Which NSAID Is Safest? A Comparison of Health Risks

No single NSAID earns the title of “safest” across the board, because each one trades risk in one organ system for relative safety in another. Celecoxib tends to be gentlest on the stomach, naproxen has long been considered the friendliest to the heart, and ibuprofen at low over-the-counter doses sits somewhere in the middle on most measures. The real question is which risks matter most for your body, your medications, and your medical history. Sorting through the evidence drug by drug and organ by organ is the only honest way to compare them.

The Heart Risk Landscape

Cardiovascular safety has driven the biggest NSAID debates since the withdrawal of rofecoxib (Vioxx) in 2004. The conventional wisdom that followed was simple: COX-2-selective drugs are bad for the heart, and naproxen is good. The evidence since then has been more nuanced. Naproxen’s low COX-2 selectivity does correlate with a lower cardiovascular risk than most other NSAIDs, and reviews continue to support that association.¹ The European Society of Cardiology has endorsed naproxen at 500 mg per day or less as the NSAID with the least harmful cardiovascular profile, with ibuprofen at 1,200 mg per day or less considered a safe alternative as well.²

But the largest randomized trial to directly compare the big three head-to-head, known as the PRECISION trial, muddied that clean narrative. It enrolled over 24,000 arthritis patients at elevated cardiovascular risk and found that celecoxib was statistically non-inferior to both naproxen and ibuprofen for major cardiovascular events. In the on-treatment analysis, the rate of primary cardiovascular events was roughly 1.7% with celecoxib, 1.8% with naproxen, and 1.9% with ibuprofen.³ A secondary analysis of that same trial found that when all major toxicity was combined, including cardiovascular, gastrointestinal, and kidney events, celecoxib users fared best. Compared with celecoxib, naproxen users had about a 20% higher rate of major toxicity and ibuprofen users about a 38% higher rate.⁴

A separate Bayesian meta-analysis using individual patient data from real-world databases found that all common NSAIDs raise the risk of heart attack even during the first week of use. The odds ratios for a one-to-seven-day course were roughly 1.24 for celecoxib, 1.48 for ibuprofen, 1.50 for diclofenac, and 1.53 for naproxen.⁵ Celecoxib did come in with the lowest point estimate, though the confidence intervals overlapped. Higher doses pushed the risk up further. One consistent finding: the risk did not appear to keep climbing beyond the first month, suggesting that the danger is front-loaded rather than cumulative.

Heart failure is a different story from heart attacks. A meta-analysis on heart failure exacerbation found that conventional NSAIDs as a class raised the risk by about 39%. Celecoxib showed a similar trend, while rofecoxib, before its withdrawal, roughly doubled the risk.⁶ If you already have heart failure or are at high risk for it, every NSAID carries meaningful danger. In patients with hypertension, a population-based study found no statistically significant difference in major cardiovascular events between COX-2-selective and nonselective NSAIDs overall.⁷

Gastrointestinal Bleeding and Ulcers

The stomach and upper GI tract have always been the most visible battleground for NSAID side effects. Here, the drugs vary enormously. A systematic review pooling observational data ranked celecoxib as carrying the lowest risk of upper GI complications, with a relative risk of about 1.45 compared with non-users. Ibuprofen came next at roughly 1.84. From there the numbers climb: diclofenac around 3.3, meloxicam about 3.5, naproxen around 4.1, and piroxicam about 7.4. Ketorolac, used mainly for short-term acute pain, had a relative risk above 11.⁸ High doses roughly doubled or tripled the risk compared with low doses across the board.

A more recent meta-analysis focused specifically on GI bleeding confirmed the same hierarchy. Celecoxib had the lowest odds of GI bleeding among all NSAIDs studied, while ibuprofen had the lowest significant risk among the non-selective drugs. Ketorolac again topped the list as the most dangerous for bleeding.⁹ This creates an uncomfortable tradeoff: naproxen, often favored for cardiovascular safety, is one of the harder drugs on the stomach. In patients who have already had an upper GI bleed and also carry cardiovascular risk, a randomized trial found that celecoxib plus a proton pump inhibitor cut recurrent bleeding to about 5.6% over 18 months, compared with 12.3% for naproxen plus a proton pump inhibitor.¹⁰

The Lower GI Problem That PPIs Cannot Fix

Most discussions of NSAID gut damage focus on the stomach and duodenum, where acid plays a role and where proton pump inhibitors can help. But NSAIDs also injure the small intestine further downstream, and that damage is a different beast. No drug has been proven to prevent or treat NSAID-induced enteropathy.¹¹ In fact, PPIs may actually make lower GI injury worse. An animal study found that PPIs shifted gut bacteria in ways that aggravated NSAID-induced small intestinal damage.¹² This is a practical concern: patients on long-term NSAIDs who also take a PPI for stomach protection may assume their entire GI tract is covered. It is not.

