You are here

Know Your Bias

Author: Richard Simcock
Published on February 23, 2021

In 1847 Hungarian physician Ignaz Semmelweis clearly demonstrated that hand washing prevented deaths of surgical patients. Extraordinarily his colleagues refused to believe his theories and ignored this simple lifesaving intervention for decades. The biases of his peers stood resolute and Semmelweiss was ultimately committed to an asylum where, in cruel irony, he died of sepsis.

The ‘Semmelweis reflex is now a name for ignoring new advice that contradicts current paradigms.  It is an example of a (fatal) cognitive bias.  These biases may impede any medical development and the Comprehensive Geriatric Assessment (CGA) is no exception. Despite decades of high-level evidence and experience the CGA is not yet universally accepted as a necessity. There are financial, workforce and systemic constraints to CGA implementation but ahead of all of these are the cognitive biases we are all vulnerable to.  Sun Tzu famously urged “Know thy enemy and know yourself” so here is a quick view of the blocks in our path.

“Our assessments are already fine”
Anchoring bias leads to overreliance on a single aspect of the history or examination (e.g.  Performance Status (PS)). Those who think that PS alone is sufficient are making the mistake of Attribute substitution; assuming that easily calculated attributes can reasonably substitute for something more complex.  They are also biased by the ‘Law of the instrument’ by over-reliance on a single tool.

Some teams may not even use simple instruments, instead preferring gut instinct, prior results and a reliance on Stereotyping. By discounting the evidence that CGA will minimise risk for their patients they are trusting to luck. This blind-eye to probabilities is one form of Extension neglect. This same cognitive bias allows us to underestimate the scale of the problem of frailty in older patients with cancer (Scope neglect). Extension neglect also allows clinicians to focus on the (relatively) small group of patients we care whilst disregarding need in a larger group of anonymous patients.

“Our results are already good”
If cases have gone well the availability bias of those recent memories lead little room to persuade that improvement may be necessary. A run of cases without problems might lead to a Clustering illusion, sometimes described as Apophenia – seeing patterns where there are none – a term coined by psychiatrist Klaus Conrad.

Optimism bias allows some to greatly underestimate the likelihood of adverse outcomes and if the patient does have a successful therapy despite the quality of decision making then it’s easy to use Outcome bias to focus on the ends rather than the means. 

We know that routine use of CGA reduces the likelihood of therapeutic disaster but if a clinician has not previously witnessed harms then they may decline to plan for them. This form of cognitive dissonance is termed Normalcy bias.

With every case considered ‘successful’ the Confirmation bias that a CGA is not required grows. Holding this bias in the face of the mounting published evidence is an example of Continued influence effect. If the trials of CGA fail to persuade then Anchoring bias has led to a Conservatism bias.

Some teams may judge ‘success’ by only focusing on limited outcomes. Being able to overlook poor outcomes (impacts on quality of life, length of stay, rehabilitation) and focus on simple survival metrics is an Availability bias . This survivorship bias is made possible by not collecting data on the outcomes that also matter to make the invisible visible (e.g. PROMS - Patient-reported outcome measures).

Clinicians may suffer versions of egocentric bias; overestimation of their ability to control events (Illusion of control) and an overestimation of their judgement (Illusion of validity). These egocentric biases lead to Overconfidence effect (I am personally absolutely sure of this).

“Introducing the CGA is too difficult”
Some colleagues viewed change with suspicion and dread of a more difficult future; Declinism in action.  When there are multiple options for improvement, sticking with the status quo is ‘default effect’ version of Framing effect. Being averse to loss and avoiding the disruption of change feed into Status quo bias  and System justification bias.

Teams may not want to change because they are justly proud in the thing that they built (IKEA effect).

Colleagues may be unsure that the effort of CGA will bring a positive outcome, they are showing the Ambiguity effect, part of the Prospect theory discussed by Daniel Kahnemann.

The prominence of CGA at the 2020 ASCO meeting may lead some to think that CGA is ‘new’   (ignorant of previous decades of work). This misunderstanding is an example of Recency illusion.

The CGA presentations at ASCO fought for prominence with the newest and brightest oncology treatments. When Availability bias leads us to focus on the excitement of immunotherapy over the prosaic simplicity of CGA we become victims of Salience bias.

Some have the ‘double trouble’ bias of simultaneously overestimating their ability to treat an older person safely whilst underestimating how simple introducing CGA into practice may be. This bias was described by Festinger as the Hard–easy effect.

Semmelweiss provided a foundation and when he was stopped Lister and Pasteur continued to build until bias could no longer stand in its’ way.  Evidence and continued implementation are the tools to chip away at the bias of those who don’t believe.  I hope you recognise your own biases and are able to bust the biases of others!