Use of Diuretics and Risk of Incident Gout: A Population-Based Case–Control Study

Authors

  • Saskia Bruderer,

    1. University of Basel and University Hospital Basel, Basel, Switzerland
    Search for more papers by this author
  • Michael Bodmer,

    1. University of Basel, Basel, Switzerland, and Bern University Hospital, Inselspital, Bern, Switzerland
    Search for more papers by this author
  • Susan S. Jick,

    1. Boston Collaborative Drug Surveillance Program, Lexington, Massachusetts, and Boston University School of Public Health, Boston, Massachusetts
    Search for more papers by this author
  • Christoph R. Meier

    Corresponding author
    1. University of Basel and University Hospital Basel, Basel, Switzerland
    2. Boston Collaborative Drug Surveillance Program, Lexington, Massachusetts, and Boston University School of Public Health, Boston, Massachusetts
    • Basel Pharmacoepidemiology Unit, University Hospital Basel, Spitalstrasse 26, CH-4031 Basel, Switzerland. E-mail: Christoph.Meier@usb.ch

    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Error in Attribution of Data Source of a Cited Study in the Article by Bruderer et al (Arthritis Rheumatol, January 2014) Volume 66, Issue 2, 427, Article first published online: 27 January 2014

Abstract

Objective

Use of diuretics has been associated with an increased risk of gout. Data on different types of diuretics are scarce. We undertook this study to investigate the association between use of loop diuretics, thiazide or thiazide-like diuretics, and potassium-sparing agents and the risk of developing incident gout.

Methods

We conducted a retrospective population-based case–control analysis using the General Practice Research Database established in the UK. We identified case patients who were diagnosed as having incident gout between 1990 and 2010. One control patient was matched to each case patient for age, sex, general practice, calendar time, and years of active history in the database. We used conditional logistic regression to calculate odds ratios (ORs) and 95% confidence intervals (95% CIs), and we adjusted for potential confounders.

Results

We identified 91,530 incident cases of gout and the same number of matched controls. Compared to past use of diuretics from each respective drug class, adjusted ORs for current use of loop diuretics, thiazide diuretics, thiazide-like diuretics, and potassium-sparing diuretics were 2.64 (95% CI 2.47–2.83), 1.70 (95% CI 1.62–1.79), 2.30 (95% CI 1.95–2.70), and 1.06 (95% CI 0.91–1.23), respectively. Combined use of loop diuretics and thiazide diuretics was associated with the highest relative risk estimates of gout (adjusted OR 4.65 [95% CI 3.51–6.16]). Current use of calcium channel blockers or losartan slightly attenuated the risk of gout in patients who took diuretics.

Conclusion

Use of loop diuretics, thiazide diuretics, and thiazide-like diuretics was associated with an increased risk of incident gout, although use of potassium-sparing agents was not.

Gout is a painful, inflammatory, acute-onset arthritis, characterized by deposition of monosodium urate monohydrate crystals in affected joints ([1, 2]). The disease is common in Western countries, with a reported prevalence of ∼1.4% in the overall UK population ([3, 4]). Gout predominantly affects men >40 years of age, and prevalence increases as individuals age ([3]). Other risk factors include obesity ([5]) and alcohol intake ([6]).

Most patients with gouty arthritis have hyperuricemia, which is considered an important risk factor ([2, 7]). Hyperuricemia mainly results from decreased uric acid excretion or increased uric acid reabsorption in the kidneys. Diuretics, including loop diuretics, most thiazide diuretics, and potassium-sparing agents (such as spironolactone or eplerenone), have been linked to hyperuricemia via a presumed mechanism of decreased renal uric acid excretion or increased uric acid reabsorption ([8-10]). Volume contraction and direct effects on urate transporters in the proximal tubule have been proposed as potential explanations ([10]). Studies have demonstrated that hydrochlorothiazide increases urate absorption by inhibition of organic anion transporter 4 ([11]), and the reduction of urate secretion by multidrug resistance protein 4 (similar to furosemide) has also been seen ([9]). Since diuretics predispose individuals to hyperuricemia, use of these drugs has repeatedly been associated with an increased risk of gouty arthritis ([5, 10, 12-21]).

However, although most studies have demonstrated an increased risk of gouty arthritis among those who used diuretics, the magnitude of the observed risks varied considerably among studies. Most investigators studied diuretics as a class of drugs rather than as individual drugs ([5, 13, 14, 16-19, 21-23]), and in many studies, findings related to different types of diuretics were based on small numbers of exposed patients (n < 50) ([12, 15, 20]). In addition, no association of the disease with the duration of diuretic use has yet been reported. Finally, potential confounding factors, such as concomitant treatment with antihypertensive drugs, acetylsalicylic acid (ASA), cyclosporine, or pyrazinamide, as well as comorbid conditions, such as hypertension, chronic kidney disease, congestive heart failure (CHF), and diabetes mellitus (which all have been linked to an increased risk of gout [[3, 24]]), have not routinely been controlled for in previous studies. We therefore conducted an observational study to investigate the association between use of different types of diuretics and the risk of developing incident gouty arthritis.

PATIENTS AND METHODS

Data source

Data were derived from the General Practice Research Database, a large, UK-based primary care database that was established in 1987. It encompasses data on some 7 million patients registered with selected general practitioners (∼7% of the UK population is represented in the database) ([25-27]). The individuals enrolled in the database are representative of the UK population with regard to age, sex, geographic distribution, and annual turnover rate ([3, 28, 29]). General practitioners have been trained to record medical information for research purposes using standard software and coding systems. The General Practice Research Database holds anonymized information regarding demographics and patient characteristics, as well as lifestyle variables, such as body mass index (BMI), smoking status, and alcohol consumption, and information on symptoms, medical diagnoses, referrals to consultants, and hospitalizations. General practitioners generate drug prescriptions electronically using a coded drug dictionary; therefore, prescriptions include the name of the preparation, route of administration, dose of a single unit, and number of units prescribed. The database has been described in detail elsewhere ([30, 31]) and has been validated extensively ([26, 32-35]). Data from the General Practice Research Database have been used in numerous epidemiologic studies ([27, 30, 31, 36, 37]), including studies pertaining to gout ([3, 38, 39]). The Independent Scientific Advisory Committee for Medicines and Healthcare products Regulatory Agency database research approved this study.

Study population

Case patients

Using READ codes (a standard clinical terminology system), we identified all patients who were ≥18 years of age and who had received a first-time diagnosis of gout between 1990 and 2010. For this study, the index date is the date at which the first diagnosis of gout was received.

