When it comes to questions of life, death, or lotto, what would you consider reasonable odds?
Countless numbers continue buying Powerball tickets despite the chances of a jackpot being vanishingly slim, but how do we make sense of ultra-low probabilities in the context of health?
These may not be typical questions encountered in the dental surgery, but as detection and treatment of disease grow ever more sophisticated, conceptualising and explaining the long shot will be essential, explain Hana Berhane, Estie Kruger and Marc Tennant in the latest installment of #TalkingTeeth.
Hana Berhane, Estie Kruger and Marc Tennant write:
Humans are not good at calculating the probability of events of ultra-low prevalence. It is not something we face regularly, and thus is not something we have practiced. For example, what is the probability of being killed by lightning? What is the chance of a coconut falling from a tree and killing someone? These are not things we think about so often.
But, in modern times with massive improvements in health, we are down to these sorts of probabilistic levels of calculation for many conditions that can have substantial public impact. Today, humans are required to make these sorts of calculations as they go about their daily lives, to weigh up their personal odds.
For example, is it safer for me to drive to work or to catch the train? What is the risk of me dying of polio? What is the risk of a severe consequence from vaccinating my kids? What are the chances of me winning the lotto tonight? These are all significant daily scenarios we face with ultra-low probabilities. Every action we contemplate is guided by a risk-benefit calculation, and the level of “comfort” we are willing to accept varies greatly between individuals and situations.
Let’s take the scenario of speeding to the airport, for example. Many worry greatly of the possibility that a flight they are on may crash, when in reality, the risk of aviation accidents in modern air travel is measured in the order of tens of millions. Contrast this with the probability of a speed-related traffic accident — one in 200,000 (depending on how you calculate it) — which is many, many times higher. Yet the vast majority of people tend to underestimate the latter and overestimate the former.
More specifically, in health, we have seen some conditions plummet in terms of both morbidity and mortality. Prostate cancer is one complex example. In the USA, this is the most common malignancy in males (after skin cancer), with more and 150,000 diagnoses annually. However, 92% of cases are localised and have a death rate of 0% at 5 years and just 2% at 10 years. In gross terms, then, males in the USA have a 1/1000 chance each year of being diagnosed with prostate cancer, and for every 100 such cases, 98 will live beyond a decade. We are now getting to ultra-low probabilities that are hard to fathom.
Another such condition that strikes fear into the heart of any parent is childhood leukaemia. During the 1980s this was the most fatal cancer in children under the age of 15 years in the United States, but a look beyond the headlines tells us more. Around 4 in 100,000 kids will be diagnosed with leukaemia (compared with 20 in 100,000 for all cancers in children), but thanks to modern care 3 out of these 4 will survive, even thrive. In real terms, the risk of dying in childhood from leukaemia is now at the same level as being attacked by a shark while surfing in Australia, and is trending lower still.
Such ultra-low probabilities can prove difficult to comprehend and weigh in decision-making.
When the odds are forensic
What can we do to help us better understand and contrast such long-shot scenarios? At low probability (measured in the thousands), diagrammatic representations are commonly used to convey risk to patients. But as we move to ultra-low risk we need other ways of communicating these levels of chance, and of gauging a “comfort” threshold for risk-benefit calculations.
This year our team looked at the problem of forensic dental identification, and the issue of ultra-low risk probabilities came to the fore. We asked people to consider the following question:
“If you have a piece of a jaw as the only part of your daughter’s/son’s mortal remains, what odds would your family expect that this was (or was not) your loved one?”
It was interesting to hear people’s views. But, what is the answer? Probabilities of 1 in 10, or 1 in 100 were considered crazy and dismissed out of hand. At 1 in 1,000 some were happy, but many were not, and — interestingly — many people remained unhappy at 1 in 10,000. What people accept as a reasonable level of ultra-low probability in such a scenario differs vastly, and are likely to be influenced by both social and cultural norms as well as the specifics of the question.
As health practitioners, it is vital that we appreciate how patients view such ultra-low probabilistic events and develop tools to help them understand the relative risks and consequences. This issue will only become more prominent as we progress to more effective systems of prevention and treatment.
Research scholar Hana Berhane and Professors Estie Kruger and Marc Tennant are from the International Research Collaborative Oral Health and Equity, School of Human Sciences, The University of Western Australia.