Obesity and younger age at gout onset in a community-based cohort

Authors


Abstract

Objective

Obesity is associated with gout risk. It is unclear whether obesity is associated with a younger age at gout onset. We examined whether obesity is related to age at gout onset and quantified the risk of incident gout by obesity status in the Campaign Against Cancer and Heart Disease (CLUE II) study, a longitudinal community-based cohort.

Methods

CLUE II began in 1989 as a cohort study of residents living within or surrounding Washington County, Maryland. Followup questionnaires queried whether each participant had been diagnosed as having gout by a health care professional. Among participants with gout, we assessed whether obesity was related to age at disease onset. We also ascertained the 18-year risk of incident gout according to obesity status (body mass index ≥30 kg/m2) at baseline with cumulative incidence ratios (RRs) and 95% confidence intervals (95% CIs) from Poisson regression.

Results

Among the study population (n = 15,533), 517 persons developed incident gout. The prevalence of obesity at baseline was 16.2%. The overall mean age at gout onset was 59.3 years. The onset of gout was 3.1 years (95% CI 0.3, 5.8) earlier in those who were obese at baseline and 11.0 years earlier (95% CI 5.8, 16.1) in participants who were obese at age 21 years, as compared with the nonobese participants. The 18-year adjusted RR of gout in obese participants compared with nonobese participants was 1.92 (95% CI 1.55, 2.37).

Conclusion

Obesity is not only a risk factor for incident gout but is associated with an earlier age at gout onset.

INTRODUCTION

In the US, the incidence and prevalence of both obesity and gout is rising (1–3). Obesity has been associated with an increased risk of gout (4–7). However, 2 of the studies that identified this gout risk factor were conducted among male health care professionals (4, 5) and another included only a few women with gout (7). These findings may not be generalized to women or to persons at risk of gout in the community setting. Community-based cohort studies are few in number, yet necessary to characterize the risk of gout associated with obesity in both sexes and across a range of ages.

Although there is an established association between body weight, obesity, and gout, the influence of obesity on the age at gout onset is not defined. Obesity has been shown to be associated with an earlier onset of other chronic diseases, including diabetes mellitus (8). Hospital and clinic-based studies have found that the age at gout onset was younger in men than in women (9, 10), differed across populations (11, 12), and was younger in people with a family history of gout (12, 13). However, the age at gout onset has not to date been studied in a US community-based population.

An improved understanding of the impact of obesity in relation to the age at gout onset will translate into better estimation of the risk of gout among patients who present with acute arthritis. As the prevalence of obesity continues to rise in the US, it is important to quantify the risk of gout across the range of body mass index (BMI) values in both women and men, and to understand the impact of obesity on the age at gout onset in the community setting. Therefore, we estimated the risk of incident gout by baseline and early adult obesity status over 18 years of followup, as well as the impact of obesity on the age at gout onset. We conducted this study among 15,533 men and women in the Campaign Against Cancer and Heart Disease (CLUE II) study, a longitudinal community-based cohort with valuable information on body weight, and prospective ascertainment of gout diagnoses.

Significance & Innovations

  • Obesity is not only a risk factor for incident gout but also was associated with an earlier age at gout onset.

  • Participants who were obese in early adulthood developed gout earlier than those who were not obese.

MATERIALS AND METHODS

Ascertainment of exposures and outcome.

CLUE II is a community-based cohort undertaken to identify risk factors for cancer and cardiovascular disease. At its inception in 1989, the cohort enrolled individuals ages 13–87 years, who resided within or surrounding Washington County, Maryland. A baseline health assessment including a questionnaire and examination with blood pressure and phlebotomy was performed at cohort entry. Followup health-history questionnaires were administered to participants in 1996, 1998, 2000, 2003, and 2007.

