QPPV.COM

Statins and Tendon Rupture: A Signal Evaluation

An evidence-based evaluation of the proposed association between statins and tendon rupture, showing how positive case reports and rechallenge findings can be weighed against larger epidemiological studies and important confounding factors.

Audio Lesson 9 min
Knowledge Assessment Test your understanding of this article. Take the assessment →

Statins and Tendon Rupture: A Signal Evaluation

Introduction

Some pharmacovigilance signals become convincing because different evidence streams point in the same direction.

Others are more difficult.

A plausible adverse reaction may be reported repeatedly, individual cases may contain positive rechallenge, and a biological mechanism may appear credible. Yet larger epidemiological studies may fail to demonstrate an increased risk.

The proposed association between statins and tendon disorders is a useful example of this second situation.

Statin-associated tendon injury has been described in spontaneous reports and case reports for many years. Some reports contain particularly interesting features, including recurrence after re-exposure. Experimental work has also generated plausible hypotheses concerning tendon extracellular matrix and cellular processes.

But the epidemiological evidence has not consistently demonstrated that statin therapy increases the risk of tendon rupture in the general population.

This makes the topic valuable for pharmacovigilance training.

The question is not:

"Can statins cause tendon rupture?"

The better question is:

"How strong is the evidence that statin exposure causes clinically important tendon injury, and what conclusions can reasonably be drawn from the different evidence streams?"

This article evaluates that question using published clinical, pharmacovigilance, epidemiological and mechanistic evidence.

The purpose is not to declare a simple yes or no.

It is to demonstrate how a signal should be evaluated when the evidence is internally mixed.

The Signal

The proposed safety signal can be expressed as:

Statin exposure may be associated with tendinopathy and, in some patients, tendon rupture.

Reported tendon events include:

The proposed association has attracted attention because tendon disorders can be clinically significant and because statins are widely used.

A rare adverse effect can therefore have meaningful public-health implications if exposure is widespread.

However, widespread exposure also creates a major pharmacovigilance challenge:

A large number of patients taking a medicine will inevitably experience common musculoskeletal events that are unrelated to the medicine.

That makes background incidence and confounding particularly important.

Why This Signal Is Difficult

The signal contains several features that initially support causality.

There are:

But there are also important counterarguments:

This is exactly the type of signal where the quality and design of evidence matter more than the number of reports.

Signal Evaluation

1. What Did the Early Case Reports Show?

Early literature consisted largely of individual case reports describing tendinopathy or tendon rupture during statin treatment.

These reports were important because they generated a plausible safety hypothesis.

A case report can be particularly informative when it contains:

However, case reports also have major limitations.

There is no denominator.

We generally do not know how many people took the statin without developing tendon injury.

There may also be selective publication.

A striking case is more likely to be published than the thousands of patients who took the same drug without experiencing a tendon disorder.

Therefore:

Case reports are excellent for signal generation but usually insufficient for estimating risk.

2. The French Pharmacovigilance Case Series

A particularly useful example is the analysis of 96 spontaneous reports collected through the French Pharmacovigilance Centers between 1990 and 2005.

The reports included:

The reported statins included atorvastatin, simvastatin, pravastatin, fluvastatin and rosuvastatin.

Tendon disorders occurred most often during the first year after statin initiation.

Most importantly from a causality perspective, seven patients were re-exposed to the statin and all seven reportedly experienced recurrence of tendinopathy.

This is a striking finding.

A positive rechallenge is one of the strongest clues available in an individual case.

But the interpretation still requires caution.

Seven rechallenges are not a population risk estimate.

The cases were selected because they were already suspected to be drug-related.

This introduces substantial selection and reporting bias.

The correct conclusion is therefore not:

"Seven out of seven rechallenges prove that statins cause tendon rupture."

The better conclusion is:

"The recurrence observed after rechallenge provides important supportive evidence for causality in those individual cases, while the selected nature of the reports prevents direct estimation of population-level risk."

That distinction is fundamental.

3. Why Rechallenge Matters

Imagine a patient develops Achilles tendinopathy after starting a statin.

That is weak evidence by itself.

Now suppose symptoms improve after the statin is withdrawn.

The evidence becomes more persuasive.

Now suppose the statin is restarted and the same tendon symptoms recur.

The causal argument becomes substantially stronger for that individual patient.

This is because the observation is no longer simply:

Exposure → event

It becomes:

Exposure → event → withdrawal → improvement → re-exposure → recurrence

The repeated temporal pattern is difficult to dismiss as simple coincidence.

However, the same caution used in other signal evaluations applies here.

Rechallenge should not be deliberately performed merely to prove causality when recurrence could expose a patient to clinically significant harm.

The pharmacovigilance value comes from clinically occurring rechallenges.

4. Does Rechallenge Prove a Class Effect?

Not necessarily.

The French pharmacovigilance series included several statins, and tendon disorders have been reported with different members of the class.

That supports the possibility of a class effect.

But several questions remain.

Are all tendon events caused by the same biological mechanism?

Is the risk the same for every statin?

Does dose matter?

Does lipophilicity matter?

Are particular patients predisposed?

Does the underlying hyperlipidaemia contribute independently?

These questions cannot be answered from spontaneous reports alone.

The observation that multiple statins have been implicated therefore supports a class-level hypothesis, but it does not establish identical risk across the class.

5. Time to Onset

Time to onset is another important component of causality assessment.

In the French pharmacovigilance analysis, tendon manifestations occurred most often within the first year after initiation, although events were not restricted to this period.

This pattern is potentially compatible with an adverse drug reaction.

But it is not specific.

Tendon disease can develop at any time in a population with age-related degeneration, physical activity, metabolic disease and other risk factors.

Therefore, time to onset must be interpreted together with:

Temporal association alone is not enough.

6. What Does the Epidemiology Show?

This is where the signal becomes substantially more complicated.

A 2009 case-control study examined 93 patients with tendon rupture and 279 age- and sex-matched controls.

Overall, statin use was not significantly associated with tendon rupture.

The adjusted odds ratio was:

OR 1.10 (95% CI 0.57–2.13).

A predefined subgroup analysis suggested an association in women but not men, with an adjusted OR of 3.76 (95% CI 1.11–12.75) in women.

The overall result therefore did not support a general association between statin therapy and tendon rupture.

The female subgroup finding is interesting, but it should not be automatically treated as proof of a sex-specific adverse reaction.

Subgroup estimates can be unstable, particularly when the number of events is relatively small.

This is a classic pharmacovigilance lesson:

A statistically positive subgroup does not automatically overturn a neutral overall study.

The subgroup should generate a hypothesis that can be tested in other datasets.

7. A Much Larger Cohort Study

A particularly informative later study used the UK Clinical Practice Research Datalink to investigate Achilles and biceps tendon rupture among new statin users.

The study used propensity-score matching and followed patients aged 45 years and older.

The crude hazard ratio was:

HR 1.32 (95% CI 1.21–1.44).

At first glance, that appears concerning.

But the association attenuated dramatically after adjustment:

Adjusted HR 1.02 (95% CI 0.92–1.12).

The propensity-score-matched estimate was:

HR 0.95 (95% CI 0.84–1.08).

The study also found null results across analyses stratified by:

The authors concluded that statin use did not increase the risk of Achilles or biceps tendon rupture.

This study provides an excellent lesson in confounding.

8. What Happened to the Crude Association?

The crude analysis suggested increased risk.

After adjustment, the association essentially disappeared.

Why?

Because statin users and non-users are not interchangeable populations.

Statin users may differ in:

Some of these characteristics may independently influence tendon injury.

Therefore:

A crude association can represent differences between the populations rather than an effect of the medicine.

This is why pharmacovigilance evaluation should not stop at:

"The exposed group had more events."

The important question is:

"More events than whom, after accounting for important differences between the groups?"

9. Hyperlipidaemia as a Potential Confounder

One of the particularly interesting issues in this signal is that the underlying disease may itself be related to tendon pathology.

Hyperlipidaemia has been associated with tendon structural abnormalities, including changes involving the Achilles tendon.

This creates a difficult causal problem.

Consider three possibilities:

Model A

Statin → tendon injury

Model B

Hyperlipidaemia → tendon injury

Model C

Hyperlipidaemia → statin treatment

and

Hyperlipidaemia → tendon injury

In Model C, statin use can appear associated with tendon injury even if the statin itself is not the cause.

This is a classic example of confounding by indication or disease-related confounding.

It is one reason why the neutral adjusted cohort findings are important.

10. What Does the Systematic Review Evidence Show?

A 2016 systematic review evaluated the clinical evidence concerning statins and tendinopathy.

Only four eligible studies were identified:

The authors found no positive association between statin therapy and tendon rupture in the overall populations studied.

They concluded that there was insufficient evidence to consider statins an established causal risk factor for tendon rupture in the general population.

This conclusion is important because it illustrates the difference between:

signal existence

and

causal confirmation.

The presence of case reports means the signal deserves evaluation.

It does not mean that the causal association has been established.

11. What Does More Recent Evidence Add?

More recent reviews continue to show a mixed picture.

A 2025 systematic review of clinical evidence included 12 studies covering more than one million patients.

The review reported:

The authors concluded that statins do not universally increase tendon rupture risk and that the observed effects may depend on factors such as:

This reinforces the importance of not collapsing several different outcomes into one broad category.

"Tendon disorder" is not a single endpoint.

Tendinopathy, tendinitis, tendon pain and complete rupture may have different causes and different risk factors.

12. Signal Versus Confirmed Adverse Reaction

This distinction deserves explicit emphasis.

A pharmacovigilance signal is information suggesting a new potentially causal association or a new aspect of a known association that warrants further investigation.

A signal does not necessarily mean that causality has been established.

The statin–tendon question demonstrates this well.

The evidence supports:

But the evidence does not convincingly establish:

The correct pharmacovigilance conclusion must preserve that distinction.

Pharmacological Plausibility

What Is Known?

Statins inhibit HMG-CoA reductase and reduce hepatic cholesterol synthesis.

Cholesterol and other lipid-related pathways are relevant to cell membranes and tissue biology.

Tendon tissue depends on:

It is therefore biologically plausible that substantial alterations in lipid-related cellular processes could influence tendon biology.

But plausibility is not proof.

Proposed Extracellular Matrix Mechanisms

One hypothesis concerns matrix metalloproteinases (MMPs) and extracellular matrix remodelling.

Tendon integrity depends on tightly regulated synthesis and degradation of extracellular matrix components.

Experimental observations have suggested that statins can influence pathways involved in matrix turnover.

A case report involving tendon rupture during simvastatin/ezetimibe treatment proposed altered MMP activity as a possible mechanism.

This is an interesting hypothesis.

It is not equivalent to demonstrating that MMP alteration causes clinically relevant tendon rupture in statin-treated humans.

That distinction should remain explicit.

What Is Not Established?

The precise molecular mechanism linking statin exposure to clinically significant tendon rupture remains uncertain.

It has not been demonstrated that:

Therefore, the mechanistic evidence should be described as plausible and hypothesis-generating, rather than established.

This is an important general pharmacovigilance principle:

Mechanistic plausibility can support a signal without converting a hypothesis into a demonstrated causal mechanism.

A Critical Alternative Explanation: The Disease Itself

The relationship between lipid metabolism and tendon disease creates an especially important interpretive problem.

If hyperlipidaemia is associated with tendon abnormalities, patients with more severe lipid disorders may simultaneously have:

This can create a non-causal association between statin treatment and tendon events.

The signal therefore requires careful adjustment for the underlying metabolic condition.

This is one reason why the disappearance of the association after adjustment in the large cohort study is highly informative.

What Would We Expect If Statins Were a Major Cause of Tendon Rupture?

A useful way to test a signal is to consider the counterfactual prediction.

If statins caused a substantial increase in tendon rupture risk in the general population, we would expect appropriately designed studies to demonstrate some combination of:

The large propensity-score-matched cohort study did not show these patterns.

The adjusted estimates remained close to the null across several subgroup analyses.

