Chronic kidney disease (CKD) is a major, increasing, health issue in the UK, and worldwide. A key indicator of future CKD is the albumin to creatinine ratio (ACR), which measures the amount of protein escaping from the kidney into the urine. It is regularly obtained through laboratory tests on patients' urine samples, ordered by GPs. There is some evidence that high values of ACR also predict future heart problems and stroke and may be associated with a reduced life expectancy. The strength of the association between ACR, and of recent increases in ACR, and major adverse health outcomes in the UK is unknown, and will be determined in the present study. Such information will enable more effective provision of health care resources and consequently reduce morbidity and delay death. We will estimate associations between ACR, and short-term changes in ACR, and future health by linking data from GP consultations, where ACR is measured, with data on (for example) heart disease from hospital records and death certificates. Since there are several other factors, e.g. smoking, that predispose to heart disease and other diseases of interest, we will account for these factors (also recorded at GP consultations) in our statistical analyses.
We will include patients with at least one ACR measure to estimate the association between ACR (after a log transformation, based on our prior work) and (for example) CVD using Cox proportional hazards models. We shall examine the patterns of association using splines and by estimating hazard ratios between ordinal groupings of ACR. Adjustments will be made for confounding variables and subgroup analyses undertaken. The UK population attributable risk for each of these outcomes associated with micro- and macro-albuminuria will be estimated. We will select subjects in CPRD who have at least two ACRs within about 3 years (the exposure window) to estimate the change in ACR by fitting a linear regression line for each subject. We will then use similar methods to estimate associations of change in ACR with CVD, but adjusting additionally for (1) the first ACR; (2) the last ACR within the window. We will alternatively estimate change in ACR as the observed relative difference between the first ACR and that closest to 3 years later, for those with ACR measured after (approximately) 3 years. The length of the exposure window will be varied in sensitivity analyses. Missing values will be imputed using multiple imputation by chained equations.
Renal outcomes will be derived as in our previous work in CPRD, using an algorithm incorporating death certificates, inpatient diagnostic or procedural codes, and primary care diagnostic/laboratory test results, and applying standard CKD definitions and creatinine-based CKD Epidemiology Collaboration (CKD-EPI) estimated GFR (eGFR) formulae25. CKD stages G4-G5 will be accepted if there were at least two eGFR measurements <30mL/min/1.73m2, spaced by at least 90 days, with no eGFR result >30mL/min/1.73m2 in the intervening period. The ESRD outcome will comprised those who died with mention of ESRD, or underwent kidney transplantation or maintenance dialysis (which was distinguished from acute dialysis by a record of CKD stage 5, permanent arteriovenous dialysis access or peritoneal dialysis). We have previously validated this algorithm using UK-renal registry data. CVD will be defined using primary care and linked HES and ONS mortality data using CALIBER definitions. Cardiovascular mortality will be defined as a death with Underlying Cause of Death attributed to a vascular cause (ICD-10 codes I00-I99 excluding I85) or sudden death (R96). Sensitivity analyses will include any mention of one of these codes on the death certificate.
Mark Woodward - Chief Investigator - The George Institute for Global Health
Mark Woodward - Corresponding Applicant - The George Institute for Global Health
Andrew S. Levey - Collaborator - William B Schwartz Division of Nephrology Tufts Medical Center
Brandon Neuen - Collaborator - The George Institute for Global Health
Clare Bankhead - Collaborator - University of Oxford
Colin Baigent - Collaborator - Johns Hopkins School of Medicine
Josef Coresh - Collaborator - Johns Hopkins School of Medicine
Kunihiro Matsushita - Collaborator - Johns Hopkins School of Medicine
Margaret Smith - Collaborator - University of Oxford
Morgan Grams - Collaborator - Johns Hopkins School of Medicine
Richard Hobbs - Collaborator - University of Oxford
Will Herrington - Collaborator - University of Oxford
Yingying Sang - Collaborator - Johns Hopkins School of Medicine
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation