Contribution of obesity to the rise in incidence of rheumatoid arthritis

Authors


Abstract

Objective

Obesity is an underrecognized risk factor for rheumatoid arthritis (RA). In recent years, both the prevalence of obesity and the incidence of RA have been rising. Our objective was to determine whether the “obesity epidemic” could explain the recent rise in the incidence of RA.

Methods

An inception cohort of Olmsted County, Minnesota residents who fulfilled the 1987 American College of Rheumatology criteria for RA in 1980–2007 was compared to population-based controls (matched on age, sex, and calendar year). Heights, weights, and smoking statuses were collected from medical records. Obesity was defined as a body mass index ≥30 kg/m2. Conditional logistic regression was used to assess the influence of obesity on developing RA. Population attributable risk was used to estimate the incidence of RA in the absence of obesity.

Results

The study included 813 patients with RA and 813 controls. Both groups had extensive medical history available prior to the incidence/index date (mean 32.2 years), and ∼30% of each group was obese at the incidence/index date. The history of obesity was significantly associated with developing RA (odds ratio 1.24, 95% confidence interval 1.01–1.53; adjusted for smoking status). In 1985–2007, the incidence of RA rose by an increment of 9.2 per 100,000 among women. Obesity accounted for 4.8 per 100,000 (52%) of this increase.

Conclusion

Obesity is associated with a modest risk for developing RA. Given the rapidly increasing prevalence of obesity, this has had a significant impact on RA incidence and may account for much of the recent increase in the incidence of RA.

INTRODUCTION

Obesity is a controversial risk factor for rheumatoid arthritis (RA). A link between obesity and RA is plausible, since biologic mechanisms of inflammation are present in adipose tissue, and these may be linked to chronic systemic inflammation (1). In fact, important advances in the understanding of adipokines have elucidated their crucial role as mediators of inflammation and immune response, which are implicated in the pathophysiology of rheumatic diseases, such as RA (2–4). While several studies have examined the potential influence of obesity on the development of RA, the results have been inconsistent. Two large US studies found no association between obesity and the risk of RA, while 2 European studies found significant, sizable risks for RA associated with obesity (i.e., odds ratios [ORs] of 1.6 and 3.7) (5–8). Reasons for the inconsistencies include the low prevalence of obesity in studies performed >15 years ago and a lack of power to detect a modest risk for RA associated with obesity. If obesity confers even a modest risk for the development of RA, it could have a large impact on the incidence of this disease, given the epidemic increases in the prevalence of obesity in recent years (9, 10).

Following a 4-decade period of decline, the incidence of RA has been on the rise since 1995 (11, 12). The cause of this recent rise in incidence is unknown. It is most likely that changes in environmental risk factors account for the rising incidence because genetic factors do not change so quickly in populations. The list of potential environmental influences on the development of RA is extensive, but obesity deserves special consideration due to its recent dramatic increase in prevalence (9, 10). The aim of this study was to assess the influence of obesity on the risk of developing RA, and to determine whether obesity could explain the increasing incidence of RA.

SUBJECTS AND METHODS

Study design.

This retrospective, population-based study of residents of Olmsted County, Minnesota was conducted using the resources of the Rochester Epidemiology Project, a population-based medical records linkage system that allows ready access to the complete (inpatient and outpatient) medical records from all community medical providers (13). An incidence cohort of all residents of Olmsted County, Minnesota ages ≥18 years who first fulfilled the 1987 American College of Rheumatology classification criteria for RA between January 1, 1980 and December 31, 2007 was identified (11, 14). For each patient, the earliest date of fulfillment of ≥4 American College of Rheumatology criteria for RA was considered the RA incidence date. For each patient with RA, a control of the same age (±1 year), sex, and calendar year without RA on the incidence date of the patient with RA was randomly selected from the Olmsted County population. The index date for each control was defined as the RA incidence date of the corresponding patient with RA.

The presence of risk factors at the RA incidence/index date was ascertained by medical record review, including smoking status (never, current, or former) and body mass index (BMI). Obesity was defined as a BMI ≥30 kg/m2, and for those patients who were not obese at the RA incidence/index date, the history of obesity prior to the RA incidence/index date was also assessed. For these patients, the earliest date of obesity during adulthood (age ≥18 years) was collected. For the patients with RA, the results of all rheumatoid factor (RF) and anti–citrullinated protein antibody (ACPA) tests were also collected. The Institutional Review Boards of the Mayo Clinic and the Olmsted Medical Center approved this study.

Statistical methods.

Descriptive statistics (means, proportions, etc.) were used to summarize the characteristics of the cases and controls. Conditional logistic regression models were used to examine the association between potential risk factors (obesity and smoking) and case/control status.

Population attributable risk (PAR) is defined as the proportion of disease that could be prevented by eliminating a causal risk factor. It is a function of the magnitude of the effect and the prevalence of the risk factor, which is estimated as PAR = P(F)(RR − 1)/[P(F)(RR − 1) + 1] using a formula by Levin, where P = probability, F = risk factor, and RR = relative risk of the risk factor for the development of RA (15). The OR obtained from conditional logistic regression was used as an estimate of RR.

Overall incidence rates for each sex were age adjusted to the US white population in the year 2000. Age- and sex-specific incidence rates were calculated by using the number of incident cases as the numerator and population estimates based on decennial census counts as the denominator, with linear interpolation used to estimate population size for intercensal years (16). Trends in incidence rates were examined using Poisson regression methods with smoothing splines for age and calendar year (17). The age-, sex-, and calendar year–specific PAR values were used to estimate the proportion of incidence cases attributable to obesity in each year. The annual incidence of RA in the absence of obesity was estimated by removing these cases from the numerator.

RESULTS

The study population included 813 patients who fulfilled the criteria for RA in 1980–2007 and 813 controls without RA. The mean age of both groups was 55.9 years at the incidence/index date and 68% were women. Among the patients with RA, 67% were RF and/or ACPA positive. Both groups had extensive medical history available prior to the incidence/index date (mean 32.2 years). Patients with RA were more likely to be current or former smokers compared to controls (P = 0.002). A history of obesity was somewhat more common in cases than in controls (40% versus 36%; P = 0.052), whereas obesity at the incidence/index date was similar in cases and controls (30% versus 28%; P = 0.41) (Table 1). Among patients with a history of obesity who were not obese at the incidence/index date (84 cases and 61 controls), the first recorded obesity was a median of 17.1 years prior to the incidence/index date (25th, 75th percentiles 5.4, 31.0 years), but the length of time the patients were obese was not collected. However, the cases had a higher BMI at the incidence/index date on average (mean 28.2 versus 27.2 kg/m2; P = 0.004), with 95% of cases and 85% of controls categorized as overweight at the incidence/index date.

Table 1. Characteristics of 813 patients with incident RA in 1980–2007 and 813 controls without RA*
 RA (n = 813)Control (n = 813)P
  • *

    Values are the number (percentage) unless otherwise indicated. RA = rheumatoid arthritis; BMI = body mass index.

Age, mean ± SD years55.9 ± 15.755.9 ± 15.70.99
Women556 (68)556 (68)1.0
Length of medical record prior to the index date, mean ± SD years32.2 ± 17.532.2 ± 17.90.96
Smoking at the index date  0.002
 Current178 (22)144 (18) 
 Former271 (33)234 (29) 
Obesity (BMI ≥30 kg/m2) at the index date244 (30)229 (28)0.41
Obesity (BMI ≥30 kg/m2) at or prior to the index date328 (40)290 (36)0.052

Conditional logistic regression models revealed that history of obesity was a significant risk factor for the development of RA (OR 1.24, 95% confidence interval [95% CI] 1.00–1.52; P = 0.046), whereas obesity at the incidence/index date did not reach statistical significance (OR 1.10, 95% CI 0.88–1.38). The history of obesity remained significant after adjustment for smoking status (OR 1.24, 95% CI 1.01–1.53; P = 0.041). An interaction between smoking status and obesity was not apparent (P > 0.6). The association between history of obesity and development of RA was similar for both sexes. The association between history of obesity and development of RA was somewhat stronger among patients ages <60 years (OR 1.32, 95% CI 1.00–1.72) than those ages ≥60 years at the incidence/index date (OR 1.15, 95% CI 0.82–1.61). The association between history of obesity and development of RA was marginally stronger among RF/ACPA–positive cases (OR 1.28, 95% CI 0.99–1.65) than among RF/ACPA–negative cases (OR 1.17, 95% CI 0.80–1.71). Adjusting the models for history of low BMI (<20 kg/m2) and overweight (BMI 25–29 kg/m2) in order to compare patients with a history of obesity to patients with a normal BMI revealed little change in the association between history of obesity and development of RA (OR 1.31, 95% CI 1.04–1.65).

Table 2 shows the prevalence of obesity (and history of obesity) in the cases and controls across the study period according to age and sex. These estimates demonstrate dramatic increases in the prevalence of obesity over time across all age groups and both sexes. These findings closely reflect national trends (9).

Table 2. Prevalence of obesity in patients with and without RA according to sex, age, and calendar year of the index date*
 1980–19891990–19992000–2007
  • *

    Values are the percentage of patients with obesity at the index date/percentage of patients with history of obesity prior to the index date. RA = rheumatoid arthritis.

RA   
 Female sex   
  Age 18–39 years8/1512/2432/38
  Age 40–59 years19/3334/4154/62
  Age ≥60 years13/3723/3335/46
 Male sex   
  Age 18–39 years22/2210/1023/46
  Age 40–59 years32/3232/3840/42
  Age ≥60 years17/3136/4833/51
Non-RA   
 Female sex   
  Age 18–39 years19/2724/3224/24
  Age 40–59 years19/2133/3946/54
  Age ≥60 years10/2216/3844/46
 Male sex   
  Age 18–39 years11/110/1042/42
  Age 40–59 years15/2520/2035/40
  Age ≥60 years14/2627/3349/65

Using the estimated OR of 1.24 for history of obesity, the proportion of the risk of RA that can be attributed to obesity (PAR) was 1.2%, 5.7%, and 10.7% for an obesity prevalence of 5%, 25%, and 50%, respectively.

Figure 1 shows the annual incidence of RA per 100,000 population in residents of Olmsted County, Minnesota in 1980–2007 according to sex. The age-, sex-, and calendar year–specific PARs were used to estimate the annual incidence of RA in the absence of obesity. For men, obesity had little effect on the incidence of RA, since the actual age-adjusted incidence in 2007 among men was 26.9 per 100,000 population and the estimated incidence in the absence of obesity was 24.6 per 100,000 population. Therefore, among men, the proportion of the risk of RA that can be attributed to obesity was only 2.3 per 100,000 population.

Figure 1.

Annual incidence of rheumatoid arthritis per 100,000 population (pop) in residents of Olmsted County, Minnesota in 1980–2007 according to sex. Solid lines show the trends in the incidence adjusted to the US white 2000 population. Broken lines show the estimated incidence in the absence of obesity.

Among women, the incidence of RA in the absence of obesity in 1985 (45.4 per 100,000 population) is very similar to the actual incidence rate in 1985 (46.6 per 100,000 population), due to the low prevalence of obesity in that year (15%). There is, however, a wide discrepancy between the actual incidence of RA among women and the estimated incidence in the absence of obesity increases in more recent years. The actual age-adjusted RA incidence in 2007 among women was 55.8 per 100,000 population and the estimated incidence of RA in the absence of obesity was 49.8 per 100,000 population. Hence, the increase in RA incidence from 1985 to 2007 was 9.2 per 100,000 (from 46.6 to 55.8 per 100,000 population) and the increase in RA incidence in the absence of obesity was only 4.4 per 100,000 (from 45.4 to 49.8 per 100,000 population). Therefore, among women, the proportion of the risk of RA that can be attributed to obesity is 4.8 per 100,000. In other words, obesity could explain 52% of the increase in the incidence of RA among women observed in 1995–2007.

Sensitivity analyses were performed to examine the variability in the estimated influence of obesity on the incidence of RA among women. In this cohort, the estimated OR for obesity was 1.24 (95% CI 1.01–1.53). If the RR for obesity on the development of RA was only 1.01, then obesity could only explain 18% of the increase in RA incidence among women. However, if the true RR of obesity for the development of RA was as large as 1.50, then obesity could explain 82% of the increase in RA incidence among women.

Because smoking is an environmental factor linked to the development of RA, we examined the potential influence of smoking on the rising incidence of RA. Both current (OR 1.48, 95% CI 1.14–1.94) and former (OR 1.40, 95% CI 1.11–1.76) smoking were found to be significant risk factors for RA. While the prevalence of current smoking declined in both sexes and all age groups, particularly between 1980 and 1995, the prevalence of former smoking increased during the same time period. In the non-RA cohort, 50% of subjects in 1980–1989 had ever smoked (26% were current smokers and 24% were former smokers). In 2000–2007, 47% had ever smoked (only 17% were current smokers and 30% were former smokers). Thus, the prevalence of ever smoking remained relatively constant over time. Therefore, smoking cannot explain the rising incidence of RA among women.

DISCUSSION

The prevalence of obesity has risen dramatically in recent years. History of obesity is a significant risk factor for the development of RA, even after adjustment for smoking status. Our findings indicate that obesity could explain 52% of the recent rise in the incidence of RA. Although smoking is a stronger risk factor for RA in our cohort (OR >1.4), smoking cannot explain the increase in the incidence of RA because the prevalence of smoking has not increased in recent years. While the relative effect of obesity on the development of RA did not appear to differ in men versus women and the prevalence of obesity was similar (and in some age groups, higher) in men versus women, the absolute impact of obesity on the trends in the incidence of RA was smaller in men than in women due to the lower incidence of RA among men.

