Article Text

Extended report
Are allopurinol dose and duration of use nephroprotective in the elderly? A Medicare claims study of allopurinol use and incident renal failure
1. Jasvinder A Singh1,2,3,
2. Shaohua Yu2
1. 1Medicine Service, Birmingham VA Medical Center, Birmingham, Alabama, USA
2. 2Division of Epidemiology, Department of Medicine School of Medicine, School of Public Health, University of Alabama at Birmingham (UAB), Birmingham, Alabama, USA
3. 3Department of Orthopedic Surgery, Mayo Clinic College of Medicine, Rochester, Minnesota, USA
1. Correspondence to Dr Jasvinder A Singh, University of Alabama, Faculty Office Tower 805B, 510 20th Street S, Birmingham, AL 35294, USA: Jasvinder.md{at}gmail.com

## Abstract

Objective To assess the effect of allopurinol dose/duration on the risk of renal failure in the elderly with allopurinol use.

Methods We used the 5% random Medicare claims data from 2006 to 2012. Multivariable-adjusted Cox regression analyses assessed the association of allopurinol dose/duration with subsequent risk of developing incident renal failure or end-stage renal disease (ESRD) (no prior diagnosis in last 183 days) in allopurinol users, controlling for age, sex, race and Charlson–Romano comorbidity index. HRs with 95% CIs were calculated. Sensitivity analyses considered a longer baseline period (365 days), controlled for gout or used more specific codes.

Results Among the 30 022 allopurinol treatment episodes, 8314 incident renal failure episodes occurred. Compared with 1–199 mg/day, allopurinol dose of 200–299 mg/day (HR 0.81; 95% CI 0.75 to 0.87) and ≥300 mg/day, 0.71 (0.67 to 0.76), had significantly lower hazard of renal failure in multivariable-adjustment model, confirmed in multiple sensitivity analyses. Longer allopurinol use duration was significantly associated with lower hazards in sensitivity analyses (365-day look-back; reference, <0.5 year): 0.5–1 year, 1.00 (0.88, 1.15); >1–2 years, 0.85 (0.73 to 0.99); and >2 years, 0.81 (0.67 to 0.98). Allopurinol ≥300 mg/day was also associated with significantly lower risk of acute renal failure and ESRD with HR of 0.89 (0.83 to 0.94) and 0.57 (0.46 to 0.71), respectively.

Conclusions Higher allopurinol dose is independently protective against incident renal failure in the elderly allopurinol users. A longer duration of allopurinol use may be associated with lower risk of incident renal failure. Potential mechanisms of these effects need to be examined.

• Gout
• Outcomes research
• Epidemiology

## Introduction

Allopurinol is the most commonly used urate-lowering therapy (ULT).1 ,2 The enzyme xanthine oxidase is responsible for the oxidation of hypoxanthine and xanthine in the purine metabolism pathway, resulting in the production of uric acid. Allopurinol (a purine itself) reduces uricemia by acting as a competitor substrate for xanthine oxidase, which results in increased excretion of hypoxanthine and xanthine, products upstream of uric acid in the purine metabolism.

