Healthcare 9 min read

Panic buttons and fall detection: the honest version.

A button is only as good as the person at the other end of it. Here is what actually works for someone living alone, what a retirement village needs instead, and what fall detection can and cannot do.

By Frank Guo · Technology & Product Leadership, addanode

TL;DR — These are two different products and the search term hides it. A person living independently at home needs a personal alarm with a monitored response — the device is the cheap part, the answering service is the whole product, and without one the button rings in an empty house. A hospital, retirement village or frail care facility needs something else entirely: a facility system where the alarm reaches on-duty staff with the caller's location, escalates when nobody acknowledges, and leaves a record of how long it took. On fall detection, be sceptical of every datasheet: automatic detection catches many falls but not all, and raises alarms for things that were not falls. It is a useful safety net underneath a button, never a replacement for one.

First, which problem are you solving?

South Africans searching for this land in one of two situations, and the right answer is genuinely different.

A parent living alone at home A facility with staff on duty
Who answersA monitoring centre, or family on a call listOn-duty staff, by name and shift
What "location" meansA home address already on fileWhich room, corridor or garden — right now
What you are buyingA monitored service with a monthly feeInfrastructure plus an escalation process
Proof requiredPeace of mind for a familyA record for regulators, insurers and families
Who we helpNot us — see the buying guidance belowThis is what we build

If you are buying for a relative at home, we are not your supplier and it would be unhelpful to pretend otherwise. What you want is a monitored personal alarm from a provider who answers the call, and the questions worth asking are in the next section. If you run a village, a frail care unit or a hospital, skip to the section after it.

The button is one link in five A personal alarm is a chain, and the device is the cheapest link in it. Whether it is worn at all, who answers, whether answering turns into a response, and how anybody gets through the door are the four links that decide the outcome — and they are the ones the marketing says least about. It is being worn The largest single cause of failure, and not a technical one pressed The button is The cheap part, and most of the marketing answers Somebody A monitored centre, or a name on a call list responds Somebody Answering is not responding arrives Somebody Getting to the door is what families discover too late Four of the five links are the response. The response is the thing actually being bought.
The button is one link in five Worn, answered, responded to, arrived at — the device is the cheap link. It is being worn The largest single cause of failure, and not a technical one The button is pressed The cheap part, and most of the marketing Somebody answers A monitored centre, or a name on a call list Somebody responds Answering is not responding Somebody arrives Getting to the door is what families discover too late The response is what is actually being bought.
The chain the six questions below are really about. The device is the cheapest link and the only one the marketing describes — whether it is worn at all is the largest single cause of failure, and it is not a technical problem.

Buying a personal alarm for someone at home: six questions

Most of the marketing is about the pendant. Almost none of the value is.

  • Who physically answers, and in what time? A device that phones a family member who is in a meeting is not a safety system. Ask whether there is a 24-hour monitored centre, what the answer-time commitment is, and what happens if nobody on the call list picks up.
  • What happens next? Answering is not responding. Is there an agreement with an ambulance service or armed response? Who has a key? Getting to the door is the part families discover too late.
  • Does it work in the garden and the bathroom? Most falls happen in the bathroom, and many happen outside. A base unit with a limited radius around it is a common and expensive disappointment.
  • What happens when the power and the network are out? In South Africa this is a weekly question, not a theoretical one. Ask about battery backup on the base unit and whether it falls back to a mobile network.
  • Will it be worn? An alarm on a bedside table protects nobody, and a device that is uncomfortable, ugly or shaming will end up there. This is the single largest cause of failure and it is not a technical problem.
  • What is the total monthly cost, and what is the exit? The device price is usually the small number. Ask about monitoring fees, contract length and whether the equipment is yours.

