Skip to main content
Download PDF

The absolute risk of gout by clusters of gout-associated comorbidities and lifestyle factors—30 years follow-up of the Malmö Preventive Project

Arthritis Research & Therapy volume 22, Article number: 244 (2020) Cite this article

Abstract

Background

Gout is predicted by a number of comorbidities and lifestyle factors. We aimed to identify discrete phenotype clusters of these factors in a Swedish population-based health survey. In these clusters, we calculated and compared the incidence and relative risk of gout.

Methods

Cluster analyses were performed to group variables with close proximity and to obtain homogenous clusters of individuals (n = 22,057) in the Malmö Preventive Project (MPP) cohort. Variables clustered included obesity, kidney dysfunction, diabetes mellitus (DM), hypertension, cardiovascular disease (CVD), dyslipidemia, pulmonary dysfunction (PD), smoking, and the use of diuretics. Incidence rates and hazard ratios (HRs) for gout, adjusted for age and sex, were computed for each cluster.

Results

Five clusters (C1–C5) were identified. Cluster C1 (n = 16,063) was characterized by few comorbidities. All participants in C2 (n = 750) had kidney dysfunction (100%), and none had CVD. In C3 (n = 528), 100% had CVD and most participants were smokers (74%). C4 (n = 3673) had the greatest fractions of obesity (34%) and dyslipidemia (74%). In C5 (n = 1043), proportions with DM (51%), hypertension (54%), and diuretics (52%) were highest. C1 was by far the most common in the population (73%), followed by C4 (17%). These two pathways included 86% of incident gout cases. The four smaller clusters (C2–C5) had higher incidence rates and a 2- to 3-fold increased risk for incident gout.

Conclusions

Five distinct clusters based on gout-related comorbidities and lifestyle factors were identified. Most incident gout cases occurred in the cluster of few comorbidities, and the four comorbidity pathways had overall a modest influence on the incidence of gout.

Background

The development of gout is influenced by multiple genetic and environmental risk factors and interactions among these [1, 2]. The symptoms of gout arise from the body’s innate immune response to monosodium urate crystals that accumulate in the joints, tendons, and other tissues as a result of hyperuricemia (HU), which is defined as serum urate (SU) level > 408 μmol/L [3]. The global prevalence of gout is gradually increasing and ranges from 1 to 4% in developed countries [4]. In Sweden, regional studies have shown a prevalence of approximately 1.7% and that gout is 2 to 3 times more common in men than in women [5,6,7].

Gout is associated with several comorbidities and lifestyle factors. The increase in gout prevalence over the last decades is partly attributed to aging of the general population and the rise of comorbid conditions such as obesity and associated insulin resistance, as well as changing lifestyle and dietary practices [3]. The prevalence of comorbid conditions such as hypertension, obesity, CVD, diabetes, dyslipidemia, and chronic kidney disease (CKD) is significantly higher in people with gout than in the general population [8]. A recent study from the USA indicated an increase in gout incidence over 20 years and a multifold increase in many associated comorbidities in patients with gout during this time period. The relative increase was 3-fold for renal disease and dyslipidemia, 4-fold for diabetes, 30% for hypertension, and 50% for obesity [9]. The close correlation between several of these comorbidities makes it a challenge to examine their individual importance as risk markers for HU and gout using the traditional epidemiological analyses. Furthermore, HU and gout may also be potential risk factors for hypertension, CVD, CKD, and the metabolic syndrome [8, 10]. In studies using Mendelian randomization (MR) analyses, there is some support for obesity as a causal risk factor for HU and gout [11], whereas such studies have so far failed to support causality between HU and/or gout and other risk factors including CVD, CKD, and diabetes [12,13,14].

Gout has also been associated with several lifestyle factors including alcohol consumption and smoking. Whereas increased alcohol consumption has been suggested to elevate SU concentrations and to act as a trigger of acute gout attacks [15, 16], the relative importance of smoking is slightly controversial with different studies providing conflicting results [17, 18]. Epidemiological studies, examining the effect of different diuretics on the risk of gout, collectively support that such medication could increase the incidence of gout [19, 20].

Since these factors (comorbidities and lifestyle) often are closely associated with each other, they may represent a few pathophysiological pathways rather than being individually important predictors. Identifying clusters of such factors may thus lead to a better understanding of the potential pathways involved in the risk of gout. Two studies have previously reported four to five phenotype clusters in prevalent cohorts of gout patients of European ancestry [21, 22]. However, the identification of such clusters (generalizability of the pattern analysis) has neither been assessed in the general population, nor to what extent they predict future gout.

The purposes of our observational, prospective study were to (1) identify clusters of gout-related baseline comorbidities and lifestyle factors among the participants of a population-based health survey and (2), for each of these identified clusters, determine absolute and relative risks for developing gout during 30 years of follow-up.

Methods

Study cohort

The study used baseline and follow-up data from the Malmö Preventive Project (MPP) cohort (Additional file 1: Figure S1). The MPP was a population-based screening program for cardiovascular risk factors, alcohol abuse, and breast cancer that started in 1974 at the Department of Preventive Medicine, University Hospital, Malmö, Sweden. Individuals born in Malmö certain years and residents of the city were invited for a clinical examination, questionnaire, and blood sampling. A total of 33,346 (22,444 men and 10,902 women) individuals, aged 26 years and over, participated and were screened at the baseline between 1974 and 1992. During the first half of the period (1974–1982), mostly men and, during the second half (1982–1992), mostly women were invited to participate. The vast majority of the participants were Caucasians of Scandinavian origin. More details on the cohort, individual questionnaire, and data recruitment for the MPP cohort are provided elsewhere [23].