That substantially weakens the argument for a large population-level effect.

It does not necessarily exclude:

This distinction is critical.

How Should the Individual Rechallenge Cases Be Weighed?

This is perhaps the most interesting part of the signal.

We have two apparently conflicting observations.

Observation 1

Some individual patients developed tendon symptoms during statin exposure and experienced recurrence after re-exposure.

Observation 2

Large adjusted observational studies do not show a meaningful overall increase in tendon rupture risk.

These observations can coexist.

A plausible interpretation is that:

Statins may rarely contribute to tendon disorders in susceptible individuals without producing a large measurable increase in tendon rupture at the population level.

That remains a hypothesis rather than a definitive conclusion.

But it demonstrates why pharmacovigilance should avoid binary reasoning.

The choice is not simply:

"Statins cause tendon rupture"

versus

"Statins never cause tendon problems."

The evidence may support a much narrower conclusion.

Signal Strength by Evidence Type

Evidence source Finding Interpretation
Spontaneous reports Tendinopathy and rupture reported with several statins Signal-generating
Case series 96 reports; several tendon phenotypes Supports investigation
Rechallenge Recurrence in 7/7 rechallenged patients in one series Strong individual-case evidence
Early case-control study No overall association Does not support population-level effect
Large propensity-score cohort Adjusted HR 1.02; PS-matched HR 0.95 Strong evidence against a major overall rupture risk
Systematic review Insufficient evidence for established causal relationship Limits causal conclusion
Mechanistic studies Several plausible pathways Supports biological plausibility
Recent clinical review Mixed findings across tendon outcomes Relationship remains heterogeneous

The table illustrates an important principle.

Different evidence streams should not be averaged mechanically.

They should be interpreted according to what each study can actually establish.

The Importance of Outcome Definition

"Tendon injury" is too broad a term for a rigorous signal evaluation.

Consider:

These events are not interchangeable.

A medicine could potentially influence one endpoint without materially affecting another.

For example, evidence suggesting an association with some forms of tendinopathy would not automatically establish an increased risk of complete tendon rupture.

Therefore, a signal evaluation should ask:

What exact event is being evaluated?

This is particularly important when comparing spontaneous reports with epidemiological studies.

Confounding and Bias Checklist

For a statin–tendon signal, a rigorous evaluation should consider at least:

Age

Tendon degeneration and rupture become more common with increasing age.

Physical activity

Mechanical loading and sports participation can strongly influence tendon injury.

Corticosteroids

Corticosteroid exposure is a well-recognised alternative explanation for some tendon events.

Diabetes

Diabetes may affect tendon structure and injury risk.

Hyperlipidaemia

The underlying lipid disorder may itself be associated with tendon abnormalities.

Renal disease

Renal disease may be associated with both treatment patterns and musculoskeletal complications.

Sex

Some studies have reported different estimates in men and women.

Tendon site

Achilles, rotator cuff, biceps and quadriceps tendons may not have identical risk factors.

Treatment duration

A true drug effect could vary with duration, but epidemiological evidence has not consistently demonstrated such a pattern.

Dose

A dose-response relationship would strengthen causality, but consistent evidence has not been established.

A signal evaluation that ignores these factors risks attributing background disease to the medicine.

What Does a Positive Rechallenge Mean in This Context?

A positive rechallenge should increase suspicion substantially at the individual-case level.

But it should not automatically establish a population-level effect.

Why?

Because the probability of observing an apparently positive rechallenge depends on:

A strong rechallenge case can therefore be:

highly informative for that patient

while simultaneously being:

insufficient to estimate population risk.

This distinction is one of the most useful lessons from this signal.

Signal Prioritisation

How should a pharmacovigilance team prioritise the signal?

Several factors increase interest:

Several factors reduce confidence in a large population-level effect:

Therefore, the signal is worth evaluating, but the evidence does not justify automatically treating statin therapy as a confirmed major risk factor for tendon rupture in the general population.

What Would a Pharmacovigilance Team Review?

A practical signal assessment should examine several layers.

Individual Cases

For each relevant case:

The quality of the case narrative matters.

