Linkage, evaluation and analysis of National Electronic Healthcare Data : Application to providing enhanced blood-stream infection surveillance in paediatric intensive care

Purpose: National monitoring of variation in the quality of infection control in paediatric intensive care units (PICUs) requires comparisons of risk-adjusted rates. To inform the development of a national monitoring system, we evaluated the effects of risk-adjustment and outcome definition on comparisons of blood-stream infection (BSI) rates in PICU, using linkage of risk-factor data captured by national audit (PICANet) with laboratory records of BSI. Methods: Admission data for two children’s hospitals 2003–2010 were extracted from PICANet and linked using multiple identifiers with laboratory BSI records. We calculated trends of PICU-acquired BSI, defined as BSI occurring between at least 2 days after admission until up to 2 days following discharge. In one PICU, we compared rates of all PICU-acquired BSI with clinically significant PICU-acquired BSI submitted to the national surveillance system. Results: Of 20,924 admissions, 1,428 (6.8 %) were linked to 1,761 PICU-acquired BSI episodes. The crude incidence rate-ratio for PICU-acquired BSI between PICUs was 1.15 [95 % confidence interval (CI) 1.05–1.26] but increased to 1.26 (1.14–1.39) after risk-adjustment. Rates of PICU-acquired BSI were 13.44 (95 % CI 12.60–14.28) per 1,000 bed-days at PICU 1 and 18.05 (95 % CI 16.80–19.32) at PICU 2. Of PICU-acquired BSI at PICU 2, 41 % was classified as clinically significant. Rates of PICU-acquired BSI decreased by 10 % per year between 2003 and 2010 for skin organisms and 8 % for non-skin organisms. Conclusions: Risk-adjustment and standardisation of outcome measures are essential for fair comparisons of BSI rates between PICUs. Linkage of risk-factor data and BSI surveillance is feasible and could allow national risk-adjusted monitoring.