Recent data have shown that allopurinol use is associated with an improved renal function (or slowing its decline or reduction in proteinuria) in both randomised trials3–8 and observational studies.9–11 Studies of animal models of renal injury have also reported a beneficial effect of allopurinol on renal function,12–16 further supporting this association. Additional evidence comes from the observation that gout, a disease characterised by hyperuricemia, is an independent risk factor for progression of renal failure.17 Given long experience with the safety profile of allopurinol and the evidence to date, a $24.3 million federally funded randomised study is under way to assess whether allopurinol can prevent renal function loss in diabetes.18 One in 10 American adults, or roughly 20 million, have chronic kidney disease (CKD). The incidence of CKD in the USA elderly increased significantly from 2% in 2000 to 4.5% in 2008 and the prevalence increased from 18.8% in 1988–1994 to 24.5% to 2003–2006.19 Thus, CKD is associated with significant population burden. While many studies have shown that allopurinol use is associated with an improvement in renal function,3–11 it is not known whether the magnitude of this benefit varies by allopurinol dose or the duration of use. Therefore, our objective was to examine this question among elderly Americans. We hypothesised that in elderly patients without underlying renal disease and with allopurinol use (1) allopurinol dose and (2) the duration of allopurinol use will each be independently associated with the risk of incident renal failure. We also explored whether allopurinol dose and duration were associated with the risk of acute renal failure and end-stage renal disease (ESRD) including haemodialysis. ## Methods ### Study design and data sources We conducted a retrospective cohort study using Medicare claims data from 2006 to 2012 on a national 5% random sample of Medicare beneficiaries, obtained from the Centers for Medicare and Medicaid Services Chronic Condition Data Warehouse. For each beneficiary, data included information on demographics and insurance coverage; claims for inpatient, outpatient, skilled nursing facility, non-institutional provider, home health, hospice, durable medical equipment services; and claims for prescription drugs. ### Eligible population and outcomes Eligible subjects for the cohort study were Medicare beneficiaries who were 65 years of age or older (age for becoming eligible for the Medicare programme); lived in the USA; were enrolled continuously in traditional Medicare fee-for-service and pharmacy coverage (parts A, B and D) and not enrolled in Medicare Advantage Plan (a health maintenance organization (HMO) plan); and who were newly treated with allopurinol. We defined new episode of treatment with allopurinol as a filled allopurinol prescription after a baseline period of 183 days during which no allopurinol prescription was filled. Baseline patient characteristics were derived from the Medicare denominator file, which included age, gender, race/ethnicity, residence (north-east, south, midwest, west and California). Comorbidity scores in the baseline period for each episode were derived using the Charlson–Romano comorbidity index score, developed for claims data analysis.20 This comorbidity index includes myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, diabetes, as well as other illnesses such as dementia, liver disease, pulmonary disease, rheumatic disease, etc. The main outcome of interest was incident renal failure, that is, the first occurrence of renal failure during the follow-up, with an absence of this diagnosis in the previous 183 days. We identified renal failure by examining the claims for International Classification of Diseases, ninth revision, common modification (ICD-9-CM) codes, 582.xx, 583.xx, 585.xx, 586.xx or 588.xx. For each episode, follow-up began on the earliest allopurinol initiation date during the study period and ended on the earliest of first date of renal failure, losing full Medicare coverage, the date of death or end of the study (31 December 2012). Patients could contribute multiple allopurinol treatment episodes during different time periods. Episodes with patients having renal failure diagnosis within baseline period (183 days before the episode initiation date) were not included for main analysis. We also explored the association of allopurinol use with acute renal failure (ICD-9-CM, 584.xx) or haemodialysis (585.6 or procedure code 39.95), using similar baseline exclusion of renal disease, as described above. ### Allopurinol exposure A filled prescription of allopurinol was characterised as allopurinol exposure/treatment based upon the days' supply for prescription in pharmacy records and a stock carry over not exceeding 30 days after the end of days' supply of allopurinol for prescription. Patients were considered exposed for 30 days after the end of days' supply to capture the attributable events. This 30-day grace period was allowed for two reasons: (1) most patients have some extra days of medication supply due to imperfect adherence and (2) because some residual beneficial effects of allopurinol could extend beyond the last day of use. ### Potential predictors of renal failure and covariates We examined two predictors, the allopurinol dose and the duration of allopurinol use. The average daily allopurinol dose was calculated as the mean allopurinol use per day for each continuous allopurinol episodes and categorised as 1–199, 200–299 and ≥300 mg/day. Continuous allopurinol episode ended after 30 days of the end of allopurinol exposure. If there were more than 30 days between prescription fills, a new allopurinol exposure started and the duration of allopurinol treatment was calculated. Duration of allopurinol use was categorised