Background: Duration free of central line-associated bloodstream infection (CLABSI) in a hospital may vary by type of patient population. We estimated patients median time to CLABSI by intensive care unit (ICU) type among acute care hospitals.
Methods: The study population was ICU patients whose CLABSI data were reported to National Healthcare Safety Network (NHSN) in 2016 under the reporting requirement of the Centers for Medicare and Medicaid. The unit of analysis was ICU location, not an individual patient. We conducted counting process survival analysis method to compute time (day) to a CLABSI beginning from Day 1 of first reporting month in 2016 in a given ICU location. Once a CLABSI occurred in a location, the start time of follow-up was reset to Day 1 after the date of event. The Cox regression method was used to explore the hospital and location-level characteristics that are potentially associated with the daily hazard of CLABSI for an ICU. We also assessed the proportionality hazard assumption of these factors. Adjusting for the vector of means of covariates, we then estimated median time to CLABSI by ICU location type, which is defined as follow-up time (days) by which 50% of events have happened in a given ICU type.
Results: In 2016, 6,935 ICUs at 3,384 hospitals reported CLABSI data to NHSN, with a total of 10,985 CLABSIs and 2,449,361 follow-up time in days. Factors associated with an increased daily hazard of CLABSI were the following: admission to a hospital with a large bed size, major teaching status and admission to a patient care location with a higher device utilization ratio (Table-1). Adjusted survival curves showed that median time to event (median CLABSI-free time) among ICUs ranged from 66 days (level-III neonatal ICU), 90 days (burn units) to 275 days (oncology units) and 284 days (cardiothoracic units) (Table-2, Figure-1).
Conclusion: The study demonstrated that ICUs with level-III care for neonatal patients and ICUs with burn patients were least likely to achieve the target of zero infection in a defined period and may warrant further targeted interventions. Similar research to investigate infection control performance through estimating median infection-free time is needed beyond ICUs and across multiple HAI types and facility settings.
See more of: Oral Abstract Session