For Clinicians | Antibiotic Resistance in 2026

What Actually Changed Since You Trained

By Dr. Jamie Wilkey, PharmD — Director of Clinical Strategy, Jase
Medically reviewed by Kristen Carpenter, PA-C — Clinical Advisory Board Member

Most of us were trained on a model of antibiotic resistance that’s about a decade out of date. Not wrong, exactly. Just no longer the whole picture. And three of the things we tell patients with the most confidence are the three things that have shifted the most.

This is a refresher, not an alarm. We’re going to walk through what’s updated in antibiotics in the last decade…where the resistance you treat at the bedside really comes from, why “finish the full course” stopped being the guidance, and what resistance actually is at the population level, because that last one is where the standby-antibiotic question finally gets a clean answer.

Where does the resistance you treat actually come from?

Let’s start with the scale, because the threat is a big deal. The GRAM study, now the gold-standard global model, attributes about 1.14 million deaths directly to resistant infections in 2021 and projects more than 39 million cumulative deaths from 2025 to 2050 if we hold the current course.¹ In the US, the CDC’s standing figure is more than 2.8 million resistant infections and over 35,000 deaths a year, and that’s before you count C. difficile.²

So the problem is real. The question is where this resistance comes from, and this is the first place the old model misleads. You’ve seen the figure that around 80% of antibiotics in this country go to animals. It’s true by volume. It’s also where most people stop, and stopping here is the mistake, because volume is not the same as the resistance you fight at the bedside.

Here’s the clearer way to hold it: agriculture matters most for the bacteria you get from food, and least for the resistant infections you actually admit. Farm use is a real driver of resistant foodborne enterics, the non-typhoidal Salmonella and Campylobacter that NARMS tracks across human, animal, and retail-meat isolates.³ It’s a minor player in the urgent threats that keep you up at night: CRE, C. diff, and drug-resistant gonorrhea are overwhelmingly driven by human prescribing and healthcare transmission.⁴

This isn’t a clean “humans, not farmers” story, and you should be wary of anyone selling it that way. Livestock-associated MRSA is real. The best transmission models still credit animal use with a modest share of human resistance, not none.⁵ But “modest and bug-specific” is the true shape of it, and it points stewardship attention back where it does the most good: the prescriptions written by people like us.

Does “finish the full course” still hold up?

For most common infections, no. This is the shift that surprises people most, because “always finish the course, even if you feel better” is something we’ve all said a thousand times, and we said it on the theory that stopping early breeds resistance. That rationale never had much evidence under it. A 2017 BMJ analysis said so in its title: the antibiotic course has had its day.⁶

What replaced it is the principle of shortest effective duration. A 2025 systematic review found that 85% of duration trials, 267 of 315, showed shorter courses were non-inferior to longer ones.⁷ IDSA’s 2025 complicated-UTI guideline moved the same direction, toward shorter treatment.⁸ The reasoning is plain: every extra day of antibiotic is another day of selection pressure on the patient’s own flora, with no added benefit once the infection is handled.

Three cautions so this doesn’t get oversold. First, and the one that matters most at the counter: shorter does not mean stop when you feel better, and it isn’t a blanket rule. Each shorter duration was validated in its own trial, for one specific infection, and the prescriber sets it for that indication. It tracks cure, not symptoms. Strep is the clean example: most patients feel fine inside a day, but the standard amoxicillin course still runs the full ten, because the goal is eradicating the organism and heading off rheumatic complications, not just clearing the sore throat. Second, the evidence is strongest for common, uncomplicated infections. Severe disease and resistant-organism infections are still under-studied, and “shorter” there is not yet settled. Third, the claim that holds up is that shorter courses are non-inferior for cure and reduce exposure. The further claim, that shorter courses demonstrably lower resistance, is plausible but weaker, and we shouldn’t state it as proven.

The practical translation for the counter: patients still follow the specific prescription in front of them. “Shorter is fine” is a guideline-level change in how we prescribe, not a license for patients to freelance the stopping point on their own.

What is resistance, actually?

This is the shift that reframes the standby-antibiotic question, so it’s worth being precise. Resistance is not an individual phenomenon. A patient’s body does not “get used to” an antibiotic the way it might habituate to caffeine. Resistance happens at the level of bacterial populations: a drug applies selection pressure, the susceptible bugs die off, the resistant ones are left to multiply, and resistance genes move between bacteria on plasmids and other mobile elements.⁹ The patient is the environment. The bacteria are what changes.

