Article Text

Hypersensitivity reactions with allopurinol and febuxostat: a study using the Medicare claims data
  1. Jasvinder A. Singh1,2,
  2. John D. Cleveland1
  1. 1 Departments of Medicine and Epidemiology, University of Alabama at Birmingham, Birmingham, Alabama, USA
  2. 2 Medicine Service, Birmingham Veterans Affairs Medical Center, Birmingham, Alabama, USA
  1. Correspondence to Dr Jasvinder A. Singh, University of Alabama at Birmingham, Birmingham, AL 35294, USA;{at}


Objective To assess the risk of hypersensitivity reactions (HSRs) with allopurinol and febuxostat in a population-based study.

Methods We used the 5% Medicare beneficiary sample (≥65 years) from 2006 to 2012 to identify people with a newly filled prescription for allopurinol, febuxostat or colchicine. We used multivariable-adjusted Cox regression analyses to compare the hazard ratio (HR) of incident HSRs with allopurinol or febuxostat use versus colchicine use; separate analyses were done in people exposed to allopurinol. Propensity-matched analyses (5:1) compared hazards with allopurinol versus febuxostat.

Results Crude incidence rates of HSRs were as follows: allopurinol, 23.7; febuxostat, 30.7; and colchicine, 25.6 per 1000 person-years. Compared with colchicine, allopurinol, febuxostat and febuxostat+colchicine were associated with significantly higher HRs of HSRs, 1.32 (95% CI: 1.10 to 1.60) and 1.54 (95% CI: 1.12 to 2.12) and 2.17 (95% CI: 1.18 to 3.99), respectively. In propensity-matched analyses, febuxostat did not significantly differ from allopurinol; HR for HSRs was 1.25 (95% CI: 0.93 to 1.67). Compared with allopurinol start dose <200 mg/day, allopurinol start dose ≥300 mg/day, diabetes and female sex were associated with significantly higher hazard of HSRs, 1.27 (95% CI: 1.12 to 1.44), 1.21 (95% CI: 1.00 to 1.45) and 1.32 (95% CI: 1.17 to 1.48), respectively. The majority (69%) of HSRs occurred in the outpatient setting.

Conclusions Compared with colchicine, allopurinol and febuxostat similarly increased the risk of HSRs. Allopurinol and febuxostat did not differ from each other. In allopurinol users, starting dose, female sex and diabetes increased this risk, findings that need further study.

  • gout
  • epidemiology
  • outcomes research

Statistics from

Key messages

What is already known about this subject?

  • In two previous studies, the risk factors for allopurinol hypersensitivity reactions (HSRs) were female sex, age ≥60 years, renal or cardiovascular comorbidities in a Taiwanese study, and a higher allopurinol start dose in both Taiwanese and New Zealand study. No population-level comparative risk data of HSRs associated with allopurinol and febuxostat are available in non-Taiwanese.

What does this study add?

  • In propensity-matched analyses using the 5% US Medicare data, we found that compared with colchicine, both allopurinol and febuxostat were associated with significantly increased hazards of HSRs, and febuxostat did not significantly differ from allopurinol regarding the hazard of HSRs.

  • In allopurinol users, allopurinol starting dose, diabetes and female sex increased the risk of HSRs.

How might this impact on clinical practice or future developments?

  • Our study provides an estimate of the HSRs with urate-lowering therapies (ULTs) in general US population 65 years or older and shows that the two most commonly used ULTs are associated with a similar hazard of HSRs.

  • Future studies need to examine as to why a higher allopurinol start dose, diabetes and female sex increase the risk of HSRs.


