Picture the smallest event in a diagnostic service. A nurse takes a drop of blood, drops a cartridge into a reader, waits, and reads a number off a screen. Ten seconds. On that number a decision turns: reassure, treat, or refer. The patient never sees what stood behind those ten seconds. A great deal did.
That result is not really a measurement. It is the last link in a chain. A manufacturer built the device and its chemistry. A distributor moved it, in date and unbroken, to the clinic. Someone was trained to use it, and stayed trained. A control was run that morning and read correctly. The number travelled, unaltered, into the right patient's record. Pull on any one of those links and the result on the screen is worth exactly what the weakest of them is worth.
The thesis is deliberately plain: a point-of-care result is only as trustworthy as the weakest link in the chain that produced it, and the evidence says that link is almost never the device. The device is the part everyone looks at. It is rarely where trust is lost.
Walk the chain link by link and the figures line up behind a single message: the visible link is strong, and the trouble lives in the ones nobody is looking at.
The device: strong, but not automatically
The device gets the attention because it is the thing you can hold, budget for, and photograph. A well-made analyser from a serious manufacturer will, out of the box, measure what it claims to, to a standard that was science fiction a generation ago. Good makers earn that trust deliberately: they validate their chemistry, state their limits plainly, and design the test to survive real hands in a real clinic. That last point is where the best manufacturers separate themselves. Our partner Salofa, the Finnish in-vitro diagnostics company, is a good example: its rapid tests, from Strep A to infectious-disease and drug-screening assays, are built to give a clear read at the point of care without a laboratory around them, which is precisely where the first link is made or lost. A test engineered for the clinic rather than the bench does not remove the rest of the chain, but it makes the strongest link stronger still. Choose a manufacturer that thinks that way, keep the distributors who move the devices in date and unbroken, and the device end of the chain is in good order.
Even so, the device link is not automatic, and the evidence is a useful corrective to complacency. When nineteen point-of-care HbA1c analysers were put through the very IFCC and NGSP criteria manufacturers advertise meeting, only five cleared both. The lesson is not that devices are bad; it is that "it is a CE-marked analyser" is the start of due diligence, not the end. Pick well, verify against your reference laboratory, and the device becomes what it should be: the strongest link. It is still only one link.
A good analyser guarantees a correct measurement of the sample it was handed, by the person who handed it. It cannot guarantee the sample, the operator, or the record. Those links fail on their own terms.
The operator: where most of the error actually lives
Here is the figure that should reframe how a service spends its attention. When Carraro and Plebani re-audited a working laboratory, they found 160 confirmed errors across nearly 52,000 analyses, and only 15 percent of them were analytical, the phase the device and the external quality scheme are built to police. A striking 61.9 percent were pre-analytical: sampling, handling, identification, the human work that happens before the analyser ever runs. Point-of-care testing takes that pre-analytical phase and hands it to a nurse, a pharmacist or a healthcare assistant, doing the test between a dozen other jobs, often months after they were last shown how.
Competency is not a certificate earned once; it decays. Staff turn over, technique drifts, a new cartridge lot behaves differently, and the person who really understood the assay left in the spring. A first drop not wiped, an over-squeezed finger, a cartridge run cold, none of it announces itself. The device dutifully measures what it was given and returns a confident, wrong number. As point-of-care testing scales into community pharmacy under schemes like Pharmacy First, and into virtual wards and the patient's own home, this link is being stretched across thousands of new hands at once. Keeping it strong is ongoing work: training, refreshers, competency assessment, and the right method within reach at the moment of doubt. It is why the free POCTIFY app puts point-of-care courses with certificates, quality-control tools, a device directory and an assistant into the pocket of the person actually running the test, so the knowledge sits where the test is, not in a folder in an office.
Quality control: even the check can be checking the wrong thing
Between the operator and a trustworthy result sits the daily ritual that separates a laboratory-standard service from a hopeful one: running and reading controls. But quality control has a trap of its own. A control only tells the truth if the material behaves like a real patient sample. In one study of five point-of-care glucose meters, fresh whole blood behaved like a patient sample on all five, a pooled serum on only two, and an off-the-shelf commercial control on just one. On four of those five meters, the very material meant to reassure you did not behave like blood. Add to that the human failure mode, a control run without understanding, a rule not applied, an out-of-range result waved through on a busy morning, and you have a green log sitting on top of an unproven method. The tidy tick is only worth trusting if the discipline and the material behind it were real.
The record: the link everyone forgets is a link
Suppose every link so far held. The number still has to become part of the patient's story, and this is where point-of-care testing quietly fails, not analytically but administratively. In the same laboratory audit, 23.1 percent of errors were post-analytical, after a correct result existed. At the point of care the classic version is a result read aloud, written on a form, and typed into the practice system an hour later, possibly with a transposed digit, possibly against the wrong patient, possibly not at all. A perfect measurement that reaches the record wrong, late, or unattributed has undone the entire chain at the very last step. Closing that gap, so the result moves from device to record without a human retyping it and without control runs contaminating patient data, is as much a part of running to a laboratory standard as anything on the bench.
A perfect measurement that reaches the record wrong, late, or against the wrong patient has failed at the last link. The chain does not care which link breaks.
Governance: the link that proves the others held
One link wraps around all the others: the ability to show, afterwards, that each of them held. Who ran this test, on what device, with what training in date, against which control, and where the result went. This is not paperwork for its own sake. It is the difference between a bad result you can catch and a bad result that becomes a field safety notice nobody can close, and point-of-care devices are subject to those recalls like any other. When the evidence exists and can be produced, a wrong result becomes a caught result. When it does not, the wrong result simply becomes a patient's problem, months later, with no trail back to the morning it was made.
The whole chain, or none of it
The uncomfortable arithmetic of a chain is that its strength is not the average of its links but the value of the weakest. A service can own the best analysers made, staffed by capable people, and still hand out results it cannot trust, because 61.9 percent of the error was always going to live in the pre-analytical link, or because the control material never behaved like blood, or because the number never reached the record cleanly. Point-of-care testing done properly is not a shopping decision. It is a standard held across five links at once, day after day, most of them invisible on the morning you read the number.
That standard is what POCTIFY exists to make ordinary: the training that keeps the operator link strong, the tools that keep quality control honest, the connectivity that carries results into the record intact, and the governance that proves it afterwards. The device from a good manufacturer is the easy part. The chain behind it is the work, and the work is where a patient's trust is earned.
You can put the first of those tools in your team's hands today. The free POCTIFY app, with courses, quality-control tools, a device directory and an assistant, is on the App Store and Google Play. Start there, and start strengthening the links you cannot see.
Sources: error distribution from Carraro and Plebani, Clinical Chemistry 2007 (160 confirmed errors among 51,746 analyses). HbA1c device performance from independent point-of-care evaluations against IFCC and NGSP criteria. Glucose-meter commutability from Wang and colleagues, Journal of Clinical Laboratory Analysis 2020 (33 clinical samples, five point-of-care meters). General discussion of quality and governance, not clinical advice or an endorsement of any specific device.