Finally, the results of my randomized controlled trial are in.

Just to recap, the question I tried to answer was whether knowing that having a gene related to obesity (FTO) would prompt people to take action to prevent weight gain. I tried to answer this using the ‘gold-standard’ method for this kind of question: The randomised controlled trial. I randomly (by chance) assigned over 1,000 students from UCL to one of two groups. One group received a leaflet with seven tips which would help them to prevent weight gain. The leaflet was based on Habit Theory (more about this here). The other group received the same leaflet, plus obesity gene feedback for one gene (FTO) which told them whether they were at ‘higher’ (AT/AA variant) or ‘lower’ (TT variant) genetic risk for weight gain. I found out their genetic risk using DNA from their saliva (they all had to be willing to spit into a tube!).

One month later I sent both groups a questionnaire asking about their intentions to prevent weight gain, and any activities they were engaged in relating to weight gain prevention (e.g. eating slowly, controlling portion size, avoiding snacks, avoiding sweet drinks, exercising). They also completed a measure about their readiness to control their weight based on the stage of change theory.

Although only 279 participants responded to my questionnaire, the study had still sufficient statistical ‘power’ to draw some meaningful conclusions. We statistically controlled for factors which could potentially explain differences between groups; in this case age, gender and BMI.

Earlier studies have shown that genetic feedback can influence behaviour change intentions, regardless of whether the actual result is ‘low’ or ‘high’ risk. This might be because the results give personal feedback, which may itself be motivating. This is why we thought that gene feedback (vs. no feedback) would have an effect on people. And we were right – participants who received genetic feedback in addition to their weight control leaflet were more likely to think about taking some action to prevent weight gain. In particular, people who were already overweight (BMI < 25kg/m2) and received genetic feedback were more likely to report that they had started to do something to prevent weight gain than overweight people who did not receive gene feedback.

We then looked at differences between ‘higher risk’, ‘lower risk’, and ‘no feedback’ groups. Participants who received a ‘higher’ genetic risk result were more likely to report that they were thinking about doing something to control weight gain, or that they had started than people who received ‘no feedback’. There was a small difference between people who had ‘higher’ and ‘lower’ genetic risk results. Importantly, people who got ‘lower risk’ results were just as likely to think about preventing weight gain than those receiving ‘no feedback’. However, when we looked at whether people had actually followed the weight gain prevention behaviours outlined in the leaflet, there was virtually no difference between groups; most people were not following any of the behaviours despite their intentions.

This is the first trial that has had enough participants to show any group differences with some certainty. It also aimed to show effects in a ‘real world’ scenario, with young, healthy people who were largely unaware of their genetic risk. However it also had some very important weaknesses.

We did not assess people’s weight control intentions when they enrolled in the study because it would logistically have been quite challenging, so we couldn’t see if people’s intentions had changed. We also used only one question to make assumptions about their weight control intention. This is not such a good idea, because people sometimes give random answers, and self-report has its own problems – in hindsight it would have been better to use more questions because that allows us to check whether people answer consistently. Another limitation was that we could not have a ‘no treatment’ control group who received neither leaflet nor gene feedback. This was mainly because our study used lots of first year students who all lived in halls together; therefore, there would have been a high chance that people assigned to a ‘control group’ would have read the leaflet anyway. In addition, lots of people did not return the questionnaire. Although we expected this, it limits what we can actually say about how most students would react. People were more likely to enrol in the study if they were not overweight, and were less likely to answer the questionnaire if they were overweight at the start of the study. This means that our results may be different for these students compared to the wider student population, but we don’t know for sure. Lastly, and perhaps most importantly, I only chose to give them feedback on one (albeit well-established) obesity gene – although we know that there are hundreds of genes which influence body weight. This means that it might not be very meaningful for an individual to know whether or not they have just one of these genes – they may have many others. However, I was mainly interested whether gene feedback could ‘in principle’ be used to help people starting to prevent weight gain early, or whether it had any negative effects.

What to make of this? The study showed that FTO feedback can influence weight gain prevention intentions, but has no effect on actual behaviour. Sadly, showing that interventions change intentions but not behaviour is common in behaviour change research. In fact, it is so common that it has a name: The ‘intention-behaviour-gap’. I am sure that most people will be familiar with the concept: You really want to do something (i.e. going to the gym, or cleaning the bathroom), but then, for one reason or another, you fail to follow through with it. In that sense, findings from the study are in good company, since lots of other studies have shown similar things, be it on the effects of genetic test feedback, or on other topics. Unfortunately, researchers are as yet not very good in explaining how to bridge the ‘intention-behaviour-gap’. This is why we thought that genetic test feedback could be a novel way – especially since it is very compelling and rational to assume that once a person knows about their elevated risk for a condition, that they would take steps to prevent it. However, as it is so often the case with human behaviour, it seems that it is not so straightforward. A more optimistic explanation is that participants did not feel the need to act on their results at this point in time – after all, most had a healthy weight – but would keep the results in mind and take action should they gain weight. Since genetic testing for common, complex conditions is still relatively novel, data on the long-term behavioural effects is still lacking.

The good news is that a ‘lower’ risk result did not result in ‘complacency’ – the false assumption that weight gain is not possible with a ‘lower’ FTO gene result. People seem to have a pretty good idea that many genes, and the environment, act together to influence weight gain, so regardless of their result they were motivated to think about preventing weight gain as a consequence of getting feedback.

