Secondary Logo

Journal Logo

Original Articles

The relationship between rheumatoid arthritis and diabetes mellitus: a systematic review and meta-analysis

Tian, Zixinga; Mclaughlin, Johna; Verma, Arpanab; Chinoy, Hectorc,,d,,e; Heald, Adrian H.a,,f

Author Information
Cardiovascular Endocrinology & Metabolism: June 2021 - Volume 10 - Issue 2 - p 125-131
doi: 10.1097/XCE.0000000000000244

Abstract

Introduction

Rheumatoid arthritis (RA) is an autoimmune and inflammatory disorder characterized by persistent synovial inflammation resulting in damage to cartilage and underlying bone [1]. The prevalence of this disease is 0.5–1.0% of adults in developed countries [2]. Accumulating evidence showed that inflammation emerges as a key factor in the onset and progression of diabetes [3]. Systemic inflammation associated with RA might contribute to the risk of developing diabetes in the future. Markers of active inflammation, such as CRP, are associated with an increased risk of diabetes in people with RA. Other traditional risk factors for type 2 diabetes mellitus (T2DM) are also highly prevalent among people with RA [4] and may contribute to the higher risk of diabetes. The prevalence of metabolic syndrome in people with RA is high [5]. As a result of chronic pain, swelling and stiffness of the joints, physical inactivity is common in people with RA, which in turn contributes to T2DM [6] through decreased calorie burning.

In summary, inflammation is implicated in the development of diabetes. Some traditional risk factors of T2DM are also highly prevalent among people with RA. Therefore, the incidence of diabetes could be higher in people with RA. Several observational studies investigated the relationship between RA and DM incidence; however, results have been conflicting [7,8]. A previous meta-analysis reviewed cohort studies (five eligible studies) before January 2014 and found that people with RA had a statistically significant increased risk of diabetes [relative risks (RR), 1.24; 95% confidence intervals (CI), 1.14–1.35] [9]. Since then, three further cohort studies have been published [10–12]. We aim to update the systematic review and meta-analysis.

Method

The meta-analysis was conducted according to the report Meta-analysis of Observational Studies in Epidemiology [13], and the protocol was registered with PROSPERO (CRD42020172084).

Search terms and strategy

The PICO/S tool was deployed to define the scope of the literature. We included cohort studies and nested case-control studies in adult humans (population) and compared the incidence of diabetes (outcome) between people with RA (intervention/exposure) and the general population (control).

A comprehensive search was conducted in Medline (via Ovid), Embase (via Ovid) and Web of Science core collection from inception to 10 March 2020. The search was limited to English language and human studies. The search terms were a combination of thesaurus terms and free-text terms, including RA, diabetes, cohort study or nested case-control study, and human. These were based on the search strategy developed for Medline but revised appropriately for the other two databases (Supplementary Contents 1–3, Supplemental digital content 1, http://links.lww.com/CAEN/A29). Search alerts were created in the databases to notify new studies according to the outlined search strategy. We also manually checked reference lists of the identified studies for potentially missing eligible studies. For eligible conference abstracts, we further searched the author to find if there were related full texts.

Inclusion and exclusion criteria

Two independent reviewers (A.H. and Z.T.) assessed the eligibility of studies according to the inclusion and exclusion criteria below. Disagreements were resolved by a third independent investigator (J.M.).

Inclusion: (1) cohort studies or nested case-control studies; (2) to compare general population with people with RA (case-control design); (3) to provide enough data to calculate RR and CIs; (4) to consider the incidence of diabetes mellitus as a specific outcome event.

Exclusion: (1) review studies, editorials/letters, case reports; (2) studies with the main outcome as gestational diabetes mellitus. If there was more than one publication from the same study population, only the study from the most recent publication was included.

