Predictive markers for nosocomial infections in multiple trauma: comparison of 13 inflammation parameters

Objective

To estimate the capability of predicting nosocomial infections of the inflammation parameters: polymorphonuclear leukocyte (PMN) migration, white blood count, C-reactive protein, PMN elastase, neopterin, IL-6 and IL-8, malondialdehyde, plasma antioxidant capacity, procalcitonin, lactate and cortisol.

Setting

The ICU of a surgical department.

Methods

Twenty-six multiple trauma patients were enrolled in the study, nine of whom developed infections during their hospital stay. Inflammation parameters were measured daily until discharge from the ICU.

PMN migration was determined in fresh, whole blood with a novel ready-for-use membrane filter assay. The percentage of PMNs migrating from the blood into a filter upon F-Met-Leu-Phe stimulation was relevant. The other parameters were measured with conventional methods. (commercially available methods).

Statistics

For each parameter, cutoffs between the groups of infected patients before occurrence of infection and noninfected patients were determined with receiver–operator characteristic curves, and patients were assigned to one of two alternatives with different specifications: did values occur that were beyond the critical cutoffs at least on 2, 3, or 4 days; or at least on 2, 3, or 4 days in sequence, and vice versa: did such values not occur in a patient? Contingency tables were set up for the yes–no decisions for each specification in the infected and noninfected groups, and sensitivity/specificity were calculated. Significance threshold was P < 0.05 (Fisher's exact test).

Results

The specifications with the highest significance were taken as relevant. PMN migration below a cutoff of 6% at least 2 days in sequence occurred before infection in eight of the nine infected patients, but only in three of the 17 noninfected patients (i.e. a sensitivity of 88% and a specificity of 82%, P = 0.0008). Fever ≥ 38°C for 3 days in sequence showed a specificity of 94%, but a sensitivity of only 55% (P = 0.009). The other parameters had no significant discriminative power.

Conclusions

Among a variety of related parameters, PMN migration proved to be a sensitive predictive marker for infections. Impaired PMN migration indicates impending infections. Early recognition of an infection risk may help to initiate aggressive antimicrobial therapy before clinical manifestation of infection, thus improving therapeutic success.