[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Table 1.  Patient Characteristics
Patient Characteristics
Table 2.  Comorbid Conditions
Comorbid Conditions
Table 3.  Concurrent Use of QT-Prolonging Drugs
Concurrent Use of QT-Prolonging Drugs
Table 4.  Occurrence of Cardiac Events
Occurrence of Cardiac Events
Table 5.  Outcomes Associated With Use of Azithromycin Compared With Amoxicillin
Outcomes Associated With Use of Azithromycin Compared With Amoxicillin
1.
Patel  H, Calip  GS, DiDomenico  RJ, Schumock  GT, Suda  KJ, Lee  TA.  Prevalence of cardiac risk factors in patients prescribed azithromycin before and after the 2012 FDA warning on the risk of potentially fatal heart rhythms.   Pharmacotherapy. 2020;40(2):107-115. doi:10.1002/phar.2355PubMedGoogle ScholarCrossref
2.
Ohtani  H, Taninaka  C, Hanada  E,  et al.  Comparative pharmacodynamic analysis of Q-T interval prolongation induced by the macrolides clarithromycin, roxithromycin, and azithromycin in rats.   Antimicrob Agents Chemother. 2000;44(10):2630-2637. doi:10.1128/AAC.44.10.2630-2637.2000 PubMedGoogle ScholarCrossref
3.
Hancox  JC, Hasnain  M, Vieweg  WVR, Crouse  ELB, Baranchuk  A.  Azithromycin, cardiovascular risks, QTc interval prolongation, torsade de pointes, and regulatory issues: a narrative review based on the study of case reports.   Ther Adv Infect Dis. 2013;1(5):155-165. doi:10.1177/2049936113501816 PubMedGoogle Scholar
4.
Ray  WA, Murray  KT, Hall  K, Arbogast  PG, Stein  CM.  Azithromycin and the risk of cardiovascular death.   N Engl J Med. 2012;366(20):1881-1890. doi:10.1056/NEJMoa1003833 PubMedGoogle ScholarCrossref
5.
US Food and Drug Administration. FDA drug safety communication: azithromycin (Zithromax or Zmax) and the risk of potentially fatal heart rhythms. Updated February 14, 2018. Accessed February 21, 2018. https://www.fda.gov/Drugs/DrugSafety/ucm341822.htm
6.
Rao  GA, Mann  JR, Shoaibi  A,  et al.  Azithromycin and levofloxacin use and increased risk of cardiac arrhythmia and death.   Ann Fam Med. 2014;12(2):121-127. doi:10.1370/afm.1601 PubMedGoogle ScholarCrossref
7.
Chou  H-W, Wang  J-L, Chang  C-H, Lai  C-L, Lai  M-S, Chan  KA.  Risks of cardiac arrhythmia and mortality among patients using new-generation macrolides, fluoroquinolones, and β-lactam/β-lactamase inhibitors: a Taiwanese nationwide study.   Clin Infect Dis. 2015;60(4):566-577. doi:10.1093/cid/ciu914 PubMedGoogle ScholarCrossref
8.
Svanström  H, Pasternak  B, Hviid  A.  Use of azithromycin and death from cardiovascular causes.   N Engl J Med. 2013;368(18):1704-1712. doi:10.1056/NEJMoa1300799 PubMedGoogle ScholarCrossref
9.
Trifirò  G, de Ridder  M, Sultana  J,  et al.  Use of azithromycin and risk of ventricular arrhythmia.   CMAJ. 2017;189(15):E560-E568. doi:10.1503/cmaj.160355 PubMedGoogle ScholarCrossref
10.
Almalki  ZS, Guo  JJ.  Cardiovascular events and safety outcomes associated with azithromycin therapy: a meta-analysis of randomized controlled trials.   Am Health Drug Benefits. 2014;7(6):318-328.PubMedGoogle Scholar
11.
Gorelik  E, Masarwa  R, Perlman  A, Rotshild  V, Muszkat  M, Matok  I.  Systematic review, meta-analysis, and network meta-analysis of the cardiovascular safety of macrolides.   Antimicrob Agents Chemother. 2018;62(6):e00438-18. doi:10.1128/AAC.00438-18 PubMedGoogle Scholar
12.
Hansen  L. The Truven Health MarketScan databases for life sciences researchers. March 2018. Accessed March 1, 2018. https://www.ibm.com/watson-health/about/truven-health-analytics
13.
Healthcare Cost and Utilization Project (HCUP). Clinical Classifications Software (CCS) for ICD-9-CM. Agency for Healthcare Research and Quality. Updated March 6, 2017. Accessed March 1, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp
14.
Woosley  RL, Heise  CW, Gallo  T, Tate  J, Woosley  D, Romero  KA. QTdrugs List. Accessed March 1, 2018. https://www.crediblemeds.org/
15.
Psaty  BM, Siscovick  DS, Weiss  NS,  et al.  Hypertension and outcomes research: from clinical trials to clinical epidemiology.   Am J Hypertens. 1996;9(2):178-183. doi:10.1016/0895-7061(96)00015-5 PubMedGoogle ScholarCrossref
16.
De Bruin  ML, van Hemel  NM, Leufkens  HGM, Hoes  AW.  Hospital discharge diagnoses of ventricular arrhythmias and cardiac arrest were useful for epidemiologic research.   J Clin Epidemiol. 2005;58(12):1325-1329. doi:10.1016/j.jclinepi.2005.04.009 PubMedGoogle ScholarCrossref
17.
Tamariz  L, Harkins  T, Nair  V.  A systematic review of validated methods for identifying ventricular arrhythmias using administrative and claims data.   Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):148-153. doi:10.1002/pds.2340 PubMedGoogle ScholarCrossref
18.
Hennessy  S, Leonard  CE, Freeman  CP,  et al.  Validation of diagnostic codes for outpatient-originating sudden cardiac death and ventricular arrhythmia in Medicaid and Medicare claims data.   Pharmacoepidemiol Drug Saf. 2010;19(6):555-562. doi:10.1002/pds.1869 PubMedGoogle ScholarCrossref
19.
Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.   Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663cc PubMedGoogle ScholarCrossref
20.
Rassen  JA, Doherty  M, Huang  W, Schneeweiss  S. Pharmacoepidemiology toolbox. Accessed June 1, 2019. https://www.drugepi.org/
21.
Austin  PC.  Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies.   Pharm Stat. 2011;10(2):150-161. doi:10.1002/pst.433 PubMedGoogle ScholarCrossref
22.
Austin  PC.  Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research.   Commun Stat Simul Comput. 2009;38(6):1228-1234. doi:10.1080/03610910902859574 Google ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 5,379
    Citations 0
    Original Investigation
    Pharmacy and Clinical Pharmacology
    September 15, 2020

