705-719-3996 705-719-3998
It came to my attention recently, that there was a publication, in a journal called Gut, in the January, 2022 issue, entitled "Proton pump inhibitors and risk of gastric cancer: Population-based cohort study".
I am providing this detailed analysis, as well as a preliminary more simplistic summary, to supplement the paper already on my website, (under "health information", and under "helpful documents", called "acid reflux"), in which there are several pages of reassurance about the long-term safety of PPIs (proton pump inhibitors, the most common drugs used for control of acid reflux, bleeding risk from ulcers and blood thinners, and treatment for Barrett's esophagus to prevent esophageal cancer).
Summary, (without the excruciating scientific details)
This paper shows a statistical risk for gastric cancer in patients on longstanding PPI therapy. In my opinion, this is a minor statistical risk of no clinical significance. The study benefits from huge numbers, but in fact the statistical power related to those numbers allows the finding of a very small absolute risk seemingly related to being on these pills. Life is full of risk, and if the PPI is being taken for the right reason, and was recommended, or reviewed, by a specialist, then the risks of stopping the PPI far exceed this very small risk for gastric cancer.
Detailed analysis
This paper is written by senior colleagues in gastroenterology and in statistics, at McGill University, and makes the following comments:
1) all previous studies were limited by important shortcomings, that could lead to an exaggeration of the reported risk between the use of PPIs and gastric cancer
2) the use of PPIs is associated with a 45% increased risk of gastric cancer.
3) In light of the of use of PPIs, physicians should regularly reassess the necessity of ongoing treatment
Obviously, this publication, and the results themselves, may be concerning to a large number of patients who are on PPI therapy, for a variety of reasons, and who have been previously reassured, by me and many other higher-ranking experts in the gastroenterology field, about the long-term safety of these drugs. This paper certainly is a high-quality publication, in a high-quality journal, by a high-quality group of experts, at a leading university in Canada.
The paper uses a large research database from the United Kingdom. The methods of this study appear to be beyond criticism. A large number of patient records were reviewed, starting 1 year after a starting point of January 19, 1990, upto an endpoint of April 2018.
The numbers of records were huge, which gives a statistically positive result that could be considered of questionable clinical importance. Small studies can miss clinically important results due to inherent weakness of statistics in those cases (type II error), and very large studies can find a statistically positive result that could be, as stated previously, of questionable clinical significance (type I error).
In this situation, the main result is that, starting with nearly 1 million patients in each group, at 10 years, there were still over 150,000 patients in each group available for analysis, and of those patients, the risk of stomach cancer in the PPI group was 45% greater than the risk of stomach cancer in the H2 blocker group - this is called the relative risk, which is very different from the absolute risk.
The true incidence of stomach cancer in the PPI group appears to be approximately 0.2% over 10 years, which is approximately 1 in 500 patients, over 10 years, which is approximately 1 in 5000 patients per year (although the annual risk might be lower initially and slowly increase with prolonged duration of therapy). Another way to look at the incidence of cancer is that it is in the range of 23-30 patients, per 100,000 patient-years (100,000 patient-years is 100,000 patients followed for 1 year, or 10,000 patients followed for 10 years), with the lower numbers in the range, being for H2 blockers, and the higher number in the range being for PPIs. 25 patients per 100,000 patient-years is again approximately 1 in 4000 patients per year.
In the conclusion of the paper, they make the comment that "the absolute risk remains low".
There is a statistical measure which relates the absolute risk reduction (ARR), or increase (ARI), to another number called the NNT (number needed to treat), or for bad outcomes, as opposed to advantages, the NNH ie "number needed to harm". Essentially 1 divided by the ARR, gives the NNT (number needed to treat in order for 1 person to receive benefit) or 1 divided by the ARI, gives the NNH, the number needed to treat in order to produce 1 negative outcome. This study suggests that the NNH is approximately 2100 at 5 years and 1200 at 10 years.
Overall, I find this paper has little if any clinical significance, except for generating anxiety in patients who have to stay on a PPI. We continue to recommend that patients on PPIs should be reviewed, at least every 2-5 years, and if they are not on a PPI for Barrett's esophagus, (which requires continuous PPI therapy to prevent esophageal cancer), most patients should be taken off the PPI every year or 2 to see if they still need it. I am a strong believer in aggressive attempts at dietary and lifestyle management for reflux disease requiring PPIs. However many patients on PPIs are taking them to prevent significant bleeding, often related to blood thinners, older age, heart disease, and other complications or conditions, which make the risks of stopping the PPI much more significant than this very small risk of stomach cancer.
There will be increasing interest in this paper, and I suspect there will be increasing critical analysis from other experts in the field.