As healthcare communicators, we have an important role to play in protecting and improving health. Our job is to create clear, accessible and engaging communication that encourage people to make healthier choices, improve the patient experience, and reduce patient safety risks.
by Michael Mbanunu and George London
For a long time, many health communications have lacked true inclusivity, being designed for a "default" audience, often assumed to be white, male, cisgender, heterosexual and non-disabled. Thankfully, we’ve become a lot more understanding on why inclusive communications matters. However, when we design our messages for one audience, we can unintentionally make it less relevant and accessible for another.
In this blog, we will look at some examples of how assumptions and “default patient” profiling reduces the safety and effectiveness of healthcare communications – and how AI is helping to overcome these issues.
Assumptions and the “default patient” problem
Health communications often make assumptions about how audiences see, interpret or apply information.
Have a think about common phrases for symptoms found on health websites, patient leaflets and clinical guidance such as "blue lips","red rash","flushed skin" or "pale appearance"; if you work in healthcare communicators you’ve probably used at least one of these phrases at some point. While these descriptions may seem native to say a white European person, they rely on a shared visual understanding that may not apply to all groups equally.
When we fail to account for differences in understanding or relevancy, it can result in delayed symptom recognition, lower health literacy, and increase health inequalities.
Case 1: Race and visual assumptions
For decades, clinical imagery, medical textbooks and health awareness campaigns have focused on how skin conditions appear on lighter skin tones, with less than 5% of images showing darker tones.
This matters a lot because conditions such as rashes, eczema, meningitis, and psoriasis look very different on lighter skin then on darker skin.
“Growing up I often saw chickenpox depicted on white skin in the media and in health care books. The common imagery of chickenpox is often represented as red spots on pale white skin. Being young, I just assumed that chickenpox wouldn’t affect me. In fact, I don't recall ever seeing different skin conditions shown on darker skin in the media growing up, which made me feel unseen.”
Michael Mbanunu (Marketing & Communications Assistant)
When patients and healthcare professionals are mostly exposed to images of conditions on lighter skin, recognising symptoms can become more difficult, which could lead to delayed diagnoses and poorer health outcomes.
Case 2: Gender and the "normal" patient
Health information is often influenced by a medical evidence base that has often not included women in clinical research. As a result, our understanding of symptoms such as cardiovascular disease often revolves around male symptoms i.e. awareness campaigns and educational materials. The downsides of public health messaging only showing male symptoms can lead to unrecognised symptoms and higher rates of misdiagnosis for women.
An additional challenge we must consider is how non-binary or transgender people might engage with communications. The assumption that, all people undergoing cervical screening would relate to a service targeted as “for women” risks excluding transgender men. There’s even some evidence that this language-based exclusion can cause distress for some trans people creating a barrier for them when accessing screening.
Case 3: Neurodiversity and inclusive language
Finally, another audience that we may not capture in our health communications is neurodivergent people. Some, neurodivergent people may find vague or ambiguous language more difficult to interpret, especially when instructions rely on personal judgement rather than being direct. Phrases such as "drink plenty of fluids" or "rest when needed" may seem clear to a neurotypical audience, but they leave room for ambiguity which can harm understanding for others.
As communicators, we often use metaphors or allegories to help our audiences understand complex or unfamiliar topics. Yet these tools can have the opposite effect for some neurodiverse groups; people with autism sometimes report finding non-literal language more difficult to interpret.
These examples all highlight a common challenge for us. We want our messages to be relevant, engaging and accessible – but how should we do this without alienating or disadvantaging certain groups? How should we decide who is in our “default audience” – should we choose the most underserved groups, or base our decision on statistically average characteristics for our audience?
On the face of it, it seems some form of unsatisfactory compromise is inevitable. Either we go for a “middle ground” of messages that are generic and bland, or we take a creative approach which is relatable and engaging for some people but excludes others.
But – what if there was a different way to approach how we deliver healthcare messaging?
