Language for action: A case study with Prostate Cancer UK

Often discourse analysts work with high quality qualitative data. The data we get from quant surveys doesn’t have that gold standard richness. For many quant researchers open ended comments are illustrative or just coded up without really examining the language.

This is a classic story of ‘beautiful constraint’ – creativity forced by limitations. The traditional take on Prostate Cancer UK’s brief would be qualitative then quantitative research.  This wasn’t going to work here – we had to think laterally.

A knotty problem

As with all cancers, the pandemic saw sharp declines in prostate cancer test referrals. 14,000 men were predicted to have undiagnosed prostate cancer that needed treatment, – a third of the overall drop in undiagnosed cancers since the pandemic.

Prostate Cancer UK launched an initiative to find the 14,000 ‘missing men’ at risk of prostate cancer, directing them to an online risk checker, highlighting risk factors and encouraged those most at risk to talk to a doctor about a test.

But creating a message that is both clinically sound and easy to understand is a maddening task, because it doesn’t conform to the norm. Convention tells us how to catch cancer early: check yourself, watch for symptoms, go for screening. Prostate cancer often doesn’t have symptoms until already spread, and there’s no national screening programme. There’s a blood test, but you have to ask for it. And we can’t tell you to ask for it, because it’s not right for all men. Clinical standards say men must make their own informed decision.

The wrong message could cause unnecessary biopsies, overtreatment and worse outcomes. This pushed Prostate Cancer UK towards lengthy ‘know your risk’ messages that didn’t aim to influence behaviour. Needless to say, this resulted in competing views from healthcare and marketing stakeholders.

The beautiful constraint

We were up against a full house of the typical constraints:

1. Its ambition was vast – get into the weeds of claims and communication language, provide hard data as evidence and predict real life behaviour.
2. Speed – not your average qual-quant timeline with reflection and stakeholder alignment.
3. Cost - as a charity, budget was inevitably limited.

So we combined behavioural quant techniques with discourse analysis, in four strands of work:

1. Workshop – instead of filtering and optimising messages in qual, we finessed them to make them more dynamic and less ambiguous, workshopped with stakeholders to agree a shortlist.
2. Online panel study: Blending behavioural, ‘system 1’ techniques with considered, ‘system 2’ elements combined to gauge attention, interpretation, behaviour and memorability.
3. Open-ended questions: a fundamental part of the survey design to provide vital data points for our discourse analysis.
4. A follow up questionnaire to assess behavioural impact and give confidence that people wouldn’t flock to their GP unnecessarily and bypass the risk checker.

There were multiple learnings to guide the team forwards, including:

The way risk is framed impacts how we process and retain information. We landed on ‘1 in 8 men will get prostate cancer’. We could not only demonstrate its impact and memorability, but the discourse also showed us why: people used much more language of personal relevance, such as ‘my loved ones’ ‘I hope I’ in responding to it.

If we talk about 1 in 8 men, and 1 in 4 black men, white men perceive themselves as low risk. If we change the language to 1 in 8 men and layer in a list of even higher risks, people retain the pertinent information to them.

People don’t read and retain information beyond 10 words. We timed their response and tested what they remembered, so we could evidence the need for much shorter messages.

Comprehension is harder when it’s counter cultural. Analysis highlighted influence of cultural expectation around testing and cancer – people expect to see ‘get tested’ messages, so frequently misinterpret risk checker messages. 

The perils of hedging language. In striving for accuracy, organisations often add words like ‘can’ or ‘may’, particularly in healthcare. But these magnify uncertainty in the reader, lead to indecision and hamper comprehension. Having hard data helps align stakeholders on the need for definitive language.

Results

This work formed a crucial strand of evidence which helped secure a £1.9M partnership campaign with the NHS for Feb/Mar 2022.  Results showed referrals for suspected prostate cancer in March were 28% higher than the pre-pandemic average…the highest rate of NHS urological referrals per working day ever seen.

This reached nearly 5% of the 11,508,321 UK men 45+ – 442,067 learned they’re at increased risk of prostate cancer.

It gave Prostate Cancer UK a set of shared principles about language. It serves as a great demonstration that quantitative open-ends can be a fundamental part of the study design, and get you much further in the face of ever-more-pressing balance of time and budget.

See the programme and headline speakers for Insight Alchemy 2023 – MRS annual conference.