What it is:not a disease, it's the decision framework you run every single time you pick an antibiotic. Confirm there's actually an infection, figure out (or predict) the bug, match a drug to that bug while respecting the patient in front of you, then reassess daily and narrow.
The core problem:you almost always have to commit to a drug before you know exactly what organism you're fighting. That guess is empiric therapy. Once culture and sensitivity data come back, you switch to definitive therapy, ideally narrower, cheaper, and less toxic than what you started.
What you do about it:pick the narrowest agent that reliably covers the likely (or confirmed) pathogen at the site of infection, dosed for this patient's organs, then de-escalate the moment culture data lets you.
Every antibiotic decision collapses into four questions: what bug is this (or likely to be)? Where is it sitting? Can my drug actually get there? And when can I narrow?If you can answer those four, you don't need to memorize the whole drug-of-choice table, you can reconstruct most of it.
The whole chapter is really a workflow: confirm infection is present, identify the pathogen if you can, start presumptive (empiric) therapy considering the patient and the drug, monitor the response, then step down to definitive therapy once you know exactly what you're treating.
| Stage | When | Goal |
|---|---|---|
| Empiric | Started before the organism is identified, sometimes before infection is even fully confirmed | Cover the most likely pathogens for that site and that patient, broadly enough not to miss |
| Definitive | Once culture, sensitivity, or both come back | Narrow to the smallest-spectrum, least-toxic, cheapest agent that still works |
A positive culture is not the same thing as an infection.Once you get a Gram stain or culture back, you still have to decide whether what grew is the true pathogen, a contaminant introduced during collection, or just normal flora from a poorly collected specimen (think a "clean catch" urine that wasn't very clean). Treating a contaminant is a documented source of unnecessary antibiotic exposure.
Almost every drug-class decision downstream traces back to one structural fact: what does the organism's wall look like, and does it even have one.
| Type | Wall structure | Why it matters for drug choice |
|---|---|---|
| Gram-positive | Thick, multilayered peptidoglycan. No outer membrane. | Beta-lactams and vancomycin get straight to the peptidoglycan target |
| Gram-negative | Thin peptidoglycan layer plus an outer membrane containing lipopolysaccharide (LPS) | That outer membrane blocks some drugs outright and is also how these organisms trade resistance genes so easily |
| Atypicals | No peptidoglycan layer at all, and stain as neither Gram-positive nor Gram-negative | Beta-lactams have nothing to bind, so they never work here no matter the dose |
| Anaerobes | Variable, defined by oxygen tolerance not wall type | Don't require (and some die in) oxygen; most anaerobic infections are treated empirically since they're hard to culture |
If a patient with community-acquired pneumonia isn't improving on amoxicillin, ask whether the cause could be Mycoplasma pneumoniae, Chlamydia pneumoniae, or Legionella pneumophila. None of the three have a peptidoglycan wall, so a beta-lactam was never going to touch them. You need something that hits ribosomes or DNA gyrase instead: a macrolide, doxycycline, or a respiratory fluoroquinolone.
Common organisms by category, and the drug classes typically aimed at each:
| Category | Common examples | Typical drug classes |
|---|---|---|
| Gram-positive | Staphylococcus, Streptococcus, Enterococcus | Penicillins (± beta-lactamase inhibitor), cephalosporins, vancomycin, SMX/TMP, respiratory fluoroquinolones, tetracyclines, carbapenems, macrolides |
| Gram-negative | Escherichia coli, Pseudomonas aeruginosa | Aminopenicillins, BL/BLI combinations, cephalosporins, SMX/TMP, aminoglycosides, fluoroquinolones, carbapenems |
| Anaerobes | Bacteroides fragilis, Clostridioides difficile | Metronidazole (usual drug of choice), clindamycin, BL/BLI combinations, cefoxitin, cefotetan, moxifloxacin, tigecycline, carbapenems |
| Atypicals | Mycoplasma pneumoniae, Chlamydia pneumoniae, Legionella pneumophila | Doxycycline, azithromycin, clarithromycin, fluoroquinolones |
Before you chase a bug, make sure there's a bug to chase. Fever and an elevated white count both have long lists of non-infectious causes, so read them in context.
