Mortality in Patients With Gout Treated With Allopurinol: A Systematic Review and Meta-Analysis
Abstract
Objective
Urate-lowering therapy (predominantly allopurinol) is highly effective as a treatment for gout, but its wider long-term effects remain unclear. This systematic review and meta-analysis aimed to ascertain the association between mortality and the use of allopurinol in patients with gout.
Method
Medline, Embase, CINAHL, and the Cochrane Library were searched from inception to August 2018. Articles eligible for inclusion used a cohort design and examined cardiovascular or all-cause mortality in patients diagnosed with gout and prescribed allopurinol. Information on study characteristics, design, sample size, and mortality risk estimates were extracted. Article quality was assessed using the Newcastle-Ottawa Scale. Included articles were described in a narrative synthesis and, where possible, risk estimate data were pooled.
Results
Four articles reported a hazard ratio (HR) risk estimate for all-cause mortality in patients with gout using allopurinol, and 2 of these also reported cardiovascular mortality. Two articles found allopurinol to be protective in patients with gout, 1 found no statistically significant association, and 1 found no statistically significant effect of escalation of allopurinol dosage on all-cause or cardiovascular-related mortality. Data pooling was possible for all-cause mortality and found no association between allopurinol use in patients with gout and all-cause mortality compared to patients with gout not using allopurinol (adjusted HR 0.80 [95% confidence interval 0.60–1.05]).
Conclusion
There was no significant association between all-cause mortality and allopurinol use in people with gout. However, the number of included studies was small, suggesting that further studies are needed.
INTRODUCTION
SIGNIFICANCE & INNOVATIONS
- We found no significant association between all-cause mortality and allopurinol use in people with gout.
- Further studies taking into account allopurinol dose and achievement of target serum urate levels are required.
Treating gout should be straightforward due to the availability of safe, effective, long-term treatment to lower urate levels (urate-lowering therapy [ULT]), allowing dissolution of existing urate crystals and prevention of new crystal formation, leading to the cessation of gout flares (6, 7). International guidelines recommend that ULT be offered to all patients with gout and initiated upon confirmation of diagnosis, once any current flare has abated (8, 9). Allopurinol is the first-line ULT and should be initiated at a low dose (≤100 mg daily), followed by uptitration in 100-mg increments until urate levels are suppressed below 360 μmol/liter (6 mg/dl). Despite clear guidelines and benefits, only 30% of patients are prescribed allopurinol, and of those, only 40% have treatment escalated to achieve the target serum urate level of <360 μmol/liter, suggesting that many patients with gout could receive better ULT (10).
In addition to its success in treating gout, other benefits of allopurinol have been suggested in patients with kidney and cardiovascular diseases. The drug has been shown to be associated with a decreased likelihood of renal events (initiation of dialysis, doubling serum creatinine, ≥50% decrease in estimated glomerular filtration rate) in two-thirds of patients with chronic kidney disease (11). Improvements in cardiovascular function include increased peripheral blood-flow due to improved endothelial function in patients with chronic heart failure (12). However, despite these improvements in morbidity, whether the benefits of allopurinol extend to reducing premature mortality in patients with gout remains unclear. In patients with hyperuricemia (the precursor to gout), the use of allopurinol has been estimated to be associated with a 25% lower risk of mortality during follow-up compared with untreated patients (13, 14).
Despite guidelines recommending earlier prescription of ULT (8, 9) and the reported benefits on comorbidities, the use of allopurinol to treat gout remains suboptimal. Though the reasons behind this hesitancy are multifaceted, 1 contributing factor relates to the apprehension of patients and clinicians to initiate life-long treatment without a clear understanding of the long-term effects (1). Because the overall balance of potential benefit and risk in the role of allopurinol on mortality in patients with gout remains unclear, this systematic review and meta-analysis examined the association between the use of allopurinol in patients with gout and cardiovascular or all-cause mortality.
MATERIALS AND METHODS
A systematic review of research literature was conducted. Medical literature databases were searched to identify articles that included patients with gout treated using allopurinol and that reported the risk of cardiovascular or all-cause mortality in their sample. Meta-analysis was used to determine pooled risk estimates of mortality. The protocol for this systematic review and meta-analysis was registered on PROSPERO (ID CRD42017056011) and the systematic review was undertaken following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.
