Risk of Major Adverse Cardiovascular Event Following Incident Hospitalization for Acute Gout: A Western Australian Population-Level Linked Data Study

Abstract

Background

Cardiovascular disease is the most common cause of death in people with gout. Acute inflammation, which is a characteristic of gout, may have a mechanistic role in major adverse cardiovascular events (MACEs). We aimed to examine the relationship between admissions to a hospital with acute gout and MACEs in a large population-based data set.

Methods

We extracted data from the Hospital Morbidity Data Collection and Death Registrations of the Western Australian Rheumatic Disease Epidemiology Registry. We identified patients admitted to hospital with incident acute gout and who had admissions or a death record because of MACEs. We compared the risk of MACEs during the postdischarge period (1-30 days after acute gout admission) and control period (365 days prior to admission and 365 days after the postdischarge period) using a self-controlled case-series (SCCS) design, which is a within-person design that controls for time-invariant patient-specific confounding. We performed conditional fixed-effects Poisson regression to obtain rate ratios (RRs).

Results

We identified 941 patients (average age: 76.4 years; SD: 12.6; 66.7% male) with an incident acute gout admission and documented MACEs during the control and/or postdischarge periods. Of the 941 patients, 898 (95%) experienced MACEs during the combined control period (730-day period) and 112 (12%) during the postdischarge period (30-day period). The rates of MACEs during the total control and postdischarge periods were 0.84 and 1.45 events per person-year, respectively. Regression analysis confirmed increased rate during the postdischarge period (RR: 1.67; 95% CI: 1.38-2.03) compared with the control period. Sensitivity analyses indicated that our results were robust in relation to known limitations of the SCCS design.

Conclusion

We report an increased risk of MACEs in the first 30 days after an incident hospital admission with acute gout, suggesting a temporal association between acute inflammation and subsequent MACEs in patients with gout.

INTRODUCTION

Cardiovascular disease is the largest contributor to the increased mortality seen in gout (1). The link between gout and cardiovascular risk is now established. However, the mechanism by which cardiovascular risk arises in patients with gout is poorly understood. Noncausal associations have focused on the association between comorbidities, such as metabolic syndrome and obesity, causing oxidative stress and endothelial dysfunction (2,3). Causal theories directly implicate intracellular uric acid (UA) as pro-oxidant and pro-inflammatory (4), contributing to the development of hypertension and atherosclerotic plaques in people with gout (5–7). Interestingly, the link between chronic asymptomatic hyperuricemia (AH) and cardiovascular mortality remains less clear, with conflicting evidence as to whether AH increases cardiovascular risk independent of traditional risk factors (8–16). The discrepancy may be related to generally high levels of UA seen in people with gout than AH but may also be related to systemic inflammation seen in people with gout. For example, recent studies have investigated the role of inflammation resulting from the deposition of urate in vascular tissue in the aetiopathogenesis of major cardiovascular events in people with gout (17).

It is well established that inflammation is critical in the development of atherosclerosis, with recent studies suggesting an association between acute intense systemic inflammation seen during infection with influenza and acute myocardial infarction (18). We theorized that, similar to that reported for acute influenza, intense and acute systemic inflammation associated with gout flares is a determinant of cardiovascular risk in these patients. Investigating whether people with gout have an increased risk of major adverse cardiovascular events (MACEs) in the short term after a severe acute attack may provide weight to the theory that acute severe inflammation is a major contributor to MACEs seen in people with gout.

In this study, we examined the risk of experiencing a MACE during the 30-day postdischarge period (hereafter postdischarge period) for an acute gout hospital admission (index admission) compared with the 365 days following the postdischarge period or 365 days prior to index admission.

MATERIALS AND METHODS

The Western Australia (WA) Department of Health Human Research Ethcis Committee (approval no. 2016/24) approved the Western Australian Rheumatic Disease Epidemiological Registry (WARDER) and this project. This study invokes a waiver of consent because the participants are unidentifiable.

Data sources

We accessed the following unit-record linked data from WARDER: 1) Hospital Morbidity Data Collection (HMDC) covering all public and private hospitalizations in WA from 1980 to 2014 and 2) WA Death Registrations covering all deaths in the state from 1980 to 2015. These data sets are maintained by the WA Department of Health and linked through the WA Data Linkage System (WADLS) (19). WARDER contains all hospital admissions between 1980 and 2014 for patients with principal or secondary discharge diagnosis of systemic autoimmune rheumatic disease, gout, or osteoarthrosis (Supplementary Table 1).

Study population

The study population comprised patients aged 25 years or older who were admitted between 1991 and 2012 for an incident (ie, first-ever) acute gout hospital admission (index admission) using a fixed 10-year look-back for each patient to remove prevalent cases. We excluded patients who died during the index admission. Patients were identified from the principal diagnosis field of the HMDC: 274 and M10 for the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and the International Classification of Diseases Tenth Revision, Australian Modification (ICD-10-AM), respectively.

Study observation period

Figure 1 shows our study observation period, which began 1 year before the index admission. The following two control periods were used: 1) 1 year before the index admission (control period 1) and 2) 1 year after the 30-day postdischarge period (control period 2), which together form the combined control period.

Figure 1. Design of self-controlled case series depicting the control period and post discharge period. Control period 1 + control period 2 = combined control period

Study outcome

The study outcome was number of MACEs (composite of acute coronary syndrome, stroke, heart failure, or cardiovascular death [codes in Supplementary Table 2]) identified during the postdischarge period compared with the combined control period. We used a 28-day interval between MACEs in which any recurrent event, which occurred within 28 days, were not counted, as they were not considered a new event (20–22).

Explanatory variables

Patient sociodemographics, identified at index admission, were determined from the HMDC. Medical history and Charlson comorbidity index were determined from all diagnosis fields in the HMDC using a fixed 10-year look-back from the index admission, with codes from Quan et al (23). In addition to conditions included in the Charlson comorbidity index, we also included coronary heart disease (CHD) (ICD-9-CM: 410-414; ICD-10-AM: I20-I25) in the medical history.

Statistical analysis

We expressed the occurrences of MACEs during the postdischarge and combined control periods as per person-year by summing of the number of MACEs and dividing by the respective exposure times for each period. We used the self-controlled case-series (SCCS) study design (24) to ascertain the risk of occurrence of each outcome during the postdischarge period compared with the combined control period. The SCCS design studies the temporal association between an exposure (ie, admission for acute gout) and an adverse event (ie, MACE) using only data from patients who experienced the exposure and adverse event. This study design minimizes confounding because each patient acts as their own control, thereby controlling for identified and unidentified confounders specific to each patient that do not vary over time (25).

The SCCS uses conditional Poisson regression to model for each outcome, in which rate ratios (RRs) represent the rate of an outcome during the postdischarge period (30 days) compared with the rate during the combined control period (730 days). Only patients with gout who experienced one or more MACEs during the study observation period were included in the regression analysis.

We also performed subgroup analyses using the following characteristics at index admission: admission year (1991-2005, 2006-2012), season of the year (summer, autumn, winter, spring), age (≤65 years vs. >65 years), sex, medical history, and Charlson comorbidity index (0-3 points vs. 4+ points). We evaluated interaction terms to determine the significance between strata in each subgroup.

In addition, we conducted three sets of sensitivity analyses. Firstly, to test the assumption that the occurrence of the event of interest must not censor or affect the length of the observation period, we excluded those who died during the postdischarge period or during control period 2. Secondly, we performed a sensitivity analysis using 90-day interval between MACEs. Thirdly, we excluded those with length of stay (LOS) at index admission that was more than 30 days.

A two-tailed value of P < 0.05 was considered statistically significant. All analyses were performed using Stata 15.1 (StataCorp, LLC).

Ethical consideration

Human Research Ethics Committee approval for the project was obtained from the WA Department of Health (approval no. 2016/24).

RESULTS

We identified 1000 patients hospitalized for their incident acute gout and experienced one or more MACEs during the study observation period. We excluded 13 patients who died during the incident admission; seven of these patients died because of MACEs. The average age of the remaining 987 patients was 75.9 years (SD = 12.4) with 85% aged 65 years or older, 67% (n = 666) were male, 57% (n = 253) with a history of MACEs, 64% (n = 633) with CHD, 21% (n = 207) with heart failure, 23% (n = 225) with cerebrovascular disease, and 31% (n = 307) with diabetes at incident admission (Table 1).

Table 1 Characteristics of patients identified with first-ever hospitalization for acute gout in Western Australia between 1991 and 2012 and included in the self-controlled case series study design (N = 987)

Characteristics	N (%) or mean ± SD
Admission year
1991-2005	598 (60.6)
2006-2012	389 (39.4)
Season of the year
Summer	227 (23.0)
Autumn	214 (21.7)
Winter	259 (26.2)
Spring	287 (29.1)
Age (y) at index admission	75.9 ± 12.4
Sex
Male	666 (67.5)
Medical history at index admissiona
MACE	563 (57.0)
CHD	633 (64.1)
Heart failure	207 (21.0)
Cerebrovascular disease	225 (22.8)
Cancer	125 (12.7)
Chronic lung disease	316 (32.0)
Diabetes	307 (31.1)
Kidney disease	348 (35.3)
Peripheral vascular disease	203 (20.6)
Charlson comorbidity index at index admission
0-1 point	208 (21.1)
2-3 points	303 (30.7)
4-5 points	274 (27.8)
6+ points	202 (20.5)

Abbreviations: CHD, coronary heart disease; MACE, major adverse cardiovascular event.

^aUsing a 10-year lookback from the index admission.

There were 120 and 941 MACEs during the postdischarge and combined control period, respectively, resulting in rates of 1.48 and 0.85 events per person-year (Table 2). The risk of experiencing MACEs was significantly higher in the postdischarge period compared with the combined control period (RR: 1.70; 95% CI: 1.41-2.04).

Table 2 Number of patients and rate of MACEs in the SCCS main analysis and stratification by patient characteristics at index admission (N = 987)

[Table omitted. See PDF]

Abbreviations: CHD, coronary heart disease; CI, confidence interval; MACE, major adverse cardiovascular event; ref, reference; RR, rate ratio; SCSS, self-controlled case-series.

^aBased on interaction term.

In the secondary analyses (Table 2), effect modification (P < 0.05) was observed across levels of history of kidney disease (RR: 2.06; 95% CI: 1.64-2.57 for those with no history of kidney disease and RR: 1.21; 95% CI: 0.86-1.68 for those with history of kidney disease; P = 0.009) and Charlson comorbidity index (RR: 2.19; 95% CI: 1.70-2.81 for those scoring 0-3 points and RR: 1.33; 95% CI: 1.01-1.75 for those scoring 4+ points; P = 0.009).

In a series of sensitivity analyses (Supplementary Table 3), our results remained robust with exclusion of those who died during the postdischarge period or control period 2 (RR: 1.44; 95% CI: 1.11-1.88), when we only counted MACEs that occurred more than 90 days apart (RR: 1.67; 95% CI: 1.39-2.02), and with exclusion of those with LOS more than 30 days at index admission (RR: 1.59; 95% CI: 1.29-1.96).

DISCUSSION

In this study, we found a temporal association between an incident hospitalization for acute gout and MACEs, in that that the risk of MACEs was 1.7-times higher during the postdischarge period than the combined control period. Apart from history of kidney disease and Charlson comorbidity index, our findings were consistent across levels of age, sex, medical history, and Charlson comorbidity index and remained robust with exclusion of those who died during the postdischarge period or control period 2, when we only counted MACEs that occurred more than 90 days apart and when we excluded those with LOS more than 30 days. Although we excluded 13 patients who died during their incident admission (seven because of MACEs), we do not feel that this would have changed our findings significantly as the risk remained similar to the main findings when we excluded those who died during the postdischarge period or control period 2. It is uncertain why a lower increase in relative risk was seen in those with kidney disease or a Charlson comorbidity score greater than 3. It may be because these patients may be at a very high risk of MACEs, regardless of gout flare, and that a gout flare has little impact on risk of MACEs in these patients. However, this phenomenon was not seen in patients with a history of MACEs, who are presumably at the highest risk of repeat MACEs. This remains somewhat unexplained and should be examined in other cohorts to confirm the finding.

This study provides support to the hypothesis that systemic inflammation seen in gout is associated with the increased risk of MACEs in that population. Inflammation is critical in the development of atherosclerosis. A temporal association between acute influenza and acute myocardial infarction has been demonstrated, implicating acute systemic inflammation in the aetiopathogenesis of MACEs (3). In vivo studies have shown that infection with influenza results in atherosclerotic plaque infiltration with inflammatory cells, platelet aggregation, and thrombosis (1). This acute inflammatory stimulus is thought to cause plaque destabilization and acute coronary syndromes. Therapeutic trials also support inflammation being a driving force behind cardiovascular risk. In the CANTOS trial (26), canakinumab, a human monoclonal antibody against interleukin-1β and potent anti-inflammatory, was tested in patients with a previous myocardial infarction and an elevated level of highly sensitive C Reactive Protein. A lower incidence of the primary end point (a composite of nonfatal myocardial infarction, nonfatal stroke, and cardiovascular death) was observed compared with placebo (27). It is intuitive to assume that the inflammatory processes associated with coronary atherosclerosis initiation/progression and its complications can be modulated by anti-inflammatory molecules, as seen in the CANTOS trial, with the most plausible postulated mechanism by which anti-inflammatory agents reduce adverse cardiovascular outcome being from stabilizing the plaque (ie, by reducing the number of high risk plaques) (26). Given interleukin-1β is the cytokine driving acute inflammation in gout, this interventional study of an anti-interleukin-1β is of interest in the mechanism of cardiovascular disease (CVD) in gout.

