Adjusted smoothed hazard functions which shows time variation of infection incidence from the day of first R administration according to the 4 different risk factors included in the final model. This is: lymphocyte counts at nadir (A); HIV sero-status (B); GVHD (C); type of malignancy (D). The hazard function for each specific risk factor is adjusted for the other 3. As reported in the multivariate analysis the graph shows that different risk factors accounted for increased risk of infections in different time periods. In particular, patients in the lowest quartile of lymphocytes counts at nadir were more likely to develop a clinical infection throughout the study period (A); being HIV positive and being affected from aggressive lymphoma seemed to be associated with clinical infections only in the first six months (B and D) while patients who experienced GVHD were at increased risk of clinical infection from day 183 up to the end of follow-up (C).