Substance and claimed effects

Fluoroquinolones are a broad-spectrum synthetic antibiotic class derived from the original quinolone scaffold, defined by the addition of a fluorine atom that dramatically expanded gram-negative and atypical coverage. The class includes the agents most readers will be prescribed by name: ciprofloxacin, levofloxacin, moxifloxacin, and (less often now) ofloxacin, norfloxacin, gemifloxacin, and delafloxacin. Mechanism: inhibition of bacterial DNA gyrase (topoisomerase II) and topoisomerase IV, which lethally fragments bacterial DNA during replication Aldred et al. 2014. Coverage spans most aerobic gram-negatives (including Pseudomonas aeruginosa for ciprofloxacin and levofloxacin), atypicals (Legionella, Chlamydia, Mycoplasma), some gram-positives (moxifloxacin against S. pneumoniae), and intracellular pathogens (Bartonella, anthrax, plague, tularemia). Reader-facing concerns this entry covers holistically: (1) what these drugs actually treat well; (2) tendon rupture, especially Achilles; (3) aortic aneurysm and dissection; (4) peripheral neuropathy, sometimes permanent; (5) central nervous system reactions (insomnia, anxiety, delirium, seizures, suicidality); (6) other class effects (QT prolongation, dysglycemia, C. difficile colitis, photosensitivity); and (7) the prescribing posture readers should adopt when handed a script — push back, ask for alternatives, escalate concerns.

Evidence by addressing question

Mechanism

Fluoroquinolones kill bacteria by trapping DNA gyrase and topoisomerase IV in covalent complexes with cleaved DNA; the unrepaired double-strand breaks are bactericidal Aldred et al. 2014. Selective toxicity comes from the structural differences between bacterial gyrase and the mammalian topoisomerases — but the gap is not absolute. At high intracellular concentrations these drugs also inhibit human mitochondrial topoisomerase II, impairing mitochondrial DNA replication and producing reactive oxygen species Hangas et al. 2018. This off-target mitochondrial hit is the leading mechanistic hypothesis for the disabling, multi-system, post-exposure syndrome that the literature now labels fluoroquinolone-associated disability (FQAD) — tendon, peripheral nerve, CNS, and connective-tissue toxicity emerging in patients without obvious pre-existing risk Michalak et al. 2017.

Tendon-specific mechanism: ciprofloxacin and other fluoroquinolones up-regulate matrix metalloproteinase-2 in tenocytes, accelerating degradation of type I collagen in tendon tissue Tsai et al. 2011. The same MMP-driven extracellular-matrix degradation is plausibly load-bearing for the aortic aneurysm and dissection signal — aortic media is collagen-elastin scaffold, and the class signature on aortic tissue mirrors the tendon signature. CNS mechanism: fluoroquinolones antagonize GABA-A receptor binding and potentiate NMDA receptors, lowering seizure threshold and producing the agitation / insomnia / delirium phenotype seen across all class members Tomé and Filipe 2011.

Evidence — bacterial coverage

The class is genuinely effective against the pathogens it covers — there is no disagreement about whether fluoroquinolones cure susceptible infections. Excellent oral bioavailability (~99% for ciprofloxacin and levofloxacin) lets them substitute one-for-one with IV therapy in many clinical scenarios. Strong indications where they remain first-line or near-first-line in guidelines: complicated UTI and acute pyelonephritis, bacterial prostatitis (one of the few drug classes that penetrates prostatic tissue), traveler's diarrhea in regions without high resistance, anthrax post-exposure prophylaxis, plague, multidrug-resistant tuberculosis, gonorrhea (where susceptibility documented), Pseudomonas infections in non-cystic-fibrosis patients Stahlmann and Lode 2010. The 2016 FDA action specifically restricted use for uncomplicated UTI, acute sinusitis, and acute bronchitis exacerbations — not because the drugs don't work, but because the risk-benefit balance shifts against them when alternatives exist FDA 2016.

