Scientific deep-dive

GLP-1s, Pneumonia and Sepsis: A Better-Built Cohort

Among 331,863 matched adults with type 2 diabetes, GLP-1 users had 40% less new pneumonia over a year than DPP-4 inhibitor users. The design is about as good as observational research gets, which is the point.

By Nora Bissett · Pricing Editor
Editorially reviewed. Not medical advice · How we verify contentLast reviewed
6 min read·1 citation

Among 331,863 matched adults with type 2 diabetes, those starting a GLP-1 drug had a 40% lower rate of new pneumonia and a similar reduction in severe sepsis over a year, compared with those starting a DPP-4 inhibitor.[1] That is a large effect from observational data, but the way this study was built makes it harder to dismiss than most.

What was found

Using a global federated database of electronic medical records, researchers compared adults with type 2 diabetes newly prescribed one drug class against those newly prescribed DPP-4 inhibitors, matched one to one on measured characteristics and followed for twelve months.[1]

Incident pneumonia and severe sepsis against DPP-4 inhibitors, after 1:1 propensity score matching. Twelve months of follow-up.
ComparisonPatientsPneumoniaSevere sepsis
GLP-1 receptor agonists vs DPP-4i331,863HR 0.60 (0.58–0.62)HR 0.61 (0.59–0.63)
SGLT2 inhibitors vs DPP-4i352,687HR 0.75 (0.73–0.78)HR 0.75 (0.73–0.77)

Both newer classes came out ahead of the older one, on both outcomes, with narrow confidence intervals. The consistency across two different drug classes and two different infections is part of what makes the pattern interesting.

Why this design deserves more trust than most

A comparison group can manufacture a result. In the cancer cohorts, one study reported 91% lower mortality against insulin users and nothing at all against a modern comparator. This study is built to avoid exactly that.

  • Active comparator. The comparison is another diabetes drug, not no treatment. Both groups have diabetes and both are being actively managed, which removes a great deal of the difference that sinks weaker studies.
  • New user design. Everyone is starting their drug, so the comparison is not between new patients and people who have already tolerated a drug for years.
  • Propensity score matching, one to one, on measured characteristics.
  • Very large samples, which is why the confidence intervals are as tight as they are.
This is roughly the best an observational study can be built. It still cannot rule out that clinicians choose newer, more expensive drugs for patients who are healthier in ways records do not capture (frailty, functional status, whether someone attends appointments). Those unmeasured differences are exactly what would produce a lower infection rate.

What the number does not say

A 40% reduction sounds like a reason to take the drug for infection protection. It is not, for three separate reasons.

  1. The comparator is another drug, not nothing. This says GLP-1 users had fewer infections than DPP-4 inhibitor users. It does not establish what would happen against no treatment at all.
  2. These are relative figures. Pneumonia and severe sepsis are uncommon over twelve months in this population, so a 40% relative reduction is a small absolute one.
  3. Twelve months is short for an outcome shaped by long-term glycemic control and body weight, both of which move slowly.

The authors’ own conclusion is that further research and focused randomized trials are warranted, which is the right place to leave it.[1]

Is there a plausible mechanism?

Several, none proven. Better glycemic control reduces infection risk in diabetes and is well established. Obesity impairs respiratory mechanics and immune function, and these drugs reduce it. GLP-1 receptors appear in lung tissue and there is preclinical work on anti-inflammatory effects. A protective association has also been reported for asthma, in a Taiwanese cohort of people with type 2 diabetes, though it faded as the asthma grew more severe.

Plausibility counts for something, but it is also what makes this kind of result risky. A finding with an obvious mechanism is easier to believe than one without, so it needs the same scrutiny, not less.

Frequently Asked Questions

In a study of 331,863 matched adults with type 2 diabetes, those starting a GLP-1 had a 40% lower rate of new pneumonia over twelve months than those starting a DPP-4 inhibitor, with a similar reduction in severe sepsis. It is observational, and the authors call for randomized trials.
No. No GLP-1 is approved for infection prevention, the comparison was against another diabetes drug rather than no treatment, and pneumonia and sepsis are uncommon over a year, so a 40% relative reduction is a small absolute difference.
Because of how it was built: an active comparator, a new-user design, one-to-one propensity matching and very large samples. That combination removes much of the bias that ruins weaker observational work. It cannot remove unmeasured differences in how sick patients are.
SGLT2 inhibitors did, at a smaller magnitude, with hazard ratios of 0.75 for both pneumonia and severe sepsis against DPP-4 inhibitors. Both newer classes came out ahead of the older one on both outcomes.
Several plausible routes: better glycemic control lowers infection risk in diabetes, obesity impairs respiratory mechanics and immune function, and GLP-1 receptors are present in lung tissue. None of these is established as the explanation, and a plausible mechanism makes a finding easier to believe rather than more likely to be true.

References

  1. 1.Henney AE, Riley DR, Hydes TJ, et al. Comparative estimate of glucose-lowering therapies on risk of incident pneumonia and severe sepsis: an analysis of real-world cohort data Thorax. 2024. PMID: 39645259.

GLP-1s and Daytime Sleepiness

A cohort found more diagnoses of excessive daytime sleepiness, at a 21% relative increase, which is 0.4% in absolute terms at one year. Four other sleep outcomes showed nothing.

5 min read

Stopping Ozempic or Wegovy: Side Effects, Weight Regain and Keeping It Off

Stopping Ozempic or Wegovy causes no withdrawal syndrome on either label. Appetite returns as semaglutide clears over about five weeks, and a 2026 BMJ meta-analysis found regain of about 0.8 kg a month. What helps keep weight off, and what women should know.

10 min read

Eating Too Little on a GLP-1: What the Deficiency Data Show

Across 480,825 adults, more than 60% took in less calcium and iron than their estimated requirements, vitamin D deficiency reached 13.6% at a year, and ferritin ran 26–30% below an active comparator.

7 min read

GLP-1s and Your Other Medications: What Actually Interacts

Digoxin, warfarin and oral contraceptives came through interaction studies needing no dose adjustment. The warfarin problem is real but dietary, and taking other pills with oral semaglutide cuts its absorption by 34%.

7 min read

GLP-1 Drugs and Opioid Use Disorder: What the Records Data Shows

Across 503,747 people with opioid use disorder, a GLP-1 or GIP prescription carried an adjusted overdose rate ratio of 0.60. In methadone-treated patients it was associated with a higher rate of remission. No randomized trial exists.

6 min read

Elamipretide (SS-31): The Phase 3 Trial That Missed Both Endpoints

Elamipretide has what almost no research-use peptide has: a completed pivotal phase 3 trial in 218 people. It missed both primary endpoints, with a six-minute walk difference of -3.2 meters and a p value of 0.69.

5 min read

Where to get GLP-1 online, safely: sellers our editors have checked

These are telehealth sellers our editors have checked. For each one we hold a price, the form the drug comes in, and the states it reaches.

No insurance needed · vetted by our editors

Some of the links on this page earn us money. If you sign up with a provider after following one, that provider may pay GLP Watchdog a commission. Learn more

7.4

Wellorithm

Compounded semaglutide at a flat $247 every 28 days, any dose

8.5

Gala

Moving between compounded and brand without changing seller

7.9

Luvo Health

Moving between compounded and brand without changing seller

About this analysis

GLP Watchdog publishes independently. Each seller in our dataset is checked against that seller's own website, and the dataset is updated continuously. Read our full methodology →