Brief ReportsValidation of a multicenter computer-based surveillance system for hospital-acquired bloodstream infections in neonatal intensive care departments☆
Section snippets
Methods
The medical centers in New York, NY, participating in NYARP include Beth Israel, Maimonides, Memorial Sloan-Kettering, Montefiore, Mt Sinai, and the New York Presbyterian Hospital (Columbia and Cornell University campuses). These medical centers include 14 acute care hospitals containing 6500 beds and discharging approximately 270,000 patients a year. Hospital size ranges from 229 to 1500 beds. Data were collected from January 2000 through December 2002. The NYARP system detects all positive
Results
The BSIs detected by both studies are shown in Table 1. The sensitivity and specificity of the NYARP data collection system were 79% and 96%, respectively. Analysis of the data excluding BSIs caused by coagulase-negative staphylococci (CoNS) improved sensitivity and specificity to 84% and 99%, respectively. The positive predictive value of NYARP was 58% and, after excluding CoNS, improved to 84%. Before and after exclusion of CoNS, negative predictive values were 98% and 99%, respectively.
Discussion
There is increasing interest in using automated electronic surveillance to augment traditional infection control practices. Although computerized approaches hold great promise, these systems collect vast amounts of data that need to be validated and managed. Few studies have assessed the sensitivity and specificity of surveillance conducted by electronic clinical information sources, particularly in NICUs. By comparing the NYARP system's electronically collected data with that of data collected
Acknowledgements
The authors gratefully thank Elaine Larson, RN, PhD, for the use of the HHS data and Brian Currie, MD, MPH, for the use of the NYARP data.
References (6)
- et al.
Status of infection surveillance and control programs in the United States, 1992-1996
Am J Infect Control
(2000) Surveillance, reporting, automation, and interventional epidemiology
Infect Control Hosp Epidemiol
(2003)- et al.
Association rules and data mining in hospital infection control and public health surveillance
J Am Med Inform Assoc
(1998)
Cited by (26)
Advances in electronic surveillance for healthcare-associated infections in the 21st Century: A systematic review
2013, Journal of Hospital InfectionCitation Excerpt :In order to increase method validity, step-wise algorithm adjustments were performed in several studies. Graham et al. used microbiological and patient demographic data to identify BSI in neonatal patients.37 They found that by removing coagulase-negative staphylococci isolates, the results from their electronic method closely approximated traditional, ‘gold standard’ surveillance.
Informatics and Epidemiology in Infection Control
2011, Infectious Disease Clinics of North AmericaCitation Excerpt :Work is furthest along with central line–associated bloodstream infections (CLABSIs) because surveillance for these infections is the most amenable to straightforward rules. Early work by several investigators was promising for the detection of all nosocomial bloodstream infections3–5; however, these studies were not catheter specific. Trick and colleagues6 demonstrated that a relatively simple electronic rule-based system for determining CLABSI performed at least as well as surveillance by infection preventionists, with a sensitivity of 81% and a specificity of 72% compared with investigator review.
Validity of electronic surveillance systems: a systematic review
2008, Journal of Hospital InfectionCitation Excerpt :Tables I–IV summarise each article included in this review. Most (21; 87%) of the included studies focused on nosocomial infections including surgical site infections (SSIs),9–13 central venous catheter (CVC)-related infections,4,14,15 postpartum infections,6,16,17 bloodstream infections,14,18 pneumonia,19 and urinary tract infections.14 Nosocomial outbreaks or clusters, rather than individual cases, were investigated in two studies.5,20
Economics of infection control surveillance technology: Cost-effective or just cost?
2008, American Journal of Infection ControlCitation Excerpt :The quality of evidence must be viewed in light of inherent strengths and weakness of the chosen study design. Evidence supporting the effectiveness of computerized surveillance systems for identifying patients with various types of nosocomial infections in different hospital settings has been previously reported.16-23 These studies include evaluations of systems designed to track bloodstream infections, pneumonias, and urinary tract infections among patients admitted to select medical and surgical wards and adult and neonatal intensive care units, as well as studies conducted in whole hospitals.
Comparisons of health care-associated infections identification using two mechanisms for public reporting
2007, American Journal of Infection ControlCitation Excerpt :However, a limitation of the method is the need for trained infection control professionals, which is resource intensive. Increasingly, automated mechanisms linking laboratory reports and electronic clinical records such as medication orders for the identification of HAI are becoming available.16,17 For CLA-BSI, the most promising results are automated surveillance computer algorithms augmented by manually determining whether a central line was in place.
Developing information technology for infection prevention surveillance
2010, Critical Care Medicine
- ☆
Supported by National Institutes of Health grant 1RO1 NR05197-01, “Staff Hand Hygiene and Nosocomial Infections in Neonates.”
Presented in abstract form at the Annual Meeting of the Society for Healthcare Epidemiology of America in Salt Lake City, Utah, on April 6-9, 2002.