    Comparison of Cardiac Events Associated With Azithromycin vs Amoxicillin

    Author Affiliations
    • 1Department of Pharmacy Systems, Outcomes, and Policy, College of Pharmacy, University of Illinois at Chicago
    • 2Flatiron Health, Inc, New York, New York
    • 3Department of Pharmacy Practice, College of Pharmacy, University of Illinois at Chicago
    • 4Center for Health Equity Research and Promotions, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    • 5Department of Medicine, Veterans Affairs Pittsburgh Healthcare System, Pittsburgh, Pennsylvania
    JAMA Netw Open. 2020;3(9):e2016864. doi:10.1001/jamanetworkopen.2020.16864
    Key Points

    Question  Using a large health care database, what is the risk of cardiac events in azithromycin users compared with amoxicillin users?

    Findings  In a cohort study of 2 141 285 episodes of azithromycin use that were matched to episodes of amoxicillin use with high-dimensional propensity scores, no difference in the rate of cardiac events occurred at 5, 10, and 30 days after therapy initiation. However, concurrent use of QT-prolonging drugs with azithromycin was associated with 40% greater odds of cardiac events compared with amoxicillin.

    Meaning  Azithromycin and amoxicillin appear to pose similar odds of cardiac events except for concurrent use of QT-prolonging drugs with azithromycin; therefore, clinicians should use amoxicillin or another antibiotic among such patients.

    Abstract

    Importance  Conflicting evidence exists on the association between azithromycin use and cardiac events.

    Objective  To compare the odds of cardiac events among new users of azithromycin relative to new users of amoxicillin using real-world data.