The Regional Ethics Committee at Lund University (Dnr 85/2004) provided ethical approval for this follow-up study.

Endpoint retrieval of gout

Date of the gout diagnosis was defined as the first visit with a registered diagnosis of gout [using International Classification of Diseases (ICD) versions 8, 9, and 10 (Additional file 1: Table S1)] to physicians in primary health care (from 1998), specialized inpatient care (from 1974), or outpatient care (1998 onward). The date for gout diagnosis was obtained through linkage of MPP cohort with regional (The Skåne Healthcare Register) and national health care (National Patient Register) registers. For the longitudinal analyses, follow-up started at the time point of baseline screening, which could vary between 1974 and 1992. Participants were followed until the date of first gout diagnosis, death, migration from the area, or December 31, 2014, whichever occurred first. The individuals were thus followed for a period of up to 30 years (mean 28.2 ± 8.4 years).

For this study, the baseline data from the health examination performed anytime between 1974 and 1992 for various gout-related comorbidities and lifestyle factors were used from the MPP cohort. Obesity was defined as body mass index (BMI) > 30 kg/m2. Kidney dysfunction was defined as a glomerular filtration rate (eGFR) of less than 60 mL/min/1.73m2. Diabetes mellitus was defined as single fasting blood glucose ≥ 6.7 mmol/L (according to contemporary reference levels in the 1970s) or a history of DM. Individuals with repeated systolic blood pressures ≥ 160 mmHg and/or diastolic blood pressures > 100 mmHg, or those who reported being on antihypertensive therapy, were considered to have hypertension [23]. Prevalent CVD was defined as a history of angina pectoris, myocardial infarction, or concomitant medication for heart disease. Dyslipidemia was defined by the presence of either hypercholesterolemia (serum cholesterol > 6.1 mmol/L) or hypertriglyceridemia (serum fasting triglyceride > 2.6 mmol/L). Pulmonary dysfunction (PD) was defined by measured spirometry values of the forced vital capacity (FVC) and the forced expiratory volume within 1 s (FEV1) as FEV1/FVC < 70% of predicted values [24]. Exposure to smoking status was considered present if subjects were current smokers or had a recent history of at least 10 years of smoking. The use of diuretics was based on the information provided by the participants in the study questionnaire. The information for SU and liver enzyme concentrations was also measured at baseline at the Department of Clinical Chemistry in Malmö [23]. The term “abnormal liver enzymes” was defined by the presence of serum alanine aminotransferase > 1.1 μkat/L (for men) and > 0.75 μkat/L (for women) and/or serum gamma-glutamyl transferase > 1.3 μkat/L (for men) and > 0.75 μkat/L (for women). Alcohol consumption behavior (risk behavior) was assessed using the Malmö-modified brief Michigan Alcohol Screening Test (Mm-MAST) comprising nine questions regarding alcohol habits [25]. According to previous validation studies, a score of ≥ 2 was classified as a risk drinking behavior [26]. All individuals who reported to have prevalent gout (n = 11) before or at the time of inclusion in the MPP were excluded from further analyses.

Statistical analysis

The study mainly involved cluster and proportional hazard analyses in the MPP cohort using the baseline and follow-up data, respectively. Summary data were expressed as mean ± SD for age and BMI, and as frequency and percentages for all categorical variables.

Based on the availability of complete data at baseline for a sufficient number of individuals and based on factors previously being identified as associated with gout in cross-sectional cluster analyses [21, 22], a set of nine variables was selected for the cluster analysis that included gout-related comorbidities and lifestyle factors. Due to the substantial proportions with missing data (questions not asked and tests not performed some years) and reduction in the total variance explained, respectively, for alcohol consumption behavior and information for abnormal liver enzymes, these variables were not included in the main cluster analysis. However, their demographics were calculated for each cluster post-analysis. The individuals included in the cluster analysis were compared with the rest of the population (with incomplete data, not included in the analyses) for age, sex, BMI, and SU levels to evaluate representativeness (Additional file 1: Table S2).

The selected variables were first arranged into homogeneous clusters to obtain a meaningful inter-variable relationship structure using the ClustOfVar package in R v3.5.2 [27]. This package can efficiently cluster categorical variables using principal component analysis based on the correlation ratio of variables with the first component (the cluster center) to create a hierarchy. In addition, a factorial analysis with oblique rotation (direct oblimin; assumes correlated factors) was conducted to establish a factor structure for the selected variables. Only the factors with eigenvalue > 1 were extracted. The highest loading value of a variable (comorbidity or lifestyle) was the inclusion criterion for that particular variable in any factor. The Kaiser-Meyer-Olkin (KMO > 0.5) test was performed to assess the suitability of the data for the factor analysis. Bartlett’s test for sphericity was used to compare the variable correlation matrix to the identity matrix using a standard p < 0.05 as a cutoff. The factorial analysis was carried out in SPSS Statistics v25 (IBM Corp., Armonk, NY, USA).

The individual observations were also clustered using the hierarchical agglomerative clustering in R v3.5.2 [28]. The analysis was not forced to provide a fixed number of clusters. Ward’s minimum variance criterion was applied to group the observations based on similarities in comorbidity and lifestyle variables. This method minimizes the total within-cluster variance. The clustering algorithm started with each cluster as a singleton, i.e., cluster containing one single point/observation. The algorithm progressed via finding another cluster to pair with while keeping a minimum increase in total within-cluster variance sum of squares at each step. The process of pairing iterated until all observation points were members of one single cluster.

For all identified clusters, the cumulative incidence as well as the incidence rates of gout was computed with 95% confidence intervals (95% CIs). Using the largest cluster as reference, the Cox proportional hazard model was applied, and both unadjusted and age- and sex-adjusted HRs with 95% CIs were calculated. Student t test was used to calculate sex-stratified differences of mean SU concentration for gout case-control groups in each cluster. The cumulative incidence for eight predefined gout-related comorbidities (see Additional file 1: Table S1 for ICD-8, ICD-9, and ICD-10 codes to define these comorbidities) before gout diagnosis was also calculated. The end of follow-up was defined as the date of the first diagnosis for the comorbidity of interest or gout, death, migration from the area, or study end date, i.e., December 31, 2014 (whichever occurred first).

Results

A subset of 22,057 MPP individuals (screening period 1975–1992) had sufficiently complete data and were eligible for the further classification process. Individuals in this subset were mostly males (66%), with the mean for age and BMI of 46.8 ± 5.5 years and 24.6 ± 3.5 kg/m2, respectively. The comorbidities were overall not very frequent, as expected, considering the mean age at baseline and that data was obtained from a general population sample (Table 1). Those included in the analyses had similar values for age, BMI, and urate, but tended to be older (46.8 vs. 43.4 years) and less often men (66% vs. 70%) compared to the rest of the MPP cohort (n = 11,278) (Additional file 1: Table S2).

Table 1 Characteristics of the individuals in the MPP cohort overall and for the five identified clusters of gout-related comorbidities
Full size table

Clustering of comorbidities and lifestyle variables

The variables selected for clustering included seven gout-related comorbidities: obesity, kidney dysfunction, DM, hypertension, CVD, dyslipidemia, PD, one lifestyle factor (smoking), and the use of diuretics. The variables were not completely uncorrelated; however, the correlation was very low, never exceeding 0.20 (Additional file 1: Figure S2). Initial clustering indicated three clusters with discrete patterns: CV-1 that grouped CVD, kidney dysfunction, hypertension, and use of diuretics; CV-2 clustered DM, obesity, and dyslipidemia, whereas PD and smoking formed CV-3 (Fig. 1). Factorial analysis extracted three factors (F1, F2, F3) accounting for 41.3% of the variance and confirming the results of cluster analysis. F1 combined cardiovascular risk factors and renal disease (CVD, kidney dysfunction, hypertension, and use of diuretics), F2 combined variables associated with pulmonary etiologies (PD and smoking), and F3 consisted of variables associated with the insulin resistance (DM, obesity, and dyslipidemia) (Additional file 1: Table S3). The correlation matrix indicated a weak correlation (p < 0.009) and a good separation between the factors. The KMO statistic was 0.58, with Bartlett’s test being significant (p < 0.0001), indicating that factor analysis was appropriate.

Fig. 1
figure 1

Dendrogram illustrating the results of variable clustering using the ClustOfVar package. Variables with the closest proximity were grouped, where proximity was based on similarity/difference in the pattern of individual’s response to any variable. Variables are represented by horizontal lines, and the length of horizontal lines represents the degree of similarity between variables. The “COV-based distance” on the x-axis indicates values for a distance metric between variables and clusters calculated by ClustOfVar (COV). Each color (blue, green, and red) represents a different cluster

Full size image

Clustering of data observations

The same set of selected variables (comorbidities and lifestyle/medication factors) were used to cluster individual observations. The clustering analyses resulted in five different clusters: C1 to C5 (Fig. 2). The baseline characteristics of these clusters are given in Table 1.

Fig. 2
figure 2

Dendrogram illustrating the results of cluster analysis in data observations (n = 22,057). The vertical axis on the graph represents the distance between clusters, while the horizontal axis represents the observations and clusters. Each vertical bar represents a subject and a cluster, and joining of two clusters is represented by the fusion of two vertical bars. The vertical position of the fusion, shown by short horizontal lines, gives the distance between the two clusters. C1 to C5 represent cluster numbers 1 to 5

Full size image

The first cluster (C1; n = 16,063) was the largest, which entailed 73% of the population with a mean age of 46 years and low frequencies for comorbidities. Almost all individuals in this cluster were healthy, i.e., none had obesity, kidney dysfunction, DM, CVD, or dyslipidemia. Only small proportions had hypertension (14%) or PD (15%). This cluster was labeled as “few comorbidities.”

Individuals in C2 (n = 750, 3.4%) were the oldest with a mean age of 51 years. All individuals in this cluster had kidney dysfunction (100%), and none had DM, CVD, or used diuretics. This cluster was labeled as “CKD/kidney dysfunction.”

In C3 (n = 528; 2.4%, mean age 48 years), all individuals had CVD (100%), and the group had the highest proportion of PD (22%), smoking (74%), and alcohol risk behavior (40%). None in this cluster had DM or used diuretics. The cluster was labeled as “CVD and lifestyle.”

Cluster C4 (n = 3673; mean age 47 years) represented the second-largest proportion (17%) of the clustered observations. Individuals in C4 were mostly males (75%) and had the highest mean BMI (25.9 kg/m2) and the largest proportions with obesity (34%) and dyslipidemia (74%) [hypercholesterolemia: 51% and hypertriglyceridemia: 32%]. This cluster was labeled “obesity and dyslipidemia.”

Cluster C5 (n = 1043; mean age 48 years) had by far the highest occurrence of DM (51%), hypertension (54%), and use of diuretics (52%), and the highest percentage of those with abnormal liver enzyme levels (16%). C5 was labeled “diabetes and hypertension.”

Risk of developing gout

For the subset of 22,057 MPP participants (mean follow-up of 28.1 (95% CI 27.9 to 28.2) years) included in the cluster analysis, a total of 910 (4.1%) developed incident gout after a mean follow-up of 27.5 years (range between clusters, 23.4 to 28.2; Table 2). The absolute risk of incident gout, and unadjusted and age- and sex-adjusted HRs are given in Table 2. The majority of incident gout cases during follow-up came from the largest cluster, C1 (“few comorbidities,” 60%) followed by C4 (“obesity and dyslipidemia,” 26%). The incidence of gout was highest for C2 (276.8 per 100,000 person-years at risk), and the highest relative risk for gout development was seen for C2 [HR (95% CI) 2.31 (1.73 to 3.07)] and C3 [HR (95% CI) 2.41 (1.63 to 3.58)]. However, all four clusters (C2–C5) identified by different comorbidity patterns resulted in a 2- to 3-fold increased risk for incident gout compared to C1 (Table 2).

Table 2 Incidence and hazard ratios (HRs) for gout during follow-up and mean SU of diagnosed patients with gout in the five observation-clusters of the MPP cohort
Full size table

Sex-stratified SU level comparisons between those developing gout and those not during follow-up showed the highest differences for C5 men [Δ (95% CI) 80.3 μmol/L (41.6 to 119.0)] and C2 women [Δ (95% CI) = 73.7 μmol/L (27.6 to 119.7)], while the differences were lowest for C3, in both men [Δ (95% CI) 47.7 μmol/L (11.0 to 84.4)] and women [Δ (95% CI) 30.1 μmol/L (− 13.2 to 73.4)] (Additional file 1: Table S4).

Incidence of gout-related comorbidities in the five clusters

Comorbidities that are considered to be well-documented risk factors/markers for HU and gout, such as obesity, CKD, and diagnosed alcoholism, had much lower cumulative incidence (2.4%, 19.4%, and 0.4%, respectively) in the “few comorbidities” cluster (C1) compared to other clusters, even when combining those identified at baseline and with a diagnosed disease during follow-up (Table 3). Also, for other comorbidities considered to be associated with HU or gout, C1 had numerically lower (diabetes, dyslipidemia, CVD, hypertension) cumulative incidence compared to other clusters (Table 3), although considerable proportions were diagnosed with hypertension (57%) or CVD (46%) before gout diagnosis. During the follow-up, the cumulative incidence of several comorbidities remained comparatively higher (CKD, 31.1%; dyslipidemia, 17.8%; and COPD, 11.1%) in C5 than in other clusters.

Table 3 Within clusters’ cumulative incidence for various gout-related comorbidities for gout incident cases during follow-up (n = 910)
Full size table

Discussion

In a population-based setting, we identified four clusters with increased occurrence of gout-related comorbidities and one with very few such comorbidities. The latter entailed the vast majority of diagnosed gout disease (60%), whereas the relative risks for the four clusters characterized by various combinations of comorbidities were 2- to 3-fold higher. Two studies to date have performed cluster analyses to categorize comorbidities in prevalent gout patients in European populations, using cross-sectional data [21, 22]. Our study is the first to identify gout-related comorbidities’ clusters as predictors of gout longitudinally in a general population, and to compare the absolute and relative risk for incident gout in such clusters.

A cluster with few comorbidities was also defined in European studies performed in prevalent gout patients, although the proportions were lower, 12% [21] and 36% [22], respectively, compared to ours. In the present study, comorbidities were determined on average 27.5 years before gout diagnosis. None in this cluster characterized by “few comorbidities” had kidney dysfunction, DM, obesity, or dyslipidemia at baseline, and only a minority developed such comorbidities before gout diagnosis. Nevertheless, 3.4% of this cluster was diagnosed with gout during follow-up and constituted 60% of the incident gout cases in our study. This supports the concept that other pathways, not defined by comorbidities, are of major importance for the development of clinical gout. These entail both genetic (in particular for HU) and non-genetic (e.g., diet and environment) pathways. Genome-wide association studies have identified a number of genes directly associated with SU levels and their possible contribution to gout [29, 30]. While the genetics of progression from HU to gout are yet poorly understood, studies have shown genetic variants to account for considerable variance in SU levels (27–41%) [29] and gout risk (30%) [31] in Europeans. The risk attributed to the diet may also be important [32, 33], although there are conflicting results regarding its effect size [34]. There could also be a genetic component to gout independent of other comorbidities. This argument is supported by a more recent study that reported similar genetic risk of gout among three subgroups stratified for BMI [35].

The prevalence of kidney dysfunction was 100% in C2, while none in this group had diabetes or CVD. This cluster has similarities to one of those identified by Bevis et al. [22] in prevalent gout, where 97% had CKD, but very low frequencies of CVD and obesity. In our study, this group characterized by reduced kidney function in midlife had the highest relative risk (HR 2.74) of incident gout, which is in line with the general perception that one out of every ten individuals with CKD develops gout at some stage [36]. We also found that the cumulative incidence of other comorbidities remained comparatively low in this cluster pre-gout diagnosis, except for CVD that was diagnosed in 45% among those who subsequently developed gout group. This is in line with the literature that described cardiovascular complications as a common finding among people with CKD [37, 38].

Associations for CVD with HU and gout have been shown repeatedly in observational studies [39, 40]. Our analysis also revealed a cluster of individuals (C3) who all had CVD, whereas the proportions in other clusters varied from minimal to nil. Interestingly, C3 was the cluster with the least difference in SU between those being later diagnosed vs. not with gout. Studies using the urate-associated genetic variants as instruments (MR studies) have not supported a causal relationship between urate levels and cardiovascular outcomes so far [41, 42]. In addition, causal evidence for cardiovascular disease as a risk factor/marker for gout is also lacking. This cluster also had the highest proportions of subjects who were smokers or had alcohol consumption risk behavior. In C3, the cumulative incidence of other comorbidities remained fairly low pre-gout diagnosis, except for hypertension and dyslipidemia. This cluster seems to represent the combination of a few components of the metabolic syndrome and cardiovascular disease together with lifestyle-related risk factors as possible predictors for gout, for which, so far, a causal relationship is less supported compared to clusters 2 and 4.

The second-largest cluster (C4, 17%) in our population had the highest prevalence of obesity (35%) and dyslipidemia (74%) (hyperlipidemia per se), and an absence of kidney dysfunction, diabetes, or CVD. This is in line with the findings of Richette et al. [21], who identified a cluster in prevalent gout patients with a high occurrence of obesity and with none having diabetes, CVD, or renal failure. C4 also contributed with the second-largest fraction (26%) of incident gout cases in our study. Obesity is a well-recognized risk factor for HU and gout, both based on observational studies [43, 44] and MR studies, having adiposity-associated genetic traits as the exposure for urate levels [45, 46] and gout risk [11]. As expected, C4 entailed the highest cumulative incidence for diabetes (26%) and CVD (50%) before gout diagnosis, compared to other clusters. While insulin resistance can be hypothesized to be a common underpinning link between these three conditions, obesity itself can also be a strong risk factor for cardiovascular events and diabetes. A meta-analysis that included data from over 880,000 individuals from multiple ethnic origins indicated that obesity (determined by BMI-associated genetic variants) could causally increase the risk of type 2 diabetes and coronary artery disease [47].

In C5, all subjects had at least one comorbidity and it was the only cluster that included individuals with DM (51%) and had subjects treated with diuretics (52%). Furthermore, it had the highest baseline prevalence and cumulative incidence (among those developing gout) of hypertension as well as high occurrence of many other comorbidities. Hypertension is common in type 2 diabetes patients, and coexistence of these two conditions worsens the clinical outcomes in humans, especially with increased risks for heart and end-stage renal disease [48]. Mendelian randomization studies have so far not demonstrated a causal relationship between HU and type 2 diabetes [14], although positive associations between insulin resistance-related variants and high urate concentrations and gout risk have repeatedly been shown [49, 50]. The previous two studies [21, 22] in prevalent gout patients also identified clusters with the combination of various comorbidities, albeit with lower proportions of CVD and CKD than in the C5 of our study. Considering the high prevalence of diabetes at baseline, it is not unexpected that among subjects diagnosed with gout during follow-up, this cluster had the highest cumulative incidence of hypertension and dyslipidemia. In C5, it may thus be that a combination of interlinked metabolic conditions, including hypertension, insulin resistance, and cardio-renal diseases, as well as PD favored the onset of gout. However, the exclusive occurrence of subjects at baseline treated with diuretics, a well-known predictor of new-onset as well as recurrence of established gout [43, 51], suggests that this may also be a major contributor to incidence gout via secondary HU in this cluster.

The study has several strengths. First, we were able to largely reproduce at a population level the clusters of gout-related comorbidities previously identified only in prevalent gout patients [21, 22]. Second, the study is based on a large population-based cohort which strengthen the generalizability of these clusters. Third, loss to follow-up was minimal due to the almost complete national Swedish register data. Fourth and most importantly, the longitudinal approach with assessment of comorbidity patterns before gout diagnosis, which has not been used in previous cluster analyses of comorbidities in prevalent gout patients [21, 22], gave the unique opportunity to study both the absolute and relative risk for gout in relation to these comorbidity clusters.

There are also possible limitations of the study. First, there may be misclassification of gout diagnosis. However, previous validation studies suggest a high positive predictive value for gout diagnosis in the Swedish health care registers [52, 53]. Second, those not included in the analyses due to missing baseline covariates may have introduced a selection bias. Since those not included in the analyses had similar BMI and SU to those in the analyses, this is unlikely to have caused major effects on the results. Third, longitudinal information on some exposures (i.e., diuretics, BMI, and SU) during the follow-up would have been useful.

Conclusions

This population-based study in middle-aged subjects demonstrated the occurrence of five clusters of gout-related comorbidities in the general population. Although clusters representing various comorbidity pathways significantly increased the relative risk for incident clinical gout, the vast majority of gout patients came from the cluster reflecting few comorbidities. These findings support that major determinants for gout include factors other than comorbidities, such as genetics or other environmental factors.

Availability of data and materials

Please contact author for data request.

Abbreviations

BMI:

Body mass index

C1 to C5:

Cluster 1 to 5

CI:

Confidence interval

CKD:

Chronic kidney disease

COPD:

Chronic obstructive pulmonary disease

CV-1 to CV-3:

Variable cluster 1 to 3

CVD:

Cardiovascular disease

DM:

Diabetes mellitus

eGFR:

Estimated glomerular filtration rate

F1 to F3:

Factor 1 to 3

FEV1:

Forced expiratory volume within 1 s

FVC:

Forced vital capacity

HR:

Hazard ratio

HU:

Hyperuricemia

ICD:

International Classification of Diseases

KMO:

Kaiser-Meyer-Olkin

Mm-MAST:

Malmö-modified brief Michigan Alcohol Screening Test

MPP:

Malmö Preventive Project

MR:

Mendelian randomization

PD:

Pulmonary dysfunction

SD:

Standard deviation

SU:

Serum urate

References

  1. Robinson PC, Horsburgh S. Gout: joints and beyond, epidemiology, clinical features, treatment and co-morbidities. Maturitas. 2014;78:245–51.