A hypersensitivity reaction (HSR) is an important, but rare/uncommon adverse event associated with the use of allopurinol,1 the most commonly used urate-lowering therapy (ULT) for the treatment of gout.2 3 A recent study using the nationwide Taiwanese data showed that the annual incidence rate of allopurinol cutaneous reactions was 4.68 per 1000 new users.4 Female sex, older age (≥60 years), renal or cardiovascular comorbidities and initial allopurinol dosage exceeding 100 mg/day were risk factors.4 In a recent retrospective study from New Zealand, a higher allopurinol start dose was significantly associated with a higher risk of allopurinol hypersensitivity syndrome (AHS),5 a severe form of HSR. Reviews indicated that HLA 58*01, Han Chinese/European ancestry and renal failure were other potential risk factors for AHS.6 7 The same group of authors also showed that in people who tolerate allopurinol, allopurinol dose can be safely increased including patients with renal failure.8 The findings of these studies have not been reproduced. Similar studies have not been conducted in populations from the USA or the European countries. Comparative studies of allopurinol versus febuxostat are limited to some ethnicities only.9

Our study objective was to use a nationally representative sample of older US adults to: (1) examine the crude incidence rates of HSRs with allopurinol, febuxostat or colchicine; (2) assess whether compared with colchicine, allopurinol or febuxostat use were independently associated with an increased risk of HSRs and whether allopurinol differed from febuxostat; and (3) evaluate whether starting allopurinol dose and comorbidity are significantly associated with a higher risk of HSRs in allopurinol users. We hypothesised that compared with febuxostat or colchicine, allopurinol use will significantly increase the risk of HSRs, this risk will be the highest in the initial few months of use and that initial allopurinol dose and the presence of renal failure will be risk factors.


Study cohort

We used the 5% random sample of Medicare Beneficiaries (≥65 years) from 2006 to 2012 to perform this study, obtained from the Centers for Medicare and Medicaid Services Chronic Condition Data Warehouse, and widely used for epidemiological research by others10 11 and us.12 13

We used the beneficiary summary file and Medicare Part D file to identify beneficiaries enrolled in Medicare fee-for-service with pharmacy coverage (parts A, B and D; files contain all insurance claims for each beneficiary) and not enrolled in a Medicare Advantage Plan, lived in the USA during the study period (2006–2012) and filled a new prescription of allopurinol or febuxostat or colchicine, three common medications used for gout and/or hyperuricaemia. We used an incident (or new) user design rather than a prevalent user design, which reduces bias by avoiding adjustment for characteristics that may be in the causal pathway and allows capture of both early and late events.14

We reported methods and results as recommended in the Strengthening of Reporting in Observational studies in Epidemiology (STROBE) statement.15

Exposure of interest: new treatment with allopurinol or febuxostat

All prescription claims, including drug name, dose and supply, were obtained from Medicare Part D file. A beneficiary began a new allopurinol (or febuxostat or colchicine) treatment episode by filling an allopurinol (or febuxostat or colchicine) prescription, provided they had not filled an allopurinol (or febuxostat or colchicine) prescription in the previous 365 days. This baseline 365-day clean period applied to all initial prescription episodes, whether they were the initial exposure or subsequent switching to another medication.

We assigned days of exposure for each allopurinol (or febuxostat or colchicine), calculated based on the days’ supply variable provided in the Medicare Part D file and also included a 30-day residual/grace period. Continuous allopurinol (or febuxostat or colchicine) episode ended after 30 days of the end of allopurinol (or febuxostat or colchicine) exposure. If there were >30 days between prescription fills, new allopurinol (or febuxostat or colchicine) exposure started. Thus, a patient receiving a single 90-day prescription was considered exposed for 120 days, which included a 30-day residual period as medication adherence is usually imperfect, and the residual period accounts for any residual biological effects of medications. If a patient had prescriptions for more than one of these three drugs, then they were considered exposed to all prescribed drugs on a given day.

The main predictor of interest was new allopurinol or febuxostat use, with new colchicine use as the reference category, as colchicine is a common non-ULT medication (with a different mechanism of action) used by patients with gout. We assessed all allopurinol (<200, 200–299 and ≥300 mg/day), febuxostat (40 and 80 mg/day) and colchicine (0.6, 0.6-<1, 1-<2 and ≥2 mg/day) start doses, considering only the first prescription. We calculated the daily allopurinol (or febuxostat or colchicine) dose as the mean daily use for each continuous allopurinol, febuxostat or colchicine treatment episode. For each allopurinol, febuxostat or colchicine treatment episode, we categorised the duration of use by 30-day intervals.

