Surveillance of ICU-acquired infections in Belgium: 2008 reference data

Introduction

This paper aims to provide reference data for pneumonia (PN) and bloodstream infections (BSI) acquired in Belgian ICUs taking patients' characteristics into account. This information could enlighten policy decisions. This 2008 set of indicators is the first yielded by the ICU Belgian surveillance (NSIH-ICU) since 2001.

Methods

The NSIH-ICU is a voluntary, patient-based reporting system set up in 1992 to follow ICU-acquired infections and guide prevention. It establishes a national risk-adjusted benchmark for infection rates, antibiotics use and invasive device-use ratios through uniform case definitions, data-collection methods, data entry and analysis. Only patients staying more than 2 days in the ICU are included. These aggregated database means are derived from the NSIH-ICU 2008 report.

Results

In 2008, 4,355 patients and 35,802 patient-days were reported by 19 ICUs. The age mean was 67.2 years; mean length of stay 8.1 days; mean SAPS II score 43.8; the proportion of patients with antibiotics at admission was 47.1%; medical admissions represented 63.4%, 4.2% were immunocompromised patients; trauma was reported in 6.7% of cases. Overall ICU mortality was 8.5%. The proportion of intubated patients was 44.4%, with central venous catheters (CVC) 68.3% and with urinary catheters 58.3%. Invasive-device use rates were: 358.5 intubation-days, 746.5 CVC-days and 644.2 urinary catheter-days per 1,000 patient-days. The ICU-acquired PN/100 patients was 9.8, 16.5/1,000 patient-days and the intubation-associated PN was 11.1/1,000 intubation days. ICU-acquired BSI was 3.6/100 patients, 5.3/1,000 patient-days and the catheter-related BSI was 4.3/1,000 catheter-days.

Conclusions

The 2008 report provided Belgian hospitals with comparative ICU-acquired infection data adjusted for patients' intrinsic and extrinsic infection risks. The participation changed across time and could cause a selection bias. Outlier verification is ongoing and efforts to increase the participation are being made. Indicator rates could be related to protocol modification, demographic trends and changes in patient mix linked to restructuration of medical services or in clinical practice patterns. Some are vulnerable to continuous medical and nursing in-service training, for example at surveying the antibiotics use and improving invasive devices use. Further studies could assess to what extent modifications in clinical practice and/or external interventions might cause changes in the patient mix.