Eur J Nutr (2014) 53:1591–1601 DOI 10.1007/s00394-014-0766-0
REVIEW
Body mass index and the risk of gout: a systematic review and dose–response meta-analysis of prospective studies Dagfinn Aune • Teresa Norat • Lars J. Vatten
Received: 14 February 2014 / Accepted: 2 September 2014 / Published online: 11 September 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose Greater body fatness has been associated with increased risk of gout in several studies; however, the strength of the association has differed between studies, and it is not clear whether the association differs by gender. We conducted a systematic review and meta-analysis of prospective studies to clarify the association between adiposity and risk of gout. Methods PubMed and Embase were searched up to August 30, 2013. Summary relative risks (RRs) were calculated using a random effects model. Results Ten prospective studies of body mass index (BMI) and gout risk with 27,944 cases and 215,739 participants were included (median follow-up 10.5 years). The summary RR for a 5 unit increment was 1.55 [95 % confidence interval (95 % CI) 1.44–1.66, I2 = 67 %] for all studies combined. The heterogeneity was explained by one study, which appeared to be an outlier. The summary RR per 5 BMI units was 1.62 (95 % CI 1.33–1.98, I2 = 79 %) for men and 1.49 (95 % CI 1.32–1.68, I2 = 30 %) for women, pheterogeneity = 0.72. The relative risks were 1.78, 2.67, 3.62, and 4.64 for persons with BMI 25, 30, 35, and 40 compared with persons with a BMI of 20. BMI in young Electronic supplementary material The online version of this article (doi:10.1007/s00394-014-0766-0) contains supplementary material, which is available to authorized users. D. Aune L. J. Vatten Department of Public Health, Faculty of Medicine, Norwegian University of Science and Technology, Trondheim, Norway D. Aune (&) T. Norat Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, St. Mary’s Campus, Norfolk Place, Paddington, London W2 1PG, UK e-mail: [email protected]
adulthood, waist-to-hip ratio, and weight gain from age 21–25 to midlife were also associated with increased risk, but few studies were included in these analyses. Conclusions Greater body mass index increases risk of gout. Further studies are needed on adiposity throughout the life course, waist-to-hip ratio, and weight changes in relation to gout as there were few studies that had published on these exposures. Keywords Body mass index Waist-to-hip ratio Weight gain Gout Meta-analysis
Introduction Gout is the most common inflammatory arthritis in men, affecting 8.3 million American adults in 2007–2008 and has a prevalence of 5.9 and 2 % in US men and women, respectively [1]. Gout is caused by elevated levels of uric acid in the blood, which crystallizes and deposit in joints, tendons and surrounding tissues and causes considerable morbidity by causing severe pain. Hyperuricemia (elevated blood concentrations of uric acid) is the underlying cause of gout, and the risk of gout increases 10–20-fold among persons with hyperuricemia [2]. However, because hyperuricemia is much more common than gout, additional factors are likely to contribute to the development of gout. Historically, gout has been known as the ‘‘disease of kings’’ as it is linked with an affluent lifestyle, and currently, the prevalence of gout is increasing in part due to population aging, changes in dietary and lifestyle factors, and increasing rates of overweight and obesity [3–5]. Overweight and obesity have been associated with increased risk of gout in several [6–15], but not all studies [16]; however, the size of the associations has varied from a
123
1592
30 % increase in the risk (15) up to a threefold increase in the risk (8, 10–13). Further, a few studies have assessed the association between abdominal adiposity, as measured by waist circumference or waist-to-hip ratio (WHR), and risk of gout [8, 13, 15] and reported increased risk, and a few other studies also reported an association between weight changes and gout [7, 8, 13]. However, it is unclear whether confounding by other risk factors including hypertension, alcohol, and dietary factors could explain the findings relating adiposity to increased risk of gout. The prevalence of gout in women is considerably lower than among men [1]. Higher estrogen concentrations have been associated with lower uric acid levels and may partly explain this gender difference. Whether or not there is a difference between men and women with regard to the risk of gout associated with adiposity is, however, unclear as studies generally reported increased risk in both genders [6–10, 14, 15]. For these reasons, we conducted a systematic review and meta-analysis of prospective studies of the association between BMI, WHR, weight changes, and the risk of gout to clarify the strength of the association, the shape of the dose–response relationship, potential confounding by other risk factors, and potential effect modification by gender.
Methods
Eur J Nutr (2014) 53:1591–1601
One cohort provided two publications [13, 14], and we used the publication with the largest number of cases [14] for the overall dose–response analysis, but for the analysis stratified by gender and for the nonlinear analysis, we used the other publication [13] because the publication used for the main analysis did not provide enough details for these analyses. One cohort [7] could not be included in the analysis of weight gain because it only provided a high versus low comparison with two categories of weight gain (3 categories of weight gain are needed to conduct dose– response analyses). The study by Choi et al. conducted analyses using both baseline and updated information on BMI. For the main analysis, we used the results from the analysis using updated BMI data, but we conducted a sensitivity analysis using the baseline information as well. A list of the excluded studies and reasons for exclusion can be found in Online Resource 1. Data extraction The following data were extracted from each study: The first author’s last name, publication year, country where the study was conducted, the study name, follow-up period, sample size, gender, age, number of cases, assessment method of anthropometric factors (measured vs. selfreported), exposure variable, level of the exposure variable, RRs and 95 % CIs, and variables adjusted for in the analysis.
