Cedars-Sinai: 650,000 IBS Patients, Two Drug Classes Flagged

This article summarizes findings from a peer-reviewed observational study and is for informational and educational purposes only. It does not constitute medical advice, diagnosis, or a recommendation to start, change, or discontinue any medication. Drug names are referenced for news context. Patients should not modify any prescription regimen — including antidepressants, antidiarrheals, or any other medication discussed — without consulting their licensed healthcare provider. Note that two of the study's authors disclose consulting and equity relationships with manufacturers of drugs in the class the study did NOT flag for elevated mortality (Bausch Health/rifaximin, Ardelyx/tenapanor); readers should weigh that disclosure when interpreting the study's clinical-practice framing.

Consider a common prescription arc. A twenty-something walks into a primary-care visit with months of abdominal pain, bloating and unpredictable bowel habits, is diagnosed with irritable bowel syndrome, and leaves with a prescription — often a low-dose tricyclic antidepressant or, for the diarrhea-predominant subtype, an older antidiarrheal. She takes it, finds it helps, and stays on it. Not for a year. For twenty.

That arc is ordinary; the evidence base underneath it is not. Most of what clinicians know about the safety of medications used for IBS comes from randomized trials whose follow-up windows are measured in months. The patients are measured in decades. The gap between the two is the subject of a new paper in Communications Medicine from a team at Cedars-Sinai, and it is the gap the paper tries to close — by looking at what happened to more than 650,000 U.S. adults with IBS, drawn from nearly two decades of electronic health records.

Two drug classes stand out in the results. Long-term use of antidepressants was associated with a 35% higher risk of death than among unexposed IBS patients, and use of the two older antidiarrheals — loperamide (the active ingredient in over-the-counter Imodium) and diphenoxylate (the prescription drug commonly paired with atropine as Lomotil) — was linked to roughly double the risk of death, according to the Cedars-Sinai-led team. The story of the paper, though, is not the two flagged classes. It is the structural hole in evidence those flags have filled.

The Long-Term Data Vacuum

Irritable bowel syndrome is among the most common chronic gastrointestinal conditions seen in primary care. It is also among the most treated. The drugs prescribed to manage it span several pharmacologic lineages — antispasmodics, antidiarrheals, laxatives and motility agents, gut-selective antibiotics, FDA-approved IBS-specific agents, and, off-label at high volume, antidepressants used for their effect on visceral pain.

The gap the Cedars-Sinai team identified is one of temporal mismatch. "Many patients are diagnosed with IBS at a young age and may remain on medications for years," Ali Rezaie, medical director of the GI Motility Program at Cedars-Sinai and the paper's senior author, told his institution's newsroom. "However, most clinical trials of these medications last less than a year, so we know very little about their long-term safety."

That sentence describes a problem with a long history in pharmacoepidemiology. Regulatory approval for any drug — IBS treatments included — hinges on trials designed to show efficacy over a defined observation window, typically weeks to a handful of months. Rare adverse events, events that accumulate over years, and events that occur in patients excluded from trial inclusion criteria are all invisible at approval. They become visible later, if at all, through mechanisms like the one this paper uses: retrospective examination of large, longitudinal, real-world data.

What a 650,000-Person EHR Study Can Do

Electronic health records are not a replacement for trials, but they address a specific complementary question: what happens to people who actually take a drug, for as long as they actually take it, in the messy conditions of routine care? At the scale the Cedars-Sinai investigators drew on — more than 650,000 U.S. adults with an IBS diagnosis over close to two decades of records, per the Cedars-Sinai and EurekAlert summaries — even modest excess event rates become statistically detectable.

The choice of endpoint matters. Where shorter trials measure symptoms — pain scores, stool-frequency scales, quality-of-life instruments — a multi-decade EHR study can measure something the trials never see: death from any cause, the most objective and least gameable outcome in all of clinical research. That choice narrows what the study can say (it cannot speak to symptom benefit) and sharpens what it can (it can speak to the long-term all-cause safety ceiling of each treatment class).