Data sources, searches, and study selection
Four electronic bibliometric databases were searched for articles (Embase, Medline, CINAHL, and Cochrane Studies) (see Supplementary Table 1, available on the Arthritis Care & Research website at http://onlinelibrary.wiley.com/doi/10.1002/acr.24205/abstract). These were required to fulfil the following eligibility criteria: 1) the study sample was formed from adults with a diagnosis of gout, 2) allopurinol was used to treat gout, 3) risk estimates of all-cause or cardiovascular mortality were reported, and 4) the study used a cohort design. Cohort studies were targeted specifically because their populations are more likely than randomized controlled trials (RCTs) to be representative of the general population and of normal courses of treatment, therefore increasing the likelihood that this systematic review and meta-analysis produces a generalizable result. Case–control and cross-sectional studies were excluded because they would not describe outcomes over time. No restrictions were imposed on the time periods for publication, with medical literature databases searched from inception to August 2018. There were no language restrictions, but if translational facilities were not available for an article, it was excluded.
Data extraction
Data were extracted by 2 authors (CAH and JAP) with the main data including demographic information (age, sex, country of origin, etc.), study sample size, numbers of patients with gout, study setting (e.g., primary care), exposures (e.g., allopurinol), mortality outcome (e.g., all-cause, cardiovascular), definition of gout, and method of adjusted risk estimates regarding the association between gout treated with allopurinol and cardiovascular and all-cause mortality risk estimates.
Quality assessment
All articles finally included in the systematic review were quality appraised independently by 2 assessors (CAH and JAP). Any disagreement on initial scoring was discussed, and if the difference could not be agreed on, the decision was arbitrated by a third reviewer (ER). Methodologic quality was assessed using the Newcastle-Ottawa Scale for cohort studies (15).
Meta-analysis
Where a sufficient number of articles (≥3) were identified, a random-effects meta-analysis was used to pool reported mortality risk estimates along with their 95% confidence intervals (95% CIs). Heterogeneity was assessed by I2. The meta-analysis was undertaken in Stata software, version 14.
RESULTS
Literature search
From 362 articles identified by the initial literature search, 90 duplicates were removed. The titles of the remaining 272 articles were screened, after which 37 articles remained. After an abstract review of these, 32 articles were excluded. The full text of the remaining 5 articles was reviewed in full, and a final 4 articles were deemed to fulfil the inclusion criteria (Figure 1).
Sample characteristics
Four articles examined all-cause mortality (16-19) and 2 of these (16, 19) also examined cardiovascular mortality in the same population. Of the articles included in the review, 1 study population was from Taiwan, 1 was from the US, and the other 2 were from the UK. All 4 articles estimated the risk of mortality using HRs (16-19) (Table 1).
| Author, year (ref.) | Population | Study period | Sample size | Incident gout, no. | Male, % | Mean ± SD age, years | Adjusted HR (95% CI) |
|---|---|---|---|---|---|---|---|
| Chen et al, 2015 (16) | MJ Health Screening Center database, Taiwan | 1997–2002 | 1,457 | 286 | 89 | 52.7 ± 15.4 | 0.39 (0.22–0.70) |
| Coburn et al, 2018 (19) | US Department of Veterans Affairs Health Administration | 1999–2010 | 111,694 | 6,428† | 99.7 | 64.4 ± 10.5 | 1.05 (0.96–1.15)‡ |
| Dubreuil et al, 2015 (17) | The Health Improvement Network, UK | 2000–2010 | 9,590 | 483 | 69 | 67§ | 0.81 (0.70–0.92) |
| Kuo et al, 2015 (18) | Clinical Practice Research Datalink, UK | 1995–2013 | 19,549 | 3519 | 72 | 64 (52–73)¶ | 0.99 (0.87–1.12)# 1.01 (0.92–1.09)** |
- * 95% CI = 95% confidence interval; HR = hazard ratio; ref. = reference.
- † Gout patients receiving dose escalation.
- ‡ This HR represents the risk of all-cause mortality in patients with gout treated with escalating doses of allopurinol compared to patients with gout on a constant dosage of allopurinol.
- § No SD was reported.
- ¶ Median (interquartile range).
- # 1-year landmark analysis.
- ** 3-year landmark analysis.
The Taiwanese study by Chen et al (16) sourced its cohort from the medical insurance data from MJ Health Screening Center, which contained 49,460 individuals age >17 years who had consultations since 1996. Gout was defined using International Classification of Diseases, Ninth Revision (ICD-9) codes for cases identified between 1997 and 2002. Kuo et al (18) used a UK primary care data source, the Clinical Practice Research Datalink, and defined incident gout by Read codes between 1995 and 1999. Dubreuil et al (17) used a different UK primary care data source (The Health Improvement Network), defining gout by Read codes between January 2000 to May 2010. The study by Coburn et al (19) sourced its cohort from the US Department of Veterans Affairs Health Administration between 2001 and 2008 and defined gout by its ICD-9 definition. Unlike the previous 3 studies, Coburn et al focused specifically on the effect on risk of all-cause and cardiovascular mortality of increasing allopurinol dosage in patients.