If resistance were something a person develops, then holding antibiotics, or having taken them before, would be the hazard all by itself. It isn’t. The lever is use, and specifically inappropriate use: the wrong drug, the wrong indication, a subtherapeutic dose, treating something that was never bacterial.

One caution for credibility, because clinicians will think it: even appropriate, indicated use carries an ecological cost. It selects for resistance in bystander gut flora every time. The point is not that correct prescribing is free. It’s that the benefit outweighs the cost when the drug is right and the indication is real. Inappropriate use is all cost and no benefit. That’s the whole game.

So where do standby antibiotics fit?

We were trained to prescribe inside a closed system: one patient, one prescriber, one chart, one local pharmacy that knew about all three. Now patients get antibiotics through telehealth, mail-order, pharmacies abroad, and the leftover stashes in the cabinet, with or without us. So the real question isn’t whether people will have antibiotics on hand. It’s whether what they have is the right drug, for a real indication, with instructions. And the logic follows straight from the last section: if the lever is use, then keeping the right course on a shelf is not what drives resistance. Misuse is. Possession isn’t.

This isn’t a fringe idea, either. Advance provision is built into medicine everywhere access is the real constraint. Ships without a physician aboard are required to carry antibiotics for crew self-treatment under WHO and maritime labor conventions.¹⁰ Expedition medicine sends them along when evacuation is days out. The US government piloted home doxycycline “MedKits” for anthrax post-exposure.¹¹

No standby provision program has published resistance outcomes. The precedent is real and the logic holds, but we can’t cite a study proving our specific model is resistance-neutral. So the defensible ground is narrow on purpose, and we’d rather claim less and hold it firmly: standby antibiotics make sense where a patient can reliably recognize the problem and access is the binding constraint. Doxycycline for a tick bite or travel into tick country. The worked-up recurrent-UTI patient who knows her pattern. A disaster or remote setting where care is unreachable for days. Travel where the local drug supply is counterfeit or degraded, which is itself a resistance driver, so a quality-assured course carried in is the safer call, not the reckless one.¹²

Where it does not belong: the sore throat, the cough, the sinus pressure that’s viral the overwhelming majority of the time.

The safeguards that make it stewardship

What separates a standby course from a leftover stash isn’t the molecule. It’s everything wrapped around it. Done right, each of these is a stewardship lever. Together they’re the standard a standby program has to clear to count as stewardship at all, the bar we hold our own work to:

• Right drug, chosen deliberately: lead with WHO Access-group agents (amoxicillin-clavulanate, doxycycline, metronidazole) over Watch-group drugs that carry more collateral damage. The molecule is a stewardship decision before the box is ever opened.
• A documented indication: a specific condition the drug is matched to, not “antibiotics, just in case.”
• Clear dosing and the shortest effective duration, spelled out, so nobody is guessing or rationing.
• Explicit “when NOT to use this”: the criteria that point the patient back to care instead of to the kit. With no clinician at the moment of use, this written guidance carries the weight, so it has to be unambiguous.
• A path back to a clinician: these complement primary care, they don’t replace it. Use what’s on the shelf only when real care isn’t reachable.

That’s the standard, not a snapshot of any kit on the shelf today, ours included. The “when not to use” guidance is the piece we’re still building toward, because getting it unambiguous with no clinician in the room is the hardest part of the whole thing, and we’d rather name the bar than pretend we’ve already cleared it.

Charting the grey space

Right now, care is all-or-none. A patient either reaches a prescriber while it still matters, or they’re left with a leftover stash and a search bar. There’s no sanctioned middle: no clinician-built step that prepares someone for the few things they can reliably recognize, before access fails. That gap is where we work, and the category has a name we use on purpose: appropriate medical preparation. The right drug, a real indication, clear instructions for when to use it, prescribed ahead of the moment access disappears.

This complements primary care, it does not replace it, and for anything outside that narrow band the answer is still a clinician. If you’d rather not prescribe in this space yourself, you can send patients to us at Jase.com. 

