Clinical vs Statistical Prediction in Dentistry

Paul Meehl is the hero of my hero. Daniel Kahneman says that he had more or less memorised Meehl's "disturbing little book" clinical vs statistical prediction when he was a student. The findings in this book are striking and increasingly relevant as data capture and technology progress.  

It sharply criticises the reliability of human decision making. Adhering to strict, rule based protocols, assuming sufficient data exists to develop these protocols in the first place, is markedly superior to reliance on intuitive "human" judgements for decision making. In a comprehensive literature review spanning half a century across medicine and psychology, conducted by psychologist William Grove in the year 2000, it was found that rule-based protocols outperformed human intuitive judgement 46% of the time, while intuitive judgement was superior only 8% of the time, with no significant difference in the remaining cases. However, even the absence of difference leans in favour of rule-based decision-making, considering its simplicity, cost-effectiveness, and consistency. Meehl's foresight was remarkable, with his book being published in 1954.

There are no strict algorithmic rules in dentistry, yet. This is because there are few quantitative data sets in dentistry. We rely on clinical assessments, photos, radiography and other imaging to make dental decisions. There is no number based data like ECG, blood tests etc to help create these rule based protocols that could create a system like Meehl suggests is so superior.  

When dealing with subjective data sets with no strict rules to follow, the least we can do is try to improve accuracy by not relying on an individual for intuitive judgments. Second opinions, or group discussions should be championed.

Further reading