    Design, Setting, and Participants  This retrospective cohort study used data from Truven Health Analytics MarketScan database from January 1, 2009, to June 30, 2015. Patients receiving either amoxicillin or azithromycin and enrolled in a health care plan 365 days before (baseline period) the dispensing date (index date) were included in the study. Patients were matched 1:1 on high-dimensional propensity scores. Data were analyzed from October 1, 2018, to December 31, 2019.

    Exposures  New use of azithromycin compared with new use of amoxicillin.

    Main Outcomes and Measures  The primary outcome consisted of cardiac events, including syncope, palpitations, ventricular arrhythmias, cardiac arrest, or death as a primary diagnosis for hospitalization at 5, 10, and 30 days from the index date. Logistic regression models were used to estimate odds ratios (ORs) with 95% CIs.

    Results  After matching, the final cohort included 2 141 285 episodes of each index therapy (N = 4 282 570) (mean [SD] age of patients, 35.7 [22.3] years; 52.6% female). Within 5 days after therapy initiation, 1474 cardiac events (0.03%) occurred (708 in the amoxicillin cohort and 766 in the azithromycin cohort). The 2 most frequent events were syncope (1032 [70.0%]) and palpitations (331 [22.5%]). The odds of cardiac events with azithromycin compared with amoxicillin were not significantly higher at 5 days (OR, 1.08; 95% CI, 0.98-1.20), 10 days (OR, 1.05; 95% CI, 0.97-1.15), and 30 days (OR, 0.98; 95% CI, 0.92-1.04). Among patients receiving any concurrent QT-prolonging drug, the odds of cardiac events with azithromycin were 1.40 (95% CI, 1.04-1.87) greater compared with amoxicillin. Among patients 65 years or older and those with a history of cardiovascular disease and other risk factors, no increased risk of cardiac events with azithromycin was noted.

    Conclusions and Relevance  This study found no association of cardiac events with azithromycin compared with amoxicillin except among patients using other QT-prolonging drugs concurrently. Although azithromycin is a safe therapy, clinicians should carefully consider its use among patients concurrently using other QT-prolonging drugs.

    Introduction

    Azithromycin is a widely prescribed antibiotic. More than 30 million prescriptions of azithromycin are dispensed every year in the United States.1 Compared with other macrolide antibiotics that carry some adverse cardiac risks, azithromycin was considered safest because of its lower arrhythmogenic activity.2

    More than 10 years after the drug’s approval in the United States, an early signal from case reports described QT interval prolongation, torsades de pointes (TdP), and ventricular tachycardia after the use of azithromycin.3 Since then, several retrospective, observational studies have investigated the risk of cardiac events with outpatient azithromycin use. In 2012, a study by Ray et al4 found a 2.9-fold higher risk of cardiac deaths within 5 days of an initial dispensing of azithromycin compared with amoxicillin. The risk was higher among patients with a history of cardiovascular disease. As a result, the US Food and Drug Administration (FDA) issued a warning to caution prescribing azithromycin in patients with known risk factors, such as existing QT interval prolongation, TdP, electrolyte abnormalities, uncompensated heart failure, bradycardia, and concomitant use of drugs that prolong the QT interval.5 Since then, 4 more studies6-9 have examined the cardiac risks with azithromycin by measuring cardiac deaths, cardiac events such as ventricular arrhythmias, or both. Three of 5 studies4,6-9 found an increased risk of deaths or arrhythmias with azithromycin. Two meta-analyses10,11 evaluating evidence from randomized clinical trials and observational studies found no significant increase in cardiac events. In a recent study,1 we showed that the prevalence of cardiac risk factors among azithromycin users remained similar before and after the FDA warning. Thus, despite the FDA warning, prescribing practices do not seem to have been modified. This outcome may be owing to the inconsistency in the results from the previous studies, which indicates a need to continue to evaluate this potential association.

    Our present objective was to investigate the association of cardiac events among patients treated with azithromycin and amoxicillin using an active-control, new-user design. Moreover, we sought to evaluate the risk in prespecified subgroups that include elderly patients with cardiovascular disease, patients using concurrent QT-prolonging drugs before and after the FDA warning from May 2012, and the presence of risk factors identified from previous studies.6-9

    Methods

    This cohort study was a population-level analysis of nationally representative databases of prescriptions dispensed from outpatient pharmacies in the United States. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The University of Illinois at Chicago investigational review board deemed that this study was exempt from review and informed consent owing to use of deidentified patient data. Patients prescribed azithromycin or amoxicillin for an acute bacterial infection from outpatient settings were included.

