Objective Atrial fibrillation (AF) is the most common arrhythmia associated with cardiovascular disease and mortality. Recent studies suggest an association between inflammation, hyperuricaemia and AF, but little is known whether gout is associated with AF risk.
Methods Using data from a US commercial insurance plan (2004–2013), we conducted a cohort study to evaluate the risk of incident AF in patients with gout versus osteoarthritis. Patients with gout or osteoarthritis were identified with ≥2 diagnoses and ≥1 dispensing for gout or osteoarthritis medications. Incident AF was defined as a new AF diagnosis and a new dispensing for anticoagulants or antiarrhythmics. The risk of incident AF in gout was also compared with the non-gout group.
Results We identified 70 015 patients with gout and 210 045 with osteoarthritis, matched on age, sex and index date. The mean age was 57 years, and 81% were men. Over the mean 2-year follow-up, the incidence rate of AF per 1000 person-years was 7.19 in gout and 5.87 in osteoarthritis. The age, sex and index date-matched HR of AF was 1.23 (95% CI 1.14 to 1.32) in gout versus osteoarthritis. In a multivariable Cox regression, adjusting for age, sex, comorbidities, medications and healthcare usage, the HR of AF in gout was 1.13 (95% CI 1.04 to 1.23). When compared with non-gout, the multivariable HR of AF in gout was also increased (HR 1.21, 95% CI 1.11 to 1.33).
Conclusions In this large population-based cohort study, gout was associated with a modestly increased risk of incident AF compared with osteoarthritis and non-gout after adjusting for other risk factors.
- Cardiovascular Disease
Statistics from Altmetric.com
Atrial fibrillation (AF) is the most common arrhythmia affecting over five million in the USA and associated with substantial morbidity, including stroke and congestive heart failure and all-cause and cardiovascular mortality.1 ,2 There are increasing data supporting the role of inflammation in the development and maintenance of AF.3 ,4 Chronic medical conditions such as diabetes, obesity and coronary artery disease are associated with systemic inflammation and increased levels of serum inflammatory biomarkers such as reactive oxygen species, transforming growth factor-β, C reactive protein (CRP), pro-inflammatory cytokines, including tumour necrosis factor (TNF)-α, interleukin (IL)-1, IL-2 and IL-6.4 ,5 Increased levels of inflammatory biomarkers, abnormal infiltration of mononuclear lymphocytes in atrial tissue and necrosis of the adjacent myocytes are noted in patients with AF, leading to the hypothesis linking inflammation with AF.3 ,4 ,6–8 Furthermore, several clinical trials show beneficial effects of anti-inflammatory drugs such as statins, colchicine and steroids in reducing the risk of postoperative AF.4 ,9–12
Gout is a common inflammatory arthritis triggered by the deposition of monosodium urate crystals in joints. It is well established that urate crystals induce an inflammatory response leading to the release of TNF-α, IL-1, IL-6 and IL-8 by activating the nucleotide oligomerization domain (NOD)-like receptor protein 3 inflammasome.13 ,14 The association between gout, hyperuricaemia and comorbid conditions such as hypertension, chronic kidney disease and cardiovascular disease is also well described.15–18 Although the causal link cannot be confirmed, a number of large prospective cohort studies show significantly increased risks of developing myocardial infarction and stroke in patients with hyperuricaemia or gout, independent of traditional cardiovascular risk factors.19–22 Recently, a few cross-sectional or cohort studies report an increased risk of sinus tachycardia, AF and left atrial thrombus in patients with hyperuricaemia.23–25
To date, it remains unknown whether gout is associated with the risk of AF. The main objective of this study was to assess the rate of incident AF in patients with gout compared with those without gout in a large population-based cohort study.
We conducted a cohort study using claims data from United HealthCare, a large commercial US health plan, for the period from 1 January 2004 to 31 December 2013. This data source has been described in detail elsewhere.26 In brief, the study database contains longitudinal information on medical and pharmacy claims. In addition, on a subset of beneficiary claims, the data were linked to laboratory test results provided by two large national laboratory providers. Patient-informed consent was not required as the dataset was de-identified to protect subject confidentiality. The study protocol was approved by the Institutional Review Board of the Brigham and Women's Hospital.
Adult patients aged ≥40 years who had ≥2 visits coded with the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9 CM) code, 274.0X, 274.8X and 274.9 for gout were eligible for the gout group. We defined the index date for the ‘gout group’ as the date of the first dispensing of a gout-related drug after ≥1 year of continuous health plan enrolment and ≥2 visits for gout. Gout-related drugs were allopurinol, febuxostat, colchicine, probenecid, pegloticase, selective and non-selective non-steroidal anti-inflammatory drugs (NSAIDs) and systemic and intra-articular steroids. To ensure that we capture incident cases of AF, we excluded patients who had a diagnosis of arrhythmia or cardiac surgery, or who used antiarrhythmics or anticoagulants in the 365-day period prior to the index date. Nursing home patients were also excluded as we did not have drug-dispensing data from a nursing home stay.
We selected two different non-gout comparison groups aged ≥40 years. The primary comparison group was the ‘osteoarthritis group’ based on ≥2 visits coded with the ICD-9 code 715.XX for osteoarthritis. The osteoarthritis group was used as the primary comparison group to minimise confounding by obesity, comorbidities and health care usage intensity as these characteristics are similar to those of the gout group. The index date for the osteoarthritis group was the date of the first dispensing of opioids, selective or non-selective NSAIDs after ≥2 physician visits coded for osteoarthritis. The secondary comparison group was the ‘non-gout group’ and included patients who had no diagnosis of gout at baseline, and had at least two physician visits after ≥12 months of continuous health plan enrolment. The index date for the non-gout patients was the date of the first dispensing of any prescription drugs after ≥2 physician visits. We applied the aforementioned exclusion criteria in both comparison groups. The osteoarthritis and non-gout groups were not allowed to use allopurinol, febuxostat, colchicine, pegloticase or probenecid in the 365-day period prior to or on the index date. The osteoarthritis and non-gout group were then matched to the gout group on age, sex and index date (±30 days) with a 3:1 ratio.
Patients in both comparison groups were followed from the index date to the first of any of the following censoring events: development of AF, development of gout (for both osteoarthritis and non-gout groups), insurance disenrolment, admission to nursing home, end of the study period, or death.
The primary outcome was defined as a combination of an inpatient or outpatient diagnosis of AF (ICD-9 427.3x), and a new dispensing of anticoagulants or antiarrhythmics within 30 days after the AF diagnosis date. We also used two secondary more stringent definitions of AF outcomes. One outcome was defined with an inpatient discharge diagnosis code of AF. This algorithm has been previously validated and had both sensitivity and specificity greater than 90%.27 Another outcome required a combination of an inpatient discharge diagnosis of AF and a new dispensing of anticoagulants or antiarrhythmics prescription for an anticoagulant within 30 days after the inpatient discharge date.
We assessed a number of predefined variables potentially related to gout, healthcare use or development of AF using data from the 365 days before the index date. These variables were age, sex, comorbidities, use of medications, healthcare usage intensity indicators and physician order of outpatient laboratory tests listed in table 1. We also calculated a comorbidity score to quantify patients’ comorbidities based on ICD codes.28 Outpatient laboratory data such as acute phase reactants (ie, high erythrocyte sedimentation rate (ESR) or CRP vs normal levels) and uric acid levels were also included in a subgroup of all three study groups.
Baseline characteristics of the gout and osteoarthritis groups were compared. We calculated the incidence rates (IR) and rate ratios of AF with 95% CIs in age, sex and index-date matched groups.29 Kaplan–Meier curves were plotted for the cumulative incidence of AF in the gout and osteoarthritis groups. We tested for an interaction between gout and sex by including the product term in a multivariable Cox proportional hazards model, and did not find a statistically significant interaction (p=0.8). To estimate the HRs of AF-associated gout versus osteoarthritis, we conducted a number of separate partially adjusted Cox models, including age, sex (model 1), and age, sex, comorbidity score and number of unique prescriptions (model 2). The final, fully adjusted Cox model included nearly 40 covariates listed in table 1.