Several well-conducted studies have examined the potential effect of obesity on the development of RA, with inconsistent results (Table 3). Symmons et al reported a large effect of obesity on the development of RA in Norfolk, UK (OR 3.74) (8). Significant effects with an OR for obesity of approximately 1.5 were reported by Pedersen et al and Voigt et al (7, 18). In contrast, no effect of obesity was found in the Nurses' Health Study, the Iowa Women's Health Study, or the UK General Practice Research Database (5, 6, 19). The bulk of the evidence seems to point toward a modest increased risk of RA associated with obesity, likely with an OR <1.5, as reported in our study. Therefore, the latter studies that were conducted using data from 1976–1997, at a time when the prevalence of obesity was approximately 10%, may have been underpowered to detect a modest risk (e.g., OR <1.5) associated with a risk factor with low prevalence. In fact, in order for a study to have 80% power to detect an association with an OR of 1.5 for a risk factor with a prevalence of 10%, a sample size of 957 cases and 957 controls would be needed. Given the rapidly increasing prevalence of obesity, a reexamination of obesity as a risk factor for RA is warranted.

Table 3. Obesity and the risk of RA*
Author, year (ref.)Study population (year)Sample sizeStudy designPrevalence of obesityResults
  • *

    RA = rheumatoid arthritis; RR = relative risk; 95% CI = 95% confidence interval; BMI = body mass index; OR = odds ratio; ACPA = anti–citrullinated protein antibody; GPRD = General Practice Research Database.

Hernandez Avila et al, 1990 (5)Nurses' Health Study (1976–1984)217 RA among 116,779 womenProspective cohort studyNot reportedRR 1.1 (95% CI 0.7–1.9) for BMI >29 kg/m2 vs. BMI <21 kg/m2
Voigt et al, 1994 (18)King County, WA (1986–1991)349 RA women, 1,457 controlsCase–control25% with BMI >25.83 kg/m2OR 1.5 (95% CI 1.0–2.3) among premenopausal women and OR 1.4 (95% CI 0.7–2.6) among postmenopausal women for BMI >25.83 kg/m2 vs. BMI <20.43 kg/m2
Symmons et al, 1997 (8)Norfolk, UK (1994–1995)165 cases, 178 controlsCase–control15% cases, 8% controlsOR 3.74 (95% CI 1.14–12.27) for BMI ≥30 kg/m2 vs. BMI <25 kg/m2, adjusted for smoking and social class
Cerhan et al, 2002 (6)Iowa Women's Health Study (1986–1997)158 RA among 31,336 women ages 55–69 yearsProspective cohort study25% with BMI >29.2 kg/m2RR 1.01 (95% CI 0.65–1.56) for BMI >29.2 kg/m2 vs. BMI <23.4 kg/m2
Pedersen et al, 2006 (7)Denmark (1998–2003)515 cases, 769 controlsCase–control9% cases, 7% controlsOR 1.57 (95% CI 1.01–2.44) for BMI ≥30 kg/m2 vs. BMI 18.5 to <25 kg/m2 ACPA positive: OR 1.34 (95% CI 0.79–2.28) ACPA negative: OR 3.22 (95% CI 1.73–5.98)
Rodriguez et al, 2009 (19)UK GPRD (1996–1997)559 cases, 4,234 controlsCase–control9.8%OR 0.95 (95% CI 0.68–1.34) for BMI ≥30 kg/m2 vs. BMI 20 to <25 kg/m2

The inconsistent results of the impact of obesity on the development of RA stem from its modest effect, which has led to underrecognition of obesity as a risk factor for RA. In fact, obesity has been nearly forgotten in recent years, as several review articles listing environmental risk factors for RA have not mentioned it (20, 21). Articles speculating on potential causes for the recent increase in the incidence of RA have made only passing mentions of obesity, while addressing an extensive list of potential environmental influences on the development of RA, including cigarette smoking, use of oral contraceptives, breastfeeding, vitamin D insufficiency, infections, immunizations, socioeconomic status, occupational exposures, alcohol use, and air pollution, among others.