Search strategy Statistical methods PubMed and Embase databases were searched for eligible studies from their inception to 30 of August 2013 using the following search terms: ‘‘body mass index OR BMI OR overweight OR obesity OR anthropometry OR fatness OR body fatness OR abdominal fatness OR abdominal obesity OR waist circumference OR waist-to-hip ratio OR risk factor OR risk factors’’ AND ‘‘gout.’’ The reference lists of the relevant studies were further inspected for potentially relevant publications. We followed standard criteria for meta-analyses of observational studies [17]. Study quality was assessed using the Newcastle-Ottawa scale [18]. Study selection We included prospective or retrospective cohort studies, case-cohort studies, or nested case–control studies of the association between BMI, waist circumference, or WHR and risk of gout. Estimates of the relative risk (hazard ratio, risk ratio) had to be available with the 95 % confidence intervals (95 % CI), and for the dose–response analysis, a quantitative measure of the exposure variable and the total number of cases and person-years had to be available in the publication. We identified 10 cohort studies (11 publications) that could be included in the analysis of BMI [6–16].
123
Summary RRs and 95 % CIs for a 5 unit increment in BMI (kg/m2) and for a 0.1 unit increment in WHR, and per 5 kg of weight gain were calculated using a random effects model [19]. These increments are consistent with previous meta-analyses of adiposity and other diseases [20, 21]. The average of the natural logarithm of the RRs was estimated, and the RR from each study was weighted by the inverse of its variance and then unweighted by a variance component, which corresponds to the amount of heterogeneity in the meta-analysis. A two-sided p \ 0.05 was considered statistically significant. For studies which reported results separately for men and women, we combined the sexspecific estimates using a fixed-effects model to generate an estimate for both genders combined before including the study in the overall analysis. The method of Greenland and Longnecker [22] was used for the dose–response analysis, and study-specific linear trends and 95 % CIs were calculated from the natural logs of the RRs and CIs across categories of anthropometric measures. The method requires that the distribution of cases and person-years or non-cases and the RRs with the variance estimates for at least three quantitative exposure categories are known. We estimated the distribution of
Eur J Nutr (2014) 53:1591–1601
cases or person-years in studies that did not report these, but reported the total number of cases and person-years (See supplement in reference number [20] for details). The mean BMI, waist circumference of WHR level, in each category was assigned to the corresponding relative risk for each study, and when these measures were reported as ranges, we estimated the midpoint in each category by calculating the average of the lower and upper cut-off point. When extreme categories were open ended, we used the width of the adjacent interval to calculate an upper or lower boundary, but we used 18.5 as the lower bound when the upper cut-off was \25, consistent with definition of normal weight by the World Health Organisation. We also conducted nonlinear dose–response analyses between BMI, and WHR, weight gain and gout using fractional polynomial models [23]. The best fitting second-order fractional polynomial regression model was determined and was defined as the model with the lowest deviance. To test for nonlinearity, we used a likelihood ratio test to assess the difference between the nonlinear and linear models [23]. Subgroup and meta-regression analyses were conducted to investigate potential sources of heterogeneity, and heterogeneity between studies was quantitatively assessed by the Q test and I2 [24]. Publication bias was assessed by inspecting the funnel plots for asymmetry and with Egger’s test [25] and Begg’s test [26], with the results considered to indicate potential publication bias when p \ 0.10. Sensitivity analyses excluding one study at a time were conducted to clarify whether the results were simply due to one large study or a study with an extreme result. The statistical analyses were conducted using Stata version 10.1 (StataCorp, College Station, Texas, USA).
Results We identified ten prospective studies (eleven publications) [6–16] that were included in the analyses of BMI and risk of gout (Table 1, Fig. 1). Two studies [8, 13] were included in the analyses of BMI at age 21–25, WHR, and weight gain and gout. Characteristics of the included studies are provided in Table 1. Seven of the studies were from the USA, and one study was from the United Kingdom and two from Taiwan. The duration of follow-up ranged from 5 to 52 years (median = 10.5 years). BMI Ten prospective studies [6–16] were included in the overall dose–response analysis of BMI and gout incidence and included a total of 27,944 cases among 215,739 participants. The summary RR for a 5-unit increment in BMI was 1.55 (95 % CI 1.44–1.66, I2 = 67.2 %, pheterogeneity \ 0.0001)
1593