| Sign | What it tells you |
|---|---|
| Fever | Body temperature controlled above ~37°C (98.6°F) oral. Many things besides infection cause it, drugs included. |
| WBC count | Normal is 4,000–10,000 cells/mm³. Bacterial infection typically raises granulocytes (neutrophils, sometimes with immature "bands" on smear), but rarely above 30,000–40,000. |
| Relative lymphocytosis | Even with a normal total WBC, points toward tuberculosis, viral, or fungal disease rather than acute bacterial infection |
| Local signs | Pain, swelling, erythema, tenderness, purulent drainage. Only visible if the infection is superficial or in a bone/joint. |
| Deep infection clues | For meningitis, pneumonia, endocarditis, or UTI you need the fluid itself: neutrophils in spinal fluid, sputum, or urine strongly suggest a bacterial process |
A patient can be neutropenic(low, not high, neutrophils) after infection onset. That's an abnormal response and it's a bad prognostic sign, not evidence they're fine. Never use "the WBC isn't that high" as your only reassurance in a sick-looking patient, especially one who's immunosuppressed.
Fever that starts around the same time a new drug was started, and disappears promptly when it's stopped (with no other explanation and temperature staying normal afterward), is drug-induced fever, not an infection you failed to treat. Don't reflexively broaden antibiotics for a fever that's actually a drug reaction.
Sample infected material (blood, sputum, urine, stool, wound or sinus drainage, aspirated abscess or cellulitic fluid) beforeantibiotics go in, whenever it's practical. Blood cultures belong in any acutely ill, febrile patient. Gram stain can reveal bacteria directly; an acid-fast stain picks up mycobacteria or actinomycetes.
Giving even one dose of antibiotic before the culture is drawn can suppress pathogen growth enough to produce a false-negative culture, or distort the cellular and chemical makeup of the fluid you're testing. This is one of the most common, avoidable reasons a workup goes nowhere.
Once you have an isolate, the lab reports a minimum inhibitory concentration (MIC), the lowest drug concentration that stops visible growth, and translates it into an interpretation: Susceptible, Intermediate, or Resistant, based on breakpoints specific to that drug.
| Drug | MIC | Interpretation |
|---|---|---|
| Ampicillin | >32 | R |
| Amoxicillin/clavulanate | 16 | R |
| Cefazolin | 16 | R |
| Ceftriaxone | 12 | R |
| Piperacillin/tazobactam | <4 | S |
| Cefepime | <1 | S |
| Meropenem | <0.25 | S |
| Tobramycin | 8 | I |
| Ciprofloxacin | 1 | S |
| Levofloxacin | 1 | S |
| Sulfamethoxazole/trimethoprim | >320 | R |
Notice the pattern above: resistant to the older aminopenicillin, the beta-lactamase inhibitor combo, first-gen and third-gen cephalosporins, and SMX/TMP, but susceptible to the antipseudomonal BL/BLI, fourth-gen cephalosporin, a carbapenem, and both fluoroquinolones. That specific shape of resistance is exactly why you don't guess your way to a definitive drug, you either wait for it or start broad and narrow once it's in hand.
Doxycycline at MIC 0.5 is not"better" than linezolid at MIC 2. MICs are unique to each drug's own potency and pharmacokinetics, they only mean something compared to that drug's own breakpoint. MIC numbers also drift as local resistance patterns change, you're not expected to memorize specific values.
An antibiogramis your hospital's own cumulative susceptibility data for common bug-drug pairs. Local antibiogram data should drive your empiric choice over national compilations or your own memory of "what usually works," because resistance patterns are genuinely regional and can shift year to year.
Once you know (or suspect) the bug, the "right" drug still depends on the patient carrying it.
