Estimating the risk of acute kidney injury associated with use of diuretics and renin angiotensin aldosterone system inhibitors: A population based cohort study using the clinical practice research datalink

The reported incidence of acute kidney injury (AKI) in community-dwelling adults and hospital inpatients varies significantly depending on the criteria used [1]. A recent meta-analysis concluded that worldwide, one in five adults and one in three children experience an episode of AKI during an inpatient admission [2]. Studies in high-income countries have reported an incidence of AKI of 522/100,000 people per year in the community, [3] and up to 22.7/100 in an inpatient setting [4]. The incidence of AKI is likely to be increasing [3, 5] due to an ageing population with increased comorbidity and polypharmacy.

There is significant morbidity, mortality and economic cost associated with AKI. A meta-analysis of adverse outcomes following AKI conducted in 2009 [6] found the risks of chronic kidney disease (CKD) and end-stage kidney disease (ESKD) following a single episode of AKI to be 7.8 and 4.9/100 patient-years, respectively. Even mild AKI (a rise in serum creatinine of less than or equal to 25%) was associated with a 70% increase in mortality. In 2014 the financial burden associated with AKI in the United Kingdom (UK) was estimated to be £1.02 billion, just over 1% of the annual National Health Service budget [7].

AKI may result from reduced kidney perfusion, intrinsic renal disease or obstructive causes, with the first of these accounting for 75% of AKI episodes in hospital settings [8]. Risk factors include increasing age, sepsis, hypotension and chronic conditions (diabetes mellitus, congestive cardiac failure (CCF), CKD, atherosclerotic peripheral vascular disease, liver disease) [9]. Certain medications, including non-steroidal anti-inflammatories (NSAIDs), diuretics and agents that inhibit the renin-angiotensin-aldosterone (RAAS) axis have also been suggested to increase the risk of AKI in epidemiological studies, [10‐12] however the absolute risk of AKI amongst these individuals is unknown.

The absolute risk of AKI monsgt maintenance users of RAAS inhibitors and diuretics is unknown. This study aims to determine the absolute and relative risk of AKI in maintenance users of RAAS inhibitors and diuretics in a “real-world” setting of community-dwelling comorbid adults.

This study demonstrated a significantly increased risk of AKI in individuals taking RAAS inhibitors or diuretics in the presence of any additional antihypertensive medication. There was, however, no association between AKI risk and the number of antihypertensive medications in the unexposed group. Despite a significantly increased relative risk of AKI attributable to RAAS inhibition/diuretic use in the exposed group, the absolute risk of AKI in these individuals was small. Further, as demonstrated in the PERR analysis, we observed an increasing risk of AKI in the months preceding initiation of RAAS inhibition which has not previously been reported: Increased risk in the exposed group was seen in those with the longest duration of diagnosis of chronic condition. This could be explained by a worsening of clinical condition warranting initiation of further medication. Alternatively, it could imply reverse causation whereby having an AKI episode results in an increased likelihood of exposure to RAAS inhibitors or diuretics, which needs consideration in future observational studies of AKI risk.

The association between RAAS inhibitor/diuretic use and AKI risk demonstrated in this study is consistent with previous literature [10, 11, 22] and can be explained biologically. RAAS inhibitors reduce tone in the efferent arteriole of the glomerulus and thus impair renal autoregulation, maintenance of intraglomerular pressure and glomerular filtration. Without directly targeting the RAAS system, medication that lowers pre-glomerular pressure by reducing the effective circulating volume either by volume contraction (i.e. diuretics) is also likely to increase the risk of AKI [23]. Glomerular filtration pressure can also be very dependent on RAAS-mediated chronic tone in the efferent arteriole in people with CCF or renal artery stenosis, and those with disruptions to the renal microcirculation, [24] making them further sensitive to such effects.

Although this study demonstrated an increased risk of AKI with RAAS inhibition/ diuretic use, the absolute risk of AKI in the exposed group was low (2.5 and 1.7 events per 1000 person years in people taking RAAS inhibitors/ diuretics and those not, respectively). With a small or even moderate effect of RAAS inhibition/diuretic on the risk of AKI the NNT will therefore be high. In the subset of individuals with a creatinine >124umol/L included in the HOPE randomised trial [25], the NNT with Ramipril to prevent one myocardial infarction or one stroke over 4.5 years was 20.4 and 16.1 respectively. In the Acute Infarction Ramipril Efficacy study, a NNT of 17.5 over 6–15 months was found to prevent all-cause mortality in patients with heart failure following myocardial infarction [26]. There is therefore substantial evidence to support a beneficial impact of RAAS inhibition in the prevention of ischaemic heart disease, CCF, cerebrovascular disease, proteinuric CKD, ESKD, all-cause and cardiovascular mortality [25‐28]. As a result, RAAS inhibitors have become the second most commonly prescribed medication in General Practice (GP) in the UK, accounting for 6% of all prescriptions [29].

Given this, any public health or individual patient level recommendations about the initiation or temporary/ permanent discontinuation of RAAS inhibitors/ diuretics must consider both the potential risk of harm as well as benefit. The risk of AKI associated with RAAS inhibition must be weighed against the potential benefits for each individual, including delayed progression of CKD, [30] reduction in cardiovascular events (myocardial infarction, cerebrovascular events, resistant heart failure following myocardial infarction) and mortality (all-cause and cardiovascular-specific) [25, 26, 31]. The risk benefit ratio is likely to be influenced by level of comorbidity as well as concurrent medications [10]. In the general population of the UK, the benefits of RAAS inhibition are likely to outweigh the disadvantages in terms of AKI. However, when prescribing RAAS inhibitors and diuretics, physicians should be aware that the risk of AKI increases with number of additional hypertensive medications and diuretic use. The mechanism of the increased risk is likely to relate to actual or relative reduction in effective circulating volume; volume contraction and hypotension should therefore be avoided.

This study has several strengths and limitations. The size of the cohort and geographical spread of recruitment should enable the results to be generalisable to all primary care populations in England. The exposed and unexposed groups were well-matched, reducing bias, and the use of multiple statistical models with similar results increase confidence in the effect estimate. In addition, the results of this study support the majority of existing literature in this area. However, the diagnosis of AKI was based upon HES data/READ codes rather than serum biochemistry, and AKI rates will therefore be underestimated. The exclusion of individuals with an AKI event prior to the index date or within 6 weeks of initiation of RAAS inhibitors or diuretics was designed to minimise baseline differences in AKI risk between groups. However by doing so we may have inadvertently excluded those with a higher risk of AKI. Medication dose and dose changes were not explored in this analysis. This may be of particular relevance to RAAS inhibitors and diuretics where a dose-dependent impact on renal function may be seen, and physiological changes in measured function may therefore be interpreted as AKI events. Further alternative explanations for our findings include the possibility of increased testing of serum biochemistry in the exposed group resulting in higher detection of AKI. As with most observational studies there is the possibility of residual confounding in the Cox and propensity-matched analyses though this should have been less of an issue in the PERR analysis.

This study supports previous literature regarding an increased risk of AKI in individuals prescribed RAAS inhibitors and diuretics [10, 11, 22]. However, the results suggest the absolute increased risk in AKI is low and limited to patients prescribed concurrent antihypertensive medications. Further, the absolute risk of AKI associated with RAAS inhibitors and diuretics is low and likely to be significantly outweighed by the multitude of beneficial effects demonstrated from RAAS inhibition in the literature, [25, 30, 31] . Given that the impact of sick day rules on patients’ adherence to RAAS inhibitor/ diuretic treatment is unknown, interventional studies are required to inform practice that will optimise all patient outcomes, not just AKI-related ones.