The importance of distinguishing between bacterial colonisation versus infection in assessing risk factors

Abstract number: 1134_04_198

Tacconelli E., Cataldo M.A., Manno D., De Angelis G., Fadda G., Cauda R., D’Agata E.M.C.


The majority of published studies use infection or clinical cultures as the outcome of interest when assessing risk factors for antimicrobial-resistant pathogens. We hypothesize that risk factors differ between colonization and infection.


Patients with Pseudomonas aeruginosa (PA) bloodstream infections (BSI) were identified during a prospective surveillance study (2002–2004) in a large University hospital, and pts with and without PA rectal colonization, respectively, were identified from a cross-sectional rectal screening of 1036 hospitalized pts. Two case-control studies were performed which included 87 randomly selected pts with PA-BSI (group-1), 29 with PA positive rectal screening (group-2) and 87 randomly selected pts with negative rectal screening (group-3). Exclusion criteria for group-2 and -3 were the detection of a clinical culture positive for PA during hospitalization.


Mean age ±SD of population was 58 ± 16 in group-1 and 66 ± 8 y.o. in group-2 (P = ns vs group-3) while the mean length of hospitalization before entering the study was 27 ± 44 in group-1 (P < 0.01) and 35 ± 37 in group-2 (P < 0.01). Resistance to the following drugs was detected more frequently among infecting than colonising strains: imipenem (28% vs 24%, P = ns), ciprofloxacin (55% vs 24%, P < 0.01), ceftazidime (32% vs 31%, P = ns), gentamicin (26% vs 17%, P = ns) , and piperacillin (18% vs 10%, P = ns). Multidrug resistance (2 classes of drugs) was reported in 50% of PA in group-1 vs 27% in group-2 (P = 0.04). At multivariate regression analysis, prolonged length of hospitalization (>17 days; group-1, OR = 2.8; group-2, OR = 3.7) and previous use of cephalosporins (group-1, OR = 6.9; group-2, OR = 9.3) and aminoglycosides (group-1, OR = 35.1; group-2 OR = 12.4) were independent risk factors for both PA colonisation and BSI (P < 0.05). Patients with PA-BSI had a higher Charlson score (>3; OR = 20.5) and were more likely to have had a ICU admission (OR = 5.8) and quinolones therapy within 30 days (OR = 4.3) (P < 0.01). The presence of wheel chair or bed-bound status was independently associated with PA colonisation (OR = 10.6, P < 0.01) but not with BSI.


This study provides novel information that may be used to better design future case-control or cohort studies analysing the epidemiology of antibiotic-resistant infections. Taking into consideration the differences between colonization or infection as the outcome of interest may reduce bias and help better design hospital interventions.

Session Details

Date: 01/08/2007
Time: 00:00-00:00
Session name: XXIst ISTH Congress
Location: Oxford, UK
Presentation type:
Back to top