Exclusion criteria

We excluded all patients with <3 years of recorded history in the database prior to the index date, as well as all patients with any recorded cancer diagnosis (except nonmelanoma skin cancer) or a human immunodeficiency virus infection prior to the index date. Additionally, we excluded all patients who were diagnosed as having hemochromatosis, osteoarthritis, septic arthritis, or rheumatoid arthritis within the 180 days that preceded the index date or within the 90 days that followed the index date. Similar case definitions of gout have been used and validated in previous studies based on General Practice Research Database data ([3, 38, 39]).

Control patients

From the base population, one control patient without any evidence of gout was matched at random to each gout case. Cases and controls were matched for calendar time (same index date), age (same year of birth), sex, general practice, and number of years of active history in the General Practice Research Database prior to the index date. We applied the same exclusion criteria to control patients as to case patients.

Definition and classification of diuretic use

We assessed the records of cases and controls to determine their use of different types of diuretics prior to the index date. We classified the diuretics into 4 groups according to World Health Organization (WHO) classification: loop diuretics, thiazide diuretics, thiazide-like diuretics, and potassium-sparing diuretics.

Exposed patients were classified as follows, based on the date the last prescription was issued: “current users” (last prescription issued 1–180 days prior to the index date), “past users” (last prescription issued >180 days prior to the index date), or “non-users” (no prescription issued prior to the index date). Most diuretics are available in packages of 90 tablets or more; therefore, we chose a cutoff date of 180 days prior to the index date to increase the likelihood of properly separating current users of diuretics from past users. Duration of exposure was classified as follows, based on the number of recorded prescriptions for the various types of diuretics prior to the index date: short-term duration of use (1–9 prescriptions), intermediate duration of use (10–19 prescriptions), or long-term duration of use (≥20 prescriptions).

Covariates and sensitivity analyses

For both cases and controls, we assessed whether arterial hypertension, chronic kidney disease, ischemic heart disease, CHF, transient ischemic attack (TIA)/stroke, diabetes mellitus, or dyslipidemia had ever been recorded prior to the index date. Furthermore, we assessed the independent associations of antihypertensive drugs (beta-blockers, angiotensin-converting enzyme [ACE] inhibitors, angiotensin II receptor blockers [ARBs], or calcium-channel blockers), organic nitrates, statins, pyrazinamide, cyclosporine, and low-dose ASA prior to the index date. Additionally, we classified cases and controls according to their smoking status (nonsmoker, current smoker, past smoker, or unknown), BMI (<25 kg/m2, 25–29.9 kg/m2, 30–59.9 kg/m2, or unknown), alcohol consumption (never, current, past, or unknown), and alcohol use (1–9 units per week, 10–19 units per week, ≥20 units per week), and we assessed these covariates as potential confounders.

In a predefined sensitivity analysis, we restricted the analysis to cases and controls who had received only one type of diuretic or fixed combinations thereof. In addition, to further address potential bias by indication, we stratified our analyses by arterial hypertension, chronic kidney disease, and CHF. These comorbidities are all linked to an increased risk of gout and may be important confounders of the association of interest. Finally, we assessed the risk of gout in association with use of diuretics in the subset of cases (and their controls) who were treated with nonsteroidal antiinflammatory drugs (NSAIDs), colchicine, or uricosuric/uricostatic drugs within 7, 30, and 90 days of the index date, respectively.

Statistical analysis

Conditional logistic regression analysis was performed using SAS statistical software (version 9.3; SAS Institute) to calculate relative risk estimates as odds ratios (ORs) with 95% confidence intervals (95% CIs). P values less than 0.05 (2-sided) were considered significant. In univariate analysis, we explored the association of arterial hypertension, chronic kidney disease, ischemic heart disease, CHF, TIA/stroke, diabetes mellitus, dyslipidemia, smoking status, BMI, alcohol consumption, as well as use of ACE inhibitors, beta-blockers, calcium-channel blockers, nitrates, pyrazinamide, cyclosporine, statins, and low-dose ASA, with the risk of gout. We tested the association of each of these potential confounders in multivariate analyses and included them in the final model if they altered the association of diuretic use with the risk of gout by >10%. To avoid overadjustment, we did not include arterial hypertension, chronic kidney disease, and CHF in the final model, since diuretics are routinely used to treat these medical conditions. However, we stratified our analyses by these important comorbidities.

RESULTS

The study population encompassed 91,530 patients with a first-time diagnosis of gout and 91,530 matched control patients. Of these, 25.9% were female, and 74.1% were male. The mean ± SD age at the index date was 69 ± 14.5 years in women and 59 ± 15.3 years in men. Time of active history in the database prior to the index date was 11.4 ± 5.3 years for cases and controls. Being overweight or obese was associated with an increased risk of gout, as was alcohol consumption (where the risk increased as alcohol use increased), while current smoking was associated with a decreased risk of gout.

Comorbidities such as arterial hypertension, dyslipidemia, chronic kidney disease, CHF, ischemic heart disease, and TIA/stroke were all associated with an increased risk of incident gout (Table 1). Analysis of treatment with most antihypertensive drugs, except losartan and calcium-channel blockers (which were both associated with decreased relative risks of gout), revealed increased gout risks. Cyclosporine therapy was also associated with an increased risk of gout, while neither low-dose ASA nor statin use was associated with an altered risk (Table 2).