Most antibiotics need renal adjustment eventually. The ones that generally don't: nafcillin, oxacillin, ceftriaxone, clindamycin, azithromycin, erythromycin, moxifloxacin, doxycycline, tigecycline, and rifampin. Three quick rules that resolve most exam questions: moxifloxacinis the only fluoroquinolone that skips renal dosing, ceftriaxoneis the only cephalosporin that skips it, and clarithromycinis the one macrolide that actually doesneed it.
Tissue penetration varies enormously by site. The CNS is the best-studied example, drugs that don't achieve meaningful cerebrospinal fluid concentrations should be avoided for meningitis (or given directly into the CSF) regardless of how good their in-vitro activity looks. Other sites where penetration data actually changes practice: urine, synovial fluid, and peritoneal fluid.
Two different kill patterns, and the dosing strategy for each is basically the opposite of the other.
| Pattern | Example classes | What predicts success | Dosing implication |
|---|---|---|---|
| Time-dependent | Beta-lactams (penicillins, cephalosporins) | Time the concentration stays above the MIC (T>MIC) | More frequent dosing or extended/continuous infusions beat a single big dose |
| Concentration-dependent | Aminoglycosides, fluoroquinolones | Peak concentration relative to MIC, or AUC/MIC | A big peak dosed less often works, thanks to a lingering "post-antibiotic effect" where the drug keeps suppressing growth even after levels fall below MIC |
Time-dependent: keep the level up longer.That's why you'll see extended-infusion piperacillin/tazobactam or cefepime. Concentration-dependent: hit it hard, less often.That's why aminoglycosides are frequently dosed once daily instead of every 8 hours.
This is the condensed, high-yield version of the drugs-of-choice table. Treat it as a reference for definitive therapy once an organism is confirmed and susceptible, not as a substitute for your local antibiogram when choosing empirically.
| Organism | Drug(s) of choice | Alternatives |
|---|---|---|
| Gram-Positive Cocci | ||
| Enterococcus faecalis, serious infection | Ampicillin or penicillin G (± gentamicin or ceftriaxone) | Vancomycin, daptomycin, linezolid |
| Enterococcus faecalis, UTI | Ampicillin, amoxicillin | Fosfomycin, nitrofurantoin |
| Enterococcus faecium | ID consult recommended | Vancomycin, linezolid, daptomycin, eravacycline, omadacycline |
| MSSA | Nafcillin, oxacillin, cefazolin | Daptomycin, SMX/TMP, clindamycin, BL/BLI |
| MRSA, serious infection | Vancomycin, daptomycin | Linezolid, ceftaroline |
| MRSA, SSTI/CAP | Doxycycline, SMX/TMP | Clindamycin, linezolid, oritavancin, tedizolid, telavancin, dalbavancin |
| Group A strep (S. pyogenes) | Penicillin G (± clindamycin or linezolid) | Erythromycin, azithromycin, clarithromycin |
| Group B strep (S. agalactiae) | Penicillin G, ampicillin, amoxicillin | Cephalexin, clindamycin, vancomycin, azithromycin |
| Viridans group strep | Penicillin G | Ceftriaxone, cefotaxime, vancomycin, doxycycline |
| S. pneumoniae, penicillin-susceptible | Penicillin G, ampicillin, amoxicillin | Ceftriaxone, doxycycline |
| S. pneumoniae, penicillin-resistant | Ceftriaxone, vancomycin | Levofloxacin, moxifloxacin, vancomycin, linezolid, ceftaroline |
| Gram-Negative Cocci | ||
| Moraxella catarrhalis | Ampicillin/sulbactam, amoxicillin/clavulanate | SMX/TMP, doxycycline, azithromycin, ceftriaxone |
| Neisseria gonorrhoeae | Ceftriaxone | Gentamicin + azithromycin |
| Neisseria meningitidis | Penicillin G, ceftriaxone | Moxifloxacin, ampicillin |
| Gram-Positive Bacilli / Anaerobes | ||
| Clostridium perfringens | Penicillin G (± clindamycin) | Metronidazole, ceftriaxone, ampicillin, pip/tazo, carbapenem |
| Clostridioides difficile | PO vancomycin, fidaxomicin | Metronidazole (add-on for fulminant disease) |
| Bacteroides spp. | Metronidazole | BL/BLI, meropenem, imipenem, cefoxitin |
| Gram-Negative Bacilli | ||
| Escherichia coli | Ceftriaxone | Cefepime, BL/BLI, fluoroquinolone, SMX/TMP, cephalexin, nitrofurantoin (cystitis), carbapenem |
| Klebsiella pneumoniae | Ceftriaxone, BL/BLI | Cefepime, carbapenem, fluoroquinolone |
| Enterobacter spp. | Cefepime, meropenem, imipenem/cilastatin | SMX/TMP, amikacin, pip/tazo, fluoroquinolone, tigecycline |
| Proteus mirabilis | Ceftriaxone | Penicillin G, BL/BLI, cefepime |
| Acinetobacter spp. | Cefepime, meropenem, imipenem/cilastatin, ampicillin/sulbactam | Amikacin, fluoroquinolone, minocycline, pip/tazo, tigecycline, SMX/TMP |
| Pseudomonas aeruginosa | Cefepime, meropenem, amikacin, tobramycin, imipenem/cilastatin, pip/tazo | Ceftazidime, ciprofloxacin, levofloxacin, aztreonam |
| Haemophilus influenzae | Ampicillin/sulbactam, ceftriaxone (ampicillin alone if beta-lactamase negative) | SMX/TMP, azithromycin, fluoroquinolone, carbapenem |
| Legionella spp. | Levofloxacin, moxifloxacin, azithromycin | Erythromycin, ciprofloxacin |
| Salmonella typhi | Ceftriaxone | Ciprofloxacin, levofloxacin, SMX/TMP, carbapenem |
| Stenotrophomonas maltophilia | SMX/TMP | Minocycline, levofloxacin (never as monotherapy) |
| Miscellaneous | ||
| Chlamydia pneumoniae | Azithromycin, clarithromycin, doxycycline | Levofloxacin, moxifloxacin |
| Mycoplasma pneumoniae | Azithromycin, clarithromycin, doxycycline | Levofloxacin, moxifloxacin |
| Treponema pallidum | Penicillin G | Ceftriaxone |
| Multidrug-Resistant Gram-Negatives | ||
| ESBL Enterobacterales, non-urinary | Carbapenem | SMX/TMP, fluoroquinolone (if susceptible) |
| ESBL Enterobacterales, pyelonephritis | SMX/TMP | Carbapenem, levofloxacin, ciprofloxacin |
| ESBL Enterobacterales, cystitis | Nitrofurantoin | SMX/TMP, levofloxacin, ciprofloxacin |
| KPC-carbapenemase Enterobacterales | Meropenem/vaborbactam, ceftazidime/avibactam, imipenem/cilastatin/relebactam | Cefiderocol |
| Metallo-beta-lactamase Enterobacterales (VIM, NDM, IMP) | Ceftazidime/avibactam + aztreonam | Cefiderocol |
| OXA-48 Enterobacterales | Ceftazidime/avibactam | Cefiderocol |
| Pan-resistant Pseudomonas | Ceftolozane/tazobactam | Ceftazidime/avibactam, imipenem/cilastatin/relebactam, cefiderocol |
All of this assumes in-vitro susceptibility is confirmed. Oral step-down for the MDR gram-negatives is reasonable once (1) susceptibility to the oral agent is proven, (2) the patient is afebrile and hemodynamically stable, (3) source control has been achieved, and (4) there's no reason to doubt GI absorption.
Daptomycindoesn't achieve useful CNS concentrations, skip it for meningitis. Clindamycinis not an acceptable alternative for bloodstream or CNS infections. For Group A strep with toxin-mediated disease, both clindamycin and linezolid add antitoxin activity, the choice between them comes down to the patient in front of you.
Stacking antibiotics is a deliberate tradeoff, not a default "more is safer" move.
Combination therapy brings increased cost, more drug toxicity, and the risk of superinfection with an even more resistant organism. Some combinations are outright antagonistic, for example pairing a drug that induces beta-lactamase production with a partner drug that beta-lactamase happens to destroy.
Five organisms show up disproportionately on exams and on the wards. Know the pattern, not just the drug list.
Gram-positive cocci in clusters. Most often causes skin and soft tissue infections (cellulitis, abscess); less commonly pneumonia or bone/joint infections. Naturally resistant to most penicillins and cephalosporins. Colonization in healthy people, especially healthcare workers, is common and does not by itself need treatment.
| Setting | Options |
|---|---|
| Inpatient / serious | Vancomycin, daptomycin (not for pneumonia), ceftaroline, linezolid, tedizolid, delafloxacin, telavancin, oritavancin, dalbavancin, tigecycline |
| Outpatient | SMX/TMP, doxycycline, clindamycin, linezolid, tedizolid, delafloxacin |
Ceftarolineis the one cephalosporin (fifth generation) with real MRSA activity, otherwise assume the whole class misses it.
Gram-negative rod, ubiquitous in the environment, and capable of causing sepsis, pneumonia, UTI, and bone/joint infections.
Options: piperacillin/tazobactam, cefepime, ciprofloxacin, levofloxacin, ceftazidime, ceftazidime/avibactam, ceftolozane/tazobactam, aztreonam, aminoglycosides, imipenem/cilastatin, meropenem, doripenem, delafloxacin.
Every carbapenem covers Pseudomonas except ertapenem.That single exception is a favorite distractor: if a question needs Pseudomonas coverage and offers ertapenem as the answer, it's wrong.
E. faecalis and E. faecium, gram-positive cocci, mostly a hospital-acquired problem. Associated with UTIs, skin infections, and intra-abdominal infections.
Options: linezolid, tedizolid, daptomycin, tigecycline, and quinupristin/dalfopristin.
No cephalosporin covers Enterococcus, period, regardless of vancomycin susceptibility. And quinupristin/dalfopristin only works against vancomycin-resistant E. faecium, not E. faecalis, don't reach for it if the report says faecalis.
Anaerobic, gram-positive, spore-forming rod. Highest risk in patients 65 and older with recent antibiotic exposure, long-term care or hospital residents, and anyone immunosuppressed or with a prior C. diff episode.
| Agent | Role |
|---|---|
| PO vancomycin or fidaxomicin | First-line per 2021 SHEA/IDSA guidance, for both nonsevere and severe disease |
| IV metronidazole | Added on top of oral/rectal vancomycin for fulminantdisease, especially with ileus |
| Rifaximin | Option for recurrence after a course of PO vancomycin |
Clindamycin carries the single highest riskof triggering C. diff of any antibiotic class, with an odds ratio around 20 compared to no antibiotic exposure (most other classes run closer to 5). It carries a black box warning for exactly this reason. Tell patients to call if diarrhea persists more than a week after finishing any antibiotic course.
These are defined by the resistance mechanism, not the organism species, so the same E. coli or Klebsiella can fall into any row below depending on what enzyme it's carrying.
| Mechanism | Preferred agent | Alternative |
|---|---|---|
| ESBL, outside urinary tract | Carbapenem | SMX/TMP or fluoroquinolone if susceptible |
| ESBL, pyelonephritis | SMX/TMP | Carbapenem, levofloxacin, ciprofloxacin |
| ESBL, cystitis | Nitrofurantoin | SMX/TMP, levofloxacin, ciprofloxacin |
| KPC carbapenemase | Meropenem/vaborbactam, ceftazidime/avibactam, or imipenem/cilastatin/relebactam | Cefiderocol |
| Metallo-beta-lactamase (VIM, NDM, IMP) | Ceftazidime/avibactam + aztreonam | Cefiderocol |
| OXA-48 carbapenemase | Ceftazidime/avibactam | Cefiderocol |
Notice the pattern: as resistance mechanisms stack, the answer keeps sliding toward newer beta-lactam/beta-lactamase-inhibitor combinations or cefiderocol, the older single carbapenems stop being reliable once a carbapenemase enzyme is in play.
