Long-Term PPI Use — Body Handbook

ნაწლავები · §88

Long-Term PPI Use

Omeprazole and its cousins — Prilosec, Nexium, Protonix, Prevacid — are among the most effective drugs ever made for heartburn, and among the most over-used. They are designed for eight-week courses; tens of millions of people are on year five. The cost of long-term use is invisible in any single month and real over a decade: B12 and magnesium drain, bone density softens, kidney function bends downward, gut infections become easier to catch. Worst of all, the drug fights back when you try to stop — your stomach floods with rebound acid for weeks and most people, thinking the original problem has returned, give up and refill. The way out is a structured taper with eyes open to the trap.

Decide · As-needed Evidence Moderate თავი ნაწლავები

The drug is cheap, fast, and worth taking for the right reasons — a severe erosion, a healing ulcer, Barrett's esophagus. The case to revisit is everyone who started one years ago for indigestion and never came off. Decades of unindicated daily use bend the kidney, fracture, and infection curves the wrong way, and the major specialist bodies have shifted from "safe for life" to "reassess every eight weeks." The taper is real work — the rebound makes you feel like the original problem is back — but for the modal long-term user, the cost of staying on quietly compounds.

A proton pump inhibitor is exactly what the name says. The cells lining your stomach pump hydrogen ions into the lumen through a molecule called the H+/K+ ATPase — the proton pump. The drug binds it permanently. New pumps have to be made before any acid comes back, which is why one pill lasts a day despite the drug being out of your blood in two hours. After three to five days of dosing, acid output drops by eighty to ninety-five percent.

That's also exactly why long use stops being a free ride. Stomach acid is not just there to torture you when you eat pizza too late. It runs a quiet shortlist of jobs: pulling B₁₂ off the protein it rides in on so your intestine can grab it; reducing iron to a form you can absorb; dissolving the calcium carbonate in your supplements; sterilising whatever bacteria you swallowed with lunch. Knock it out for a decade and each of those jobs degrades. The body adapts to the drug, too. With acid suppressed, the hormone that tells the stomach to make more acid — gastrin — rises and stays high. Gastrin is a growth signal to the acid-making cells. They multiply. When the drug eventually comes off, a larger-than-baseline acid factory restarts on a stomach that's now hyper-reactive. That's the rebound, and it's the single most important thing to understand about coming off the drug.

What the long-term cost looks like

The short-term safety profile is excellent — these drugs got famous because they reliably heal ulcers and put out reflux fires within days. The argument is entirely about what years of daily use does. Here the evidence is large, mostly observational, mostly pointing the same way, and contested at the edges.

Kidneys. A 13-year cohort of about ten thousand American adults found PPI users had fifty percent more chronic kidney disease than non-users, with the risk rising the longer they took the drug Lazarus et al. 2016. A Veterans Affairs study of 125,000 new users hit the same pattern, harder: roughly double the risk of end-stage kidney failure compared with people taking the older heartburn pill (famotidine), and over half of the kidney injuries showed up without any warning event you'd notice along the way Xie et al. 2017.

Bones. A British case-control study of more than thirteen thousand hip fractures found long-term PPI users broke their hips forty-four percent more often Yang et al. 2006. A later pooled analysis of one and a half million people put hip fracture at twenty-six percent more and spine fracture at fifty-eight percent more in chronic users Zhou et al. 2016. The FDA added a fracture warning to the label more than a decade ago.

Gut infections. The stomach's acid bath sterilises about half a litre of bacterial slop a day. Turn it off and ingested spores survive. PPI users get Clostridium difficile — the antibiotic-associated gut infection that produces violent diarrhoea and sometimes ICU-grade colitis — at almost double the rate of non-users Janarthanan et al. 2012, replicated in a Danish nationwide cohort Inghammar et al. 2021. Community-acquired pneumonia is similar — about a quarter more common, with the strongest signal in the first month of starting the drug Eom et al. 2011. Beyond outright infection, the same missing acid lets gut bacteria creep upstream into the small intestine where they don't belong — the overgrowth (SIBO) behind a lot of the new bloating and irregularity that turns up after years on the drug.

The vitamin B₁₂ and magnesium story. The acid you suppressed was doing the first step of B₁₂ absorption. Two-plus-year users develop measurable B₁₂ deficiency two-thirds more often than non-users, with the strongest signal at higher doses Lam et al. 2013. Magnesium is the one the FDA flagged with its own safety bulletin in 2011, after enough hospital cases of seizures and dangerous heart rhythms from PPI-driven hypomagnesemia accumulated to be undeniable FDA 2011. Neither deficiency is dramatic in any one year — they creep in, present as the kind of tiredness and brain fog that everyone blames on age and overwork.

