Sometimes doing hospital nephrology rounds is like being a medical Cassandra; we know when and why a patient will develop acute kidney injury (AKI), but are unable to avoid it. Contrast nephropathy, cross clamping the aorta, and CABG all run a high risk of causing AKI, yet despite our foreknowledge, to a large degree we are unable to prevent it. What is even more frustrating is we have not even nailed down the most basic elements of preventative care, like should ACE inhibitor (ACEi) and ARBs be stopped prior to surgery?

Recently published by AJKD, Yacoub et al presents a meta-analysis that examines whether RAS blockade is beneficial or detrimental in cardiac surgery. They looked at outcomes of AKI and death. After careful vetting of the corpus of medical literature, they found 29 studies for analysis. None of these studies were prospective trials. None were RCT pitting RAS blockade versus a placebo or usual care. All but four of the studies were retrospective.

For the outcome, they examined 23 studies representing 69,027 patients, of which 7000 developed AKI. ACEi/ARB were harmful with an OR of 1.17. By calculation, this is a number needed to harm (NNH) of 32.9; you would need to let 33 patients go for CT surgery on an ACEi/ARB to see one additional episode of AKI. There were 18 evaluable studies for mortality representing 54,418 patients of which 1,792 died. ACEi/ARB were harmful with an OR of 1.20. This is a NNH of 199.6; you would need to let 200 patients go for CT surgery on an ACEi/ARB to see one additional death.

The data on AKI is a bit hard to interpret because the individual studies had such widely variant definitions of acute kidney injury. The 23 studies used 17 unique definitions of AKI. Thankfully, the mortality data is a bit more homogeneous with only 3 definitions of mortality: hospital mortality, 30-day mortality, and 90-day mortality. Additionally, the authors were unable to reconcile bias by indication. Not surprisingly, diabetics and patients with hypertension were more likely to be on RAS blockade. They describe individual studies showing harm from RAS blockade that disappeared with multivariate analysis:

Rady and Ryan studied the association of use of ACEi on post–cardiac surgery mortality. Univariate analyses showed that ACEi use was associated with increased odds of death. However, multivariate analysis failed to show this association. This subtlety was not possible with the meta-analysis. Similarly, some studies used propensity scoring to adjust for increased illness in patients receiving RAS blockade, but this statistical crutch was not available in the meta-analysis.

In 2009, Miceli et al analyzed more than 10,000 patients and found that preoperative use of ACEi was associated with increased odds of mortality (OR, 2.00; 95% CI, 1.17-3.42). Although patients on ACEi/ARB therapy were sicker than those not on RAS-blockade, they found similar results after performing propensity score-matched group analyses to reduce selection bias and potential confounding.

Yacoub et al do a nice job discussing the limitations of their analysis, but despite that do not call for additional research to examine this question in an appropriately powered, placebo controlled, double blind study. Instead, the authors embrace their conclusions without reservation:

“This meta-analysis shows that preoperative RAS-blocker use is associated with an increased incidence of AKI and mortality after cardiac surgery. We propose that RAS blockers be withheld at least 48 hours preoperatively in patients undergoing coronary artery bypass grafting and/or valvular surgery.”

One cannot read these strong conclusions and recommendations without thinking of the last time physicians prematurely recommended changing blood pressure medications prior to surgery, by adding preoperative β-blockade. In a recent meta-analysis in Heart, Bouri et al estimate that clinical practice guidelines based on fraudulent research were resulting in 10,000 extra deaths a year in the UK alone from over-use of β-blockers. This sobering statistic should caution anyone making sweeping recommendations based on retrospective data.

Joel Topf, MD
eAJKD Advisory Board member

To view the article abstract or full-text (subscription required), please visit AJKD.org.