Two classes crossed that ceiling. Long-term antidepressant use carried a 35% higher mortality association. Loperamide and diphenoxylate together were linked to approximately double the mortality of unexposed IBS patients, per ScienceDaily's coverage. Two classes did not cross it. The FDA-approved IBS-specific drugs, as a group, showed no increased mortality association. Neither did the antispasmodics — the hyoscyamine-and-dicyclomine lineage that has been in clinical use for so long that it predates modern trials entirely, per the EurekAlert summary.

The pattern is worth pausing on. The drugs most directly approved for the condition by modern regulatory processes came out clean. The flagged classes are, in a different sense, foreign to the condition: antidepressants are an off-label borrowing from psychiatry, and loperamide and diphenoxylate are pre-IBS-era antidiarrheals originally engineered from the opioid-receptor pharmacology of the 1960s and 1970s.

The Mechanism Hypothesis — Why These Classes, Not Others

The paper's authors are careful on the causal question. The study does not establish that these medications directly cause death; it establishes that patients who take them long-term have higher observed mortality than otherwise-similar patients who do not. The distinction is not academic. It governs what the number means for any individual patient and what the clinical response should be.

That said, a plausible mechanism framework has been offered. The excess mortality may reflect higher rates of adverse outcomes — specifically cardiovascular events, falls and stroke — that were reported as more frequent among exposed patients. Each of those three categories maps to known pharmacology in the flagged classes.

Antidepressants — particularly the tricyclic class commonly used at sub-psychiatric doses for IBS pain — have long-recognized cardiovascular signal (QT-interval prolongation, arrhythmia risk at higher exposures) and anticholinergic burden that contributes to falls in older patients. SSRIs and SNRIs carry their own cerebrovascular associations, particularly bleeding-related, that have been debated in the neurology literature for decades.

Loperamide and diphenoxylate are older agents with μ-opioid receptor activity constrained to the gut under normal dosing. At high doses, or in interaction with drugs that disrupt the efflux pumps holding them out of the central nervous system, both have been implicated in cardiac conduction events — an issue prominent enough that the U.S. Food and Drug Administration has issued warnings on loperamide misuse specifically. Falls are a plausible downstream effect of opioid-receptor activity in older users even at therapeutic doses.

None of this demonstrates that the flagged classes are causing the excess mortality observed in the paper. It does suggest that, if the association turns out to be at least partly causal, the physiology through which it would operate is already sketched in existing pharmacology literature.

The Confounding-by-Indication Problem

The single most important caveat on a finding of this shape is confounding by indication — the statistical shadow cast when a drug gets prescribed preferentially to sicker patients, whose underlying illness rather than the drug accounts for the excess events. IBS, as a diagnosis, is heterogeneous enough that this concern is non-trivial.

Patients prescribed long-term antidepressants for IBS-related pain may, on average, differ from unexposed IBS patients in ways other than the prescription itself: higher burden of co-morbid chronic pain, higher rates of anxiety and depression as independent diagnoses, higher contact with the medical system, higher baseline mortality risk for reasons unrelated to the drug. Patients reliant on loperamide or diphenoxylate may have the more severe end of IBS-D, or coexisting conditions — inflammatory bowel disease overlap, post-infectious diarrhea, functional diarrhea variants — that themselves influence life expectancy.

A 650,000-patient sample gives statistical power to adjust for measured confounders. It does not make the problem of unmeasured confounding vanish. The paper is best understood as a high-confidence description of an association, with residual confounding as the dominant alternative explanation to direct drug effect. The authors' own framing is explicit on the causation limit.

The Disclosure Context

One structural feature of the study deserves a candid read. EurekAlert's disclosure paragraph notes that senior author Ali Rezaie serves as a consultant for Bausch Health and Ardelyx, and holds equity in Gemelli Biotech and Good LFE; co-author Mark Pimentel reports consultant and grant relationships with Bausch Health.