Dementia — contested. A 2016 German claims study lit the firestorm: forty-four percent more incident dementia in older adults on regular PPIs Gomm et al. 2016. Later analyses did not replicate it cleanly — a US prospective cohort and a 2020 meta-analysis of eleven studies both came up empty after adjustment Goldstein et al. 2017 Khan et al. 2020. But a 2023 study tracking cumulative years on the drug instead of any-use found a signal at very long exposures — beyond about four and a half years of cumulative use, dementia risk went up by a third Northuis et al. 2023. The honest reading: probably small or zero at typical short courses, possibly real at decade-scale exposure.

The trap that keeps people on it

Here is the single most consequential fact about long-term PPI use, and almost nobody on the drug has been told it. Stopping the drug causes heartburn — even in people who never had heartburn before they started.

The cleanest demonstration of this took healthy adults with no reflux, gave them eight weeks of esomeprazole, and stopped it. Forty-four percent developed heartburn, regurgitation, or dyspepsia in the weeks after stopping — symptoms they did not have before the trial began Reimer et al. 2009. This is the rebound the mechanism section described — your enlarged acid-making machinery firing on a stomach that's gone hyper-reactive. It lasts roughly four to eight weeks and then settles, replicated in a blinded discontinuation trial of real reflux patients Niklasson et al. 2010.

The trap writes itself. Patient takes a PPI for the burning. Patient tries to stop a year later. The burning comes back, harder than before. Patient concludes: I clearly need this drug, my reflux is back, restart. Patient is now on the drug for the next twenty years. The conclusion is wrong — the burning is the drug, not the disease — but no one told them, so they made the only reasonable inference they could from the data in front of them.

A few other widely-repeated half-truths worth puncturing:

"My acid is the problem." Acid is what hurts; the real problem is the valve at the top of your stomach not closing properly. PPIs let the reflux happen — they just neutralise what comes up. This is why some people on a PPI still have a chronic cough, hoarseness, or asthma flares from non-acid reflux: the drug didn't fix the mechanical problem.
"Famotidine (Pepcid) is just a weaker version." Different drug class, different mechanism — blocks histamine's signal to the acid-making cells instead of jamming the pump. Less profound acid suppression, far less rebound, much shorter safety dossier of concerns. A useful step-down tool, not a downgrade.
"It's available without a prescription, so it's safe long-term." The OTC label says a fourteen-day course, no more than three courses a year. The drug was approved for the public on the strength of short-term safety — not the long-term safety question this article is about.

What ten unnecessary years on it actually look like

The reader the stakes apply to is not the rare extreme case — it's the modal long-term user. Someone who started a PPI in their late thirties for chest burning after big meals, refilled it through a couple of doctor changes, and is now fifty and has been on it for twelve years without anyone re-asking why. Most of what follows isn't dramatic in any given year. That's the point.

Year by year, you don't notice. No symptom announces a quietly rising creatinine. No symptom announces a slowly falling B₁₂. Magnesium creeping toward the low end of normal makes you a little crampier, your sleep a little more restless, your patience a little shorter — and you blame work and age. Your kidney function in your fifties is still well inside the lab's "normal" range, just not where it would have been.

The decade compounds. By your early sixties, the calcium and bone story starts to bite. A wrist break from a fall that wouldn't have broken anyone at forty. A vertebral fracture detected on imaging done for something else. Your partner notices you're getting smaller. Hospital admissions for pneumonia happen earlier and recover slower. A course of antibiotics for a sinus infection ends in a bad bout of C. difficile and a hospitalisation that takes months to fully recover from — the kind of event that frequently marks the start of irreversible functional decline in older adults. Your kidneys, looked at properly, are in early-stage chronic disease.

The numbers behind that picture. Not every long-term user follows this trajectory. None of these risks are certain, and the relative-risk increases in the literature — half again the rate of chronic kidney disease, forty-four percent more hip fractures, double the rate of C. difficile — translate to modest absolute risk for an individual in any one year and substantial absolute risk over decades of unindicated use. The version of you who reviewed the prescription at year two and tapered off is the version who arrives at sixty-five with the kidney function, bones, and gut you would have had anyway.

How to actually get off

If you have been on a PPI for more than eight weeks, the question to bring to your doctor is: do I still have a reason to be on this? The 2022 expert deprescribing guidance from the American Gastroenterological Association is the operational basis for everything below Targownik et al. 2022.

For most people without a clear long-term reason — see the next section for the people who should stay on — there are three things to get right.

Fix the mechanical problem at the same time. The PPI was masking a valve that doesn't close well. Raise the head of the bed about six inches on blocks (pillows don't work). Stop eating two to three hours before lying down. Make the evening meal the smaller one. Lose weight if you have it to lose — every five-kilo drop measurably reduces reflux frequency. Stop smoking. These are the things the drug was substituting for.

Expect the rebound and ride it out. The burning will come back in week one or two, sometimes worse than before you started. This is the drug's withdrawal effect, not your reflux returning. It clears in four to eight weeks. If you give in and restart at full dose during that window, you are back where you began. Know this in advance and keep antacids within reach for the bad days.