For the present study, the population was restricted to 16,103 of the original CLUE II participants who answered the gout query on the 2000, 2003, or 2007 questionnaire, since these were the only questionnaires to query the participants about gout. The data set was further restricted to participants who self-reported white race due to the limited number of nonwhite participants (n = 165, 1%). Additionally, participants with prevalent gout were similarly excluded from the study population. Prevalent gout was defined as self-reported age at gout diagnosis younger than the age at cohort entry, or self-reported year of gout diagnosis prior to 1989 (n = 405, 3%). We defined the age at gout onset as the participant's reported age at physician diagnosis at the first affirmative report of gout. All participants provided written consent at cohort entry in 1989. The study was approved by the Institutional Review Board at the Johns Hopkins Bloomberg School of Public Health.

At baseline, participants were sent a questionnaire to self-report demographic data, including health status, sex, race, age, height, weight, weight at age 21 years, and treatments received for high cholesterol and hypertension. Additionally, blood pressure and cholesterol levels were measured at the baseline assessment. Alcohol consumption was ascertained by a self-reported food frequency questionnaire. The questionnaire collected information regarding the frequency of beer consumption (12-ounce can or bottle), wine or wine coolers (1 medium glass), and liquor (1 shot). The 3 alcohol intake variables were categorized as 1) never or less than once per month, 2) once per month to once per week, and 3) once per week or more.

BMI at baseline was calculated as weight (kg) divided by height (meters) squared, using the 1989 self-report of weight and height. Additionally, we calculated BMI at age 21 years, based on the baseline-reported values for height and weight at age 21 years.

Followup questionnaires were sent to the CLUE II participants in 2000, 2003, and 2007 (response rates: 63.5%, 62.2%, and 55.6%, respectively). Each of these 3 followup questionnaires queried the participants as to whether they had ever received a diagnosis of gout by a health care professional. Additionally, participants were asked the year of diagnosis (1989 or before, 1990–1994, or 1995 or after) on the 2000 questionnaire, as well as the age at diagnosis on the 2003 and 2007 questionnaires. We defined an incident case of gout for any participant who self-reported gout on at least one of the 2000, 2003, or 2007 followup questionnaires and who self-reported the onset of gout after 1989, or at an age greater than the age at cohort entry. Notably, age at gout onset was found to be reliable (r = 0.85), and self-reported gout status has been found to be reliable (3-year reliability κ = 0.73) and sensitive in this population (sensitivity 84%) (14).

Analysis of age at gout onset in relation to obesity status.

We assessed the association between the various anthropometric measures and age at gout onset among the participants who reported the age at gout onset on at least one of the questionnaires. Specifically, we compared the age at gout onset by obesity status, both at baseline and at age 21 years, for all participants who responded to the 2003 or 2007 questionnaire and who self-reported the age at onset. If there were any discrepancies in the reported age at onset, the first reported age was used. The 2000 questionnaire was not used because participants self-reported the categorical year of onset rather than the age at onset. To compare the mean age at gout onset, t-tests were used. Additionally, we examined whether the age at gout onset differed with BMI modeled as a continuous variable, both at cohort entry and at age 21 years, using linear regression analysis. Both models were adjusted for sex, baseline blood pressure, and beer, wine, and liquor intake, since these factors were associated with the age at gout onset.

Analysis of the association between BMI, obesity, and gout.

Using all participants who responded to the gout query, we examined the association of known risk factors for gout, including age, sex, high blood pressure, elevated serum cholesterol level, and beer, wine, and liquor intake with baseline obesity in this population. We then assessed the 18-year risk of incident gout associated with baseline obesity with cumulative incidence ratios (RRs), referred to as the risk of gout. Ninety-five percent confidence intervals (95% CIs) were derived from Poisson regression analyses. We modeled the RR rather than the odds ratio (OR) using Poisson regression, since gout is not a rare disease in a community cohort of adults, and we reported robust SEs. However, results were similar when run as a logistic regression. To allow analysis of all reported cases of gout across the 3 followup surveys (2000, 2003, and 2007), the primary analysis focused on the cumulative incidence of gout (ever versus never). We considered baseline BMI as a continuous variable and present these results in 5 kg/m2 increments as the exposure of interest. Next, obesity in 1989 was modeled as a categorical variable, defined as those participants having a BMI value ≥30 kg/m2. BMI at baseline was grouped into 4 categories: normal weight (BMI <25 kg/m2), overweight (BMI 25–29.9 kg/m2), class I obesity (BMI 30–34.9 kg/m2), and class II or III obesity (BMI ≥35 kg/m2) (15). There were too few participants with a BMI <18.5 kg/m2 to categorize BMI as underweight. Obesity status at age 21 years was also examined, using the standard cutoff for obesity of BMI at ≥30 kg/m2.

In multivariable analyses, we adjusted the above models for known risk factors of gout that may confound the obesity and gout association. These risk factors included sex, age, baseline alcohol intake, baseline measures of blood pressure and cholesterol levels, and treatment for high blood pressure and hypercholesterolemia. All analyses were performed using SAS software, version 9.1.

RESULTS

The study population consisted of 15,533 participants, 39.3% of whom were male. The mean ± SD BMI at cohort entry in 1989 was 25.8 ± 4.8 kg/m2. At baseline, 47.8% of the participants were of normal weight, whereas 36.0% of the participants were overweight. In 1989 at the initial examination, 2,508 of the participants were obese, corresponding to a 16.2% baseline prevalence of obesity. Other population demographics are displayed in Table 1.

Table 1. Risk factors for gout among 15,533 Campaign Against Cancer and Heart Disease participants by baseline (1989) obesity status*
Baseline risk factorsFull cohortNonobese populationObese population
  • *

    Values are the number (percentage) unless otherwise indicated. Obesity defined as body mass index ≥30 kg/m2.

  • P < 0.001 comparing obese to nonobese.

Men6,100 (39.3)5,190 (39.9)907 (36.2)
Age, mean ± SD years47.0 ± 15.346.7 ± 15.748.9 ± 12.9
Cholesterol, mean ± SD mg/dl204.8 ± 39.2203.0 ± 38.9214.6 ± 39.5
Blood pressure, mean ± SD mm Hg   
 Systolic124.8 ± 16.3123.3 ± 16.1132.5 ± 15.4
 Diastolic78.7 ± 9.577.8 ± 9.383.7 ± 9.1
Treated hypertension2,393 (15.4)1,660 (12.7)731 (29.1)
Treated hypercholesterolemia581 (3.7)477 (3.1)104 (4.2)
Beer   
 Never or <1/month9,208 (70.5)7,630 (69.3)1,572 (77.1)
 1/month to 1/week1,918 (14.7)1,677 (15.2)241 (11.8)
 ≥1/week1,932 (14.8)1,707 (15.5)225 (11.0)
Wine   
 Never or <1/month10,191 (77.9)8,441 (76.4)1,744 (85.9)
 1/month to 1/week2,281 (17.4)2,037 (18.5)244 (12.0)
 ≥1/week607 (4.6)565 (5.1)42 (2.1)
Liquor   
 Never or <1/month10,530 (80.7)8,797 (79.9)1,727 (84.7)
 1/month to 1/week1,635 (12.5)1,414 (12.8)221 (10.8)
 ≥5/week890 (6.8)798 (7.3)92 (4.5)

The mean ± SD BMI at age 21 years was 22.1 ± 3.5 kg/m2. There were 449 participants (3.1%) who were obese at age 21 years. In addition, 882 participants did not self-report their weight at age 21 years. Overall, the participants reported lower BMI levels at age 21 years, corresponding to few participants described as being overweight and obese in young adult life.

Baseline obesity status was statistically associated with all the measured gout risk factors, suggesting that these factors may confound the association of obesity and gout (Table 1). Women and older participants were more likely to be obese at baseline. Additionally, obese participants had higher mean cholesterol levels and blood pressure values, and were more likely to be treated for these conditions. Finally, beer and wine intake was associated with a decreased prevalence of obesity at baseline.