Evidence — tendon rupture and tendinopathy

Largest signal in the class. Multiple population-based cohort and case-control studies, replicated across countries and decades. Khaliq and Zhanel's review pooled the early data and quantified relative risk of Achilles tendon rupture at roughly 3–4× background in current users, with the highest absolute incidence in patients over 60 and those on concurrent corticosteroids Khaliq and Zhanel 2003. The Danish national cohort confirmed an adjusted rate ratio for Achilles rupture of 3.1 (95% CI 2.0–4.8) in current fluoroquinolone users versus the background population Sode et al. 2007. The Italian case-control study by Corrao et al. found roughly 4× increased odds with greatest effect in the first month after exposure Corrao et al. 2006. Persson and Jick's analysis of the UK CPRD database quantified the magnitude of the corticosteroid interaction: tendon rupture rate jumps from ~1 per 10,000 prescriptions in fluoroquinolone alone to ~10 per 10,000 with concurrent oral steroids, a 9-fold further elevation on top of the class-baseline risk Persson and Jick 2019. Daneman's longitudinal Ontario cohort tied the same signal to other collagen-tissue events including retinal detachment Daneman et al. 2015. Time course: onset typically within days to weeks of starting therapy, with the Achilles the dominant site (rotator cuff, patellar, biceps, hand tendons also documented). Athletes are a particularly important population — the rupture can occur with normal training load, sometimes weeks after the course ends Lewis and Cook 2014. Bidell and Lodise's pharmacovigilance analysis suggested levofloxacin carries the highest tendinopathy signal among the modern agents, but the FDA black-box treats this as a class effect across all systemic fluoroquinolones Bidell and Lodise 2016.

Evidence — aortic aneurysm and dissection

Second major collagen-tissue signal, recognized later than tendon. Lee et al.'s Taiwanese case-control study first put the signal in JAMA Internal Medicine: adjusted odds ratio 2.43 (95% CI 1.83–3.22) for aortic aneurysm or dissection during current fluoroquinolone use, with greatest effect at high cumulative dose Lee et al. 2015. Pasternak et al. replicated in the Swedish national registry: hazard ratio 1.66 (95% CI 1.12–2.46) for aortic aneurysm or dissection in fluoroquinolone users versus amoxicillin users within 60 days of treatment start Pasternak et al. 2018. Lee 2018's prospective JACC analysis added the dose-response dimension and the post-exposure tail (elevated risk persisting weeks after the course ends) Lee et al. 2018. The signal triggered the 2018 FDA Drug Safety Communication restricting use in patients with aortic aneurysm, known atherosclerotic vascular disease, hypertension, Marfan syndrome, Ehlers-Danlos syndrome, and the elderly FDA 2018.

The picture is not unanimous. Gopalakrishnan et al.'s active-comparator analysis using US Medicare data — designed to control for the confounding-by-indication that may inflate the signal (sicker patients are more likely to receive fluoroquinolones and more likely to have undiagnosed aortic disease) — found a more modest hazard ratio of 1.20 (95% CI 1.17–1.24) versus azithromycin, suggesting prior estimates may overstate the absolute effect Gopalakrishnan et al. 2020. Brown et al.'s 2023 JAMA Cardiology study within UK CPRD found a relative incidence of 2.09 versus co-amoxiclav in the first 60 days but no increase versus a non-antibiotic comparator group, raising the prospect that part of the observed effect is the underlying-infection confounder rather than the drug itself Brown et al. 2023. Author's working position (carried into the article): the signal is real and the mechanism (collagen-elastin matrix degradation) is plausible from tendon parallels, but the absolute event rate is small and the relative risk is in the 1.2–2.5x range, not 5–10x. Patients with known aortic risk factors should treat the contraindication as hard; the broader population should weight this signal as one input alongside the dominant tendon and CNS signals.

Evidence — peripheral neuropathy

Etminan et al.'s pharmacoepidemiologic study in Neurology using a US claims database found roughly 1.83x increased risk of incident peripheral neuropathy in current oral fluoroquinolone users versus non-users (95% CI 1.49–2.27) Etminan et al. 2014. The 2013 FDA action upgraded the label warning to flag that the neuropathy can be permanent — onset within days, persistence over years FDA 2013. Phenotype: distal symmetric, axonal sensory or sensorimotor neuropathy; burning, tingling, allodynia, numbness; weakness in severe cases. Some patients recover over months once the drug is stopped; a minority do not. Mechanism unclear; mitochondrial dysfunction in dorsal-root-ganglion neurons is the leading hypothesis, consistent with the broader FQAD mitochondrial-toxicity framework Michalak et al. 2017. The FDA explicitly recommends stopping the drug at the first neuropathy symptom and switching to a non-fluoroquinolone FDA 2013.

