1940
ESTABLISHING A RENAL DATA OBSERVATORY USING HEALTHCARE DATA LINKAGE METHODOLOGY
Black C, Marks A, Robertson L,Wilde K, Simpson WG, Prescott GJ, Smith WCS & Fluck N
Chronic Disease Research Group & Department of Renal Medicine, University of Aberdeen & NHS Grampian
INTRODUCTION: Chronic kidney disease (CKD) affects 8-10% of the adult population, rising to more than 50% of those over 85yrs. Understanding the epidemiology, disease progression and burden in terms of health service needs is important for informing healthcare planning. To be able to identify vulnerable patient groups and provide tailored care stratified by long term risks offers the potential to target care to optimise health benefits.The aim of this work was to establish a Renal Data Observatory enabling us to link healthcare data from various sources to study CKD.
METHODS: The Grampian Data Safe Haven (DaSH) provides a secure environment for storing, indexing and linking and analysis of healthcare data. A 10 year extract of the NHS regional biochemistry laboratory system for measures of renal function formed the core of the data platform. Using the unique patient identifier assigned to all people in Scotland registered with a general practitioner, data from a range of different sources could be linked to support bespoke study specific data extracts. Data linkages used a mixture of deterministic and probabilistic methods to enable linkage to hospital episode data, socioeconomic and geographical data, Scottish Renal Registry and death registry data. The GLOMMS-II cohort (with complete capture of all individuals with a low(<60) eGFR and a large sample of those with normal renal function) forms part of this renal observatory.
RESULTS: After data cleaning, validation and exclusions (e.g. non-residents) there were 4,302,216 creatinine, 130,326 albumin:creatinine ratio and 31,532 protein:creatinine ratio results for 417,295 individuals from 1st July 1999 to 30th June 2009. The proportion with an estimated Glomerular Filtration Rate (eGFR) of <60 ml/min/1.73m2 increased steeply with age (Fig 1). From this data platform, the GLOMMS II cohort of 70,780 people was established from a baseline in 2003 and, using record linkage, followed up for six years. GLOMMS II includes a sample of individuals with normal renal function as well as all people with CKD (~19,000) and those on renal replacement therapy (RRT). Using GLOMMS II, the adjusted incidence rate ratio comparing those with CKD to normal renal function for RRT, all-cause mortality and myocardial infarction was 436.9, 1.22 and 1.95 respectively. Being male, worse kidney function and proteinuria were associated with increased risk of poor outcomes. A simple clinical tool was constructed to describe outcomes at five years. A logistic regression prediction model for RRT performed well (c statistic 0.9581) classifying 98% of patients correctly. A planning tool was developed as a simple spreadsheet.
CONCLUSION:Having established the methodology and demonstrated the value for understanding CKD, work is underway to establish ongoing updates and new linkages to dispensing data, unique birth cohorts and maternity and neonatal data providing a rich resource for understanding renal disease. The renal data observatory provides a powerful platform for future renal research.