Historical Signal Evaluation: Fluoroquinolones and Aortic Aneurysm or Dissection

Introduction

The association between systemic fluoroquinolone exposure and aortic aneurysm or aortic dissection is a useful example of how a pharmacovigilance signal can evolve when the evidence is concerning but does not initially provide a definitive causal estimate.

The signal developed primarily through epidemiological studies rather than through a single decisive clinical trial. Observational studies reported an increased risk of aortic aneurysm or dissection following fluoroquinolone exposure, while non-clinical evidence provided biological plausibility. The evidence was then considered by the European Pharmacovigilance Risk Assessment Committee (PRAC), which recommended changes to product information in 2018.

The case is particularly instructive because the strength of the evidence has not been identical for all outcomes or across all studies.

Some analyses have reported increased risks for the combined outcome of aortic aneurysm or dissection. Other analyses have found stronger evidence for aortic aneurysm than for aortic dissection. Later systematic reviews have also highlighted substantial limitations in the observational evidence, including residual confounding and low or very low certainty in some analyses.

This makes the case valuable for pharmacovigilance professionals.

The appropriate question is not simply:

Do fluoroquinolones cause aortic aneurysm or dissection?

A more useful signal-evaluation question is:

What did the evidence demonstrate at each stage, what uncertainties remained, and was the regulatory response proportionate to the evidence available at the time?

Why This Is a Useful Signal-Evaluation Case

A historical signal evaluation should reconstruct the evidence as it developed.

The key questions are:

1. What was the safety concern?
2. What evidence first suggested an association?
3. How strong was the initial evidence?
4. What alternative explanations existed?
5. Was the association reproduced?
6. Was there a credible temporal relationship?
7. Was there biological plausibility?
8. Were particular patients at greater risk?
9. Did later evidence strengthen or weaken the association?
10. How did regulators interpret the totality of evidence?
11. What changed in product information?
12. What should a pharmacovigilance professional learn from the case?

The fluoroquinolone example is particularly useful because it demonstrates that regulatory action and scientific certainty are not identical concepts.

A regulator may determine that the available evidence justifies a warning or other risk-minimisation measure even though uncertainty remains about the exact magnitude of risk or the contribution of alternative explanations.

1. Define the Signal Correctly

The regulatory signal concerned fluoroquinolones for systemic and inhaled use and the potential association with:

• aortic aneurysm;
• aortic dissection.

The distinction between these outcomes matters.

An aortic aneurysm is an abnormal dilation or enlargement of the aorta. Aortic dissection involves a tear in the aortic wall with separation of its layers.

They are related vascular conditions, but they are not interchangeable clinical outcomes.

A signal evaluation that combines them without examining them separately can therefore obscure important differences in the evidence.

The fluoroquinolones included in the PRAC signal assessment were:

• ciprofloxacin;
• flumequine;
• levofloxacin;
• lomefloxacin;
• moxifloxacin;
• norfloxacin;
• ofloxacin;
• pefloxacin;
• prulifloxacin;
• rufloxacin.

The regulatory assessment was conducted at class level.

This creates an important pharmacovigilance question:

Was the available evidence sufficiently consistent to justify a class-level regulatory response?

That question should be distinguished from the separate question of whether the magnitude of risk was identical for every individual fluoroquinolone.

2. Why the Signal Was Plausible

Fluoroquinolones had an established association with certain connective-tissue-related adverse reactions, particularly tendinopathy and tendon rupture.

This raised a mechanistic question about whether effects on connective tissue or extracellular matrix could also affect the structural integrity of the aortic wall.

Non-clinical evidence subsequently provided additional biological plausibility.

PRAC considered non-clinical evidence alongside epidemiological evidence when evaluating the signal.

Biological plausibility does not establish clinical causality by itself.

Its value is supportive.

A mechanistic hypothesis becomes more persuasive when it is consistent with:

• clinical observations;
• temporal association;
• epidemiological findings;
• exposure patterns;
• susceptible populations;
• and other independent evidence.

The fluoroquinolone case therefore illustrates an important principle:

Mechanistic plausibility should strengthen an evidence assessment, not replace it.

3. Early Epidemiological Evidence

The signal did not arise from one definitive study.

Several observational investigations reported an association between fluoroquinolone exposure and aortic aneurysm or dissection.

Among the influential studies considered during the regulatory assessment were epidemiological studies by Lee and colleagues, Daneman and colleagues, and Pasternak and colleagues.

These studies were important because aortic aneurysm and dissection are relatively uncommon outcomes.

Randomised clinical trials of antibiotics are generally not designed or powered to detect small increases in rare vascular events.

Observational databases therefore became an important source of evidence.

However, observational studies also create important limitations.

Patients receiving fluoroquinolones are not randomly assigned to treatment.

They may differ from comparator patients with respect to:

• age;
• infection severity;
• cardiovascular disease;
• vascular risk;
• healthcare utilisation;
• concomitant medicines;
• comorbidity;
• hospitalisation.

These differences can produce confounding.

Therefore:

An adjusted epidemiological association is evidence of an association. It is not automatically evidence that the exposure caused the outcome.

That distinction remains essential throughout the evaluation.

4. The 2017 Meta-Analysis

By 2017, enough observational evidence had accumulated for a systematic review and meta-analysis.

Singh and Nautiyal reviewed controlled studies available through February 2017.

Their analysis reported statistically significant associations between fluoroquinolone exposure and both:

• aortic dissection;
• aortic aneurysm.

The pooled estimates were:

These findings increased concern because the estimated associations were substantial.

However, the analysis was based on a small number of observational studies.

The result therefore needed to be interpreted in the context of:

• study design;
• heterogeneity;
• outcome definition;
• comparator selection;
• residual confounding;
• exposure classification;
• and the rarity of the outcomes.

A pooled observational estimate does not eliminate the limitations of the underlying studies.

The appropriate interpretation was therefore:

The available evidence provided a credible safety signal requiring further investigation.

It was not:

The meta-analysis definitively proved that fluoroquinolones cause aortic aneurysm and dissection.

5. Temporal Association

Temporal association is an important component of any drug-event evaluation.

Several studies examined the risk of aortic events following relatively recent fluoroquinolone exposure.

This is relevant because a very long interval between exposure and outcome would make a direct causal interpretation less persuasive unless there were a compelling biological explanation.

The later systematic-review literature generally focused on relatively short exposure windows, particularly 30- and 60-day periods.

For example, a 2019 systematic review and meta-analysis of four controlled observational studies reported more than a two-fold increased risk of the combined outcome within 60 days of fluoroquinolone exposure.

The adjusted relative risk was:

2.14 (95% CI 1.93–2.36).

For aortic aneurysm alone, the reported adjusted relative risk was:

2.23 (95% CI 2.01–2.45).

The analysis did not demonstrate a statistically significant association for aortic dissection alone.

That distinction is important.

It would therefore be inappropriate to reduce the evidence to the statement:

Fluoroquinolones double the risk of aortic dissection.

The evidence was more specific than that.

The combined endpoint was associated with increased risk, while the evidence for the individual components differed.

6. Aneurysm Versus Dissection

The distinction between aneurysm and dissection became increasingly important as additional evidence accumulated.

The 2019 meta-analysis found a significant association for aortic aneurysm but did not find a statistically significant association for aortic dissection alone.

A 2022 systematic review and meta-analysis including more than 53 million participants similarly reported:

• aortic aneurysm: HR 1.84 (95% CI 1.10–2.48);
• aortic dissection: HR 1.09 (95% CI 0.96–1.25).

The authors concluded that fluoroquinolone use was more strongly associated with aortic aneurysm than with aortic dissection.

These later results do not invalidate the earlier signal.

They refine it.

This is exactly what signal evaluation should accomplish.

A signal may initially be defined broadly because the available evidence is limited.

As evidence accumulates, the phenotype may become:

• narrower;
• more specific;
• differently quantified;
• or differently interpreted.

A good pharmacovigilance system should allow that refinement.

7. Confounding by Indication

Confounding by indication is one of the most important methodological concerns in this signal.