Study outcome

The outcome of interest for our study was incident HSR, defined using a validated algorithm. The main analysis follows the definition of potential HSR similar to that by Wright et al 16 using eosinophilia (288.3), arthropathy associated with HSR (713.6), other anaphylactic reaction (995.0), unspecified adverse effect of unspecified drugs, medicinal and biological substance not elsewhere classified (995.2), or allergy, unspecified, not elsewhere classified (995.3). We added a baseline exclusion for codes E930 to E949 to increase the specificity of HSR. For sensitivity analyses, we included another definition of incident HSR, per Strom et al 17 based on the presence of codes 995.2, 995.0 or 995.3. Our outcome of overall HSRs is different than AHS that has been studied previously.


Vital statistics (birthdate, date of death), and monthly entitlement indicators (A/B/C/D) were obtained from the Medicare claims data. We assessed several important covariates at baseline, including the patient demographics (age, gender, race/ethnicity), medical comorbidity and the use of medications for cardiovascular diseases from Medicare denominator and other claims files. We assessed the Charlson-Romano comorbidity index score, a validated weighted comorbidity index developed for claims data analysis18 adapted from the Charlson index.19 Cardiovascular medications (statins, beta-blockers, diuretics and angiotensin-converting enzyme (ACE)-inhibitors) served as markers of cardiovascular disease.

Patient and public involvement

The study question was informed by patients in our gout clinic who questioned whether there were any differences in the risk of HSRs with gout medications. There was no formal patient involvement in the study design.

Statistical analyses

We compared baseline characteristics of new medication exposure episodes with versus without incident HSRs and calculated crude incidence rates of HSRs for new allopurinol (or febuxostat or colchicine) episodes. As noted in the methods, we tested two different definitions to assess incidence rates, Wright et al 16 and Strom et al,17 with and without the additional E codes for baseline exclusion. The Wright definition with E codes16 was used for all other analyses.

For assessing comparative risk of HSR with allopurinol or febuxostat versus colchicine, we performed multivariable-adjusted Cox proportional hazard regression analyses adjusted for demographics (age, gender and race), cardiovascular medications (statins, beta-blockers, diuretics and ACE inhibitors) and Charlson comorbidity index to control for differences between patients exposed to allopurinol versus febuxostat versus colchicine. We used the Huber-White ‘Sandwich’ variance estimator to account for correlations between observations from the same patient.20 We calculated the HR of new HSRs for new allopurinol or febuxostat use versus colchicine use (reference category); sensitivity analyses limited to people with gout were done.

We performed propensity-matched analyses comparing new allopurinol with new febuxostat use, in a 5:1 ratio. Propensity matching included age, gender, race, Charlson-Romano comorbidity score, region, each Charlson-Romano comorbidity, risk factors for coronary artery disease and current use of medications for cardiovascular diseases (statins, beta-blockers, diuretics and ACE inhibitors). We explored the association of daily dose and duration of use with the risk of HSRs in people with the same propensity of receiving either drug, allopurinol or febuxostat.

We also performed separate multivariable-adjusted analyses in allopurinol exposed people to assess the effect of allopurinol start dose and duration of use and whether specific comorbidities such as coronary artery disease, heart failure, renal failure, liver disease, diabetes and malignancy were associated. We assessed these comorbidities, as they may impact the metabolism of ULT drugs, and/or concomitant medications, which might have drug–drug interactions with ULT. We used SAS 9.0 (Cary, North Carolina, USA) to perform the analyses. A p-value <0.05 was considered statistically significant.


Study cohort characteristics

Of 39 261 people who contributed 66 178 new episodes of allopurinol/febuxostat/colchicine use, 1038 ended in a HSR; the majority (69%) occurred in an outpatient setting. The mean age was 77 years, 50% were men, 77% white and the mean Charlson-Romano score was 2.6 (table 1). Episodes with versus without HSRs had similar characteristics except higher comorbidity (table 1).

