Prilosec, Chronic Kidney Disease, and Fuzzy Numbers
This week the major media outlets published results of a JAMA Internal Medicine study linking proton pump inhibitor (PPI, medicines like Prilosec) use to chronic kidney disease (CKD). Not long ago a flawed study linked PPI use to heart attacks, utilizing exaggerated relative risk numbers and failing to establish a causal relationship between the drug and the outcome. Now a new study shows that people who use PPIs have a 20-50% higher risk of chronic kidney disease. Therefore we must ask: does PPI use really cause kidney disease, and if so, how serious is it?
Without a doubt the use of PPIs is excessive among the American population. People take these drugs indefinitely for symptoms that may well be resolved through diet or more mild medicines. As this article’s authors state: More than 15 million people use PPIs at a cost of over $10 billion, and “up to 70% of these prescriptions are without indication”. That may be true, but do these drugs cause kidney disease, as this article suggests, and if so, is that kidney disease dangerous?
This study used two groups of people to reach its conclusions. In the main (but smaller) group studied prospectively, an excess of 4 people developed CKD in a year out of 1000 PPI users compared to 1000 non-users. In the other group, the excess was 2 people out of 1000. To split the difference, we can say that there were 3 excess cases of CKD out of 1000 PPI users per year. (see the BRCT above) In other words, in a year, 99.7% people using PPIs did not develop CKD. Assuming the risk is cumulative over years (which we do not know), in 10 years 3% of long term users could be prone to CKD, a number much smaller than the 20-50% risk reduction touted by the press and the study's authors.
There are two other issues at play here. First, can we presume that there is a cause-effect relationship between PPIs and CKD? The answer is, this study shows no causality. Among PPI users more were overweight and had higher blood pressure. The researchers tried to control for this, but it is not certain they could. Also, it is possible that people with CKD were more prone to acid reflux and thus used more PPI drugs, thus reversing the cause-effect relationship. Given that so few excess PPI users developed CKD compared to non-users, any of these, or other, explanations would explain the difference.
But an even more important question is whether we should even care if more people develop CKD with PPI use. Surprisingly, the answer is likely no. Two recent studies in JAMA (2015) show that the vast majority of people with CKD do not go on to develop any serious kidney problems. In fact, by current criteria for CKD, 50% of people over the age of 70 have CKD, while only 0.7/1000 of them go on to develop serious kidney issues. Therefore, even if 3/1000 excess people using PPIs develop CKD, only .7/1000 of those 3 will likely develop any kidney problems. The number of people at risk is so small as to not even be clinically meaningful.
It is very important that doctors and patients have access to absolute risk/benefit numbers and a clinically realistic understanding of how significant the side effects may be when assessing the use of PPI drugs. Articles like the one published by JAMA Internal Medicine can help uncover potential links between drugs and side effects as a preliminary exercise. But this article in no way determines cause and effect, while the authors’ use of exaggerated relative risk (20-50% increased risk of kidney disease among PPI users) is clearly deceptive, especially when we realize just how small the absolute risk is, and just how few people with chronic kidney disease develop any serious medical consequences. What is very disturbing is that the national news media again simply echoed the seemingly impressive results of this study without asking any tough questions about their significance. What is even more disturbing is that an accompanying editorial to this article listed many potential side effects from PPIs using only very exaggerated relative risk numbers (in all the cases they cite, the absolute number of people harmed with PPIs is very small), and claiming that there are cause-effect relationships when few have been proven.
In the end, we have a lot of reasons to doubt the utility of PPIs and to question the safety of long term use, but we need not rely on nebulous data to frighten patients into making a decision that requires much more sensible and fact-based shared decision making.