The drivers of cardiovascular risk in gout are likely multifactorial and complex. Noncausal associations have been theorized, including the observation that levels of UA are associated with cardiovascular risk factors, including metabolic syndrome, hypertension, obesity, and renal (5,6). Causal theories directly implicate intracellular UA as being pro-oxidative and pro-inflammatory (28). UA may have direct effects on atherosclerotic plaque development through stimulation of cyclooxygenase-2, thromboxane, and increasing platelet activity (7,9,10). However, the theory that UA is causal is complicated by large observational studies that have not always confirmed that the relationship between AH (as opposed to gout) is associated with an increased cardiovascular risk (11,14,18). For example, the Framingham Heart Study found serum UA levels were not an independent risk factor for cardiovascular death after adjustment for CVD risk factors (12). Similarly, the smaller, but very well characterized, Fremantle Diabetes Study found no association between UA and cardiovascular outcomes (13). Similarly, large Mendelian randomization studies have shown no relationship between hyperuricemia and coronary artery disease (27,29). Hyperuricemia alone may not completely explain the increased CVD risk seen in gout. Admittedly, it may be difficult to independently demonstrate a sizeable impact of UA as a risk factor for CVD independently from traditional cardiovascular risk factors such as obesity, hypertension, and renal disease, given the close relationship among these factors (30).

Given the growing evidence to suggest an association between immune-mediated inflammatory conditions and infections and CVD (18,31–33), the role of systemic inflammation as a contributor to cardiovascular risk in gout should be explored. Gout is characterized by both chronic low-grade inflammation and acute exacerbations of intense inflammation. In this study, we aimed to focus on the potential impact of acute inflammation. We hypothesized that similar to acute influenza, hospitalization with acute gout would be associated with poor cardiovascular outcomes in the short term. The findings of this study provide support to this hypothesis and raise questions as to whether intense suppression of inflammation in acute gout may improve outcomes for people with gout.

The strength of this study includes the use of routinely collected whole-population administrative data to identify and determine outcomes of patients hospitalized for acute gout. This reduces issues with selection bias as well as reporting or recall bias in terms of experiencing a cardiovascular event. Limitations of this study relate to the SCCS design. Firstly, one assumption of the SCCS is that recurrent adverse events must be independent, in that the occurrence of one cardiovascular event must not alter the probability of a subsequent event occurring. In our study, we conducted a sensitivity analysis including only cases with an incident MACE during the observation period that showed similar estimates to those seen with the full sample. Secondly, the SCCS assumption that the occurrence of the event of interest, such as death from a cardiovascular cause, should not censor or affect the length of the observation period. We overcame this by excluding patients who died during the postdischarge period and control period 2. In both of these sensitivity analyses addressing these limitations, our findings remained robust with respect to the main results. This study also lacks data regarding pharmacotherapy for acute gout in this cohort, which may confound the results. Routine pharmacotherapy for acute gout might encompass colchicine, nonsteroidal anti-inflammatory drugs (NSAIDs), steroids, and occasionally anti-interleukin-1β therapies (anti-IL-1) such as anakinra or canakinumab. Canakinumab has been shown to be efficacious in the secondary prevention of MACEs; however, the evidence of benefit of colchicine in secondary prevention is conflicted (26,34). Regardless, if these agents reduce MACEs in the short term, and patients with gout in this cohort received these therapies, then our study would underestimate the association between admission with acute gout and MACEs. Although NSAIDs and steroids might be considered to have a negative effect on cardiovascular outcomes, in gout their use is usually short term, and little data exist to support a negative effect in the short term. The study also lacks reliable smoking data for the cohort, which is recognized as a limitation.

This whole-of-population study found a temporal association between gout and MACEs, in that people admitted to hospital with an incident acute attack of gout had a 70% higher risk of experiencing a MACE during the first 30 days after discharge, compared with the 365 days following the postdischarge period or 365 days prior to admission. The findings provide independent validation of recent similar work. Further investigation into the relationship between acute inflammation and MACEs in gout is required, including the impact of anti-inflammatory therapy.

ACKNOWLEDGMENTS

The authors would like to acknowledge the support of the Western Australian Data Linkage Branch and data custodians of the Hospital Morbidity Data Collection (HMDC) and Western Australian Deaths Registrations from the Western Australian Department of Health for providing the linked HMDC and death data.

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 published. Derrick Lopez 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

Lopez, Dwivedi, Nossent, Preen, Murray, Raymond, Inderjeeth, Keen.

Acquisition of data

Nossent, Preen, Raymond.

Analysis and interpretation of data

Lopez, Murray.