After eight weeks off, reassess. Most people without an erosive disease or Barrett's land somewhere acceptable — occasional symptoms manageable with on-demand famotidine and lifestyle. If symptoms are constant and disabling at the twelve-week mark, that is information: see a gastroenterologist, get scoped, and find out what the actual disease is. The honest answer for a small fraction of long-term users is that they do have severe reflux and should go back on the drug, this time with a real indication on the chart.

When you should not try to come off

One drug-interaction note worth knowing: if you take clopidogrel (Plavix) after a heart attack or stent, ask about pantoprazole or rabeprazole specifically — omeprazole and esomeprazole interfere with how clopidogrel is activated. The effect is clinically modest but the safer choice is free.

What comes back

The honest version is that successful deprescribing rarely produces a dramatic before-and-after moment. The harms you avoided are statistical and slow. The recovery is the same. But it does come.

Weeks one through eight: the worst of it, then settled. The rebound burning peaks in the first two weeks off, eases through the second month, and is mostly gone by week eight. You spend that stretch with antacids at the bedside and a healthier dinner schedule. The people around you mostly don't notice anything is happening.

Months three to six: the quiet wins. Magnesium recovers first — the cramps you'd written off as a normal part of getting older soften. Sleep gets a little less restless. If you'd drifted into B₁₂ depletion, the afternoon fatigue you'd blamed on work eases by month four or six (full recovery takes longer if you were deep into deficiency). For older adults specifically, mood lifts a notch as the B₁₂ pool refills — the apathy and short fuse that everyone had explained away.

Years out: the curves you didn't bend. This is the payoff that exists only in counterfactual — the kidney function still inside normal at sixty-five, the hip that didn't break at seventy-two, the bout of C. difficile from that course of antibiotics that stayed mild instead of putting you in hospital. These are the years you do not lose. Hard to feel grateful for an admission you didn't have, but those are the entries on the ledger.

One thing to set expectations about: most of the benefit is in not adding more years on the drug. The decade you already did is mostly done. The decade ahead is the one this changes.

Adjacent topics worth looking into

Reflux and the LES mechanics. The valve at the top of your stomach is the thing the PPI was masking. Most of the lifestyle moves above belong there as a standalone topic.
H. pylori screening. If you've never been tested and have a long heartburn history, get tested. Eradicating it cures a substantial share of long-term acid-suppression dependence.
Vitamin B₁₂ testing in older adults. Routinely missed; cheap; high yield in anyone over sixty, on metformin, or with any neurological vagueness.
Magnesium status. Not on routine bloodwork. Worth asking for explicitly if you've been on a PPI for years, especially alongside a diuretic.
Bone density screening. Earlier and more frequent for long-term PPI users, particularly women approaching menopause and men over sixty-five.

Related in the handbook

Worsens / aggravates

— Long-term acid suppression slowly starves you of B12; on a PPI for years, get your level and an active marker checked.
— Years on a PPI quietly cut B12 absorption, one of the long-term costs to plan for.
— Long-term acid suppression cuts B12 absorption, and low B12 is one driver of preventable hearing decline. Test before you blame your ears on age.
— Years of acid blockers bend kidney function the wrong way — a reason to reassess if you've been on one a long time.
— Stomach acid helps you absorb iron — block it for years and iron can quietly run low.
— Years on a PPI can quietly drop your magnesium — the FDA flags it — so long-term users should have the level checked.
— Long-term PPI use is linked to thinner bones and more fractures. If osteoporosis is on the table, ask whether you still need it.
— Long-term acid suppression lets gut bacteria creep upstream — one reason SIBO shows up after years on a PPI.
— Years on a PPI quietly lower zinc absorption alongside B12; another cost to factor in.

Helps

— PPIs aren't only for heartburn — they put a real share of eosinophilic esophagitis into remission.

— A PPI started years ago and never stopped is the textbook case for an annual medication review.
— Reflux is the reason most people start a PPI — and the lifestyle fixes are how many can taper off it.
— Before settling into long-term acid suppression, it's worth ruling out and treating H. pylori.

Substance and claimed effects

Proton pump inhibitors (PPIs) — omeprazole, esomeprazole, pantoprazole, lansoprazole, rabeprazole, dexlansoprazole — irreversibly inhibit the gastric H+/K+ ATPase on parietal cells, suppressing gastric acid secretion by 80–95% at standard doses. They are among the most prescribed drugs in the world; in the US, an estimated 7–15% of adults take a PPI in any given year and a substantial fraction take them daily for years, often without an active indication review Heidelbaugh et al. 2012. Approved short-term indications (4–8 weeks): erosive esophagitis, peptic ulcer disease, H. pylori eradication, Zollinger-Ellison, NSAID-related ulcer prophylaxis in high-risk patients. Long-term indications: Barrett's esophagus, severe erosive reflux that relapses on withdrawal, chronic NSAID/anticoagulant use with high bleeding risk.