All articles were cohort studies and used matching based on propensity scores. Chen et al and Dubreuil et al followed up their patients from exposure onward (date of diagnosis for Chen et al and initiation of allopurinol for Dubreuil et al). Kuo et al and Coburn et al both used landmark analysis to avoid immortal time bias. Kuo et al only included patients who were alive by the landmark time points (1 year and 3 years); this method excludes the initial time period immediately after gout diagnosis, reducing the possibility of conferring an unfair survival advantage on the allopurinol-treated group. Coburn et al used 2 models; in model 1, they followed up patients from exposure, and in model 2, they followed up patients after a 2-year landmark.
Quality assessment results
All 4 articles included representative patients with gout and assessed exposures and outcomes using secure methods (medical records), and employed appropriate methods to compare subjects, with and without gout, to avoid confounding by indication affecting the veracity of results. All 4 studies also employed propensity score matching. In addition, 3 of the 4 articles used methods to attempt to negate immortal time bias; Chen et al used time-index matching between patients and controls and Coburn et al used an analytical method that involved only following up patients who were alive 2 years after allopurinol initiation. Kuo et al employed a landmark analysis method that only followed up patients who were alive after 1 year post-allopurinol initiation and then 3 years after initiation. Loss to follow-up was minimal and was accounted for in analyses.
Risk of all-cause mortality
Chen et al and Dubreuil et al both found allopurinol to have a protective effect on all-cause mortality in patients with gout. Chen et al reported an adjusted HR of 0.39 (95% CI 0.22–0.70) (allopurinol was slightly more protective against all-cause mortality than the use of any ULT medication, with HR 0.47 [95% CI 0.29–0.79]), and Dubreuil et al reported an adjusted HR of 0.81 (95% CI 0.70–0.92). However, Kuo et al found no association between the use of allopurinol in patients with gout and all-cause mortality, with HR 0.99 (95% CI 0.87–1.12) for 1-year landmark analysis and HR 1.01 (95% CI 0.92–1.09) for the 3-year landmark analysis, the latter of which was included in the pooled analysis. Finally, Coburn et al reported an HR for all-cause mortality in patients with gout for whom allopurinol dosage was increased, compared with patients with gout using a constant dose. The researchers reported a significant increase in all-cause mortality for model 1 (propensity score matching HR 1.08 [95% CI 1.01–1.17]) and a nonsignificant HR for model 2 (inclusion of 2-year landmark analysis HR 1.05 [95% CI 0.96-1.15]). However, because these HRs were based on stratification by dose, their inclusion in the pooled analysis was not possible. The pooled adjusted HR for all-cause mortality calculated from the 3 applicable cohorts was 0.80 (95% CI 0.60–1.05), and heterogeneity was statistically significant (87.6%; P < 0.001) (Figure 2).
Risk of cardiovascular mortality
Chen et al reported a protective effect of allopurinol on cardiovascular mortality, finding an HR in patients with gout treated with allopurinol of 0.37 (95% CI 0.01–0.48) compared to non-allopurinol users. Coburn et al initially reported an association between increased cardiovascular-related mortality in those with escalated allopurinol dose compared to those with a stable dose for model 1 (HR 1.08 [95% CI 0.97–1.21]), but no association remained in model 2 (HR 1.05 [95% CI 0.92–1.20]). Due to the sparsity of data related to cardiovascular mortality, we were unable to conduct pooled analysis for this outcome (see Supplementary Table 2, available on the Arthritis Care & Research website at http://onlinelibrary.wiley.com/doi/10.1002/acr.24205/abstract).
DISCUSSION
Our systematic review and meta-analysis of 3 studies showed no significant association between the use of allopurinol and all-cause mortality in patients with gout. The results of studies into cardiovascular mortality were contradictory and limited (preventing data pooling).
Our findings are not consistent with reports of statistically significantly decreased mortality associated with allopurinol use in hyperuricemic patients and a protective effect against cardiovascular and chronic kidney disease in patients with gout (11-14, 20). Though not directly comparable, such findings supported our initial hypothesis that a reduction in mortality for patients with gout using allopurinol would be observed. There are, however, some important differences between the studies in our meta-analysis and those that have previously shown protective effects of allopurinol. Notably, the studies included in our review used observational data from clinical practice, where allopurinol dosage is commonly insufficient to lower urate significantly (only 40% having treatment escalated to achieve the target serum urate level [10]).