It will now be important to find out how we can get better at communicating gene results to people, so they may have some impact on behaviour –genomics is undoubtedly here to stay, so this will be an important task for the future.
Article reference: Meisel SF, Beeken RJ., van Jaarsveld CHM., & Wardle J Genetic susceptibility testing and readiness to control weight: results from a randomized controlled trial in university students. Obesity, 23, 2, 305-312. DOI: 10.1002/oby.20958
http://onlinelibrary.wiley.com/doi/10.1002/oby.20958/full

Have you ever made a New Year’s resolution? If so, you may have been assured – usually by a well-meaning supporter of your attempted transformation – that you only have to stick with your resolution for 21 days for it to become an ingrained habit. The magic number 21 creeps up in many articles about forming a new habit or making a change, but little is known about the origins of the ’21 days’ claim.

Psychologists from our department have devoted extensive time and effort to find out what it takes to form ‘habits’ (which psychologists define as learned actions that are triggered automatically when we encounter the situation in which we’ve repeatedly done those actions).

We know that habits are formed through a process called ‘context-dependent repetition’.  For example, imagine that, each time you get home each evening, you eat a snack. When you first eat the snack upon getting home, a mental link is formed between the context (getting home) and your response to that context (eating a snack). Each time you subsequently snack in response to getting home, this link strengthens, to the point that getting home comes to prompt you to eat a snack automatically, without giving it much prior thought; a habit has formed.

Habits are mentally efficient: the automation of frequent behaviours allows us to conserve the mental resources that we would otherwise use to monitor and control these behaviours, and deploy them on more difficult or novel tasks. Habits are likely to persist over time; because they are automatic and so do not rely on conscious thought, memory or willpower.  This is why there is growing interest, both within and outside of psychology, in the role of ‘habits’ in sustaining our good behaviours.

So where does the magic ’21 days’ figure come from?

We think we have tracked down the source. In the preface to his 1960 book ‘Psycho-cybernetics’, Dr Maxwell Maltz, a plastic surgeon turned psychologist wrote:

‘It usually requires a minimum of about 21 days to effect any perceptible change in a mental image. Following plastic surgery it takes about 21 days for the average patient to get used to his new face. When an arm or leg is amputated the “phantom limb” persists for about 21 days. People must live in a new house for about three weeks before it begins to “seem like home”. These, and many other commonly observed phenomena tend to show that it requires a minimum of about 21 days for an old mental image to dissolve and a new one to jell.’ (pp xiii-xiv)

How anecdotal evidence from plastic surgery patients came to be generalised so broadly is unclear.  One possibility is that the distinction between the term habituation (which refers to ‘getting used’ to something) and habit formation (which refers to the formation of a response elicited automatically by an associated situation) was lost in translation somewhere along the line. Alternatively, Maltz stated elsewhere that:

‘Our self-image and our habits tend to go together. Change one and you will automatically change the other.’ (p108)

Perhaps readers reasoned that, if self-image takes 21 days to change, and self-image changes necessarily lead to changes in habits, then habit formation must take 21 days. Although ‘21 days’ may perhaps apply to adjustment to plastic surgery, it is unfounded as a basis for habit formation. So, if not 21 days, then, how long does it really take to form a habit?

Researchers from our department have done a more rigorous and valid study of habit formation (Lally, van Jaarsveld, Potts, & Wardle, 2010). Participants performed a self-chosen health-promoting dietary or activity behaviour (e.g. drinking a glass of water) in response to a once-daily cue (e.g. after breakfast), and gave daily self-reports of how automatic (i.e. habitual) the behaviour felt. Participants were tracked for 84 days. Automaticity typically developed indistinct pattern: initial repetitions of the behaviour led to quite large increases in automaticity, but these increases then reduced in size the more often the behaviour was repeated, until automaticity plateaued. Assumed that the point, at which automaticity is highest, is also the point when the habit has formed, it took, on average, 66 days for the habit to form. (To clarify: that’s March 6^th for anyone attempting a New Year’s resolution.)

Interestingly, however, there were quite large differences between individuals in how quickly automaticity reached its peak, although everyone repeated their chosen behaviour daily: for one person it took just 18 days, and another did not get there in the 84 days, but was forecast to do so after as long as 254 days.

There was also variation in how strong the habit became: for some people habit strength peaked below the halfway point of the 42-point strength scale and for others it peaked at the very top. It may be that some behaviours are more suited to habit formation – habit strength for simple behaviours (such as drinking a glass of water) peaked quicker than for more complex behaviours (e.g. doing 50 sit-ups) – or that people differ in how quickly they can form habits, and how strong those habits can become.

The bottom line is: stay strong. 21 days is a myth; habit formation typically takes longer than that. The best estimate is 66 days, but it’s unwise to attempt to assign a number to this process. The duration of habit formation is likely to differ depending on who you are and what you are trying to do. As long as you continue doing your new healthy behaviour consistently in a given situation, a habit will form. But you will probably have to persevere beyond January 21^st.

Benjamin Gardner and Susanne Meisel

(www.ucl.ac.uk/hbrc/gardnerb)

References

Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology, 40, 998-1009. (http://onlinelibrary.wiley.com/doi/10.1002/ejsp.674/abstract)

Maltz, M. (1960) Psycho-cybernetics. NJ: Prentice-Hall.