Risk of incident gout.

Between 1989 and 2007, 517 CLUE II participants developed gout; there were 185 women and 332 men with incident gout. Therefore, the overall 18-year cumulative incidence of gout in this community-based population was 3.3%, corresponding to a cumulative gout incidence of 5.4% among the men and 2.0% among the women in the CLUE II cohort. Additionally, gout patients were more likely to be male (RR 2.78; 95% CI 2.33, 3.31). There were 392 participants with gout (76% of the total gout cases) who reported both their age at the onset of gout and having gout on the 2003 or 2007 followup questionnaires. The overall mean ± SD age at gout onset in the population was 59.3 ± 12.9 years (median 59, range 25–90 years). For men, the mean ± SD age at gout onset was 57.8 ± 12.4 years, and for women it was 62.2 ± 13.3 years (Figure 1). For men, the distribution of ages appears to have a non-Gaussian distribution, suggesting that there may be different risk factors for gout in earlier adulthood and later adulthood. As such, men developed gout at a younger age (on average ∼4 years earlier) than did the women in the study.

Figure 1.

The density plot of the age distribution of gout onset for men and women. Density represents the percentage of Campaign Against Cancer and Heart Disease participants that developed gout at a given age.

Age at gout onset.

The age at gout onset was younger in participants who were obese compared to those who were not obese at baseline (57.1 versus 60.2 years; P = 0.03 by t-test). However, obesity at baseline was marginally associated with a difference in the age at onset of gout for women and was associated with a significantly younger age of disease onset for men (Table 2). At age 21 years, obesity was similarly associated with a younger age at gout onset, a difference of 11 years in those who were obese compared to those who were normal weight or overweight (49.2 versus 60.1 years; P < 0.001). Moreover, this association persisted when examined separately for both men and women.

Table 2. Age at gout onset according to BMI and obesity status, at baseline and age 21 years, in the Campaign Against Cancer and Heart Disease cohort*
 TotalWomenMen
No.Age, yearsNo.Age, yearsNo.Age, years
  • *

    Values are the mean ± SD unless otherwise indicated. Obesity defined as a body mass index (BMI) ≥30 kg/m2. 95% CI = 95% confidence interval.

  • P < 0.01 by t-test.

  • P < 0.001 by t-test.

Baseline BMI      
 Not obese27160.2 ± 12.48362.5 ± 13.218859.2 ± 11.9
 Obese12157.1 ± 13.64561.6 ± 13.67654.5 ± 13.0
 Difference (95% CI) 3.1 (0.3, 5.8) 0.9 (−4.0, 5.9) 4.7 (1.40, 8.0)
BMI at age 21 years      
 Not obese36560.1 ± 12.511963.3 ± 12.824658.6 ± 12.1
 Obese2449.2 ± 10.5849.6 ± 12.21648.9 ± 8.7
 Difference (95% CI) 11.0 (5.8, 16.1) 13.6 (4.4, 22.9) 9.7 (3.6, 15.8)

Although the sample size was small for the comparison of age at onset by obesity status, we additionally observed a younger age at onset for participants who were overweight or obese at age 21 years in a post hoc sensitivity analysis (data not shown).

The linear model to assess the association between BMI modeled as a continuous variable with the age at gout onset suggested that increasing values of BMI translate into an earlier age at disease onset. After adjustment for sex, blood pressure, and alcohol intake, a 5-kg/m2 change in BMI was associated with a 1.11-year decrease in the age at gout onset (P = 0.13). For every 5-kg/m2 increase in BMI at age 21 years, there was a 4.5-year adjusted decrease in the age at gout onset (P < 0.001).

Association of obesity with incident gout.

