Fraser, Caroline Isabel; (2020) Linking electronic health records to evaluate infection rates in neonatal units in England. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Fraser_10105578_Thesis_sig-removed.pdf Download (15MB) | Preview

Abstract

Background: Bloodstream infection (BSI) in babies admitted to neonatal units (NNUs) is associated with mortality and morbidity. There is no single data source to measure national BSI rates in NNUs in England. Methods: I created a system for monitoring BSI rates in NNUs in England by linking electronic health records (the National Neonatal Research Database; NNRD) for babies in 110 NNUs to national infection surveillance data between March 2010 and June 2017. I evaluated how risk-adjusted BSI rates varied by regional neonatal network and over time using multilevel Poisson regression. I used the system to support evaluations of two interventions to prevent BSI. First, I assessed change in BSI rates following the adoption of peripherally inserted central venous catheter (PICC) care bundles using interrupted time series regression. Second, I evaluated the generalisability and applicability of BSI rates in a large pragmatic trial of antimicrobial-impregnated PICCs (the PREVAIL trial) to BSI rates in all babies receiving PICCs in NNUs. Finally, I estimated the proportion of all BSI in NNUs that occurred while a PICC was present to determine the contribution of PICCs to overall BSI. Results: I linked 3.3% of babies in NNUs (11,247 / 340,146) to at least one BSI which improved ascertainment compared to using NNRD alone (2.2%), but underestimated rates due to incomplete identifiers. Risk-adjusted BSI rates were similar across networks, but varied between NNUs within networks. From 2010 to 2017, risk-adjusted BSI rates per 1000 days of intensive and high dependency care decreased by 2.59% (95% CI: -4.23%, -0.93%) annually; the rate in 2017 was 1.52 (95% CI: 0.99, 2.04). Risk-adjusted rates of BSI per 100 admissions decreased by 3.39% (95% CI: -5.00%, -1.76%) annually; the rate in 2017 was 1.63 (95% CI: 1.11, 2.15). BSI per 1000 central line days remained stable at 2.25 (95% CI: 2.14, 2.34). I found no evidence of a change in BSI rates following adoption of PICC care bundles. BSI rates from the PREVAIL trial were generalisable and applicable to other babies receiving PICCs in NNUs. Of all BSI in NNUs, 46% occurred while a PICC was in situ. Conclusions: Accurate linkage of NNU electronic health records to infection surveillance data enables evaluation of risk-adjusted BSI rates for nearly all NNUs in England. These data can be used to evaluate interventions to reduce BSI among for all babies in NNUs, not only those with PICCs.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item