Literature

The literature review should distinguish:

Do not combine them into one undifferentiated evidence pool.

Epidemiology

Important questions include:

Mechanism

The mechanistic assessment should distinguish:

Established

from

plausible

from

speculative.

This prevents mechanistic language from becoming stronger than the clinical evidence.

A Two-Level Conclusion

Evidence conclusion

The available evidence establishes a credible pharmacovigilance signal concerning tendon disorders during statin therapy, and individual cases—including reports with recurrence after rechallenge—provide supportive evidence of causality in some patients.

However, available epidemiological evidence does not demonstrate a consistent increase in the risk of tendon rupture in the general statin-treated population. In a large propensity-score-matched cohort study, the adjusted estimates were close to the null, including across analyses by sex, age, treatment duration and dose.

The evidence therefore does not support describing statin therapy as a well-established major population-level cause of tendon rupture.

Mechanistic conclusion

Several biological mechanisms have been proposed through which statins could influence tendon extracellular matrix or cellular processes.

These mechanisms are biologically plausible, but the precise pathway responsible for clinically significant tendon injury in humans has not been established.

The mechanistic evidence therefore supports continued investigation rather than definitive mechanistic attribution.

Overall Pharmacovigilance Assessment

The most defensible interpretation is that the evidence supports the following layered conclusion:

Tendon disorders have been reported during statin therapy, and some individual cases contain strong temporal and rechallenge evidence suggesting a causal relationship. However, large adjusted epidemiological studies do not demonstrate a consistent increased risk of tendon rupture in the general statin-treated population. The possibility of rare, patient-specific susceptibility or associations with particular tendon disorders cannot be excluded.

This is deliberately narrower than saying:

"Statins cause tendon rupture."

It is also more scientifically useful than saying:

"There is no association."

The evidence does not justify either extreme.

What This Signal Teaches About Pharmacovigilance

1. A positive rechallenge is powerful but local

A convincing rechallenge can strongly increase causal confidence for an individual case.

It does not automatically quantify population risk.

2. Crude epidemiology can mislead

An apparent association may disappear after adjustment.

The difference between a crude HR of 1.32 and an adjusted HR of 1.02 is an excellent demonstration of confounding.

3. Disease and treatment can be linked

When the underlying disease may itself contribute to the event, confounding by indication becomes particularly important.

4. Do not overinterpret subgroup findings

A statistically positive subgroup can be hypothesis-generating without being definitive.

It should be replicated before being treated as an established effect.

5. Define the event precisely

Tendinopathy is not the same as tendon rupture.

A signal assessment should specify the exact phenotype being evaluated.

6. Mechanistic plausibility is not proof

An attractive molecular mechanism can support a signal.

It cannot substitute for appropriate clinical and epidemiological evidence.

7. Rare susceptibility remains possible

A null population-level study does not prove that no individual patient can experience a drug-related event.

Population-level causality and individual-case causality are related but distinct questions.

8. Evidence should be weighted, not counted

Ten weak reports do not necessarily outweigh one well-designed epidemiological study.

Conversely, one neutral cohort does not necessarily invalidate a compelling individual rechallenge.

The correct approach is to ask what each evidence source contributes.

Key Takeaways

The proposed association between statins and tendon rupture provides an excellent example of a pharmacovigilance signal that remains scientifically interesting despite inconsistent evidence.

The main observations are:

  1. Tendon disorders have been reported during exposure to several statins.

  2. A French pharmacovigilance case series identified 96 spontaneous reports, including tendinitis and tendon rupture.

  3. Seven patients in that series were rechallenged, with recurrence of tendinopathy reported in all seven. This provides important individual-case evidence, although the selected nature of spontaneous reports prevents population-level risk estimation.

  4. An early case-control study found no overall significant association between statin use and tendon rupture.

  5. A large propensity-score-matched cohort study initially showed a crude association, but the estimate attenuated to approximately the null after adjustment.

  6. The large cohort study found no consistent effect across sex, age, treatment duration or dose.

  7. Hyperlipidaemia and other patient characteristics may confound the association because they can influence both the likelihood of receiving a statin and the underlying risk of tendon abnormalities.