Table 1. Demographic and clinical characteristics of the patients with gout and the age- and sex-matched control patients*
VariableNo. (%) of cases (n = 91,530)No. (%) of controls (n = 91,530)Crude OR for incident gout (95% CI)PAdjusted OR for incident gout (95% CI)P
  1. Patients without evidence of gout served as controls. Controls were additionally matched for general practice, calendar time, and years of active history in the database. The odds ratios (ORs) were adjusted for the following variables: body mass index (BMI), smoking status, alcohol consumption, hypertension, chronic kidney disease, congestive heart failure (CHF), and ischemic heart disease. 95% CI = 95% confidence interval; NA = not applicable; TIA = transient ischemic attack.
  2. aOne unit is equivalent to 10 ml of pure ethanol (8 gm of ethanol).
Sex      
Male67,823 (74.1)67,823 (74.1)NANANANA
Female23,707 (25.9)23,707 (25.9)NANANANA
Age group, years      
18–291,721 (1.9)1,711 (1.9)NANANANA
30–397,242 (7.9)7,250 (7.9)NANANANA
40–4913,548 (14.8)13,552 (14.8)NANANANA
50–5917,344 (19.0)17,316 (18.9)NANANANA
60–6919,544 (21.4)19,586 (21.4)NANANANA
70–7919,595 (21.4)19,590 (21.4)NANANANA
>7912,536 (13.7)12,525 (13.7)NANANANA
BMI group, kg/m2      
12.0–18.4440 (0.5)1,003 (1.1)0.67 (0.60–0.76)<0.0010.76 (0.67–0.87)<0.001
18.5–24.915,912 (17.4)25,050 (27.4)1 (referent)NA1 (referent)NA
25.0–29.931,027 (33.9)26,518 (29.0)1.89 (1.84–1.94)<0.0011.72 (1.67–1.77)<0.001
30.0–60.023,304 (25.5)11,645 (12.7)3.32 (3.22-3.43)<0.0012.73 (2.64–2.83)<0.001
Unknown20,847 (22.8)27,314 (29.8)1.09 (1.06-1.13)<0.0011.42 (1.37–1.48)<0.001
Smoking status      
Nonsmoker38,532 (42.1)36,776 (40.2)1 (referent)NA1 (referent)NA
Current smoker13,031 (14.2)16,388 (17.9)0.74 (0.72–0.77)<0.0010.76 (0.74–0.79)<0.001
Former smoker26,625 (29.1)20,348 (22.2)1.33 (1.30–1.36)<0.0011.10 (1.07–1.13)<0.001
Unknown13,342 (14.6)18,018 (19.7)0.62 (0.60–0.64)<0.0010.91 (0.87–0.95)<0.001
Alcohol use, units/weeka      
Never used10,648 (11.6)11,902 (13.0)1 (referent)NA1 (referent)NA
Current use      
Unknown15,816 (17.3)15,995 (17.5)1.12 (1.08–1.16)<0.0011.17 (1.12–1.22)<0.001
1–918,574 (20.3)20,280 (22.2)1.08 (1.04–1.12)<0.0011.16 (1.11–1.20)<0.001
10–1911,261 (12.3)9,460 (10.3)1.47 (1.41–1.53)<0.0011.63 (1.56–1.71)<0.001
≥2016,171 (17.7)8,255 (9.0)2.48 (2.38–2.58)<0.0012.83 (2.70–2.96)<0.001
Unknown19,060 (20.8)25,638 (28.0)0.79 (0.76–0.82)<0.0011.08 (1.02–1.13)0.004
Comorbidities      
Hypertension39,890 (43.6)23,904 (26.1)2.61 (2.55–2.67)<0.0012.05 (2.00–2.10)<0.001
Diabetes mellitus7,555 (8.3)6,416 (7.0)1.20 (1.16–1.25)<0.0010.72 (0.69–0.75)<0.001
Dyslipidemia13,798 (15.1)9,064 (9.9)1.72 (1.67–1.78)<0.0011.20 (1.15–1.24)<0.001
Chronic kidney disease14,328 (15.7)7,490 (8.2)4.03 (3.85–4.22)<0.0012.83 (2.69–2.98)<0.001
CHF8,038 (8.8)2,503 (2.7)3.93 (3.73–4.13)<0.0013.18 (3.01–3.36)<0.001
Ischemic heart disease16,584 (18.1)10,487 (11.5)1.84 (1.79–1.89)<0.0011.30 (1.26–1.34)<0.001
Stroke/TIA6,453 (7.1)4,819 (5.3)1.40 (1.34–1.45)<0.0011.10 (1.05–1.15)<0.001
Table 2. Concomitant medications taken by the patients with gout and the matched control patients (current use)*
Concomitant medicationNo. (%) of cases (n = 91,530)No. (%) of controls (n = 91,530)Crude OR for incident gout (95% CI)PAdjusted OR for incident gout (95% CI)P
  1. The odds ratios (ORs) were adjusted for the following variables: body mass index, smoking status, alcohol consumption, and use of potassium-sparing diuretics, thiazide diuretics, thiazide-like diuretics, loop diuretics, angiotensin-converting enzyme (ACE) inhibitors, beta-blockers, calcium-channel blockers, and nitrates prior to the index date. 95% CI = 95% confidence interval; ARB = angiotensin II receptor blocker; ASA = acetylsalicylic acid.
ACE inhibitor19,121 (20.9)9,759 (10.7)1.04 (1.00–1.10)0.0731.14 (1.08–1.20)<0.001
ARB (excluding losartan)4,086 (4.5)1,979 (2.2)1.07 (0.95–1.21)0.2891.24 (1.07–1.42)0.003
Losartan1,429 (1.6)896 (1.0)0.74 (0.64–0.85)<0.0010.89 (0.76–1.04)0.150
Beta-blockers19,540 (21.4)10,048 (11.0)1.60 (1.54–1.67)<0.0011.66 (1.59–1.73)<0.001
Calcium-channel blockers13,290 (14.5)9,425 (10.3)0.76 (0.73–0.80)<0.0010.87 (0.83–0.92)<0.001
Nitrates7,641 (8.4)4,381 (4.8)1.29 (1.23–1.36)<0.0011.14 (1.08–1.21)<0.001
Statins17,142 (18.7)11,502 (12.6)1.09 (1.02–1.17)0.0100.95 (0.88–1.03)0.228
Low-dose ASA15,983 (17.5)11,444 (12.5)1.01 (0.97–1.06)0.5600.97 (0.92–1.03)0.284
Pyrazinamide22 (0.0)20 (0.0)0.88 (0.46–1.67)0.6930.94 (0.46–1.91)0.855
Cyclosporine263 (0.3)39 (0.0)1.56 (0.89–2.71)0.1201.56 (0.84–2.87)0.157

Compared with past use of diuretics from the respective diuretic drug classes, current use of loop diuretics (adjusted OR 2.64 [95% CI 2.47–2.83]), thiazide diuretics (adjusted OR 1.70 [95% CI 1.62–1.79]), and thiazide-like diuretics (adjusted OR 2.30 [95% CI 1.95–2.70]) was associated with increased risks of incident gout, although current use of potassium-sparing diuretics (adjusted OR 1.06 [95% 0.91–1.23]) was not. The risk increased in current users of loop diuretics as the duration of use increased (Table 3).

Table 3. ORs for incident gout in association with current use of different types of diuretics*
DiureticsNo. (%) of cases (n = 91,530)No. (%) of controls (n = 91,530)Crude OR (95% CI)PAdjusted OR (95% CI)P
  1. The odds ratios (ORs) were adjusted for the following variables: body mass index, smoking status, alcohol consumption, and use of angiotensin-converting enzyme inhibitors, beta-blockers, calcium-channel blockers, and nitrates, as well as concomitant use of diuretic drugs of other classes prior to the index date. Loop diuretics included mainly furosemide and bumetanide, as well as torsemide. Thiazide diuretics included mainly bendroflumethiazide, as well as hydroflumethiazide, hydrochlorothiazide, chlorothiazide, polythiazide, cyclopenthiazide, and methyclothiazide. Thiazide-like diuretics included mainly indapamide and chlorthalidone, as well as metolazone, quinethazone, xipamide, mefruside, clorexolone, and clopamide. Potassium-sparing diuretics included mainly spironolactone and amiloride, as well as eplerenone and triamterene. 95% CI = 95% confidence interval; NA = not applicable.
Loop diuretics      
Never used74,177 (81.0)84,387 (92.2)0.57 (0.54–0.60)<0.0010.75 (0.71–0.80)<0.001
Current use (<180 days)      
Overall13,487 (14.7)4,136 (4.5)2.63 (2.48–2.80)<0.0012.64 (2.47–2.83)<0.001
1–9 prescriptions2,652 (2.9)1,104 (1.2)1.81 (1.66–1.98)<0.0011.95 (1.77–2.15)<0.001
10–19 prescriptions2,330 (2.6)780 (0.9)2.36 (2.14–2.60)<0.0012.36 (2.12–2.62)<0.001
≥20 prescriptions8,505 (9.3)2,252 (2.5)3.18 (2.96–3.42)<0.0013.16 (2.93–3.42)<0.001
Past use (>180 days)      
Overall3,866 (4.2)3,007 (3.3)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Never used68,835 (75.2)78,335 (85.6)0.49 (0.47–0.51)<0.0010.85 (0.81–0.88)<0.001
Current use (<180 days)      
Overall13,332 (14.6)7,203 (7.9)1.16 (1.11–1.22)<0.0011.70 (1.62–1.79)<0.001
1–9 prescriptions2,605 (2.9)1,483 (1.6)1.05 (0.97–1.13)0.2091.51 (1.39–1.64)<0.001
10–19 prescriptions2,463 (2.7)1,297 (1.4)1.15 (1.06–1.24)<0.0011.64 (1.51–1.79)<0.001
≥20 prescriptions8,264 (9.0)4,423 (4.8)1.21 (1.15–1.27)<0.0011.81 (1.71–1.92)<0.001
Past use (>180 days)      
Overall9,363 (10.2)5,992 (6.6)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Never used89,258 (97.5)90,376 (98.7)0.59 (0.53–0.65)<0.0010.95 (0.85–1.07)0.425
Current use (<180 days)      
Overall1,260 (1.4)537 (0.6)1.45 (1.25–1.67)<0.0012.30 (1.95–2.70)<0.001
1–9 prescriptions314 (0.3)130 (0.1)1.47 (1.17–1.85)<0.0012.08 (2.61–2.70)<0.001
10–19 prescriptions270 (0.3)101 (0.1)1.64 (1.27–2.10)<0.0012.36 (1.79–3.12)<0.001
≥20 prescriptions676 (0.7)306 (0.3)1.37 (1.16–1.62)<0.0012.44 (2.01–2.97)<0.001
Past use (>180 days)      
Overall1,012 (1.1)617 (0.7)1 (referent)NA1 (referent)NA
Potassium-sparing diuretics      
Never used87,452 (95.5)90,392 (98.8)0.31 (0.28–0.34)<0.0010.62 (0.55–0.69)<0.001
Current use (<180 days)      
Overall2,372 (2.6)572 (0.6)1.37 (1.20–1.57)<0.0011.06 (0.91–1.23)0.470
1–9 prescriptions769 (0.8)182 (0.2)1.37 (1.14–1.66)0.0011.14 (0.92–1.40)0.230
10–19 prescriptions516 (0.6)109 (0.1)1.57 (1.25–1.98)<0.0011.21 (0.94–1.57)0.141
≥20 prescriptions1,087 (1.2)281 (0.3)1.29 (1.10–1.52)0.0020.96 (0.80–1.15)0.623
Past use (>180 days)      
Overall1,706 (1.9)566 (0.6)1 (referent)NA1 (referent)NA

Concomitant treatment with losartan attenuated the ORs for gout in users of loop diuretics (adjusted OR 0.81 [95% CI 0.59–1.12]), thiazide diuretics (adjusted OR 0.76 [95% CI 0.56–1.02]), and thiazide-like diuretics (adjusted OR 0.85 [95% CI 0.39–1.86]), compared to past use of losartan and current use of the respective diuretic drugs. Similarly, concomitant treatment with calcium-channel blockers decreased relative risk estimates for gout in current users of loop diuretics (adjusted OR 0.64 [95% CI 0.57–0.72]) and thiazide diuretics (adjusted OR 0.82 [95% CI 0.74–0.91]) compared to past use of calcium-channel blockers and current use of the respective diuretic drugs, although the same was not true for thiazide-like diuretics (adjusted OR 1.31 [95% CI 0.95–1.80]).

In the sensitivity analysis of the mutually exclusive groups of different diuretic drug classes, current use of loop diuretics (adjusted OR 3.01 [95% CI 2.69–3.37]), combined use of loop diuretics and thiazide diuretics (adjusted OR 4.65 [95% CI 3.51–6.16]), and combined use of loop diuretics and potassium-sparing agents (adjusted OR 4.53 [95% CI 2.96–6.93]) yielded substantially increased relative risk estimates (Table 4). When the analysis was restricted to gout patients treated with NSAIDs, colchicine, uricosuric drugs, or uricostatic drugs, results did not materially differ from those of the main analysis (data not shown). Stratification by presence or absence of arterial hypertension did not alter results when compared to the results of the main analysis (data not shown). Stratification by presence or absence of chronic kidney disease also did not meaningfully alter the relative risk estimates (Table 5). Finally, when we stratified by presence or absence of CHF, relative risk estimates remained increased in users of loop diuretics and thiazide-like diuretics; however, among patients who used thiazide diuretics, the significant increase in gout risk overall was not observed in the subgroup with CHF (Table 6).