Fluoroquinolones are prescribed to treat infections.

The underlying infection may itself be associated with:

• systemic inflammation;
• physiological stress;
• hospitalisation;
• changes in cardiovascular risk;
• increased healthcare contact;
• diagnostic imaging;
• detection of previously unrecognised vascular disease.

Patients receiving fluoroquinolones may also differ systematically from patients receiving other antibiotics.

This means that comparison with a completely untreated population may not adequately address confounding.

Active comparators can be more informative.

For example, comparing fluoroquinolone-treated patients with patients receiving another antibiotic for a similar clinical indication can reduce some forms of confounding by indication.

It does not eliminate confounding.

It simply addresses a different part of the problem.

Therefore, a signal evaluator should ask:

What is the most clinically appropriate comparator for the question being investigated?

rather than:

Which comparator produces the largest statistical association?

8. Active-Comparator Evidence

Later meta-analyses included studies using alternative antibiotics as comparators.

A 2020 systematic review and meta-analysis examined seven observational studies comprising approximately 2.85 million participants.

When fluoroquinolone exposure was compared with beta-lactam antibiotic exposure, the pooled odds ratio for aortic aneurysm, dissection or rupture was:

1.56 (95% CI 1.37–1.79).

This estimate was lower than the association observed when fluoroquinolones were compared with non-treatment.

That difference is instructive.

It demonstrates how the estimated magnitude of a drug-event association can depend substantially on comparator selection.

The review also assessed the risk of bias and concluded that the certainty of evidence was very low.

This is an important qualification.

A statistically significant result is not synonymous with high-quality evidence.

The question is not simply whether the confidence interval excludes one.

The question is whether the study design provides a credible estimate of the causal effect.

9. Duration and Exposure

Some observational analyses reported evidence suggesting that longer fluoroquinolone treatment was associated with greater risk.

This can be potentially supportive of causality because a relationship between exposure and outcome can strengthen a causal hypothesis.

However, duration-response relationships in observational studies are difficult to interpret.

Patients receiving longer antibiotic courses may differ from patients receiving shorter courses in:

• infection severity;
• comorbidity;
• treatment setting;
• healthcare exposure;
• concomitant medication;
• underlying disease.

Therefore:

A duration-response relationship is supportive evidence, but it should not automatically be interpreted as a causal dose-response relationship.

The same principle applies to cumulative exposure.

Signal evaluators should distinguish between:

• pharmacological dose-response;
• treatment-duration association;
• cumulative exposure;
• and differences in patient characteristics.

10. Susceptible Populations

Aortic aneurysm and dissection have important baseline risk factors.

These include:

• increasing age;
• hypertension;
• established atherosclerosis;
• previous aneurysm;
• family history of aneurysmal disease;
• connective-tissue disorders;
• other vascular disease.

PRAC specifically highlighted higher-risk patients in its product-information recommendation.

The proposed SmPC wording advised careful benefit-risk assessment and consideration of alternative therapeutic options in patients with:

• a positive family history of aneurysm disease;
• pre-existing aortic aneurysm;
• previous aortic dissection;
• Marfan syndrome;
• vascular Ehlers-Danlos syndrome;
• Takayasu arteritis;
• giant cell arteritis;
• Behcet's disease;
• hypertension;
• known atherosclerosis.

This is an important example of the difference between relative and absolute risk.

If baseline risk is substantially higher in a particular patient, a similar relative increase can produce a larger absolute increase.

Therefore, the clinical significance of a pharmacovigilance signal cannot be determined from the relative effect estimate alone.

11. Evidence Available at the Time of Regulatory Action

Historical signal evaluation requires particular discipline.

The regulatory decision should be reconstructed using the evidence available at the time.

PRAC's September 2018 signal recommendation states that it considered:

• epidemiological studies;
• non-clinical studies;
• responses from marketing authorisation holders.

The epidemiological evidence cited included studies by Lee et al., Daneman et al. and Pasternak et al.

The non-clinical evidence included work by LeMaire et al.

PRAC concluded that marketing authorisation holders for systemic and inhaled fluoroquinolones should submit a variation to amend product information.

This is the critical historical decision point.

The later literature should not be used to pretend that PRAC had evidence that had not yet been published.

At the same time, a modern pharmacovigilance review should not ignore later evidence.

The two questions are different:

Historical question

Was the regulatory action reasonable based on the evidence available in 2018?

Current question

How should the association be understood in light of the evidence accumulated since 2018?

Keeping these questions separate prevents hindsight bias.

12. What PRAC Actually Recommended

PRAC recommended amendments to the product information.

The proposed SmPC wording stated that epidemiological studies reported an increased risk of aortic aneurysm and dissection following fluoroquinolone exposure, particularly in older people.

It recommended that fluoroquinolones be used only after careful benefit-risk assessment and consideration of other therapeutic options in patients with predisposing conditions or risk factors.

The proposed wording also advised patients to seek immediate medical attention for sudden:

• abdominal pain;
• chest pain;
• back pain.

This regulatory language is important because it demonstrates how uncertainty was translated into risk minimisation.

PRAC did not simply state that every fluoroquinolone inevitably causes aortic disease.

The wording described epidemiological evidence, identified higher-risk populations and provided a clinical response.

That is a more nuanced regulatory conclusion than an absolute causal assertion.

13. Was the Regulatory Action Proportionate?

The evidence supports the view that regulatory action was proportionate to the concern identified.

The outcome under consideration was serious and potentially fatal.

The available evidence showed a recurring epidemiological association.

There was biological plausibility.

There were identifiable patient characteristics associated with higher baseline risk.

Fluoroquinolones also have therapeutic alternatives for many indications.

Against this background, targeted risk minimisation was a reasonable regulatory option.

Importantly, the action was not equivalent to demonstrating a large absolute risk across the entire treated population.

The regulatory decision can therefore be understood as:

The evidence was sufficient to justify communicating and managing a potentially serious risk, particularly in patients with relevant predisposing factors, despite remaining uncertainty about the exact magnitude and causality of the association.

That distinction is central to good pharmacovigilance reasoning.

14. Later Evidence: Did the Signal Strengthen or Weaken?

Later evidence provides a useful test of the historical conclusion.

The subsequent literature did not produce a single uniform estimate.

A 2019 meta-analysis found a more than two-fold association for the combined outcome within 60 days, with a stronger signal for aneurysm than dissection.

A 2020 systematic review found an association even when comparing fluoroquinolones with beta-lactam antibiotics, but assessed the certainty of evidence as very low.

A 2021 systematic review and meta-analysis also reported statistically significant associations, although it identified substantial methodological limitations.

A 2022 systematic review and meta-analysis found increased risk for aortic aneurysm or dissection within 30 and 60 days, but rated the certainty of evidence as very low for the 30-day analysis and low for the 60-day analysis.

Another 2022/2023 systematic review and meta-analysis including more than 53 million participants found:

The later evidence therefore generally continued to support concern about aortic aneurysm, while providing less consistent evidence for aortic dissection considered separately.

More recent evidence continues to show that the association is an area of scientific interest rather than a completely settled quantitative question.

A 2025 meta-analysis of 11 cohort studies involving approximately 82 million participants reported a pooled HR of 1.20 for aortic aneurysm or dissection, 1.47 for aneurysm and 1.12 for dissection.

These estimates were smaller than several earlier estimates.

The direction of the evidence remained compatible with an association, but the magnitude varied substantially by study design, comparator and outcome.

This is exactly why a historical signal should not be represented by a single headline number.

15. What the Later Evidence Changes

Later evidence does not necessarily overturn the 2018 regulatory assessment.

Instead, it changes the level of precision with which the association can be described.

The current evidence supports several conclusions:

The association remains biologically and epidemiologically plausible

Multiple observational studies and meta-analyses continue to report an association.

The magnitude is uncertain

Effect estimates vary considerably.

Aortic aneurysm appears more consistently associated than aortic dissection

Several later analyses support this distinction.

Residual confounding remains important

The evidence base is predominantly observational.

Absolute risk remains important

The outcomes are rare, and baseline vascular risk varies substantially between patients.