    Data Sources

    We used data from the Truven Health Analytics MarketScan database,12 a private sector health data resource that reflects the health care experience of enrollees covered by the health benefit programs for employers. Data are collected from more than 100 different health insurance companies and are nationally representative of commercially insured enrollees and their dependents as well as Medicare-eligible beneficiaries with employer-provided Medicare supplemental plans.

    Study Cohort

    We identified patients who were dispensed 1 or more azithromycin or amoxicillin prescriptions from January 1, 2009, through June 30, 2015. Multiple short-term dispensings (at least 30 days apart) of an index therapy for a single patient during the study period were treated as unique episodes. Based on the possible mechanism of action, the risk of cardiac events with azithromycin is expected to be acute and transient; thus, any subsequent exposure to the drug would still pose a similar risk regardless of prior exposures. Episodes were included if the patient had at least 365 days of continuous enrollment in a health plan (baseline period) before the dispense date of 1 index medication (index date). Episodes were excluded if the patient had a missing enrollment identification; if the patient was exposed to amoxicillin, azithromycin, clarithromycin, and erythromycin in the 30 days before and 5 days after the index date (follow-up period); or if the supply of the index therapy was greater than 14 days (as possible maintenance therapy).

    Follow-up Period

    Patients were followed up to 5, 10, and 30 days after the index date. Patients were censored if they disenrolled from the health plan or at the end of the follow-up period, whichever occurred first.

    Covariates

    Patient characteristics, including age, sex, geographic region, and insurance type, were collected at the index date. Baseline clinical conditions were categorized using Clinical Classifications Software.13 The software collapses diagnosis and procedure codes from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), which contains more than 14 000 diagnosis codes and 3900 procedure codes.

    Drugs that prolong the QT interval or induce TdP were identified from a publicly available, well-established, and widely recognized list from the CredibleMeds website (see eTable 1 in the Supplement for the full list).14 The drugs were categorized into 24 different therapeutic classes and into known, possible, and conditional risk groups. Any QT-prolonging drugs prescribed within 90 days before the index date were considered. Concurrent use was defined based on an overlap of the days of supply (with a 15% allowed gap15) of QT-prolonging drugs with the index date of the azithromycin or amoxicillin prescription.

    Study End Point

    The outcome was measured as a primary diagnosis during a hospitalization or first-listed emergency department visit based on ICD-9-CM codes for ventricular arrhythmia, palpitation, long QT syndrome, cardiac arrest, syncope, or death (codes 426.82, 427, 427.1, 427.2, 427.4, 427.5, 427.9, 427.41, 427.42, 780.2, 785.1, 798, 798.1, 798.2, or 798.9). This outcome definition was validated in previous studies with positive predictive values greater than 80%.16,17 We also applied an algorithm developed by another validation study to identify the outcomes originating from outpatient settings.18

    Propensity Score

    We used high-dimensional propensity scores to control for any potential confounding between cohorts.19,20 Propensity scores were generated using a semiautomated approach in which 500 covariates were empirically selected across domains of inpatient and outpatient diagnosis, procedures, and drug dispensing. The covariates were ranked in order of their magnitude of association with the exposure. Furthermore, we included a prespecified list of the following variables in the final-exposure propensity score model: age, sex, calendar year, concurrent use of QT-prolonging drugs, and previously identified factors associated with cardiac events with azithromycin use (history of syncope, cardiac dysrhythmias, nonspecific chest pain, and use of antiarrhythmic agents, antiemetics, antidepressants, loop diuretics, and angiotensin-converting enzyme inhibitors). We compared models with and without the prespecified variables by examining the concordance statistic. We also assessed for multicollinearity by checking the variance inflation factor for each prespecified variable. We matched the cohorts 1:1 on high-dimensional propensity scores using the nearest neighbor matching approach with a caliper width of 0.2 SDs of the logit of the high-dimensional propensity score.21 We then assessed the balance of characteristics between cohorts using standardized differences (>10% was considered a significant difference between the groups).22 The eFigure in the Supplement shows the distribution of high-dimensional propensity scores between before matching and after matching.

    Subgroups

    We examined an a priori list of subgroups to determine whether the risk of cardiac events between azithromycin and amoxicillin varied within these groups. We evaluated the risk among the subgroups of elderly patients (aged ≥65 years); patients with a history of syncope, cardiac dysrhythmias, or nonspecific chest pain; and patients with concurrent use of QT-prolonging drugs. A separate subgroup included patients with specific concomitant use of QT-prolonging medicines, including antiarrhythmic agents, antiemetics, antidepressants, loop diuretics, or angiotensin-converting enzyme inhibitors. Last, we examined the risk among patients with and without baseline cardiovascular disease. For each subgroup, the cohorts of patients receiving azithromycin and amoxicillin were matched based on their propensity scores.

    Statistical Analysis

    Data analysis was performed from October 1, 2018, to December 31, 2019. For efficiency, we used a random sample of 25% of the total cohort to generate high-dimensional propensity scores. The unit of analysis consisted of episodes of index therapies. We used the logistic regression model to examine the association of cardiac risks with azithromycin and amoxicillin. We generated odds ratios (ORs) with 95% CIs, with amoxicillin as the reference cohort. We conducted several sensitivity analyses to evaluate the impact on study findings. First, we restricted our outcome definition to cardiac events identified only from inpatient or emergency department visits. Second, instead of using episodes or multiple dispensing of index therapies as our unit of analysis, we evaluated risk among unique patients by examining their first dispensing of index therapy. Third, we examined the risk before and after the FDA warning from 2012. Fourth, we evaluated the outcome at both 10 and 30 days. Fifth, we relaxed our exclusion criterion of a supply of index therapy of more than 14 days. To address the issue of multiple comparisons (n = 13), we applied the Benjamini-Hochberg procedure and presented the findings in eTable 3 in the Supplement. All statistical analyses were completed using SAS Enterprise Guide, version 7.1 (SAS Institute Inc). Two-sided P < .05 indicated significance.

    Results

    The original cohort included more than 44 million episodes of azithromycin and amoxicillin use. After taking a random sample of 25%, 9 507 450 users of azithromycin and amoxicillin (25% of each cohort) were eligible for inclusion. After matching with high-dimensional propensity scores at a 1:1 ratio, the final cohort included 2 141 285 episodes of each index therapy (N = 4 282 570). The combined cohorts included episodes of patients with a mean (SD) age of 35.7 (22.3) years, of whom 52.6% were female and 47.4% were male (Table 1). Except for the preferred provider organization insurance category (standardized difference, 0.18 [>10%]), all of the baseline characteristics were similar between the amoxicillin and azithromycin cohorts. The duration of therapy was 5 days for 87.3% of azithromycin episodes and less than 10 days for 93.2% of amoxicillin episodes.

    More than one-third of episodes (34.5%) included patients with a history of disease of the circulatory systems (Table 2). Among such episodes, 20.8% had hypertension, 11.1% had nonspecific chest pain, and 7.3% had cardiac dysrhythmias. More than one-half of episodes (53.5%) included patients with a history of respiratory infection, and almost one-quarter (23.7%) included patients with a mental illness. The prevalence of these conditions was similar between azithromycin and amoxicillin users. More than one-fifth of episodes (20.4%) included at least 1 concurrent use of a QT-prolonging drug, and 6.6% used at least 2 concurrent drugs with the index therapies (Table 3). Antidepressants (7.3%) and opiate agonists (2.3%) were the most commonly prescribed drugs. The use of at least 1 QT-prolonging drug with known risk was 6.0%; with possible risk, 5.5%; and with conditional risk, 12.8%. The use of concurrent QT-prolonging drugs was similar between the amoxicillin and azithromycin cohorts.

    Within 5 days after treatment initiation, the occurrence of cardiac events among the overall cohort was 0.03% (n = 1474), or 3.4 events per 10 000 episodes (Table 4). The most frequent diagnoses of cardiac events included syncope at 70.0% and palpitations at 22.5%. Seven hundred eight cardiac events occurred in amoxicillin users and 766 in azithromycin users within 5 days. The prevalence of cardiac events within 10 days was 4.9 events per 10 000 episodes; within 30 days, 10.0 events per 10 000 (eTable 2 in the Supplement).

    Compared with amoxicillin, the odds of cardiac events within 5 days with azithromycin were similar (OR, 1.08; 95% CI, 0.98-1.20) (Table 5). Similar findings were observed when extending the follow-up period to 10 (OR, 1.05; 95% CI, 0.97-1.15) and 30 (OR, 0.98; 95% CI, 0.92-1.04) days because there was no association between azithromycin use and cardiac events when compared with amoxicillin. In all subgroups, the rate of events was higher in the azithromycin group; however, none of the events were statistically significant. One exception was when the analysis was restricted to patients using QT-prolonging medications. In this subgroup, the azithromycin users had a 40% increased risk of cardiac events during the 5 days of follow-up when compared with amoxicillin users (OR, 1.40; 95% CI, 1.04-1.87). The results of the sensitivity analyses of unique patients before and after the FDA warning, cardiac events identified from inpatient diagnosis only, and inclusion of patients regardless of their supply duration were consistent with the findings from our base-case analysis.

    Discussion

    This study contributes evidence to the association of cardiac events with azithromycin. Using real-world data from an extensive claims database, we found no increased odds of cardiac events with azithromycin within 5 days after treatment initiation, but a small, clinically relevant association was found among patients concurrently using a QT-prolonging drug. Elderly patients, those with a history of cardiovascular disease, and those with other baseline risk factors had numerically higher odds of cardiac events with azithromycin.

    The evidence from previous studies has also suggested a small increase in the absolute risk of cardiac arrhythmias with azithromycin, but such risk, when compared with that of another antibiotic, is negligible. In 2017, Trifirò et al9 conducted a study using data from 7 European countries and found no increase in cardiac arrhythmias with azithromycin compared with amoxicillin. Our findings are consistent with those of Trifirò et al9 but not with those of 2 other studies that similarly examined the risk among the US Veterans Health Administration6 and Taiwanese cohorts.7 Compared with other studies, we included a younger cohort (on average, 10 years younger) and did not exclude patients based on their comorbid conditions to make our findings more generalizable. We also included the cardiac events syncope and tachycardia in our outcome definition because we strongly believe both conditions are intermediaries in the sequelae of cardiac arrhythmias. We also controlled for the previously identified factors associated with cardiac events among azithromycin users. Last, we used the high-dimensional propensity score method, whereas previous studies used a traditional propensity scoring method or matching to a control group for confounding. Such differences in cohort, definitions, and study methods could explain the inconsistencies in the findings between the studies.

    More than 200 drugs are associated with prolonging the QT interval.14 Such drugs may confound the association between azithromycin and cardiac events. Thus, similar to the study by Trifirò et al,9 we controlled for the confounding effects of concurrent use of QT-prolonging drugs with azithromycin. In the subgroups, including episodes with concomitant use of QT-prolonging drugs, we found a significantly higher risk with azithromycin. Although we were unable to examine the risk of cardiac events by each therapeutic class of QT-prolonging drugs, we found that most episodes included patients who were prescribed QT-prolonging antidepressants. This finding suggests that cardiac risk may be modified by the presence of any QT-prolonging drug and not just antiarrhythmic drugs, as previously thought. In a sample of the cohort, we explored the likelihood of patients receiving a QT-prolonging medication with azithromycin vs amoxicillin. We found that patients were more likely to be prescribed QT-prolonging drugs with azithromycin. Although we were not able to ascertain the reasons for such findings, accounting for the use of QT-prolonging drugs when examining the risks associated with azithromycin may be necessary. We hypothesize that the confounding effect of other QT-prolonging drugs may help explain and resolve the inconsistent findings from previous studies.

    Strengths and Limitations

    Our study has several strengths. Given a large cohort without many inclusion and exclusion criteria, our findings are likely to be generalizable to the entire population in the United States. We used a robust method and an extensive list of previously identified confounders to control for confounding.

