September 11, 2011

TeleSynth: Making Spoken Medication Reminders Intelligible

Posted in Uncategorized at 2:37 pm by mariawolters

Many people who have gone on a health regime, which can involve taking medications, changing their diet, or exercising regularly, fail to stick to the programme. The consequences can be serious, sometimes even fatal (or, in the case of forgetting the contraceptive pill, natal.)

Researchers and practitioners in medicine, pharmacy, pharmacology, public health, psychology, anthropology, engineering, and information technology are working hard to understand why people fall off the wagon, and what might help them climb back on.

One way to support people in remembering health-related behaviours is to remind them. The question that motivates much of my work is: How can we design effective auditory reminders (speech, tunes, noises) that are pleasant to listen to?

I don’t claim that auditory reminders are the solution to all problems – instead, within the massive community of people that work on helping people remember, I focus on the particular reminder modality (hearing) that I know best. When my colleagues Marilyn McGee-Lennon, Steve Brewster, and I asked hundreds of people to tell us how they would like to receive reminders, and what they would like to be reminded of, most wanted visual reminders, but around a third said that they wanted to hear their reminder messages – either as speech or as a little tune. (McGee-Lennon et al., 2011) My mission is to ensure that those people who want to hear their reminders get audio messages that are as pleasant as well-designed as possible.


As part of this research programme, I ran a small study with my colleagues Brian McKinstry, Christine DePlacido, Christine Johnson, and Vasilis Karaiskos that had the following aims:

  1. To determine how well older people with and without hearing problems can understand medication reminders when they hear them under adverse circumstances (e.g., high background noise, bad phoneline)
  2. To get older people’s views on how to design telecare reminders.

We focused on medications because medication names are extremely tricky words to understand and remember. They are infrequent, latinate, and long. Yet, replacing the names with descriptions such as “the little blue pill” is not safe. Many people with chronic conditions are on generic medications that are produced by several different companies – packaging and pills look different depending on the manufacturer to the other.

In our study, we worked with older people, because this age group is more likely to be on medication, and may have quite complex medication regimes. Older people also have a much broader range of sensory and cognitive ability than younger people – therefore, testing a solution with older people is a good way of making it accessible to a large proportion of the population.

How was the study done?

We tested how well people could remember medication reminders that were generated by one human and two computer voices. Reminders were either for one or four medications at a time. In some of the four-medication reminders, the medication names were repeated, in some, an explanation was added. In the reminders, we used both over-the-counter and prescription medications, and we checked beforehand to what extent participants were familiar with the medications.

Some examples of reminders:
“Please remember to take the following medication: Aspirin.”
“Please remember to take the following four medications: Paracetamol, Aspirin, Corsodyl, and Metformin.”
“Please remember to take the following four medications: Paracetamol, Aspirin, Corsodyl, and Metformin. I repeat: Paracetamol, Aspirin, Corsodyl, and Metformin.”
“Please remember to take the following four medications: Paracetamol, for your pain, Aspirin, to thin your blood, Corsodyl, for your mouth ulcer, and Metformin, for your diabetes.”

Participants were recruited from four general practices in Edinburgh. First, they rated how natural the voices sounded and provided comments in a short interview. Then, they heard 72 reminders. Each reminder was followed by a short sentence, such as “Lucy sees four blue ships.” After they had repeated the sentence, they were asked to pick out the medications they’d heard from a list of 24 medication names, complete with potential indications. Only 12 of those occurred in the reminders, the remaining 12 were distractors.

Participants heard reminders in one of four conditions: high background noise / bad simulated telephone line, high background noise / good simulated telephone line, low background noise / bad simulated telephone line and low background noise / low simulated telephone line.

Before the voice tests, we assessed participants’ hearing and asked them to do several memory, attention, and recognition tasks. 56 participants completed the study, and 44 of these fulfilled the inclusion criteria (no hearing aids, no hearing loss due to middle ear hearing loss, no severe hearing loss).

What did we find?

Overall, people could remember one medication fairly well, but only if they already knew it. Medications were more difficult to remember when there was a lot of background noise, but there was no difference between the voices.

When hearing four medications, people mostly only remembered the one or two they knew. The human voice had a slight advantage over the computer-generated voices. Medications were remembered a little better when they were explained, and best when the names were repeated.

When we spoke to our participants after they’d completed the task, the feedback was very similar to what we found in the 2011 study. Some liked the idea of getting reminders over the phone, some preferred getting texts, while others liked to tick things off on paper. It was important for people to be able to review the information they’d gotten and ask for further information if they needed it. People who took many medications managed because they wove them tightly into their daily routine.

What does this mean in practice?

It is feasible to give people spoken reminders over the telephone even if they have mild hearing problems, but these reminders need to be about something they already know, and they should be reinforced by other measures such as compartment boxes. It also helps to talk to people about their routine – if this is more or less regular, finding a way to integrate the medications with it might be all they need.

Finally, since people prefer to receive reminders in different ways, reminder systems should use different channels, e.g. texting and phoning.

Over to you!

So, what do you think these results mean for your practice? Is this relevant to what you do or too esoteric? Would you like more information about this research? What should we have done differently, and what would you like us to look at next?


I will give a talk about TeleSynth at the Health Informatics Scotland conference on Monday, September 12, 2011; we are also writing up the results for a journal.


TeleSynth was funded by the Health Services research programme of the Scottish Chief Scientist Office; it is part of the TeleScot family of studies. The official TeleScot page on the project summarises the context and design.

You can find more background about our work on home and telecare reminders on the web site of the EPSRC project MultiMemoHome.