ÌÇÐÄvlog

Key Points

QuestionÌý What is the risk of peripheral neuropathy following exposure to fluoroquinolone or amoxicillin-clavulanate antibiotics?

FindingsÌý This nested case-control study of 5357 incident peripheral neuropathy cases and 17 285 matched controls showed that current use of systemic fluoroquinolone antibiotics increased the risk of peripheral neuropathy by 47%, causing an additional 2.4 cases per 10 000 patients per year of current use. Risk appeared to be greater with cumulative exposure, among men, and among those older than 60 years of age but was not significantly associated with amoxicillin-clavulanate.

MeaningÌý Health care professionals should consider the potential risk of peripheral neuropathy when prescribing fluoroquinolone antibiotics.

ImportanceÌý Peripheral neuropathy has been associated with systemic fluoroquinolone exposure, but risk has been poorly quantified.

ObjectiveÌý To calculate relative and absolute risk estimates for the association of fluoroquinolone exposure with peripheral neuropathy and to examine how risk may be affected by timing of fluoroquinolone exposure and by other risk factors.

Design, Setting, and ParticipantsÌý This nested case-control study used anonymized data from all patients routinely registered with general practices in The Health Improvement Network database, a large primary care population database in the United Kingdom, from January 1, 1999, to December 31, 2015. Data analyses were conducted January 8, 2018. The cohort consisted of 1 338 900 adults issued 1 or more prescriptions of fluoroquinolone (34.3%) or amoxicillin-clavulanate (65.7%) antibiotics. Adults with incident peripheral neuropathy were matched (on age, sex, general practice, and calendar time) with up to 4 controls by using incidence density sampling selected from a cohort prescribed oral fluoroquinolone or amoxicillin-clavulanate antibiotics. Incidence rate ratios of peripheral neuropathy were calculated for fluoroquinolone and for amoxicillin-clavulanate exposure and compared with nonexposure among patients without diabetes, with sensitivity analyses testing the consistency of the results. Population mean–adjusted rate differences were then estimated, including the number needed to harm for various durations of fluoroquinolone therapy.

ExposuresÌý Current and cumulative exposure to oral fluoroquinolone or amoxicillin-clavulanate antibiotics.

Main Outcomes and MeasuresÌý Incident peripheral neuropathy cases recorded in electronic medical records.

ResultsÌý In total, 5357 patients with incident peripheral neuropathy (mean [SD] age, 65.6 [14.7] years; 2809 women [52.4%]) were matched to 17 285 controls (mean [SD] age, 64.4 [15.2] years; 9485 women [54.9%]) without diabetes. Current oral fluoroquinolone exposure was associated with an increased relative incidence of peripheral neuropathy compared with nonexposure (adjusted incident rate ratio, 1.47; 95% CI, 1.13-1.92). Risk increased by approximately 3% for each additional day of current fluoroquinolone exposure and persisted for up to 180 days following exposure. No significant increased risk was observed with oral amoxicillin-clavulanate exposure. The absolute risk with current oral fluoroquinolone exposure was 2.4 (95% CI, 1.8-3.1) per 10 000 patients per year of current use. The number needed to harm for a 10-day course was 152 083 patients (95% CI, 117 742-202 778) and was greatest among men and among patients older than 60 years.

Conclusions and RelevanceÌý The results of the present study suggested that oral fluoroquinolone therapy was associated with an increased risk of incident peripheral neuropathy that may depend on the timing of the exposure and the cumulative dose. Health care professionals should consider these potential risks when prescribing fluoroquinolone antibiotics.

Fluoroquinolones are widely prescribed antibiotics commonly used for the treatment of urinary tract, respiratory tract, and gastrointestinal tract system infections, with variation in drug usage among different countries.^1,2 However, fluoroquinolones may cause adverse drug reactions, as listed in their product information. These adverse drug reactions include peripheral neuropathy, the frequency of which is either unknown or poorly quantified.³ Peripheral neuropathy encompasses single and multiple mononeuropathies as well as polyneuropathy.⁴ It has a diverse manifestation, usually involving sensory disturbances affecting the nerves leading to hypoesthesia or hyperesthesia, affecting functional ability and quality of life. The incidence of peripheral neuropathy in the general population varies according to the etiology, with estimates ranging from 1 to 61 per 1000 person-years in the general population.^5,6 Common causes of peripheral neuropathy include diabetes, alcohol abuse, HIV infection, and drugs.^7,8

Peripheral neuropathy was first identified as an adverse drug reaction to fluoroquinolones from spontaneous reports, but the exact mechanism is uncertain. In the United States and Europe, fluoroquinolones have undergone regulatory safety reviews to investigate the risk of long-term, persistent, or irreversible reactions commonly affecting muscles, joints, and the nervous system.^9,10 This concern led the US Food and Drug Administration to strengthen warnings for this class of products in 2016 and restrict their use for certain clinical indications, while warnings about peripheral neuropathy risk were implemented as early as 2013.¹¹

In Europe, the safety review has been conducted by the European Medicines Agency Pharmacovigilance Risk Assessment Committee.⁹ Although the risk of peripheral neuropathy is recognized, there are limited data quantifying both relative and absolute risk from fluoroquinolone exposure. To our knowledge, only 1 observational study has been conducted on this topic, a case-control study using US administrative claims data, which reported an increased risk of peripheral neuropathy in current users of fluoroquinolones.¹² The aim of the present study was to further quantify the relative and absolute risk of peripheral neuropathy associated with fluoroquinolone exposure and to investigate potential factors associated with this risk.

The Health Improvement Network (THIN) database contains longitudinal electronic patient records extracted from more than 500 general practices across the United Kingdom covering approximately 6% of the UK population. Data are largely representative of the UK population in terms of age, sex, deprivation status, and geographical distribution.¹³ A primary care database, THIN contains general practitioners, diagnostics, prescriptions, and some patient lifestyle-related information. Diagnoses, symptoms, procedures, and other relevant health information are coded into the database using the Read code clinical classification system. This system has been the standard approach to recording patient data in UK primary care since 1990, and it is a hierarchical classification system linked to the International Classification of Diseases.¹⁴ Data quality control measures available in THIN include the Acceptable Mortality Reporting date, which are dates specific to each practice and define the date from which computerized recording of mortality data has reached an acceptable standard.¹⁵ The present study was approved by the THIN Scientific Review Committee (protocol No. 17THIN087) per the standard terms and conditions for use of anonymized THIN data, which does not require the need to obtain informed consent from individual patients. The study was registered in the EU PAS Register (identifier ).

A nested case-control study design was used to evaluate the risk of incident peripheral neuropathy. The study population consisted of adults at least 18 years of age in the THIN database between January 1, 1999, and December 31, 2015, who were issued at least 1 prescription of oral amoxicillin-clavulanate or oral fluoroquinolone antibiotic therapy. Amoxicillin-clavulanate was chosen so that controls were sampled from a more representative population prescribed antibiotics to circumvent confounding by indication or by severity. Given that there is a high incidence of peripheral neuropathy associated with diabetes in the general population and that infective diabetic foot complications inherently associated with peripheral neuropathy are frequently treated with amoxicillin-clavulanate, patients with diabetes were excluded from the primary analysis. Patients with a medical history of peripheral neuropathy prior to cohort entry were excluded to evaluate incident cases.

Cohort entry was defined as the date of the first oral amoxicillin-clavulanate or oral fluoroquinolone prescription after the latest of the following criteria: start of the study period (January 1, 1999), Acceptable Mortality Reporting date of the general practice, patient’s 18th birthday, and date of registration with a general practice plus 1 year. All participants were required to have at least 1 year of observation prior to cohort entry. Cohort exit was defined by the earliest of the following criteria: occurrence of the outcome, deregistration from the general practice, death, date of last data collection from the general practice, and end of the study period (December 31, 2015).

Cases comprised patients with an incident recording of peripheral neuropathy defined by the first ever Read code for peripheral neuropathy in the patient’s electronic medical record. The date of the first event occurring after cohort entry was the index date for case participants. The list of Read codes is given in eTable 1 in the Supplement.

Up to 4 controls were randomly selected and matched to each case on age decile, sex, general practice, and calendar year of cohort entry by using incidence density sampling. The risk-set date for controls was the index date for cases. With incidence density sampling, controls are a selection of person-moments from individuals who have not experienced the event at or before the index date. In this regard, controls may be selected more than once, and people who subsequently become cases may be selected as controls at earlier time points. In total, 185 cases of peripheral neuropathy (1.7%) were unmatched following the first round of matching criteria. These cases were included matched on age decile, sex, and calendar year of cohort entry only. Sensitivity analyses were then conducted excluding these risk sets from the primary analysis.

Current exposure to oral fluoroquinolone antibiotics (ciprofloxacin, moxifloxacin, levofloxacin, norfloxacin, and ofloxacin) and to amoxicillin-clavulanate was measured by identifying 1 or more prescriptions issued within 30 days prior to the index date (excluding the day of the index date). In a secondary analysis, we extended the exposure risk windows up to 90 days and then 91 to 180 days prior to the index date. Cumulative antibiotic exposure was measured as the total number of days of oral fluoroquinolone or oral amoxicillin-clavulanate exposure within each risk window. Cumulative days of exposure was calculated by dividing the prescription quantity information by the standard administration schedules for each antibiotic as follows: oral amoxicillin-clavulanate 3 times a day; oral levofloxacin, moxifloxacin, and norfloxacin once a day; and ciprofloxacin and ofloxacin twice a day.¹⁶ Prescriptions for oral suspension formulations were calculated taking into account quantity information recorded in milliliters. Cumulative exposure was examined as a continuous variable.

The following confounders were considered in the statistical model.

• Demographic factors: age, sex, and general practice (inherent in the matching criteria).

• Lifestyle factors: smoking, body mass index (BMI) calculated as weight in kilograms divided by height in meters squared, and history of alcohol abuse.

• Comorbidity: The Charlson comorbidity score was defined using the clinical conditions of myocardial infarction, congestive heart failure, peripheral vascular disease, dementia, cerebrovascular disease, chronic lung disease, connective tissue disease, peptic ulcer, chronic liver disease, diabetes with or without complications, hemiplegia, moderate or severe liver disease, chronic kidney disease, cancer, cancer with metastasis, and AIDS.¹⁷ In the primary analysis (patients without diabetes), the Charlson comorbidity score was based on all conditions except diabetes with or without complications, with higher scores indicating a greater number of comorbidities.

• Exposure to nitrofurantoin, phenytoin, or metronidazole therapy (defined by a prescription within 90 days of the index date), which may cause peripheral neuropathy.

• Medical history of other conditions associated with peripheral neuropathy, including Lyme disease, Sjögren syndrome, systemic lupus erythematosus, Charcot-Marie-Tooth, amyloidosis, and shingles.

Conditional logistic regression was used to calculate odds ratios for incident peripheral neuropathy and fluoroquinolone and amoxicillin-clavulanate exposure compared with no exposure. Using an incidence density sampling approach, odds ratios are estimators of incidence rate ratios (IRR). Adjusted rate differences were calculated for all patients (including people with diabetes) for significant associations providing an absolute measure.¹⁸ We then used adjusted rate differences to estimate the number of patients needed to treat to cause 1 additional case of peripheral neuropathy (number needed to harm) for different durations of fluoroquinolone therapy. Because concomitant exposure to other drugs may increase the risk of peripheral neuropathy, the interaction between fluoroquinolone exposure and oral phenytoin, nitrofurantoin, and metronidazole exposure was further evaluated on a multiplicative scale.

Sensitivity analyses were conducted as follows: (1) modeling individual components of the Charlson comorbidity score; (2) excluding risk sets not matched by general practice; (3) undertaking a complete case analysis; (4) including people with diabetes; and (5) stratifying by new and prevalent users. Multiple imputation was used to impute missing data on BMI (7.3% missing) and smoking status (1.8% missing). The imputation model included all variables relating to clinical characteristics, outcomes, medications, and fluoroquinolone or amoxicillin-clavulanate exposure. Multiple imputation used fully conditional specification, with linear regression for continuous variables (BMI) and logistic regression for categorical variables (smoking status), with 5 imputations analyzed using the Rubin rules.¹⁹ Analysis was conducted January 8, 2018, using SAS Enterprise Guide, version 6.1 (SAS Institute Inc) and Stata, version 14 (StataCorp). A 2-sided P < .05 was considered statistically significant.

The cohort consisted of 1 338 900 adults issued 1 or more prescriptions of fluoroquinolone (34.3%) or amoxicillin-clavulanate (65.7%) antibiotics (mean [SD] age, 52.8 [19.5] years; 57% female) without a diagnosis of peripheral neuropathy at cohort entry. A total of 11 224 incident peripheral neuropathy cases were identified and matched to 42 316 controls. People with diabetes recorded on or prior to the index date were then identified and excluded from the population, resulting in 5357 incident peripheral neuropathy cases (mean [SD] age, 65.6 [14.7] years; 2809 women [52.4%]) matched to 17 285 controls (mean [SD] age, 64.4 [15.2] years; 9485 women [54.9%]) (Table 1). The median (interquartile range) duration was 10 (9) days for fluoroquinolone exposure and 7 (7) days for amoxicillin-clavulanate exposure.

The relative incidence of peripheral neuropathy was significantly elevated with current oral fluoroquinolone exposure (adjusted IRR [aIRR], 1.47; 95% CI, 1.13-1.92) but not with amoxicillin-clavulanate exposure (aIRR, 1.10; 95% CI, 0.86-1.40) (Table 2). Risk increased by approximately 3% for each additional day of current fluoroquinolone exposure, but not with increasing amoxicillin-clavulanate exposure, within the risk window. The secondary analysis assessing the association with recent exposure found that the relative incidence of peripheral neuropathy was significantly elevated within 1 to 90 days of systemic fluoroquinolone exposure (aIRR, 1.29; 95% CI, 1.05-1.57) and within 91 to 180 days of systemic fluoroquinolone exposure (aIRR, 1.25; 95% CI, 1.03-1.51), with the greatest risk observed when exposure occurred in both risk periods (aIRR, 1.68; 95% CI, 1.20-2.35) (Table 3).

Within the model, there was a significant association between peripheral neuropathy and history of alcohol abuse, amyloidosis, Lyme disease, Sjögren syndrome, and shingles. The adjusted relative incidence of peripheral neuropathy was also significantly elevated with increasing BMI (aIRR, 1.01; 95% CI, 1.00-1.01) and Charlson comorbidity scores (aIRR, 1.19; 95% CI, 1.16-1.21) as well as in current smokers (aIRR, 1.14; 95% CI, 1.04-1.26) and in those with exposure to phenytoin (aIRR, 1.62; 95% CI, 1.09-2.42) or nitrofurantoin (aIRR, 1.40; 95% CI, 1.13-1.75) therapy (Table 4). No significant interaction was observed between current fluoroquinolone and concomitant phenytoin, nitrofurantoin, or metronidazole therapy. Following the exclusion of people with other conditions strongly associated with peripheral neuropathy (namely, amyloidosis, Lyme disease, Sjögren syndrome, systemic lupus erythematosus, shingles, and alcohol abuse), the association between incident peripheral neuropathy and current oral fluoroquinolone exposure was stronger (aIRR, 1.64; 95% CI, 1.27-2.12) (eTable 2 in the Supplement).

