Skip to main content
  • Meeting abstract
  • Published:

ICU comparative survey of bacteremia in a teaching hospital

Many studies have established that nosocomial infection rates are more important in intensive care units (ICU). The objective of the study was to evaluate frequency and severity of imported and acquired nosocomial bacteremia in the different ICU of a teaching hospital in relation to bacteremia occurring in the entire hospital.


The prospective study was carried out from 1 November 1995 to 30 April 1996 in a 1837 bed teaching hospital including 121 ICU beds. Bloodstream infection (BSI) criteria were defined according to CDC criteria. All data were collected by a medical and nursing team from positive blood cultures isolated in the microbiology department, and the recruitment used an evaluation schedule. All infection data was validated by an external investigator and analyses were performed on EPI INFO.


Of patients with nosocomial bacteremia, 28.9% (69/239) needed a stay in ICU. The ICU represent 6.6% of the total beds, 10.9% of the admissions, and 6.5% of the hospital days. Nosocomial BSI complicated 1.68 per 100 admissions to the ICU during the study period with an incidence density of 3.61 per 1000 days of care. There were no significant differences in patient characteristics, except for their origin: transferring of another hospital increased the risk for patients admitted in ICU [Relative risk (RR) × 2.6, P < 0.001].

Clinically, hypothermia occurred nearly only in ICU (RR × 17, P < 0.001), but neutropenia ≤ 1500 was less present (3% versus 14%, P < 0.05).

Only respiratory tract infection multiplied by 5 (P < 0.0l) the risk of BSI in ICU patients comparatively to patients of other units. There was no significant difference for the other sources: intravascular devices, surgical wound, gastrointestinal tract and skin infections. On the contrary, genitourinary sources of BSI were more frequent outside ICU (with a protective effect in ICU: RR = 0.53. P < 0.01).

The patient was more frequently already treated with antibiotics in ICU when BSI occurred (49.3% versus 34.7%, P < 0.01).

The average length of stay in hospital increased by 8 days for patients needing a stay in ICU. The mortality rate also increased and was multipled by 3 (P < 0.01). The mortality was not bound with the underlying disease (Mac Cabe, surgical patients, ASA). SAPS II above 40 the day of the bacteremia was a predictive value of early mortality.


BSI prevention policy must take place in a total nosocomial infection control and must not be restricted to sole ICU.


This work was supported by the Scientific Committee of Option-Hôpital, with a grant of Roussel-Uclaf Company.

Author information

Authors and Affiliations


Rights and permissions

Reprints and permissions

About this article

Cite this article

Daumal, F., De Cagny, B., Defouilloy, C. et al. ICU comparative survey of bacteremia in a teaching hospital. Crit Care 1 (Suppl 1), P036 (1997).

Download citation

  • Published:

  • DOI: