Methotrexate and DMARDs: the routine prescription that could kill

Clinical Rounds, the GPAtlas podcast. Musculoskeletal (MSK). 19 minutes. Published 31 August 2026. Free to listen.

Clinical source: our own Ocean article "DMARDs (Disease-Modifying Anti-Rheumatic Drugs)", last reviewed 3 June 2026.

Transcript

Sarah: Hello, and welcome to GPAtlas Clinical Rounds. I'm Sarah, and this is the podcast where we take one thing you'll genuinely see this week and work out what to do with it.

Ben: And I'm Ben. Today, the disease-modifying drugs that sit on your repeat list under shared care, and the two prescriptions that turn them lethal.

Sarah: So this one starts in the most ordinary way imaginable. Mrs Whelan is fifty-eight, she's got rheumatoid arthritis, she's on methotrexate, and she rings on a Monday morning with burning when she passes urine and going more often. Classic uncomplicated urinary tract infection. It's a two-minute phone call, it's probably a same-day appointment with a nurse or a pharmacist, and the standard empirical answer in most of England is trimethoprim. And that combination can kill her.

Ben: Three things today. The one rule that protects these patients above all others. The two co-prescriptions that cause fatal marrow failure, both of which are made in primary care for completely mundane reasons. And how to read the monitoring bloods properly, which is not just checking whether each number is inside its range.

Sarah: So the rule first, because everything else hangs off it. Our article states it as bluntly as I've seen anything stated: no up-to-date, satisfactory bloods, no prescription. That's the rule that protects the patient above all others. And the frequency is the part that surprises people, because it says to check monitoring is up to date and satisfactory before every issue. Not annually, not at review. Every issue.

Ben: And the panel you're checking is the full blood count, the urea and electrolytes, and the liver function. The schedule is typically every two weeks from starting or from a dose change, then monthly for three months, then every three months once stable, although the twenty twenty-five British Society for Rheumatology guideline supports risk-stratified and sometimes less frequent monitoring. So your local protocol may differ, and it's the local one you follow.

Sarah: And I'd add something our article says that gives you permission you might not know you have. It is reasonable to decline shared care if the arrangements are unsafe. Shared care is an agreement, not an instruction. If the monitoring isn't happening and you can't make it happen, you are allowed to say this isn't safe for us to continue.

Ben: No bloods, no prescription. Before every issue.

Sarah: Right. The two killers. And what makes them so dangerous is that neither of them happens in a rheumatology clinic. Both are made by a GP, or a nurse, or an out-of-hours clinician, for a completely reasonable, unrelated reason.

Ben: The first is methotrexate plus trimethoprim, or co-trimoxazole. Our article's word is never. The mechanism is an additive antifolate effect, and the consequence is severe, potentially fatal marrow suppression. And the reason it happens is that trimethoprim is the standard empirical choice for an uncomplicated urinary tract infection, which is the single most routine prescription in general practice.

Sarah: So think about how that consultation actually presents. It's a phone call. It might not even reach a doctor. The person taking it is looking at a urinary tract infection, not at a rheumatology problem, and the methotrexate is somewhere further down a repeat list they may not scroll to. Nothing about that interaction feels risky in the moment, and that is exactly why it's the one to build a hard habit around: any antibiotic request, check what else they're on.

Ben: The second is azathioprine plus allopurinol. Avoid it, or reduce the azathioprine to about twenty-five per cent of the dose under specialist guidance. The mechanism is xanthine oxidase inhibition, and the result is profound marrow suppression. And again, look at how it arises: a patient on azathioprine gets gout, and you start allopurinol, which is the correct treatment for gout and a catastrophe for them.

Sarah: Trimethoprim for a UTI. Allopurinol for gout. Both perfectly reasonable.

Ben: Then the dosing error, which is the other way these patients die. Methotrexate is once weekly. Our article says to agree the day with the patient, write it in full on the prescription, supply the patient-held booklet, and reinforce that it is never taken daily. And accidental daily instead of weekly dosing is listed as a risk of fatal overdose needing urgent assessment.

Sarah: And the ways that error happens are all administrative rather than clinical. A dose written as a number without the word weekly. A patient discharged from hospital with a new list. Somebody moving into a care home, where the drug chart gets rewritten by hand. A relative filling a dosette box. In every one of those, the safeguard isn't your knowledge, it's what's actually written on the prescription.

Ben: Folic acid goes alongside it: five milligrams once weekly, on a different day to the methotrexate. Regimens vary, and some give five milligrams on the six non-methotrexate days. It reduces the gastrointestinal, mucosal and hepatic side effects. The key detail is the different day, because folic acid on the same day is a common and pointless error.

Sarah: Weekly, written in full, on an agreed day. Folic acid on a different one.

Sarah: So the bloods come back and I have to decide whether to sign the repeat. What are the numbers that stop me? And these are worth knowing because they're the ones you'll be looking at on a screen with a queue outside.

Ben: White cells below three point five, or neutrophils below one point six, and some protocols use two point zero. Transaminases, so AST or ALT, above a hundred, or more than three times the upper limit of normal, and if they're at twice the upper limit you reduce the dose and repeat. An unexplained fall in albumin below thirty. An MCV above a hundred and five, where you check B12, folate and thyroid function. And a creatinine rise of more than thirty per cent from baseline, or an eGFR below sixty, where you recheck and discuss.

Sarah: And that renal one matters more than it looks, because it isn't a separate problem. Methotrexate is renally cleared, so a falling eGFR raises the methotrexate level. The patient who gets a diarrhoeal illness, or starts an ACE inhibitor, or gets dehydrated in a heatwave, can move from a stable dose to a toxic one without anybody changing the prescription.

Ben: And then the skill that isn't in any threshold, which our article states directly: a steady downward drift in neutrophils or albumin, or a creeping rise in transaminases, can be the first sign of toxicity even while every value is still technically normal. So you're reading a trend across several results, not a single number against a range. A neutrophil count that's gone from six to four to two point eight is telling you something, and all three of those are normal.

Sarah: Read the trend. Three normal results in a row can still be a warning.

Ben: Three symptoms that mean withhold the drug and act the same day. An unexplained sore throat or mouth ulcers, which is possible neutropenia: urgent full blood count, withhold the DMARD. New or worsening breathlessness or a persistent dry cough, which is possible pneumonitis, especially with methotrexate: withhold, chest X-ray, refer. And unexplained bruising or bleeding, on the same footing.

Sarah: The pneumonitis one is the one I find hardest, because a dry cough in winter is the most ordinary presentation there is. And the instruction is to withhold and investigate rather than to watch and wait, which feels disproportionate right up until it isn't.

Ben: And a rule about restarting that's easy to breach with good intentions: do not restart a DMARD after a monitoring breach or toxicity without specialist advice. So if you've withheld it, you don't quietly resume it once the bloods look better. That's a rheumatology decision.

Sarah: Sore throat, ulcers, or a dry cough. Withhold and check.

Ben: Vaccines, where there's a change that's caught a lot of people out. Live vaccines are contraindicated on most DMARDs, so MMR, yellow fever, BCG, varicella, oral typhoid and the nasal live influenza vaccine. But Shingrix, the current UK shingles vaccine, is non-live, and it is recommended for the severely immunosuppressed. It is not contraindicated.

Sarah: Which is the opposite of the old reflex. Shingles vaccine used to be the one you didn't give these patients, and now it's one you should be actively offering them, because they're precisely the group at risk. And there's a related one: pause methotrexate for two weeks after a COVID vaccine.

Ben: And an exposure that needs same-day action: a chickenpox or shingles contact in a non- immune patient is a risk of severe varicella and needs urgent assessment for post-exposure prophylaxis. So that phone call from a grandmother whose grandchild has just come out in spots is a clinical problem, not a reassurance call.

Sarah: Live vaccines out. Shingrix in, and actively offered.

Ben: Pregnancy, which has two halves and the second one has changed. Methotrexate is teratogenic and must be stopped three months before conception, with effective contraception essential. Mycophenolate is strongly teratogenic. Leflunomide has such a long half-life that it needs a cholestyramine washout. And cyclophosphamide is in the same group.

Sarah: But the paternal advice has been superseded, and this is one where we may still be giving out-of-date information. Current British Society for Rheumatology guidance classes low- dose methotrexate, so twenty-five milligrams a week or less, along with azathioprine, leflunomide, mycophenolate and most biologics, as compatible with a partner's pregnancy. Cyclophosphamide remains the exception. So a man on methotrexate whose partner is trying to conceive does not automatically have to stop, and telling him he does is its own kind of harm.