Evidence — CNS reactions

Tomé and Filipe's pharmacovigilance review of EU and US adverse-event databases catalogued the spectrum: insomnia (most common, often dose-related), anxiety, agitation, confusion, delirium, hallucinations, psychosis, depression, mania, suicidal ideation, and seizures Tomé and Filipe 2011. Mechanism: GABA-A receptor antagonism and NMDA receptor potentiation lower seizure threshold and produce excitatory CNS effects. Most reactions resolve on discontinuation, but case reports document persisting psychiatric symptoms (depression, anxiety) for months to years post-exposure, especially in the FQAD population Michalak et al. 2017. The 2016 FDA action added "mental health side effects" as a black-box, including disturbances in attention, disorientation, agitation, nervousness, memory impairment, and serious disturbances in mental abilities (delirium) FDA 2016. Seizure risk is highest in patients with prior seizure history, structural CNS lesions, renal impairment (drug accumulation), and concurrent NSAID use (which potentiates the GABA-A antagonism).

Evidence — other class effects

Dysglycemia. Chou et al.'s Taiwanese cohort of diabetic patients found significantly increased risk of severe hypoglycemia and hyperglycemia requiring hospitalization with all three of levofloxacin, ciprofloxacin, and moxifloxacin versus macrolides, with moxifloxacin showing the strongest hypoglycemia signal Chou et al. 2013. The 2018 FDA action formalized this as a class-wide warning, particularly hazardous in diabetic patients on insulin or sulfonylureas FDA 2018. Mechanism: fluoroquinolones block ATP-sensitive potassium channels in pancreatic beta cells, triggering insulin release.

QT prolongation. All fluoroquinolones prolong the QT interval to varying degrees (moxifloxacin > levofloxacin > ciprofloxacin). Clinically significant torsades de pointes is uncommon but documented; risk concentrates in patients with congenital long-QT, electrolyte derangements (hypokalemia, hypomagnesemia), or concurrent QT-prolonging drugs (antiarrhythmics, certain antipsychotics, methadone) Stahlmann and Lode 2010.

Clostridioides difficile colitis. Pépin et al. documented fluoroquinolones as the dominant antibiotic risk factor for CDI during the Quebec outbreak of the hypervirulent NAP1/027 strain, with adjusted hazard ratio ~3.4 versus non-exposed; the class overtook clindamycin and cephalosporins as the leading driver of healthcare-associated CDI in many North American centers Pépin et al. 2005. The risk persists for weeks after the course ends as gut microbiome recovers.

Photosensitivity, drug interactions. Fluoroquinolones cause UVA-mediated phototoxic skin reactions (lomefloxacin and sparfloxacin worst; ciprofloxacin and levofloxacin milder but real). Major interactions: divalent and trivalent cations (calcium, magnesium, iron, aluminum — antacids, dairy products, multivitamins) chelate the drug and gut absorption drops by 50–90%; warfarin (INR elevation, bleeding); theophylline (toxicity from CYP1A2 inhibition); sulfonylureas and insulin (hypoglycemia); NSAIDs (potentiated CNS excitation, lower seizure threshold) Stahlmann and Lode 2010.

Contraindications

Hard contraindications (do not give): known hypersensitivity; concurrent tizanidine (ciprofloxacin); myasthenia gravis (boxed warning — fluoroquinolones can precipitate severe exacerbation including respiratory failure) FDA 2016. Pediatric use is restricted given the historical concern about cartilage toxicity from animal data, though human evidence is more reassuring than initially feared; FDA permits use in children for narrow indications (complicated UTI, anthrax exposure, plague) when the benefit clearly outweighs alternatives. Pregnancy: avoid unless no alternative; same cartilage-development concerns. Lactation: minimal data; generally avoided. Strong relative contraindications drawn from the regulatory actions: known aortic aneurysm or family history; uncontrolled hypertension; connective tissue disease (Marfan, Ehlers-Danlos); active or prior tendinopathy; concurrent systemic corticosteroids; prior fluoroquinolone-associated adverse event; age over 60 with multiple risk factors; renal impairment requiring dose adjustment; congenital long-QT syndrome or concurrent QT-prolonging drugs FDA 2018 EMA 2019.

