Severe bacterial infections are associated with mortality of about 30%. Patients with moderate to severe bacterial infections given early and appropriate empirical antibiotic treatment are at a lesser risk for a fatal outcome, with odds ratios ranging from 1.6 to 6.9. However only about 2/3 of patients worldwide are given early and appropriate empirical antibiotic treatment. About 40% of patients treated with antibiotics are given superfluous treatment.

TREAT is a computerized decision support system for antibiotic treatment in inpatients with common bacterial infections. TREAT is based on a state of the art stochastic model of the domain (a causal probabilistic network) and uses a cost benefit model for antibiotic treatment, including costs assigned to future resistance. It was tested in a randomized controlled trial in 3 countries and shown to improve the percentage of appropriate empirical antibiotic treatment while at the same time reduce hospital stay and the use of broad-spectrum antibiotics. The main limitation of TREAT is inherent in the limited information available within hours of presentation.

A second attractive approach to improve antibiotic treatment is to use techniques that do not depend on cultures, and thus shorten the time to identification of the pathogen to a few hours only. The LightCycler® SeptiFast test from Roche performs in vitro nucleic acid amplification test for pathogens causing bloodstream infections.

The purpose of the clinical trial is to show that the combined system TREAT/PCR assays will improve the outcome of inpatients with moderate to severe bacterial infections, while at the same time reducing the use of broad-spectrum antibiotics, with no or little additional costs. A secondary objective will be to assess the sensitivity and specificity of whole blood PCR, using TREAT as the reference standard.

Inclusion Criteria:

The rationale of the study will be to address patients for whom antibiotic treatment is, or should be, considered. Thus we will include consecutive patients detected in medical departments according to the following criteria:

1. Patients from whom blood cultures were drawn or

2. All patients for whom antibiotic treatment (but not prophylaxis or intra-peritoneal, inhalation or local antibiotics therapy ) is started within the last 96 days or

3. Patients fulfilling the CDC diagnostic criteria for any infection (see appendix C) or

4. Patients with temperature ≥38.3 measured orally (or 38 axillary or 38.5 rectal) on a single measurement or ≥38 / < 36 on at least 2 consecutive measurements separated at least 1 hr apart and one of the next criteria:

   (a) heart rate >90 beats/minute; (b) respiratory rate >20 breaths/minute or partial pressure of CO2 <32 mm Hg; and (c) white blood cell count >12,000/ L, <4000 L, or >10% immature (band) forms. or

5. Patients with shock compatible with septic shock. As defined as hypotension (arterial blood pressure <90 mmHg systolic, or 40 mmHg less than patient's normal blood pressure) for at least 1 h despite adequate fluid resuscitation; Or Need for vasopressors to maintain systolic blood pressure >90 mmHg or mean arterial pressure >70 mmHg and a suspected source of infection.

Exclusion Criteria:

1. HIV positive patients with opportunistic infection (suspected or proven)and / or AIDS defining illness in the last 6 months prior to data collection.
2. Children<18 years.
3. Suspected travel infections.
4. Suspected tuberculosis.
5. Pregnancy.
6. We will exclude patients not able to sign informed consent or not having a legal guardian willing to do so.