The historical regulatory warning remains understandable

The seriousness of the outcome, the evidence available in 2018 and the availability of alternative antibiotics provided a reasonable basis for targeted risk minimisation.

The later evidence therefore refines rather than simply invalidates the original signal evaluation.

16. A Structured Evidence Assessment

A practical evidence matrix can help prevent overinterpretation.

This illustrates a central principle:

Signal evaluation is an integration exercise rather than a contest between individual studies.

17. Signal Detection Versus Signal Evaluation

The case also demonstrates the distinction between signal detection and signal evaluation.

Signal detection asks:

Is there enough evidence to justify investigation of a potential new or insufficiently characterised association?

Signal evaluation asks:

After reviewing the available evidence, what is the most defensible interpretation and what action, if any, is warranted?

The fluoroquinolone case progressed beyond signal detection.

The evidence was sufficient for a formal regulatory assessment.

That assessment incorporated:

• epidemiology;
• non-clinical evidence;
• clinical seriousness;
• susceptible populations;
• therapeutic alternatives;
• and uncertainty.

This distinction matters operationally.

A signal should not be closed merely because the evidence is not yet definitive.

Conversely, the existence of a signal does not automatically mean that the product's benefit-risk balance has become unacceptable.

The decision depends on the totality of evidence and the available risk-management options.

18. What a QPPV Should Take From This Case

18.1 Define the decision question

A signal evaluation should identify the decision that needs to be made.

Examples include:

• Is the association credible?
• Is there enough evidence to update the safety profile?
• Are particular patients at increased risk?
• Does product information require amendment?
• Is additional data collection warranted?
• Has the evidence changed sufficiently to reconsider the benefit-risk balance?

Without a clear decision question, signal evaluations can become literature summaries rather than regulatory assessments.

18.2 Define the phenotype precisely

"Aortic events" is not sufficiently precise.

The evaluator should distinguish:

• aneurysm;
• dissection;
• rupture;
• pre-existing disease;
• newly diagnosed disease;
• incidental findings.

The outcome definition can materially affect the estimated association.

18.3 Use appropriate comparators

A comparator should be selected based on the clinical question.

An untreated population may answer a different question from an active antibiotic comparator.

The evaluator should understand what sources of confounding each comparator addresses and what limitations remain.

18.4 Separate relative risk from absolute risk

A relative risk of approximately two does not mean that an individual patient has a 50% or even 2% probability of experiencing an aortic event.

The baseline risk may be very low.

Conversely, the absolute increase may be more important in a patient with substantial pre-existing vascular risk.

Risk communication should therefore avoid presenting relative measures without clinical context.

18.5 Treat uncertainty as evidence

Uncertainty is not a defect that needs to be hidden.

A high-quality evaluation should explicitly record:

• what is established;
• what is probable;
• what remains uncertain;
• what alternative explanations remain plausible;
• what evidence would materially change the conclusion.

This makes the evaluation more useful for future reassessment.

18.6 Distinguish regulatory action from proof of causality

A regulatory warning does not necessarily mean that every component of a causal hypothesis has been conclusively established.

Regulatory action may be justified because:

• the event is serious;
• the signal is credible;
• susceptible patients can be identified;
• safer alternatives exist;
• risk minimisation is feasible.

That is different from claiming that a precise causal effect has been quantified.

19. What the SmPC Wording Teaches

The final regulatory language is itself an important pharmacovigilance teaching tool.

The wording communicates an epidemiological association rather than making an absolute statement that fluoroquinolones invariably cause aortic disease.

It then translates that evidence into practical actions:

• careful benefit-risk assessment;
• consideration of alternative treatments in higher-risk patients;
• recognition of predisposing conditions;
• urgent evaluation of relevant symptoms.

This illustrates how product information can preserve scientific uncertainty while still communicating a clinically meaningful risk.

For a pharmacovigilance professional, an important question is therefore:

Does the proposed product-information wording accurately reflect both the evidence and its uncertainty?

Overly strong wording may overstate causality.

Overly weak wording may fail to communicate a serious potential risk.