(Fig. 2a). The summary RR ranged from 1.50 (95 % CI 1.44–1.55), when the study by Williams [9] was excluded, to 1.58 (95 % CI 1.46–1.70), when the study by Chen et al. [15] was excluded. Heterogeneity was also reduced when the study by Williams (9) was excluded, I2 = 10.4 %, pheterogeneity = 0.35, and it appeared to be an outlier. There was no evidence of publication bias with Egger’s test, p = 0.56 or with Begg’s test, p = 0.59 or by inspection of the funnel plot (Online resource 2). Although there was evidence of nonlinearity in the association between body mass index and gout, pnonlinearity = 0.003, with a slight flattening of the curve at higher BMI levels, the association appeared to be linear over most of the BMI range investigated (Fig. 2b, Online Resource 2). Compared with a person with a BMI of 20, the RRs were 1.78, 2.67, 3.62, and 4.64 for a person with a BMI of 25, 30, 35, and 40 (Online Resource 3). BMI at age 21–25, waist-to-hip ratio, and weight change Two studies [8, 13] reported on BMI in young adulthood (age 21–25 years), WHR and weight change (weight gain) in relation to risk of gout and included 836 cases among 53,413 participants. The summary RR per 5-units increase in BMI at age 21–25 was 1.57 (95 % CI 0.96–2.57, I2 = 89 %, pheterogeneity = 0.002) (Online Resource 4a). There was no evidence for a nonlinear association between BMI at age 21–25 and risk of gout, pnonlinearity = 0.99 (Online Resource 4b). The summary RR per 0.1-unit increase in WHR was 1.82 (95 % CI 1.44–2.29, I2 = 29 %, pheterogeneity = 0.24) (Fig. 3a). The test for nonlinearity was marginally significant, pnonlinearity = 0.06, with a steeper increase in risk at higher levels of BMI (Fig. 3b, Online resource 5). Compared with a WHR of 0.870, a WHR of 1.000 was increased more than twofold (Online resource 5). The summary RR per 5 kg of weight gain from age 21–25 to baseline was 1.28 (95 % CI 1.20–1.37, I2 = 0 %, pheterogeneity = 0.35) (Fig. 4a). There was no evidence of a nonlinear association between weight gain and risk of gout, pnonlinearity = 0.13 (Fig. 4b, Online resource 6). Subgroup and sensitivity analyses The association between BMI and gout persisted in subgroup analyses stratified by study characteristics such as duration of follow-up, study quality score, geographic location, number of cases, and adjustment for confounding factors including hypertension, renal failure or insufficiency, diuretic use, intake of alcohol, meat, fish, and fruit and vegetable intake, although there were few studies in some of these subgroups (Table 2). The association also persisted in both genders with summary RRs of 1.62 (95 % CI 1.33–1.98) for men and 1.49 (95 % CI 1.32–1.68) among
123
123 1961–1970—NA, 14.9 years follow-up
The Normative Aging Study
The Johns Hopkins Percursors Study
The Kinmen Study
Health Professionals Follow-up Study
The National Runners’ Health Study
Framingham Heart Study
Campion et al. [6], USA
Roubenoff et al. [7], USA
Lin et al. [16], Taiwan
Choi et al. [8], USA
Williams [9], USA
Bhole et al. [10], USA
1950–2002, 52 years follow-up
1991–1993—NA, 7.74 years follow-up
1986–1998, 12 years follow-up HPFS
1991–1992— 1996–1997, 5 years follow-up
1948–1964—1989, 29 years follow-up
Follow-up period
Study name
Authors, country/ region
2,476 women and 1951 men, age 29–62 years: 304 cases (104/200 w/m)
28,990 men, median age 44.6 years: 228 cases
5
8
Measured
8
Self-reported
Self-reported
5
Self-reported
223 asymptomatic hyperuricemic men, age NA: 42 cases 47,150 men, age 40–75 years: 730 cases
7
9
Study quality score
Measured
Measured
Assessment of weight and height
1,216 men, mean age 22.2 years: 60 cases
21–81 years: 84 cases
2,046 men, age
Study size, gender, age, number of cases
1.00 1.44 (0.88–2.37) 2.74 (1.65–4.58) 1.00 1.76 (1.22–2.54) 2.90 (1.89–4.44)
\25 25–29.9 C30 \25 25–29.9 C30
BMI, men
BMI, women
1.18 (1.14–1.23) 1.06 (1.04–1.07)
Per cm
1.99 (1.49–2.66) Waist circumference
1.39 (1.02–1.90) ?30 or more Per unit
1.14 (0.84–1.56)
?5 to ?9 ?10 to ?19
1.00 0.86 (0.58–1.27)
-4 to ?4
?20 to ?29
0.73 (0.44–1.22) 1.16 (0.68–1.97)
1.82 (1.39–2.39) -10 to -5
1.35 (1.01–1.79) 0.98–1.39 -10 or more lbs
1.11 (0.83–1.50) 0.95–0.97
C30
0.92–0.94
1.23 (1.00–1.52) 1.66 (1.06–2.60)
25–29.9
1.00
1.00 (0.81–1.23)
23–24.9
1.02 (0.76–1.38)
1.00
21–22.9
0.70–0.88
0.92 (0.74–1.15)
\21
0.89–0.91
1.79 (1.27–2.13)
1.65 (1.27–2.13)
25–29.9
2.30 (1.30–4.06)
0.97 (0.73–1.30)
23–24.9
C35
1.00
21–22.9
30–34.9
0.48 (0.23–1.00)
\21
1.95 (1.44–2.65)
25–29.9
2.33 (1.62–3.36)
1.31 (0.94–1.83)
23–24.9
2.97 (1.73–5.10)
1.00
21–22.9
C35
0.85 (0.43–1.68)
\21
30–34.9
1.01 (0.72–1.22)
1.12 (1.01–1.24)
1.11 (1.04–1.19)
RR (95 % CI)
Per unit
Per unit
Per unit
Description of categories
BMI
Weight change since age 21 years, lbs
Waist-to-hip ratio
BMI at age 21 years
Baseline BMI
Updated BMI
BMI
BMI at age 35
BMI
Exposure
Table 1 Prospective studies of body mass index, waist circumference, waist-to-hip ratio, weight change, and risk of gout
Age, education, alcohol, hypertension, use of diuretics, blood glucose level, blood cholesterol level, menopausal status (women)
Age, meat, fish, alcohol, fruit, aspirin, hypertension, physical activity