  8. Biological mechanisms involving extracellular matrix remodelling and cellular pathways are plausible but remain incompletely established.

  9. The evidence therefore supports a credible historical safety signal and possible rare individual susceptibility, but does not establish a major population-level causal relationship between statins and tendon rupture.

  10. The most important lesson is methodological: a pharmacovigilance assessment should be capable of concluding that a signal is plausible in individual patients while simultaneously finding insufficient evidence for a substantial population-level risk.

References

  1. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP) Module IX – Signal management.

  2. Marie I, DelafenĂŞtre H, Massy N, Thuillez C, Noblet C; Network of the French Pharmacovigilance Centers. Tendinous disorders attributed to statins: a study on ninety-six spontaneous reports in the period 1990-2005 and review of the literature. Arthritis Rheum. 2008;59(3):367-372. doi:10.1002/art.23309. PMID: 18311771.

  3. Beri A, Dwamena FC, Dwamena BA. Association between statin therapy and tendon rupture: a case-control study. J Cardiovasc Pharmacol. 2009;53(5):401-404. doi:10.1097/FJC.0b013e3181a0ce8b. PMID: 19454900.

  4. Spoendlin J, Layton JB, Mundkur M, Meier C, Jick SS, Meier CR. The risk of Achilles or biceps tendon rupture in new statin users: a propensity score-matched sequential cohort study. Drug Saf. 2016;39(12):1229-1237. doi:10.1007/s40264-016-0462-5. PMID: 27677637.

  5. Teichtahl AJ, Brady SR, Urquhart DM, Wluka AE, Wang Y, Shaw JE, Cicuttini FM. Statins and tendinopathy: a systematic review. Med J Aust. 2016;204(3):115-121.e1. doi:10.5694/mja15.00806. PMID: 26866552.

  6. Deren ME, Klinge SA, Mukand NH, Mukand JA. Tendinopathy and tendon rupture associated with statins. JBJS Rev. 2016;4(5):e4. doi:10.2106/JBJS.RVW.15.00072. PMID: 27490216.

  7. Marie I, DelafenĂŞtre H, Massy N, Thuillez C, Noblet C. Tendinous disorders attributed to statins: pharmacovigilance case series and literature review. Arthritis Rheum. 2008;59(3):367-372.

  8. Sendzik J, Klosterhalfen B, van Osch GJVM, et al. [Mechanistic literature concerning statins and tendon extracellular matrix]. Relevant experimental literature should be interpreted as mechanistic evidence rather than direct proof of clinical causality.

  9. Anzillotti G, Vespasiano F, Ă–ttl F, Conte P, Minelli M, Raspugli GF, Svensson Di Giorgio S, Valente R, Kon E, Di Matteo B. Statins: bitter enemies of tendons or not? A systematic review of clinical evidence. Eur Rev Med Pharmacol Sci. 2025;29(10):457-469. doi:10.26355/eurrev_202510_37466. PMID: 41182304.

  10. Gaida JE, Alfredson H, Kiss ZS, et al. Is higher serum cholesterol associated with altered tendon structure or tendon pain? A systematic review. Br J Sports Med. 2015. Review of the relationship between lipid abnormalities and tendon structure.

Revision History

Related Articles

Proton Pump Inhibitors and Hypomagnesaemia: A Signal Evaluation

A practical signal evaluation of proton pump inhibitors and hypomagnesaemia, showing how dechallenge, rechallenge, class effects, latency and mechanistic evidence can strengthen causal assessment.

SSRIs and Postpartum Haemorrhage: A Signal Evaluation

A practical signal evaluation of SSRI exposure and postpartum haemorrhage, examining the literature, exposure timing, biological plausibility, confounding, absolute risk and the evidence for causality.

Signal Evaluation: Fluoroquinolones and Tendon Injury and Tendon Rupture

A historical signal-evaluation case study showing how fluoroquinolone-associated tendon injury evolved from clinical reports into a well-characterised safety concern involving timing, Achilles tendon injury, corticosteroid exposure, patient susceptibility and class-wide regulatory restrictions.