We conducted multivariable Cox regression analyses stratified by several important risk factors for AF, including obesity, diabetes and history of cardiovascular disease. In addition, to examine the risk of AF associated with gout versus osteoarthritis, adjusting for elevated ESR/CRP levels at baseline, we conducted subgroup analyses in patients with baseline ESR/CRP levels available. We also conducted subgroup analyses, adjusting for baseline serum uric acid levels among patients in whom we had a baseline uric acid level available. Lastly, among patients with gout in whom we had a baseline uric acid level measured, we assessed the risk of AF in relation to their serum uric acid levels.
All the main and subgroup analyses were repeated for the comparison between the gout and non-gout groups. The proportional hazards assumption was assessed by testing the significance of the interaction term between exposure (ie, gout or osteoarthritis) and time, and was not violated in any models.30 All analyses were done using SAS V.9.3 Statistical Software (SAS Institute, Cary, North Carolina, USA).
Figure 1 displays the study cohort selection process. There were initially 303 400 patients with at least one gout diagnosis and 1 828 333 with at least one osteoarthritis diagnosis after a 365-day enrolment period. After applying the inclusion and exclusion criteria, we selected three patients with osteoarthritis matched to each patient with gout on age, sex and the index date. Our final study cohort includes 70 015 patients with gout and 210 045 patients with osteoarthritis. The mean (SD) follow-up time was 2.1 (1.8) years for patients with gout and 2.0 (1.8) years for patients with osteoarthritis. The most common reason for censoring besides the development of outcome was disenrolment from the plan (63% in gout and 59% in osteoarthritis) followed by the end of database period (33% in gout and 32% in osteoarthritis), nursing home admission (3% in both groups) and death (<1% in both groups). Three per cent of patients with osteoarthritis were censored because they developed gout during the follow-up time.
For the comparison between gout and non-gout groups, there were 91 976 patients with gout and 275 928 with no gout, matched on age, sex and the index date (see online supplementary appendix table 2). The mean (SD) follow-up time was 2.0 (1.8) years for patients with gout and 2.0 (1.8) years for patients with no gout.
Baseline characteristics of the gout and osteoarthritis groups are presented in table 1. The mean (SD) age was 56.8 (9.0) years, and men comprised 81.4% in both groups. Patients with gout generally had a greater frequency of comorbidities such as hypertension, cardiovascular disease, diabetes, chronic kidney disease and hyperlipidaemia compared with the osteoarthritis group. The proportion of patients with obesity at baseline was 10.9% in the gout group and 9.7% in the osteoarthritis group. Among patients with gout, 55.2% used allopurinol or febuxostat and 23.6% colchicine at baseline. Of those, 10.5% used colchicine within 30 days from the index date. Use of antihypertensives, diuretics, lipid-lowering drugs and steroids were more common in gout than osteoarthritis, while use of bisphosphonates, opioids and selective NSAIDs were more common in osteoarthritis than gout. Healthcare usage intensity was similar between the groups. Among patients with baseline ESR/CRP levels available (n=15 174, 5.4% of the total cohort), 35.5% of gout and 23.4% of osteoarthritis patients had elevated ESR/CRP. Among patients with baseline uric acid levels available (n=20 622, 7.4% of the total cohort), the mean (SD) uric acid level (in mg/dL) was 7.4 (2.0) in gout and 5.9 (1.4) in osteoarthritis. Baseline characteristics of the gout and non-gout groups are shown in online supplementary appendix table 1.