    Our findings are to be interpreted within the context of observational studies and their limitations. First, retrospective observational studies are always prone to unmeasured confounding and measurement error. Despite the large denominator, the power of the study was low. We used data from administrative health care claims, which are primarily used for billing and not research purposes. We were unable to ascertain essential risk factors, including race, smoking status, use of over-the-counter drugs, abnormal electrolyte levels, recent QT-interval prolongation, or body mass index because these variables were not available in the database. Because we could not measure these factors, we used high-dimensional propensity scores in our study such that proxies of these unmeasured confounders might be included, resulting in balance across the exposure groups. Although we were able to measure the outcome of death using ICD-9-CM codes, we know this would underestimate the number of deaths because we lacked information on death occurring in settings other than during an inpatient hospitalization. From our database, we did not have access to the reasons for death (cardiac vs no cardiac). As a result, our study findings should be interpreted in the context of risk of cardiac events and not death.

    Conclusions

    In this large cohort study including more than 4 million episodes of both azithromycin and amoxicillin use, we found no increased odds of cardiac events with azithromycin in the overall cohort but a significantly higher odds among patients prescribed concurrent QT-prolonging drugs with azithromycin. Alternative antibiotic therapies should be considered among these patients.

    Back to top
    Article Information

    Accepted for Publication: July 3, 2020.

    Published: September 15, 2020. doi:10.1001/jamanetworkopen.2020.16864

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Patel H et al. JAMA Network Open.

    Corresponding Author: Haridarshan Patel, PharmD, PhD, Department of Pharmacy Systems, Outcomes, and Policy, College of Pharmacy, University of Illinois at Chicago, 833 S Wood St, MC, Chicago, IL 60610 (haripatel86@gmail.com).

    Author Contributions: Drs Patel and Lee had full access to all of the data in the study and take 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: All authors.

    Drafting of the manuscript: Patel, DiDomenico, Schumock, Suda.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Patel, Calip, Schumock.

    Obtained funding: Patel.

    Administrative, technical, or material support: Patel, Lee.

    Supervision: Suda, Lee.

    Data definitions: Suda.

    Conflict of Interest Disclosures: Dr Calip reported current employment with Flatiron Health, Inc, which is an independent subsidiary of the Roche Group. Dr DiDomenico reported receiving personal fees from ABIOMED and grants from CSL Behring outside the submitted work. No other disclosures were reported.

    Disclaimer: The opinions expressed are those of the authors and do not represent those of the Department of Veterans Affairs or the US Government.