Table 4. ORs for incident gout in association with current use of combinations of diuretics of different drug classes*
DiureticsNo. (%) of cases (n = 91,530)No. (%) of controls (n = 91,530)Crude OR (95% CI)PAdjusted OR (95% CI)P
  1. The odds ratios (ORs) were adjusted for the following variables: body mass index, smoking status, alcohol consumption, and use of angiotensin-converting enzyme inhibitors, beta-blockers, calcium-channel blockers, and nitrates prior to the index date. Loop diuretics included mainly furosemide and bumetanide, as well as torsemide. Thiazide diuretics included mainly bendroflumethiazide, as well as hydroflumethiazide, hydrochlorothiazide, chlorothiazide, polythiazide, cyclopenthiazide, and methyclothiazide. Thiazide-like diuretics included mainly indapamide and chlorthalidone, as well as metolazone, quinethazone, xipamide, mefruside, clorexolone, and clopamide. Potassium-sparing diuretics included mainly spironolactone and amiloride, as well as eplerenone and triamterene. 95% CI = 95% confidence interval; NA = not applicable.
None51,433 (56.2)70,034 (76.5)NANANANA
Loop diuretics      
Current use4,303 (4.7)1,644 (1.8)3.29 (2.96–3.67)<0.0013.01 (2.69–3.37)<0.001
Past use1,070 (1.2)1,216 (1.3)1 (referent)NA1 (referent)NA
Loop and thiazide diuretics      
Current use469 (0.5)103 (0.1)5.03 (3.84–6.60)<0.0014.65 (3.51–6.16)<0.001
Past use352 (0.4)382 (0.4)1 (referent)NA1 (referent)NA
Loop and thiazide-like diuretics      
Current use34 (0.0)9 (0.0)4.59 (1.47–14.30)0.0093.66 (1.12–12.02)0.032
Past use12 (0.0)14 (0.0)1 (referent)NA1 (referent)NA
Loop and potassium-sparing diuretics      
Current use867 (1.0)182 (0.2)5.22 (3.49–7.80)<0.0014.53 (2.96–6.93)<0.001
Past use65 (0.1)67 (0.1)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Current use9,732 (10.6)5,605 (6.1)1.91 (1.79–2.04)<0.0011.90 (1.77–2.04)<0.001
Past use2,851 (3.1)3,029 (3.3)1 (referent)NA1 (referent)NA
Thiazide and potassium-sparing diuretics      
Current use39 (0.0)24 (0.0)1.14 (0.46–2.78)0.7821.22 (0.48–3.06)0.679
Past use22 (0.0)14 (0.0)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Current use567 (0.6)282 (0.3)2.09 (1.55–2.82)<0.0012.08 (1.53–2.85)<0.001
Past use139 (0.2)131 (0.1)1 (referent)NA1 (referent)NA
Potassium-sparing diuretics      
Current use61 (0.1)48 (0.1)1.39 (0.79–2.42)0.2521.13 (0.63–2.03)0.674
Past use57 (0.1)61 (0.1)1 (referent)NA1 (referent)NA
Other combinations19,457 (21.3)8,685 (9.5)NANANANA
Table 5. Risk of gout in association with the use of different types of diuretics, stratified by the presence or absence of chronic kidney disease*
 No. (%) of casesNo. (%) of controlsCrude OR (95% CI)PAdjusted OR (95% CI)P
  1. For chronic kidney disease, 14,328 cases and 7,490 controls are represented. For no chronic kidney disease, 77,202 cases and 84,040 controls are represented. The odds ratios (ORs) were adjusted for the following variables: body mass index, smoking status, alcohol consumption, and use of angiotensin-converting enzyme inhibitors, beta-blockers, calcium-channel blockers, and nitrates, as well as the concomitant use of diuretic drugs of other classes prior to the index date. Loop diuretics included mainly furosemide and bumetanide, as well as torsemide. Thiazide diuretics included mainly bendroflumethiazide, as well as hydroflumethiazide, hydrochlorothiazide, chlorothiazide, polythiazide, cyclopenthiazide, and methylclothiazide. Thiazide-like diuretics included mainly indapamide and chlorthalidone, as well as metolazone, quinethazone, xipamide, mefruside, clorexolone, and clopamide. Potassium-sparing diuretics included mainly spironolactone and amiloride, as well as eplerenone and triamterene. 95% CI = 95% confidence interval; NA = not applicable.
Chronic kidney disease      
Loop diuretics      
Never used6,707 (46.8)5,064 (67.6)0.84 (0.76–0.93)<0.0011.01 (0.91–1.13)0.821
Current use (<180 days)      
Overall6,313 (44.1)1,506 (20.1)3.01 (2.69–3.37)<0.0013.02 (2.67–3.40)<0.001
1–9 prescriptions870 (6.1)277 (3.7)2.08 (1.75–2.47)<0.0012.26 (1.88–2.72)<0.001
10–19 prescriptions955 (6.7)236 (3.2)2.83 (2.36–3.38)<0.0012.92 (2.41–3.53)<0.001
≥20 prescriptions4,488 (31.3)993 (13.3)3.35 (2.97–3.78)<0.0013.30 (2.90–3.76)<0.001
Past use (>180 days)      
Overall1,308 (9.1)920 (12.3)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Never used7,105 (49.6)4,003 (53.4)0.51 (0.48–0.54)<0.0011.01 (0.93–1.09)0.886
Current use (<180 days)      
Overall3,616 (25.2)1,746 (23.3)1.16 (1.08–1.25)<0.0011.46 (1.33–1.61)<0.001
1–9 prescriptions418 (2.9)196 (2.6)0.89 (0.74–1.08)0.2361.50 (1.21–1.86)<0.001
10–19 prescriptions487 (3.4)240 (3.2)0.89 (0.75–1.06)0.2081.34 (1.10–1.62)0.003
≥20 prescriptions2,711 (18.9)1,310 (17.5)0.99 (0.90–1.08)0.7661.48 (1.34–1.65)<0.001
Past use (>180 days)      
Overall3,607 (25.2)1,741 (23.2)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Never used13,493 (94.2)7,126 (95.1)0.83 (0.70–0.99)0.0421.15 (0.94–1.40)0.168
Current use (<180 days)      
Overall405 (2.8)154 (2.1)1.29 (0.99–1.67)0.0591.95 (1.45–2.61)<0.001
1–9 prescriptions94 (0.7)38 (0.5)1.25 (0.81–1.92)0.3171.52 (0.93–2.48)0.092
10–19 prescriptions72 (0.5)31 (0.4)1.00 (0.62–1.60)0.9831.38 (0.82–2.33)0.223
≥20 prescriptions239 (1.7)85 (1.1)1.42 (1.04–1.94)0.0292.40 (1.69–3.41)<0.001
Past use (>180 days)      
Overall430 (3.0)210 (2.8)1 (referent)NA1 (referent)NA
Potassium-sparing diuretics      
Never used12,240 (85.4)7,087 (94.6)0.36 (0.31–0.43)<0.0010.59 (0.49–0.71)<0.001
Current use (<180 days)      
Overall1,228 (8.6)212 (2.8)1.25 (1.00–1.56)0.0561.32 (1.04–1.68)0.022
1–9 prescriptions375 (2.6)56 (0.7)1.40 (1.00–1.95)0.0491.38 (0.97–1.97)0.073
10–19 prescriptions258 (1.8)35 (0.5)1.53 (1.02–2.29)0.0381.76 (1.15–2.69)0.010
≥20 prescriptions595 (4.2)121 (1.6)1.09 (0.84–1.41)0.5321.17 (0.88–1.55)0.280
Past use (>180 days)      
Overall860 (6.0)191 (2.6)1 (referent)NA1 (referent)NA
No chronic kidney disease      
Loop diuretics      
Never used67,470 (87.4)79,323 (94.4)0.55 (0.52–0.59)<0.0010.74 (0.69–0.79)<0.001
Current use (<180 days)      
Overall7,174 (9.3)2,630 (3.1)2.20 (2.03–2.38)<0.0012.15 (1.98–2.34)<0.001
1–9 prescriptions1,782 (2.3)827 (1.0)1.67 (1.50–1.85)<0.0011.77 (1.58–1.98)<0.001
10–19 prescriptions1,375 (1.8)544 (0.6)1.97 (1.75–2.22)<0.0011.84 (1.61–2.09)<0.001
≥20 prescriptions4,017 (5.2)1,259 (1.5)2.68 (2.44–2.93)<0.0012.57 (2.33–2.84)<0.001
Past use (>180 days)      
Overall2,558 (3.3)2,087 (2.5)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Never used61,730 (80.0)74,332 (88.4)0.50 (0.48–0.53)<0.0010.82 (0.78–0.86)<0.001
Current use (<180 days)      
Overall9,716 (12.6)5,457 (6.5)1.28 (1.22–1.36)<0.0011.61 (1.52–1.71)<0.001
1–9 prescriptions2,187 (2.8)1,287 (1.5)1.12 (1.03–1.22)0.0061.42 (1.30–1.55)<0.001
10–19 prescriptions1,976 (2.6)1,057 (1.3)1.25 (1.15–1.37)<0.0011.57 (1.43–1.73)<0.001
≥20 prescriptions5,553 (7.2)3,113 (3.7)1.37 (1.29–1.46)<0.0011.72 (1.60–1.84)<0.001
Past use (>180 days)      
Overall5,756 (7.5)4,251 (5.1)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Never used75,765 (98.1)83,250 (99.1)0.57 (0.50–0.65)<0.0010.93 (0.80–1.07)0.318
Current use (<180 days)      
Overall855 (1.1)383 (0.5)1.55 (1.29–1.86)<0.0012.09 (1.72–2.56)<0.001
1–9 prescriptions220 (0.3)92 (0.1)1.61 (1.22–2.14)<0.0012.01 (1.47–2.74)<0.001
10–19 prescriptions198 (0.3)70 (0.1)1.90 (1.40–2.59)<0.0012.25 (1.61–3.12)<0.001
≥20 prescriptions437 (0.6)221 (0.3)1.41 (1.14–1.74)0.0022.07 (1.64–2.62)<0.001
Past use (>180 days)      
Overall582 (0.8)407 (0.5)1 (referent)NA1 (referent)NA
Potassium-sparing diuretics      
Never used75,212 (97.4)83,305 (99.1)0.37 (0.33–0.42)<0.0010.69 (0.60–0.79)<0.001
Current use (<180 days)      
Overall1,144 (1.5)360 (0.4)1.39 (1.17–1.66)<0.0011.26 (1.05–1.52)0.015
1–9 prescriptions394 (0.5)126 (0.1)1.38 (1.08–1.75)0.0101.29 (1.00–1.68)0.053
10–19 prescriptions258 (0.3)74 (0.1)1.52 (1.13–2.03)0.0051.25 (0.91–1.72)0.162
≥20 prescriptions492 (0.6)160 (0.2)1.35 (1.08–1.68)0.0081.24 (0.98–1.58)0.074
Past use (>180 days)      
Overall846 (1.1)375 (0.4)1 (referent)NA1 (referent)NA
Table 6. Risk of gout in association with the use of different types of diuretics, stratified by the presence or absence of congestive heart failure*
 No. (%) of casesNo. (%) of controlsCrude OR (95% CI)PAdjusted OR (95% CI)P
  1. For congestive heart failure, 8,038 cases and 2,503 controls are represented. For no congestive heart failure, 83,492 cases and 89,027 controls are represented. The odds ratios (ORs) were adjusted for the following variables: body mass index, smoking status, alcohol consumption, and use of angiotensin-converting enzyme inhibitors, beta-blockers, calcium-channel blockers, and nitrates, as well as the concomitant use of diuretic drugs of other classes prior to the index date. Loop diuretics included mainly furosemide and bumetanide, as well as torsemide. Thiazide diuretics included mainly bendroflumethiazide, as well as hydroflumethiazide, hydrochlorothiazide, chlorothiazide, polythiazide, cyclopenthiazide, and methylclothiazide. Thiazide-like diuretics included mainly indapamide and chlorthalidone, as well as metolazone, quinethazone, xipamide, mefruside, clorexolone, and clopamide. Potassium-sparing diuretics included mainly spironolactone and amiloride, as well as eplerenone and triamterene. 95% CI = 95% confidence interval; NA = not applicable.
