Serum immunoglobulins in acute lymphoblastic leukaemia.
M. M. Reid
DOI: https://doi.org/10.1136/adc.65.12.1379-b
1990-12-01
Archives of Disease in Childhood
Abstract:SIR,-I have read with great interest the paper by Shaw and Dear. ' I think that this paper in fact addresses a very important and often neglected subject; I have however some misgivings about their results and their conclusions. The qualitative expressions of probability used in the paper are obviously vague, but in real life they can be better defined by the context in which they are used, by the way words are uttered, by gestures, non-verbal language, prior and subsequent communications on the same subject, etc. I do not think that a physician, speaking to a mother, would say only 'Your baby is jaundiced, and will probably need phototherapy' (statement 1), and vanish. Judging an isolated statement is not representative of interpreting a real communication, so the conclusion that the use of these words can produce misunderstandings might stem from the rather artificial study context, and might not be warranted. I am surprised that doctors differ in a systematic way from mothers in the interpretation of the qualitative expressions of probability used. These expressions are taken from everyday language, and have no specialised meaning in medicine. The differences between doctors and mothers could in part be due to artefactual factors: in particular, doctors could be more accustomed to the use of scales and to probability axioms. Using a 10 point scale as in the paper, it is natural to think that something having 0-3 (30%) probability to happen receives a '3' score, something having 0-8 probability an '8' score, etc. This requires a knowledge of the axiom that probability can assume values between 0 and 1 only, and this would not be so 'natural' among most people. In fact, Shaw and Dear report that the more educated mothers behave more similarly to doctors than the less educated mothers do. So, the problem could not lie in different interpretation of the same expressions, but rather in different ability to use the scale (that is, an artefact). Finally, although I agree that in important decisions verbal expressions are to be avoided, I am not convinced that using numerical expressions of probability per se would resolve the problem. In fact, both doctors2 and lay people3 seem to err frequently in handling numerical estimates of probability and important principles of quantitative inference. After reading the Shaw and Dear paper, I carried out a small experiment: I asked 12 mothers of healthy neonates in our ward to answer the following written question: 'If the probability that a neonate undergo phototherapy for jaundice is 45%, do you think it more likely that he will undergo phototherapy or not?'. I obtained only four correct answers. I'm not claiming that this small sample dismisses the use of numerical probability expressions altogether, but it raises the possibility that also in numerical communication the context may be important, and the 'message sent' may differ from the 'message received'. So, in the absence of further evidence, I suspect that numerical expressions of probability also do not convey information unambiguously between doctors and mothers. The question remains open.