Studies that have shown a protective effect of allopurinol dosage on the risk of cardiovascular events often involves dosage of >600 mg/day, compared to the more common 100–300 mg/day found throughout normal primary care gout management. Also in the case of RCTs (11), dosage of ULT was managed, observed, and escalated in a more systematic way than in the cohort studies included in our review. Possibly, therefore, the nonsignificant protective effect reported by our meta-analysis is related to the fact that lower dosages infrequently facilitate the achievement of target serum urate levels in patients, and there are frequently lower levels of compliance and treatment observation in the general population compared to RCT populations.
Our findings support the existing body of evidence on the short-term safety of allopurinol (21, 22), because our included articles used large, nationally representative data sets and provided a combined sample of >10,000 patients with gout in which to examine all-cause mortality. In particular in the UK, where the majority of patients with gout are managed in primary care, Kuo et al and Dubreuil et al formed the principal weighting within the meta-analysis, with data from 2 different primary care data sets. A key methodologic difference between the studies is the use of landmark analysis by Kuo et al to address the potential for immortal time bias, and this methodologic difference may well be the cause of the disagreement between the 2 studies regarding risk. Though Chen et al demonstrated a protective effect of allopurinol use, their sample was small, and they did not include landmark analysis. However, they attempted to avoid immortal time bias by matching for the index date of ULT prescription using a propensity score (16). Possibly the difference in reported effects between the study of Chen et al and the other 3 studies in this systematic review is due to the difference in populations.
The pooled HR of 0.80 with its 95% CI of 0.60 to 1.05 could suggest a possible small protective effect of allopurinol; however, statistical significance was not reached, and the 2 largest of the 3 included studies contributed the greatest weighting in the meta-analysis and had HRs closest to 1. Further large studies into the effect of allopurinol on both all-cause and cause-specific mortality in patients with gout are needed. Our findings are complicated by the results of Coburn et al, which showed an increase in the risk of all-cause mortality in patients with gout whose dosage was escalated, although these associations became nonsignificant upon closer matching of patients with dose escalation to patients without dose escalation.
Given the protective effects of allopurinol found in RCTs and several cohort studies, further research in this area to produce a more cohesive and conclusive view of the association between treatment of gout with allopurinol and mortality is essential. Consideration should be given to the effect of allopurinol on mortality in specific subgroups, such as men and women and those with different comorbidities or tophaceous gout. Also of high importance in this research would be the effects of treatment adherence, because this adherence is so low in patients with gout that it may be undermining not just the primary aims for allopurinol, but also possible secondary positive outcomes, such as a lower risk of early all-cause mortality.
We are unable to draw any conclusions on any potential role of allopurinol use in cardiovascular mortality in patients with gout. Only 2 articles were identified and results were varied, so further research is required. However, from 1 of these articles (19), the consideration of allopurinol dose arises as an important issue in the matter of the role of allopurinol on mortality in patients with gout. Coburn et al found no significant difference between either all-cause or cardiovascular mortality in those patients who had their dose of allopurinol escalated over 2 years and those whose dose remained stable. To address the fact that the majority of patients with gout using allopurinol never reach target serum urate levels, the researchers performed a sensitivity analysis using only those patients who did reach the guideline target levels. Within this subsample, they found that for all-cause mortality there remained a similar HR (not reported); however, for cardiovascular mortality, though not significant, they now found a reduction in risk of 7% (HR 0.93 [95% CI 0.76–1.14]). The role of allopurinol and its dose on the risk of premature mortality (particularly cardiovascular) in patients with gout using allopurinol requires much further study.
This is the first systematic review to examine the association between patients with gout treated with allopurinol and cardiovascular or all-cause mortality. Our search criteria were extensive (not limited by language) and included cohort studies from large, nationally representative samples using data over similar time periods to provide a more generalizable picture of the role of allopurinol on mortality in patients with gout. Risk estimates for all-cause mortality from different studies were pooled. The principal limitations of our review are the small number of articles available and statistical heterogeneity in the pooled analysis. However, despite the low number of studies, those included in this systematic review are of high methodologic quality, having factored in methods for avoiding immortal time bias and confounding by indication.
Our systematic review and meta-analysis did not find a significant association between allopurinol use and cardiovascular or all-cause mortality. However, the small number of studies suitable for inclusion and the evidence from the wider literature that allopurinol may have cardiovascular and renal benefits suggest that further studies into the effect of allopurinol use on mortality in people with gout are required, particularly regarding the role of allopurinol dose and the importance of reaching target serum urate levels.
Acknowledgment
The authors thank the School of Primary, Community and Social Care, Keele University for its support.
AUTHOR CONTRIBUTIONS
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Mr. Hay had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design
Hay, Prior, Roddy.
Acquisition of data
Hay, Prior, Roddy.
Analysis and interpretation of data
Hay, Prior, Belcher, Mallen, Roddy.
