Defining the problem in behavioural terms

Hi, this is Fabiana Lorencatto again– health psychologist and research lead at the University of College London Centre for Behaviour Change. In this brief module, I’m going to provide an introduction to defining our quality improvement problem of interest in behavioural terms. As you will recall from my introductory module, there are three key steps in a behavioural science approach to designing Behaviour change interventions. The first is defining, precisely and clearly, what is it that you are trying to change.

This may seem obvious and simplistic, but it is an absolutely essential first step that, if done right, will make it much easier to conduct the subsequent steps in the process of conducting a behavioural diagnosis, to understand Behaviour and context, and then selecting intervention strategies that are tailored to your problem of interest. Being precise about what is it that we actually wish to change will enable much more focused intervention design and also facilitate evaluation of our intervention. Yet, as I will discuss shortly, it is often overlooked, poorly done, and reported.

When it comes to the starting point for designing interventions and defining what is it that we want to change, a common pitfall is that we tend to think about or define the problem in terms of the outcome we want to achieve, or change.

For example, we may wish to reduce infection rates or improve infection prevention control. These are absolutely worthwhile outcomes to try and achieve. However, the issue from a Behaviour change intervention design point of view is that these are not, in themselves, behaviours. Rather, these are outcomes, or the results, of a number of different discrete behaviours. For instance, infection rates for infection control may be the result of hand washing or hand hygiene behaviours of hospital staff, whether staff are effectively using protective clothing, disinfecting and cleaning surfaces and instruments, maintaining appropriate isolation procedures, et cetera.

These are all different actions or behaviours which will likely interact with one another and contribute, in turn, to the broader outcomes of interests such as infection rates and control. Achieving these outcomes will, therefore, require changing one or more of these underlying behaviours. So when designing a Behaviour change intervention, it is important to first map out the systems of behaviours, underpinning your outcome of interest, and select which will be the target of the Behaviour change intervention. We can, then, conduct a behavioural diagnosis to understand what are the barriers and enablers to the selected Behaviour? For example, what are the barriers and enablers to hand hygiene in hospital staff?

We can, then, choose a Behaviour change intervention strategy to target the identified barriers and enablers to that specific Behaviour, such as the factors influencing hand hygiene. This will hopefully bring about change in that Behaviour, such as improving hand hygiene, which will in turn also improve the clinical health outcome of interest we were initially focusing on– so infection rates and control. Defining the outcome of interest in behavioural terms and breaking the problem down into smaller constituent behaviours is particularly important for antimicrobial stewardship, which is arguably a highly complex Behaviour. Indeed, antimicrobial stewardship has been termed an umbrella concept, or Behaviour, representing, or covering, a number of discrete set behaviours and actions.

For instance, in addition to the infection prevention control behaviours that might contribute to antimicrobial stewardship, antibiotic prescribing is one key Behaviour that is often the focus of antimicrobial stewardship programmes or interventions. Antibiotic prescribing is a Behaviour, but it can be further deconstructed into different sub behaviours. It involves multiple actions to be performed from ordering and conducting diagnostic tests to determine the presence and type of infection– decision making around type of antibiotic, dose, route, duration– timely administration of the antibiotic, recording these decisions in patient notes and charts. And then subsequently reviewing the antibiotic prescription to decide whether to continue, to switch, or stop, or de-escalate the antibiotic– these are all discrete actions or behaviours.

These actions may be performed at different time points in the patient care pathway or continuum by different individuals– such as nurses, pharmacists, microbiologists, junior versus more senior physicians, et cetera– and also across different settings, from primary care, care homes, to secondary care. And even within care settings as well, such as across different wards within hospitals– the factors that influence these different actions are likely to differ. For instance, the factors influencing the use of diagnostic tests are likely to be different from those factors influencing decision making around whether to continue or de-escalate an antibiotic. The influences on Behaviour will also differ across different individuals/actors, time points, and settings.

Therefore, we need to be precise about what is it that we are interested in– which Behaviour performed by who, where, and when– so we can conduct a more informative and focused behavioural diagnosis in the second stage of the intervention development process. Therefore, the first step in any intervention development process should be to take your outcome of interest and break it down into the smaller behaviours underpinning it. Map out the system of interacting behaviours underpinning your outcome of interest. Sometimes, the Behaviour of interest will be relatively obvious and well-defined, such as reducing smoking in adolescence. But more often than not, there is scope to be more specific.

However, the problem is that when you map out the system of behaviours, you often end up with a very long list of behaviours that you could target with your intervention. As I mentioned, different behaviours will have different influences. So if your intervention targets multiple behaviours, then it will need to address different influences and, thus, require different intervention strategies. This means you risk ending up with a highly complex intervention, which may not be feasible to do in practise. As such, we advise that you start small and build incrementally by first narrowing down to one or two key behaviours to intervene on more intensively rather than trying to change everything at once.

This can be in terms of picking one or two behaviours, or one or two specific actor groups, or specific settings to intervene on– such as specific specialties or wards in the hospital rather than targeting the whole hospital at once. You can narrow down the list of candidate behaviours by considering factors, such as the likely impact of the Behaviour on the outcome of interest. If you increase that Behaviour, is it likely to make a difference to the outcome you want to achieve? Think about how feasible that Behaviour is to address, given your resources and priorities. How acceptable is that Behaviour likely to be to those whose Behaviour you’re trying to change?

Is the candidate Behaviour interlinked closely with any other behaviours, whereby, if you change that Behaviour, there may be spillover or knock-on effects on other related behaviours, et cetera? Once you have decided on which specific Behaviour or behaviours you wish to change, it is important to then describe these as precisely as possible. You’re much more likely to succeed if you are specific. It makes it easier to conduct a more informative and focused behavioural diagnosis, and it also makes it easier to measure the Behaviour as a basis for evaluation. For example, being more physically active, handwashing, or antibiotic prescribing are all behaviours, but they are not specific.

Taking physical activity as a general example– we are much more likely to achieve Behaviour change if we have a more specific target Behaviour such as being physically active outside for 30 minutes three times a week. The factors influencing this type of physical activity are likely to differ from those influencing other physical activity behaviours such as reducing sedentary Behaviour in the workplace or even running a marathon. Being more physically active is also likely to pose different barriers / enablers for different population groups, such as children versus younger or older adults.

Therefore, if we want to more effectively understand what is driving physical activity, and what would it take to be more physically active, then we need to be more specific about what type of physical activity, where, and from whom we want to change. We can then design interventions tailored to the specific Behaviour, group and setting of interest. You’ll also facilitate an evaluation and measurement of physical activity. It is much easier to measure whether someone has been physically active for 30 minutes three times a week than it is to measure being physically more active in general. The same applies for antimicrobial stewardship.

Simply saying we designed an intervention to improve antimicrobial stewardship or even antibiotic prescribing is not sufficiently behaviorally specific and can be quite hard to measure and understand. We need to be precise about which behaviours, whether it be prescribing, reviewing, switching, or stopping antibiotics, for instance, by whom– i.e. which health care professional roles– where and when, that we are interested in understanding and targeting for change. As such, a helpful hubristic, or rule of thumb, we recommend for defining your Behaviour of interest in behavioural terms is to think about who needs to do what differently to whom, where, and when, or how often?

These points are summarised by the TACTA principle of intervention design, which will be explained in further detail by Professor Jill Francis in the next presentation.