Predicting Outcomes in Emergency Medical Admissions Using a Laboratory Only Nomogram

Figure 2

The 30-day in-hospital mortality was related to the underlying level of albumin or urea at time of admission. The decile of each predictor variable was related to the 30-day mortality rate; the risk estimate was derived from the logistic regression multivariable model and was adjusted for Charlson Comorbidity, Chronic Disabling Disease, Sepsis and Deprivation Status. We used margins to estimate the average marginal effect. The cutpoints for albumin were 31, 34, 36, 38, 39, 40, 42, 43, and 46 and for urea were 2.8, 3.6, 4.2, 4.9, 5.5, 6.3, 7.4, 9.1, and 12.7.