This entry covers prolonged PPI use beyond a clear indication and its downstream consequences: impaired absorption of B₁₂, magnesium, calcium, and iron; increased risk of C. difficile and other enteric infections plus community-acquired pneumonia; bone density loss and hip/spine/wrist fracture risk; acute interstitial nephritis and accelerated chronic kidney disease; a contested dementia signal; and rebound acid hypersecretion on withdrawal that traps people on the drug. The scoped consequences span health_short_term (infection, deficiency symptoms), longevity (CKD progression, fracture-related mortality), energy/focus/mood (B₁₂ and magnesium pathways), beauty_cumulative (bone, skin, hair via micronutrient status), and meaningful effort_burden in the act of deprescribing.

Evidence by addressing question

mechanism

The parietal cell secretes H⁺ into the gastric lumen via the H+/K+ ATPase (the proton pump). PPIs are prodrugs activated in the acidic secretory canaliculus where they bind cysteines on the pump's alpha-subunit covalently. Suppression is therefore irreversible per pump molecule — recovery requires synthesis of new pumps, which is why a once-daily dose suppresses acid for ~24 hours despite a plasma half-life under 2 hours. Steady-state acid suppression takes 3–5 days.

Downstream from acid suppression, several mechanisms drive the observed harms Vaezi et al. 2017:

Nutrient malabsorption. Vitamin B₁₂ is released from dietary protein by gastric acid and pepsin; without that first step, even normal intrinsic factor can't bind and absorb it. Serum B₁₂ falls measurably within ~2 years of daily use Lam et al. 2013. Non-heme iron requires acid for reduction to Fe²⁺; calcium-carbonate dissolution is acid-dependent (calcium citrate is not). Magnesium absorption mechanism is less clear — transcellular TRPM6/7 channels in the colon depend on luminal pH and microbiota.
Loss of the acid barrier. Gastric pH normally <3 sterilises ingested microbes. PPIs raise pH to 4–7, allowing oral/oropharyngeal flora to colonise the proximal small bowel (SIBO) and ingested pathogens (C. difficile spores, Salmonella, Campylobacter) to survive transit. Aspiration of acid-suppressed gastric contents seeds the lower airway with bacteria — the pneumonia mechanism.
Hypergastrinemia and rebound. Acid suppression removes feedback inhibition on antral G cells. Gastrin rises 2–4× within weeks. Gastrin is trophic to enterochromaffin-like (ECL) cells, expanding the histamine-secreting machinery. When the PPI is withdrawn, the expanded ECL/parietal mass plus restored pump function produces acid output above pre-treatment baseline for 4–8 weeks — rebound acid hypersecretion. This was demonstrated in PPI-naive healthy volunteers: 8 weeks of esomeprazole produced acid-related symptoms (heartburn, regurgitation, dyspepsia) in 44% of the active arm versus 15% of placebo during the post-withdrawal period, with zero patients having symptoms before randomisation Reimer et al. 2009. This is the mechanism that traps people who started a PPI for transient heartburn.
Kidney injury. PPIs cause acute interstitial nephritis (AIN) — an idiosyncratic T-cell mediated hypersensitivity that can present indolently, without rash or eosinophilia, and progress to permanent loss of renal function before recognition. Beyond AIN, chronic exposure is associated with CKD progression through mechanisms still being worked out (possibly subclinical AIN, possibly magnesium-related, possibly vascular).

evidence

Kidney. The Atherosclerosis Risk in Communities (ARIC) cohort found PPI users had a 50% higher relative risk of incident CKD over 13 years versus non-users, dose-dependent and unchanged after extensive confounder adjustment Lazarus et al. 2016. A US VA cohort of 125,000 new PPI users versus H2-blocker users showed PPIs were associated with a 28% higher risk of CKD progression and a 96% higher risk of end-stage renal disease, with risk rising monotonically with cumulative duration — and notably, more than half of the PPI users who developed kidney injury had no preceding AKI, undercutting the "you'd notice" defense Xie et al. 2017. A Stockholm cohort replicated the CKD-progression finding in patients with baseline kidney disease Klatte et al. 2017. Confounding by indication remains a serious limitation; the COMPASS RCT (described below) did not find a renal signal but was underpowered for chronic endpoints.

Fractures. The UK GPRD case-control study of >13,000 hip fractures found long-term (>1 year) PPI use was associated with adjusted odds ratio 1.44 for hip fracture, dose- and duration-dependent Yang et al. 2006. A 2016 meta-analysis of 18 studies (1.5M participants) found pooled relative risks of 1.26 for hip fracture, 1.58 for spine fracture, and 1.33 for any-site fracture Zhou et al. 2016. The FDA added a fracture warning to PPI labelling in 2010-2011. Mechanism is contested: calcium-absorption hypothesis is the textbook account but BMD changes in PPI users are small; some signal may come through hypomagnesemia (magnesium is a cofactor for vitamin D activation and PTH secretion) or direct osteoclast effects.