The 18-year unadjusted risk of developing gout was more than 2 times higher (RR 2.26; 95% CI 1.89, 2.72) in obese participants compared to the nonobese participants. The risk of incident gout increased for every 5-unit increase in BMI (RR 1.51; 95% CI 1.42, 1.59). In the categorical BMI analysis with normal baseline BMI as the comparator, the risk of incident gout was 2.62 (95% CI 2.11, 3.25) in participants who were overweight, 4.01 (95% CI 3.13, 5.14) for those who were class I obese, and 3.35 (95% CI 2.33, 4.82) in those who were class II or III obese.

In the 3 adjusted models, baseline obesity, categorical BMI, and continuous BMI were each associated with incident gout (Table 3). In the age- and sex-adjusted model 1, obesity remained associated with incident gout. Further adjustment, as portrayed in model 2, to control for potential confounding by beer, wine, and liquor intake, did not significantly change the results. In the final fully-adjusted model 3, the risk of gout was nearly 2 times higher (RR 1.92; 95% CI 1.55, 2.37) for participants who were obese, compared to those who were not obese at baseline. In addition, there was an increased risk of gout for every 5-unit increase in BMI at baseline. The risk of developing gout for class I obese participants was similar to that of those who were class II or III obese, arguing for a dose response for categorical BMI that plateaus at BMIs >30 kg/m2. Additionally, an increased risk of gout was also observed for those participants who were obese at age 21 years. In the adjusted models, the risk of incident gout was nearly twice as high for participants who were obese at age 21 years (RR 1.82; 95% CI 1.21, 2.73) compared to those who were not obese.

Table 3. Adjusted cumulative RRs of incident gout by BMI (kg/m2), and obesity status, at baseline and age 21 years in the Campaign Against Cancer and Heart Disease cohort*
 Model 1Model 2Model 3
 RR (95% CI)RR (95% CI)RR (95% CI)§
  • *

    Obesity is defined as body mass index (BMI) ≥30 kg/m2. RR = incidence ratio; 95% CI = 95% confidence interval.

  • Model 1: age and sex adjusted.

  • Model 2: model 1 plus beer, wine, and liquor intake.

  • §

    Model 3: model 2 plus measures of blood pressure and cholesterol, and treatment for high blood pressure and high cholesterol.

  • P < 0.001 for trend.

Obesity at baseline2.25 (1.84, 2.75)2.23 (1.82, 2.73)1.92 (1.55, 2.37)
Categorical baseline BMI   
 Normal weight1.001.001.00
 Overweight2.03 (1.61, 2.58)2.01 (1.59, 2.56)1.88 (1.47, 2.39)
 Class I obesity3.34 (2.54, 4.39)3.31 (2.52, 4.36)2.86 (2.15, 3.80)
 Class II and III obesity3.51 (2.36, 5.20)3.49 (2.35, 5.19)2.72 (1.80, 4.12)
5-unit (kg/m2) change in baseline BMI1.53 (1.43, 1.65)1.53 (1.42, 1.65)1.44 (1.33, 1.56)
Obesity at age 21 years2.06 (1.38, 3.07)2.06 (1.38, 3.08)1.82 (1.21, 2.73)

In sex-stratified adjusted analyses, no differences were observed in the association of obesity and gout between the sexes (male RR 1.98, 95% CI 1.52, 2.58; female RR 1.78, 95% CI 1.25, 2.54), nor were the sex-stratified analyses different for obesity at age 21 years and gout (male RR 1.55, 95% CI 0.91, 2.64; female RR 2.29, 95% CI 1.22, 4.30). There was no sex and obesity interaction for obesity at baseline and early adult obesity. Therefore, there is not a difference in the associations observed between obesity incident gout and sex.

To evaluate for survival bias, we analyzed whether prevalent gout was related to mortality. These sensitivity analyses suggest that prevalent gout was not related to reporting on the 2003 and 2007 questionnaires, or 3- and 7-year mortality. Additional sensitivity analyses suggested that the results were similar when the ORs were calculated using logistic regression (OR 2.01; 95% CI 1.61, 2.52, for obese versus nonobese) and when incident rate ratios (IRs) were calculated using Poisson regression and imputation for categorical onset (IR 1.99; 95% CI 1.59, 2.48, for obese versus nonobese) suggesting that the results are robust to the analytic method. However, the Poisson regression for the RR is the most comprehensive analysis because it does depend on the reported onset and is not inflated due to the prevalence of gout in this population.

