NSAIDs Tied to HF Hospitalization in Patients With Diabetes

The overall message is to be cautious about NSAIDs and limit their use as much as possible, says Craig Beavers.

NSAIDs Tied to HF Hospitalization in Patients With Diabetes

In patients with diabetes and no history of heart failure (HF), the use of nonsteroidal anti-inflammatory drugs (NSAIDs) may increase the risk of hospitalization for incident HF, according to results of a Danish registry study.

Individuals at greatest risk had not previously used NSAIDs, were age 80 or older, and had elevated hemoglobin (Hb) A1c levels, researchers found.

The association is not new, but it adds another group of patients to the list of those who may be vulnerable on NSAIDs. In a 2013 meta-analysis from the Coxib and Traditional NSAID Trialists’ (CNT) collaboration, selective COX-2 inhibitors, diclofenac, and ibuprofen were all associated with a near doubling of HF hospitalization risk. Additionally, a large European analysis linked current use of ibuprofen, naproxen, or diclofenac to a 20% greater risk of being hospitalized for HF.

When the Danish investigators, led by Anders Holt, MD (Copenhagen University Hospital–Herlev and Gentofte, Hellerup, Denmark), performed explanatory subgroup analyses to understand whether the increased risk seen in their study could be attributable to worsening of subclinical structural HF, or a transient impact on the renal system that caused fluid buildup, they saw that patients with normal HbA1c levels had no increased HF hospitalization risk.

“This suggests that the combined effects of hyperglycemia and NSAID exposure may lead to endothelial dysfunction, resulting in HF,” Holt and colleagues write in the Journal of the American College of Cardiology. However, individuals with decreased estimated glomerular filtration rate (eGFR) also were not found to have increased risk of HF with NSAID use.

“If the risk of being hospitalized with first-time HF following short-term NSAID exposure is substantial in subgroups with elevated HbA1c levels, at least more so than in subgroups with decreased kidney function, it seems that a likely pathophysiological explanation could be attributed to a ‘demasking’ of subclinical HF caused by type 2 diabetes mellitus,” they add.

Commenting on the findings for TCTMD, Craig Beavers, PharmD (University of Kentucky, Lexington), said while it is possible that NSAIDs increase HF risk in diabetic patients with no known HF, the demasking scenario makes sense.

“The NSAID use could have put some patients who had heart failure but didn’t know it into an exacerbation, and then they were diagnosed with heart failure at that particular juncture,” he explained.

The overall message is to be cautious about NSAID use and limit it as much as possible, said Beavers. Although NSAIDs are widely available over-the-counter (OTC), patients in the study were taking prescription-strength NSAIDs, primarily ibuprofen. But the results may be generalizable to non-prescription-takers too, since many patients exceed recommended OTC doses and likely don’t tell their clinician.

“Potentially there are some NSAIDs that may be a little bit more appropriate from a cardiovascular perspective. In theory, meloxicam may [confer] lower risk, but they all carry some risk of cardiovascular effects. It’s important to be judicious, be cautious, and look for alternative agents, if possible,” he advised.

Similarly, in an accompanying editorial, Hassan Khan, MD, PhD (Norton Healthcare, Louisville, KY), and Setor K. Kunutsor, MD, PhD (Leicester General Hospital, United Kingdom), say patients “should be treated only for the shortest duration of time and with the lowest dose necessary to provide relief.”

More Than ‘Just’ Fluid Overload

For the analysis, Holt and colleagues studied 331,189 patients (median age 62 years; 44% female) who received a diagnosis of type II diabetes between 1998 and 2021. Approximately one-third of the group had hypertension, 8.3% had atrial fibrillation, 20% were on a beta-blocker, 41.6% a renin-angiotensin-aldosterone-system (RAAS) inhibitor, 43% a statin, and 25% an antiplatelet.

Within 1 year of being enrolled, 16% patients claimed at least one prescription for ibuprofen, celecoxib, diclofenac, or naproxen. At a median follow-up of 5.9 years, there were 23,308 first HF hospitalizations. Among those hospitalized, nearly 40% were women and the median age was 76 years.

Use of NSAIDs was associated with heightened risk of HF hospitalization (OR 1.43; 95% CI 1.27-1.63), with similar results seen across patients who were exposed to NSAIDs for periods ranging from 14 to 42 days. No increased risk was seen in patients younger than age 65, but an increased risk was associated with NSAIDs used concomitantly with RAAS inhibitors and diuretics, the so-called “triple whammy” (OR 1.41; 95% CI 1.16-1.71).

Another important finding was that at 5 years after discharge for first-time HF hospitalization, the all-cause mortality rates were comparable in those who did and did not have an NSAID history, “suggesting that NSAID-associated HF hospitalizations were more than ‘just’ fluid overload,” Holt and colleagues conclude.

According to Khan and Kunutsor, the study is limited by its inability to explore for a class effect due to too few events for users of celecoxib and naproxen.

“This is clinically relevant given that the various NSAIDs reversibly inhibit the enzyme cyclooxygenase (COX) in both of its isoforms, COX-1 and COX-2; however, their COX selectivity and associated cardiovascular risk vary. Evidence suggests that naproxen, a nonselective NSAID, is associated with the lowest risk of cardiovascular events, whereas diclofenac is associated with the highest cardiovascular risk among nonselective NSAIDs,” they write.

The study also doesn’t allow for a dose-response analysis, the editorialists add.

“Large registry data are great because you have a large population and typically a good ability to collect data points and information, but there's limitations in terms of not having data on everything that could be contributing  to . . . or driving this association,” said Beavers. “This should be considered hypothesis-generating to some degree unless or until there’s a really well-designed study that manages all the possible confounders.

  • Holt, Khan, Kunutsor, and Beavers report no relevant conflicts of interest.