When the patient is the operator

Somewhere in the country this morning, a woman on warfarin pricked her own finger at her kitchen table, dropped a bead of blood onto a strip, read a number off a small screen, and adjusted her dose. She has done this for years. Her anticoagulation service selected her, trained her, signed her off as competent, and reviews her results. If her number drifts, someone who understands it will notice. This is one of the oldest and safest examples of a patient acting as their own laboratory operator, and it works because a whole quiet apparatus of governance sits behind it.

Now picture two more scenes from the same morning. A man discharged onto a virtual ward the previous evening straps on a pulse oximeter, taps through an app, and uploads a set of readings before his tea goes cold. A few streets away, a woman who bought a bowel cancer test online reads two faint lines, decides they are probably nothing, and puts the box in the recycling. Both people are now operators of diagnostic tests. Neither has been trained the way our profession trains an operator. Neither is running quality control. And in at least one of those cases, nobody is guaranteed to see the result.

My argument is this. Our governance frameworks for point of care testing quietly assume a trained professional is holding the device. Competency assessment, quality control, adherence to the instructions for use, escalation of abnormal results: every one of those pillars was designed around a member of staff. Home diagnostics, hospital at home and direct to consumer testing move the operator role onto the patient, and they are scaling faster than our governance unless we deliberately redesign it. Done well, this is the biggest opportunity in decentralised testing for a generation. Done carelessly, it is a slow leak of clinical risk into people's homes.

The model that already works, and why

Start with what goes right, because it tells us what safety actually requires. Patient self-testing and self-management of the international normalised ratio has a long track record, and the evidence is not soft. The Cochrane review that pooled 28 randomised trials, nearly nine thousand patients, found that self-testing cut thromboembolic events to a relative risk of 0.58 against usual care. Where patients also self-adjusted their dose, so-called self-management, clots fell further to 0.47, and death from any cause to 0.55. Major bleeding did not rise. This is about as good as decentralised testing gets.

Look closely at that chart and it argues against the obvious reading of itself. Bare self-monitoring reduced clots but did not move mortality; it was self-management, the version with more structure, dose algorithms, education and clinical backup, that was associated with the survival benefit. The device is much the same kind in both arms. What differs is the pathway around it. INR self-testing did not succeed because the meter is foolproof or because patients are unusually diligent. It succeeded because someone selects suitable patients, trains and signs them off, reviews the numbers and intervenes when they wander. The patient holds the device; the clinical service still holds the accountability. The operator changed. The governance did not disappear, it was rebuilt around the home.

That distinction is the entire argument of this piece. The hazard is not that patients hold devices. The hazard is assuming that because a device is now portable, cheap and easy, the governance can be thin.

The accountability chain, and where it snaps

In the laboratory, and in well run hospital point of care programmes, accountability is unbroken. A named committee owns the service. A coordinator owns competency and quality. Manufacturers, procurement, clinical leads and operators each hold a defined link. When something fails, you can trace it back along the chain to a person and a process. Home and remote testing stretches that chain across the front door, and at exactly the point where it crosses the threshold, the traditional owner of quality disappears.

Take each pillar of point of care governance and ask a simple question: who owns it when the patient is the operator?

• Competency. In a hospital we assess operators, record it, and re-verify it. Who trains the patient at home, who confirms they can actually do it, and who notices when their technique drifts a year later?
• Quality control. A ward glucose meter runs internal QC and sits in an EQA scheme. A self-test kit typically has neither. No daily control, no Levey-Jennings chart, no external comparison. The only quality signal is the result itself, and by then it has already been believed.
• Adherence to the instructions for use. Professionals are held to the manufacturer's method: sample volume, timing, storage, expiry. At home, a strip may have sat on a sunny windowsill, the reading taken at ninety seconds instead of thirty, the finger squeezed too hard. The device performs to specification; the test does not.
• Result escalation. On a ward, an abnormal result triggers a response by design. At home, it triggers whatever the patient decides to do with it, which may be nothing, or panic, or a search engine.

None of these pillars is impossible to rebuild for the home. But none rebuilds itself. Left to default, the patient-operator node in that chain is a link with no keeper.

The hazard is not that patients hold devices. The hazard is assuming that because a device is now portable and easy, the governance can be thin.

The new wave, and how it stretches the model

The reason this matters now is scale and speed. Virtual wards and hospital at home have gone from pilot to core delivery: by December 2024 England was running around 12,700 virtual ward beds at roughly three-quarters occupancy, and NHS England's stated ambition is 40 to 50 beds per 100,000 people, some 24,000 in total. Remote monitoring of long term conditions is following the same curve. And a large, growing direct to consumer market sells finger-prick and swab tests straight to the public, outside any clinical relationship at all. Each of these stretches the INR model in a different direction, and the governance answers differ.

