Nosocomial infection rates in US children's hospitals' neonatal and pediatric intensive care units

Background: Few data are available on nosocomial infections (NIs) in US chi ldren's hospitals' neonatal or pediatric intensive care units. The Pediatri c Prevention Network (PPN) was established to improve characterization of N Is in pediatric patients and to develop and test interventions to decrease NI. Methods: Fifty participating children's hospitals were surveyed in 1998 to determine NI surveillance methods used and neonatal intensive care unit (NI CU) and pediatric intensive care unit (PICU) 1997 NI rates. Data were colle cted on standardized forms and entered and analyzed by using SPSS for Windo ws. Restults: Forty-three (86%) children's hospitals returned a completed quest ionnaire. All reported conducting NICU and PICU NI surveillance (range, 2-1 2; median, 12 months). Nineteen children's hospitals provided NICU NI rate data in one or more formats suitable for comparison. Denominators used for NICU NI rate calculations varied: 17 reported overall NI by patient-days; 1 9 reported bloodstream infection (BSI) by central venous catheter (CVC)-day s, and 8 reported BSI by patient-days. Sixteen (16) children's hospitals re ported NICU BSI data stratified by CVC-days and birth-weight cohort, and ve ntilator-associated pneumonia (VAP) by birth weight cohort was reported by 12. Twenty-four children's hospitals reported PICU NI rate data in one or m ore formats suitable for comparison. Denominators used for PICU NI rate cal culations also varied: 20 reported overall NI rates by patient-days: 23 rep orted BSI rates by CVC-days, and 10 reported BSI rates by patient-days; 24 reported VAP by ventilator-days; and 15 reported urinary tract infections ( UTIs) by urinary catheter-days. Median overall NI rates per 1000 patient da ys were 8.9 in NICUs and 13.9 in PICUs. Median NICU NI device-associated ra tes by birth weight (> 2500 g, 1501-2500 g, 1001-1500 g, and less than or e qual to 1000 gi were BSI 4.4, 4.7, 8.9, and 12.6, and VAP 0.9, 1.1, 4.9, an d 3.5, respectively. Median PICU NI rates per 1000 device days were 6.5 for BSI, 3.7 for VAP; and 5.4 for UTI. Conclusions: The number of months that NICU or PICU NI surveillance was con ducted varied among hospitals. Reported NICU and PICU NI rates varied by ho spital; some reported overall NI rates, and others focused on one or more p articular sites of infection leg, BSI or pneumonia). Many did not provide N ICU device-associated rates stratified by birth-weight group. Denominators used to calculate device-associated infection rates also varied, with hospi tals reporting either patient-days or device-days. These findings suggest t he need to determine reasons for variations and to identify optimal M surve illance methods at children's hospitals so that valid interhospital NI rate comparisons can be made.