Use of the Rhame and Sudderth method to estimate incidence resulted in most estimated incidence rates becoming negative values (below zero). Simulating prevalence from incidence data showed large variation in prevalence depending on the day of measurement. The predictive model best predicting incidence, with a proportion explained variance of 0.31, was the model including the mean length of hospitalization of patients with an SSI (LN), the mean interval between admission and onset of the SSI (INT) and hospital (as random effect). Adding prevalence to the prediction model did not improve the model.