    PubMed  Article  PubMed Central  Google Scholar 

  2. Major TJ, Dalbeth N, Stahl EA, Merriman TR. An update on the genetics of hyperuricaemia and gout. Nat Rev Rheumatol. 2018;14:341–53.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  3. Dalbeth N, Merriman TR, Stamp LK. Gout. Lancet. 2016;388:2039–52.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  4. Kuo C-F, Grainge MJ, Zhang W, Doherty M. Global epidemiology of gout: prevalence, incidence and risk factors. Nat Rev Rheumatol. 2015;11:649–62.

    PubMed  Article  PubMed Central  Google Scholar 

  5. Wändell P, Carlsson AC, Ljunggren G. Gout and its comorbidities in the total population of Stockholm. Prev Med. 2015;81:387–91.

    PubMed  Article  PubMed Central  Google Scholar 

  6. Dehlin M, Drivelegka P, Sigurdardottir V, Svärd A, Jacobsson LT. Incidence and prevalence of gout in Western Sweden. Arthritis Res Ther. 2016;18:164.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  7. Kapetanovic MC, Hameed M, Turkiewicz A, Neogi T, Saxne T, Jacobsson L, et al. Prevalence and incidence of gout in southern Sweden from the socioeconomic perspective. RMD open. 2016;2:e000326.

    PubMed  PubMed Central  Article  Google Scholar 

  8. Kuo C-F, Grainge MJ, Mallen C, Zhang W, Doherty M. Comorbidities in patients with gout prior to and following diagnosis: case-control study. Ann Rheum Dis. 2016;75:210–7.

    PubMed  Article  PubMed Central  Google Scholar 

  9. Elfishawi MM, Zleik N, Kvrgic Z, Michet CJ, Crowson CS, Matteson EL, et al. The rising incidence of gout and the increasing burden of comorbidities: a population-based study over 20 years. J Rheumatol. 2018;45:574–9.

    PubMed  Article  PubMed Central  Google Scholar 

  10. Ndrepepa G. Uric acid and cardiovascular disease. Clin Chim Acta. 2018;484:150–63.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  11. Larsson SC, Burgess S, Michaëlsson K. Genetic association between adiposity and gout: a Mendelian randomization study. Rheumatology. 2018;57:2145–8.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  12. Efstathiadou A, Gill D, McGrane F, Quinn T, Dawson J. Genetically determined uric acid and the risk of cardiovascular and neurovascular diseases: a Mendelian randomization study of outcomes investigated in randomized trials. J Am Heart Assoc. 2019;8:e012738.

    PubMed  PubMed Central  Article  Google Scholar 

  13. Jordan DM, Choi HK, Verbanck M, Topless R, Won H-H, Nadkarni G, et al. No causal effects of serum urate levels on the risk of chronic kidney disease: a Mendelian randomization study. PLoS Med. 2019;16:e1002725.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  14. Sluijs I, Holmes MV, Van Der Schouw YT, Beulens JW, Asselbergs FW, Huerta JM, et al. A Mendelian randomization study of circulating uric acid and type 2 diabetes. Diabetes. 2015;64:3028–36.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. Choi HK, Atkinson K, Karlson EW, Willett W, Curhan G. Alcohol intake and risk of incident gout in men: a prospective study. Lancet. 2004;363:1277–81.

    PubMed  Article  PubMed Central  Google Scholar 

  16. Tu H-P, Tung Y-C, Tsai W-C, Lin G-T, Ko Y-C, Lee S-S. Alcohol-related diseases and alcohol dependence syndrome is associated with increased gout risk: a nationwide population-based cohort study. Joint Bone Spine. 2017;84:189–96.

    PubMed  Article  PubMed Central  Google Scholar 

  17. Fanning N, Merriman TR, Dalbeth N, Stamp LK. An association of smoking with serum urate and gout: a health paradox. Semin Arthritis Rheum. 2018;47:825–42.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  18. Teng GG, Pan A, Yuan J-M, Koh W-P. Cigarette smoking and risk of incident gout in the Singapore Chinese Health Study. Arthritis Care Res. 2016;68:1135–42.

    Article  CAS  Google Scholar 

  19. Hueskes BA, Roovers EA, Mantel-Teeuwisse AK, Janssens HJ, van de Lisdonk EH, Janssen M. Use of diuretics and the risk of gouty arthritis: a systematic review. Semin Arthritis Rheum. 2012;41:879–89.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  20. Bruderer S, Bodmer M, Jick SS, Meier CR. Use of diuretics and risk of incident gout: a population-based case–control study. Arthritis Rheumatol. 2014;66:185–96.

    PubMed  Article  PubMed Central  Google Scholar 

  21. Richette P, Clerson P, Périssin L, Flipo R-M, Bardin T. Revisiting comorbidities in gout: a cluster analysis. Ann Rheum Dis. 2015;74:142–7.

    PubMed  Article  PubMed Central  Google Scholar 

  22. Bevis M, Blagojevic-Bucknall M, Mallen C, Hider S, Roddy E. Comorbidity clusters in people with gout: an observational cohort study with linked medical record review. Rheumatology. 2018;57:1358–63.

    PubMed  Article  PubMed Central  Google Scholar 

  23. Berglund G, Nilsson P, Eriksson KF, Nilsson JÅ, Hedblad B, Kristenson H, et al. Long-term outcome of the Malmö Preventive Project: mortality and cardiovascular morbidity. J Intern Med. 2000;247:19–29.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  24. Bergström U, Jacobsson LT, Nilsson J-Å, Berglund G, Turesson C. Pulmonary dysfunction, smoking, socioeconomic status and the risk of developing rheumatoid arthritis. Rheumatology. 2011;50:2005–13.