Taking lessons in inclusivity from fast fashion
In healthcare, we tend to look at inclusivity as an issue of morality. There are economic arguments for tackling some health inequalities, of course, but the underlying drive for equity in health is a shared belief in fairness; that nobody should experience poorer health outcomes simply because of their ethnicity, their gender or any other characteristic beyond their own control.
This values-led approach has its advantages – but it is not the only way to approach inclusivity.
Take, for example, online fast fashion brands. Looking for inspiration on inclusivity from an industry notorious for worker exploitation might seem counterintuitive, but there is method to our madness.
In healthcare communications, measurement of how engaging and understandable your communications are, and whether they contribute towards actually changing behaviours is usually difficult. Did more staff get their flu jab because you chose a perfectly inclusive photo to include in that all-department email, or was the increase down to one of a million other factors outside our direct control as communicators?
The world of fast fashion is different. With tight margins, understanding the effect of every creative decision or penny spent on marketing counts. Digital advertising, in particular, has become a world where data and analytics are king.
In this environment, inclusivity is often given a different label: personalisation. Personalisation is not an entirely unfamiliar concept to healthcare communicators, but digital advertising and eCommerce are industries at the cutting edge – using a range of innovative technologies and techniques to ensure that advertisements are tailored to maximise their impact on the individual.
Machine learning and personalised messaging: a healthcare case study
One of the most interesting of these technologies is machine learning based creative optimisation. Machine learning is a subset of AI which uses algorithms to identify patterns in data and use these patterns to make predictions or decisions.
The idea behind this approach is relatively intuitive. Instead of considering an advert as a single “unit”, it breaks it down into different component parts, such as:
Visual elements (such as videos or photos)
Primary copy (such as the post copy for a social media advert)
Microcopy (such as URL previews or call to action buttons)
For each of these components, instead of having one option aimed at a “default” audience, several options can be supplied. The digital advertising platform then mixes and matches the different components for each individual user, creating hundreds (or thousands) of possible variations of the creative.
The variations shown to each individual user are not down to chance; the advertising platforms use their rich data about individual user preferences and machine learning algorithms to select the creative components they think will resonate most with the individual. You can find out more about how this (fairly complicated) process works from a technical perspective here.
We recently completed a project funded by the National Institute for Health and Care Research (NIHR), exploring how a range of innovative social media advertising techniques (including machine learning based creative optimisation) could be applied to recruiting participants for clinical research studies.
Research recruitment is a perfect candidate for exploring personalisation, because choosing to take part in research is a deeply personal decision. The motivating factors for taking part (or barriers to participation) are highly varied; there is no “default patient” and no universally relevant “silver bullet” message to encourage participation.
We were particularly interested in whether we could use machine learning based creative optimisation to improve the relevance and inclusiveness of our messaging for global majority audiences – who are often underrepresented in research.
Our approach – using Meta’s Dynamic Creative Optimisation tool – meant that instead of choosing a single message or visual to try and engage our audience, we could give the platform a range of options and find the best fit for each individual person.
The results were compelling:
Compared to a control group using standard creatives, our Dynamic Creative Optimisation adverts had a 41% higher engagement rate;
These ads also had a 21% lower cost-per-click;
The ads significantly increased the overall volume of participants onboarding and completing the studies we worked with, while also significantly improving the ethnic diversity of participants.
We’re now planning on applying these techniques more broadly to other research recruitment activities and other behaviour change campaigns.
AI is often criticised for taking the human touch out of communications. Sometimes that is true, but based on what we’ve seen it also has some real potential for overcoming one of the most stubborn challenges faced by healthcare communicators.
Perhaps, in the future, we’ll forget about the compromises of deciding on a “default patient” – and get used to personalising every message to the individual.
Michael Mbanunu and George London deliver communications for Health Innovation Network South London
*Sign up for the comms2point0 eMag*
The comms2point0 eMag features exclusive new content, free give-aways, special offers, first dibs on new events and much, much more.
Sound good? Join over 4.4k other comms people who have subscribed. You can sign up to it right here
AND if you’re an in-house comms pro you can join the comms2point0 comms lifetime WhatsApp group here.
Photo by Dylann Hendricks | 딜란 on Unsplash
