Objective The aim of this study was to explore the risk of incident gout in patients with type 2 diabetes mellitus (T2DM) in association with diabetes duration, diabetes severity and antidiabetic drug treatment.
Methods We conducted a case-control study in patients with T2DM using the UK-based Clinical Practice Research Datalink (CPRD). We identified case patients aged ≥18 years with an incident diagnosis of gout between 1990 and 2012. We matched to each case patient one gout-free control patient. We used conditional logistic regression analysis to calculate adjusted ORs (adj. ORs) with 95% CIs and adjusted our analyses for important potential confounders.
Results The study encompassed 7536 T2DM cases with a first-time diagnosis of gout. Compared to a diabetes duration <1 year, prolonged diabetes duration (1–3, 3–6, 7–9 and ≥10 years) was associated with decreased adj. ORs of 0.91 (95% CI 0.79 to 1.04), 0.76 (95% CI 0.67 to 0.86), 0.70 (95% CI 0.61 to 0.86), and 0.58 (95% CI 0.51 to 0.66), respectively. Compared to a reference A1C level of <7%, the risk estimates of increasing A1C levels (7.0–7.9, 8.0–8.9 and ≥9%) steadily decreased with adj. ORs of 0.79 (95% CI 0.72 to 0.86), 0.63 (95% CI 0.55 to 0.72), and 0.46 (95% CI 0.40 to 0.53), respectively. Neither use of insulin, metformin, nor sulfonylureas was associated with an altered risk of incident gout.
Conclusions Increased A1C levels, but not use of antidiabetic drugs, was associated with a decreased risk of incident gout among patients with T2DM.
Statistics from Altmetric.com
Gout is a common painful inflammatory arthritis with acute onset, characterised by deposition of monosodium urate crystals in affected joints.1 ,2 The reported prevalence in the UK is about 1.4%.3 Increasing age and male gender,3 obesity,4 alcohol intake,5 and hyperuricaemia1 ,6 are the most important risk factors for gout. Congestive heart failure (CHF), chronic kidney disease (CKD), arterial hypertension and various drug treatments, such as different types of diuretics, are also associated with a markedly increased risk of gouty arthritis.7 Diabetes mellitus is a comorbid condition to CHF, CKD and arterial hypertension, which has also been associated with an increased risk of gout in several studies.3 ,8–11 However, confounding by these comorbidities and concomitant drug treatments was not routinely controlled in these studies3 ,8–11 which were mostly based on a limited number of patients.8–11 Of note, a recent observational study from the UK found a decreased gout risk in individuals with diagnosed type 2 diabetes mellitus (T2DM) as compared to diabetes-free subjects.12 Relative risk estimates decreased with increasing diabetes duration and were lower in treated than in untreated patients with diabetes mellitus.12 It could not be determined from the information provided whether the observed risks were further modified by certain antidiabetic treatments or by diabetes disease severity.12 We therefore conducted an observational study in patients with T2DM to explore the association between diabetes duration, diabetes severity and use of different types of antidiabetic drugs and the risk of developing incident gout in T2DM patients.
Patients and methods
The data were derived from the UK-based Clinical Practice Research Datalink (CPRD) which was established around 1987 and encompasses data from 450 general practices, representative of the UK population.13 ,14 The individuals enrolled in the database are representative of the UK population with regard to age, sex, geographic distribution and annual turnover rate.3 ,15 ,16 General practitioners have been trained to record medical information for research purposes using standard software and coding systems. The CPRD holds information regarding patient demographics and characteristics, lifestyle variables, such as Body Mass Index (BMI), smoking status, and alcohol consumption, symptoms, medical diagnoses, referrals to consultants and hospitalisations. The general practitioner generates drug prescriptions directly with the computer using a coded drug dictionary. The database has been described in detail elsewhere17 ,18 and has been validated extensively.13 ,19–22 The CPRD has been the source of numerous epidemiological studies published in peer-reviewed journals, including pharmacoepidemiological studies on gout using similar case definitions.3 ,7 ,23 ,24 Gout diagnoses are recorded with high validity in the CPRD.24 The Independent Scientific Advisory Committee (ISAC) for the Medicines and Healthcare products Regulatory Agency (MHRA) database research approved the study.
Using READ codes, we identified all patients aged 18 years or older with a first-time diagnosis of gout between 1 January 1990 and end June 2012 and with an antecedent history of T2DM based on READ codes; we refer to the date of the first gout diagnosis as the ‘index date’.