Genetic factors are unlikely to explain the recent increase in RA incidence because they change in populations over centuries rather than decades. Therefore, environmental factors, or possibly epigenetic factors, are most likely responsible for the recent changes in the incidence of RA. The strongest environmental factors influencing the development of RA are arguably cigarette smoking and use of oral contraceptives. However, the prevalence of these factors has been stable in recent years, so they are unlikely to explain the recent rise in RA incidence.

The mechanism by which obesity may lead to RA is unknown, but several mechanisms have been postulated. One plausible mechanism is the association between obesity and chronic inflammation. The amount of adipose tissue increases during weight gain, and adipocytes produce adipocytokines and inflammatory cytokines, including adiponectin, leptin, tumor necrosis factor, interleukin-6, C-reactive protein, and many others (1, 22). The major adipocytokines have immune-modulatory properties and impact inflammation (2–4). This is an active area of research, and both adipocytokines and inflammatory cytokines are implicated in the pathophysiology of rheumatic diseases, such as RA. A second possibility involves the association between obesity and vitamin D deficiency. Obese patients often have vitamin D deficiency (23, 24). Merlino et al reported that vitamin D intake was inversely associated with the development of RA (25). Others have found associations between vitamin D deficiency and the development of autoimmune diseases, such as multiple sclerosis and type 1 diabetes mellitus, but these results have not been consistent across studies (26, 27). Third, the relationship between obesity and sex hormones could also influence the development of RA. Obese men and women have higher levels of estrogens and androgens (28–30). Given the sex bias in RA incidence and the protective effect of oral contraceptives, sex hormones are thought to play a role in the development of RA, which might be modified by obesity (31). Furthermore, estrogen levels are positively correlated with adipocytokines (32). Finally, obesity has been linked to development of psoriatic arthritis, and Soltani-Arabshahi et al postulate that the link between obesity and autoimmune diseases could be driven by a genetic variation that predisposes patients to both conditions (33). Therefore, while there are many plausible connections between obesity and RA, further research is needed to understand the complex mechanisms whereby obesity influences the development of RA.

One question of interest is whether obesity comes first or whether the mechanisms that cause RA are already in place when obesity develops. This question is difficult to answer. However, in our study, obesity at the incidence/index date was not significantly associated with development of RA, despite small differences in patients fulfilling the definition for history of obesity versus obesity at the incidence/index date (10% of RA patients and 8% of control subjects met the definition of history of obesity, but were not obese at the incidence/index date). This finding, along with the potential associations between obesity and RA discussed above, would suggest that obesity manifests prior to the development of RA. However, the possibility of a common antecedent to both obesity and RA cannot be excluded. Therefore, further research is needed to address this question.

Furthermore, several studies have reported a protective effect of increasing body mass for radiographic joint damage and mortality in RA (34–38). These findings may seem to contradict our findings of an association between obesity and the development of RA. However, patients with RA may experience weight loss associated with their disease severity, which can lead to rheumatoid cachexia.

Our study has several strengths, including its population-based design and standardized approach to case identification, as well as the availability of extensive medical history prior to the RA incidence/index date. The population of Olmsted County, Minnesota is 90% white, suggesting that the results of our study may not be generalizable to other, more racially diverse populations. However, our trends in the prevalence of obesity over time were remarkably similar to those in the US population reported by the National Health and Nutrition Examination Survey (9, 39). In addition, as with any observational study, our study precludes determination of causality. However, randomizing patients to obesity or not is infeasible. Therefore, we cannot exclude the possibility that our findings of an association between obesity and the development of RA may not be causal due to confounding by another related factor. For instance, altered energy expenditure resulting from loss of muscle mass and increase in fat mass, referred to as rheumatoid cachexia, may play a role. In addition, BMI may not accurately reflect abdominal adiposity, so a measure of abdominal fat mass, such as waist circumference, would have been useful to further elucidate the relationship between adiposity and the development of RA. However, a measure of abdominal fat mass was not available in this retrospective study because this information is not routinely collected during clinical care. Furthermore, data on the length of time that patients were obese were not collected. However, the majority of patients with a history of obesity who were no longer obese at the incidence/index date remained overweight.

In conclusion, obesity is associated with a modest risk for developing RA. Given the recent rapid increase in the prevalence of obesity, this risk factor appears to have a significant impact on the incidence of RA and may account for a large proportion of the recent increase in incidence of RA among women. These findings suggest that unless the obesity epidemic is controlled, the incidence and prevalence of RA will continue to rise, leading to increasing demands for rheumatologic care.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Gabriel 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. Crowson, Matteson, Gabriel.

Acquisition of data. Gabriel.

Analysis and interpretation of data. Crowson, Matteson, Davis, Gabriel.

Ancillary