as 1–180 days, 181–365 days, >1 to 2 years and longer than 2 years. Covariates included age at the start of each episode, sex, race/ethnicity, the Charlson–Romano comorbidity score and the use of beta-blockers, ACE inhibitors, statins and diuretics. ### Statistical analysis The primary analysis included allopurinol episodes with episodes of incident renal failure (ie, no renal failure in the prior baseline period). We compared demographic and clinical characteristics between allopurinol episodes during which patients did versus did not develop incident renal failure in the follow-up period. Cox proportional hazard regression models assessed the association of average daily allopurinol dose and of allopurinol use duration with incident renal failure, using hazard ratio (HR). Univariate and multivariable-adjusted models included age, gender, race, Charlson–Romano comorbidity score and the use of medications for cardiovascular diseases (beta-blockers, ACE inhibitors, statins and diuretics), allopurinol use and the duration of allopurinol use. Since patients could possibly contribute more than one episode of new allopurinol use, we accounted for these correlations using the Huber–White ‘sandwich’ variance estimator and calculated robust standard errors for all estimates.21 Four sets of sensitivity analyses were conducted by (1) extending the baseline period from 183 days (as in the main analysis) to 365 days; (2) adjusting the main model for gout, a potential risk factor for renal failure; (3) limiting to ICD-9-CM codes 582.xx and 585.xx, more specific codes for renal failure; and (4) using the exact dose of allopurinol rather than the average daily allopurinol dose. Exploratory analyses examined the association of allopurinol dose and use duration with acute renal failure or ESRD. ## Results ### Patient and cohort characteristics We identified 30 022 allopurinol treatment episodes with no history of renal failure during baseline period meeting eligibility criteria. Among these episodes, 8314 episodes ended with incident renal failure events (of which 5731 cases occurred during the allopurinol exposure) and 21 708 episodes ended without incident renal failure (figure 1); respective mean ages were 77 and 75 years (table 1). Overall, the treatment episodes were very similar between allopurinol episodes with and without renal failure except a higher Charlson–Romano comorbidity score (mean of 1.91 vs 1.38) in the episodes with versus without incident renal failure (table 1). Table 1 Demographic and clinical characteristics of episodes of allopurinol use ending or not ending in incident renal failure Figure 1 Flow chart of study cohort of incident allopurinol users from 2006 to 2012 with baseline of 183 days. DC, District of Columbia; HMO, health maintenance organization; NE, number of episodes eligible. We had 10.9 million person-days of follow-up for allopurinol exposure. Rates of incident renal failure by allopurinol doses 1–199, 200–299 and ≥300 mg/day were 66, 49 and 43 per 100 000 person-days (table 2). Crude incidence rates of renal failure by allopurinol use duration are shown in table 2. Table 2 Incidence rate of renal failure in allopurinol-exposed patients by dose and duration ### Allopurinol dose, allopurinol use duration and incident renal failure Univariate Cox proportional hazard regression analysis indicated that allopurinol daily dose, older age, female gender, black race, Charlson-Romano comorbidity index, beta-blocker use and diuretic use were associated with a higher risk of incident renal failure (table 3). Compared with allopurinol dose of 1–199 mg/day, higher allopurinol doses of 200–299 mg/day (HR 0.77; 95% CI 0.72 to 0.83) or ≥300 mg/day (HR 0.65; 95% CI 0.62 to 0.69) were associated with significantly lower hazard of incident renal failure. Table 3 Association of allopurinol dose and use duration with incident renal failure After multivariable adjustment for age, race, gender and Charlson–Romano comorbidity score (model 1, table 3), compared with allopurinol dose of 1–199 mg/day, both allopurinol doses of 200–299 mg/day (HR 0.81; 95% CI 0.75 to 0.87) and ≥300 mg/day (HR 0.71; 95% CI 0.67 to 0.76) were associated with significantly lower hazards of incident renal failure. Subsequent regression model that additionally adjusted for allopurinol use duration and the use of statins, beta-blockers, diuretics and ACE inhibitors confirmed the robustness of these associations with no change in HR or statistical significance (model 2, table 3). Sensitivity analysis that extended the baseline period to 365 days confirmed the association of higher allopurinol daily doses with incident renal failure as noted in the main analyses with minimal differences, 200–299 mg/day (HR 0.84; 95% CI 0.76 to 0.92) and ≥300 mg/day (HR 0.76, 95% CI 0.71 to 0.82) (table 4). In this analysis, we also found that compared with 1–180 days (<0.5 year) year of use, longer allopurinol use durations (>1 year) were now associated with significantly lower hazards of incident renal failure: 0.5–1 year, 1.00 (95% CI 0.88 to 1.55); >1–2 years, 0.85 (95% CI 0.73 to 0.99); and >2 years, 0.81 (95% CI 0.67 to 0.98), unlike the main analysis (table 4). Sensitivity analyses that adjusted for a diagnosis of gout showed that allopurinol dose, but not use duration, was associated with incident renal failure, with essentially no attenuation of HRs, similar to the main analysis (see online supplementary appendix 1). Sensitivity analysis limiting the definition of renal failure to specific codes for chronic renal failure (582.xx, 585.xx only) showed no attenuation of HR for allopurinol dose or duration of use (see online supplementary appendix 2). Sensitivity analysis using the exact use of the allopurinol rather than the average daily use of allopurinol yielded very similar results (data not shown). Table 4 Sensitivity analysis: association of allopurinol dose and use duration with incident renal failure (baseline period, 365 days instead of 183 days) ### Supplementary appendix 1 ### Supplementary appendix 2 ### Allopurinol and the risk of incident acute renal failure or ESRD Multivariable-adjusted analyses showed the allopurinol doses, 200–299 and≥300 mg/day, were associated with 0.91 (95% CI 0.85 to 0.99) and 0.89 (95% CI 0.83 to 0.94) hazards of acute renal failure compared with <200 mg/day allopurinol dose (see online supplementary appendix 2). Higher allopurinol dose was protective against the risk of ESRD; respective HRs were 0.77 (95% CI 0.60 to 0.99) and 0.57 (95% CI 0.46 to 0.71). The duration of allopurinol use was not significantly associated with reduced risk of acute renal failure or ESRD (see online supplementary appendix 2). ## Discussion In this retrospective cohort study using the 5% Medicare data, we examined whether allopurinol dose and duration were independently associated with a lower hazard ratio for incident renal failure in the elderly who were allopurinol users. Higher allopurinol dose was associated with a lower hazard for incident renal failure, with a dose–response relationship. Allopurinol use duration was associated with a protective effect in sensitivity analysis, but not in the main analysis. In exploratory analyses, allopurinol dose was associated with a lower hazard ratio for acute renal failure and haemodialysis. These observations are novel and merit further discussion. In this large sample of US elderly 65 years or older, using an incident user design, we found that higher average daily allopurinol use was independently associated with a lower risk of incident renal failure. Several randomised trials3–7 and observational studies9–11 have shown that allopurinol use is associated with a lower risk of CKD or better renal function. A meta-analysis of randomised trials of ULT concluded that ULT use was associated with a decrease in serum creatinine and an increase in estimated glomerular filtration rate (eGFR).22 The novel contributions of our study are the findings related to allopurinol dose and duration and the risk of incident renal failure. The associations were independent of several other factors, including race, age, gender, comorbidity and the use of common medications for cardiovascular disease (statins, beta-blockers, diuretics, ACE inhibitor). We found that higher allopurinol doses were associated with lower risk of incident renal failure in allopurinol users. We noted a dose–response relationship—allopurinol doses of 200–299 and ≥300 mg/day were associated with HRs of 0.81 and 0.71 in the main multivariable-adjusted analyses and 0.83 and 0.76 in the sensitivity analyses. Our study finding fulfils two additional Bradford–Hill's criteria for causation versus association23 for this relationship, that is, the biological gradient (allopurinol dose effect) and temporality (incident user design with allopurinol exposure preceding the outcome, renal failure). Other criteria, including consistency in observations, plausibility, coherence between clinical and animal models, are supported by the previous evidence.3–11 A biological rationale for this benefit can be postulated based on several potential mechanisms: (1) anti-ischaemic effect of allopurinol due to free radical scavenging12 ,24 ,25 and (2) observations that hyperuricemia is associated with proinflammatory state26 and renal endothelial dysfunction27 and therefore urate-lowering with allopurinol may lead to reduction of proinflammatory state. Our data did not have serum urate data, a study limitation. Therefore, we could not assess as to what proportion of the variability in renal failure prevention was related to urate-lowering versus an independent non-urate-lowering action of allopurinol. There are several important clinical implications of these findings. Our findings strengthen the emerging literature regarding the association of allopurinol use with better renal outcomes in patients with hyperuricemia. Several randomised trials showed that allopurinol use led to either improvement in renal function or slowing of progression of renal disease or proteinuria.3–8 Therefore, we did not focus on comparing allopurinol use versus non-use in our study. Instead, we focused on studying the association of allopurinol dose and duration with incident renal failure. Allopurinol, a cheap and affordable ULT, costs only a few cents a pill in the USA and is available for a$4 co-pay at several pharmacies.28 Allopurinol is the most commonly used treatment for hyperuricemia in settings of gout, the most common inflammatory arthritis in adults;29–31 allopurinol leads to lowering of serum uric acid and its use is thus associated with reduction in gouty tophi and gout flares.32–34 Given the suboptimal rates of allopurinol initiation and a low allopurinol adherence in gout1 ,35 ,36 and renal-protective effect of allopurinol doses 200 mg/day or higher shown in our study, now there is an additional rationale for implementing initiatives to optimise allopurinol use in patients with gout. Although our study included all allopurinol users, not just patients with gout, findings may be generalisable to gout since the most common indication for allopurinol is gout. Our study findings raise the question whether the risk–benefit ratio of treating asymptomatic hyperuricemia (without gout) with allopurinol changes with this evidence. While allopurinol is generally well tolerated, rare hypersensitivity reactions occur.37 ,38 We believe that if these findings are replicated in future studies, a reassessment of risk–benefit ratio for using allopurinol in asymptomatic hyperuricemia may be needed. Separate studies in people with asymptomatic hyperuricemia are needed.

