PulmCrit: PPIs are safe and effective for GI prophylaxis... the end.

REVISE is the latest multicenter RCT on the use of PPIs for GI prophylaxis in critical illness. I would view this as the triquel following SUP-ICU and PEPTIC. In order to understand REVISE in context, let's briefly review SUP-ICU and PEPTIC trials, before discussing REVISE.

Part I: SUP-ICU (2018)

This was a large multicenter RCT evaluating the use of PPIs for GI prophylaxis. The findings included:

No difference in mortality (the primary endpoint).
PPIs reduced GI bleeding.
PPIs didn't increase the risk of C. difficile or pneumonia.

I found this to be a useful study because it eliminates fears that PPIs might increase the risk of pneumonia or C. difficile (blog here).

The concept that GI prophylaxis would improve mortality was delusional (for reasons discussed here), so I wasn't perturbed that PPIs didn't reduce mortality.

Part II: PEPTIC (2020)

This was a large multicenter RCT comparing PPIs vs. H2-blockers for GI prophylaxis.

The primary endpoint was mortality. There was no statistically significant difference in mortality, but there was a trend towards increased mortality in the PPI group. Analyzing the raw data shows this mortality difference has a p-value of 0.13:

People got really jazzed about this mortality trend, but it never really made sense that PPIs were killing people. Given my low pretest probability that three tablets of PPIs were assassinating people, I didn't get too excited about a p-value of 0.13. (For more discussion about this, I argued in a blog here that the mortality trend was noise).

If we can look past the mortality endpoint, there is some useful information in this trial:

First, PPIs were more effective at reducing bleeding as compared to H2-blockers. That's important. That's their job.

Also, there's a little more data here that PPIs don't increase the risk of pneumonia or C. difficile.

Part III: REVISE trial (2024)

This is a large RCT of PPI vs. placebo for GI prophylaxis among intubated patients.

PPIs reduced the rate of clinical GI bleeding, as they do in every trial (no big surprise). The authors also looked at clinically relevant GI bleeding, and PPIs improved that endpoint too. Once again, PPIs are good at their job.

This confirms the importance of stress ulcer prophylaxis in a modern critical care setting, where patients are generally receiving enteral nutrition.

The more important news here is…. drumroll…. there was a trend towards lower mortality in the PPI group. Now, I doubt that PPIs are reducing mortality in a statistically reproducible fashion. However, this finding is useful because it helps assuage concerns that PPIs are increasing mortality.

The study also confirms that PPIs don't increase the risk of pneumonia or C. difficile:

The authors performed a variety of subgroup analyses. PPIs didn't increase mortality among any subgroup of patients. There were also no statistically significant differences in the mortality endpoint between any of the various subgroups.

Meta-analysis including all trials

Wang et. al. performed a meta-analysis including these RCTs and others. They found that PPIs had no effect on mortality:

That's a scientifically sound and definitive result.

And it's totally neutral.

But it's also totally boring.

To spice things up a little, the authors combined two trials and risk-stratified patients on the basis of disease severity. When you do this, there is still… no statistically significant difference in mortality:

The authors of this trial are worried about this data showing a trend towards increased mortality when PPIs are given to sicker patients within these two studies.

Once again, I doubt that this mortality trend is real. This mortality trend is inconsistent with the findings of the REVISE trial – the newest, largest, and most robust RCT of PPI-vs-placebo. The REVISE trial didn't find any increase in mortality among patients with high disease severity (or any subgroup for that matter).

It feels like we are chasing an endless, morphing series of statistical trends. The PEPTIC trial suggested that PPIs killed people. The REVISE trial and the Wang meta-analysis completely disprove that assertion. Full stop. This should be a moment to pause, introspect, and realize that we've been getting really jazzed about statistical noise.

But nope. There's always another subgroup to chase after!

In all seriousness, this illustrates the perpetual struggle of establishing medication safety. It's impossible to ever prove that any medication is safe. There is always some non-zero probability that the medication is dangerous. Attempting to fully prove that any medication causes no serious side effects is a perpetual, Sisyphean struggle. You can sometimes disprove one complication, but meanwhile you find evidence of another potential harm. Repeat ad infinitum.

Until the next trial comes out, I don't believe that we should base our clinical practice upon a statistically insignificant finding derived from pooling two subgroups together in a meta-analysis.

my recommendation: use PPIs for GI prophylaxis

I favor the use of PPIs for GI prophylaxis for the following reasons:

PPIs are more effective than H2-receptor antagonists (based on the PEPTIC trial).
PPIs don't require dose adjustment for renal failure, which make it easy to systematically ensure that patients are receiving the proper dose of medication.
H2-receptor antagonists may be associated with delirium (some more on this here).
PPIs have now been investigated in numerous high-quality RCTs and the data is overall very reassuring in terms of an excellent safety profile.

So my preference would be just to use 40 mg/day of a PPI for any patient requiring GI prophylaxis.

There's no reason to waste brain power on this. We can pontificate about other, more interesting things on rounds. Or talk about the latest movie. Maybe they're going to make a fourth Matrix movie. I mean, literally any topic would probably more interesting than stress ulcer prophylaxis.

PPIs are highly effective in reducing the rate of stress ulceration (more effective than H2-receptor antagonists).
The REVISE trial shows that mortality was 2% lower in patients treated with PPIs. Overall, there is no convincing theory or evidence that PPIs affect mortality.
PPIs do not increase the risk of C. difficile infection or pneumonia.
People keep on trying to make this topic exciting, but it's not. The boring truth is that PPIs are highly effective and have an outstanding safety profile.