Nobody likes fuzziness in data.
In our guts, a number is a number, and we should be able to trust it. After all, if we have a set of numbers and a problem, there is a right answer and a wrong answer, correct?
Unfortunately, in real life this is not true. Every measurement has an uncertainty. Every test has an error rate. To think clearly in medicine, we must identify and embrace the probability cloud around every number and work with it.
But we turn away from this imperative. Maybe it’s because the ideas of uncertainty and error in measurement are introduced late in the educational process, long after our mathematical instincts are formed. MIT thinks that’s a problem, especially in the time of Covid-19.
But a good medical intern knows that if the blood potassium level is ridiculously high, it should be repeated. And a good hospital nurse knows that when the blood pressure readout looks wrong, they should recheck it by hand.
The same problem occurs in home blood pressure monitoring. There’s an error rate. But that doesn’t mean home BP monitoring isn’t useful, it is. We just have to assess and think about the data we receive.
Our working estimate at Brook is that in about 15 percent of patients, there is a nontrivial error in home blood pressure readings compared to manual blood pressure readings. This is mostly due to a combination of arm size and blood vessel elasticity. To get slightly technical for a minute, home BP cuffs measure the pressure waveform inside the arm as the cuff is inflated. An algorithm then uses changes in that pressure curve to estimate systolic and diastolic blood pressure values. Those algorithms are proprietary to each manufacturer – it’s a black box, from our point of view.
But this brings up systemic issues. “Ambulatory” blood pressure monitoring is the gold standard for the diagnosing blood pressure. This is when a patient wears a device continuously for a few days, and has their blood pressure automatically checked every 15 minutes. These monitors analyze either the pressure waveform, as with home machines, or the Korotkoff sounds that a clinician listens to. Both methods have errors. So what is the clinical significance of the intrinsic error of our gold standard?
At any rate, practically speaking, when blood pressure readings “don’t look right,” the machine should be checked for accuracy by a clinician. This is best practice.
As everywhere in medicine, careful evaluation and decision-making are required. There is no substitute for thinking.
Brook’s recommendations on home BP testing are here.
on March 3, 2022.
Mike is the Chief Medical Officer at Brook. Formerly CMO of United Memorial Medical Center at Rochester Regional Health, Medical Director for Innovation and Science at Independent Health, and a partner in Endion Hospitalists. He has degrees from Dartmouth College, Columbia University, the University of Liverpool, and SUNY Buffalo, where he is currently Associate Clinical Professor of Medicine.