Fingertip Pulse Oximeter

Details

A fingertip oximeter that estimates blood oxygen saturation and pulse rate, with the things that make it wrong printed on the same page as the things that make it right. Every oximeter on the shelf says plus or minus 2 percent. Almost none say what that number is. It is Arms, a root mean square error measured against arterial blood samples across a controlled desaturation study. It is a statistic, not a limit. If a device is 2 percent Arms then roughly a third of its readings are more than 2 percent from the truth and about one in twenty is more than 4 percent out. A reading of 94 on a good oximeter is compatible with a true value of 90 to 98. The second thing is skin. Pulse oximeters read light through tissue and melanin absorbs light, so they tend to overestimate saturation on darker skin, most at low saturations, which is exactly where an error costs the most. This is documented and it is almost never printed. Our validation panel was recruited across a measured range of skin pigmentation and the bias by group is in the report: on the most pigmented group the device read on average 1.4 percent high at true saturations between 70 and 80 percent, against 0.4 percent on the least pigmented. That gap does not disappear because a pack does not mention it. And cold hands, nail polish and mehndi all defeat it, which in India is worth saying out loud.

• SpO2 accuracy plus or minus 2 percent Arms from 70 to 100 percent. Below 70 percent the accuracy is not specified, by anyone, and the number should not be relied on
• Pulse rate 30 to 250 beats per minute, plus or minus 2 beats or 2 percent, whichever is larger. Resolution 1 percent and 1 beat
• Reading settles in about 8 seconds, with a perfusion index shown so you can see whether the signal was worth trusting
• Skin pigmentation bias measured and published by group, not asserted as tested on all skin tones
• Four-way rotating OLED with a plethysmograph waveform. The waveform is not decoration: a ragged trace means the number under it is unreliable
• Two AAA cells, user replaceable, about 30 hours. Fits fingers 8 to 25 mm, so it suits children above about four
• It measures. It is not a diagnostic instrument, it is not a screening test, and it cannot tell you why a number is what it is
• Cold fingers, nail polish, gel nails, mehndi, shivering, movement and bright overhead light all degrade it. Warm the hand, use a bare finger, sit still, and shade the sensor