Bausch Health manufactures rifaximin — marketed as Xifaxan — which is among the FDA-approved IBS-D treatments in the class the paper found not to be associated with increased mortality. Ardelyx markets tenapanor (Ibsrela), an FDA-approved IBS-C drug in the same cleared class. The disclosure is standard, formally declared, and does not invalidate the finding. It does, however, mean that the two drug classes implicated in elevated mortality are unrelated to the authors' commercial relationships, while the class cleared of elevated mortality contains products from companies the authors are compensated by. A reader is entitled to hold that geometry in mind while evaluating the paper's clinical-practice framing, and the authors are entitled to be presumed honest in having disclosed it.

What the Paper Does Not Do

Several non-claims are worth stating explicitly, because this kind of finding tends to be compressed in secondary coverage.

It does not show that any individual antidepressant, loperamide, or diphenoxylate caused any individual patient's death. It shows that, across a cohort of more than 650,000 people, the groups exposed to those classes had higher mortality than the unexposed groups after the kinds of adjustments routinely applied to EHR studies.

It does not recommend discontinuing antidepressants. Rezaie's own language is emphatic on this point: "IBS patients should not panic, but they do need to understand and weigh the small but meaningful risks when considering long-term treatments," he told the press. Abrupt discontinuation of antidepressants has its own documented harms, and is not the paper's implication.

It does not quantify over-the-counter loperamide use. Electronic health records capture prescriptions and documented clinical interactions; they substantially undercount OTC-only drug exposure. The loperamide signal in the paper is therefore almost certainly a conservative estimate of total real-world exposure. Whether that biases the mortality association up or down depends on subtleties of how OTC use differs from prescribed use.

It does not resolve the mechanism. Cardiovascular events, falls, and stroke are plausible routes; the paper's press coverage describes them as more frequent among exposed patients, not as the proven mediators of the mortality finding.

The Clinical-Practice Nudge

The practice implication the authors point toward is incremental rather than revolutionary: identify the underlying drivers of each patient's IBS, lean on the evidence-based options with the strongest safety ledgers, and resist settling into decade-long reliance on a single class, per the Cedars-Sinai summary of Rezaie's comments.

Read against the paper's own taxonomy, that nudge has a direction. If the FDA-approved IBS-specific drugs and the antispasmodics are the cleared classes, the pattern suggests a reallocation of long-term management toward those agents, combined with periodic reassessment of whether off-label antidepressant use for pain modulation and chronic antidiarrheal reliance are still the right tools for each specific patient. The alternative reading — that the cleared classes are the safer long-term option because they were designed for the condition and have their own targeted-trial evidence base — is consistent both with the paper's findings and with the authors' disclosed commercial ties, which is why both readings need to be held at once.

For patients on long-term antidepressants for IBS-related pain, the paper is not a stop sign. It is a suggestion to re-examine, with a physician, whether the original indication still applies, whether the dose has drifted, whether there are non-pharmacologic options (dietary therapy, gut-directed hypnotherapy, cognitive-behavioral therapy targeted at GI symptoms) that were not previously exhausted, and whether newer IBS-specific agents might carry the pain-modulating effect without the cardiovascular and anticholinergic ledger.

The Prior-Art Ladder — Credit Where It Belongs

This paper is the largest and longest of its kind, but it is not the first to raise the signal in either of the flagged classes. Cardiovascular concerns with tricyclic antidepressants are older than modern pharmacoepidemiology; they date to the drugs' original psychiatric use. Loperamide cardiotoxicity at high doses has been the subject of FDA communications for years. Falls associated with anticholinergic burden are well studied in the geriatric-pharmacy literature.

What the Cedars-Sinai paper adds is scale and chronicity. It is one thing for a pharmacology textbook to list tricyclic antidepressants as QT-prolonging; it is another for a 650,000-patient EHR study to show that, at population scale, over two decades, long-term users of antidepressants prescribed in the IBS context carry a 35% higher all-cause mortality than unexposed IBS patients, per the News-Medical summary. The latter does not replace the former. It promotes a pharmacology caveat into a population-level mortality signal.