Table 1

Demographic and clinical characteristics of incident HSRs in people with allopurinol or febuxostat or colchicine exposure

Crude incidence rates of HSRs

Crude incidence rates of HSRs were as follows: allopurinol, 23.9; febuxostat, 30.5; and colchicine, 25.7 per 1000 person-years (table 2). Crude incidence rates were higher in people using medication combinations, particularly febuxostat+colchicine and febuxostat+colchicine+allopurinol, 56.8 and 89.1 per 1000 person-years, respectively (table 2).

Table 2

Crude incidence rate of incident hypersensitivity reactions with allopurinol, febuxostat or colchicine exposure†

Crude incidence rates of HSRs were higher for allopurinol start dose ≥300 mg/day compared with lower doses, but not for higher febuxostat start dose (table 2). Rates were highest in the first 30 days of exposure to allopurinol, 73.9 and febuxostat, 60.6/1000 person-years, with a progressive reduction over time. However, HSR rates remained fairly stable for colchicine up to 120 days, with a slight reduction for use >120 days (table 2).

Incidence rates varied little, when we used alternate definitions for HSRs except for the Strom et al definition with no events for the simultaneous use of allopurinol and febuxostat (online supplementary table S1).

Multivariable-adjusted comparative risk of HSRs with allopurinol or febuxostat compared with colchicine

In multivariable-adjusted analyses, compared with colchicine, the use of allopurinol, febuxostat or febuxostat+colchicine combination was associated with significantly higher HRs of HSRs, 1.32 (95% CI: 1.10 to 1.60) and 1.54 (95% CI: 1.12 to 2.12) and 2.17 (95% CI: 1.18 to 3.99), respectively (table 3). Women had 1.28-fold higher HR of HSRs (table 3). When we limited the analyses to people with gout (n=883; fewer cases), all findings were replicated except that allopurinol exposure was no longer significantly associated (table 3; model 2).

Table 3

Multivariable-adjusted association of allopurinol or febuxostat use and other risk factors with incident hypersensitivity reactions

Propensity-matched comparative risk of HSRs with febuxostat compared with allopurinol

Most differences were eliminated/reduced after propensity matching, and the mean propensity scores in 5:1 propensity-matched allopurinol to febuxostat exposures were the same (online supplementary table S2). In a propensity-matched analysis, febuxostat did not significantly differ from allopurinol, the HR of HSRs was 1.25 (95% CI: 0.93 to 1.67) (table 4).

Table 4

Propensity-matched* association of allopurinol or febuxostat (5:1 match) with the hazard of incident hypersensitivity reactions

In separate exploratory analyses, compared with allopurinol daily dose <200 mg/day, febuxostat 40 mg/day was associated with a significantly higher hazard of HSRs with HR of 1.43 (95% CI: 1.03 to 1.98) (online supplementary table S3); other allopurinol or febuxostat doses were not significantly associated.

Allopurinol start dose and duration versus the risk of HSRs

In multivariable-adjusted analyses limited to allopurinol users, compared with allopurinol start dose of <200 mg/day, allopurinol start dose of ≥300 mg/day was associated with a significantly higher hazard of HSRs, 1.27 (95% CI: 1.12 to 1.44) (table 5). Compared with the use duration of 1–30 days, longer use durations were associated with non-significantly HRs (table 5). Of the six prespecified comorbidities, only diabetes was significantly associated with a higher risk (table 5).

Table 5

Association of various risk factors with incident hypersensitivity reactions in patients who received allopurinol


In this study of older Americans, we examined HSRs in patients newly exposed to allopurinol or febuxostat compared with colchicine. Our study examined all HSRs, not just AHS. Compared with colchicine, both allopurinol and febuxostat use were associated with significantly higher hazards of HSRs. In propensity-matched analyses, febuxostat was not significantly different from allopurinol regarding the risk of new HSRs. Female sex, diabetes and a higher allopurinol start dose were also risk factors. Several findings merit further discussion.

