Thomas Frost: What do we mean by bias


How many doctors would go one step further and talk to their patients about research bias? discusses Thomas Frost


In 1969, Enid Balint first spoke of ‘patient-centred medicine’, and the idea that every patient “has to be understood as a unique human being”.[i] Nearly fifty years later, healthcare has evolved from “doctor knows best” towards a model of patient ownership. At the same time, medicine has established a base of evidence for itself, growing substantially each year. As these two elements meet and conversations with patients about evidence become routine, the question of depth remains unclear.

Many patients will have a basic understanding of how some research works. Doctors could talk about studies and trials with patients; they might even discuss ‘evidence quality’. But how many doctors would go one step further and talk to their patients about research bias? And would it be worth it? I would argue yes.

The earliest recorded use of the word ‘Bias’ appears in the 16th century, from the game of bowls.[ii] Bowls were filled with varying amounts of lead on one side, giving them a ‘bias’ that meant they would travel a curved path in the direction of that bias.  Shakespeare’s Queen in Richard II complained that a game of bowls “twill make me think the world is full of rubs [uneven patches of ground] and that my fortune runs against the bias”.[iii] The analogies eventually stuck, and the idea of a bias diverting something persists today (even if the mental image might be lost).

We can find bias almost everywhere. We have cognitive biases, mental shortcuts tricking us into seeing the world in a certain light. As an example, which of these rare events is more likely to happen to you – dying in a plane crash, or being struck by lightning? Take a moment before you answer. You might want to be on the lookout for availability bias – mistaking awareness of an event for its likelihood.[1] Patients often experience this same bias when they report “similar symptoms” to a relative or acquaintance recently diagnosed with a rare illness.

If people are biased, what about research carried out by people? As an example – drugs for depression are amongst the most commonly prescribed medicines in general practice and in the US, they generate sales of billions of dollars annually.  And this week a team from the University of Oxford reported results from a mega-review of them stating their benefits. Yet, through a combination of rhetoric bias and publication bias, the usefulness of these drugs in anything but severe depression has been enormously oversold. Trials with negative data were either written in a manner to suggest positive results or simply not written up at all.  Both of those are diversions from the truth any patient could grasp.

These biases and much more have been, and are, ever present in medical research. David Sackett raised this in his 1979 paper ‘Bias in analytic research’.[iv] He called for ‘an annotated catalog of bias…’, which ‘…should include a useful definition, a referenced example illustrating magnitude and direction and direction of its effects, and a description of the appropriate preventive measures, if any’. Forty years on, the Oxford Centre for Evidence-Based Medicine has taken up the challenge.

The Catalogue Of Bias will be handy for research scientists and clinicians when preparing and appraising evidence but I also think it could be so much more. If patients can understand one or two personal biases, why not biases affecting the medical research they may see and act upon in news headlines? A communication tool designed to help patients with conversations about study quality could be invaluable. And if ‘good’ evidence is defined by its control of bias, then a toolkit of biases aimed at the layperson might be just the place to start.

[1] If you fly six times a year, the risks are the same. According to the TORnado and storm Research Organisation, on average 32 people are struck by lightning each year in the UK (~1 in 2 million). Aviation consultants To70 put the current rate of fatal aviation accidents at 1 per 12 million commercial flights… or ~1 in 2 million over six flights.

Author: Thomas Frost is, University of Oxford final Year Medical Student

Conflicts of interest: none reported


[i] Balint, E. “The Possibilities of Patient-Centered Medicine.” The Journal of the Royal College of General Practitioners 17.82 (1969): 269–276. Print.

[ii] “bias, adj., n., and adv.” OED Online, Oxford University Press, January 2018, Accessed 26 February 2018.

[iii] Shakespeare, William, and Charles R. Forker. “King Richard II.” London: Arden Shakespeare, 2002. Print.

[iv] Sackett, D. “Bias in analytic research.” Journal of Chronic Diseases 32:1-2 (1979): 51-63.