The results of sensitivity analyses were similar to those observed in the main analysis (eTables 2 and 3 in the Supplement). The association between incident peripheral neuropathy and fluoroquinolone exposure within 180 days of the index date was not higher among new users compared with prevalent users (IRR, 1.21; 95% CI, 0.98-1.50 vs IRR, 1.35; 95% CI, 1.14-1.59). The association between incident peripheral neuropathy and current fluoroquinolone exposure remained significant when patients with diabetes were included in the analysis, although a significant association with amoxicillin-clavulanate exposure was observed as hypothesized.

Adjusted rate differences of peripheral neuropathy per 10 000 patients per year of current oral fluoroquinolone use are presented overall and by age and by sex (eTable 4 in the Supplement). Fluoroquinolone exposure was associated with an additional 2.4 (95% CI, 1.8-3.1) peripheral neuropathies per 10 000 patients per year of current fluoroquinolone use and varied by age and sex. The overall population-mean number needed to harm varied from 304 167 (95% CI, 235 484-405 556) patients for a 5-day course of fluoroquinolone therapy to 152 083 patients (95% CI, 117 742-202 778) for a 10-day course to 54 315 (95% CI, 42 051-72 421) patients for a 28-day course and was greatest among men and among patients older than 60 years (Table 5).

This study used a population cohort to investigate the association between oral fluoroquinolone exposure and risk of incident peripheral neuropathy. A significantly increased relative incidence of peripheral neuropathy within 30 days of oral fluoroquinolone therapy was found that remained significant up to 180 days following exposure. Relative incidence may increase by approximately 3% with each additional day of current oral fluoroquinolone use. By contrast, amoxicillin-clavulanate exposure was not associated with a significant increased risk of peripheral neuropathy.

Few articles have been published attempting to quantify the risk of peripheral neuropathy associated with fluoroquinolone exposure. The present analysis was undertaken to support the European Medicines Agency Pharmacovigilance Risk Assessment Committee evaluation on the risk of long-term, persistent, or irreversible adverse effects of fluoroquinolones.⁹ To our knowledge, only 1 observational study investigating the risk of peripheral neuropathy with fluoroquinolone antibiotics has been published: a case-control study using US administrative claims data, which reported an 83% significantly increased risk of peripheral neuropathy with current oral fluoroquinolone use, although a brief report has observed a 34% increased risk of carpal tunnel syndrome associated with fluoroquinolone use.^12,20 These data sources did not include information on the potentially important confounding factors that our study adjusted for, such as smoking status, BMI, alcohol abuse, and certain comorbidities, that may partly explain differences in the size of the observed relative risk estimates. Our study provided data on the risk according to timing, and we observed that the risk of incident peripheral neuropathy may remain elevated up to at least 180 days following oral fluoroquinolone exposure. Peripheral neuropathy risk may increase by approximately 3% for each additional day of fluoroquinolone exposure when modeled as a continuous variable within the treatment episode, suggesting that shorter treatment regimens are one potential way of reducing such risks. Indeed, we estimated the absolute risk of peripheral neuropathy, information that is not currently included in the product information of medicinal products in the class. Although we calculated a population mean–adjusted rate difference of 2.4 additional peripheral neuropathy cases per 10 000 patients per year of current use, absolute risk was greater in men and in those older than 60 years of age. We estimated the number needed to harm for different durations of fluoroquinolone therapy, for age, and for sex. This information may potentially support an update to the summary of product information regarding the frequency of events.