Infection. C. difficile: meta-analysis of 42 studies showed PPIs nearly doubled the risk (OR 1.74 community-acquired, 1.74 hospital-acquired) Janarthanan et al. 2012; a Danish nationwide cohort confirmed roughly 2× community-associated CDI risk Inghammar et al. 2021. Community-acquired pneumonia: pooled OR 1.27 in a meta-analysis of 31 studies, with the highest risk in the first 30 days of starting therapy Eom et al. 2011. SIBO is harder to attribute mechanistically but consistently more common in chronic PPI users.

B₁₂. Kaiser Permanente case-control: 2-or-more-year PPI use carried OR 1.65 for clinical B₁₂ deficiency, with stronger association at higher doses and in women and younger patients Lam et al. 2013. Clinical relevance is dose- and diet-dependent — vegetarians and the elderly (already at higher baseline risk) are the populations where deficiency becomes symptomatic.

Magnesium. The FDA issued a safety communication in 2011 after multiple case reports of severe symptomatic hypomagnesemia (cardiac arrhythmias, tetany, seizures) requiring drug discontinuation FDA 2011. A meta-analysis of 9 observational studies estimated pooled relative risk ~1.43 for hypomagnesemia in PPI users Cheungpasitporn et al. 2015. Risk is elevated in patients also on diuretics (loop or thiazide). Magnesium is not detected on routine chemistries unless ordered explicitly.

Dementia. The 2016 German AOK claims analysis found a 44% higher hazard of incident dementia in adults >75 on regular PPIs versus non-users Gomm et al. 2016. This signal has not consistently replicated. A US prospective study of ~10,000 adults found no association between PPI use and incident dementia or mild cognitive impairment after adjustment, including for cumulative duration Goldstein et al. 2017. A 2020 meta-analysis of 11 studies found no overall association Khan et al. 2020. The 2023 ARIC cumulative-exposure analysis re-opened the question: among ~5,700 dementia-free older adults followed for ~5 years, >4.4 cumulative years of PPI use carried HR 1.33 for incident dementia, with no signal at shorter exposures Northuis et al. 2023. The current state is: the signal is small, possibly real at very long durations, almost certainly not causal at typical short-term exposures, and easily confounded by gastrointestinal symptoms that themselves predict cognitive decline.

The RCT counterweight. The COMPASS sub-study randomised 17,598 patients to pantoprazole 40mg daily or placebo for a median 3 years Moayyedi et al. 2019. No statistically significant increase in any of the cohort-reported harms except enteric infection (OR 1.33 for any enteric infection). No signal for CKD, dementia, fracture, MI, pneumonia, COPD exacerbation, gastric cancer, chronic kidney disease, or death — though the trial was almost certainly underpowered for slow endpoints like CKD progression and dementia, and the median follow-up is shorter than the cumulative exposure that drives observational signals. This is the strongest piece of evidence on the "PPIs are safer than the headlines" side — but it isn't a clean acquittal, because the absence of effect for slow-to-develop endpoints at 3 years tells you less than a 13-year observational cohort with the same number.

protocol

Deprescribing is the actionable behavior. The 2022 AGA Clinical Practice Update gives the operational framework Targownik et al. 2022:

Identify candidates. Any patient on a PPI >8 weeks who does not have a clear long-term indication: Barrett's esophagus, severe/erosive esophagitis (LA grade C/D), Zollinger-Ellison, idiopathic chronic peptic ulcer, chronic high-dose NSAIDs in a high-bleeding-risk patient, or recent ulcer bleeding on antithrombotic therapy.
Plan the taper, don't quit cold. Cold-turkey withdrawal triggers rebound symptoms in ~40% of patients Reimer et al. 2009 — which they then attribute to "my reflux coming back," reach for the PPI again, and conclude they need it forever. The recommended approach: halve the dose for 2–4 weeks, then take alternate days for 2–4 weeks, then stop. Some clinicians switch to H2-blockers (famotidine 20–40mg) as a bridge during the taper, then taper the H2-blocker. Either approach mitigates rebound.
Treat the underlying symptom. The reason most people are on a PPI is reflux. Weight loss (when applicable — every 5kg drops reflux symptom frequency), bed-head elevation, smaller evening meals, avoiding late eating (2–3 hours pre-supine), and stopping smoking address mechanism. Alginates (Gaviscon Advance, sodium alginate-based) and on-demand antacids cover breakthrough.
Reassess at 4 and 12 weeks. Symptoms recurring on a regular basis is an indication to re-evaluate, not necessarily to restart at full dose. On-demand or step-down protocols are options.

