Mortality, length of stay, bloodstream and respiratory viral infections in a pediatric intensive care unit

Abstract Objectives We investigated whether diagnostic categories and presence of infections were associated with increased mortality or length of stay (LOS) in patients admitted to a pediatric intensive care unit (PICU). Methods A retrospective study of all PICU admissions between October 2002 and April 2016 was performed. Oncologic vs nononcologic, trauma/injuries vs nontraumatic, infectious (gram-positive, gram-negative, fungal bloodstream infections, common respiratory viruses) vs noninfectious diagnoses were evaluated for survival and LOS. Results Pediatric intensive care unit admissions (n = 2211) were associated with a mortality of 5.3%. Backward binary logistic regression showed that nonsurvival was associated with leukemia (odds ratio [OR], 4.81; 95% confidence interval [CI], 2.2-10.10; P P P = .003), encephalitis (OR, 6.93; 95% CI, 3.27-14.67; P P P P = .039), and pneumococcal disease (OR, 3.26; 95% CI, 1.21-8.75; P = .019). Stepwise linear regression revealed that LOS of survivors was associated with bloodstream gram-positive infection ( B = 98.2; 95% CI, 75.7-120.7; P Conclusions Patients with diagnoses of leukemia, lymphoma, cardiomyopathy/myocarditits, encephalitis, and comorbidity of bloodstream infections and pneumococcal disease were significantly at risk of PICU mortality. Length of stay of survivors was associated with bloodstream gram-positive infection. The highest odds for death were among patients with leukemia/lymphoma and bloodstream coinfection. As early diagnosis of these childhood malignancies is desirable but not always possible, adequate and early antimicrobial coverage for gram-positive and gram-negative bacteria might be the only feasible option to reduce PICU mortality in these patients. In Hong Kong, a subtropical Asian city, none of the common respiratory viruses were associated with increased mortality or LOS in PICU.