We used oral amoxicillin-clavulanate exposure as a negative control because, in the United Kingdom, amoxicillin-clavulanate is a broader-spectrum antibiotic used for more severe types of infection and may be less subject to unmeasured confounding compared with the more commonly chosen comparator of amoxicillin. Similar to fluoroquinolones, amoxicillin-clavulanate is associated with an increased risk of Clostridium difficile infection, which resulted in similar changes to antibiotic guidelines in the United Kingdom during the study period. Although we observed classic associations with conditions linked to peripheral neuropathy, such as alcohol abuse, we also observed that people with increasing BMI, current smokers, or those prescribed oral phenytoin or nitrofurantoin therapy were also associated with an increased relative incidence. Unlike the multiplicative interactions between fluoroquinolones and oral corticosteroids increasing the risk of tendon rupture, we observed no significant interaction between oral fluoroquinolone exposure and phenytoin, nitrofurantoin, or metronidazole therapy.²¹

We performed a large nested case-control study matching patients on their general practices so that controls were more likely to have similar socioeconomic deprivation status and health care physician prescribing behavior to reduce confounding by indication. Although small numbers of cases were included unmatched on general practices, sensitivity analyses showed that these unmatched cases had negligible influence on the results. Despite adjustment for multiple confounders, risk of residual unmeasured confounding remains possible. For example, certain medicines used in the secondary care setting may cause peripheral neuropathy but are not recorded in THIN. Few types of peripheral neuropathy were recorded, and therefore it was not possible to determine risk associated with potential subtypes that would require further study. The present study used a United Kingdom population; thus, it remains unknown whether risk varies among different countries, although a similar increased risk was observed in a US study.¹² There was no strong evidence that new users of fluoroquinolones were at higher risk than prevalent users, potentially suggesting a more complicated mechanism of toxicity that depends on both current, recent, and cumulative exposure, and further studies on this aspect may be warranted. Similarly, we focused on fluoroquinolones as a class, and risk associated with individual fluoroquinolone products requires further evaluation. Finally, we were unable to determine the duration of peripheral neuropathy episodes associated with fluoroquinolone exposure to ascertain how long these cases persisted and whether they were irreversible, a topic that would have been useful given the recent European regulatory referral procedure.

Although harm may be prevented by reducing fluoroquinolone exposure through antibiotic stewardship interventions, this practice may not be possible or appropriate for all patients, particularly in those with more serious infections such that the benefit-risk balance may differ. Although this risk may be considered a rare event among individual patients, a large number of people are treated with fluoroquinolones with relatively little attention given to this risk among clinical guidelines for the management of infections, suggesting that the risk may not be well known among health care professionals. This lack of knowledge may have implications for special patient groups, such as those with comorbidity and with different absolute risks that may vary. Peripheral neuropathy is only one potential adverse effect of fluoroquinolone exposure. Other potential adverse effects of tendon rupture and aortic aneurysm have been reported in addition to the association of fluoroquinolone exposure with other infection-related problems, including Clostridium difficile infection and antibiotic resistance, that may result in treatment failure and delays in clinical recovery.^21-26

The results of the present study suggested that oral fluoroquinolone antibiotic exposure was associated with an increased risk of peripheral neuropathy. This information may be useful to health care professionals for assessing the benefit-risk of oral fluoroquinolone use and communicating these risks to patients.

Accepted for Publication: February 7, 2019.

Corresponding Author: Daniel Morales, PhD, Division of Population Health and Genomics, University of Dundee, Mackenzie Building Kirsty Semple Way, Dundee DD2 4BF, United Kingdom (d.r.z.morales@dundee.ac.uk).

Published Online: April 29, 2019. doi:10.1001/jamaneurol.2019.0887

Author Contributions: Dr Morales had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: Morales, Pacurariu, Pinheiro, McGettigan, Kurz.

Drafting of the manuscript: Morales, Pacurariu, Pinheiro, McGettigan.

Critical revision of the manuscript for important intellectual content: Morales, Pacurariu, Slattery, McGettigan, Kurz.

Statistical analysis: Morales, Pacurariu, Slattery.

Administrative, technical, or material support: Pacurariu, Pinheiro, Kurz.

Supervision: Kurz.

Conflict of Interest Disclosures: None reported.

Disclaimer: The views expressed in this article are those of the authors and may not be not be understood or quoted as reflecting the views of the European Medicines Agency or one of its committees or working parties.