Niklasson's double-blind placebo-controlled discontinuation trial showed rebound symptoms persist for ~4 weeks after stopping, then fade Niklasson et al. 2010. The literature converges: most patients can come off; the 2–8 weeks after stopping are the critical window where they need to be told the burning is the drug's withdrawal effect, not the original disease returning.

contraindications

Stopping is contraindicated in: Barrett's esophagus (PPI suppression is the standard of care to reduce dysplasia progression risk); recent variceal/peptic ulcer bleed within 12 months on anticoagulants or antiplatelets; Zollinger-Ellison syndrome; severe erosive esophagitis (LA grade C/D) without proven healing on endoscopy; eosinophilic esophagitis being treated with PPI; chronic NSAID use with prior ulcer or in older adults. These patients should stay on the PPI; the long-term safety risks are dominated by the bleeding/disease-progression risks of stopping.

Drug interactions: clopidogrel-PPI interaction is mostly clinically modest but matters for omeprazole/esomeprazole specifically (CYP2C19 inhibition reduces clopidogrel activation); pantoprazole or rabeprazole are preferred in patients on clopidogrel. Methotrexate clearance is reduced by PPIs (relevant in high-dose oncology dosing).

misconceptions

Several myths dominate the lay conversation:

"My acid is the problem." Acid is the symptom; the lower esophageal sphincter's incompetence is the disease. PPIs let the reflux happen — they just neutralise what's being refluxed. Hence the failure of PPIs to fix the cough, hoarseness, and asthma that some reflux patients have: non-acid reflux still happens.
"PPIs cause heartburn when I stop because I need them." No — they cause heartburn when you stop because you took them. Rebound acid hypersecretion was demonstrated in healthy volunteers with no baseline reflux Reimer et al. 2009. The taper-then-tolerate-rebound script is necessary to escape this trap.
"Famotidine (Pepcid) is just a weaker PPI." Different drug class (H2-receptor antagonist). Different mechanism (blocks histamine's stimulation of acid; doesn't suppress the pump). Less profound suppression, no rebound to the same degree, fewer long-term safety associations. A useful step-down tool.
"Over-the-counter means safe long-term." OTC labelling specifies 14-day courses, max 3 courses per year. The drug being OTC reflects acute symptom relief, not chronic-use safety.

failure-modes

Where deprescribing breaks down:

Cold turkey without coaching. Patient stops, gets rebound at days 5–14, restarts, decides they need the drug.
No address of LES mechanics. Patient tapers but keeps eating large late dinners and lying down; symptoms recur and they conclude they have "real" reflux.
Indication drift. Started for an upper-GI bleed prophylaxis in the hospital, never stopped on discharge. Studies repeatedly find >50% of inpatient-initiated PPIs continued without indication at one year.
The "small dose feels fine" trap. Long-term low-dose maintenance still produces the harms (the risk associations track cumulative duration more than dose at typical ranges).

practicalities

PPIs are cheap (generic omeprazole ~$5/month OTC in the US), available without prescription, and effective for the symptom they target. This is why the prescribing inertia is so strong: they work, they're cheap, the harms are statistical and delayed. The cost of staying on is small and visible; the cost of stopping is immediate (rebound) and visible; the cost of long-term use is invisible. This explains the population behaviour.

The deprescribing conversation requires a clinician for anyone with a possible long-term indication (Barrett's, prior bleeding, chronic NSAIDs). For uncomplicated patients started on a PPI for "burning sensation" or "indigestion" 6 months to 5 years ago, the taper can usually be self-managed with primary-care backup.

stakes

The stakes are the cumulative, slow consequences of an unindicated PPI continued for years to decades: a kidney function trajectory that bends downward earlier and steeper than it should; a B₁₂ deficiency that surfaces as fatigue, peripheral tingling, or cognitive slowing in your 60s; a magnesium status low enough to fuel cramps, palpitations, and poor sleep; a hip or vertebral fracture in your 70s that wouldn't have happened; recurrent C. difficile after a course of antibiotics; perhaps a slightly higher dementia risk at very long cumulative exposure. None of these are dramatic in any single year of use. They aggregate. The stakes specifically apply to readers whose PPI use has drifted past its original indication — which is the modal long-term user.

payoff

For the modal long-term user without indication: a successful taper restores baseline acid output within 4–8 weeks; B₁₂ and magnesium recover over months; the elevated kidney/fracture/infection risks roll back toward baseline (the observational data on discontinuation is thinner than on initiation, but risk does not appear to persist indefinitely after stopping). The felt change is rarely dramatic — most people don't notice the deficiency-related symptoms in real time because they crept in slowly. The gain is removing a slow cost.

audience

Two populations carry the highest absolute risk-of-harm from chronic PPI use: older adults (already at higher baseline fracture, CKD, B₁₂-deficiency, pneumonia, and CDI risk — PPI multiplies all of these) and CKD patients (further accelerates progression, AIN can be catastrophic). Younger long-term users (35–55) carry meaningful relative risk increases but lower absolute risk; the case to deprescribe is strongest based on cumulative exposure they're accumulating for the decades ahead.

