Acute kidney injury: an acceptable risk of treatment with renin-angiotensin system blockade in primary care?

Background: Use of renin-angiotensin system (RAS) blockade has become increasingly widespread driven by evidence-based guidance. There is concern about the role of these agents in the genesis of avoidable acute kidney injury (AKI). Objectives: To investigate the association between AKI and use of RAS blockade. Design: Multilevel hierarchical analysis of a large cohort of patients registered with UK general practitioners. Setting: Primary care practices in East and West Kent, United Kingdom. Patients: 244,715 patients from 27 practices. Measurements: Demographic, clinical, biochemical and prescription data. Methods: Analyses of data acquired between 02/3/2004 and 17/04/2012 using multilevel logistic regression to determine the relationship between AKI and use of RAS blockade; further analysed by indication for treatment with RAS blockade. Results: Sufficient serum creatinine data were available to define AKI in 63,735 patients with 208,275 blood test instances. In 95,569 instances the patient was prescribed a RAS antagonist of which 5.4% fulfilled criteria for AKI. The unadjusted odds ratio (OR) for AKI in those prescribed RAS blockade was 1.93 (1.81-2.06, 95%CI) falling to 1.11 (1.02-1.20, 95%CI) when adjusted for age, gender, co-morbidity, GFR category, proteinuria, systolic blood pressure and diuretic therapy. In patients with an evidence-based indication there was no difference in absolute risk of AKI. However, prescription of RAS blockade in the absence of indication appeared to be associated with greater risk of AKI. When analysis was repeated with AKIN2/AKIN3 as the outcome, although risk of AKI remained significant when unadjusted (OR 1.73, 95%CI 1.42-2.11, p<0.001), after full adjustment there was no increased risk (OR 0.83, 95%CI 0.63-1.09) in those taking RAS antagonists. However, when analysed by indication AKIN2/AKIN3 was significantly more likely in those prescribed RAS antagonists without indication (OR 2.04, 95%CI 1.41-2.94, p<0.001). Limitations: Observational database study. No information concerning hospitalisation. Prescribing assumptions and potential inaccurate coding. Potential survival bias; patients surviving longer will contribute more data. Conclusions: Use of RAS antagonists increased the risk of AKI, independent of common confounding variables. After correction for confounders the risk fell away and became non-significant for moderate and severe AKI. However, where there was no evidence-based indication for RAS antagonists the risk of AKI, whether mild, moderate or severe, remained greater.