Long-term survival in patients included in a randomized controlled trial of treat, a decision support system for antibiotic treatment

Objectives: TREAT is a computerized decision support system for antibiotic treatment. In a randomized controlled trial it improved significantly the percentage of appropriate empirical antibiotic treatment and shortened hospital stay, while the usage of broad-spectrum antibiotics was reduced. The trial was not powered to show significance for the difference in 1month mortality rate. In the present analysis we looked at 6month survival in one of the hospitals (Beilinson Hospital) that participated in the trial. Methods:We plotted the Kaplan-Meier survival function for all patients [intention to treat (ITT)] and for patients treated according to the TREAT advice [per protocol (PP)]. The analysis was repeated for patients with clinically or bacteriologically documented bacterial infections. Results: At Beilinson Hospital 1683 patientswere included in the study, 860 in the intervention arm and 823 in the control arm. In the ITTanalysis180 daysurvival in control patientswas71%versus74%in the intervention patients (P=0.2). In the PP analysis the survival percentages were 71% and 77%, respectively (P=0.04). In patients with bacterial infections, in the ITTanalysis180 daysurvival inthe control groupwas68%versus71%in the intervention patients (P=0.1). In the PP analysis the survival percentages were 68% versus 74% (P=0.04). Conclusions: The present data support an effect of the TREATdecision support system on 6month survival, mainly because of its benefit in patients with documented bacterial infections.