We excluded all cases with less than 3 years of recorded history in the database prior to the index date, and all patients with any recorded diagnosis of type 1 diabetes mellitus, cancer (except non-melanoma skin cancer), or a HIV infection. Additionally, to minimise the risk of misclassification of cases with incident gout, we excluded all patients with other conditions associated with joint inflammation, such as haemochromatosis, osteoarthritis, septic arthritis, or rheumatoid arthritis within 180 days preceding the index date, or within 90 days after the index date. Similar case definitions of gout have been used or validated in previous studies based on CPRD data.3 ,7 ,23 ,24
From the base population of patients with a history of T2DM, we identified at random one control patient with no evidence of gout, per case patient. Control patients were matched to cases on age (same year of birth), sex, BMI (max ±2.0 kg/m2), calendar time (same index date), general practice and number of years of active history in the CPRD prior to the index date. We applied the same exclusion criteria to control patients as to case patients.
The exposures of interest in this study were diabetes severity, duration and treatment. Diabetes severity was classified according to A1C levels recorded within 365 days prior to the index date (no A1C-level recorded, <7%, 7.0–7.9%, 8.0–8.9%, ≥9.0%). We defined diabetes duration as the time interval between the date of the first recorded diagnosis of T2DM and the index date by counting the number of days (<1, 1–3, 3–7, 7–10, >10 years). We assessed the use of different types of antidiabetic drugs based on prescriptions in the computer records. Cases and controls were classified as current users if their last prescription was within 180 days prior to the index date; or as past users if their last prescription was more than 180 days prior to the index date. We also classified them into the following groups: insulin users, metformin users, and sulfonylureas users. Due to the low number of users of other antidiabetic drugs, we did not assess their use separately. The duration of antidiabetic drug exposure was classified based on the number of recorded prescriptions for these drugs prior to the index date into 1–19 prescriptions, 20–39 prescriptions, 40–59 prescriptions, or ≥60 prescriptions.
We classified demographics and lifestyle factors, such as smoking status (non-smoker, current, past or unknown), alcohol consumption (never, current (1–9 units per week; 10–19 units per week; ≥20 units per week, unknown), past, unknown), and number of general practitioner visits ever (0–19, 20–39, ≥40) prior to the index date in cases and controls, and assessed these covariates as potential confounders. We further assessed for cases and controls whether they had arterial hypertension, CKD, CHF, ischaemic heart disease (IHD), transient ischaemic attack/stroke, dyslipidemia, or diabetes-related complications, such as diabetic nephropathy, diabetic neuropathy, or diabetic angiopathy at any time prior to the index date.
Furthermore, we assessed the independent effects of current (last prescription within 180 days prior to the index date) use of certain drugs of interest such as diuretics (loop, thiazide and thiazide-like, or potassium-sparing diuretics), β-blockers, ACE inhibitors (ACE-Is), angiotensin II receptor blockers (ARBs), calcium channel blockers (CCBs), organic nitrates, statins and low-dose acetylsalicylic acid (ASA) prior to the index date, and we assessed these covariates as potential confounders.
In a predefined sensitivity analysis, we additionally matched cases and controls on diabetes duration (±1 year) to ensure that cases and controls had an equal length of diabetes history and, therefore, equal exposure opportunity. To further address the impact of this potential exposure time-related bias on our results, we assessed diabetes duration across different exposure duration strata for all antidiabetic drugs under study. In another sensitivity analysis we assessed the risk of gout in association with A1C level stratified by diabetes duration to explore whether increasing A1C levels, irrespective of diabetes duration, were associated with an altered risk of gout. To assess whether diabetes duration or A1C levels are potential mediators of the association between antidiabetic drug use and the risk of gout, we additionally matched on diabetes duration and A1C (±1%). In another sensitivity analysis we only included patients with incident diabetes mellitus and incident antidiabetic drug use to address potential prevalent user bias. Finally, we restricted our analyses to cases and their matched controls who were treated with either non-steroidal anti-inflammatory drugs (NSAIDS), colchicine, or uricosuric or uricostatic drugs within 7, 30 and 90 days after the index date, respectively, to decrease the risk of possible outcome misclassification.
We conducted conditional logistic regression analyses using SAS statistical software V.9.3 (SAS Institute, Cary, North Carolina, USA) to calculate relative risk estimates as ORs with 95% CIs. We performed a χ2 test for trend (p<0.05) for diabetes duration and A1C levels. We explored the association between potential risk factors and the risk for gout in univariate analyses. We tested the effects of each of these potential confounders in multivariate analyses and included them in the final model if they altered the effect of antidiabetic drug use on the risk of gout by more than 10%. We a priori decided to adjust the analysis of antidiabetic drug use for smoking status, alcohol consumption, and number of general practitioner visits.
