More on Reference Ranges

I recently referred to an editorial in the Annals of Clinical Biochemistry that outlines ways to establish reference ranges and their attendant advantages and disadvantages. The authors of an editorial in January's Clinical Chemistry (55:15-17, 2009 [subscription required]) advocate "a new way of thinking about the relationship between biomarker measures and clinical decision-making". They envisage the use of lab results to generate a prediction of risk of disease, which actually sounds a lot like existing likelihood ratios. These predictions will be based on data other than simply a lab measurement, for example, they might include the patient's family and social histories, results of imaging studies, and even the patient's desire for treatment. A result might be of the form that given the results of the lab studies, imaging, history etc. the patient has X% risk of having heart disease, cancer, etc. Certainly I agree with the assessment that traditional reference ranges have important limitations. To move beyond reference ranges to presenting the clinician with more useful data, specifically data that integrate findings from the history, physical exam, imaging, treatment desires etc. will take a lot of effort, much of which will be informatics related.