Dr. Joe Marasco talked about using Nomograms to improve the diagnostic process at http://www.bio2devicegroup.org . It is largely known that medical decision making is not a consistently reliable process. This is true, despite the fact that modern medicine makes extensive use of high technology, like diagnostic imaging, surgical advances with robots, genome sequencing and more. But scientific and mathematical understanding is lagging when it comes to processing the data and using data for decision making. There are now baby steps being taken with evidence based medicine, patient empowerment, and education of clinicians, to look closely at the data. Still it is a challenge as to how one can use data and information meaningfully. Patients still often do not have the vocabulary and do not have the time to process all the information. In essence, medical diagnosis is about probabilities; it is rarely definitive.
The question then rises is, how can scientific information be incorporated along with quantitative measures of diagnostic test effectiveness into a tool that the physicians can use and explain to the patients. Bayes’ theorem, named after Thomas Bayes, who first suggested using a theorem to update beliefs, is the foundation for nomograms. Nomograms as a tool can not only facilitate the relevant calculations, but could provide an ability to quickly visually scan the information and enable communication with the patient about following steps. Nomography dates back to 1870, first developed by Maurice d’Ocagne. A nomogram is a graphical calculating device. It is a two-dimensional diagram that allows drawing of one straight line from one scale to another. A nomogram is therefore very easy to use but challenging to construct. A nomogram can consist of a set of n scales, one for each variable in the equation. Knowing the values of n-1 variables the value of the unknown variable can be found or the relationship can be better examined. Dr. Fagan had constructed a Fagan nomogram in 1975 and it was cited hundreds of times during last 35 years. However, it has not much improved and has not been used much clinically.
Marasco discussed the recent Bayes’ nomogram developed by him and his colleagues, the “New Bayes’ nomogram” published by Marasco and his colleagues, Marasco, J., Doerfler, R., and Roschier, L. 2011, “Doc, What Are My Chances? The UMAP Journal 32 (4): 279-298”. Using Prostate specific antigen (PSA) screening as an example, in the determination of if and when such testing is appropriate, Marasco demonstrated how the nomogram can be used. PSA screening has recently been heatedly discussed in the medical literature. Based on the odds and probabilities affecting different individuals’ cohorts differently, different recommendations can be derived for screening. For instance, knowing nothing about a patient, he would have about 18% chance of developing the dreaded disease. But this would dramatically change if for instance, the patient was a 50 year old African-American man. The Likelihood Ratio (LR) gives the likelihood that a given result would be expected in a given patient. When LR is not specified then the sensitivity and specificity numbers would help graphically determine LR numbers and the user of the tool can still determine the value. The key element would be matching the patient to the right cohort. In practice, sensitivity and specificity are most commonly listed parameters.
The talk was followed by interesting questions like, “can all data be equally useful?” “Would older data be less reliable and more noise?” On the other hand, “more recent data would have less noise but no history”? So there would be some interesting challenges in constructing and using a nomogram, but it would still provide a quick, easy to use tool to discuss a plan of action based slightly more on facts, and less on subjective elements of bias, comfort level, fear factor, doctor-patient relationship, and anxiety level.