    PubMed  Article  PubMed Central  Google Scholar 

  25. Kapetanovic MC, Nilsson P, Turesson C, Englund M, Dalbeth N, Jacobsson L. The risk of clinically diagnosed gout by serum urate levels: results from 30 years follow-up of the Malmö Preventive Project cohort in southern Sweden. Arthritis Res Ther. 2018;20:190.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  26. Kristenson H, Trell E. Indicators of alcohol consumption: comparisons between a questionnaire (Mm-MAST), interviews and serum γ-glutamyl transferase (GGT) in a health survey of middle-aged males. Br J Addict. 1982;77:297–304.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  27. Chavent M, Kuentz V, Liquet B, Saracco L. ClustOfVar: an R package for the clustering of variables. J Stat Softw. 2012;50:1–16.

    Article  Google Scholar 

  28. R Core T. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2018. https://www.R-project.org/.

    Google Scholar 

  29. Köttgen A, Albrecht E, Teumer A, Vitart V, Krumsiek J, Hundertmark C, et al. Genome-wide association analyses identify 18 new loci associated with serum urate concentrations. Nat Genet. 2013;45:145–54.

    PubMed  Article  CAS  PubMed Central  Google Scholar 

  30. Kanai M, Akiyama M, Takahashi A, Matoba N, Momozawa Y, Ikeda M, et al. Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases. Nat Genet. 2018;50:390–400.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  31. Cadzow M, Merriman TR, Dalbeth N. Performance of gout definitions for genetic epidemiological studies: analysis of UK Biobank. Arthritis Res Ther. 2017;19:181.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  32. Choi HK, Atkinson K, Karlson EW, Willett W, Curhan G. Purine-rich foods, dairy and protein intake, and the risk of gout in men. N Engl J Med. 2004;350:1093–103.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  33. Choi HK, Curhan G. Soft drinks, fructose consumption, and the risk of gout in men: prospective cohort study. BMJ. 2008;336:309–12.

    PubMed  PubMed Central  Article  Google Scholar 

  34. Major TJ, Topless RK, Dalbeth N, Merriman TR. Evaluation of the diet wide contribution to serum urate levels: meta-analysis of population based cohorts. BMJ. 2018;363:k3951.

    PubMed  PubMed Central  Article  Google Scholar 

  35. Tai V, Narang RK, Gamble G, Cadzow M, Stamp LK, Merriman TR, et al. Do serum urate-associated genetic variants differentially contribute to gout risk according to body mass index? Analysis of the UK Biobank. Arthritis Rheumatol. 2020;72:1184–91.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  36. Lim JJ, Fu A-C, Reasner D, Taylor DC. Prevalence of CKD and uncontrolled gout among us adults: results from nhanes 2007-2012. Am J Kidney Dis. 2017;69:A63.

    Google Scholar 

  37. Major RW, Cheng MR, Grant RA, Shantikumar S, Xu G, Oozeerally I, et al. Cardiovascular disease risk factors in chronic kidney disease: a systematic review and meta-analysis. PLoS One. 2018;13:e0192895.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  38. Yuan J, Zou X-R, Han S-P, Cheng H, Wang L, Wang J-W, et al. Prevalence and risk factors for cardiovascular disease among chronic kidney disease patients: results from the Chinese cohort study of chronic kidney disease (C-STRIDE). BMC Nephrol. 2017;18:23.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  39. Pillinger MH, Bangalore S, Klein AB, Baumgartner S, Morlock R. Cardiovascular disease and gout: real-world experience evaluating patient characteristics, treatment patterns, and health care utilization. J Manag Care Spec Pharm. 2017;23:677–83.

    PubMed  PubMed Central  Google Scholar 

  40. Pagidipati NJ, Clare RM, Keenan RT, Chiswell K, Roe MT, Hess CN. Association of gout with long-term cardiovascular outcomes among patients with obstructive coronary artery disease. J Am Heart Assoc. 2018;7:e009328.

    PubMed  PubMed Central  Article  Google Scholar 

  41. White J, Sofat R, Hemani G, Shah T, Engmann J, Dale C, et al. Plasma urate concentration and risk of coronary heart disease: a Mendelian randomisation analysis. Lancet Diabetes Endocrinol. 2016;4:327–36.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  42. Palmer TM, Nordestgaard BG, Benn M, Tybjærg-Hansen A, Smith GD, Lawlor DA, et al. Association of plasma uric acid with ischaemic heart disease and blood pressure: mendelian randomisation analysis of two large cohorts. BMJ. 2013;347:f4262.

    PubMed  PubMed Central  Article  Google Scholar 

  43. Choi HK, Atkinson K, Karlson EW, Curhan G. Obesity, weight change, hypertension, diuretic use, and risk of gout in men: the health professionals follow-up study. Arch Intern Med. 2005;165:742–8.

    PubMed  Article  PubMed Central  Google Scholar 

  44. Chen JH, Pan WH, Hsu CC, Yeh WT, Chuang SY, Chen PY, et al. Impact of obesity and hypertriglyceridemia on gout development with or without hyperuricemia: a prospective study. Arthritis Care Res (Hoboken). 2013;65:133–40.

