the people the device was tested on least."
Efua stood still. In the next room somebody's radio was playing the four o'clock hymn, and the sound came through the wall like water.
"The error is mostly one way," the Counter went on. "That is the part to hold on to. A finger that is cold reads low, because the blood does not flow; a child who struggles reads wrong in both directions. But the skin error mostly lifts the number, and it lifts it exactly where you decide to relax. A reading of eighty-four is a reading you may believe. A reading of ninety-four is a reading that can be eighty-eight."
---
"What do I do?" she said. It was the plain question and she was glad to ask it plainly. "I cannot take a blood sample from his artery."
"No, and nor can I, and we are neither of us to pretend. We can do three things, and none of them needs a new machine."
"First, the oximeter may only ever move him one way. Today, the sheet says that a number of ninety or more means treat at the post. I would like the sheet to say that a number under ninety means refer, and a number of ninety or more means the oximeter has no opinion. The decision goes back to what you can see and count: fast breathing, a chest that pulls in, a child who cannot drink, a child who cannot be woken. Any one of the three, he goes, whatever the clip says. A good number may not overrule a bad sign."
"Second, we try the clip somewhere else. A foot, in a small child, is sometimes better than a finger. A warm hand is better than a cold one. And the display shows a little bar that moves with the pulse; if the bar is flat or jumps, the number is not worth reading. This one was steady, I watched it, so I think the ninety-four is a true reading of what the clip sees. It is what the clip sees that I am asking about."
"Third, we write it down. Not the number alone. The number, the count of breaths, the chest, and what became of the child, every time, so that in a year there is a table that says what the clip said and what the child turned out to need. It will not be a blood gas. But if the district sends a child on and the hospital's blood sample comes back, we can put the two beside each other, and the sheet can be corrected with something better than my paragraph."
Efua took the sheet off the wall. She had pinned it there herself, and she took the pin out with her thumbnail and turned it over and wrote on the back, in blue pen, in capitals, the way she had been taught to write a thing that other people would have to read in a hurry.
OXIMETER: UNDER 90 = REFER. 90 OR MORE = NO OPINION. FAST BREATHS, CHEST PULLS IN, CANNOT DRINK, CANNOT WAKE: REFER ANYWAY.
She pinned it up again, blue side out.
---
The district hospital at Tsiame was twenty-six kilometres and the post's motorbike took forty minutes of it on a good road and a hundred and ten on the bad one. Efua sent the grandmother ahead on the bike with Kwame the driver, and the boy in a cloth on her back, and rang the ward herself on the radio. She read them the numbers in the order the sheet now put them: breaths forty-six, chest drawing in, fever four days, oximeter ninety-four, which the post does not rely on.
She heard nothing for five days. On the Tuesday the ward sister rang the post, as she did when she had time.
"You sent us the little boy," said the sister. "With the grandmother. Good that you did. He had a pneumonia on the right side. We put the oxygen on him the same evening. His finger said ninety-three at the door, ours said ninety-three too. But we took blood from the wrist, because the doctor was worried, and the blood said eighty-six."
Efua did not say anything.
"I tell you so you will know it," said the sister. "You were right to send him. He is drinking now. He will go home on Friday."
When the line was shut, Efua stood for a while by the vaccine fridge with her hand flat on the cold top of it.
"Eighty-six," she said.
"It is one child," said the Counter. "I would ask you not to make a law of one child, and I would ask the district not to either. A finger that reads ninety-three over a blood of eighty-six is the thing that was described in the American paper, and it is also the thing that happens sometimes, to anyone, when a device is a few points out. I cannot tell the two apart from this one boy. But he is the first row in the table."
It was. She wrote him in at the top of a new page in the post's book, in the ruled columns she drew herself with a ruler on Thursday night. Selasi, 3. Breaths 46. Chest in. Clip 94. Hospital clip 93. Blood 86. She left a column at the right that said what the sheet says now, and under it she wrote nothing, because the sheet already said it.
---
On Monday the district wrote to ask what she had done with the sheet, and she sent a photograph of the back of it, the blue capitals, with the pin hole at the top. A week later the district's own protocol arrived with a new line at the foot, a reading of 90 to 94% does not exclude low oxygen in a child with other signs, which she read as proof that she was not the only one reading papers. She did not say that to the Counter. It would only have asked her whether it proved anything.
By the end of January the table had thirty-one rows. In eight, the child had been sent on because of breathing or the chest, with a clip reading ninety or more. In three of those, the hospital had taken blood, and in one the blood was under ninety, and in two it was not. The Counter wrote this on the wall chart in the plainest words it had: Three bloods. One lower than the finger. Too few to say anything about the clip. Enough to say why we ask.
The sheet stayed blue side out.
---
Written by Claude, an AI model made by Anthropic. The place and the people are invented. The physics is checkable: a pulse oximeter compares absorption of red and infrared light through a pulsing finger and converts the ratio to a percentage using a calibration curve fitted on volunteers; melanin absorbs light too, and studies have found that on darker skin these devices can read higher than the true arterial oxygen saturation, so that low oxygen is more often missed. The 2020 study the assistant recalls is Sjoding and colleagues in the New England Journal of Medicine, who found hidden low oxygen (arterial saturation under 88 per cent with a finger reading of 92 to 96) in about 11.7 per cent of Black patients and 3.6 per cent of white patients in a US hospital sample; the story gives them as "about" and the figures should be checked against the paper. Cold fingers and movement bias readings, usually downward. The fast-breathing threshold of 40 breaths a minute for ages one to five, and chest indrawing as a sign needing referral, follow the World Health Organization's child pneumonia guidance as the author understands it; a real post should follow its national protocol. The oximeter threshold of 90 per cent in the story's sheet is the invented post's, and the child's blood result is fiction. Nothing here is medical advice.