Carriage testing in neonates and health care workers and environmental investigation were performed. An unmatched case-control study was conducted to identify risk factors for S marcescens isolation.
During the 35-month period, there were 57 neonates with S marcescens isolation in the NICU, including 37 carriers and 20 infected neonates. The prevalence rate of S marcescens isolation was 12.3 % in outbreak 1, 47.4 % in outbreak 2, and 42 % in outbreak 3. Nine of the 20 infected neonates died (45 % case fatality rate). A total of 10 pulsed field gel electrophoresis types were introduced in the NICU in various times; 4 of these types accounted for the 9 fatal cases. During outbreak 3, a type VIII S marcescens strain, the prevalent clinical clone during this period, was detected in the milk kitchen sink drain. Multiple logistic regression revealed that the only statistically significant factor for S marcencens isolation was the administration of total parenteral nutrition.
Total parenteral nutrition solution might constitute a possible route for the introduction of microorganisms in the NICU. Gaps in infection control should be identified and strict measures implemented to ensure patient safety.