Why Trust in Data Matters: Words from Sigal Atzmon on World Health Day

Much has changed since the American writer Mark Twain was alive during the 19th century. But one of his most famous quotes - “lies, damned lies and statistics” – still rings as true today.

In fact, recent research suggests that our relationship with statistics is becoming more complex than ever before. We don’t just struggle to interpret data ourselves, but also increasingly feel weighed down by a deluge of misleading information from others.

Edelman’s latest annual Trust Barometer highlights this trend very clearly. Its survey of 32,000 respondents across 28 countries shows falling trust in governments and the media.

It is business, which is now viewed as the most trusted source of reliable information globally. The findings are consistent for both high- and low-income countries.

Governments are viewed as unethical and incompetent, while there is low trust in social media and a belief that formerly shared media environments have splintered into echo chambers.?

Nearly three-quarters of respondents believe that CEOs should defend facts and expose any questionable science. This is especially marked in the US where trust in business has risen sharply as politics and the media become viewed as more polarised.?

These societal expectations place a big responsibility on companies. But we willingly embrace them here at Medix.?

Advocating for the facts lies at the heart of our mission to provide people with the best healthcare for their needs. Their lives depend on our ability to correctly interpret and deliver life-changing information and data.

Providing educational information also plays a key role in helping people to understand their own health and make changes for the better. Healthcare companies are well placed to do this alongside the traditional media.

Unlike the media, we don’t need to chase sensationalist headlines to gain the clicks that directly link them to subscribers and advertising dollars. Here is one recent example of how this can play out.?

It occurred when Oxford University academics published a case-control study looking at possible links between hormonal contraceptives and breast cancer. Many media headlines ran along similar lines - hormonal contraceptives increase breast cancer risk. Seems frightening, doesn’t it?

And it’s easy to see why since the report’s core finding was a 20% to 30% increased risk. However, the highlighted data point is a relative risk value.

It is meaningless without an absolute value. Look at this and a very different picture emerges.?

The surveyed women were diagnosed with breast before aged 50. North America’s National Cancer Institute data shows that women in this age bracket have these risks over the following 10 years:

• Age 30? - 0.49% (1 in 204)
• Age 40 - 1.55% (1 in 65)
• Age 50 – 2.4% (1 in 42)

When a 25% increase is put into this perspective, the data becomes far more reassuring. For women aged 30 to 40, it rises from 0.49% to 0.64% (using the higher end of Oxford’s assumptions).

Framing data correctly is a service that healthcare companies can provide: deploying our numerical ability and medical knowledge to provide clear and balanced information about the latest medical thinking.

But this doesn’t mean that we always get things right. There’s one type of data all humans struggle with thanks to our brain wiring: exponential growth data.?

This was starkly illuminated during the Covid-19 pandemic. The Canadian cognitive psychologist Steven Pinker captured it in his Facebook post right at the beginning on March 11, 2020.

Pinker wanted to warn people, who might then have then been feeling quite safe, at how Covid-19 could quickly skyrocket out of control. He did so by asking the following question: if a patch of weeds doubles every day and covers a field in 30 days, how long before it covers half the field?

If you answered 15 days, then you wouldn’t be alone. The correct answer is 29 days. That’s exponential growth for you.

Unsurprisingly, many policymakers and even some scientists were blindsided by Covid-19 and didn’t put precautions in place quickly enough. One reason why is that our brains tend to make deductions along linear rather than exponential lines.?

So what can we do to have more trust in data??

As individuals, it’s always good to check the underlying source to see if data has been cherry-picked or if key points have been lost in translation.

As medical professionals, we have a duty of care as part of our educational remit, to explain data in ways the layperson can understand. We should always provide full context too: if academic research is based on a small subset, we should point it out, for example.

This April, we celebrate World Health Day. “Health for all” is the theme. We can use data to bring this closer to reality. But we must use it objectively if we’re to continue wearing the mantle of trust people have placed on us.

At Medix, we make this our profession: combining data science with technology and human operations to make a real difference in people’s lives.