ACE Inhibitor and Angiotensin Receptor-II Antagonist Prescribing and Hospital Admissions with Acute Kidney Injury: A Longitudinal Ecological Study

BACKGROUND ACE Inhibitors (ACE-I) and Angiotensin-Receptor Antagonists (ARAs) are commonly prescribed but can cause acute kidney injury (AKI) during intercurrent illness. Rates of hospitalization with AKI are increasing. We aimed to determine whether hospital AKI admission rates are associated with increased ACE-I/ARA prescribing. METHODS AND FINDINGS English NHS prescribing data for ACE-I/ARA prescriptions were matched at the level of the general practice to numbers of hospital admissions with a primary diagnosis of AKI. Numbers of prescriptions were weighted for the demographic characteristics of general practices by expressing prescribing as rates where the denominator is Age, Sex, and Temporary Resident Originated Prescribing Units (ASTRO-PUs). We performed a mixed-effect Poisson regression to model the number of admissions for AKI occurring in each practice for each of 4 years from 1/4/2007. From 2007/8-2010/11, crude AKI admission rates increased from 0.38 to 0.57 per 1000 patients (51.6% increase), and national annual ACE-I/ARA prescribing rates increased by 0.032 from 0.202 to 0.234 (15.8% increase). There was strong evidence (p<0.001) that increases in practice-level prescribing of ACE-I/ARA over the study period were associated with an increase in AKI admission rates. The increase in prescribing seen in a typical practice corresponded to an increase in admissions of approximately 5.1% (rate ratio = 1.051 for a 0.03 per ASTRO-PU increase in annual prescribing rate, 95%CI 1.047-1.055). Using the regression model we predict that 1,636 (95%CI 1,540-1,780) AKI admissions would have been avoided if prescribing rates were at the 2007/8 level, equivalent to 14.8% of the total increase in AKI admissions. CONCLUSION In this ecological analysis, up to 15% of the increase in AKI admissions in England over a 4-year time period is potentially attributable to increased prescribing of ACE-I and ARAs. However, these findings are limited by the lack of patient level data such as indication for prescribing and patient characteristics.