A single-lead ECG costs between nothing and about £100, and it answers one question well. The difficulty is that people buy them to answer a different question — usually "is this chest pain my heart?" — which is the one thing they cannot do.
1. What the different devices are
| Type | How it works | Worth knowing |
|---|---|---|
| Handheld pad around £80–100 | Two thumbs on two electrodes for 30 seconds, recording roughly the equivalent of lead I | The best traces of the consumer options, because the contact is deliberate and the hands are still. Some versions take a knee or ankle contact to give six leads |
| Smartwatch often already owned | Worn on one wrist, finger of the other hand on the crown or bezel to complete the circuit | The most convenient and the one most likely to be used at the moment symptoms occur, which matters more than trace quality for an intermittent arrhythmia |
| Combined blood pressure and ECG monitor around £100–130 | An upper-arm cuff with ECG electrodes, giving both from one sitting | The most useful single purchase for someone monitoring both, because it enforces the seated, still, arm-supported position that blood pressure needs anyway — and that position also produces a cleaner trace |
| Chest-strap and patch recorders | Continuous recording over hours or days | Closer to what a clinician would order. Useful where symptoms are frequent but never present at the moment you think to record |
2. How good are they, actually
Better than most people expect at the thing they are for, and with one caveat that is usually left out of the marketing.
The BASEL Wearable Study compared five consumer devices against a simultaneous physician-interpreted 12-lead ECG in 163 patients, a third of whom were in AF at the time. Where the device gave an answer, it was a good answer: sensitivity for AF ran from 84% to 96% and specificity from 91% to 98%, with over 92% of tracings correctly classified across every device.
The caveat: about a fifth to a quarter of recordings produce no answer at all
- Inconclusive rates ran from 17% to 26% depending on the device. Only 79% of nearly a thousand recordings returned an automated diagnosis. The published sensitivity figures are calculated after the inconclusive ones are set aside — which is legitimate, and is not how they are usually quoted.
- A clinician looking at the same trace could determine the rhythm in about 99% of cases. So an inconclusive result is usually a limitation of the algorithm rather than of the recording, and it is worth keeping and showing rather than deleting and retrying.
- False positives happen. In that study, 8% of patients who were in sinus rhythm were labelled as AF by at least one device. A single "AF detected" from a consumer device is a reason to see someone, not a diagnosis.
- Repeat once, not repeatedly. In a separate cardioversion study, repeating an unclassified recording once reduced the unclassified rate substantially. A third attempt reduced accuracy and started producing false positives. If two good recordings will not classify, that is the answer — take it to a clinician.
And all of it is adult data. These algorithms are built and validated on adults, mostly older ones — mean ages in the validation studies sit in the mid-sixties, because that is where AF is. Applied to a teenager, the output is a general guide and not a paediatric standard, which is why the school chapter treats a device reading as a line for the handover rather than a finding.
3. Getting a trace worth reading
Sit down, back supported, and rest your arms
Muscle activity is the main source of artefact. A trace taken standing, or with the arms unsupported, is the commonest reason for an inconclusive result.
Still, warm, and quiet for the whole recording
Do not talk, do not shiver, and do not hold your breath either — breathe normally. Cold dry hands make poor contact; warming them or slightly moistening the fingertips helps.
Firm, steady contact — and no more
Pressing harder does not improve the signal and introduces tremor. Rest the finger; do not push.
Record when the symptom is happening
This is the whole value of owning one. A perfect trace between episodes tells you very little about an intermittent arrhythmia; a rough trace during the palpitation may be diagnostic.
Keep everything, including the inconclusive ones
Export the PDF rather than screenshotting the phone. A clinician can usually read a trace the algorithm could not, and a series of recordings around symptoms is worth far more than the best single one.
4. What the traces look like
Schematic illustrations rather than real recordings, for pattern recognition. The value of knowing these is being able to describe what you are seeing — not to make a diagnosis from it.
What these are drawn to, and what they are not. The proportions within each beat are physiological — P wave about 95 ms, PR interval about 175 ms, QRS about 90 ms, and a QT that shortens as the rate rises, which is what real hearts do. But the strips are compressed horizontally to fit several beats on a screen, so they are not at clinical paper speed and you cannot measure an interval off them with a ruler. Note in particular that in the fast and slow traces the beat itself is the same width and only the gap between beats changes — a slow heart does not depolarise slowly, and a diagram that stretches the whole beat teaches the wrong thing.
Normal sinus rhythm the baseline
P wave: atria contracting. QRS spike: ventricles contracting — the beat the device counts. T wave: the heart resetting.
Evenly spaced beats, a P wave before every QRS, and a consistent shape beat to beat.
Atrial fibrillation irregularly irregular
No clear P waves and a fine wavy baseline. The R–R gaps vary randomly rather than in a pattern.
The one pattern these devices are actually validated to catch. It raises stroke risk — flag it to a clinician.
Tachycardia fast but regular
Evenly spaced, close together. Normal shape, abnormal speed.
A device can flag “fast”. It cannot reliably tell you which kind of fast, and that distinction is the one that matters clinically.
Bradycardia slow but regular
Evenly spaced, far apart.
Can be entirely normal in a fit teenager or athlete at rest. Concerning alongside fainting, dizziness, confusion or a weak pulse.
Premature beat (ectopic) one odd early beat
One wider or taller beat arriving early, with no P wave in front of it, then a pause before the rhythm resumes.
Occasional single ones are usually harmless. Frequent or symptomatic ones need follow-up.
5. What these devices cannot do, and it is the important section
| Question | Answer |
|---|---|
| Can it rule out a heart attack? | No, and this is the one that hurts people. A heart attack is diagnosed on the pattern across twelve leads plus blood tests, and much of the electrical change happens in territory a single lead across the chest simply does not see. A normal consumer trace during chest pain means nothing at all. Chest pain is 999, every time, regardless of what any device says. |
| Can it tell me which kind of fast rhythm this is? | Not reliably. It can tell you "fast and regular" or "fast and irregular", which is genuinely useful information, but the distinction between the fast rhythms that matter needs more leads |
| Can it detect a problem when I have no symptoms? | Sometimes — that is exactly how silent AF gets found, and it is the strongest argument for owning one if you are older or have had a stroke. But screening a young person with no risk factors mostly generates false positives and anxiety |
| Can it replace a 12-lead ECG or a monitor my GP orders? | No. It can, however, capture something they would otherwise miss, which is a different and real contribution |
| Can I use it on a child? | Read the output as indicative only. The algorithms are not validated in children |
| What if it says "AF detected"? | Save the trace and see a GP within days — not an emergency by itself, and not something to ignore either. If it comes with chest pain, breathlessness, fainting or stroke symptoms, that is 999 |