alternatives

The replacement stack for the modal patient leaving PPIs: (1) Lifestyle: weight loss, head-of-bed elevation 15cm, no eating within 3 hours of supine, smaller evening meals. (2) On-demand antacids: calcium-carbonate (Tums) or aluminum/magnesium hydroxide (Mylanta) for breakthrough. (3) Alginates: sodium alginate (Gaviscon Advance) forms a raft over the gastric pool that blocks reflux mechanically rather than reducing acid; particularly effective post-meal. (4) H2-blockers: famotidine 20mg BID or 40mg at bedtime, either as the destination therapy or as a bridge during taper. (5) Anti-reflux surgery (fundoplication, magnetic sphincter augmentation) for severe disease with documented LES failure where PPIs are also failing. (6) For specific etiologies (eosinophilic esophagitis, motility disorders), targeted therapy.

history

Omeprazole was approved in 1989. The class transformed the management of peptic ulcer disease and reflux through the 1990s. The first long-term safety signals — hip fracture, C. difficile — emerged in mid-2000s observational cohorts. The 2010s produced the kidney, dementia, and CKD signals. The 2020s produced the AGA deprescribing framework, the COMPASS RCT mostly-reassurance, and the still-unresolved dementia question. The arc is the same one that has played out for many "safe" drugs from this era: short-term efficacy demonstrated cleanly, long-term harms slowly surfaced by big observational datasets, RCTs of long-term use too expensive and slow to run definitively.

The credibility range

Optimist case

The most defensible pro-PPI argument: PPIs are among the safest drugs ever brought to market. They reliably treat erosive esophagitis, eradicate H. pylori (with antibiotics), prevent NSAID ulcers, control Zollinger-Ellison, and let Barrett's patients live without progression worry. Most reported "harms" come from observational studies with substantial confounding by indication (sicker patients get prescribed PPIs and have more comorbidities, fracture risk, CKD, and dementia — the drug correlates with disease, not causes it). The COMPASS RCT — 17,598 patients, 3 years, double-blind — found essentially no signal except enteric infection Moayyedi et al. 2019. The dementia panic has not held up in better-designed analyses Khan et al. 2020. Telling patients to come off a drug that controls their symptoms based on a 1.3 hazard ratio in a database is bad medicine.

Skeptic case

The most defensible anti-overuse argument: an estimated 25–70% of long-term PPI prescriptions have no documented indication that justifies continued use Heidelbaugh et al. 2012. Even with confounding-by-indication discounted, the consistency of associations across CKD, fracture, CDI, pneumonia, B₁₂, magnesium, and the dose-response and duration-response gradients on multiple endpoints all point to real causal mechanisms (suppressed acid → infection vector; impaired absorption → deficiency; potential AIN → CKD). Mechanism is plausible for every documented association. COMPASS is reassuring at 3 years but underpowered for the chronic endpoints that drive most concern, and the median enrolled patient's cumulative exposure during the trial is shorter than the cumulative exposure that the observational signal needs. Rebound acid hypersecretion in healthy volunteers Reimer et al. 2009 is iatrogenic dependence — a mechanism that doesn't even depend on long-term harms to justify caution. The default posture should be: short courses for clear indications; reassess every patient at 8 weeks; deprescribe anyone without a defensible reason to continue.

Author's call

The skeptic case is closer to right for the modal long-term user. The optimist case is right for the patient with a clear indication (Barrett's, severe erosive disease, bleeding on antithrombotics, Zollinger-Ellison) — for them, the absolute benefit dominates. The unsettled space is the middle: the long-term low-dose user with mild reflux who took the drug for a known reason 5 years ago and has been on it ever since "just in case." For that patient — who is most patients — the evidence base supports a structured taper with rebound coaching. evidence for the harms is ~4 (multiple replicated observational signals across endpoints, RCT-level reassurance on aggregate harm but underpowered for chronic endpoints, multiple guideline bodies aligned on deprescribing). controversy is ~3 (dementia link disputed; fracture mechanism contested; CKD link strongly held by some, doubted by others).

Stakeholder and incentive map

Pharma. Most PPIs are off-patent and OTC; pharma's active incentive to push the class is small. The continuing prescribing momentum is largely cultural / habitual.
Primary care prescribers. PPIs work fast, are cheap, are covered, and have decades of perceived-safety reputation. The path of least resistance with a dyspeptic patient is a PPI script. Deprescribing requires more conversation time per patient than continuing.
Gastroenterology specialty bodies. AGA, ACG, AASLD have shifted from "safe long-term" to "use as long as needed but no longer" — the AGA 2022 deprescribing update is the clearest example Targownik et al. 2022.
Patients. Most users on long-term PPIs experience symptom control they don't want to give up; the deprescribing conversation faces a strong status-quo bias.
Counter-incentive — wellness / functional medicine. Some practitioners over-rotate on PPI risks and recommend cold-turkey discontinuation, low-stomach-acid supplementation (betaine HCl), and aggressive testing. This camp is often right that there's overuse and wrong about the magnitude of remedy needed.