Whether the duration of allopurinol use is associated with the risk of incident renal failure is not clear from our analyses. To our knowledge, there are no published data to compare our findings with. In our main analysis with 183 days as the baseline look-back period, there was no significant association, but in the sensitivity analyses with 365 days as the baseline look-back period, both allopurinol use for >1 to 2 years and >2 years were associated with 15% and 19% reduction in hazards of incident renal failure, respectively. This is an important question to explore in future studies.

An impressive finding was the association of higher allopurinol dose with a lower risk of acute renal failure and haemodialysis. Hazards of haemodialysis were 40% lower with ≥300 mg/day allopurinol compared with <200 mg/day. This is an impressive finding, consistent with the noted hazard of 0.71 for renal failure, especially since we excluded patients with renal failure in the baseline period for this analysis. Hazards were reduced for acute renal failure as well, though not with the same magnitude as seen for chronic renal failure or ESRD.

Recent data published for febuxostat must be considered while interpreting these study findings. In a randomised trial, Saag et al39 study evaluated the impact of febuxostat, compared with placebo, on renal function in 96 hyperuricemic gout subjects with moderate or severe renal impairment (eGFR ≥15 to <50 mL/min/1.73 m2) and found no improvement in renal function with febuxostat. Our study differs from the Saag et al study in important ways: (1) patient population, that is, people without previous renal failure versus people with moderate or severe renal failure; (2) intervention: allopurinol versus febuxostat; and (3) study design: observational versus randomised, placebo-controlled trial. Thus, although it is very tempting to extrapolate conclusions from one drug to another drug within a drug category, we urge caution. It is possible that renal effects of allopurinol may differ from those of febuxostat since previous randomised controlled trials (RCTs) of allopurinol with similar sample sizes showed improvement of renal function and/or preservation.3–8 The medication effect may vary by the severity of renal failure, that is, the improvement in renal function may only be evident in those with mild renal disease. A \$24.3 million randomised trial funded by the National Institute of Diabetes, Digestive and Kidney Diseases18 will compare allopurinol with placebo in diabetics and definitively answer the important question that our study cannot answer: ‘Can allopurinol independently protect kidney function and decrease chronic renal disease progression (compared with placebo)?’