Data extraction and quality assessment

For each eligible study, two researchers (A.H. and Z.T.) independently extracted the following items: (1) the first author, the year of publication, the country or region where the subjects lived; (2) characteristics of the participants, including sample size, sex and age; (3) study design and data source; (4) diagnosis criteria of RA and DM; (5) the number of outcomes in exposed and nonexposed cohorts; (6) crude and adjusted risk estimates, 95% CI; (7) adjustments. The quality of the included studies was assessed by the Newcastle–Ottawa scale [14], in which a study was judged based on section (four items, one star each), comparability (one item, up to two stars) and outcome (three items, one star each). In this meta-analysis, we graded quality as good (≥7 stars), fair (4–6 stars) and poor (<4 stars) [15]. We evaluated comparability by whether the studies had adequately adjusted for potential confounders (one star for age, one star for at least three of five confounders including a family history of diabetes, BMI or other measures of overweight/obesity, comorbidity, the use of glucocorticoids and ethnicity). The item ‘was follow-up long enough for outcomes to occur’ was evaluated by whether the study period was above 5 years.

Outcome measures

T2DM was defined when the subjects were described as having adult-onset, type II or noninsulin-dependent DM. DM is defined when the subjects were described as DM or composite DM.

Statistical analysis

The random-effects models were used to estimate the pooled RR according to heterogeneity. The I2 statistic and Q-test were used to test heterogeneity. PQ statistic ≥ 0.10 was considered to indicate no significant heterogeneity among the included studies. Publication bias was assessed using funnel plot, Egger’s test and Begg’s test. Sensitivity analyses were conducted by excluding one study at a time. Subgroup analyses were conducted based on the countries (or regions) of subjects and study designs (age and sex-matched or not). All analyses were performed using R, version 3.6.2 64 bit (R Core Team. R Foundation for Statistical Computing, Vienna, Austria) and Review Manager (RevMan), version 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark).

Results

The initial search provided 3669 articles. Forty-nine studies were included after a screen of titles and abstracts. Ultimately, five journal articles and two conference abstracts comprising 1 629 854 participants were included in our study (Fig. 1).

Fig. 1
Fig. 1:
Flowchart of study selection. RA, rheumatoid arthritis.

Most eligible studies were population-based, whereas one was hospital-based [16]. All the eligible studies were retrospective cohort in design. Sample sizes ranged from 1121 to 600 336, with a female predominance. The studies were conducted in diverse settings, including large electronic medical record databases [10–12,16], insurance claims [7,17] or a population-based RA inception cohort with a medical records linkage system [8]. Most of these studies used International Classification of Disease codes (ICD-9-CM 714.0) [7,16,17], diagnosis codes [10–12] and/or prescription of antirheumatic drugs [10,17] to identify people with RA. One study further reviewed RA cases identified by medical records based on the American Rheumatism Association 1987 classification criteria [8]. International Classification of Disease codes [7,12,16,17], diagnosis codes [10,11] and/or DM-specific drug [7,11,12,16] were used to identify the DM. Two studies also used blood glucose to define the incidence of DM during follow-up (Table 1) [8,16]. The quality of the studies was good (7–8 stars, Supplementary Content 4, Supplemental digital content 1, http://links.lww.com/CAEN/A29).