Ben: And on hydroxychloroquine, which people sometimes forget is in this family at all: no routine blood monitoring is needed, but it needs annual eye screening for retinal toxicity. Different drug, different risk, different check.

Sarah: Paternal exposure advice changed. He probably does not need to stop.

Sarah: I want to say something about how this actually plays out in a practice, because the clinical facts are the easy part. Everything we've described is a system problem wearing clinical clothes. The person who takes Mrs Whelan's call may be a receptionist routing it, then a pharmacist or a nurse prescribing from a protocol, and the protocol says trimethoprim. Nobody in that chain is being careless. The methotrexate is simply not in front of them at the moment they decide.

Sarah: So the fix isn't knowing the interaction, because most of us do know it. The fix is where in the chain the check happens. If it depends on someone remembering, it will fail eventually, because the whole point of a same-day urinary tract infection pathway is that it's fast and doesn't need a doctor. That's a good pathway. It's just blind to this one thing.

Sarah: And the same is true of the monitoring. No bloods, no prescription is a lovely rule and it only works if the person signing the repeat can see, in one glance, whether the bloods are current and satisfactory. If that takes three clicks into a different system, it will get skipped on a Friday afternoon by somebody with two hundred repeats to authorise. Which is not a moral failing, it's a predictable one.

Ben: And it's worth connecting that back to the article's line about declining shared care where arrangements are unsafe. That isn't only about the hospital's paperwork. If your own practice can't reliably see the monitoring at the point of signing, that's part of the same safety question, and it's one you can actually fix.

Sarah: Everyone in that chain knows the interaction. None of them can see the drug.

Sarah: And there's a version of this that's worth spelling out, which is what you actually put in place rather than what you know. Three things, all of them systems rather than knowledge.

Sarah: One: the monitoring has to be visible at the point of signing, not two clicks away in another system. Two: anybody who can prescribe an antibiotic in your practice needs to know the methotrexate and trimethoprim combination, and that includes whoever staffs the same-day access clinic, because that's where it will happen. And three: a gout consultation in someone on azathioprine has to stop you, which means allopurinol is the second drug to build a check around.

Ben: And behind all three, the article's own rule, which is the one that catches everything else: no up-to-date, satisfactory bloods, no prescription, checked before every issue. If that one holds, most of the rest of it is caught even when somebody forgets the specifics.

Sarah: Three systems, not three facts. The bloods rule catches the rest.

Ben: Exam corner. A fifty-eight year old woman on methotrexate for rheumatoid arthritis telephones with dysuria and urinary frequency. She is systemically well. Her last monitoring bloods were four months ago. What is the most appropriate action? A, prescribe trimethoprim, the local first-line choice. B, prescribe nitrofurantoin and check that monitoring bloods are up to date. C, prescribe trimethoprim and arrange bloods next week. Or D, withhold antibiotics until bloods are back.

Ben: It's B. Trimethoprim with methotrexate is a never, because the additive antifolate effect risks severe and potentially fatal marrow suppression, so A and C are both dangerous, and C is arguably the worst because it looks careful. D over-corrects: she has an infection that needs treating, and the answer is a safe antibiotic, not no antibiotic. And the second half of B matters as much as the first, because her monitoring is four months overdue and the rule is no up-to-date satisfactory bloods, no prescription.

Sarah: So. Three things for Monday. One. Any antibiotic request from a patient on a DMARD, look at the repeat list before you choose the drug, because trimethoprim for a urinary tract infection is the classic fatal error. Two. Check the monitoring before every issue, not at the annual review, and read the trend rather than the single number. And three. Shingrix is not contraindicated, it's recommended, so go and find the immunosuppressed patients on your list who've been declined it.

Sarah: One thing to reflect on, if you're logging this. Think about who in your practice answers the same-day urinary tract infection calls. Would the methotrexate on the repeat list stop them? The full transcript and the references are on the episode page. Ben and I are synthetic voices. The medicine isn't. See you next week.

More from Clinical Rounds

Every episode of Clinical Rounds, free to listen with a full transcript.