Hospital at home and virtual wards

Here the patient is inside a clinical pathway, which is the good news. There is a responsible team, a record, and an escalation route in principle. The stretch is operational: a team may be watching dozens of patients using devices they never physically handed over, training delivered in a rushed discharge, readings arriving through consumer-grade kit.

The pulse oximeter is a sharp illustration of how a device can meet its specification and still mislead at the kitchen table. When researchers tested five of the oximeters the NHS distributed under COVID Oximetry @home against arterial blood in more than 900 patients, the devices read on average 0.6 to 1.5 percentage points higher in people with the darkest skin than the lightest. That sounds trivial until you look at the consequence: for detecting genuinely low oxygen, the false-negative rate was 2.3 to 7.1 times higher in darker skin. The device met its specification, yet it performed measurably differently across skin tones, so this is both a measurement problem, a real bias in what the box reports, and a governance problem about who interprets the number, in what context, with what safety-netting. The MHRA is unusually direct about the answer. It does not recommend that the public use oximeters at home unless a clinician has advised it, has shown them how to take an accurate reading, and is receiving the results for review. That is the wrap-around, spelt out by the regulator: trained, connected, reviewed. It is precisely what a bare device dropped at a front door does not have.

Remote monitoring of long term conditions

This is the closest cousin to INR self-management, and the place the model most naturally extends. The ingredients for safety are known: selection, training, connected results, review. The risk is dilution. As programmes scale from hundreds of motivated, well-supported patients to tens of thousands, the temptation is to keep the device and drop the wrap-around, because the wrap-around is the expensive part. As the INR evidence showed, that is precisely the part that made the original safe.

Direct to consumer testing

This is the sharpest edge, because there may be no clinical relationship at all. In 2023 a team at the University of Birmingham did something refreshingly simple: they bought the self-tests on sale within ten miles of their campus, in supermarkets, pharmacies and wellbeing shops, and read what was actually on the box. Of 30 kits from 14 manufacturers, spanning everything from cholesterol and HbA1c to menopause, anaemia and bowel cancer, 24 claimed high accuracy. Only 12 had any accessible evidence to support that claim. And 18, three in five, carried at least one high-risk concern about safety or usability, most often unclear instructions or how to read and act on the result. The authors' conclusion was blunt: improved regulatory oversight is urgently needed to protect the public.

It is worth separating the well-governed home test from the free-for-all, because they get lumped together and should not be. The NHS bowel screening programme is also home sampling: a person collects a tiny stool sample at their own kitchen table. But the loop is closed by design. The sample goes to a laboratory, the result is measured against a defined threshold, being lowered from 120 to 80 micrograms of haemoglobin per gram of faeces to catch more cancers, and an abnormal result generates a colonoscopy referral inside a quality-assured pathway. The shop-bought bowel test read as two faint lines and dropped in the recycling is the same body fluid with none of that around it. Same sample type, opposite governance. Regulation of the device is real, and it matters: a UK self-test IVD must be designed and validated for a lay user, with evidence that a non-professional can understand and correctly interpret the result. But regulation of the device is not governance of the pathway. A compliant kit can still be used at the wrong time, read incorrectly, or produce a result that reaches no clinician and changes no decision, or changes the wrong one.

The connective tissue that keeps home testing safe

If the operator has moved, the governance has to move with them. It does not vanish; it changes shape. The useful way to picture it is a loop, because a one-way test is where the danger lives.

Nearly every safe example of patient-as-operator I can think of closes that loop, and the worrying ones tend to leave it open. The single most important design decision, more important than the device, is whether an abnormal result is guaranteed to reach a clinician and be acted upon, or merely able to. Those are not the same thing, and the gap between them is where harm tends to accumulate. Closing the loop has practical requirements. The result needs to be connected, not stranded on a screen the patient may misread and forget. Escalation needs to be defined in advance, with a named recipient and a timescale, not improvised after a frightening number appears. And the patient needs feedback, so that testing feels like part of their care rather than homework they perform into a void.

Design around the pathway, not the device

The mistake I see coming, and in places already here, is to govern the device and forget the pathway. Vendors will tell you a home device is accurate, and it may well be. Accuracy of the analyser was never the hard part, as the oximeter and the self-test shelf both show in their different ways. The hard part is everything the laboratory used to do around the box, and that responsibility has not been abolished by moving the test into a home. It has been transferred, mostly to people who have not been told they now hold it.