Table 1 - Characteristics of eligible studies
Reference Country/
region
Sample size Study period Percent of
female, %
Mean age, years Data source Crude or age/
sex-adjusted risk
Adjusted risk Adjustment
Solomon et al. [7] Canada RA 48 718 non-RA 442 033 1996–2006 RA 68 non-RA 60 RA 58 ± 16 non-RA 53 ± 17 Insurance programme of British Columbia NR HR, 1.5 (1.4–1.5) Age, sex, doctor visits, number of drugs, prior use of oral glucocorticoids, prior use of systemic immunosuppressive agents, prior use of topical glucocorticoids and comorbidity index
Schmidt et al. [12] Canada RA 26 013 non-RA 25 823 1996–2006–2010 RA 67 non-RA 67 RA 58.6 ± 17.2 non-RA 58.1 ± 17.1 Administrative health data NR HR 1.62 (1.49–1.76) Age, sex and comorbidity index (excluding DM).
Su et al. [17] Taiwan RA 3839 non-RA 596 497 1998–2009 RA 70.6 non-RA 49.9 >20a Taiwan national health insurance HR, 2.37 (2.15–2.60) HR, 2.40 (2.18–2.63) NR
Jafri et al. [10] UK 48 639 RA 364 988 non-RA 1994–2014 RA: 70.3 non-RA: 55.9 RA 60.73 ± 15.41 non-RA 51.33 ± 17.93 The health improvement network HR, 1.16 (1.12–1.21) HR, 1.21 (1.15–1.26) Age, sex, hypertension, hyperlipidaemia, smoking, BMI, heart disease, comorbidity index and healthcare utilization
Wilson et al. [11] UK RA 31 330 non-RA 31 464 1995–2015 RA 70.5 non-RA 70.5 RA 56.2 ± 16.0 non-RA 56.2 ± 16.0 UK Clinical Practice Research Datalink RR, 1.20 (1.11–1.29) NR NR
Gonzalez et al. [8] USA RA 559 non-RA 562 RA 15 non-RA 17b RA 73 non-RA 73 RA 58 ± NA non-RA 58 ± NA Rochester Epidemiology Project RR, 0.78 (0.57–1.06) NR NR
Mathew et al. [16] USA RA 1502 non-RA 7887 2001–2012 RA 70 non-RA 60 RA 55 ± NA non-RA 55 ± ±NA Electronic medical record database RR, 1.32 (1.13–1.54) HR, 1.14 (0.94–1.38) Age, sex, race, BMI, hypertension, hyperlipidaemia, ESR, number of office visits, number of drug classes, glucocorticoid and immunosuppressive agent and statin use in the year prior to cohort entry
DM, diabetes mellitus; ESR, erythrocyte sedimentation rate; HR, hazard ratio; NR, not reported; RA, rheumatoid arthritis; RR, risk ratio.
aAge range.
bMean year of follow-up, years.

The pooled RR from random-effects analysis for the association between RA and the risk of DM incidence is shown in Fig. 2, which was 1.23 (95% CI, 1.07–1.40). There was considerable between-study heterogeneity (I2 = 96%). Exploration of potential sources of heterogeneity found that there was no significant heterogeneity in different study designs (P = 0.30). However, there was significant heterogeneity in different countries or regions (P = 0.002). RA was not significantly associated DM in studies conducted in the United States of America (RR, 0.92, 95% CI, 0.52–1.63, Supplementary Content 5, Supplemental digital content 1, http://links.lww.com/CAEN/A29).

Fig. 2
Fig. 2:
Incidence of diabetes in people with rheumatoid arthritis compared with controls, calculated from a 2 × 2 table. RA, rheumatoid arthritis; CI, confidence interval; M–H random, Mantel–Haenszel random

Sensitivity analyses showed that the heterogeneity ranged from 84 to 96%, and confirmed that the association between RA and DM incidence was robust. We further performed a sensitivity analysis with studies reporting adjusted HR (Fig. 3), where the DM incidence was still higher in people with RA (HR, 1.52, 95% CI, 1.20–1.93). The funnel plot showed potential publication bias (Fig. 4), proven by Egger’s test (−3.15, P < 0.01), but not Begg’s test (−0.05, P = 1.00).

Fig. 3
Fig. 3:
Incidence of diabetes in people with rheumatoid arthritis compared with controls, generic inverse variance method. RA, rheumatoid arthritis; CI, confidence interval; M–H random, Mantel–Haenszel random.
Fig. 4
Fig. 4:
Publication bias in meta-analysis with funnel plot.

Discussion

Using both, calculation of pooled RR from a 2 × 2 table and adjusted HRs from original studies, we have confirmed an association between RA and DM, which was consistent across five included studies. Our finding strengthens the potential role of inflammation in DM on the basis of an increased incidence of this disease in people with RA.

Most eligible studies provided no distinction between different types of diabetes. Therefore we could not exclude the possibility that the data provided could represent the sum of all types of diabetes. However, considering T2DM accounts for more than 90% of all diabetes worldwide [18], and that the mean age of subjects in eligible studies was above 50 years old, it is reasonable to conclude that the papers in our meta-analysis predominantly described the link between RA and incidence of T2DM.

Accumulating evidence showed that inflammation emerges as a key factor in the onset of diabetes. The β-cell has a high density of interleukin-1β (IL-1β) receptors [19], and is susceptible to the toxic effects of innate mediators [20]. IL-1β, which is abundantly expressed in RA [21], induces the production of various cytokines and chemokines, including interleukin 6 (IL-6), interleukin-8, interleukin-33 and tumour necrosis factor-alpha (TNF-α). These mediators attract macrophages and other immune cells to the islet leading to β cell dysfunction and apoptosis, relative and absolute insulin deficiency and eventually DM [1]. Key pathways in the pathophysiology of RA include overproduction and overexpression of TNF-α and IL-6 [1]. TNF-α and IL-6 have a deleterious effect on insulin sensitivity in animal experiments, and are involved in inhibiting insulin signalling, thereby promoting insulin resistance [22], which links to T2DM.

Subgroup analysis showed heterogeneity was significant for the countries or regions of study. RA was not significantly associated with RA in studies conducted in the United States of America. The reason for this may due to a high incidence rate of DM in the control group in one study of this group [7].

Widdifield et al. [23] found only half of the people with RA were screened for diabetes risk factors, including lipid testing, glycosylated haemoglobin testing or BMI. The screening rates were similar between RA and non-RA groups. However, people with RA are at greater risk of developing DM. Thus we suggest that RA patients require enhanced screening to prevent and reduce diabetes morbidity.

We have found only two previous meta-analyses focused on the association between RA and DM [9,24]. The substudy of Boyer in 2011 included seven case-control or cross-sectional studies and found people with RA had a higher prevalence of DM (odds ratio, 1.74; 95% CI, 1.22–2.50). The second meta-analysis of Jiang, in 2015, compared the prevalence (11 studies) and incidence (five studies) of DM between people with RA and those without. Our meta-analysis has updated the relationship between RA and incidence of DM from three points as follows: (1) two of five eligible studies reporting incidence in Jiang’s article were from the same study population [8,25], so we only kept the most recent one in this study [8]; (2) we replaced another included study in Jiang’s with the study of Jafri et al. [10], which used the same data as the previous one; (3) our meta-analysis also included three new retrospective cohort studies; (4) we included both conference abstracts and full texts for potential publication bias.

Our study also has several limitations. We only included English articles. There were only seven eligible studies in our study, so the power of funnel plot and Egger test was too low to test publication bias. We could not perform further subgroup analyses to explore the heterogeneity due to the limited number of studies.

Conclusion

RA is associated with an increased risk of diabetes incidence. This finding supports the notion that inflammatory pathways are involved in the pathogenesis of diabetes. We suggest that more intensive screening and tighter management of DM risk factors should be considered in people with RA.

Acknowledgements

The views expressed in this publication are those of the authors and not necessarily those of the China Scholarship Council, the NHS, the National Institute for Health Research or the Department of Health.

Conflicts of interest

Z.T. is supported by China Scholarship Council. H.C. is supported by the National Institute for Health Research Biomedical Research Centre Funding Scheme. For the remaining authors, there are no conflicts of interest.