Our study used a representative US national sample. It is among the first studies to describe HSR incidence rates with a new prescription of allopurinol or febuxostat. Notably, febuxostat combination with colchicine and triple combination (allopurinol, febuxostat and colchicine) were associated with much higher HSR rates. HSR incidence decreased 2 months after allopurinol or febuxostat initiation. This might represent the time course of HSRs that occur more commonly earlier, or the development of immune tolerance in some other people with time; this requires further study.

In a Taiwanese population-based cohort, cutaneous adverse reaction incidence rates were higher in allopurinol versus febuxostat, 15.37 vs 3.48 per 1000 person-years.9 Although our HSR rates in allopurinol exposed were similar to those in the Taiwanese study at 24/1000 person-years, rates in febuxostat-exposed patients were much higher at 31/1000 person-years in our study. Important differences in studies were the country setting (USA vs Taiwan), population (older adults vs all ages) and the definition of outcome (all HSRs vs cutaneous adverse reactions). We caution the reader that despite a large sample size for the study, estimates for combinations are less precise and can change significantly with a difference of a few cases, which can easily happen by chance.

In adjusted analyses, compared with colchicine use, new allopurinol (HR, 1.32) or febuxostat use (HR, 1.54) were each independently associated with a higher risk of HSRs. Findings add significantly to the limited literature for febuxostat associated reactions.21 22 The similarity of the risk of HSRs between allopurinol and febuxostat is novel.

A retrospective case–control study of 54 AHS cases (defined as the presence of two of the major criteria -worsening renal function, acute hepatocellular injury or rash; or one of the major criteria and one of the minor criteria (fever, eosinophilia or leukocytosis)) found that allopurinol start dose was higher in cases versus controls, 184 s. 112 mg/day.5 Our findings add to the growing knowledge in this area.

We found in the overall analyses that compared with a starting allopurinol dose <200 mg/day, a higher allopurinol start dose of ≥300 mg/day was associated with a significantly higher hazard of HSRs in multivariable-adjusted analyses, 1.27 (95% CI: 1.12 to 1.44), but non-significantly associated in propensity-matched analyses, 1.25 (95% CI: 0.99 to 1.58). This is an interesting finding. We believe that the similarity of HRs in the two analyses indicates that the possibility of harm with a higher allopurinol start dose, but this possibility ranged from no harm, lower 95% CI of 0.99, to significant harm, higher 95% CI of 1.58 in the propensity-matched analyses. This has important clinical implications. Cautious clinicians may start allopurinol at a low dose and gradually escalate the dose to achieve target serum urate to potentially decrease the risk of HSRs and allow patients to achieve target sUA, an important disease target.

Our findings regarding the risk factors for allopurinol HSRs differ somewhat from a recent Taiwanese study, which found that female sex, age ≥60 years, initial allopurinol dose >100 mg/day, renal and cardiovascular comorbidity were significant risk factors for allopurinol-associated HSRs.4 Our study found that allopurinol dose >300 mg/day, female sex and diabetes, but not malignancy, heart failure, myocardial infarction or renal disease were significantly associated with a higher risk of allopurinol HSRs. The independent association of diabetes with a higher risk of HSRs might be due to subclinical renal disease, and needs further confirmation. Thus, both studies independently found that female gender and higher allopurinol start dose were significant predictors, increasing confidence in these results. Differences in study findings may be due to the differences in study population (US adults≥65 years vs Taiwanese general population), medications studied (allopurinol, febuxostat and colchicine vs allopurinol only), the unit of analysis (episodes vs people) and analyses (Cox vs logistic regression).