Age, total energy intake, diuretic use, history of hypertension, history of chronic renal failure, alcohol intake, fluid intake, meat intake seafood, purine-rich vegetable intake, animal protein, dairy foods
Uric acid, alcohol, diuretics
Age, hypertension
Age, hypertension, cholesterol, alcohol, glucose, systolic blood pressure, socioeconomic status
Adjustment for confounders
1594 Eur J Nutr (2014) 53:1591–1601
NCC: 24,768 cases 50,000 controls
2000–2007, 7 years
1987–1989— 1996–1998, 9 years follow-up
1987–1989–1996–1998, 9 years follow-up
The Health Improvement Network database
Atherosclerosis Risk in Communities Study
Atherosclerosis Risk in Communities Study
National Health Insurance Database
Cea Soriano et al. [12], United Kingdom
Maynard JW et al. [13], USA
McAdamsDeMarco et al. [14], USA
Chen et al. [15], Taiwan
BMI body mass index, HDL high-density lipoprotein, NA not available
1996–2002, 6.45 years follow-up
15,533 men and women, age 13–87 years: 517 cases
1989–2007, 18 years follow-up
Campaign Against Cancer and Heart Disease (CLUE 2)
McAdams DeMarco et al. [14], USA
40,513 men and women, age: 1,189 cases
10,872 men and women, age 45–64 years: 274 cases
6,263 women, age 45–65 years: 106 cases
Age 20–89 years
Study size, gender, age, number of cases
Follow-up period
Study name
Authors, country/ region
Table 1 continued
8
8
Measured
Measured
8
Measured
7
7
Self-reported
Self-reported
Study quality score
Assessment of weight and height
BMI, women, B50 years
BMI, women, age [50 years
BMI, men
BMI
Weight change between age 25 years and baseline
BMI at age 25 years
Baseline waist-tohip ratio
Baseline BMI
Baseline obesity (BMI)
BMI
BMI
Exposure
Obesity at age 21 years
C16.3
1.00 1.18 (0.73–1.91) 1.63 (0.93–2.86)
B24 [27
[27 [24–27
1.83 (1.31–2.55) 1.97 (1.48–2.62)
[24–27
1.30 (1.11–1.53)
[27
1.00
1.31 (1.14–1.51) B24
1.00
B24 [24–27
1.39 (1.25–1.54)
2.05 (1.06–3.96)
6.8–16.2
Per 5 units
1.00 1.54 (0.77–3.08)
\6.8 kg
3.36 (2.09–5.41) 2.84 (1.33–6.09)
C30
1.00
\25 25–29.9
1.56 (0.91–2.69) 2.78 (1.65–4.70)
C0.968
1.00
\0.900 0.901–0.967
2.76 (1.40–5.44) 3.90 (1.95–7.82)
1.63 (0.84–3.18)
25–29.9 C35
1.00
\25 30–34.9
1.00 2.37 (1.53–3.68)
C30
C30
\30
1.62 (1.55–1.70) 2.34 (2.22–2.47)
25–29
0.66 (0.58–0.76)
Obesity at baseline
1.00
2.23 (1.82–2.73) 2.06 (1.38–3.08)
Per 5 units
15–19
1.53 (1.42–1.65)
C35
20–24
3.31 (2.52–4.36) 3.49 (2.35–5.19)
30–34.9
1.00 2.01 (1.59–2.56)
18.5–24.9
RR (95 % CI)
25–29.9
Description of categories
Age, hyperuricemia, hypertriglyceridemia, low HDL cholesterol, high blood pressure, hyperglycemia, renal insufficiency, cigarette smoking, alcohol, central obesity
Age, sex, hypertension, race, alcohol, estimated glomerular filtration rate
Age, menopausal status, race, diabetes, hypertension, diuretic use, alcohol, organ meat, estimated glomerular filtration rate
Age, sex, alcohol, number of general practitioner visits, chronic renal failure, hypertension, heart failure, ischemic heart disease, hyperlipidemia, diabetes, nephrolithiasis, psoriasis
Age, sex, beer, wine, liquor
Adjustment for confounders
Eur J Nutr (2014) 53:1591–1601 1595
123
1596
Eur J Nutr (2014) 53:1591–1601
3661 total records in PubMed and Embase 1386 records identified in the PubMed database 2275 records identified in the Embase databse 1134 duplicates excluded 2527 unique records
2484 excluded based on title or abstract
Body mass index and gout, per 5 units
A
Relative Risk (95% CI)
Study Chen, 2013
1.35 ( 1.21, 1.52)
Maynard, 2012
1.39 ( 1.25, 1.54)
Cea Soriano, 2011
1.51 ( 1.47, 1.56)
McAdams DeMarco, 2011
1.53 ( 1.42, 1.65)
Bhole, 2010
1.68 ( 1.43, 1.98)
Williams, 2008
2.29 ( 1.93, 2.82)
Choi, 2005
1.45 ( 1.26, 1.68)
Lin, 2000
1.05 ( 0.19, 2.70)
Roubenoff, 1991
1.76 ( 1.05, 2.93)
Campion, 1987
1.69 ( 1.22, 2.39) 1.55 ( 1.44, 1.66)
Overall
43 given detailed assessment .75
1
1.5
2
3
Relative Risk
32 articles excluded: 11 not relevant exposure, outcome or data 1 no risk estimates 1 duplicate 2
REVIEW
Body mass index and the risk of gout: a systematic review and dose–response meta-analysis of prospective studies Dagfinn Aune • Teresa Norat • Lars J. Vatten
Received: 14 February 2014 / Accepted: 2 September 2014 / Published online: 11 September 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose Greater body fatness has been associated with increased risk of gout in several studies; however, the strength of the association has differed between studies, and it is not clear whether the association differs by gender. We conducted a systematic review and meta-analysis of prospective studies to clarify the association between adiposity and risk of gout. Methods PubMed and Embase were searched up to August 30, 2013. Summary relative risks (RRs) were calculated using a random effects model. Results Ten prospective studies of body mass index (BMI) and gout risk with 27,944 cases and 215,739 participants were included (median follow-up 10.5 years). The summary RR for a 5 unit increment was 1.55 [95 % confidence interval (95 % CI) 1.44–1.66, I2 = 67 %] for all studies combined. The heterogeneity was explained by one study, which appeared to be an outlier. The summary RR per 5 BMI units was 1.62 (95 % CI 1.33–1.98, I2 = 79 %) for men and 1.49 (95 % CI 1.32–1.68, I2 = 30 %) for women, pheterogeneity = 0.72. The relative risks were 1.78, 2.67, 3.62, and 4.64 for persons with BMI 25, 30, 35, and 40 compared with persons with a BMI of 20. BMI in young Electronic supplementary material The online version of this article (doi:10.1007/s00394-014-0766-0) contains supplementary material, which is available to authorized users. D. Aune L. J. Vatten Department of Public Health, Faculty of Medicine, Norwegian University of Science and Technology, Trondheim, Norway D. Aune (&) T. Norat Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, St. Mary’s Campus, Norfolk Place, Paddington, London W2 1PG, UK e-mail: [email protected]
