Septicaemia, E. coli bacteraemia, epidemiology, risk factors, health protection, public health
Objectives:
To examine the feasibility of anonymised
linked health data analysis for public health investigation of E. coli bacteraemia (ECB), a population based retrospective cohort study.
Methods:
From routine microbiology surveillance,
datasets were created of E. coli blood and urine culture records from 2005 to 2011 in Wales. The data was anonymised by the NHS Wales Informatics Service (NWIS) and electronically linked with administrative
healthcare usage datasets within the Secure Anonymised Information Linkage (SAIL) databank.
Results:
From the 10,511 distinct ECB patients
identified; 5,311 had a GP record in SAIL at any time point in the study (50.5% linkage); 97% (10,208) were linked with an inpatient event and 72% (7,567) were linked to an outpatient event during the year prior to ECB; 36% (3,754) were linked to a mortality record within 1 year post-ECB; 30% (3,249) were linked to a positive E. coli/coliform urine culture within 14 days prior to ECB; and 20% (2,079) were linked to an A&E attendance during the 3 months prior to
ECB. Limitation: SAIL began collecting A&E data on 01/04/2009. Linked health data models were created; Survival analysis and cox regression analysis of mortality post- ECB, Inadequate treatment of E. coli UTI and ECB, trimethoprim resistance in E. coli urine culture and ECB, risk factors for developing ECB in the over 65 population, a case- control study.
Conclusions:
Linked health data analyses determined risk factors for developing- and the rise of- ECB in Wales. Linked health data analysis is feasible for public health investigation.