The bottom line

The model most of us trained on is about a decade out of date. Resistance is ecological, driven by how antibiotics are used, not by who keeps them on a shelf. Agriculture owns the foodborne bugs; human prescribing owns the ones we admit. Shorter courses are the standard now for common infections. And a well-chosen standby antibiotic, with a real indication and clear instructions, belongs inside stewardship, not outside it.

We’re building this in the open, and what you see in the office and at the counter sharpens it.

Sources

1. Naghavi M et al. (GBD 2021 Antimicrobial Resistance Collaborators). Global burden of bacterial antimicrobial resistance 1990–2021: a systematic analysis with forecasts to 2050. The Lancet, 16 September 2024. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(24)01867-1/fulltext
2. CDC. Antimicrobial Resistance Facts and Stats (2019 AR Threats Report baseline: more than 2.8 million resistant infections and over 35,000 deaths a year; with C. difficile, more than 3 million infections and 48,000 deaths). https://www.cdc.gov/antimicrobial-resistance/data-research/facts-stats/index.html
3. CDC, FDA, and USDA. National Antimicrobial Resistance Monitoring System (NARMS): tracks Salmonella and Campylobacter across human, retail-meat, and food-animal isolates. https://www.cdc.gov/narms/about/index.html
4. CDC. Antibiotic Resistance Threats in the United States, 2019 (CRE, C. difficile, and drug-resistant N. gonorrhoeae driven predominantly by human prescribing and healthcare transmission). https://www.cdc.gov/antimicrobial-resistance/data-research/threats/index.html
5. Booton RD et al. One Health drivers of antibacterial resistance: quantifying the relative impacts of human, animal, and environmental use and transmission. One Health, 2021 (Thailand model: eliminating animal use yields a modest reduction in human resistant colonization). https://www.sciencedirect.com/science/article/pii/S2352771421000100 — and Tang KL et al. Restricting the use of antibiotics in food-producing animals and its associations with antibiotic resistance: a systematic review and meta-analysis. The Lancet Planetary Health, 2017 (about 24% lower resistance in humans with restriction, concentrated in people with direct animal contact). https://www.thelancet.com/journals/lancet/article/PIIS2542-5196(17)30141-9/fulltext
6. Llewelyn MJ et al. The antibiotic course has had its day. BMJ, 2017;358:j3418. https://www.bmj.com/content/358/bmj.j3418
7. Mo Y, Tan WC, Cooper BS. Antibiotic duration for common bacterial infections: a systematic review. JAC-Antimicrobial Resistance, 2025;7(1):dlae215 (85%, 267 of 315 trials, found shorter courses non-inferior). https://academic.oup.com/jacamr/article/7/1/dlae215
8. Infectious Diseases Society of America. 2025 Guidance on the Management and Treatment of Complicated Urinary Tract Infections (shorter antibiotic courses for clinically improving patients). https://www.idsociety.org/practice-guideline/complicated-urinary-tract-infections/
9. CDC. About Antimicrobial Resistance (resistance as population-level selection pressure plus horizontal gene transfer, not individual habituation). https://www.cdc.gov/antimicrobial-resistance/about/index.html
10. WHO, ILO, and IMO. International Medical Guide for Ships, and ILO Maritime Labour Convention, 2006 (ships must carry a medicine chest, medical guide, and required medicines, including antibiotics, for self-treatment without a physician aboard). https://imo-epublications.org/content/books/9789241547208
11. Institute of Medicine. Prepositioning Antibiotics for Anthrax (home doxycycline MedKits; Minneapolis–St. Paul postal pilot, 2008). https://www.ncbi.nlm.nih.gov/books/NBK190049/
12. Hall RM et al. Substandard and falsified antibiotics: neglected drivers of antimicrobial resistance? (subtherapeutic dosing from degraded or falsified product selects for resistance). https://pmc.ncbi.nlm.nih.gov/articles/PMC9394205/

For Clinicians | Standby Antibiotics and Self-Start Therapy

Guideline Case for Guided Self-Treatment

By Dr. Jamie Wilkey, PharmD — Director of Clinical Strategy, Jase
Medically reviewed by Kristen Carpenter, PA-C — Clinical Advisory Board Member

Patients stopped asking me whether they should keep leftover antibiotics years ago. They just tell me they did, usually after the fact, usually at the counter, usually with a little defiance in it: their old azithromycin from last winter went toward what felt like a UTI, aaaand they’re only bringing it up now because the symptoms didn’t budge.

Wrong drug for the likely bug. Wrong duration even if it were the right one. And the standard answer we’re all trained to give: “never self-treat, see a provider!” was never really in the running. It assumes a provider was reachable, and at 2am with no cell service, or three days out from the nearest clinic, nobody is. Patients in those moments will act on something. The only question is whether that something was written by clinicians, by a search bar, or just what is at the back of the medicine cupboard.

So today we’re going to flesh this out: what the guidelines already permit, where patient self-diagnosis is reliable (and where it really isn’t), what patients actually do when they can’t reach us, and what the guidance in their hands should look like.

Can patients accurately self-diagnose?

It depends on the condition, and the spread is wider than most of us would guess.Let’s just start at the encouraging side. In women with prior culture-confirmed UTIs, patient suspicion of a new UTI is more than 85% accurate in predicting culture-positive infection, more accurate than a urine dipstick.¹ This holds for women with an established recurrent pattern whose symptoms match prior confirmed episodes, and accuracy drops when vaginal discharge, pelvic pain, or STI exposure complicates the picture. But within those bounds, these patients know their bodies, and the data backs them.Now the other end. When women self-diagnosed a yeast infection and bought an over-the-counter antifungal, only about one in three actually had vulvovaginal candidiasis. The rest had bacterial vaginosis, mixed vaginitis, trichomoniasis, or normal findings. A prior clinician-confirmed yeast infection did not make them any more accurate the second time.²Hold both results at once. Patient intuition is real, and it is not uniform. It is condition-dependent, and patients have no way of knowing which conditions their intuition is good for. That specificity gap is exactly what a written tool has to close.

The permission already exists

The reflex is to file guided self-treatment under fringe medicine. The guidelines disagree.

• Recurrent UTIs. The AUA’s guideline lets clinicians offer select patients self-start therapy: the antibiotic waits at home, and the patient starts it when symptoms hit.³

• Travelers’ diarrhea. The CDC Yellow Book tells travelers to carry an antibiotic and start it themselves when moderate-to-severe symptoms hit. No call required.⁴

• Expedition medicine. Kits are scaled to how far the group is from care, and every drug is labeled with what it treats and how to take it.

• Your own exam room. We already prescribe ahead of the emergency: EpiPens, rescue inhalers, nitroglycerin. The patient carries the drug and uses their own judgment on the day.

Every one of these kicks in when care is out of reach. And every one of them assumes a clinician is nearby doing the guiding.

That’s the gap. We trained inside a closed system: one patient, one prescriber, one chart, one pharmacy. Nobody trained us for the patient at 2am, because she was supposed to be somebody else’s problem. There is no somebody else at 2am except maybe an ultra expensive ER with a wait a mile long. The permission for guided self-treatment exists. The plain-language tool it assumes was never built.

What medically-reviewed guidance actually looks like

This is where the med card earns its place, and the card makes the argument better than any abstraction does. Take the doxycycline card that ships in a JaseCase. The front opens with the counseling we give at the counter (take it with a full glass of water and stay upright for 30 minutes, your esophagus will thank you), then a conditions-treated table with adult dosing for each indication. The doses are not interchangeable: a single two-pill dose for Lyme prophylaxis, twice daily for 10 days for tetanus, once daily starting before travel for malaria prevention, up to 60 days for anthrax. They differ by condition, which is precisely what patients guess wrong. And the Lyme line won’t even grant that single dose unless every criterion is met: tick attached 36 hours or more, a region where blacklegged ticks are common, started within 72 hours of tick removal, no contraindications. That’s the IDSA’s own prophylaxis standard, printed where the patient can read it.⁷

The back carries the side effects worth watching for, when to avoid the drug entirely, and a red Pregnancy Category D where nobody can miss it. Scattered through the dosing table is a phrase doing quiet stewardship work: secondary treatment option. The card tells the patient when this drug is not the right first choice. For pneumonia it goes further and requires a second antibiotic alongside it, partner drug and dose spelled out: the same combination the ATS/IDSA pneumonia guideline recommends for outpatients with comorbidities.⁸ That is the part the skeptics skip: the card spends as much ink on limits as on permissions.

The stewardship objection deserves a direct answer

The strongest pushback on at-home antibiotic kits, and infectious disease colleagues have made it in print, is antimicrobial resistance: patients shooting from the hip with broad-spectrum drugs. The concern is legitimate, and it deserves engagement rather than a dodge.

But look at what the objection assumes: that the alternative to the kit is a clinic visit. For the no-access moments these kits exist for, it isn’t. The real alternatives are going without, taking whatever antibiotics turn up around the house or from a friend’s leftover stash, or buying whatever they think they need from an online pharmacy. None of those comes with the right spectrum, a full course, or any counseling behind it. Against that baseline, a condition-specific, full-course, clinician-reviewed regimen with explicit do-not-use guidance is better stewardship, not worse.

And sometimes the right answer on the card is no drug at all. The clinical review behind each card draws the fence deliberately: well-understood, self-limiting conditions with predictable treatment paths, screened ahead of time by a clinician who reviewed the patient’s history. That fence is what keeps guided self-treatment from sliding into the free-for-all the skeptics fear.

Charting the grey area in public

Step back and look at how antibiotic access actually works in this country. It is all or none. Either the patient reaches a prescriber and gets the right drug, or they are completely on their own resources: the leftover stash, the search bar, the no-questions-asked website. There is no sanctioned middle step. We built a light switch and then act surprised when patients in the dark go looking for matches.

The guidelines covered above have already sketched what the middle step looks like: a defined, conditional, clinician-controlled layer between “call your doctor” and “you’re on your own.” Self-start therapy is that layer for recurrent UTIs. Standby treatment is that layer for travelers. What we’re building at Jase is the same layer for a short list of common, well-understood infections: the clinical work happens up front, a clinician reviews the patient’s history and prescribes for defined conditions, and the card carries the guidance into the moment it’s needed. That is what appropriate medical preparation means in practice: the basics, decided ahead of time, for the moments care isn’t there.

The boundary holds on both sides. This is in no way a replacement for primary care: anything complex, chronic, or unfamiliar still belongs in the exam room, and the cards say so. And if a patient asks you what they should have on hand just in case and you would rather not chart that middle layer yourself, you can refer them to us at Jase.com. We’re a family team of physicians, PAs, and pharmacists drawing these lines carefully, and we’ll keep publishing where we draw them.

The bottom line

Organized medicine already endorses guided self-treatment when access is the constraint; the existing frameworks just assume a clinician is standing there to do the guiding. At 2am, nobody is. What we can control is whether the information in the patient’s hand that night came from clinicians or from a search bar.

Most people never think about any of this until the night it happens to them.

We are here for them then.

Sources

1. American Family Physician (AAFP), April 2016. Patient suspicion of UTI is more than 85% accurate in predicting culture-positive infection, more accurate than urine dipstick. https://www.aafp.org/pubs/afp/issues/2016/0401/p560.html
2. Ferris DG, et al. Obstetrics & Gynecology, 2002. Among women who self-diagnosed vulvovaginal candidiasis and purchased an over-the-counter antifungal, 33.7% had the condition. https://pubmed.ncbi.nlm.nih.gov/11864668/
3. AUA/CUA/SUFU, Recurrent Uncomplicated Urinary Tract Infections in Women, 2025 guideline amendment. Patient-initiated (self-start) treatment for select patients, conditional recommendation (Moderate, Grade C). https://www.auanet.org/guidelines-and-quality/guidelines/recurrent-uti
4. CDC Yellow Book, Travelers’ Diarrhea. Standby self-treatment for travelers; antibiotics reduce illness duration by approximately 1 to 2 days for susceptible bacterial pathogens. https://www.cdc.gov/yellow-book/hcp/preparing-international-travelers/travelers-diarrhea.html
5. IDSA/AAN/ACR, Prevention, Diagnosis and Treatment of Lyme Disease guideline, 2020. Single-dose doxycycline prophylaxis within 72 hours of a high-risk bite: identified Ixodes vector, highly endemic area, attached 36 hours or more. https://www.idsociety.org/practice-guideline/lyme-disease/
6. ATS/IDSA, Community-Acquired Pneumonia guideline, 2019, summarized in American Family Physician, 2020. Outpatients with comorbidities: amoxicillin/clavulanate plus a macrolide or doxycycline. https://www.aafp.org/pubs/afp/issues/2020/0715/p121.html