Chapter 13 The linkage of incident renal replacement therapy patients in England (2002-2006) to hospital episodes and national mortality data: improved demography and hospitalisation data in patients undergoing renal replacement therapy.

INTRODUCTION: Missing data has hampered the comprehensive and inclusive reporting of adjusted outcomes for patients on renal replacement therapy (RRT) captured by the UK Renal Registry (UKRR). Furthermore the information collected by the UKRR does not currently include morbidity after starting RRT, details on hospital admission rates or location of death. Linking datasets offers the opportunity to enhance existing data and describe new measures of centre performance.

METHODS: 21,633 incident patients, starting RRT between 2002 and 2006, were linked to all hospital care recorded by the Hospital Episode Statistics (HES) database and Office of National Statistics (ONS) mortality data using a secure anonymised service. Comorbidity prior to admission was determined from ICD10 coded HES admission diagnoses before the start of RRT, along with missing data on ethnicity and socioeconomic status. Location of death was determined by comparing the ONS and UKRR date of death to concurrent hospitalisations from HES.

RESULTS: 290,443 admissions, 2.2 million haemodialysis attendances, 1.5 million outpatient attendances and 11,546 ONS deaths were returned for this cohort. Coding depth improved over time and varied between centres. Following linkage 21,271 patients were suitable for analysis, with improvements in ethnicity completeness (75.5% to 98.9%) and socioeconomic status (72.0% to 98.6%). Comorbidity improved substantially from 53.7% to 98.1% with 93% concordance in those with UKRR data. Mean comorbid scores between centres was similar (0.73-1.14) but variation in the proportion of admissions under nephrology in the first 12 months and the location of death between centres was noted, suggesting differing policies, practices and coding methods.

CONCLUSIONS: Linking routine healthcare datasets with a national registry has dramatically reduced missing data and enables reporting of additional comprehensively adjusted measures of performance that allow more robust comparisons between centres. Hospitalisation frequency and associated mortality can be described in much greater detail. Linking routine datasets to national audits and registries represents an achievable, cost-effective and illuminating new way to evaluate services such as renal replacement therapy in the English NHS.