Derivation of a score to predict infection due to multidrug-resistant Pseudomonas aeruginosa: a tool for guiding empirical antibiotic treatment.

OBJECTIVES

Multidrug-resistant Pseudomonas aeruginosa (MDR-PSA) constitutes an emerging health problem. A predictive score of MDR-PSA infection would allow an early adaptation of empirical antibiotic therapy.

METHODS

We performed a single-centre case-control (1:2) retrospective study including 100 patients with MDR-PSA and 200 with a non-MDR-PSA infection. Cases and controls were matched by site of infection, clinical characteristics and immunosuppression. A point risk score for prediction of MDR-PSA infection was derived from a logistic regression model. Secondary outcomes (clinical improvement, complications and discharge) were also compared.

RESULTS

Cases with MDR-PSA infection were younger than controls (67.5 vs. 73.0 y; P = 0.031) and have more frequent cirrhosis (9% vs. 2%; P = 0.005). Independent risk factors for MDR-PSA infection were prior antibiotic treatment (80% vs. 50.5%; P < 0.001), prior colonisation with MDR bacteria (41% vs. 13.5%; P < 0.001), hospital-acquired infection (63% vs. 47%; P = 0.009) and septic shock at diagnosis (33% vs. 14%; P < 0.001). Adequate therapy was less frequent in MDR-PSA infections (31% vs. 66.5% for empirical therapy; P < 0.001). The risk score included: previous MDR-PSA isolation (11 points), prior antibiotic use (3 points), hospital-acquired infection (2 points) and septic shock at diagnosis (2 points). It showed an area under the curve of 0.755 (95% CI: 0.70-0.81) and allowed to classify individual risk into various categories: 0-2 points (<20%), 3-5 points (25%-45%), 7-11 points (55%-60%), 13-16 points (75%-87%) and a maximum of 18 points (93%).

CONCLUSION

Infections due to MDR-PSA have a poorer prognosis than those produced by non-MDR-PSA. Our score could guide empirical therapy for MDR-PSA when P. aeruginosa is isolated.