Risk of incident AF in gout
The IR of AF using the primary outcome definition was 7.19 per 1000 person-years in patients with gout and 5.87 per 1000 person-years in patients with osteoarthritis (table 2). The age, sex and index-date matched rate ratio was 1.22 (95% CI 1.13 to 1.31) in gout compared with osteoarthritis. Figure 2 illustrates Kaplan–Meier curves for the cumulative incidence of AF in the gout and osteoarthritis groups (log-rank p value <0.0001). Table 3 summarises the results from several multivariable Cox regression analyses comparing patients with gout and patients with osteoarthritis. After adjusting for age, sex, obesity, comorbidities, medications and healthcare usage intensity listed in table 1, the HR of AF associated with gout was 1.13 (95% CI 1.04 to 1.23). Multivariable Cox regression analyses stratified by obesity, diabetes or history of cardiovascular disease showed no significant heterogeneity with similar HRs between the subgroups (see table 3).
In the subgroup of patients with baseline ESR/CRP level measured, the HR of AF in gout versus osteoarthritis was 1.12 (95% CI 0.74 to 1.69) after adjustment for age, sex, comorbidity score and the number of prescription drugs. After further adjusting for elevated ESR/CRP at baseline, the HR of AF was 1.06 (95% CI 0.70 to 1.61) in gout versus osteoarthritis. Among patients with baseline serum uric acid levels available, the HR of AF was 1.33 (95% CI 0.89 to 2.00), adjusting for age, sex, comorbidity score and the number of prescription drugs and 1.19 (95% CI 0.78 to 1.83) further adjusted for serum uric acid levels at baseline. Among patients with gout who have baseline serum uric acid levels available (n=14 968), the HR of AF associated with a 1 mg/dL increase in the uric acid level was 1.05 (95% CI 0.96 to 1.16) after adjusting for age, sex, comorbidity score and the number of prescription drugs.
In the secondary comparison between patients with gout and no gout, the IR of AF using the primary outcome definition was 6.44 per 1000 person-years in patients with gout and 4.11 per 1000 person-years in patients with no gout (see online supplementary appendix table 2). The risk of incident AF associated with gout (HR 1.21, 95% CI 1.11 to 1.33) was consistently increased in patients with gout compared with those with no gout after fully adjusting for more than 35 covariates (see online supplementary appendix table 3). In the subgroup analyses, similar to the primary comparison, the HR of AF in gout versus non-gout was 1.05 (95% CI 0.64 to 1.73), adjusting for age, sex, comorbidity score, the number of prescription drugs and elevated ESR/CRP and 1.21 (95% CI 0.70 to 2.08), adjusting for age, sex, comorbidity score, the number of prescription drugs and serum uric acid levels.
In this large population-based cohort study, gout was associated with a 13% greater risk of incident AF compared with osteoarthritis and a 21% greater risk compared with non-gout. The risk of AF associated with gout did not differ between men and women. Although the causal relationship between gout and AF is yet to be determined, gout could potentially increase the risk of AF through two distinct, but possibly related, mechanisms—inflammation and hyperuricaemia. We first investigated the association between gout and AF independent of the level of systemic inflammation at baseline in a subgroup analysis, adjusting for elevated ESR or CRP at baseline. The risk of AF in gout did not appear to be elevated after further adjusting for elevated ESR/CRP. Second, as prior evidence showed, a potential relationship between serum uric acid levels and AF risk,23 ,24 we conducted a separate sensitivity analysis restricting to those with baseline uric acid levels. The risk of AF in gout was attenuated towards the null after further adjusting for baseline serum uric acid levels. Lastly, there seems to be a dose-response, although not statistically significant due to a smaller subgroup size, between serum uric acid levels and risk of AF relationship (HR 1.05 per 1 mg/dL increase in serum uric acid level, 95% CI 0.96 to 1.16). While these subgroup analyses were run in only a subgroup of patients with the laboratory data at baseline, these findings support the hypothesis that gout may increase the risk of AF due to complex inter-related mechanisms of systemic inflammation and hyperuricaemia.
This study has several strengths. First, the study cohort includes a large number of patients with gout, osteoarthritis and no gout in a population that is representative of the US commercially insured population. Second, we assessed a wide range of potential confounders at baseline, including comorbidities, medications and healthcare use characteristics. Third, we also performed several subgroup analyses to investigate potential mechanisms in which gout might increase the risk of AF. Fourth, to minimise surveillance bias and confounding by body mass index and comorbidities, we compared patients who had gout with those who had osteoarthritis, a chronic non-inflammatory or minimally inflammatory medical condition requiring frequent medical attention or pharmacological treatment. Lastly, we also showed an increased risk of AF with an HR of 1.21 in the comparison between patients with gout and no gout, similar to the results from the primary analysis.