What This Does Not Tell Us — Yet

Five open items that matter and are not yet publicly resolved.

1. Within-class differentiation for antidepressants. The press coverage treats antidepressants as a single class. Tricyclics, SSRIs and SNRIs have different pharmacology and plausibly different long-term safety ledgers. Whether the mortality signal is dose-dependent, class-specific, or uniform across the antidepressant family is a first-order clinical question that awaits the full paper and any follow-on analyses.

2. Dose-response for antidiarrheals. Loperamide prescribed intermittently for acute diarrhea is pharmacologically different from loperamide taken daily, for years, at the top of the labeled range. Whether the mortality association scales with cumulative exposure, or is present at any long-term use, is not yet publicly disaggregated.

3. The role of OTC exposure. Because loperamide is over-the-counter, the EHR cohort's exposure measure is necessarily incomplete. Whether the true population-level association is larger or smaller than the EHR signal suggests has not been quantified.

4. Replication in non-U.S. health systems. The cohort is drawn from U.S. records. IBS prescribing practice varies internationally — antispasmodic use is more common in parts of Europe and Asia, antidepressant use for IBS is more common in some systems than others. Whether the signal reproduces outside the U.S. is an empirical question.

5. The mechanism mediation. The press framing names cardiovascular events, falls and stroke as more frequent in exposed groups. How much of the excess mortality is mediated through each of those three categories — versus other causes of death not mentioned in press coverage — is the piece of information that would most sharply guide clinical response.

Implications for the Next Several Years

For patients and clinicians, the immediate implication is prescriptive review at the periodic visit: is the original indication still active, is the dose still the minimum necessary, are non-pharmacologic options still in play, and — for long-term users of the two flagged classes — is a transition to a more specifically-indicated agent justifiable? The answer will be patient-specific; the paper does not obviate individualized judgment, it just shifts the default prior.

For regulators and guideline bodies, the relevant question is whether long-term safety data at this scale should now be an ordinary expectation for chronic-disease medication classes that have been in use for decades without equivalent follow-up. The Cedars-Sinai paper is a proof-of-concept that EHR-based mortality endpoints are achievable for common-disease prescribing patterns. Other chronic conditions with similar duration-of-use profiles may be candidates for the same treatment.

For payers and health systems, the paper is an analytic template. Linking formulary data to long-term mortality through their own EHR systems is now technically demonstrated for one common condition. Whether it becomes a routine part of medication-management programs depends on institutional appetite for producing numbers of the shape this paper produced.

For industry, the finding cuts two ways. Makers of FDA-approved IBS-specific agents have a long-term-safety data point that could support clinical-practice differentiation. Makers of the flagged classes — particularly the generic antidepressant and OTC antidiarrheal supply chains — face a marginal but measurable reassessment of duration-of-use recommendations. Neither effect is immediate; both are plausible within the horizon of the next several prescribing-guideline revision cycles.

Key Takeaways

• A Cedars-Sinai-led team published in Communications Medicine on April 8, 2026 what it describes as the largest long-term real-world safety analysis of IBS treatments, drawing on nearly two decades of U.S. electronic health records covering more than 650,000 adults with an IBS diagnosis, per the EurekAlert summary.
• Long-term antidepressant use was associated with a 35% higher risk of death, and long-term use of loperamide or diphenoxylate with roughly double the risk, relative to otherwise-similar unexposed IBS patients, per the Cedars-Sinai and News-Medical summaries.
• FDA-approved IBS-specific drugs and antispasmodics — the classes specifically developed or long used for IBS — showed no increased mortality association.
• The press coverage frames cardiovascular events, falls and stroke as more frequent in exposed groups, consistent with known pharmacology for both flagged classes, though the paper is explicit that it shows association, not causation.
• The clinical implication is incremental: periodic reassessment of long-term prescriptions, not discontinuation, and a documented shift in the default prior toward IBS-specific rather than off-label pain-modulating options for chronic management.