    References
    1.
    Patel  H, Calip  GS, DiDomenico  RJ, Schumock  GT, Suda  KJ, Lee  TA.  Prevalence of cardiac risk factors in patients prescribed azithromycin before and after the 2012 FDA warning on the risk of potentially fatal heart rhythms.   Pharmacotherapy. 2020;40(2):107-115. doi:10.1002/phar.2355PubMedGoogle ScholarCrossref
    2.
    Ohtani  H, Taninaka  C, Hanada  E,  et al.  Comparative pharmacodynamic analysis of Q-T interval prolongation induced by the macrolides clarithromycin, roxithromycin, and azithromycin in rats.   Antimicrob Agents Chemother. 2000;44(10):2630-2637. doi:10.1128/AAC.44.10.2630-2637.2000 PubMedGoogle ScholarCrossref
    3.
    Hancox  JC, Hasnain  M, Vieweg  WVR, Crouse  ELB, Baranchuk  A.  Azithromycin, cardiovascular risks, QTc interval prolongation, torsade de pointes, and regulatory issues: a narrative review based on the study of case reports.   Ther Adv Infect Dis. 2013;1(5):155-165. doi:10.1177/2049936113501816 PubMedGoogle Scholar
    4.
    Ray  WA, Murray  KT, Hall  K, Arbogast  PG, Stein  CM.  Azithromycin and the risk of cardiovascular death.   N Engl J Med. 2012;366(20):1881-1890. doi:10.1056/NEJMoa1003833 PubMedGoogle ScholarCrossref
    5.
    US Food and Drug Administration. FDA drug safety communication: azithromycin (Zithromax or Zmax) and the risk of potentially fatal heart rhythms. Updated February 14, 2018. Accessed February 21, 2018. https://www.fda.gov/Drugs/DrugSafety/ucm341822.htm
    6.
    Rao  GA, Mann  JR, Shoaibi  A,  et al.  Azithromycin and levofloxacin use and increased risk of cardiac arrhythmia and death.   Ann Fam Med. 2014;12(2):121-127. doi:10.1370/afm.1601 PubMedGoogle ScholarCrossref
    7.
    Chou  H-W, Wang  J-L, Chang  C-H, Lai  C-L, Lai  M-S, Chan  KA.  Risks of cardiac arrhythmia and mortality among patients using new-generation macrolides, fluoroquinolones, and β-lactam/β-lactamase inhibitors: a Taiwanese nationwide study.   Clin Infect Dis. 2015;60(4):566-577. doi:10.1093/cid/ciu914 PubMedGoogle ScholarCrossref
    8.
    Svanström  H, Pasternak  B, Hviid  A.  Use of azithromycin and death from cardiovascular causes.   N Engl J Med. 2013;368(18):1704-1712. doi:10.1056/NEJMoa1300799 PubMedGoogle ScholarCrossref
    9.
    Trifirò  G, de Ridder  M, Sultana  J,  et al.  Use of azithromycin and risk of ventricular arrhythmia.   CMAJ. 2017;189(15):E560-E568. doi:10.1503/cmaj.160355 PubMedGoogle ScholarCrossref
    10.
    Almalki  ZS, Guo  JJ.  Cardiovascular events and safety outcomes associated with azithromycin therapy: a meta-analysis of randomized controlled trials.   Am Health Drug Benefits. 2014;7(6):318-328.PubMedGoogle Scholar
    11.
    Gorelik  E, Masarwa  R, Perlman  A, Rotshild  V, Muszkat  M, Matok  I.  Systematic review, meta-analysis, and network meta-analysis of the cardiovascular safety of macrolides.   Antimicrob Agents Chemother. 2018;62(6):e00438-18. doi:10.1128/AAC.00438-18 PubMedGoogle Scholar
    12.
    Hansen  L. The Truven Health MarketScan databases for life sciences researchers. March 2018. Accessed March 1, 2018. https://www.ibm.com/watson-health/about/truven-health-analytics
    13.
    Healthcare Cost and Utilization Project (HCUP). Clinical Classifications Software (CCS) for ICD-9-CM. Agency for Healthcare Research and Quality. Updated March 6, 2017. Accessed March 1, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp
    14.
    Woosley  RL, Heise  CW, Gallo  T, Tate  J, Woosley  D, Romero  KA. QTdrugs List. Accessed March 1, 2018. https://www.crediblemeds.org/
    15.
    Psaty  BM, Siscovick  DS, Weiss  NS,  et al.  Hypertension and outcomes research: from clinical trials to clinical epidemiology.   Am J Hypertens. 1996;9(2):178-183. doi:10.1016/0895-7061(96)00015-5 PubMedGoogle ScholarCrossref
    16.
    De Bruin  ML, van Hemel  NM, Leufkens  HGM, Hoes  AW.  Hospital discharge diagnoses of ventricular arrhythmias and cardiac arrest were useful for epidemiologic research.   J Clin Epidemiol. 2005;58(12):1325-1329. doi:10.1016/j.jclinepi.2005.04.009 PubMedGoogle ScholarCrossref
    17.
    Tamariz  L, Harkins  T, Nair  V.  A systematic review of validated methods for identifying ventricular arrhythmias using administrative and claims data.   Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):148-153. doi:10.1002/pds.2340 PubMedGoogle ScholarCrossref
    18.
    Hennessy  S, Leonard  CE, Freeman  CP,  et al.  Validation of diagnostic codes for outpatient-originating sudden cardiac death and ventricular arrhythmia in Medicaid and Medicare claims data.   Pharmacoepidemiol Drug Saf. 2010;19(6):555-562. doi:10.1002/pds.1869 PubMedGoogle ScholarCrossref
    19.
    Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.   Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663cc PubMedGoogle ScholarCrossref
    20.
    Rassen  JA, Doherty  M, Huang  W, Schneeweiss  S. Pharmacoepidemiology toolbox. Accessed June 1, 2019. https://www.drugepi.org/
    21.
    Austin  PC.  Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies.   Pharm Stat. 2011;10(2):150-161. doi:10.1002/pst.433 PubMedGoogle ScholarCrossref
    22.
    Austin  PC.  Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research.   Commun Stat Simul Comput. 2009;38(6):1228-1234. doi:10.1080/03610910902859574 Google ScholarCrossref
    ×