Liver Damage

Across the class, serious liver injury from NSAIDs is rare, with the incidence of liver-related hospitalizations estimated between 3 and 23 per 100,000 patients.¹³ Diclofenac, however, stands apart. In a large clinical trial of over 17,000 arthritis patients, about 3% developed liver enzyme elevations more than three times the upper limit of normal, and half a percent had elevations above ten times normal. These changes typically appeared within the first four to six months. Liver-related hospitalizations were uncommon, and no patients in the study progressed to liver failure or needed a transplant.¹⁴ The underlying mechanism appears to involve the formation of reactive metabolites that can damage liver cells, sometimes triggering an immune response with delayed onset.¹⁵

The practical takeaway: if you are on diclofenac for more than a few weeks, periodic liver enzyme monitoring is standard practice. Other NSAIDs can occasionally cause liver enzyme bumps, but diclofenac is the only commonly used one where routine monitoring is considered worthwhile.

Kidney Effects

All NSAIDs can harm the kidneys, and there is no clearly “safe” member of the class for renal risk. The mechanism is straightforward: kidneys rely on prostaglandins to keep blood vessels dilated and blood flowing to the organ. NSAIDs block prostaglandin production, which reduces that blood flow and can cause acute kidney injury. This is especially dangerous in people who are already dehydrated, elderly, or taking other drugs that affect kidney perfusion. The most common form of NSAID-related kidney injury is this blood-flow-mediated type, though a less common allergic-type inflammation of the kidney tissue can also occur.¹⁶

The combination of an NSAID with a diuretic and an ACE inhibitor or an angiotensin receptor blocker is known informally as the “triple whammy.” Each drug on its own affects kidney blood flow or fluid balance in a different way, and together they can push a vulnerable patient into acute kidney injury.¹⁷ This combination is common because the same patients who need blood pressure drugs often reach for an NSAID for joint pain. If you take medications for hypertension, any NSAID use needs to be brief and monitored.¹⁸

The Ibuprofen-Aspirin Interaction

Many people who take low-dose aspirin for heart protection also reach for ibuprofen when they have pain. That combination creates a specific and underappreciated problem. Ibuprofen competes with aspirin for access to the same enzyme in platelets. If ibuprofen arrives first, it physically blocks aspirin from permanently disabling the platelet, which is the whole point of cardioprotective aspirin. A pharmacokinetic study found that taking ibuprofen as little as one hour after aspirin markedly reduced aspirin’s antiplatelet effect, and taking ibuprofen up to 12 hours before aspirin essentially wiped it out entirely.¹⁹ A key study in the New England Journal of Medicine confirmed that ibuprofen blocks aspirin’s antiplatelet action when given before it, while diclofenac and acetaminophen did not interfere.²⁰

The practical fix, if you must take both, is to take aspirin first and wait at least two hours before taking ibuprofen. But be aware that naproxen, celecoxib, and several other NSAIDs have also been shown to interfere with aspirin’s antiplatelet activity in laboratory settings.²¹ Among common NSAIDs, diclofenac was one of the few that did not appear to blunt aspirin’s effect, though its own cardiovascular risk profile offsets that advantage.

Older Adults Face Amplified Risks

Roughly 20 to 40% of older adults experience adverse drug reactions in a given year, driven by multiple medications, more chronic conditions, and age-related shifts in how the body processes drugs.²² NSAIDs hit especially hard in this population because every major risk category, including GI bleeding, kidney injury, cardiovascular events, and central nervous system side effects, is more likely with age.²³ Some specific drugs warrant dose reductions in otherwise healthy older adults, including naproxen, ketoprofen, and salicylates, because their clearance slows with age. When kidney function is already reduced, additional drugs including indomethacin, sulindac, and diflunisal may also need lower doses.²⁴

For older adults who need anti-inflammatory relief, many clinicians default to celecoxib at the lowest effective dose, often paired with a PPI, because of its relatively gentle GI profile. Topical formulations offer another way to reduce systemic exposure, which matters more as kidney and liver reserve decline.

Pregnancy and NSAIDs

NSAIDs are generally considered off-limits in the third trimester because they can cause premature closure of the ductus arteriosus, a blood vessel in the fetus that normally stays open until birth. A meta-analysis found that the risk of ductal closure was roughly 15-fold higher in pregnancies exposed to NSAIDs compared with placebo or unexposed controls.²⁵ Most of the data came from indomethacin, which was widely used to delay preterm labor, but case reports document the same problem with diclofenac and other NSAIDs.²⁶ The FDA now recommends avoiding all NSAIDs after 20 weeks of gestation. Before 20 weeks, short courses are generally considered acceptable, though the evidence in early pregnancy is less clear-cut.