The appropriate language should match the regulatory conclusion.

20. Historical Evidence Versus Current Evidence

This case demonstrates why historical signal evaluations should maintain a strict temporal boundary.

A regulator should be evaluated against the evidence available at the time of its decision.

The 2018 PRAC assessment cannot be criticised because it did not incorporate studies that were published later.

At the same time, a contemporary pharmacovigilance assessment should incorporate subsequent evidence.

The evidence can therefore be divided into two layers.

Evidence available for the 2018 decision

• epidemiological studies available at that time;
• non-clinical evidence;
• clinical safety information;
• MAH responses;
• existing knowledge concerning susceptible populations.

Evidence accumulated after the decision

• additional cohort studies;
• additional systematic reviews;
• active-comparator analyses;
• analyses separating aneurysm from dissection;
• newer meta-analyses;
• continued evaluation of confounding and effect magnitude.

This separation prevents hindsight bias while allowing the signal to evolve.

21. What Evidence Would Most Reduce Uncertainty?

A critical signal evaluation should identify the evidence that would actually change the conclusion.

Useful evidence would include:

• large active-comparator cohort studies;
• robust adjustment for infection severity;
• clearly defined aneurysm and dissection outcomes;
• replication across independent databases;
• consistent analyses of exposure windows;
• analyses of susceptible populations;
• assessment of risk after treatment discontinuation;
• investigation of dose and cumulative exposure;
• mechanistic studies directly relevant to aortic-wall integrity.

The objective is not to demand perfect evidence.

It is to identify which remaining uncertainties are decision-relevant.

For example, if additional research consistently showed no association with dissection but continued to show an association with aneurysm, the signal definition should become more specific.

Similarly, if better active-comparator studies substantially reduced the effect estimate, the magnitude of concern might need reassessment.

22. Overall Causal Assessment

The available evidence supports an epidemiological association between fluoroquinolone exposure and aortic disease.

The strongest and most consistent later evidence concerns aortic aneurysm.

The evidence for aortic dissection considered separately is less consistent.

Several factors increase the credibility of the association:

• temporal association with recent exposure;
• reproducibility across observational studies;
• biological plausibility;
• findings in susceptible populations;
• persistence of association in some active-comparator analyses;
• regulatory review incorporating multiple evidence streams.

Important limitations remain:

• the evidence base is predominantly observational;
• residual confounding cannot be excluded;
• confounding by indication remains relevant;
• effect estimates vary between studies;
• absolute risk is low;
• the strength of evidence differs between aneurysm and dissection;
• some systematic reviews rate the certainty of evidence as low or very low.

The most defensible modern conclusion is therefore:

The accumulated evidence supports an association between systemic fluoroquinolone exposure and aortic aneurysm or aortic disease, with the evidence appearing more consistent for aortic aneurysm than for aortic dissection. The magnitude of any causal risk remains uncertain because the evidence is predominantly observational and susceptible to residual confounding. The evidence available to PRAC in 2018 was nevertheless sufficient to support targeted risk-minimisation measures in European product information.

This conclusion is deliberately narrower than saying that fluoroquinolones definitively cause aortic aneurysm or dissection in all exposed patients.

23. Lessons for Signal Management

Lesson 1: Define the event precisely

A combined endpoint may be useful for initial detection but should not prevent evaluation of clinically distinct outcomes.

Lesson 2: Do not confuse association with causation

Observational effect estimates are important evidence but require critical assessment of bias and confounding.

Lesson 3: Comparator selection matters

Active comparators can address some confounding that remains when treated patients are compared with untreated patients.

Lesson 4: Examine timing

A credible exposure window strengthens interpretation.

Lesson 5: Consider baseline risk

Relative risk cannot be interpreted independently of absolute baseline risk.

Lesson 6: Look for converging evidence

Epidemiology, clinical observations and biological plausibility can reinforce one another.

Lesson 7: Preserve uncertainty

A regulatory warning does not mean that every component of the causal hypothesis has been proven.

Lesson 8: Examine the regulatory wording

The final SmPC language can show how regulators translated evidence and uncertainty into practical risk minimisation.

