Food Security and Nutrition Network

Extension of the Health Belief Model

Extension of the Health Belief Model

Posted by tdavismph on 16 Mar 2017

Barrier Analysis is based on several health behavior models, in particular , the Health Belief Model (HBM). The Health Belief Model (HBM), which was originally created to investigate why some people fail to undertake preventive health measures, remains one of the most widely employed theories of health behavior. The attached 2012 paper by Orji et al discusses some of the behavioral determinants that are part of the Health Belief Model (which are also assessed in Barrier Analysis), and also how adding several determinants to the model (future consequences, self-identity, concern for appearance, perceived importance) led to a 78% increase (from 40% to 71%) in predictive capacity of the extended (compared to the original) Health Belief Model when studying food consumption behavior.

When I developed Barrier Analysis in 1990, I never intended for it to be a closed model that would not be adapted based on emerging scientific discoveries about behavior. I also have encouraged people to add on their own questions to explore hypotheses about why some people do a behavior and others do not. Is it time to consider adding other determinants to those assessed with Barrier Analysis, at least in some contexts? How do we balance the exploration of many possible determinants with the practical use of the tool (e.g., wanting to keep interviews brief)? What is your opinion?

 

 

Download
File Attachments: 

The way forward for diagnosing behavior change

Posted by btidwell on 24 Mar 2017

Tom,

Thanks for this interesting post, as well as a reminder to us all of the challenge it is to keep up with the pace of scientific progress in this area. 

I have an appreciation for Barrier Analyses, having conducted them in several countries on a variety of behaviors and finding the results useful.  I think they are well-designed to be intensely practical tools.  There has been significant progress made on understanding the brain, learning, decision making, and cognitive biases in recent years, and I think a discussion of the possibility of including these advances in some way would be fruitful. 

First, I should say that the natural successor to the BA framework has to be the RANAS model--it uses a similar Doer/Nondoer approach, is based on a similar framework of psychological constructs (Risks, Attitudes, Norms, Abilities, and Self-regulation), and while the approach is a bit more complex to apply (due to more steps and a large sample size to account for multiple comparisons), it includes a more robust process that incorporates qualitative interviews, D/ND analysis, and a Behavior Change Technique (BCT) selection process and even evaluation guidance for evaluating multiple "pilots."  It has been aggressively "marketed" by the research responsible group, and there are peer-reviewed publications as well as well-organized toolkits available describing the approach. 

I do have some basic concerns about the D/ND approach in theory.  The primary one is based on recent advances in neuroscience.  While D/ND approaches look at associations between certain attributes and behavioral performance, the causal direction (if any) is difficult to nail down.  For example, perhaps D have more supportive spouses than ND.  This may mean that spouses have helped to facilitate the change, or that the change was made for another reason and then the spouse saw the benefits and is now more supportive.  Instead, I find it more helpful to understand how we learn to do behaviors.  When we learn a new behavior, a process known as "reinforcement learning" essentially tells us if our behavior led to a desired outcome, and our brain chemically rewards us to encourage repetition of the behavior.  This is a powerful mechanism, and one of my favorite examples of its effectiveness is a simple robot programmed to move a multi-jointed arm at "random," see which motions lead to forward movement (the goal), and then preferentially select these behaviors.  Hence, it learns to "crawl" with only a goal and a map of previous experience, rather than a complicated series of instructions from a programmer (i.e., "Expert knowledge of how to crawl").  Neuroscientific studies tell us that this is essentially how we learn most behaviors (natural reflexes being a notable exception) as we grow and develop.  When and how these mechanisms fail (and when they succeed) is a major key to understanding why our behavior is so hard to change.  When we eat a Big Mac, our brain tells us to "do that again" because of the basic motivation for ingesting energy.  Even if we eat so much it makes us sick, we only learn to eat "a little less" next time, though the point at which we get sick is far beyond our need for basic sustinence.  On the other hand, washing our hands isn't inherently rewarding, as there is no direct reinforcement of "slight reduction in chance of disease in the future based on an individual instance of the action."  So we eat Big Macs and don't wash our hands.  Most of the behaviors we want to change in this way are really a failure to adapt to the modern world--rewards are delayed or unclear, environmental changes make our behaviors maladaptive, .  Our bodies are designed to live in pre-modern times, when consuming calories was a full-time job and we lived in dispersed-enough societies that we didn't spread epidemics when pathogens crossed over from animals to humans.  These "mismatches" are, I believe, the key to understanding barriers and facilitators of behavior change.

There are other specific issues that I believe are also worthy of reflection:

  • Dividing D vs ND is sometimes arbitrary, with different categories within both glossed over (i.e. if EBF means D, and one time of not BF means ND, then the differences between ND's who have colostrum-related beliefs, missed one feeding because of illness, supplement regularly with water in hot climates, and supplement with some kind of food are difficult to investigate)
  • Choosing the behavioral determinants to investigate requires either a broad list that's too difficult to emulate, or a narrowed list that's behavior-specific, but knowing how to choose the questions to ask is challenging (as you mention above)
  • Differences in "kinds" of behavior are insufficiently addressed--for example, stopping smoking is quite a different kind of behavior change than having a delivery in a health facility for many reasons
  • Once you've identified learning challenges or determinants, it's not clear how to develop an intervention other than plugging in the selected determinant into a communication (i.e., Norms -> "80% of people in your village do this behavior!"), when communications are not often the most effective method of behavior change, and much recent research points to the knowledge-behavior or intention-behavior gap.

I am obivously biased (as I'm currently pursuing a PhD in this exact thing and will be looking for work soon :] ), but I think reflection on this topic and re-engagement with academia would be a huge benefit to the donor/NGO communities and would welcome others ideas on ways forward!  I think there are approaches/theories/processes that have been developed recently (and not just from my research group :] ) from which our programs could benefit immensely.

Log in or register to post comments