| Antimicrobial | Interacts with | Effect | What to do |
|---|---|---|---|
| Aminoglycosides | Neuromuscular blockers | Additive weakness | Avoid |
| Aminoglycosides | Other nephro/ototoxins (amphotericin B, cisplatin, cyclosporine, furosemide, NSAIDs, radiocontrast, vancomycin) | Additive renal/hearing toxicity | Monitor levels and renal function |
| Metronidazole | Ethanol | Disulfiram-like reaction | Avoid completely |
| Macrolides (clarithromycin, erythromycin) | Digoxin | Increased digoxin bioavailability and reduced clearance | Monitor digoxin level, avoid if possible |
| Macrolides / ciprofloxacin | Theophylline | Decreased theophylline metabolism | Monitor theophylline level |
| Fluoroquinolones | Class Ia/III antiarrhythmics | Additive QT prolongation | Avoid |
| Fluoroquinolones / tetracyclines | Multivalent cations (antacids, iron, calcium, zinc, sucralfate, dairy) | Decreased absorption of the antibiotic | Separate dosing by 2 hours |
| Rifampin | Azoles, cyclosporine, methadone, oral contraceptives, tacrolimus, warfarin | Induces metabolism of the other agent | Avoid if possible, adjust doses if not |
| Sulfonamides | Sulfonylureas, phenytoin, warfarin | Decreased metabolism of the other agent | Monitor glucose, phenytoin level, INR |
| Isoniazid | Carbamazepine, phenytoin | Decreased metabolism of the other agent | Monitor drug levels |
| Penicillins/cephalosporins | Probenecid | Blocks renal excretion of the beta-lactam | Can be used deliberately when you want a prolonged high level |
When a patient isn't responding, work through three buckets before assuming the drug was simply "wrong."
| Cause bucket | Specific failure modes |
|---|---|
| Not actually infectious/bacterial | Disease isn't infectious in origin, or there's an undetected second pathogen in a polymicrobial infection. Lab identification or susceptibility error is possible but rare. |
| Drug-related | Wrong drug, dose, or route. Malabsorption from GI disease (short-bowel syndrome) or a drug interaction (fluoroquinolone bound by multivalent cations). Accelerated elimination in cystic fibrosis or pregnancy, larger volumes of distribution and faster clearance especially hit aminoglycoside levels. Poor penetration into "privileged sites": CNS, eye, prostate. |
| Host-related | Immunosuppression (chemo-induced granulocytopenia, HIV/AIDS) means the patient's own defenses can't finish the job even with an adequate regimen. Unaddressed source control, an abscess that needs drainage, or necrotic tissue/foreign body that needs removal, will keep the infection going no matter what antibiotic you choose. |
| Organism-related | Primary (intrinsic) resistance in the infecting organism, or resistance that develops duringtreatment. Broad community and hospital overuse of antibiotics, plus long-term suppressive therapy in immunosuppressed patients, drives this at a population level. |
If there's an abscess or dead tissue sitting there, no antibiotic regimen fixes it alone. Source control comes first, drugs are what you use after (or alongside) draining/debriding, not instead of it.
| Parameter | When | Watching for |
|---|---|---|
| WBC count and differential | Serially through treatment | Trend back toward normal, resolving left shift |
| Temperature curve | Daily | Defervescence, sustained afebrile period |
| Signs/symptoms of infection | Every assessment | Resolving local findings, improving appetite |
| Culture and sensitivity results | As soon as available | Opportunity to narrow (de-escalate) therapy |
| Imaging | As clinically indicated | Resolution of the source, or a missed collection needing drainage |
| Antimicrobial concentrations | For narrow-therapeutic-index agents (aminoglycosides, vancomycin) | Therapeutic, non-toxic levels |
Overall clinical improvement, afebrile for 8–24 hours, a decreasing WBC count, and a functioning GI tract. Meet all four and there's rarely a reason to keep the line in. Always aim for the narrowest-spectrum agentthat still covers the identified pathogen once you have susceptibilities in hand.