There were 7536 cases of incident gout in the T2DM study population, matched to 7536 diabetic, but gout-free controls. Of these, 62.5% were male, and the mean age (±SD) was 69.9±11.0 years. Alcohol consumption was associated with an increased risk of gout, while current smoking was associated with a decreased risk. Comorbidities, such as arterial hypertension, CKD, CHF and IHD were all associated with an increased risk of incident gout. The characteristics of cases and controls are displayed in table 1.
Current use of most antihypertensive drugs except CCBs, which was associated with a decreased adj. OR of 0.87 (95% CI 0.80 to 0.95), yielded increased relative risk estimates for incident gout (table 2).
The adjusted ORs for gouty arthritis in association with duration of T2DM of <1 (reference), 1–3, 3–7, 7–10 or >10 years were 0.91 (95% CI 0.79 to 1.04), 0.76 (95% CI 0.67 to 0.86), 0.70 (95% CI 0.61 to 0.81), and 0.58 (95% CI 0.51 to 0.66), respectively. Compared to a reference A1C level of <7%, the risk estimates of increasing A1C levels (7.0–7.9, 8.0–8.9 and ≥9%) steadily decreased with adj. ORs of 0.79 (95% CI 0.72 to 0.86), 0.63 (95% CI 0.55 to 0.72) and 0.46 (95% CI 0.40 to 0.53), respectively (table 3). Tests for trend were statistically significant for prolonged diabetes duration and increasing A1C levels. In the analysis where we stratified the A1C by diabetes duration, increasing A1C levels were associated with a decreasing relative risk estimate of gout irrespective of diabetes duration (table 4).
Compared to non-use, current use of insulin, metformin and sulfonylureas was associated with a decreased risk of developing gout in the main analysis; however, we did not observe a consistent duration effect (table 5). In this analysis, diabetes duration was closely similar (<3% difference) in cases and controls among corresponding exposure duration for all antidiabetic drugs studied (data not shown). In the sensitivity analysis where we additionally matched on diabetes duration, relative risk estimates were closely similar to the findings of the main analysis (table 6). Finally, when we additionally matched our analyses on A1C level, we did not observe decreased relative risk estimates for use of any antidiabetic drugs (table 6).
In the sensitivity analyses in which we included patients with incident diabetes mellitus and incident antidiabetic drug use only and in which we restricted the population to gout patients treated with NSAIDs, colchicine, uricosuric or uricostatic drugs, results did not materially differ from the main analysis (data not shown).
In this large observational study using the UK-based CPRD, we found a markedly decreased risk of incident gout among patients with increasing levels of A1C compared to patients with an A1C level <7%. The risk of gout was also decreased in association with increasing diabetes duration, a finding which is consistent with reported results of a recent population-based study in UK patients using the health improvement network database (THIN).12 Of interest, when we assessed the risk of gout in association with different A1C levels stratified by diabetes duration, increasing A1C levels irrespective of diabetes duration were associated with a decreased risk of gout. We are not aware of another observational study exploring the association between A1C as the surrogate marker for disease severity of T2DM and the risk of incident gout. Of note, Choi and Ford,25 using the US Third National Health and Nutrition Examination Survey (1988–1994), explored the association between A1C levels and serum uric acid levels, an important risk factor for gouty arthritis. Interestingly, they found that subjects with an A1C level of 6.0–6.9% had increased serum uric acid levels compared to patients with lower A1C levels. However, in patients with diagnosed diabetes mellitus and/or markedly increased A1C levels, they reported substantially decreased uric acid levels. The authors finally concluded that individuals with diabetes mellitus and markedly elevated A1C levels may be at a lower risk of hyperuricaemia and gout.25 Possible mechanistic explanations for this observed association include an uricosuric effect of glycosuria 25–27 or an impaired inflammatory response in patients with severe and long-lasting diabetes mellitus.1 By contrast with our findings, and arguing against the proposed mechanisms discussed above, several small observational studies suggested an increased risk of gout in patients with T2DM.8–11 Two studies from Taiwan 9 ,10 explored the association between gout and manifestations of the metabolic syndrome9 and trends in the manifestation of gout10 using a hospital-derived database. They found an increased risk of gout in patients with T2DM. However, the results were not adjusted for important risk factors of gout and comorbid conditions for T2DM.9 ,10 Comorbidities, such as arterial hypertension, CKD and CHF, as well as co-medications, such as different types of diuretics, have been associated with a substantially increased risk of gout in this study as well as in previous studies,7 ,28 and could, therefore, explain the positive association. In two questionnaire-based studies, one from Greece11 and one from New Zealand,8 the authors reported increased relative risks of gout in association with T2DM. Again, the results were not adjusted for important confounders. Finally, in a larger study using the General Practice Research Database (former name for the CPRD), diabetes mellitus was associated with a slightly increased risk for incident gout in univariate analysis.3 Again, no adjusted results for important confounders were reported in that study.