Congestive heart failure      
Loop diuretics      
Never used1,237 (15.4)714 (28.5)0.96 (0.82–1.12)0.5811.20 (1.01–1.42)0.036
Current use (<180 days)      
Overall6,069 (75.5)1,358 (54.3)2.62 (2.28–3.01)<0.0012.53 (2.18–2.94)<0.001
1–9 prescriptions815 (10.1)275 (11.0)1.60 (1.33–1.94)<0.0011.77 (1.45–2.17)<0.001
10–19 prescriptions1,027 (12.8)254 (10.1)2.34 (1.94–2.84)<0.0012.26 (1.84–2.76)<0.001
≥20 prescriptions4,227 (52.6)829 (33.1)3.06 (2.64−3.55)<0.0012.90 (2.48–3.40)<0.001
Past use (>180 days)      
Overall732 (9.1)431 (17.2)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Never used5,735 (71.3)1,817 (72.6)0.79 (0.70–0.89)<0.0011.01 (0.89–1.15)0.898
Current use (<180 days)      
Overall515 (6.4)194 (7.8)0.69 (0.57–0.85)<0.0010.86 (0.69–1.07)0.169
1–9 prescriptions161 (2.0)60 (2.4)0.68 (0.49–0.94)0.0190.94 (0.66–1.35)0.750
10–19 prescriptions98 (1.2)33 (1.3)0.71 (0.47–1.09)0.1210.72 (0.46–1.13)0.149
≥20 prescriptions256 (3.2)101 (4.0)0.70 (0.54–0.90)0.0070.86 (0.65–1.14)0.308
Past use (>180 days)      
Overall1,788 (22.2)492 (19.7)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Never used7,823 (97.3)2,451 (97.9)0.84 (0.60–1.19)0.3221.20 (0.83–1.74)0.333
Current use (<180 days)      
Overall50 (0.6)8 (0.3)1.64 (0.71–3.77)0.2442.51 (1.04–6.06)0.042
1–9 prescriptions12 (0.1)3 (0.1)0.96 (0.25–3.62)0.9471.14 (0.28–4.69)0.858
10–19 prescriptions8 (0.1)1 (0.0)2.63 (0.31–22.72)0.3793.71 (0.42–32.82)0.238
≥20 prescriptions30 (0.4)4 (0.2)1.93 (0.64–5.86)0.2453.33 (1.02–10.87)0.047
Past use (>180 days)      
Overall165 (2.1)44 (1.8)1 (referent)NA1 (referent)NA
Potassium-sparing diuretics      
Never used5,725 (71.2)2,102 (84.0)0.53 (0.45–0.63)<0.0010.67 (0.56–0.81)<0.001
Current use (<180 days)      
Overall1,416 (17.6)228 (9.1)1.18 (0.94–1.47)0.1491.22 (0.96–1.53)0.101
1–9 prescriptions434 (5.4)70 (2.8)1.17 (0.86–1.58)0.3261.22 (0.88–1.69)0.225
10–19 prescriptions341 (4.2)41 (1.6)1.64 (1.13–2.37)0.0101.70 (1.15–2.51)0.008
≥20 prescriptions641 (8.0)117 (4.7)1.03 (0.79–1.33)0.8521.05 (0.79–1.38)0.755
Past use (>180 days)      
Overall897 (11.2)173 (6.9)1 (referent)NA1 (referent)NA
No congestive heart failure      
Loop diuretics      
Never used72,940 (87.4)83,673 (94.0)0.60 (0.56–0.63)<0.0010.81 (0.76–0.86)<0.001
Current use (<180 days)      
Overall7,418 (8.9)2,778 (3.1)2.30 (2.14–2.47)<0.0012.27 (2.10–2.45)<0.001
1–9 prescriptions1,837 (2.2)829 (0.9)1.80 (1.63–1.99)<0.0011.90 (1.70–2.12)<0.001
10–19 prescriptions1,303 (1.6)526 (0.6)2.07 (1.83–2.33)<0.0011.97 (1.73–2.24)<0.001
≥20 prescriptions4,278 (5.1)1,423 (1.6)2.71 (4.49–2.94)<0.0012.64 (2.41–2.89)<0.001
Past use (>180 days)      
Overall3,134 (3.8)2,576 (2.9)1 (referent)NA1 (referent)NA
Thiazide diuretics      
Never used63,100 (75.6)76,518 (85.9)0.49 (0.47–0.51)<0.0010.81 (0.78–0.85)<0.001
Current use (<180 days)      
Overall12,817 (15.4)7,009 (7.9)1.30 (1.24–1.36)<0.0011.63 (1.55–1.72)<0.001
1–9 prescriptions2,444 (2.9)1,423 (1.6)1.15 (1.07–1.24)<0.0011.45 (1.34–1.58)<0.001
10–19 prescriptions2,365 (2.8)1,264 (1.4)1.26 (1.17–1.37)<0.0011.58 (1.45–1.72)<0.001
≥20 prescriptions8,008 (9.6)4,322 (4.9)1.36 (1.29–1.44)<0.0011.71 (1.61–1.81)<0.001
Past use (>180 days)      
Overall7,575 (9.1)5,500 (6.2)1 (referent)NA1 (referent)NA
Thiazide-like diuretics      
Never used81,435 (97.5)87,925 (98.8)0.58 (0.52–0.65)<0.0010.95 (0.84–1.08)0.428
Current use (<180 days)      
Overall1,210 (1.4)529 (0.6)1.53 (1.32–1.78)<0.0012.06 (1.74–2.43)<0.001
1–9 prescriptions302 (0.4)127 (0.1)1.56 (1.23–1.98)<0.0011.88 (1.44–2.45)<0.001
10–19 prescriptions262 (0.3)100 (0.1)1.71 (1.32–2.21)<0.0012.01 (1.52–2.65)<0.001
≥20 prescriptions646 (0.8)302 (0.3)1.46 (1.22–1.74)<0.0012.16 (1.77–2.63)<0.001
Past use (>180 days)      
Overall847 (1.0)573 (0.6)1 (referent)NA1 (referent)NA
Potassium sparing diuretics      
Never used81,727 (97.9)88,290 (99.2)0.42 (0.37–0.48)<0.0010.72 (0.63–0.83)<0.001
Current use (<180 days)      
Overall956 (1.1)344 (0.4)1.36 (1.14–1.62)<0.0011.27 (1.05–1.55)0.014
1–9 prescriptions335 (0.4)112 (0.1)1.40 (1.09–1.80)0.0081.26 (0.96–1.66)0.095
10–19 prescriptions175 (0.2)68 (0.1)1.23 (0.90–1.68)0.1851.11 (0.79–1.56)0.552
≥20 prescriptions446 (0.5)164 (0.2)1.38 (1.11–1.73)0.0051.36 (1.07–1.74)0.013
Past use (>180 days)      
Overall809 (1.0)393 (0.4)1 (referent)NA1 (referent)NA

