The use of graph lines to interpret the importance of risk factors for giant cell arteritis

What is it about?

Logistic regression is a statistical method for classifying the risk of having versus not having a disease based on various risk factors (predictors). The output of the logistic regression usually involves odds ratios and p values. If the odds ratio is not equal to one, and the risk factor is statistically significant, then the risk factor is presumed to have an important association with the disease outcome. However odds ratios are difficult to interpret if your risk factors are binary (yes / no) versus continuous (e.g. bloodwork values). The graph points on a Kattan nomogram are easier to understand than the odds ratios, because the length and position of the lines signifies the importance of the predictor variable to the disease outcome. We applied Kattan nomograms for a logistic regression statistical model, to predict the risk of giant cell arteritis.

Featured Image

Why is it important?

Seemingly small but statistically significant odds ratios of continuous variables such as platelets and age, may contribute more to the risk of GCA than the larger odds ratios from binary predictors such as new onset headache.

Perspectives