Misconceptions

The dominant misconception in lay use: "it's just another antibiotic." The black-box warnings make this class structurally different from amoxicillin or doxycycline — the European Medicines Agency formally restricted the class in April 2019 for use in mild to moderate infections, citing the disabling and potentially permanent adverse-effect profile EMA 2019. A second misconception: that tendon and other side effects are dose-dependent or only manifest after long courses. The cohort data show events within days of starting therapy, including 5-day courses for uncomplicated UTI Khaliq and Zhanel 2003. Third: that side effects resolve fully after discontinuation. Most do; a minority — the FQAD population — develop persistent multi-system symptoms lasting months to years Michalak et al. 2017. Fourth misconception, on the prescriber side: that the FDA "reserve for no-alternative" language is regulatory caution rather than a binding clinical posture. The 2016 communication is unusually specific that for the three indications named (uncomplicated UTI, acute sinusitis, acute bronchitis) the risks outweigh the benefits and fluoroquinolones should not be used when alternatives exist FDA 2016.

Audience — who is at especially high risk

The cohort data identifies several populations where the risk-benefit calculation tilts dramatically against fluoroquinolones: adults over 60 (3–4× elevated tendon rupture rate, also dominant aortic risk group); athletes and physically active adults (rupture during ordinary training load, sometimes weeks post-course) Lewis and Cook 2014; patients on systemic corticosteroids (~9× tendon rupture risk on top of class baseline) Persson and Jick 2019; patients with known or suspected aortic aneurysm, atherosclerotic vascular disease, hypertension, Marfan, or Ehlers-Danlos FDA 2018; myasthenia gravis (precipitation of crisis); diabetics on insulin or sulfonylureas (severe dysglycemia) Chou et al. 2013; patients with seizure disorders or structural CNS lesions; patients with prior fluoroquinolone-associated adverse event; renal impairment (drug accumulation, CNS risk).

Alternatives

The 2016 FDA action is explicit that for the three named indications — uncomplicated UTI, acute sinusitis, acute bronchitis — non-fluoroquinolone alternatives should be used when available FDA 2016. Reasonable substitutes by indication: uncomplicated UTI → nitrofurantoin 5 days, trimethoprim-sulfamethoxazole 3 days, or fosfomycin single dose (IDSA guidance); pyelonephritis → IV ceftriaxone followed by oral cephalosporin or TMP-SMX guided by culture; sinusitis → amoxicillin-clavulanate or doxycycline (or watchful waiting — most acute sinusitis is viral); bronchitis exacerbation → consideration of whether antibiotic is needed at all; community-acquired pneumonia → amoxicillin / amoxicillin-clavulanate plus a macrolide for atypicals, with doxycycline as alternative (the role of respiratory fluoroquinolones — levofloxacin, moxifloxacin — narrowed in the latest ATS/IDSA guidance); traveler's diarrhea → azithromycin where fluoroquinolone resistance is high (Southeast Asia); gonorrhea → ceftriaxone (fluoroquinolone resistance is widespread). Where fluoroquinolones remain genuinely first-line: bacterial prostatitis (few alternatives penetrate prostatic tissue), Pseudomonas infections in selected settings, anthrax / plague exposure, multidrug-resistant tuberculosis as part of a regimen.

Failure modes

Most common preventable harm patterns: (1) Prescribed for an indication with adequate alternatives (uncomplicated UTI is the modal case) without that decision being made explicitly. (2) Co-administered with oral corticosteroids without recognition of the ~9× tendon rupture risk amplification Persson and Jick 2019. (3) Continued through emergent neuropathy or tendon pain instead of stopped at first symptom FDA 2013. (4) Taken concurrently with calcium / magnesium / iron supplements or dairy products at the same time as the dose, causing absorption to drop 50–90% and therapeutic failure. (5) Prescribed for diabetes patients on sulfonylureas or insulin without dysglycemia monitoring Chou et al. 2013. (6) Continued physical training during and immediately after a course; ruptures during ordinary load have been reported weeks post-discontinuation Lewis and Cook 2014. (7) Prescribed by clinicians who treat the class as equivalent to amoxicillin; the asymmetry of the risk-benefit profile is non-obvious unless the prescriber has internalized the FDA 2016 / 2018 black-box language and the EMA 2019 restriction.

Stakes

Felt-experience anchor for the patient who accepts the prescription without pushing back, in the typical case: nothing. Most short courses (3–7 days for uncomplicated UTI or sinusitis) finish without overt adverse events. The signal lives in the tail: per the BMJ Open cohort, tendon rupture incidence is on the order of 1–5 per 10,000 prescriptions in the general population and rises 10× in patients on concurrent steroids; aortic aneurysm/dissection incidence is on the order of 1–10 per 10,000; permanent peripheral neuropathy is rarer still but documented in the thousands of FDA MedWatch reports compiled in the 2013–2018 actions Daneman et al. 2015. The asymmetric harm is what justifies the avoidance posture: when an equally effective alternative exists, accepting fluoroquinolones means accepting a small chance of a disabling, potentially permanent injury that wouldn't have happened otherwise. The FQAD population (patients with persistent multi-system disability after a single course) numbers in the thousands of documented case reports and was the explicit subject of the 2016 FDA boxed warning FDA 2016.

Out of scope

This entry is about the fluoroquinolone class itself; adjacent topics the reader may want to look into separately: general antibiotic stewardship (when is any antibiotic warranted); the gut microbiome / C. difficile consequences of broad-spectrum exposure; specific indications (uncomplicated UTI management; community-acquired pneumonia); the FQAD post-treatment syndrome (a future entry candidate if the literature continues to consolidate).

The credibility range

Optimist case

The optimist position on fluoroquinolones (defended in some infectious-disease clinical contexts and in segments of the prescribing community): these are the most bioavailable, best-penetrating oral antibiotics ever developed; for pyelonephritis, bacterial prostatitis, Pseudomonas infections, and selected respiratory infections they remain near-irreplaceable. The black-box adverse events are real but rare in absolute terms — Achilles rupture on the order of 1 per few-thousand prescriptions, aortic events on the order of 1 per low-thousands, severe neuropathy rarer still. The 2016 FDA "reserve" language overstates the risk-benefit shift for patients with the relevant indications because the alternative agents (TMP-SMX, nitrofurantoin, amoxicillin-clavulanate) carry their own meaningful adverse-event profiles (TMP-SMX hyperkalemia and Stevens-Johnson; nitrofurantoin pulmonary fibrosis on long-term use) and resistance to the alternatives is rising. The FQAD syndrome, the optimist argues, is a heterogeneous patient-reported phenomenon without a defined biomarker, conflating drug-attributable injury with unrelated incident illness and nocebo-amplified symptom clusters. Confounding by indication contaminates the aortic-aneurysm signal: patients who receive fluoroquinolones for severe UTI or pyelonephritis are sicker and more vascularly aged than the comparator population, which is why active-comparator analyses (Gopalakrishnan 2020) attenuate the effect Gopalakrishnan et al. 2020.

Skeptic case

The skeptic position (carried by patient-advocacy organizations, the FDA's 2016 boxed-warning panel, the EMA's 2019 restriction): the class shows a coherent signature of connective-tissue and mitochondrial injury — tendon, aorta, sclera, peripheral nerve, autonomic nervous system — replicated across multiple study designs and registries on multiple continents. The biological mechanism (mitochondrial topoisomerase inhibition, MMP up-regulation, GABA-A antagonism) is plausible and increasingly characterized at the molecular level. The fact that absolute rates are low does not mean the harm is low when integrated over tens of millions of prescriptions annually — even a 1-in-10,000 disabling event accumulates to thousands of disabled patients per year. The disabling cases are not rare enough to be dismissed as background noise; the FDA's 2016 action was explicitly responding to a cumulative adverse-event signal that no single trial would have surfaced. The "alternatives have their own side effects" rebuttal is true but irrelevant when comparing class profiles: nitrofurantoin does not cause aortic dissection or permanent peripheral neuropathy. The widespread prescribing of fluoroquinolones for uncomplicated UTI (where 3 days of TMP-SMX or 5 days of nitrofurantoin work as well) is the clearest case of negligent overuse in modern outpatient prescribing. FQAD as a syndrome may be heterogeneous, but the FDA and EMA both treat it as real enough to drive label changes.

Author's call

The convergent regulatory action across FDA (2008, 2013, 2016, 2018) and EMA (2019), the replicated cohort signals across countries and decades, and the biologically plausible mechanism push the entry toward an avoidance posture with explicit carve-outs for indications where the class remains first-line. The article's frame for the reader: this is a "decide with your clinician" entry, not a "do" or "avoid" reflex. The default posture when handed a fluoroquinolone prescription should be to ask: is there an alternative for this indication, and if not, do I match the high-risk audience criteria that should push us further toward alternative or specialist consultation? Meta evidence is 5 (replicated cohorts, regulatory action, mechanism); meta controversy is 3 (active debate over the magnitude of the aortic signal, the legitimacy of FQAD as a defined syndrome, and the appropriate boundaries of "reserve" prescribing). The benefit dimensions are mostly 0 — fluoroquinolones don't lift mood, energy, focus, sleep, or beauty; the health_short_term dimension gets a small positive score (1) reflecting that the drug does cure the infection it's prescribed for in the narrow subset of cases where it's the right choice, but this is not the dimension the article leans on.

Stakeholder and incentive map

• Generic manufacturers — the patents on ciprofloxacin and levofloxacin are long expired; the class is cheap, generic, and high-volume. Manufacturer marketing pressure is essentially zero now. The persistent overprescribing is driven by clinician habit and convenience, not active marketing.
• Prescribing clinicians — fluoroquinolones are convenient (once-daily dosing, broad coverage, 99% bioavailability so oral-IV switch is seamless, low cost). The convenience is the prescribing driver. Adopting the FDA 2016 posture requires retraining a habit; not all primary care has caught up.
• Patient advocacy — patient-led organizations (Quinolone Vigilance Foundation, Floxie Hope, the "floxed" community on social media) have been a meaningful force pushing the FDA on FQAD recognition. Their messaging skews catastrophic, but the underlying signal is real and was the proximate driver of the 2016 boxed warning.
• Regulatory bodies — FDA has taken sequential restrictive actions (2008 tendon, 2013 neuropathy, 2016 disabling/MH, 2018 aorta/dysglycemia); EMA followed in 2019 with a class-level restriction. Health Canada and TGA Australia have similar restrictions. The regulatory direction is unambiguous.
• Counter-pressure — IDSA and certain ID specialists pushed back on aspects of the 2016 FDA action, arguing the language was too broad and could compromise treatment of indications where alternatives are inferior (e.g., pyelonephritis in some practice settings). The pushback shaped the carve-out language but didn't reverse the restriction.

Population variability

• Age — tendon rupture and aortic risk both rise sharply over 60; the EMA explicitly flags elderly use as a relative contraindication EMA 2019.
• Concurrent corticosteroid — tendon rupture rate is ~9× higher with concurrent oral steroids; this is the single largest amplifier and a near-absolute relative contraindication Persson and Jick 2019.
• Connective tissue disease — Marfan, Ehlers-Danlos, prior aortic aneurysm: hard avoidance population.
• Diabetes — severe dysglycemia risk; particularly in patients on insulin or sulfonylureas; moxifloxacin worst offender for hypoglycemia Chou et al. 2013.
• Renal function — fluoroquinolones are renally eliminated; impaired clearance accumulates the drug and amplifies CNS risk (seizure, delirium). Dose adjustment per creatinine clearance is mandatory.
• Myasthenia gravis — boxed-warning contraindication; can precipitate respiratory crisis.
• Athletes — case series document tendon rupture during ordinary training load, sometimes weeks after the course; non-athletes are also at risk but symptoms more likely to be attributed to overuse rather than the drug Lewis and Cook 2014.
• Pediatric — historical concern about cartilage toxicity (from juvenile animal data); current human evidence more reassuring; narrow approved indications.
• Pregnancy and lactation — generally avoided.
• Genetic polymorphism — case series of FQAD patients have looked at variants in mitochondrial function, antioxidant response, and connective-tissue genes; no defined biomarker yet identifies the susceptible subpopulation.

Knowledge gaps

• FQAD biomarker. No prospective study identifies the patients who will develop the disabling multi-system phenotype before exposure. Without it, the avoidance posture has to be applied class-wide rather than precision-targeted to the susceptible population.
• Aortic signal magnitude. Active-comparator and non-antibiotic-comparator analyses (Gopalakrishnan 2020, Brown 2023) attenuate the signal versus the original case-control estimates. The "true" hazard ratio adjusted for confounding by indication is uncertain — likely in the 1.2–2.0 range rather than the 2.4 of Lee 2015, but exact pinning would require a prospective trial unlikely to be ethical.
• Reversibility of neuropathy. Proportion of patients who recover fully versus those left with persistent symptoms remains undefined; case series and the FDA 2013 action treat "potentially permanent" as the operating assumption, but a rigorous prevalence estimate doesn't exist.
• Mitochondrial mechanism quantification. The mitochondrial topoisomerase inhibition story is biologically clean but the dose-response in human tissue (versus in vitro at supratherapeutic concentrations) is not well established.
• Pediatric long-term tendon data. The earliest concerns about juvenile cartilage toxicity have not fully resolved in long-term follow-up of pediatric users; data are still accumulating.
• What would change the call: a prospective biomarker for FQAD susceptibility (would enable precision avoidance), a head-to-head trial of fluoroquinolone vs. alternative for the high-volume indications showing equivalent efficacy and lower harm (would consolidate the alternative-first posture), or a rigorous mechanism-targeted antidote (would shift the risk-benefit balance back toward acceptance).