Lesson 9: Separate historical and current evidence

A historical regulatory decision should be evaluated using the evidence available at the time, while current assessments should incorporate subsequent evidence.

Lesson 10: Re-evaluate signals

Signal evaluation is not necessarily permanently closed.

New evidence can:

• strengthen an association;
• weaken an association;
• identify susceptible populations;
• narrow the phenotype;
• alter the estimated magnitude;
• or change the appropriate regulatory response.

Lesson 11: Separate signal detection from signal evaluation

The threshold for investigating a potential signal is different from the threshold for changing the benefit-risk assessment or product information.

Conclusion

The fluoroquinolone–aortic aneurysm/dissection signal demonstrates how pharmacovigilance operates when evidence is concerning, clinically important and biologically plausible, but not perfectly definitive.

The signal developed primarily through epidemiological investigation.

Early studies and meta-analyses reported substantial associations. Subsequent research continued to identify an association but also demonstrated important variation in effect estimates and greater consistency for aortic aneurysm than for aortic dissection.

PRAC's 2018 assessment illustrates the regulatory consequence of this type of evidence.

The Committee considered epidemiological and non-clinical evidence and recommended changes to product information. The resulting risk-minimisation language identified higher-risk patients, required careful benefit-risk consideration and provided advice concerning symptoms that could indicate an acute aortic event.

The regulatory response therefore did not depend on proving a single precise causal risk estimate.

It reflected the seriousness of the event, the available evidence, the clinical characteristics of susceptible patients and the feasibility of targeted risk minimisation.

For pharmacovigilance professionals, the most important lesson is methodological.

A strong signal evaluation does not ask only:

"Is the signal real?"

It asks:

"What does the totality of evidence actually support, how certain are we, what alternative explanations remain, which patients are most affected, and what action is proportionate to the evidence?"

That is the difference between detecting a signal and evaluating one.

References

1. European Medicines Agency. PRAC recommendations on signals adopted at the 3–6 September 2018 PRAC meeting. Fluoroquinolones for systemic and inhaled use — aortic aneurysm and dissection. EMA/PRAC/595691/2018.

2. European Medicines Agency. Quinolone- and fluoroquinolone-containing medicinal products — Article 31 referral procedure.

3. Singh S, Nautiyal A. Aortic Dissection and Aortic Aneurysms Associated with Fluoroquinolones: A Systematic Review and Meta-Analysis. American Journal of Medicine. 2017;130(12):1449–1457.e9. PMID: 28739200.

4. Pasternak B, Inghammar M, Svanström H. Fluoroquinolone use and risk of aortic aneurysm and dissection: nationwide cohort study. BMJ. 2018;360:k678. PMID: 29519881.

5. Gopalakrishnan C, et al. Fluoroquinolones and the Risk of Aortic Aneurysm or Aortic Dissection: A Systematic Review and Meta-Analysis. Journal of the American Heart Association. 2019. PMID: 30947680.

6. Ribeiro NV, et al. Fluoroquinolones Are Associated With Increased Risk of Aortic Aneurysm or Dissection: Systematic Review and Meta-analysis. Seminars in Thoracic and Cardiovascular Surgery. 2021;33(4):907–918. PMID: 33181305.

7. Wee I, et al. The association between fluoroquinolones and aortic dissection and aortic aneurysms: a systematic review and meta-analysis. Scientific Reports. 2021;11:11073. PMID: 34040146.

8. Chen C, et al. Do fluoroquinolones increase aortic aneurysm or dissection incidence and mortality? A systematic review and meta-analysis. Frontiers in Cardiovascular Medicine. 2022;9:949538. PMID: 36017083.

9. Fatima K, et al. Fluoroquinolones and the risk of aortic aneurysm or aortic dissection: an updated systematic review and meta-analysis including 53,651,283 patients. Journal of Cardiovascular Medicine. 2023;71(5):485–493. PMID: 36468763.

10. Meta-analysis of cohort studies examining the risk of aortic aneurysm or aortic dissection associated with fluoroquinolone use, including studies available through March 2025. PMID: 41678806.