ICU acquired late pneumonia: epidemiological, clinical, bacteriological and histological aspects of a 3 years study

Objective

To evaluate the incidence and the characteristics of ICU acquired late pneumonia and to correlate the clinical diagnosis of pneumonia to histological findings; to identify and emphasize the risk factors of its onset.

Subject and methods

ICU acquired late pneumonia was defined by the presence of new and persistent infiltrates to the chest X-ray (appeared at least 72 h after admission) in addition to at least one of the following: a) purulent sputum; b) T° >38°C or <36°C; c) white blood cells count >12 000 or <4 000/mm³. Tracheal aspirate, BAL or PSB were performed to obtain microbiological samples. We evaluated 573 patients, 380 males (66%) and 193 females (34%), consecutively admitted in a medical and surgical ICU from 1994 to 1996; the median age was 64 years. Patients ventilated at the admission were 395 (69%); 229 patients (40%) were admitted in ICU after a surgical procedure. The Apache II and the SAPS I scores at the admission were respectively >16 in 46% of patients and >9 in 76% of patients. The overall ICU mortality was 33%. We observed 112 episodes of pneumonia (crude incidence 19%): 52% of patients had undergone a previous surgical procedure; Apache II score was >16 in 63% of patients while Saps I score was >9 in 93%. Pneumonia lethality was 49%. All the dead patients were ventilated. In 47% postmortem examination was performed. Chi square test with Yates correction, T Student's test and Fisher's exact test were performed for statistical analysis.

Results

The incidence of pneumonia was higher in males than in females (22% vs 13%; P < 0.05); its frequency was higher in surgical than in medical patients (25% vs 15%; P < 0.01) and in ventilated patients than in not ventilated (26% vs 4%; P < 0.000001). Incidence in patients with Apache II >16 was 27% (P < 0.0001 vs Apache II <16); in patients with SAPS I >9 frequency was 24% (P < 0.00001 vs SAPS <9). The mean length of ICU stay (LOS) of patients with pneumonia was 36 days, while LOS of patients without pneumonia was 13 days (P < 0.0000000001). In 94 patients (84%) we obtained positive microbiological samples; gram positive germs were observed in 44% of isolates, gram negative in 38%, fungi in 17%. Post-mortem examination confirmed the diagnosis of pneumonia in 50% of cases; acute pulmonary oedema, ARDS and pulmonary infarction were the most common causes of misdiagnosis.

Conclusions

Pneumonia is one of the most common nosocomial infections, particularly in ventilated critically ill patients, and it is burdened by high lethality. Our data evidence that male sex, mechanical ventilation, high severity disease indexes scores, previous surgery and length of ICU stay are important risk factors for the onset of late pneumonia. Relationship between ARDS and pneumonia is strict, as ARDS may be both the result or the cause of pneumonia. How best to diagnose ICU acquired late pneumonia is still under debate, considering the frequent discrepancy between clinical and histological diagnosis.