Count the steps before the first reading

The median age of a patient enrolled with Ardenwick is 73, and about a third are older than 78. Many live alone. Many have arthritis in their hands or reduced vision, and a meaningful share have some hearing loss or mild cognitive impairment. When product teams in our industry talk about designing for this population, the conversation usually turns quickly to large type, high contrast and simple screens. Those things matter. In my experience they are also the smaller part of the problem.

The larger part is the number of things that have to go right before a first reading reaches a nurse. I have come to think of accessibility in remote monitoring as a count of steps and dependencies, and to judge a kit by how short that count is. A beautifully designed app that requires six prerequisites is less accessible than a plain device that requires none.

Every dependency is a step

In early 2021, when I joined Ardenwick, our standard kit was a Bluetooth scale and blood pressure cuff paired with an app on the patient's own smartphone, or on a tablet we supplied when the patient had none. On paper the process had five steps. When our researchers sat in patients' kitchens and counted, they found nearer twenty: charging the phone, finding the app store, remembering a password that a grandchild had set up years earlier, accepting a permissions request, switching on Bluetooth, keeping the phone within range of a scale that lived in the bathroom, and accepting an operating system update that quietly reset the pairing. Each was a place where an older person could stop, and many did. Among patients aged 75 or over, only half sent a first reading within thirty days of their kit arriving.

We moved to cellular devices that arrive already linked to the patient's record. The patient opens the box, puts the scale on the floor and steps on it. There is no phone, no password and no pairing, and the step count fell from about twenty to three. Within a year, first-reading rates among patients over 75 rose to 87 per cent, and the median time from delivery to first reading fell from nine days to three.

The lesson I took was not that apps are bad. Some patients prefer them, and we kept the option. The lesson was that every dependency we added was one the patient had to manage, often without help, and that our own process maps had hidden most of them. We now count steps in real homes for every change to the kit or to the patient's printed material, and a proposal that adds a step has to say what the step buys.

The second user in the kitchen

The researchers noticed something else. In about a third of the homes they visited, the person who unpacked the kit was not the patient. It was a daughter, a son, a spouse or a neighbour. Often that person also fitted the cuff, read the instructions aloud and answered the telephone when a nurse rang about a high reading. Our product had been designed for one user. In practice it had two.

Designing for the second user raises questions that type size never does. Who is allowed to see the readings? Who receives the call when the nurse is concerned? What happens when the patient and the caregiver disagree? HIPAA allows a patient to involve family members and others in her care, and our customers, who are the covered entities, set their own policies on how that involvement is recorded. Our job was to make the patient's choice easy to make, easy to see and easy to change.

Since 2023 a patient can name up to two caregivers at enrolment, with the clinic confirming the patient's agreement. Each caregiver has a separate login that shows only what the patient has agreed they may see. The nurse's screen shows, beside every contact, whether that person is the patient or a named caregiver and when the permission was last confirmed. It sounds administrative, and it is. It also ended a recurring confusion in which nurses were unsure whether they could discuss a reading with whoever answered the phone.

Hearing loss deserved its own work. Many of our patients struggle with telephone calls, and a call about a worrying weight is not a moment for misunderstanding. Patients can now choose a text message with a call-back time, a call with live captions on a supplied tablet, or a call to a named caregiver. The choice is stored with the patient's record and shown to the nurse before she dials.

Who pays for fewer steps

None of this is free, and I do not want to describe it as though it were. The cellular kit cost us US$154 against US$88 for the Bluetooth kit, plus a monthly data charge. Our gross margin fell from 62 per cent to 56 per cent in the year we changed the default, and our CFO was right to point out that the reimbursement our customers received had not changed by a cent. Caregiver accounts added identity checks, support calls and a permissions model our engineers had to rebuild twice. Home research visits cost several hundred dollars each.

I think the cost belongs with the vendor, at least at first, and I argued that to our board. A patient who never sends a reading generates no clinical value and, in the end, no revenue for anyone. A kit that works on its first day is cheaper than six support calls and a nurse's time spent chasing a patient who has given up. Our margin recovered to 60 per cent by 2024 as cellular volumes rose and support costs fell, though I would have made the same decision if it had not.

There is a broader point for the industry. The published guidelines on web accessibility are necessary, and we follow them, but they were written mainly for screens. Remote monitoring happens on bathroom floors and kitchen tables, in the hands of people who did not choose this technology and may not want it. The measure that matters is how many of them send a first reading without help, and how long it takes them. If a product team cannot tell you that number for its patients over 75, it does not yet know whether its product is accessible.