Population variability

Indication-defined absolute-risk floor. Patients with Barrett's, severe erosive disease, recent ulcer bleed, or chronic high-bleeding-risk NSAID use have benefit that dominates harm at virtually any duration. The reasoning in this entry does not apply to them.
Older adults (>65) carry higher absolute risk on every endpoint — fracture, CDI, pneumonia, CKD, B₁₂ deficiency, possibly dementia. Same relative risk produces larger absolute harm.
CKD patients. Pre-existing renal disease accelerates on PPI exposure in multiple cohorts Klatte et al. 2017; the deprescribing case is strong unless GI indication is clear.
Vegetarians and the elderly. Higher baseline B₁₂ risk; PPI multiplies.
Patients on diuretics. Higher hypomagnesemia risk.
Patients on clopidogrel. Avoid omeprazole/esomeprazole specifically.
Younger patients (<40) on long-term PPI without clear indication. Lower absolute risk currently but the cumulative exposure ahead of them is what drives the deprescribing case.

Knowledge gaps

Causality vs confounding for the kidney, fracture, and dementia signals. Mendelian-randomisation studies and longer RCT follow-ups would help; both are partial and limited.
Whether risk fully resolves after discontinuation. Limited data; what exists suggests partial regression but not zero residual risk for some endpoints.
The dementia question at long cumulative exposure. The Northuis 2023 signal at >4.4 years cumulative Northuis et al. 2023 needs replication; it could be real, confounded, or chance.
Whether intermittent (alternate-day, on-demand) dosing has lower long-term risk than daily dosing. Plausible mechanistically; not formally tested.
P-CABs (potassium-competitive acid blockers, e.g., vonoprazan). Newer class with different pharmacology, less duration of marketing, unknown long-term safety profile that may or may not parallel PPIs.

Scoping calls and rationales the article doesn't say out loud:

Framing as "long-term unindicated use," not "PPIs are bad." The brief said "Proton pump inhibitor use beyond short-term indications." I chose to lean into that framing throughout — the substance under critique is duration-without-reason, not the drug class. This produces a different reading than a generic anti-PPI piece: short-term users and properly-indicated long-term users (Barrett's, ZE, severe erosive disease, NSAID prophylaxis) are explicitly the wrong audience for the deprescribing argument, and the contraindications section is unusually load-bearing.
Dementia: covered but heavily caveated. The brief named dementia as a consequence. The honest read of the literature is that the original Gomm 2016 signal didn't replicate cleanly (Goldstein 2017, Khan 2020 meta-analysis), with the 2023 ARIC cumulative-exposure analysis re-opening the question at very long durations. I covered it in evidence and gave it a focus score of 1 — not zero (because the Northuis 2023 signal and the B₁₂-cognition pathway are real), but small (because shorter exposures don't show it). A future replication of Northuis 2023 either way would change this scoring meaningfully.
Action chose decide over avoid. The entry is fundamentally a "weigh this with your clinician" piece — many readers should stay on, the taper needs supervision in higher-risk patients, and the consequences of getting the indication wrong (Barrett's progression, ulcer rebleeding on antithrombotics) are serious. avoid would have been wrong; know too passive.
Rebound was given outsized space in mechanism, misconceptions, and protocol because in my reading of the literature it is the single most causally upstream failure point for the long-term unindicated population. Patients who don't know about rebound restart the drug and stay on it forever. Patients who do know taper successfully roughly half the time.
P-CABs (vonoprazan, etc.) excluded. Different drug class, different long-term safety profile not yet characterised, marketing trajectory just starting in the US. Worth its own entry once the literature catches up.
H₂-blockers (famotidine) referenced as a tool but not the subject. The H₂-blocker story (lower long-term harm signal, less rebound, useful step-down) probably warrants its own entry — flagging as a separate-entry candidate.
Rating difficulty: longevity at 3 vs 2. The kidney and fracture signals are large in observational data and contested in RCT data (COMPASS reassuring at 3 years but underpowered for chronic endpoints). I landed at 3 because the convergence across CKD, ESRD, fracture, pneumonia, and CDI endpoints — each a real population mortality driver — is hard to dismiss as artifact even if any single one is overestimated.
Future-link candidates: GERD / LES mechanics (the underlying disease the PPI is masking), H. pylori screening and eradication, B₁₂ testing in older adults, magnesium testing, bone density screening cadence, famotidine / H₂-blockers as a class, anti-reflux surgery (fundoplication, magnetic sphincter augmentation) for severe disease.