What a facility needs instead

In a hospital, retirement village or frail care unit, the answering problem is already solved — there are staff on duty. The problems are different ones, and a consumer pendant solves none of them:

  • The alarm must find the person. A resident who falls in a corridor, a garden or someone else's unit is not in the room the button is wired to. An alarm that says only "flat 14" sends staff to the wrong place.
  • It must reach the right staff member. Broadcasting to everyone means it is nobody's job. Routing to the nearest available carer, then escalating on a timer to the duty sister and the manager, is what turns an alarm into a response.
  • It must survive not being acknowledged. The failure that hurts is not the alarm nobody heard; it is the alarm somebody silenced and then got distracted from. An escalation ladder with timers is the control for that.
  • It must leave a record. How many calls, answered how fast, by whom, and what happened when the first responder did not acknowledge. Under section 23(1) of the Older Persons Act, the operator of a registered residential facility reports annually to the Minister on compliance with prescribed service standards and measures to prevent and combat abuse of older persons — and a record beats a recollection every time that report is written.
  • It must cover staff too. The same wearable that lets a resident call for help lets a carer alone at 02:00 do the same, which is a direct response to the employer's general duty under section 8 of the Occupational Health and Safety Act to provide a working environment that is safe and without risk to health as far as is reasonably practicable.

That is the system described on our nurse call and staff safety page. The button is the cheapest component in it.

Fall detection: what it really does

Automatic fall detection works by watching acceleration and orientation — a sharp deceleration followed by a period of lying still. It is genuinely useful, and it is oversold with a consistency that should make any buyer cautious.

What it is good at: the case that matters most, which is a fall where the person is unconscious, disoriented or cannot reach the button. That is precisely the scenario a manual button cannot cover, and it is why the feature exists.

What it is not good at: being certain. A slow slide down a wall to the floor may not generate the signature. Sitting down heavily, dropping the pendant, or a wheelchair going over a threshold may generate one that looks like a fall. Every such system trades these two errors against each other — tune it to miss fewer real falls and it will raise more false ones; tune it to be quiet and it will miss more. There is no setting that eliminates both, and any supplier who implies otherwise is describing a product that does not exist.

How to use it properly: as a safety net underneath a button, not as a replacement for one. And in a facility, the false-alarm side has a specific cost worth naming — every unnecessary alarm erodes the staff response to the next real one. That is why sensitivity should be set per resident against a care assessment rather than globally, and reviewed after the first month with the actual alarm log in front of you.

A boundary we hold: a fall-detection or pulse-sensing wearable is a safety device, not a medical device. It exists to make somebody go and look. It does not diagnose, it does not monitor a condition, and it does not replace clinical observation. Anything presented as clinical measurement in South Africa falls under medical device regulation administered by SAHPRA — a different regime with different obligations, and not one we operate in or claim to.

Where the wearable actually fails

After the specifications are matched, three practical things decide whether the system works. None of them appear on a datasheet.

  • Charging. A device that needs charging every night will spend some nights on the charger while its wearer is in bed — which is when a large share of incidents happen. Longer battery life is not a convenience feature; it is a coverage feature. Ask what happens when the battery is low and who is told.
  • Cleaning. In a care setting the wearable has to be wiped down with whatever the infection prevention and control programme specifies. A band that cannot take disinfectant will either be cleaned improperly or removed. Ingress rating of IP67 or better is the practical floor.
  • Dignity. This decides adherence more than anything technical. A band introduced as "so we can get to you quickly if you need us" is worn. One introduced as tagging is not. Under the Older Persons Act a resident of a residential facility has an express right to privacy, and an approach that respects it is also the approach that gets the device worn — those two things point the same way far more often than buyers expect.

What we would ask at a survey

  • Where do your incidents actually happen — rooms, bathrooms, corridors, gardens? Cover those, in that order.
  • Who is the first responder at 02:40, and who is second when they do not acknowledge?
  • What is your current answer to "how long did it take" — and where does that answer come from?
  • Which residents does a care assessment call for automatic fall detection on, and who reviews that list?
  • Do your staff have a way to call for help that works where they are, not where the wall is?
  • What will you do in month two with the alarm log from month one?

Related reading: wander management without surveillance for the resident who leaves rather than falls, POPIA and location tracking for the compliance design, and healthcare and retirement living for the wider picture.