References

1. Scott DL, Wolfe F, Huizinga TW. Rheumatoid arthritis. Lancet. 2010; 376:1094–1108.
2. Batko B, Stajszczyk M, Swierkot J, Raciborski F, Wiland P. Prevalence and clinical characteristics of rheumatoid arthritis in Poland: first nationwide study. Arthritis and Rheumatology Conference: American College of Rheumatology/Association of Rheumatology Health Professionals Annual Scientific Meeting, ACR/ARHP; 2015; vol 67 (Suppl 10).
3. Donath MY, Dalmas É, Sauter NS, Böni-Schnetzler M. Inflammation in obesity and diabetes: islet dysfunction and therapeutic opportunity. Cell Metab. 2013; 17:860–872.
4. Blum A, Adawi M. Rheumatoid arthritis (RA) and cardiovascular disease. Autoimmun Rev. 2019; 18:679–690.
5. Chung CP, Oeser A, Solus JF, Avalos I, Gebretsadik T, Shintani A, et al. Prevalence of the metabolic syndrome is increased in rheumatoid arthritis and is associated with coronary atherosclerosis. Atherosclerosis. 2008; 196:756–763.
6. Chatterjee S, Khunti K, Davies MJ. Type 2 diabetes. Lancet. 2017; 389:2239–2251.
7. Solomon DH, Love TJ, Canning C, Schneeweiss S. Risk of diabetes among patients with rheumatoid arthritis, psoriatic arthritis and psoriasis. Ann Rheum Dis. 2010; 69:2114–2117.
8. Gonzalez A, Maradit Kremers H, Crowson CS, Ballman KV, Roger VL, Jacobsen SJ, et al. Do cardiovascular risk factors confer the same risk for cardiovascular outcomes in rheumatoid arthritis patients as in non-rheumatoid arthritis patients? Ann Rheum Dis. 2008; 67:64–69.
9. Jiang P, Li H, Li X. Diabetes mellitus risk factors in rheumatoid arthritis: a systematic review and meta-analysis. Clin Exp Rheumatol. 2015; 33:115–121.
10. Jafri K, Bartels CM, Shin D, Gelfand JM, Ogdie A. Incidence and management of cardiovascular risk factors in psoriatic arthritis and rheumatoid arthritis: a population-based study. Arthritis Care Res. 2017; 69:51–57.
11. Wilson JC, Sarsour K, Gale S, Pethö-Schramm A, Jick SS, Meier CR. Incidence and risk of glucocorticoid-associated adverse effects in patients with rheumatoid arthritis. Arthritis Care Res. 2019; 71:498–511.
12. Schmidt T, Avina-Zubieta A, Saure E, Abrahamowicz M, Esdaile J, Lacaille D. Incidence of diabetes mellitus in rheumatoid arthritis: a population based cohort study. Journal of Rheumatology Conference: 70th Annual Meeting of the Canadian Rheumatology Association, CRA; 2015; Vol 42.
13. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000; 283:2008–2012.
14. Wells GA, Shea B, O'Connell D, Peterson J, Welch V, Losos M, Tugwell P. The Newcastle-ottawa scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses.
15. Huang Y, Cai X, Mai W, Li M, Hu Y. Association between prediabetes and risk of cardiovascular disease and all cause mortality: systematic review and meta-analysis. BMJ. 2016; 355:i5953.
16. S M, X T, K HL, W MCM, A B. Rheumatoid and psoriatic arthritis are not associated with higher risk of incident diabetes mellitus. Arthritis and Rheumatism Conference: American College of Rheumatology/Association of Rheumatology Health Professionals Annual Scientific Meeting, ACR/ARHP; 2013; Vol 65 (Suppl 10).
17. Su CC, Chen IC, Young FN, Lian IB. Risk of diabetes in patients with rheumatoid arthritis: a 12-year retrospective cohort study. J Rheumatol. 2013; 40:1513–1518.
18. Federation ID. IDF Diabetes Atlas. 9th ed. 2019.
19. Mandrup-Poulsen T, Pickersgill L, Donath MY. Blockade of interleukin 1 in type 1 diabetes mellitus. Nat Rev Endocrinol. 2010; 6:158–166.
20. Donath MY, Dinarello CA, Mandrup-Poulsen T. Targeting innate immune mediators in type 1 and type 2 diabetes. Nat Rev Immunol. 2019; 19:734–746.
21. McInnes IB, Schett G. The pathogenesis of rheumatoid arthritis. N Engl J Med. 2011; 365:2205–2219.
22. Moller DE. Potential role of TNF-alpha in the pathogenesis of insulin resistance and type 2 diabetes. Trends Endocrinol Metab. 2000; 11:212–217.
23. Widdifield J, Ivers NM, Bernatsky S, Jaakkimainen L, Bombardier C, Thorne JC, et al. Primary care screening and comorbidity management in rheumatoid arthritis in Ontario, Canada. Arthritis Care Res (Hoboken). 2017; 69:1495–1503.
24. Boyer JF, Gourraud PA, Cantagrel A, Davignon JL, Constantin A. Traditional cardiovascular risk factors in rheumatoid arthritis: a meta-analysis. Joint Bone Spine. 2011; 78:179–183.
25. Doran MF, Crowson CS, Pond GR, O’Fallon WM, Gabriel SE. Frequency of infection in patients with rheumatoid arthritis compared with controls: a population-based study. Arthritis Rheum. 2002; 46:2287–2293.
Keywords:

diabetes mellitus; incidence; inflammation; meta-analysis; rheumatoid arthritis

Supplemental Digital Content

Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc.