2048 ways to take medications

What the 2048 game can teach us about our medicine-taking behaviours

Issue #17 of LFG: Learning from Games is a different take on using games in science. In particular, how a game can be used to simulate behaviours related to taking medications (or not). Read on to find out more!

Hello JOMT Reader!

Most of us have probably needed to take medications at some point in our lives. Hopefully, it was a one-time thing like Advil or Reactine, but sometimes, those medications have to be taken for longer. Antibiotics often have to be taken twice a day for 10 days.

One of the reasons medications have to be taken multiple times is because our bodies can’t or doesn’t absorb all the medication. Another reason is that our bodies try to clear away medications as part of a natural detox process. Sometimes, the medication just has a really hard time getting to where it needs to go in the body, getting lost in the traffic of the body.

So, most medications are taken in doses. Twice daily for two weeks. Once every other day for five weeks. Once daily for three days with a break of four days, repeated for eight weeks. You get the idea.

But all these different ways of taking medications means that we have to keep track of these schedules, especially if more than one medication has to be taken at once. And research shows that we’re not really good at taking medication when we’re supposed to.

How good we are at taking medication when we’re supposed to, called the adherence rate, is about 51% for patients in the US taking medication for hypertension. In some countries like Gambia, that adherence rate is a dismal 23%. That means for medications that require ten doses, only two of them are taken when they are supposed to.¹

There are many reasons why medication adherence can be low. Sometimes it’s because patients feel like they are feeling better or beliefs get in the way of taking the medication and these patients make a conscious decision not to take medications (intentional non-adherence). Some patients just forget (unintentional non-adherence). Whatever the reason, non-adherence results in the patient not getting the right amount of medication to treat their condition.

Games, and more specifically gamification, has been proposed as a way to make taking medications more fun and engaging. I’m not entirely convinced that gamification is the answer to adherence: no matter how you slice it, taking medicines is not fun even with point streaks and badges. But I’m curious to see if games can help in some way.

Researchers at University of Warwick in England decided to see if they could use games to study behaviours related to taking medications. If they could develop a model that simulates different ways we can take medicines (and conditions that prevent us from taking them), we can also study how to get better at taking medicines.

How can games simulate the act of taking medication?

Let me start by saying that the researchers’ approach to using games to study behaviours related to taking medications was brilliant. They started with an addictive game (2048, a math-lite game where you combine number tiles until you reach 2048) as a metaphor for everyday life. Interruptions to the game, where parts of the game were blurred from view, were introduced to simulate a diseased condition. Finally, treatments, including taking medications, were a series of actions that could be taken to prevent or restore the game back to the original state.

Neat, right?

A sample of the 2048 game. Players press one of the directional arrow keys to combine the same number tiles together. In this case, UP would combine the “2” tiles in the top right corner forming a new tile “4”.

A simulation like this allows researchers to change the interruptions (i.e., disease) and the actions required to get back to normal (i.e., treatments). Maybe the interruptions start off less disruptively, simulating a slow progression towards a disease state. The series of actions to correct it could vary from a simple code entry to more elaborate actions like needing to enter multiple codes or clicking on a sequence of buttons to complete the treatment.

The researchers also included a mechanism that simulated disease relapse. Not all diseases are cured as soon as the medication is taken. Some require continued treatment, otherwise, the disease could come back, sometimes even stronger or more aggressively than before. In this experiment, if the participants failed to enter their codes correctly and on time on a number of occasions, relapse could occur in the form of a blurred screen.

How did this all come together?

The goal of the experiment was to study behaviours related to taking medications. So, the researchers needed a way to measure baseline (control) behaviours and compare them against different conditions that represent the act of taking medications.

For the baseline (control) condition, participants were asked to play the 2048 game for about 15 minutes. Playing the game represented everyday life — a fun and engaging activity. Soon after starting, interruptions in the form of a partially blurred screen were introduced, simulating the onset of disease. Each person was given a code that they were required to enter once every minute (the treatment) for a total of 14 times. A pill counter in the corner of the game screen reminded participants how many times they had taken their medication (entered the codes) and how many remained.