DISCUSSION

In a community-based cohort of both men and women, we quantified the 18-year risk of incident gout associated with obesity, and found that being obese at study entry, as well as at age 21 years, were both associated with an earlier age at gout onset. Our study confirms that obesity is a strong risk factor for gout, even after accounting for known risk factors and comorbid conditions. Additionally, this is the first study to show that those who were obese at age 21 years developed gout 11 years earlier than their nonobese counterparts.

Our study confirms the previously noted association of obesity and gout (4–7) and extends the research that has been previously limited to men (7) and male health professionals (4, 5). One prospective study of male physicians noted a greater than 2-fold increase in risk of gout in those whose BMI increased more than 1.88 kg/m2 prior to age 35 years (4). The Johns Hopkins Precursors Study of male medical students found the relative risk of gout to be 1.12 per 1-unit increase in BMI at age 35 years, yet this relationship was not seen when evaluating weight at age 22 years (4). In contrast to the Johns Hopkins Precursors cohort, our study did find that early life obesity was associated with incident gout in men (4). These differences may be secondary to power since there were only 60 cases of gout in this cohort; there may not have been adequate power to detect such a difference. Additionally, the association between obesity on the development of gout was also examined in the Health Professionals Follow-up Study (5). Results similar to our own were reported according to the 3 categories of BMI common to both studies: 25–29.9, 30–34.9, and ≥35 kg/m2. Although we did not adjust for dietary factors, we found that the risk of gout was similar to Choi et al (5), who reported a relative risk of 1.95, 2.33, and 2.97, respectively. Additionally, this study found the risk of gout for those who were obese at age 21 years was 1.66 times greater compared to those with normal BMI, which was similar to our results about early adult obesity. Unlike this study of male health professionals, we were unable to test short-term versus long-term effects of obesity because this was not the focus of our study. Similar to our results, in a long-term followup study of gout in the Framingham cohort, obesity was associated with greater than 2-fold risk in both men and women (6).

Within the hospital-based setting, the age at gout onset has been demonstrated to be younger in men than in women (9, 10). It is also clinically important to determine whether the age at onset varies by obesity status over adulthood. One study in Taiwan reported the association of BMI and the age at first rheumatology clinic visit (13). For study participants ages 19–44 years, 45–64 years, and ≥65 years, the ORs of gout for overweight compared to normal weight gout patients were 1.70, 1.17, and 1.36, respectively. Additionally, this study found that a higher percentage of gout patients who were ages 14–44 years at their first visit were overweight than those patients with gout who first presented for medical care in mid-adult and late-adult life. However, this study did not assess the association between obesity prior to the development of gout in relation to the age at disease onset. Notably, our study quantified the association between 2 measures of adult BMI prior to gout onset, suggesting that an earlier onset of gout is in fact associated with obesity. However, we were unable to assess all the components of metabolic syndrome, and future studies should address this hypothesis in a community-based cohort.

Studies over the past decades have provided evidence that obesity causes insulin resistance, and this in turn may lead to hyperuricemia and gout. The degree of insulin resistance correlates directly with the serum urate level and inversely with the renal urate clearance (16, 17). In one trial of patients with gout, weight loss following a prescribed change in macronutrient intake was associated with a decrease in serum urate levels and in the frequency of gout attacks (18). The components of insulin resistance, hyperinsulinemia, hypertension, dyslipidemia, and obesity, were all related to hyperuricemia in one epidemiology study, although obesity was the strongest risk factor (19). Leptin has been found to be a pathogenic factor for hyperuricemia in patients with obesity and has been hypothesized to be the link between the two (20, 21). These findings have supported the argument that hyperuricemia is an integral component of the insulin resistance syndrome (19, 22).

The main strength of this study is our large sample size, with more than 15,533 participants in CLUE II. This large population-based study and the prospective nature of the study design allowed us to estimate the risk of incident gout rather than focus on prevalent gout alone. Therefore, we can be more confident that our measures of adiposity precede the development of gout. Additionally, this study is well suited to test the hypothesis that BMI at age 21 years is related to the age at gout onset because CLUE II collected health data from adult participants of all ages. Recall of early adult weight has been shown to be valid in a previous epidemiology study (23). Therefore, we were able to assess the role of early adult obesity on the age at onset without introducing bias that would occur in cohorts with an age inclusion criteria. Our sensitivity analyses suggest a limited role of survival bias.

We acknowledge that the present study defined incident gout by self-report of a physician diagnosis of gout instead of either confirmation of monosodium urate crystals from a gouty effusion or fulfillment of the American College of Rheumatology criteria (24). Additionally, the survey did not collect information on how the physician diagnosed gout for each participant. Self-report of physician-diagnosed gout may under/over represent cases. However, our results will only be affected if there is differential misclassification of gout. There is no evidence that the presence of obesity leads to an under- or overdiagnosis of gout. We have found the self-report of gout and the self-report of age at onset to be both reliable and valid (14). Additionally, community-based cohorts that collect the outcomes data through questionnaires do not routinely allow for assessment of measured clinical risk factors such as serum urate levels, hormone levels, or renal insufficiency. However, this community-based cohort study did measure cholesterol levels and blood pressure at baseline.

We were unable to adjust for the number of grams of alcohol intake, but rather we measured categorical beer, wine, and liquor intake. However, the results did not differ when we used different parameterizations of these measures of alcohol intake. Additionally, the study did not obtain the baseline menopause status of the female participants. This study did not measure weight or height, and therefore relies on self-report of height, weight, and weight at age 21 years. Previous studies have suggested that self-report of height and weight in addition to early adult weights are valid for population-based studies (23, 25). Nor did the study routinely collect multiple measures of weight to allow for analysis of time-varying BMI and obesity. Finally, like all observational studies, ours is limited by missing data. However, these missing data are unlikely to bias our results, because we only adjusted for baseline factors, which are unlikely to be missing based on obesity status. Additionally, using cumulative RRs assumes that all participants were followed until the development of gout or the end of the study, and that loss to followup does not differ by obesity status. Finally, we used age at diagnosis as a proxy for the age of first attack, and the age-at-onset analyses were limited to small sample sizes for those who were obese at age 21 years. However, we performed sensitivity analyses to test the association between increased BMI in the overweight category and observed the comparable results. Our results do not rule out the possibility that obese participants frequent their physician's office more often than nonobese participants, and they are therefore more likely to be diagnosed as having gout at a younger age.

Our study confirms an increase in the risk of developing gout in obese men and extends these findings to women. We further found that early adulthood obesity impacts the age at gout onset, with gout occurring at substantially younger ages in those who are obese at cohort entry or at body weight levels reported at age 21 years. The results of this community-based cohort contribute important clinical findings to the gout literature pertaining to the risk and age at gout onset in relation to BMI, an association particularly important in the context of the growing epidemic of obesity in the US and other countries. In the clinical setting, our study suggests that gout should be considered in the differential diagnosis of acute arthritis in both obese men and women at all ages.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Ms McAdams DeMarco 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. McAdams DeMarco, Maynard, Huizinga, Coresh.

Acquisition of data. McAdams DeMarco, Coresh.

Analysis and interpretation of data. McAdams DeMarco, Maynard, Huizinga, Baer, Köttgen, Gelber, Coresh.

Acknowledgements

We would like to thank the CLUE II participants for their ongoing participation. Additionally, we would like to thank Judith Bolton Hoffman for her help with obtaining the data.

Ancillary