This article is general guidance on selecting and deploying safety equipment. It is not medical, legal or clinical advice.

FAQ

Panic buttons and fall detection — common questions

What is the best panic button for elderly people in South Africa?

The honest answer is that the button barely matters and the response does. What you are actually choosing is a monitored service: who physically answers at 3am, within what time, what they do next, and whether anyone can get through the front door. Ask for the answer-time commitment in writing, ask what happens when nobody on the family call list picks up, and ask how the base unit behaves when the power and the network are out — which in South Africa is a weekly question. A cheaper pendant with a real response beats an expensive one without.

Where can we buy fall detection devices in South Africa?

For a person living independently at home, buy fall detection devices in South Africa as part of a monitored personal alarm service rather than as a standalone gadget — the pendant is widely available and worth little without somebody answering it, so choose the provider by their answer-time commitment and what happens after the call. For a hospital, retirement village or frail care unit, a consumer device is the wrong product: what you need is a facility system where the alarm carries the wearer's location, routes to the nearest carer, escalates on a timer and leaves a response record. That is what we supply, and we quote it per project against a site survey.

Does fall detection actually work?

Partly, and that is the correct expectation to have. It detects a sharp deceleration followed by stillness, so it is good at the case a button cannot cover — an unconscious or disoriented person. It is not reliable at every fall: a slow slide down a wall may not register, while sitting down heavily or dropping the device may register as one. Every system trades missed falls against false alarms and no setting removes both. Treat it as a safety net underneath a button, and be sceptical of any supplier who implies certainty.

Can we use consumer panic pendants in our retirement village?

You can, and many villages start there, but they solve a different problem and the gaps show up quickly. A consumer pendant assumes an off-site monitoring centre and a fixed home address; a village already has staff on duty and needs the alarm to say which room, corridor or garden the person is in right now, route to the nearest available carer, escalate on a timer when nobody acknowledges, and leave a record of how long it took. Those four things are the facility product, and none of them are features of a home pendant.

Is a fall-detection wearable a medical device?

Not as we supply it, and we are deliberate about staying on the right side of that line. It is a safety device whose purpose is to make somebody go and look. It does not diagnose, monitor a condition or replace clinical observation, and we do not present pulse sensing as clinical monitoring. Anything offered as clinical measurement in South Africa falls under medical device regulation administered by SAHPRA, which is a different regulatory regime with different obligations — if that is what a facility needs, it is a conversation with a medical device supplier.

How do we stop false alarms from wearing staff down?

Set sensitivity per resident against a care assessment rather than globally, and then review it in month two with the actual alarm log in front of you — most facilities never do the review, which is why their systems drift into being ignored. Combine that with an escalation ladder so alarms are owned by a named person rather than broadcast to everyone, and with rules that stop predictable non-events from alarming at all. Alarm fatigue is the mechanism by which these systems fail, and it is a configuration problem more than a hardware one.

What about staff — should they have buttons too?

Yes, and it is often the easier business case to make. The same wearable infrastructure lets a carer alone at 02:00, a nurse in an emergency unit or a receptionist facing an aggressive visitor call for help from where they actually are, with their location attached. Section 8 of the Occupational Health and Safety Act places a general duty on an employer to provide a working environment that is safe and without risk to health as far as is reasonably practicable, and in a health establishment regulation 20 of the norms and standards regulations imports that duty directly.

How long should the battery last?

Long enough that charging never competes with wearing. A device needing a nightly charge will spend some nights on the charger while its wearer is in bed, which is exactly when many incidents happen — so battery life is a coverage question, not a convenience one. Ask what the realistic life is in the alarm configuration you will actually run, not in a best case, and ask what happens when a battery gets low: a low-battery state that notifies staff is a feature, one that silently stops protecting somebody is a hazard.

Primary sources

The button is the cheap part. Buy the response.

Tell us where your incidents happen and who answers at 02:40. We will design the cover, the routing and the escalation around that.