NSAID-Exacerbated Respiratory Disease

About one in ten people with both asthma and chronic sinus disease with nasal polyps has a condition known as NSAID-exacerbated respiratory disease, or N-ERD.²⁷ In these patients, taking any NSAID that inhibits COX-1 can trigger severe nasal congestion, bronchospasm, and sometimes full anaphylactic-type reactions.²⁸ The reaction is not a true allergy in the immunological sense, meaning standard allergy tests will not detect it. COX-2-selective drugs like celecoxib are generally tolerated in these patients because the reaction is driven by COX-1 inhibition, but this should always be confirmed under medical supervision before assuming safety.

Topical NSAIDs as a Lower-Risk Alternative

One way to sidestep much of the systemic risk is to use a topical formulation. Topical diclofenac, for instance, delivers the drug directly to the affected joint and produces far less drug in the bloodstream. A pooled safety analysis comparing topical and oral diclofenac found significantly fewer GI side effects with the topical version (about 25% vs. 39% reporting any GI complaint) and a trend toward fewer cardiovascular events as well. Oral diclofenac also caused greater increases in liver enzymes and creatinine and greater drops in kidney function markers.²⁹ Topical formulations work best for superficial joints like knees and hands. They are less effective for deep joints like hips or for widespread pain, which limits their usefulness.

How Your Genes Can Shift the Risk

Not everyone metabolizes NSAIDs at the same rate. The liver enzyme CYP2C9 handles the breakdown of several common NSAIDs, including ibuprofen, diclofenac, and celecoxib. Genetic variations in the gene for this enzyme can slow metabolism, leading to higher drug levels in the blood and potentially greater side effects.³⁰ The variant known as CYP2C9*3 is the most clinically relevant. People who carry this variant break down ibuprofen more slowly, resulting in higher plasma concentrations and longer drug exposure.³¹

A case-control study found that among people taking moderate-to-high NSAID doses, carriers of the CYP2C9*3 variant had nearly twice the risk of upper GI bleeding compared with those carrying the normal or *2 variant.³² The Clinical Pharmacogenetics Implementation Consortium (CPIC) now has published guidelines recommending that CYP2C9 genotype be considered when prescribing NSAIDs. In practice, pharmacogenomic testing is not yet routine for NSAID prescriptions in most clinics, but it is increasingly available through direct-to-consumer genetic testing and some hospital systems. If you already know your CYP2C9 status from another medication (it also affects warfarin dosing), that information is worth sharing with your prescriber.

Protecting the Stomach When You Need an NSAID

For people who must stay on an NSAID long-term, co-prescribing a proton pump inhibitor is the most common protective strategy. A Cochrane review found that PPIs reduced the risk of endoscopic gastric ulcers by about 60% in chronic NSAID users. Double-dose H2 receptor blockers showed similar effectiveness and were better tolerated than misoprostol, the older prostaglandin analog that is effective but causes diarrhea frequently enough that many patients stop taking it.³³ The European Society of Cardiology recommends that clinicians always consider whether a patient on an NSAID needs a PPI, regardless of which NSAID is chosen.²

The overall clinical principle behind all of this is individualized risk assessment. A person with a history of stomach ulcers but a healthy heart faces a completely different set of tradeoffs than someone with coronary artery disease and an iron stomach. Guidelines consistently recommend choosing the lowest effective dose for the shortest necessary duration, and tailoring the specific NSAID to the patient’s GI and cardiovascular risk profile.³⁴ Diclofenac, despite being widely prescribed worldwide, is increasingly difficult to justify as a first-line choice: its cardiovascular risk is on par with COX-2-selective drugs, its GI risk is moderate-to-high, and its liver toxicity signal is the strongest in the class. The European Society of Cardiology position paper specifically noted that there is no evidence diclofenac relieves pain better than less hazardous alternatives.

Low-Dose Over-the-Counter Use

Much of the risk data discussed above comes from prescription-strength doses taken over weeks or months. Low-dose, short-duration over-the-counter ibuprofen use for common aches and headaches sits in a different risk category. Adverse reactions to ibuprofen appear to be closely tied to dose and duration, and the safety concerns that apply to chronic prescription-dose therapy do not cleanly translate to popping a 200 mg tablet for a headache once in a while. At OTC doses, ibuprofen is associated with the lowest risk of GI complications among common NSAIDs and is considered relatively mild even in overdose scenarios. The story changes when people start treating OTC ibuprofen as a daily habit at higher-than-labeled doses, which is more common than the labeling assumes. If you find yourself reaching for ibuprofen more days than not, that pattern deserves a conversation with a doctor rather than another trip to the medicine cabinet.