Our study has several limitations, which must be considered while interpreting study findings. Despite the use of a validated algorithm to define HSR,16 we are aware that the use of diagnostic codes may be associated with some misclassification bias, as well as underdocumentation or overdocumentation of HSR. The validation of Wright et al algorithm was in inpatient or emergency room setting16 and of Strom et al included all cases17; our study that used all settings and might have had lower PPV of the algorithm than that studied by Wright et al.16 Due to observational study design, residual confounding is possible. However, it is unlikely that a large prospective cohort consisting of 50–100 000 patients using these medications will be performed; therefore, a retrospective database study is one of the most practical ways to examine rare events such as the HSR. These findings are not generalisable to younger people, as our study sample consists of older adults, ≥65 years. Our results only apply to 30-day or longer use of these medications; future studies can examine associations for a shorter duration of exposure, more common with colchicine versus other medications. Study strengths include the use of a representative US population, the availability of clinical and pharmacy fill data, the use of robust observational study design and analytic methods, and confirmation of study findings using several sensitivity analyses.


We described the crude incidence rates of HSRs in patients with new exposure to allopurinol, febuxostat or colchicine. In this study of a representative US population, we found that compared with colchicine, both allopurinol and febuxostat use was associated with a higher risk of HSRs. In propensity-matched analyses, febuxostat did not differ significantly from allopurinol regarding the risk of HSRs. In allopurinol users, start dose >300 mg/day was associated with a higher risk of HSRs, compared with lower doses. In allopurinol users, diabetes, but not a renal failure, and female sex were independent risk factors for incident HSRs. Our findings provide new knowledge regarding HSRs to these commonly used medications for the treatment of gout that can inform clinicians and patients alike starting these medications for the treatment of gout and other disorders.


The authors thank Dr Jeffrey Curtis of the UAB Division of Rheumatology, who permitted us to reuse the 5% Medicare data. They thank patients at the University of Alabama gout clinic for asking us the question regarding the relative risk of allergic reactions with allopurinol versus febuxostat, which prompted them to conduct this study. This manuscript is based on the work previously presented at a conference and published as a conference abstract presented at the 2017 ACR Annual Meeting:


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Josef S Smolen

  • Contributors JAS: designed the study, developed study protocol, reviewed analyses and wrote the first draft of the paper. JC: performed the data abstraction and data analyses. All authors revised the manuscript, read and approved the final manuscript.

  • Funding This material is the result of work supported by research funds from the Division of Rheumatology at the University of Alabama at Birmingham and the resources and use of facilities at the Birmingham VA Medical Center, Birmingham, Alabama, USA. The funding body did not play any role in design, in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.

  • Competing interests JAS has received consultant fees from Crealta/Horizon, Medisys, Fidia, UBM LLC, Trio health, Medscape, WebMD, Clinical Care options, Clearview healthcare partners, Putnam associates, Focus forward, Navigant consulting, Spherix, Practice Point communications, the National Institutes of Health and the American College of Rheumatology. JAS owns stock options in Amarin pharmaceuticals and Viking therapeutics. JAS is on the speaker’s bureau of Simply Speaking. JAS is a member of the executive of OMERACT, an organization that develops outcome measures in rheumatology and receives arms-length funding from 12 companies. JAS serves on the FDA Arthritis Advisory Committee. JAS is the chair of the Veterans Affairs Rheumatology Field Advisory Committee. JAS is the editor and the Director of the UAB Cochrane Musculoskeletal Group Satellite Center on Network Meta-analysis. JAS previously served as a member of the following committees: member, the American College of Rheumatology’s (ACR) Annual Meeting Planning Committee (AMPC) and Quality of Care Committees, the Chair of the ACR Meet-the-Professor, Workshop and Study Group Subcommittee and the co-chair of the ACR Criteria and Response Criteria subcommittee.

  • Patient consent for publication Not required.

  • Ethics approval The University of Alabama at Birmingham’s Institutional Review Board approved this study and all investigations were conducted in conformity with ethical principles of research (X1201207004). The IRB waived the need for informed consent of patients for this database study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available. These data can be obtained from the Centers for Medicare and Medicaid Services (CMS) Chronic Condition Data Warehouse. The authors are ready to share the data with colleagues, after obtaining appropriate permissions from the Centers for Medicare and Medicaid Services (CMS) Chronic Condition Data Warehouse and the University of Alabama at Birmingham (UAB) Ethics Committee, related to HIPAA and Privacy policies.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.