Reference intervals, etc.

The Gaussian (‘Normal’) distributionDrug therapeutic ranges in plasmaGentamicinSome important drug interactionsHaematology reference intervalsBiochemistry reference intervalsOnce upon a time, in a famous hospital named R— in the middle of England, there lived a crusty old surgeon and a brilliant young house officer. The surgeon issued infallible and peremptory edicts such as “All my patients with a haemoglobin less than 100 must be transfused.” Everyone did as the surgeon said (this was a long time ago) except for the wily house officer who understood statistics, sampling error, and the play of chance. One day she was rung up by the haematologist who asked her “Why have you requested 3 blood counts on Mrs Wells today? One is enough. You are wasting our resources!” “Not so,” said the house officer. “The first Hb was 98, the second was 97 and the third was 101g/L. I knew if I was persistent, I stood a good chance of preventing an unnecessary transfusion. She is a patient of Mr X.” The two conspirators smiled at each other down the telephone, and no more was said. Of course the right way of dealing with this problem is through clinical governance and dialogue with the surgeon. But the point remains: numbers are elastic, despite, on occasion, being given to 3 decimal places. Don’t believe in them as absolute entities, and don’t believe that the normal range is anything other than arbitrary; think before you act: think statistically. ...