To investigate whether using different cutoff dates to define current diuretic drug use could impact our findings, we tested different cutoff dates (90 days and 180 days prior to the index date). No meaningful difference was observed, but since packages could last for >90 days, we used 180 days as the cutoff in all analyses to increase the likelihood of properly separating current from past use (data not shown).

DISCUSSION

Using the General Practice Research Database, we explored the risk of incident gout in association with the use of different diuretic drugs. Current use of loop diuretics was associated with a markedly increased risk of incident gout compared to past use. Use of thiazide and thiazide-like diuretics was also associated with a significantly increased risk of developing incident gout, although use of potassium-sparing diuretics was not. These findings were further strengthened by the results obtained from the mutually exclusive model and from the analysis that was restricted to cases with recorded pharmacologic treatment of gout (results of which were similar to those found in the main model). Of interest, combined use of diuretics of different drug classes, namely, loop diuretics in combination with thiazide diuretics or potassium-sparing agents, further increased the relative risk estimates. Of note, the risk increase in association with different types of diuretics was most pronounced in individuals who used loop diuretics and was further increased by concomitant use of other diuretic drugs. This observation is consistent with the proposed mechanism of hypovolemia-induced increases in renal reabsorption of urate ([8-10]).

Our results did not materially differ when the analyses were stratified by the presence or absence of important indications for diuretic drug use, such as arterial hypertension or chronic kidney disease. However, in the analysis stratified by presence or absence of CHF, use of thiazide diuretics was no longer associated with an increased risk in patients with CHF. Taken together, these findings indicate that residual confounding by indication by arterial hypertension or chronic kidney disease does not seem to explain our findings. However, a recorded diagnosis of CHF must be considered as a confounder of the association between the risk of gout and the use of thiazides (but not other types of diuretics).

We observed a decreased relative risk estimate for gout in current users of losartan or calcium-channel blockers in combination with loop diuretics or thiazide diuretics. This observation may be of clinical importance for patients receiving diuretic therapy; treatment with losartan and/or calcium-channel blockers may be considered in patients with an increased risk of gout where appropriate.

To our knowledge, this large population-based study of >90,000 patients with incident gout is the first to assess the association of current versus past use of diuretics of different groups on the risk of gout. In a recent systematic review, Hueskes et al ([10]) reported on the results of 2 randomized controlled trials ([12, 13]) and 12 population-based studies that explored the association between diuretics and gout. Randomized controlled trials that were based on small numbers of participants were limited to use of bendrofluazide (n = 30) ([12]) and the combination of hydrochlorothiazide/triamterene (n = 7) ([13]); a markedly increased risk of incident gout in patients treated with these drugs was seen. Of the 12 available population-based studies, only the study by Gurwitz et al ([15]) demonstrated the relative risks of developing gout among individuals who used diuretic drugs of a specific class (thiazide diuretics), while no other studies differentiated between the various types of diuretics. All but one study demonstrated an increased risk of gout in association with diuretic drug use, although relative risk estimates varied considerably ([10]). In the study by Janssens et al ([22]), which included 70 cases of incident gout, the reported incidence rate ratio of gout in association with use of diuretics was 0.6 (95% CI 0.2–2.0) after adjustment for cardiovascular comorbidities such as hypertension, CHF, and myocardial infarction. However, the number of participants in that study was small, the reported 95% CIs were wide, and results for specific diuretic drug classes were not reported.

Potential confounding is an important issue when exploring the risk of gout in association with use of diuretics, since these drugs are used to treat medical conditions (such as arterial hypertension, chronic kidney disease, and CHF) that have themselves been linked to an increased risk of gout ([3, 40]). In addition, concomitant treatment with medications from other drug classes, such as antihypertensive drugs, has also been linked to an altered risk of developing gout, as seen in this study as well as in another ([40]). In our study, adjusting the analyses for potential confounders markedly attenuated relative risk estimates, but relative risk estimates remained significantly increased among current users of different types of diuretics. Furthermore, relative risk estimates were slightly increased among current users of loop diuretics as the number of prescriptions increased, and relative risk estimates remained increased in analyses stratified by presence or absence of arterial hypertension, CHF, or chronic kidney disease, with the exception of use of thiazide diuretics in cases and controls with recorded CHF. Taken together, these findings likely suggest that loop, thiazide, and thiazide-like diuretics play a role in the development of incident gout. However, the causality of such a relationship cannot be proven in an observational study.

In our study and in accordance with the findings of other studies, being overweight or obese ([5, 24]), consuming alcohol ([6, 24]), and having comorbidities, such as hypertension, chronic kidney disease, and CHF ([3, 40]), were associated with an increased risk of incident gout. Use of ACE inhibitors and beta-blockers was associated with marginally increased risks of incident gout, while use of losartan and calcium-channel blockers was associated with slightly decreased risks. These findings were similar to the results reported by Choi et al in a recently published General Practice Research Database–based study ([40]).

Our large population-based study has several strengths. We studied a large number of cases with incident gout in a well-established validated primary care database ([26, 32-35]). Furthermore, we studied different types of diuretics and the role of the duration of diuretic use, and we conducted various sensitivity analyses that yielded consistent findings. We further addressed the role of important potential confounders such as BMI, alcohol consumption, comorbidities, and/or concomitant drug therapy. Since information on diseases and drug exposure was prospectively entered in the General Practice Research Database in the absence of any study hypothesis, recall bias is not an issue. Lastly, exclusion of all patients with <3 years of recorded history in the database prior to the index date reduced the risk of including prevalent rather than incident gout cases.

Some limitations of our study have to be acknowledged. Misclassification of some gout cases may have occurred, although previous studies (based on a limited number of cases [i.e.,[38]]) have shown that gout diagnoses are recorded with high validity in the General Practice Research Database ([38]), and similar case definitions have been used in other studies ([38, 39]). However, gout diagnoses are often made based on clinical presentation and are rarely confirmed in routine clinical practice by analysis of aspirated joint fluid for evidence of urate crystals. To minimize misclassification, we excluded subjects with differential diagnoses such as osteoarthritis, arthropathy due to hemochromatosis, septic arthritis, or rheumatoid arthritis that were recorded around the index date. Nevertheless, misclassification could, in theory, distort our findings toward the observed increased risk of gout among users of different diuretic drug classes through the introduction of a diagnostic bias (i.e., from a general practitioner who may be aware of an association between use of diuretics and gout). However, it is rather implausible that such a diagnostic bias accounts for the substantially increased relative risk estimates we observed. Furthermore, presence of such a bias would also be expected in individuals who used potassium-sparing agents.

Residual confounding by indication by chronic kidney disease and CHF, which are causally linked to development of hyperuricemia, cannot be excluded in the current study despite every effort to minimize such a bias. Of note, we decided not to include these comorbidities in the final model due to concerns regarding overadjustment ([41]). However, we adjusted for use of drugs indicated for treatment of these comorbidities, and we stratified our analyses by the most important comorbidities, namely, arterial hypertension, chronic kidney disease, and CHF. By including “past users” as the reference group, we intended to further minimize bias by indication.

We did not adjust for all potential risk factors for gout, since, for example, dietary habits or physical activities ([2, 5]) are not routinely recorded in the General Practice Research Database. However, we adjusted for BMI, a factor that is related to physical activity and dietary habits. We were unable to assess race/ethnicity because this information is also not consistently available in the General Practice Research Database. However, as 86% of individuals living in the UK are white ([42]), our results are most likely representative of that same demographic. Finally, we could not address potential confounding by socioeconomic status; however, we partially controlled for this potential confounder by matching cases and controls from the same general practice, since it is likely that patients from the same neighborhood see the same general practitioner. In summary, this large observational study provides evidence that current use of loop diuretics, thiazide diuretics, and thiazide-like diuretics is associated with a substantially increased risk of incident gout.

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. Meier 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. Bruderer, Bodmer, Meier.

Acquisition of data. Jick, Meier.

Analysis and interpretation of data. Bruderer, Bodmer, Meier.

Acknowledgments

We thank Pascal Egger for programming and technical support.

Ancillary