adulthood, waist-to-hip ratio, and weight gain from age 21–25 to midlife were also associated with increased risk, but few studies were included in these analyses. Conclusions Greater body mass index increases risk of gout. Further studies are needed on adiposity throughout the life course, waist-to-hip ratio, and weight changes in relation to gout as there were few studies that had published on these exposures. Keywords Body mass index Waist-to-hip ratio Weight gain Gout Meta-analysis
Introduction Gout is the most common inflammatory arthritis in men, affecting 8.3 million American adults in 2007–2008 and has a prevalence of 5.9 and 2 % in US men and women, respectively [1]. Gout is caused by elevated levels of uric acid in the blood, which crystallizes and deposit in joints, tendons and surrounding tissues and causes considerable morbidity by causing severe pain. Hyperuricemia (elevated blood concentrations of uric acid) is the underlying cause of gout, and the risk of gout increases 10–20-fold among persons with hyperuricemia [2]. However, because hyperuricemia is much more common than gout, additional factors are likely to contribute to the development of gout. Historically, gout has been known as the ‘‘disease of kings’’ as it is linked with an affluent lifestyle, and currently, the prevalence of gout is increasing in part due to population aging, changes in dietary and lifestyle factors, and increasing rates of overweight and obesity [3–5]. Overweight and obesity have been associated with increased risk of gout in several [6–15], but not all studies [16]; however, the size of the associations has varied from a
123
1592
30 % increase in the risk (15) up to a threefold increase in the risk (8, 10–13). Further, a few studies have assessed the association between abdominal adiposity, as measured by waist circumference or waist-to-hip ratio (WHR), and risk of gout [8, 13, 15] and reported increased risk, and a few other studies also reported an association between weight changes and gout [7, 8, 13]. However, it is unclear whether confounding by other risk factors including hypertension, alcohol, and dietary factors could explain the findings relating adiposity to increased risk of gout. The prevalence of gout in women is considerably lower than among men [1]. Higher estrogen concentrations have been associated with lower uric acid levels and may partly explain this gender difference. Whether or not there is a difference between men and women with regard to the risk of gout associated with adiposity is, however, unclear as studies generally reported increased risk in both genders [6–10, 14, 15]. For these reasons, we conducted a systematic review and meta-analysis of prospective studies of the association between BMI, WHR, weight changes, and the risk of gout to clarify the strength of the association, the shape of the dose–response relationship, potential confounding by other risk factors, and potential effect modification by gender.
Methods
Eur J Nutr (2014) 53:1591–1601
One cohort provided two publications [13, 14], and we used the publication with the largest number of cases [14] for the overall dose–response analysis, but for the analysis stratified by gender and for the nonlinear analysis, we used the other publication [13] because the publication used for the main analysis did not provide enough details for these analyses. One cohort [7] could not be included in the analysis of weight gain because it only provided a high versus low comparison with two categories of weight gain (3 categories of weight gain are needed to conduct dose– response analyses). The study by Choi et al. conducted analyses using both baseline and updated information on BMI. For the main analysis, we used the results from the analysis using updated BMI data, but we conducted a sensitivity analysis using the baseline information as well. A list of the excluded studies and reasons for exclusion can be found in Online Resource 1. Data extraction The following data were extracted from each study: The first author’s last name, publication year, country where the study was conducted, the study name, follow-up period, sample size, gender, age, number of cases, assessment method of anthropometric factors (measured vs. selfreported), exposure variable, level of the exposure variable, RRs and 95 % CIs, and variables adjusted for in the analysis.
Search strategy Statistical methods PubMed and Embase databases were searched for eligible studies from their inception to 30 of August 2013 using the following search terms: ‘‘body mass index OR BMI OR overweight OR obesity OR anthropometry OR fatness OR body fatness OR abdominal fatness OR abdominal obesity OR waist circumference OR waist-to-hip ratio OR risk factor OR risk factors’’ AND ‘‘gout.’’ The reference lists of the relevant studies were further inspected for potentially relevant publications. We followed standard criteria for meta-analyses of observational studies [17]. Study quality was assessed using the Newcastle-Ottawa scale [18]. Study selection We included prospective or retrospective cohort studies, case-cohort studies, or nested case–control studies of the association between BMI, waist circumference, or WHR and risk of gout. Estimates of the relative risk (hazard ratio, risk ratio) had to be available with the 95 % confidence intervals (95 % CI), and for the dose–response analysis, a quantitative measure of the exposure variable and the total number of cases and person-years had to be available in the publication. We identified 10 cohort studies (11 publications) that could be included in the analysis of BMI [6–16].
123
Summary RRs and 95 % CIs for a 5 unit increment in BMI (kg/m2) and for a 0.1 unit increment in WHR, and per 5 kg of weight gain were calculated using a random effects model [19]. These increments are consistent with previous meta-analyses of adiposity and other diseases [20, 21]. The average of the natural logarithm of the RRs was estimated, and the RR from each study was weighted by the inverse of its variance and then unweighted by a variance component, which corresponds to the amount of heterogeneity in the meta-analysis. A two-sided p \ 0.05 was considered statistically significant. For studies which reported results separately for men and women, we combined the sexspecific estimates using a fixed-effects model to generate an estimate for both genders combined before including the study in the overall analysis. The method of Greenland and Longnecker [22] was used for the dose–response analysis, and study-specific linear trends and 95 % CIs were calculated from the natural logs of the RRs and CIs across categories of anthropometric measures. The method requires that the distribution of cases and person-years or non-cases and the RRs with the variance estimates for at least three quantitative exposure categories are known. We estimated the distribution of
Eur J Nutr (2014) 53:1591–1601
cases or person-years in studies that did not report these, but reported the total number of cases and person-years (See supplement in reference number [20] for details). The mean BMI, waist circumference of WHR level, in each category was assigned to the corresponding relative risk for each study, and when these measures were reported as ranges, we estimated the midpoint in each category by calculating the average of the lower and upper cut-off point. When extreme categories were open ended, we used the width of the adjacent interval to calculate an upper or lower boundary, but we used 18.5 as the lower bound when the upper cut-off was \25, consistent with definition of normal weight by the World Health Organisation. We also conducted nonlinear dose–response analyses between BMI, and WHR, weight gain and gout using fractional polynomial models [23]. The best fitting second-order fractional polynomial regression model was determined and was defined as the model with the lowest deviance. To test for nonlinearity, we used a likelihood ratio test to assess the difference between the nonlinear and linear models [23]. Subgroup and meta-regression analyses were conducted to investigate potential sources of heterogeneity, and heterogeneity between studies was quantitatively assessed by the Q test and I2 [24]. Publication bias was assessed by inspecting the funnel plots for asymmetry and with Egger’s test [25] and Begg’s test [26], with the results considered to indicate potential publication bias when p \ 0.10. Sensitivity analyses excluding one study at a time were conducted to clarify whether the results were simply due to one large study or a study with an extreme result. The statistical analyses were conducted using Stata version 10.1 (StataCorp, College Station, Texas, USA).
Results We identified ten prospective studies (eleven publications) [6–16] that were included in the analyses of BMI and risk of gout (Table 1, Fig. 1). Two studies [8, 13] were included in the analyses of BMI at age 21–25, WHR, and weight gain and gout. Characteristics of the included studies are provided in Table 1. Seven of the studies were from the USA, and one study was from the United Kingdom and two from Taiwan. The duration of follow-up ranged from 5 to 52 years (median = 10.5 years). BMI Ten prospective studies [6–16] were included in the overall dose–response analysis of BMI and gout incidence and included a total of 27,944 cases among 215,739 participants. The summary RR for a 5-unit increment in BMI was 1.55 (95 % CI 1.44–1.66, I2 = 67.2 %, pheterogeneity \ 0.0001)