    Article  Google Scholar 

  45. Lyngdoh T, Vuistiner P, Marques-Vidal P, Rousson V, Waeber G, Vollenweider P, et al. Serum uric acid and adiposity: deciphering causality using a bidirectional Mendelian randomization approach. PLoS One. 2012;7:e39321.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  46. Oikonen M, Wendelin-Saarenhovi M, Lyytikäinen L-P, Siitonen N, Loo B-M, Jula A, et al. Associations between serum uric acid and markers of subclinical atherosclerosis in young adults. The cardiovascular risk in Young Finns study. Atherosclerosis. 2012;223:497–503.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  47. Riaz H, Khan MS, Siddiqi TJ, Usman MS, Shah N, Goyal A, et al. Association between obesity and cardiovascular outcomes: a systematic review and meta-analysis of Mendelian randomization studies. JAMA Netw Open. 2018;1:e183788.

    PubMed  PubMed Central  Article  Google Scholar 

  48. Khangura DS, Salam MW, Brietzke SA, Sowers JR. Hypertension in diabetes. In: Feingold KR, Anawalt B, Boyce A, et al., editors. Endotext. South Dartmouth: MDText.com, Inc.; 2000.

    Google Scholar 

  49. Wang B, Meng D, Wang J, Jia Z, Zhoub S, Liu S, et al. Positive correlation between Beta-3-Adrenergic Receptor (ADRB3) gene and gout in a Chinese male population. J Rheumatol. 2011;38:738–40.

    PubMed  Article  PubMed Central  Google Scholar 

  50. Fatima T, Altaf S, Phipps-Green A, Topless R, Flynn TJ, Stamp LK, et al. Association analysis of the beta-3 adrenergic receptor Trp64Arg (rs4994) polymorphism with urate and gout. Rheumatol Int. 2016;36:255–61.

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  51. Kahn AM. Effect of diuretics on the renal handling of urate. Semin Nephrol. 1988;8:305–14.

    CAS  PubMed  PubMed Central  Google Scholar 

  52. Dehlin M, Stasinopoulou K, Jacobsson L. Validity of gout diagnosis in Swedish primary and secondary care-a validation study. BMC Musculoskelet Disord. 2015;16:149.

    PubMed  PubMed Central  Article  Google Scholar 

  53. Dehlin M, Landgren AJ, Bergsten U, Jacobsson LT. The validity of gout diagnosis in primary care: results from a patient survey. J Rheumatol. 2019;46:1531–4.

    PubMed  Article  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable

Funding

The study has received grants from the Swedish Rheumatism Association, Kockska Foundations, Skåne University Hospital donation funds, and Alfred Österlund Foundation. This study was supported by the City of Malmö, the Swedish Medical Research Council, and the Swedish Heart and Lung Foundation. Open Access funding provided by Lund University.

Author information

Author notes

  1. Lennart T.H. Jacobsson and Meliha C. Kapetanovic shared the last authorship.

Affiliations

  1. Department of Rheumatology and Inflammation Research, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

    Tahzeeb Fatima, Mats Dehlin & Lennart T. H. Jacobsson

  2. Lund Arthritis Research Group, Lund University, Lund, Sweden

    Tahzeeb Fatima

  3. Department of Clinical Sciences, Lund University, Malmö, Sweden

    Peter M. Nilsson

  4. Rheumatology, Department of Clinical Sciences, Lund University, Malmö, Sweden

    Carl Turesson

  5. Department of Rheumatology, Skåne University Hospital, Malmö, Sweden

    Carl Turesson

  6. Department of Medicine, University of Auckland, Auckland, New Zealand

    Nicola Dalbeth

  7. Department of Clinical Sciences Lund, Section of Rheumatology, Lund University and Skåne University Hospital, Lund, Sweden

    Meliha C. Kapetanovic

Authors
  1. Tahzeeb Fatima
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Peter M. Nilsson
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Carl Turesson
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Mats Dehlin
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Nicola Dalbeth
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Lennart T. H. Jacobsson
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Meliha C. Kapetanovic
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

TF, LTHJ, and MCK helped to design the study, oversee its execution, generate and analyze and/or interpret the data/results, and write the manuscript. PMN, CT, MD, and ND helped to interpret the results and revised the manuscript for important intellectual content. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Tahzeeb Fatima.

Ethics declarations

Ethics approval and consent to participate

The Regional Ethics Committee at Lund University (Dnr 85/2004) granted ethical approval for this study.

Consent for publication

Not applicable

Competing interests

TF and MCK declare no competing interests. PMN declares modest honoraria for lecturing from Merck, AstraZeneca, Boehringer, and Novo Nordisk. CT has received consulting fees from Roche; speaking fees from Abbvie, Bristol Myers-Squibb, Pfizer, and Roche; and an unrestricted research grant from Bristol Myers-Squibb. MD has received consulting fees from Grünental. ND declares personal fees (all < $10,000) from Hengrui, Selecta, Arthrosi, AstraZeneca, Horizon, Dyve Biosciences, Abbvie, and Janssen. LTHJ has received lecture and consulting fees from Pfizer, Abbvie, Novartis, and Eli-Lily.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1.

Supplemental material. Figure S1 to S2 and Tables S1 to S4.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Fatima, T., Nilsson, P.M., Turesson, C. et al. The absolute risk of gout by clusters of gout-associated comorbidities and lifestyle factors—30 years follow-up of the Malmö Preventive Project. Arthritis Res Ther 22, 244 (2020). https://doi.org/10.1186/s13075-020-02339-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13075-020-02339-0

Keywords

  • Gout
  • Urate
  • Epidemiology
  • Comorbidities
  • Clusters
  • Risk

Arthritis Research & Therapy

ISSN: 1478-6362

Contact us