Neonatal Laboratory-Confirmed Bloodstream Infection And Clinical Sepsis: Definitions Matter

Background: Six monthly point-prevalence surveys of healthcare-associated infections (HAI) were conducted in all neonatal intensive care units (NICU) in Finland between November 2008 and May 2009. Clinical sepsis (CS) and laboratory-confirmed bloodstream infections (LCBI) were the most common HAIs detected. The US surveillance network NHSN, using CDC definitions, and the German network NEO-KISS, using modified CDC definitions, report LCBI and CS adjusted for device utilization days. Both networks have reported an overall central line (CL) utilization ratio of 24% (CL-days/patient-days).

Objective: To compare the number of LCBI and CS cases fulfilling the definitions used by NHSN and NEO-KISS, and to assess the CL use in Finnish NICUs, in order to evaluate the options for future national surveillance, and to consider possibilities for comparing the Finnish HAI rates with those reported from other countries.

Methods: The data for the repeated prevalence surveys were collected by standardized forms. A form was filled for each patient present at the NICUs during each day of survey, including e.g. birth date, gender, date of hospitalization, birth weight, gestational age, use of invasive devices and procedures (recorded daily during a 7-day period prior to the day of survey), and date of onset and type of  HAI according to the CDC definitions version 1988. Additionally, a separate form for collecting data on clinical signs and symptoms and laboratory findings was filled for each patient diagnosed with sepsis by the clinician. These clinical data were reviewed to confirm the classification according to the NHSN criteria (CDC definitions version 2008), and to see which proportion of these LCBI and CS cases would have been detected by using the NEO-KISS definitions. A LCBI or a CS was considered as potentially CL-associated, if a CL was present 48 hours or less before HAI onset.

Results: A total of 1281 forms were obtained from 23 NICUs during the six surveys. A total of 6316 patient-days and 339 CL-days were recorded within the 7-day period prior to the day of survey (CL utilization ratio, 5.4%). Of the 164 HAIs detected by the CDC definitions, 13 (8%) were LCBIs and 87 (55%) were CS; the same LCBI and CS fulfilled criteria of both versions of the CDC definitions. Of these LCBIs 12 (92%), and of the CS cases 57 (66%), fulfilled the NEO-KISS criteria. Of all patients, 69 (5%) had been exposed to a CL during the 7-day period prior to survey. Of the LCBI or CS cases, 10 (10%) had a CL exposure, and 5 (5%) of the infections were potentially CL-associated.

Conclusions: The number of LCBI was similar, but the number of CS differed by one third depending on the HAI definitions used, suggesting that the choice of definitions is crucial. Our CL utilization ratio seemed low compared to those of NHSN and NEO-KISS.  Thus, the CL use policy in Finnish NICUs might be different, but this remains to be verified by incidence surveillance.