There are several limitations in this study. First, there is a possibility of residual confounding due to unmeasured risk factors such as body mass index, abdominal circumference, baseline cardiac function and severity of comorbidities such as heart failure or chronic obstructive pulmonary disease.1 ,31 Second, there is a potential for exposure (ie, gout vs osteoarthritis) and outcome misclassification as we relied on a combination of diagnosis and procedure codes. However, the algorithm that we used to identify patients with gout has a positive predictive value of >99% in a separate chart-review study that we recently conducted using an electronic medical record database linked to claims data (unpublished). For the outcomes, we used three different algorithms to define AF, which used previously validated diagnosis codes27 with and without a dispensing of new anticoagulants or antiarrhythmics, and found similar results across those definitions. Third, because we required patients with gout and osteoarthritis to have at least two visits and a dispensing for gout-related or osteoarthritis-related medications, patients with mild or minimally symptomatic gout or osteoarthritis who did not need to take a drug were not included. Thus, our results may not be generalisable to such patients. Fourth, the statistical power of the subgroup analyses was limited due to the relatively small number of patients with baseline ESR/CRP or uric acid levels.
Our findings have important implications for future studies aimed at reducing the risk of AF. While the magnitude of the association between gout and AF is small, it may still have an important impact on the gout population as AF is the most common arrhythmia with substantial consequences such as stroke and heart failure. First, several studies have examined the role of colchicine in reducing AF development or recurrence after cardiac surgery over the past 5 years.10 ,32 In a placebo-controlled clinical trial, a 3-month course of colchicine 0.5 mg twice daily was effective (OR 0.38, 95% CI 0.18 to 0.80) in preventing recurrent AF in patients with paroxysmal AF who underwent pulmonary vein isolation procedure.32 The multicentre, double-blind, randomised trial, COlchicine for the Prevention of the Post-pericardiotomy Syndrome (COPPS), showed that patients who started colchicine 1.0 mg twice daily 3 days after cardiac surgery and continued a maintenance dose of 0.5 mg twice daily for 1 month had a 45% reduced risk of developing postoperative AF compared with placebo.10 Since colchicine is commonly used in patients with gout, future research may examine the effect of colchicine on primary or secondary prevention of AF among patients with gout. Second, a previous case–crossover clinical trial of allopurinol was effective in lowering blood pressure in adolescent patients with newly diagnosed hypertension.33 Future studies may explore the role of urate-lowering therapy in reducing AF risk as hyperuricaemia is causally related to gout and associated with cardiovascular disease.
In conclusion, gout was associated with a modestly increased risk of incident AF compared with osteoarthritis and non-gout after adjusting for a number of traditional risk factors. Our results support the hypothesis that gout increases the risk of AF via systemic inflammation combined with hyperuricaemia, although a dose–response relationship between the level of serum uric acid and the risk of AF needs to be further studied. Future research should confirm this finding in a different setting, and may examine the role of gout treatment, controlling systemic inflammation and/or lowering uric acid levels in the primary or secondary prevention of AF.
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.
- Data supplement 1 - Online tables
Handling editor Tore K Kvien
Contributors All authors conceived and designed the study, interpreted the data and critically revised the manuscript for important intellectual content. SCK drafted the paper. SCK has 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.
Funding SCK is supported by the NIH grant K23 AR059677. DHS is supported by NIH K24 AR055989.
Competing interests SCK receives salary support from unrelated grants to Brigham and Women's Hospital from Pfizer, AstraZeneca and Lilly. DHS receives salary support from unrelated grants to Brigham and Women's Hospital from Lilly, Pfizer, Amgen and Astra Zeneca.
Ethics approval Brigham and Women's Hospital IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
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.