Our study has several strengths. The sample size was large, the number of outcome events was reasonable (>5000) and the population was representative. Sensitivity analyses increased our confidence in these findings. We used 5% Medicare claims data that allowed adjustment for important confounders/covariates.

The study findings must be interpreted considering limitations. Findings should not be generalised to the entire US population, only to the US elderly, 65 years and older. Residual confounding is possible, due to the study design, despite our efforts to control for key factors. We used claims database definition of renal failure, which might lead to misclassification. However, this definition for renal failure has been shown to be valid previously40 and used commonly previously,41 ,42 though some studies have shown lower accuracy.43 ,44 Misclassification of incident renal failure can also occur due to variation in the length of the baseline period. To address this potential issue, we performed analyses with baseline periods of 183 days (main) and 365 days (sensitivity), which confirmed our finding. Reverse causation is possible, that is, patients with normal renal function might be more likely to receive allopurinol, compared with patients with early/mild renal failure, which could potentially explain the renal-protective effect observed with allopurinol. However, this is unlikely since we excluded patients with diagnosis of renal failure in the baseline period and used an incident user design. Despite a large sample size and rigorous design, given the novelty of our findings, confirmation in other populations/studies is prudent. A placebo-controlled RCT, similar to the one currently under way,18 can test the hypothesis generated from our study. We considered including non-steroidal anti-inflammatory drugs (NSAIDs) in our analyses, but decided not to do so due to the possibility of serious misclassification bias since 70% of all NSAID users in the USA use over-the-counter NSAIDs (exclusively or with prescription NSAIDs), which are not captured in any administrative database, such as Medicare.

In conclusion, this study used 5% Medicare claims data and found that allopurinol doses of 200 mg daily or higher were associated with a lower risk of incident renal failure in the elderly. The ongoing National Institutes of Health trial will answer this question for patients with type I diabetes18 in the next several years and confirm whether allopurinol use is associated with this benefit or not. Benefits were also noted for higher allopurinol dose that reduced the risk of acute renal failure and ESRD. Future studies need to replicate this finding in other population(s). Studies should also explore whether combination of allopurinol with other renal-protective medications is additive or synergistic, especially in patients with higher risk of kidney disease, for example, hypertension, diabetes, and so on.

## Footnotes

• Handling editor Tore K Kvien

• Contributors JAS: study design, protocol, analysis plan, writing the first draft of the manuscript, critical revision of the paper and the decision to submit. SY: review of analysis plan, data management, data analysis, critical revision of the paper and the decision to submit.

• Funding This material is the result of work supported by research funds from UAB Division of Rheumatology and the resources and use of facilities at the Birmingham VA Medical Center.

• Competing interests JAS has received research grants from Takeda and Savient and consultant fees from Savient, Takeda, Regeneron, Iroko, Merz, Bioiberica, Crealta, and Allergan pharmaceuticals. JAS serves as the principal investigator for an investigator-initiated study funded by Horizon pharmaceuticals through a grant to DINORA, a 501c3 entity. JAS is a member of the executive of OMERACT, an organisation that receives arms-length funding from 36 companies; a member of the American College of Rheumatology's Annual Meeting Planning Committee; and a member of the Veterans Affairs Rheumatology Field Advisory Committee.

• 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.

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

• Data sharing statement We will share this database with other colleagues after appropriate permissions from privacy and ethics committees of our institution have been obtained. There are no unpublished data from this study.

## 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.