For Clinicians | Bacterial vs Viral Infection

When Antibiotics Actually Help, and When to Wait

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

How do you know if an infection is bacterial or viral?

Every one of us has had this visit. A patient who feels awful, certain it’s bacterial, already asking for the antibiotic by name. And an exam that, if we’re being straight about it, could go either way.

Here’s the really hard part: signs alone often can’t separate bacteria from viral. Not reliably. And the labs don’t bail us out the way we’d like, either. CRP, procalcitonin, a white count, even all three together nudge the probability without handing over a clean, clear, simple verdict. Patients lean hard on yellow mucus, a single fever, the ‘it’s been five days’ clock. None of those reliably tell you a thing. They feel diagnostic but darn it, they aren’t. And they drive a whole lot of prescriptions that never had a real shot at helping.

So this isn’t going to be a bacterial-versus-viral checklist. You’ve already got one. What’s actually hard, and what actually changes outcomes, is the layer underneath: holding that uncertainty well in front of a patient who came in wanting a clean answer and a prescription.

So today we’re walking the three moves that live in this gray space. What genuinely raises suspicion. When waiting is the right call, and why the evidence backs it instead of treating it as a cop-out. And the narrow, specific place a standby antibiotic actually belongs. None of it skips the clinical judgment. All of it just makes that judgment easier to hold.

The real skill is holding the uncertainty

We were trained to find the answer. Name the bug, match the drug, move on. That reflex gets drilled in from our first courses of our advanced degrees. So when the answer won’t come, when the exam and the labs both just shrug at us, it feels like a failure of skill rather than a feature of the disease. It isn’t. Most everyday infections live in that gray, and the actual skill is staying steady inside it instead of forcing a resolution that the evidence won’t support.

And that brings us to the typical forced resolution. It’s the antibiotic prescription. We all know the pull: the patient expects it, the visit is already running long, and a script is the fastest way to send everyone home feeling like something happened that was worth their copay. Nobody sets out to over-prescribe. It’s just that “let’s treat it to be safe” is the path of least resistance, and patients have learned to expect it, which makes the next clinician’s “no” land like somebody’s holding out on them.

Holding the uncertainty well is the alternative, and it’s a real clinical act, not a soft one. It means saying out loud the thing we’d rather not: I can’t tell you for certain today whether this is bacterial, and here’s why that’s actually okay. It means giving the patient a plan for both branches instead of a pill for one. If this is viral, here’s the course it should run and the exact point where you come back. If it turns, here’s what we do then. That isn’t doing nothing. That’s doing the harder thing.

Everything that follows sits on that footing. The three moves aren’t a way around the uncertainty. They’re how you hold onto it without dropping it on the patient or papering over it with a drug.

What actually raises suspicion for a bacterial infection?

Let us start with the caveat that matters: almost everything here is probabilistic. These aren’t rule-in tests, they’re the findings that move the needle just enough to act, and they look different in every category. Here’s where the suspicion is actually earned in the infections we see most.

UTI: the symptom is the diagnosis. A positive dipstick is not a UTI. Bacteriuria without attributable symptoms (dysuria, urgency, frequency, suprapubic pain) is asymptomatic bacteriuria, and IDSA recommends against treating it in almost everyone outside pregnancy and patients headed for a urologic procedure.¹ This is the cleanest version of the whole article’s premise: a positive test is not an infection. The patient with real symptoms and a positive dip is a different person than the patient with a positive dip and nothing else, and only one of them has a UTI.

Skin and soft tissue: here you actually do get to trust your eyes, mostly. Cellulitis announces itself with spreading warmth, redness, and tenderness. The hard call usually isn’t “is it bacterial,” it’s “how sick is this person,” and that’s where the red flags matter more than the label:

• Pain out of proportion to what you’re seeing (think necrotizing, and move fast)
• Rapid progression you can track by the hour
• Systemic toxicity: fever, hypotension, a sustained tachycardia, two or more SIRS criteria

Those move a patient from “treat and recheck” to “this needs a higher level of care right now.” The flip side is just as useful: uncomplicated cellulitis should turn the corner within 24 to 48 hours of starting therapy.² If it doesn’t, the next question isn’t “stronger antibiotic,” it’s “wrong diagnosis, or something deeper.”

Strep throat: this is what Centor is for. Tonsillar exudate, tender anterior cervical nodes, absence of cough, a history of fever. Fewer than three of those and most patients need neither a swab nor a script.³ The reflex to test and treat every sore throat is exactly the low-value move the score exists to stop.

Respiratory (the big one): this is where suspicion should be hardest to earn, not easiest. Most acute sinusitis, and essentially all acute bronchitis, is viral.⁴ Antibiotics in sinusitis are for genuinely bacterial features or persistence past an observation window, not for the calendar and not for the color of what’s in the tissue. Bronchitis doesn’t get an antibiotic unless you’re actually worried about pneumonia. Cold, flu, COVID: never. This is the category that generates the most “I’m sure it’s bacterial” visits and the fewest infections that actually are.

The thread through all four: suspicion is local and specific, not a global sense that someone seems sick enough to deserve a drug. When the findings genuinely line up, you treat. When they don’t, you’ve got a second move that’s every bit as evidence-based as the first.

When is it safe to wait?

Now before you start to stress seeing this section title take a deep breath. This is the move that feels wrong until you look at the data, and then it feels obvious. Watchful waiting isn’t the absence of a plan. Done right, it is the plan, and it has a real evidence base under it.

The cleanest version is the back-up, or delayed, prescription: the patient leaves with a script in hand (or one you’ll send if needed), plus clear instructions to fill it only if they aren’t improving by a set point or they cross a specific line you’ve drawn. The Cochrane review on delayed prescribing for respiratory infections found it cut antibiotic use sharply versus prescribing immediately, with no increase in complications and patient satisfaction that holds up.⁵ The individual-patient-data meta-analysis landed in the same place: for most symptom outcomes, delayed performs about as well as immediate, and far better on the number that matters for stewardship, which is how many people actually swallow an antibiotic they never needed.⁶

What makes it work is the part we can’t skip: the return plan. Delayed prescribing is safe because the patient knows exactly what they’re watching for and exactly when to come back. Take that away and it’s just a slower prescription. And there’s a real caveat to sit with: at least one analysis tied delayed prescribing in upper-respiratory infection to a higher risk of hospital admission, a reminder that this is a tool for the borderline, self-limited case, not a blanket policy you run on everyone who walks in.⁷ The sicker the patient and the higher the stakes, the less this belongs.

Used in its lane, though, the back-up prescription is one of the most useful tools we have for the gray space. It tells the patient the truth: I don’t think you need this today, I could be wrong, and here’s how we’ll both know. It treats them as a partner in the uncertainty instead of resolving it on their behalf with a drug they probably don’t need.

And notice what it actually is, underneath. It’s a clinician deciding, ahead of time, that an antibiotic is warranted only if a specific condition is met, then putting the medication within reach for that exact moment. Hold that thought. It’s the whole idea behind the third move.

So what IS it for, down there? A short, specific list:

• Giardiasis
• Amebiasis (intestinal, and liver abscess), followed by an intraluminal agent
• C. diff, with the asterisk from a minute ago: not first-line anymore
• Intra-abdominal infection, where it’s always paired with a gram-negative agent, never flying solo

Notice what’s missing. Undifferentiated “food poisoning.” Routine traveler’s diarrhea, too. CDC’s Yellow Book puts azithromycin first-line and fluoroquinolones second for traveler’s diarrhea and metronidazole isn’t on that ladder at all. It’s reserved for Giardia, and even there it’s one option alongside tinidazole and nitazoxanide.⁸ It’s a specialist, not a generalist. Reach for it when you’ve got an anaerobe or a protozoan in your sights…not when a patient tells you their stomach is “off.”

And that narrowness, the very thing that makes it the wrong empiric pick for a vague gut complaint, is exactly why metronidazole can’t be the only gut drug in any kit worth building. Which is right where we’re headed next.

Where does a standby antibiotic actually fit?

So far, the gray space has had a clinician standing in it. Someone to examine, to decide, to be called back in 48 hours. Almost everything above assumes that person is reachable. Pull that assumption out, and the picture changes.

Think about the patient who is genuinely going to need an antibiotic and genuinely can’t get to anyone when the moment arrives. The recurrent-UTI patient with a documented workup and a clinician-set plan, who feels the exact symptoms she’s felt a dozen times, on a weekend, six hours from the nearest clinic. The traveler with a tick bite deep in the backcountry. Traveler’s diarrhea somewhere “go see someone” isn’t a real option for days. For those people, access is the binding constraint, not the diagnosis. And that is the only place a standby antibiotic earns its keep.

Notice it’s the same structure as the back-up prescription from a minute ago: a clinician deciding, in advance, that a drug is warranted only if a specific, recognizable condition is met, then putting it within reach for that moment. The back-up script solves for time. A standby kit solves for access. Same logic, same discipline, same requirement that a clinician drew the line first.

Which is exactly why that line has to be drawn narrowly, and why we’re loud about where it isn’t. A standby antibiotic is defensible only when two things are both true: self-recognition is reliable, and access is the real barrier. That points at a short list of access-tier drugs and clearly bounded indications: amoxicillin-clavulanate, doxycycline, metronidazole, for the worked-up recurrent UTI, the tick exposure, the giardiasis far from care. It points away from the respiratory complaints we spent this whole article cautioning about. Nobody should be carrying azithromycin to self-treat a cough or reaching for ciprofloxacin because their sinuses hurt. That’s the category where overdiagnosis already runs the show, and a kit there makes it worse, not better.

And we have to be candid about the risk, because pretending it away is how stewardship gets undermined. Just having antibiotics on hand is associated with using them more, and using them less appropriately, even among travelers who know better.⁸ That finding doesn’t kill the standby model. It sets the bar for it: the only version that survives that data is one where a clinician scoped the drug, the indication, and the “do not use this for X” before the patient ever walked out with it.

That discipline has a name. We call it appropriate medical preparation: the right drug, for the right indication, with a clear plan, and a clinician making the call. We’re a family team of physicians, PAs, and pharmacists, and it’s the standard we hold the JaseCase to. The kit removes the access barrier for the moment one is genuinely needed. It does not move the decision, and it is not a replacement for primary care, it’s for the times primary care isn’t reachable. If you’ve got patients who travel hard, live remote, or have a recurrent condition you’ve already worked up, you can point them to us at jase.com. And we’ll keep publishing where we draw these lines, and where we refuse to, so the framework is out in the open for the people who’d have to defend it.

The bottom line

Here’s where it all nets out. Most of the time, the patient in front of you who is certain they need an antibiotic doesn’t. The win isn’t a faster diagnosis or a stronger drug. It’s the right action at the right moment, and that action is usually one of three things (none of which is a rushed prescription): reassurance and a clear picture of what normal looks like, or symptom care plus a return plan, or “this one’s serious, let’s get you seen now.” Only in a narrow, pre-worked-out set of cases does it become “you already have what you need, and a clinician already decided when to use it.”

That’s what “having what you need” really means in the gray space. Not a pill for every bug. The judgment to tell the difference, the patience to wait when waiting is right, and the preparation so access is never the thing standing between a patient and the care they genuinely need. Get those three right and you’re practicing at the top of your training, even, and especially, on the days the answer won’t come clean.

That’s the discipline behind how we build, and how we’d ask anyone to build, a standby kit worth trusting.

Sources

1. Nicolle LE, et al. IDSA 2019 Clinical Practice Guideline for the Management of Asymptomatic Bacteriuria. Clin Infect Dis. 2019;68(10):e83-e110. Screening and treatment recommended only in pregnancy and before invasive urologic procedures; treating ASB otherwise drives inappropriate use. https://www.idsociety.org/practice-guideline/asymptomatic-bacteriuria/
2. Brown BD, Hood Watson KL. Cellulitis. StatPearls. Uncomplicated cellulitis improves within 24 to 48 hours of therapy; pain out of proportion to exam, rapid progression over hours, and systemic toxicity raise concern for necrotizing infection and warrant emergent evaluation. https://www.ncbi.nlm.nih.gov/books/NBK549770/
3. Choby BA. Diagnosis and Treatment of Streptococcal Pharyngitis. Am Fam Physician. 2009;79(5):383-390. Modified Centor: score 0-1 needs neither testing nor antibiotics; 2-3 warrants RADT or culture with treatment only if positive. https://www.aafp.org/pubs/afp/issues/2009/0301/p383.html
4. Harris AM, Hicks LA, Qaseem A. Appropriate Antibiotic Use for Acute Respiratory Tract Infection in Adults: Advice for High-Value Care From the ACP and the CDC. Ann Intern Med. 2016;164(6):425-434. No antibiotics for uncomplicated bronchitis unless pneumonia is suspected; antibiotics in rhinosinusitis only for severe or persistent (beyond ~10 days) symptoms; none for cold or influenza. https://www.acpjournals.org/doi/10.7326/M15-1840
5. Spurling GKP, Del Mar CB, Dooley L, et al. Delayed antibiotic prescriptions for respiratory infections. Cochrane Database of Systematic Reviews. 2017;9:CD004417. 11 trials, 3,555 participants; delayed prescribing reduces antibiotic use versus immediate, with no significant difference in complications and comparable patient satisfaction. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD004417.pub5/full
6. Delayed antibiotic prescribing for respiratory tract infections: individual patient data meta-analysis. BMJ. 2021. 9 RCTs plus 4 observational studies, 55,682 patients; no difference in symptom severity delayed versus immediate, safe and effective for most including higher-risk subgroups, with slightly higher symptom severity only in children under 5. https://pmc.ncbi.nlm.nih.gov/articles/PMC8080136/
7. The Safety of Delayed Versus Immediate Antibiotic Prescribing for Upper Respiratory Tract Infections. Clin Infect Dis. 2021;73(2):e394-e401. 1.82 million patients; delayed prescribing associated with higher infection-related hospital admission (adjusted HR 1.52), with delayed prescriptions not well targeted to lower-risk patients, which underscores the need for case selection and clear return precautions. https://academic.oup.com/cid/article/73/2/e394/5864470
8. Kantele A, et al. Stand-by antibiotics encourage unwarranted use of antibiotics for travelers’ diarrhea: A prospective study. Travel Medicine and Infectious Disease. 2018;23:7-13. 316 travelers; carrying standby antibiotics raised antibiotic use to 34% versus 11% in non-carriers, with the excess driven by mild-to-moderate illness. https://pubmed.ncbi.nlm.nih.gov/29894796/

For Clinicians | Metronidazole and Alcohol, C. diff, and What It Actually Treats

Three Things Worth Rechecking

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

Can you drink alcohol on metronidazole?

For me as a pharmacist, I can’t even count how many times I’ve told patients to stay off alcohol while on metronidazole. I thought that was just settled, undisputed information, especially since the FDA still carries the warning right there on the label. For more than ten years I’ve been giving that counsel, probably to thousands of patients at this point in my career. Well, dang it. Turns out the scary part of it, the dramatic reaction we’ve all warned about, was never actually proven.

That’s the thing about a drug we all reach for constantly. We carry beliefs about it that felt true the day we learned them, and some of them haven’t kept up with the evidence. Metronidazole is a perfect example. Most of us learned it cold in school, prescribe or dispense it every week, and counsel on it without thinking twice. And a few of the things we repeat about it are out of date, or were never quite right to begin with.

So today we’re talking about three of them: the alcohol reaction, where metronidazole sits for C. diff, and the idea that it’s a reasonable pick for a “stomach bug.” None of this changes how careful we are with the drug. It just gets the record straight on what the evidence actually says.

Myth 1: Alcohol plus metronidazole causes a disulfiram-like reaction

Short answer: the dangerous version of this almost certainly isn’t real.

The warning we all give traces back to one idea, that metronidazole blocks aldehyde dehydrogenase the way disulfiram does, letting acetaldehyde build up so a patient who drinks while taking this med gets flushing, vomiting, and a pounding heart. It’s a tidy mechanism, buuuuut, it just hasn’t held up. A 2002 double-blind, placebo-controlled study gave volunteers metronidazole and then ethanol and actually measured them: no rise in blood acetaldehyde, and no objective or subjective disulfiram-like symptoms.¹ Reviews that looked at the whole body of evidence landed in the same place: the reported reactions are mostly older case reports confounded by other causes, and a true disulfiram-like effect is minimal or nonexistent.²

So why does every one of us still say it?! Because it’s still written down where we learned it and still written to this day into standard references and the package insert.³ When the reference texts agree, the counsel gets passed down clinician to clinician, year after year, long after the primary evidence stopped backing it. That isn’t anyone being careless. It’s just how durable a line in a textbook is.

This information changes how we counsel rather than whether we counsel on this. The FDA label still instructs patients to avoid alcohol during treatment and for at least three days (72 hours) after the last dose.⁴ That instruction stands, and we don’t get to wave it off at the counter or in the office. What’s changed is our confidence in the reason behind it. So the better answer for the patient who asks isn’t “you’ll get violently ill if you have a drink.” It’s closer to this: the label still says skip alcohol while you’re on it and for a few days after, so that’s the guidance we give, but the old story about a severe reaction is overstated, and most patients won’t have the dramatic episode they’re picturing. Follow the label, but drop the scare tactics we picked up secondhand.

Myth 2: Metronidazole is first-line for C. difficile

Nope. Not since 2017.

For a lot of us this one is pure muscle memory. Metronidazole was the workhorse for mild-to-moderate C. diff for years, that’s where we filed it, and there it stayed. But the guidelines moved on without us. IDSA/SHEA dropped metronidazole as a preferred agent in their 2017 update and held that line in the 2021 focused update: oral vancomycin and fidaxomicin are first-line now, for initial episodes and recurrences both.⁵ Metronidazole only hangs around for non-severe disease when neither preferred agent can be obtained, and even then it’s only the fallback, not the plan.

Why the demotion? The head-to-head data stopped being kind to it. Metronidazole came out with lower cure rates and more recurrence than vancomycin, especially in the sicker patients, and the guideline writers followed the evidence where it went.⁶

So if metronidazole is still filed in your head as the C. diff drug, update that belief right now. It earned its place in the history of treating this bug…it just doesn’t sit at the front of the line anymore.

Myth 3: Metronidazole treats a “stomach bug”

Eh. Usually not the one you’re picturing.

“Stomach bug” terminology covers everything and nothing. Let’s make this simpler: metronidazole has no aerobic gram-negative activity. None. The drug has to be activated inside the organism by an anaerobic electron-transport pathway, so if the bug breathes oxygen, metronidazole walks right past it.⁷ That rules out the aerobic gram-negatives behind a lot of bacterial GI illness, and it does nothing for viruses, which is what most acute gastroenteritis actually is: self-limited and gone in a few days without an antibiotic at all.

So what IS it for, down there? A short, specific list:

• Giardiasis
• Amebiasis (intestinal, and liver abscess), followed by an intraluminal agent
• C. diff, with the asterisk from a minute ago: not first-line anymore
• Intra-abdominal infection, where it’s always paired with a gram-negative agent, never flying solo

Notice what’s missing. Undifferentiated “food poisoning.” Routine traveler’s diarrhea, too. CDC’s Yellow Book puts azithromycin first-line and fluoroquinolones second for traveler’s diarrhea and metronidazole isn’t on that ladder at all. It’s reserved for Giardia, and even there it’s one option alongside tinidazole and nitazoxanide.⁸ It’s a specialist, not a generalist. Reach for it when you’ve got an anaerobe or a protozoan in your sights…not when a patient tells you their stomach is “off.”

And that narrowness, the very thing that makes it the wrong empiric pick for a vague gut complaint, is exactly why metronidazole can’t be the only gut drug in any kit worth building. Which is right where we’re headed next.

Why the spectrum is the whole argument for building a kit right

This is where the spectrum stops being trivial and becomes the design problem. No single oral antibiotic covers the gut. Metronidazole owns the anaerobes and protozoa and is blind to everything aerobic. Ciprofloxacin covers the gram-negative aerobes metronidazole can’t touch. Azithromycin is the one CDC puts first for traveler’s diarrhea, and the agent of choice for the Campylobacter that fluoroquinolones increasingly can’t be trusted against. Three drugs, three jobs, no overlap by accident.

So when metronidazole 500mg sits in a JaseCase next to ciprofloxacin and azithromycin, that isn’t a grab-bag of “stomach stuff.” It’s the same ladder the guidelines already use, assembled on the front end so the clinical thinking is done before the patient ever needs it. Metronidazole is in there as the anaerobe-and-parasite specialist, not the do-everything stomach pill. The kit is built around the exact spectrum line we just drew.

That is what we mean by appropriate medical preparation: the right drug for the right indication, a duration the prescriber sets, and a patient who has been told plainly when not to reach for it, not for the viral stomach bug that’ll clear on its own, not for the vague complaint that needs a real workup. Done that way, a standby kit isn’t a threat to stewardship. Its stewardship moved earlier in time. The opposite of good stewardship was never a prescribed, clinician-built kit. It’s the leftover half-course in the medicine cabinet and the no-questions-asked website that fill the gap when nobody planned ahead.

None of this replaces primary care. Complex disease, chronic conditions, anything that needs a workup and an ongoing relationship belongs in the office. We work in a narrow lane: well-defined situations where the drug, the dose, and the when-not-to-use-this can all be settled in advance. If you’ve got a patient asking what’s reasonable to have on hand and you’d rather not build the answer from scratch, you can refer them to us at jase.com. And we’ll keep publishing the thinking, the spectrum calls, the duration logic, where we draw the lines, so the framework is out in the open for the medical community to push on.

The bottom line

Three things to update about a drug most of us think we know cold. The alcohol reaction is overstated, though the label instruction still stands, so we still give it. Metronidazole hasn’t been first-line for C. diff since 2017. And it’s a narrow-spectrum specialist for anaerobes and protozoa, not the answer for a vague stomach bug. Get the spectrum right and everything downstream gets easier, including the case for why a standby kit, built by clinicians with the thinking done in advance, belongs alongside good stewardship rather than against it.

That’s one drug under the lens. We’re going to keep doing this, one medication at a time. 

Sources

1. Visapää JP, Tillonen JS, Kaihovaara PS, Salaspuro MP. Lack of disulfiram-like reaction with metronidazole and ethanol. Annals of Pharmacotherapy. 2002;36(6):971-974. Double-blind study in 12 healthy volunteers; metronidazole did not raise blood acetaldehyde and produced no objective or subjective disulfiram-like reaction with ethanol. https://pubmed.ncbi.nlm.nih.gov/12022894/
2. Mergenhagen KA, Wattengel BA, Skelly MK, Clark CM, Russo TA. Fact versus Fiction: a Review of the Evidence behind Alcohol and Antibiotic Interactions. Antimicrobial Agents and Chemotherapy. 2020;64(3):e02167-19. Concludes the metronidazole-alcohol disulfiram-like reaction is poorly supported. https://pmc.ncbi.nlm.nih.gov/articles/PMC7038249/
3. Gussow L. The Myth that Metronidazole and Alcohol Cause a Disulfiram-Like Reaction. Emergency Medicine News. 2023;45(10):5. Traces the belief to a single uncontrolled 1964 case and notes it persists in standard references and the FDA package insert. https://journals.lww.com/em-news/fulltext/2023/10000/the_myth_that_metronidazole_and_alcohol_cause_a.5.aspx
4. Flagyl (metronidazole) FDA prescribing information. Instructs patients to avoid alcohol and propylene glycol-containing products during therapy and for at least 3 days (72 hours) after the last dose. (Confirm current label revision on draft day.)
5. Johnson S, Lavergne V, Skinner AM, et al. IDSA/SHEA 2021 Focused Update Guidelines on Management of Clostridioides difficile Infection in Adults. Builds on the 2017 guideline that removed metronidazole as a preferred agent; the 2021 update prioritizes fidaxomicin, with vancomycin an acceptable alternative, and reserves metronidazole for non-severe disease when preferred agents are unavailable. https://www.idsociety.org/practice-guideline/clostridioides-difficile-2021-focused-update/
6. Zar FA, Bakkanagari SR, Moorthi KMLST, Davis MB. A comparison of vancomycin and metronidazole for the treatment of Clostridium difficile-associated diarrhea, stratified by disease severity. Clinical Infectious Diseases. 2007;45(3):302-307. Vancomycin superior, with the difference concentrated in severe disease. (Supported by later pooled phase 3 RCT analysis showing metronidazole inferior to vancomycin.) https://pubmed.ncbi.nlm.nih.gov/17599306/
7. Weir CB, Le JK. Metronidazole. StatPearls. Activity requires intracellular reduction of the nitro group and is limited to anaerobic bacteria and protozoa; aerobes, including aerobic gram-negatives, are intrinsically resistant. https://www.ncbi.nlm.nih.gov/books/NBK539728/
8. CDC Yellow Book, Travelers’ Diarrhea. Azithromycin is the preferred antibiotic for bacterial travelers’ diarrhea; metronidazole is a giardiasis treatment option alongside tinidazole and nitazoxanide, not a treatment for bacterial TD. https://www.cdc.gov/yellow-book/hcp/preparing-international-travelers/travelers-diarrhea.html

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/