So the governing question for any home or remote testing programme is not "is the device good?" It is "who owns quality now, and can they prove the loop closes?" If a service cannot answer that in a sentence, it does not have a testing programme, it has a testing product, and the difference is measured in patient safety. This reframing also protects the opportunity. Supported self-testing is genuinely good medicine, as the INR numbers make plain. It gives patients agency, catches problems earlier, and takes pressure off stretched services. None of the caution above is an argument against home testing. It is an argument for doing it the way the anticoagulation services quietly proved works: train the operator, connect the result, define the escalation, keep the clinical ownership.

What this means for your service

If you commission, run or advise on any programme that puts a testing device into a patient's hands, treat the patient as an operator you are responsible for, and build accordingly.

• Name the owner of quality. Before the first kit ships, decide who owns competency, result review and escalation for patient-operators. If the answer is "the patient", you have not designed a service.
• Train and assess, do not just instruct. A leaflet is not competency. Build in a supported first use, a way to confirm the patient can perform the test, and a light-touch way to re-check technique over time.
• Replace missing quality control with something. Home kits rarely run QC. Compensate with device selection validated for lay use, periodic comparison against a laboratory or professional test where the stakes justify it, and storage and expiry guidance patients can actually follow.
• Engineer the escalation. Decide in advance what happens to an abnormal result, who receives it, in what timescale, and how the patient is told. Guaranteed, not possible.
• Interpret in context, not in isolation. The oximeter lesson generalises: teach patients and reviewers to watch trends and clinical picture, not single readings, and design the pathway so a known device limitation cannot quietly become a missed deterioration.
• Write it into your risk assessment. Where home or remote testing sits under an accredited service's quality system, ISO 15189:2022 treats it as within that system's scope, so testing outside the laboratory's walls is still the laboratory's business. Assessors increasingly expect to see such testing governed, not ignored, and the same principle is worth applying even where formal accreditation does not yet reach.

Our consultancy helps services design this governance around the pathway before a programme scales, and our training, including the POCT fundamentals course, builds the quality thinking that transfers directly to the home setting. If you want to understand what a single well-governed analyte pathway looks like, the long track record of INR self-testing is the case study worth studying, and the wider analyte library shows how differently the stakes sit across tests.

The link with no keeper

Home testing is coming whether our governance is ready or not, and mostly it should come, because supported self-testing at its best is patient care done well. But the woman at her kitchen table is safe not because she owns a good device. She is safe because a service still owns her. The task in front of us is to make sure that as the operator role moves through millions of front doors, the ownership of quality goes with it, and does not quietly stay behind in a building the patient will never visit again.

Sources and notes

This article draws on peer-reviewed evidence, published NHS data and UK regulatory guidance. The anticoagulation figures in Figure 1 are pooled relative risks from a Cochrane systematic review. The direct to consumer figures in Figure 3 are counts from a 2025 BMJ cross-sectional review of self-tests bought in UK shops in 2023. The pulse oximeter figures are from a diagnostic accuracy study of the devices used in the NHS COVID Oximetry @home scheme. Virtual ward capacity and occupancy are from NHS England's monthly statistics for December 2024 and are rounded; readers needing exact figures should consult the current NHS England release, as these update monthly.

1. Heneghan CJ and colleagues. Self-monitoring and self-management of oral anticoagulation. Cochrane Database of Systematic Reviews, 2016 (28 trials, 8,950 participants; thromboembolic events RR 0.58, self-management RR 0.47, all-cause mortality on self-management RR 0.55, major haemorrhage RR 0.95).
2. Davenport C and colleagues. Direct-to-consumer self-tests sold in the UK in 2023: cross sectional review of regulation and evidence of performance. BMJ, 2025 (30 tests from 14 manufacturers; 18 of 30 with at least one high-risk concern; 12 of 30 with accessible performance evidence).
3. BMJ Group. Many high street health tests are unfit-for-purpose and need greater regulation, warn experts. Press summary of the two Birmingham self-test reviews, 2025.
4. Martin DS and colleagues. The impact of skin tone on performance of pulse oximeters used by NHS England COVID Oximetry @home scheme. BMJ, 2026, measurement and diagnostic accuracy study (five devices, 903 patients; SpO2 read 0.6 to 1.5 points higher in the darkest versus lightest skin; false-negative rate ratio 2.3 to 7.1 for detecting low oxygen).
5. MHRA. The use and regulation of pulse oximeters. GOV.UK guidance, on home use only under clinical advice, with instruction and results review.
6. NHS England. Virtual wards, and the monthly virtual ward capacity and occupancy statistics. Programme scale and the 40 to 50 beds per 100,000 ambition.
7. NHS England. NHS to detect and prevent thousands more bowel cancers with more sensitive screening. On the FIT threshold moving from 120 to 80 micrograms of haemoglobin per gram.
8. MHRA. Common specification requirements for in vitro diagnostic devices. On the requirement that self-test results be understood and correctly interpreted by a lay user.