In our study, increasing A1C levels were associated with a significantly decreased risk of incident gout. While a uricosuric effect of poorly controlled hyperglycaemia and/or impaired inflammatory response in prolonged and poorly controlled diabetes offer potential mechanistic explanations for these findings,1 ,25–27 a potential effect of antidiabetic drug treatment on these risk estimates should also be considered. Rodriguez and coworkers12 did not report detailed results, but observed that relative risk estimates for incident gout were lower in patients with treated T2DM compared to patients who did not receive treatment. Whether this observation reflects a beneficial effect of antidiabetic drug treatment or is explained by more severe and/or prolonged diabetes mellitus remains unclear. In our study, we did not find evidence that different types of antidiabetic drugs alter the risk of gouty arthritis. Although we observed marginally decreased relative risk estimates of gout in current antidiabetic drug users in the main analysis, there was no consistent trend with increasing number of prescriptions. Finally, in the sensitivity analysis in which we explored the association between different antidiabetic drugs and gout, which was additionally matched on diabetes duration and A1C level, current antidiabetic drug treatment was no more associated with an altered risk of gout, irrespective of the specific drug used. Taken together, our results strongly suggest that increasing A1C levels, but not prolonged diabetes duration, per se, nor different antidiabetic drug treatments, explain the observed decreased relative risk estimates of gout.
This large population-based study has several strengths. First, the diagnosis of gout has been validated in this well-established primary care database,3 ,24 and similar case definitions for incident gout have been used by other authors.3 ,7 ,23 ,24 Second, we were in the position to study a large number of cases with T2DM and incident gout, to explore the role of duration and severity of T2DM, and to run various sensitivity analyses to address in-depth the potential role of different antidiabetic treatments on the risk of incident gout. Third, unlike in most former studies, we were able to address the role of important potential confounders such as age, sex and BMI by matching, and we adjusted our analyses for various important comorbidities and co-medications. Fourth, information in the CPRD is prospectively collected in the absence of any study hypothesis; therefore, recall bias is not an issue in this study. Fifth, exclusion of patients with less than 3 years of recorded history in the CPRD prior to the index date reduced the likelihood of including prevalent rather than incident gout cases. Finally, time-related biases, namely bias by different exposure opportunity (also named ‘time-window bias’) and immortal time bias were likely not an issue in this study in which we explored potential drug effects on the risk of incident gout. Furthermore, our findings were closely similar in the sensitivity analysis in which we matched cases and controls on diabetes duration.
Potential limitations of this study are possible misclassification of some gout cases, since diagnoses were mainly made by general practitioners, and not all by rheumatologists. However, a previous study has shown that gout diagnoses are recorded with high validity in the CPRD,24 and other investigators used similar case definitions.7 ,23 ,24 Furthermore, we excluded subjects with recorded differential diagnoses of gout, such as osteoarthritis, septic arthritis, arthropathy due to haemochromatosis, or rheumatoid arthritis to reduce the risk of misclassification. Furthermore, we were not able to adjust for all potential risk factors for gouty arthritis since dietary habits and physical activity 1 ,4 are not routinely recorded in the CPRD. However, by matching on BMI which is related to physical activity and dietary habits, we partly controlled for these risk factors. Additionally, we were not able to address potential confounding by socioeconomic status in-depth. However, we partially controlled for it by matching cases and controls on general practice attended, as it is likely that patients from the same neighbourhood see the same general practitioner. Finally, we were unable to assess race/ethnicity because this information is not consistently available in the CPRD. Our results are most likely representative of Caucasians, since 86% of individuals living in the UK are white.29
In summary, this large observational study provides evidence that increasing A1C levels are associated with a markedly decreased risk of incident gout in patients with T2DM. Neither use of insulin, metformin, nor sulfonylureas was associated with an altered risk of incident gout.
We thank Pascal Egger for programming and technical support.
Handling editor Tore K Kvien
Contributors All authors meet the criteria for authorship.
Funding This study was supported by an unconditional grant by the Senglet Foundation, Switzerland, but the work was independently done by the authors.
Competing interests None.
Ethics approval The Independent Scientific Advisory Committee (ISAC) for the Medicines and Healthcare products Regulatory Agency (MHRA) database research approved the study.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.