1593
(Fig. 2a). The summary RR ranged from 1.50 (95 % CI 1.44–1.55), when the study by Williams [9] was excluded, to 1.58 (95 % CI 1.46–1.70), when the study by Chen et al. [15] was excluded. Heterogeneity was also reduced when the study by Williams (9) was excluded, I2 = 10.4 %, pheterogeneity = 0.35, and it appeared to be an outlier. There was no evidence of publication bias with Egger’s test, p = 0.56 or with Begg’s test, p = 0.59 or by inspection of the funnel plot (Online resource 2). Although there was evidence of nonlinearity in the association between body mass index and gout, pnonlinearity = 0.003, with a slight flattening of the curve at higher BMI levels, the association appeared to be linear over most of the BMI range investigated (Fig. 2b, Online Resource 2). Compared with a person with a BMI of 20, the RRs were 1.78, 2.67, 3.62, and 4.64 for a person with a BMI of 25, 30, 35, and 40 (Online Resource 3). BMI at age 21–25, waist-to-hip ratio, and weight change Two studies [8, 13] reported on BMI in young adulthood (age 21–25 years), WHR and weight change (weight gain) in relation to risk of gout and included 836 cases among 53,413 participants. The summary RR per 5-units increase in BMI at age 21–25 was 1.57 (95 % CI 0.96–2.57, I2 = 89 %, pheterogeneity = 0.002) (Online Resource 4a). There was no evidence for a nonlinear association between BMI at age 21–25 and risk of gout, pnonlinearity = 0.99 (Online Resource 4b). The summary RR per 0.1-unit increase in WHR was 1.82 (95 % CI 1.44–2.29, I2 = 29 %, pheterogeneity = 0.24) (Fig. 3a). The test for nonlinearity was marginally significant, pnonlinearity = 0.06, with a steeper increase in risk at higher levels of BMI (Fig. 3b, Online resource 5). Compared with a WHR of 0.870, a WHR of 1.000 was increased more than twofold (Online resource 5). The summary RR per 5 kg of weight gain from age 21–25 to baseline was 1.28 (95 % CI 1.20–1.37, I2 = 0 %, pheterogeneity = 0.35) (Fig. 4a). There was no evidence of a nonlinear association between weight gain and risk of gout, pnonlinearity = 0.13 (Fig. 4b, Online resource 6). Subgroup and sensitivity analyses The association between BMI and gout persisted in subgroup analyses stratified by study characteristics such as duration of follow-up, study quality score, geographic location, number of cases, and adjustment for confounding factors including hypertension, renal failure or insufficiency, diuretic use, intake of alcohol, meat, fish, and fruit and vegetable intake, although there were few studies in some of these subgroups (Table 2). The association also persisted in both genders with summary RRs of 1.62 (95 % CI 1.33–1.98) for men and 1.49 (95 % CI 1.32–1.68) among
123
123 1961–1970—NA, 14.9 years follow-up
The Normative Aging Study
The Johns Hopkins Percursors Study
The Kinmen Study
Health Professionals Follow-up Study
The National Runners’ Health Study
Framingham Heart Study
Campion et al. [6], USA
Roubenoff et al. [7], USA
Lin et al. [16], Taiwan
Choi et al. [8], USA
Williams [9], USA
Bhole et al. [10], USA
1950–2002, 52 years follow-up
1991–1993—NA, 7.74 years follow-up
1986–1998, 12 years follow-up HPFS
1991–1992— 1996–1997, 5 years follow-up
1948–1964—1989, 29 years follow-up
Follow-up period
Study name
Authors, country/ region
2,476 women and 1951 men, age 29–62 years: 304 cases (104/200 w/m)
28,990 men, median age 44.6 years: 228 cases
5
8
Measured
8
Self-reported
Self-reported
5
Self-reported
223 asymptomatic hyperuricemic men, age NA: 42 cases 47,150 men, age 40–75 years: 730 cases
7
9
Study quality score
Measured
Measured
Assessment of weight and height
1,216 men, mean age 22.2 years: 60 cases
21–81 years: 84 cases
2,046 men, age
Study size, gender, age, number of cases
1.00 1.44 (0.88–2.37) 2.74 (1.65–4.58) 1.00 1.76 (1.22–2.54) 2.90 (1.89–4.44)
\25 25–29.9 C30 \25 25–29.9 C30
BMI, men
BMI, women
1.18 (1.14–1.23) 1.06 (1.04–1.07)
Per cm
1.99 (1.49–2.66) Waist circumference
1.39 (1.02–1.90) ?30 or more Per unit
1.14 (0.84–1.56)
?5 to ?9 ?10 to ?19
1.00 0.86 (0.58–1.27)
-4 to ?4
?20 to ?29
0.73 (0.44–1.22) 1.16 (0.68–1.97)
1.82 (1.39–2.39) -10 to -5
1.35 (1.01–1.79) 0.98–1.39 -10 or more lbs
1.11 (0.83–1.50) 0.95–0.97
C30
0.92–0.94
1.23 (1.00–1.52) 1.66 (1.06–2.60)
25–29.9
1.00
1.00 (0.81–1.23)
23–24.9
1.02 (0.76–1.38)
1.00
21–22.9
0.70–0.88
0.92 (0.74–1.15)
\21
0.89–0.91
1.79 (1.27–2.13)
1.65 (1.27–2.13)
25–29.9
2.30 (1.30–4.06)
0.97 (0.73–1.30)
23–24.9
C35
1.00
21–22.9
30–34.9
0.48 (0.23–1.00)
\21
1.95 (1.44–2.65)
25–29.9
2.33 (1.62–3.36)
1.31 (0.94–1.83)
23–24.9
2.97 (1.73–5.10)
1.00
21–22.9
C35
0.85 (0.43–1.68)
\21
30–34.9
1.01 (0.72–1.22)
1.12 (1.01–1.24)
1.11 (1.04–1.19)
RR (95 % CI)
Per unit
Per unit
Per unit
Description of categories
BMI
Weight change since age 21 years, lbs
Waist-to-hip ratio
BMI at age 21 years
Baseline BMI
Updated BMI
BMI
BMI at age 35
BMI
Exposure
Table 1 Prospective studies of body mass index, waist circumference, waist-to-hip ratio, weight change, and risk of gout
Age, education, alcohol, hypertension, use of diuretics, blood glucose level, blood cholesterol level, menopausal status (women)
Age, meat, fish, alcohol, fruit, aspirin, hypertension, physical activity
Age, total energy intake, diuretic use, history of hypertension, history of chronic renal failure, alcohol intake, fluid intake, meat intake seafood, purine-rich vegetable intake, animal protein, dairy foods
Uric acid, alcohol, diuretics
Age, hypertension
Age, hypertension, cholesterol, alcohol, glucose, systolic blood pressure, socioeconomic status
Adjustment for confounders
1594 Eur J Nutr (2014) 53:1591–1601
NCC: 24,768 cases 50,000 controls
2000–2007, 7 years
1987–1989— 1996–1998, 9 years follow-up
1987–1989–1996–1998, 9 years follow-up
The Health Improvement Network database
Atherosclerosis Risk in Communities Study
Atherosclerosis Risk in Communities Study
National Health Insurance Database
Cea Soriano et al. [12], United Kingdom
Maynard JW et al. [13], USA
McAdamsDeMarco et al. [14], USA
Chen et al. [15], Taiwan
BMI body mass index, HDL high-density lipoprotein, NA not available
1996–2002, 6.45 years follow-up
15,533 men and women, age 13–87 years: 517 cases
1989–2007, 18 years follow-up
Campaign Against Cancer and Heart Disease (CLUE 2)
McAdams DeMarco et al. [14], USA
40,513 men and women, age: 1,189 cases
10,872 men and women, age 45–64 years: 274 cases
6,263 women, age 45–65 years: 106 cases
Age 20–89 years
Study size, gender, age, number of cases
Follow-up period
Study name
Authors, country/ region
Table 1 continued
8
8
Measured
Measured
8
Measured
7
7
Self-reported
Self-reported
Study quality score
Assessment of weight and height
BMI, women, B50 years
BMI, women, age [50 years
BMI, men
BMI
Weight change between age 25 years and baseline
BMI at age 25 years
Baseline waist-tohip ratio
Baseline BMI
Baseline obesity (BMI)
BMI
BMI
Exposure
Obesity at age 21 years
C16.3
1.00 1.18 (0.73–1.91) 1.63 (0.93–2.86)
B24 [27
[27 [24–27
1.83 (1.31–2.55) 1.97 (1.48–2.62)
[24–27
1.30 (1.11–1.53)
[27
1.00
1.31 (1.14–1.51) B24
1.00
B24 [24–27
1.39 (1.25–1.54)
2.05 (1.06–3.96)
6.8–16.2
Per 5 units
1.00 1.54 (0.77–3.08)
\6.8 kg
3.36 (2.09–5.41) 2.84 (1.33–6.09)
C30
1.00
\25 25–29.9
1.56 (0.91–2.69) 2.78 (1.65–4.70)
C0.968
1.00
\0.900 0.901–0.967
2.76 (1.40–5.44) 3.90 (1.95–7.82)
1.63 (0.84–3.18)
25–29.9 C35
1.00
\25 30–34.9
1.00 2.37 (1.53–3.68)
C30
C30
\30
1.62 (1.55–1.70) 2.34 (2.22–2.47)
25–29
0.66 (0.58–0.76)
Obesity at baseline
1.00
2.23 (1.82–2.73) 2.06 (1.38–3.08)
Per 5 units
15–19
1.53 (1.42–1.65)
C35
20–24
3.31 (2.52–4.36) 3.49 (2.35–5.19)
30–34.9
1.00 2.01 (1.59–2.56)
18.5–24.9
RR (95 % CI)
25–29.9
Description of categories
Age, hyperuricemia, hypertriglyceridemia, low HDL cholesterol, high blood pressure, hyperglycemia, renal insufficiency, cigarette smoking, alcohol, central obesity
Age, sex, hypertension, race, alcohol, estimated glomerular filtration rate
Age, menopausal status, race, diabetes, hypertension, diuretic use, alcohol, organ meat, estimated glomerular filtration rate
Age, sex, alcohol, number of general practitioner visits, chronic renal failure, hypertension, heart failure, ischemic heart disease, hyperlipidemia, diabetes, nephrolithiasis, psoriasis
Age, sex, beer, wine, liquor
Adjustment for confounders
Eur J Nutr (2014) 53:1591–1601 1595
123
1596
Eur J Nutr (2014) 53:1591–1601
3661 total records in PubMed and Embase 1386 records identified in the PubMed database 2275 records identified in the Embase databse 1134 duplicates excluded 2527 unique records
2484 excluded based on title or abstract
Body mass index and gout, per 5 units
A
Relative Risk (95% CI)
Study Chen, 2013
1.35 ( 1.21, 1.52)
Maynard, 2012
1.39 ( 1.25, 1.54)
Cea Soriano, 2011
1.51 ( 1.47, 1.56)
McAdams DeMarco, 2011
1.53 ( 1.42, 1.65)
Bhole, 2010
1.68 ( 1.43, 1.98)
Williams, 2008
2.29 ( 1.93, 2.82)
Choi, 2005
1.45 ( 1.26, 1.68)
Lin, 2000
1.05 ( 0.19, 2.70)
Roubenoff, 1991
1.76 ( 1.05, 2.93)
Campion, 1987
1.69 ( 1.22, 2.39) 1.55 ( 1.44, 1.66)
Overall
43 given detailed assessment .75
1
1.5
2
3
Relative Risk
32 articles excluded: 11 not relevant exposure, outcome or data 1 no risk estimates 1 duplicate 2