Scope vs. brief. Brief named: bacterial coverage, tendon rupture risk, aortic aneurysm and dissection risk, peripheral neuropathy, CNS reactions, and prescribing precautions. All six are covered end-to-end. The article adds the C. difficile / dysglycemia / QT signals that round out the class's adverse-effect profile, since they bear directly on the "alternatives-first" posture the brief implies.

Action choice. Picked decide over avoid deliberately. The class still earns its prescription for bacterial prostatitis, complicated UTI, pyelonephritis, Pseudomonas, anthrax exposure, MDR-TB. A blanket avoid would mis-set the reader's posture for those cases. decide captures the "is there an alternative for this indication" framing the FDA 2016 communication itself uses.

Rating difficulties. health_short_term was the hardest call. Setting it to 0 felt dishonest — when the drug is correctly indicated it does cure the infection. Setting it to 2 or higher misrepresents the entry's centre of gravity, which is the alternatives-first push. Landed on 1: trivially positive, not the reason you'd recommend it. The pitch makes the contingency explicit. Longevity stayed at 0 — there's no positive longevity effect for the typical reader scenario; the tail of aortic dissection, severe CDI, and disabling FQAD can shorten lives but a benefit dimension can't be negative.

Controversy = 3. Reflects three live debates: (1) the magnitude of the aortic signal after Gopalakrishnan 2020 and Brown 2023 brought it down via active-comparator design; (2) the legitimacy of FQAD as a defined syndrome (no biomarker, heterogeneous case definitions); (3) the appropriate boundaries of the FDA "reserve" posture in clinical practice (IDSA pushback in 2016 was real). Not 4 — the existence of the harms is not in dispute; the regulators have acted.

Contraindications field. Included autoimmune for myasthenia gravis (the boxed-warning case); cardiac-condition for the QT and aortic-aneurysm signals; blood-thinners for the warfarin-INR interaction; kidney-disease for the renal-accumulation / CNS-toxicity dose adjustment; pregnancy and breastfeeding per the cartilage-development concern. Did not include diabetes-medication as a token because the closed vocabulary doesn't have one (the spec lists insulin/sulfonylureas under the diabetes-medication interaction risk but Chou 2013 evidence is foregrounded inline in the article and audience section instead).

FQAD framing. The article names the syndrome once in the evidence section but doesn't lean on it as the centre of the case. Reason: the syndrome is real enough that the FDA, EMA, and a substantial Michalak-style mechanism literature take it seriously, but it remains heterogeneous and case-defined rather than biomarker-defined. Building the entry's argument primarily on FQAD would expose it to a legitimate skeptic-side rebuttal; building it on the four named black-box endpoints is uncontroversial.

Excluded but considered. Retinal detachment signal (Etminan 2012 controversial, later studies attenuating); pediatric cartilage concerns (relevant to a small slice of readers); the in-detail pharmacokinetic table; specific dose-by-indication tables (clinician-territory, not reader-actionable). The pediatric note belongs in a future entry on pediatric antibiotic use if one is commissioned.

Future-link candidates. antibiotic-stewardship (general — when an antibiotic is warranted at all); c-difficile-colitis (downstream consequence of broad-spectrum exposure); uncomplicated-uti-management (the modal indication this entry pushes readers to redirect away from FQs on); community-acquired-pneumonia (the indication where respiratory FQs are narrowing in guidelines); antibiotic-microbiome-recovery (post-course gut recovery, where this class is one of the worst offenders).

Separate-entry candidate. Fluoroquinolone-associated disability (FQAD) is borderline. If the syndrome consolidates with a biomarker over the next 5 years, it warrants its own entry; for now the cluster sits inside this entry's evidence section.