Presented at the Neonatal Society 2003 Autumn Meeting.
Richards M1, Saraswatula A2, DeVega I3, Coello R4, Holmes A1, Modi N2
1 Department of Infectious Diseases, Imperial College London, UK
2 Division of Paediatrics, Obstetrics & Gynaecology, Imperial College London, UK
3 Women’s & Children’s Directorate, Hammersmith Hospitals Trust, London, UK
4 UK Health Protection Agency
Background: Neonates rank among hospitalised populations at highest risk of bloodstream infection (BSI). Prospective surveillance of health-care associated infections assists in benchmarking, performance comparison, tracking of resistance patterns, testing of novel therapies and targeted infection control measures. The UK currently has no neonatal BSI surveillance scheme.
Aims: We conducted a prospective study aiming to (1) develop a BSI surveillance scheme based upon routine data collection within neonatal units and (2) determine the major risk factors for neonatal BSI upon which to base recommendations for rate stratification. The coagulase negative staphylococcus (CoNS), a skin commensal, is the commonest isolate from neonatal blood cultures, but may represent either true BSI or contamination. We therefore also aimed to apply consistent, unambiguous criteria for the diagnosis of CoNS BSI.
Methods: Daily data have been collected on all neonatal unit in-patients at the Hammersmith and Queen Charlotte’s neonatal units since 1999. This dataset comprises 36 raw items, 23 of which are used to assign level of care (BAPM 2001). The prospective capture of additional information triggered by a positive blood culture began in January 2001 and was extended to include negative blood cultures in May 2003. The data comprised maximal C reactive protein (CRP) response in the 24-hour period following the blood culture, relevant signs from 12 predefined clinical indicators of systemic infection and culture results. Occasions when a single known pathogen was isolated were used to establish the likely sensitivity of any combination of CRP response and clinical signs, and occasions when a culture was taken but found negative were used to estimate specificity. Only commensal or mixed cultures that conformed to the best criteria were considered true BSI. We calculated the rates of episodes of BSI per thousand patient-days and performed multiple regression analyses to examine the risk induced by static (birth weight, gestational age, gender) and daily (postnatal age, level of care, receipt of breast milk, parenteral nutrition, presence of endotracheal tube, central vascular device, peripheral vascular device, other invasive device) risk factors. A “daily” risk factor was considered applicable if present during any of the 3 days prior to the day the blood culture was obtained. An identical isolate re-occurring within 2 days of the initial positive blood culture was considered the same episode. The use of the neonatal clinical database was approved by the Hammersmith Hospitals Trust Caldicott Guardian.
Results: Over the 30-month period there were 294 episodes of BSI (93 pathogens in pure culture; 41 mixed growth; 160 CoNS, pure culture). A positive CRP response plus 3 or more clinical signs had a sensitivity of 75% and a specificity of 81% for the diagnosis of BSI. Multiple regression analysis, applied to all episodes of BSI, using this definition, found only parenteral nutrition, whether administered centrally or peripherally, to be a significant independent risk factor. Level of care was a significant risk factor having allowed for gestational age (LOC 1 vs LOC 3, RR 5.73 (95% CI 2.5,13.3). The BSI rate/1000 patient days was 13.4 for infants of 23-26 weeks gestation (95% CI 9.5, 19.1) and 20.3 (95% CI 15.2, 27.1)/1000 parenteral nutrition days.
Discussion: Reliable data capture for neonatal BSI is facilitated by incorporation into routine practice and is safeguarded by avoiding reliance on specific IT systems. We have demonstrated the feasibility of conducting neonatal BSI surveillance in conjunction with the daily data capture recommended by the British Association of Perinatal Medicine. We recommend standardised neonatal specific criteria for the diagnosis of BSI attributable to CoNS or other skin flora and risk stratification by gestational age and use of parenteral nutrition. Further work is needed to establish if BAPM level of care provides a measure of composite risk stratification and is thus suitable for benchmarking within the UK.
Acknowledgements: We would like to express our gratitude to the neonatal unit nursing staff for undertaking daily data collection.