The researchers compared adherence (percentage of successful code entries) between the control condition the following four conditions:

1. Financial incentives
   In this condition, participants played the game similar to controls, but were told that they would receive rewards (about 3 US cents) for every successful code entry.

2. Reminders
   In this condition, participants played the game similar to controls, received a pop-up message on screen for a short while, reminding them to enter the code.

3. Commitment device
   In this condition, participants played the game similar to controls, but only after signing a sticker that said they would commit to entering their codes. That sticker was pasted in view of their screen as they played the game.

4. Longer symptoms
   In this condition, participants played the game similar to controls, but were faced with interruptions that took twice as long as the control condition to clear.

What did the researchers find?

Long story short, the researchers found that study participant behaviours matched those seen in the real world for patients taking medications.

Adherence rates for participants in each of the groups. Only reminders seemed to have a real effect in getting people to enter codes on time (i.e., take their medications when they are supposed to).

Compared to the control condition, only reminders seemed to increase the adherence rate. Surprisingly, neither financial incentives nor the commitment device had any significant effect to get participants to enter their codes on time.

Interestingly, longer symptoms increased the adherence rate, although not significantly, over controls. It makes sense that if interruptions to the game persisted for longer, participants would continue to enter codes to make the interruptions go away. But over the course of the short experiment (15 minutes in total), the symptoms probably didn’t last long enough (even though they were double that of the controls) to make a huge impact.

Finally, across all 509 participants, 73 experienced a relapse after they failed to enter their code on multiple occasions. The researchers expected that the adherence rate might spike upwards after experiencing an interruption in the game.

But they found the exact opposite.

Adherence rates spiked sharply downwards after a relapse. Why? When the researchers looked at when the relapse happened, they noticed that most of it occurred towards the end of the 15 minute experiment time. That means players only had two or three code entries left, and they focused more on playing the game than clearing the blurry screen.

What does this all mean?

The results from this simulated experiment were in line with what researchers were seeing in the real world with medications that people take to treat their conditions. That means this model can be used to study all sorts of different conditions without risking patient lives. It would be hard to try out different doses or different schedules of a medication on patients whose lives depended on those medications. But with this model, researchers have a safe way to test some of their ideas.

But it’s not all cures and miracles.

Models like this can lead to dangerous conclusions. In particular, the fact that adherence was higher in participants who experienced longer symptoms than controls led the authors to make the following statement:

An interesting idea that comes to mind is that maybe the pills can be made in such a way that they keep the patients feeling sick until the last day of their treatment.

Keep the patients feeling sick? That is not at all how this information should be used.

I will say that the authors did note (several sentences afterwards) that this is an extreme interpretation of their results. But it does highlight the dangers of models in general that allow for some non-ethical conclusions to be drawn.

Where do we go from here?

It’s important to remember that the goal of the experiment was to create a model that can be used to study behaviours that we wouldn’t be able to otherwise. In that respect, I think the researchers succeeded.

But it’s also crucial to keep in mind that a model doesn’t always accurately portray what happens in the real world. There are many other factors that haven’t been considered.

Like how disease burden is felt by those other than the patients themselves. Sometimes, the outside influence is a strong factor that plays into whether medication adherence.

Or like how habits are formed especially when it comes to taking medications. Taking medications on a regular schedule will feel like a chore at first, but over time, may become a habit that you don’t have to think about. The experiment was too short to address whether habit formation would have made a difference in the adherence rate.

Limitations aside, I am hopeful that research like this can lead to new insights that will make the whole process of taking medications just a little easier. It’s also great to see some of the different ways that games are being used for science!

If you want to read more about this study, you can access the article here for free: https://games.jmir.org/2024/1/e47141/

If you liked what you read, please consider giving this post a like and sharing it with your community!

1. These numbers vary wildly depending on the country and medical condition you are looking at. For some conditions, the adherence rate is very high (close to 90%); for other conditions, the rate can hover around 10%.