Rheumatology 2013;52:2251–2259

RHEUMATOLOGY doi:10.1093/rheumatology/ket293
Advance Access publication 17 September 2013

Original article
Gout as a risk factor for myocardial infarction and
stroke in England: evidence from record linkage
studies
Olena O. Seminog1 and Michael J. Goldacre1

Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023

Abstract
Objective. Some studies suggest that gout is a risk factor for cardiovascular disease. There is more
evidence about the association between gout and acute myocardial infarction (MI) than about gout and
stroke, and only limited information about risks by age group and sex. We aimed to study MI and stroke
following gout, including types of stroke, by age group and comparing men and women.
Methods. We analysed an all-England national linked dataset of hospital admissions and death records
from 1999 to 2011, and a similar dataset in the Oxford Record Linkage Study spanning 1963–98. The
occurrence of MI and stroke was estimated in cohorts of patients admitted to hospital with gout, com-
pared with MI and stroke in control cohorts, and the comparisons were expressed as rate ratios (RRs).
Results. The risk of MI and stroke was elevated, and similar, in both datasets. In the all-England dataset,
which included 202 033 hospital patients with gout, the RR for MI following gout was 1.82 (95% CI 1.78,
1.85), for all stroke 1.71 (1.68, 1.75), ischaemic stroke 1.68 (1.64, 1.73), haemorrhagic stroke 1.69 (1.61,
1.77) and stroke of unspecified type 2.00 (1.95, 2.06). Associations were stronger in younger than older
age groups, and in the younger were stronger in women than men.
Conclusion. Gout was associated with increased risk of stroke as well as MI. These findings should be
considered by clinicians and may have implications for preventive management of circulatory disease risks
in people with gout.

CLINICAL
SCIENCE
Key words: gout, MI, acute stroke, risk factors, record linkage.

Introduction
Gout has been recognized for thousands of years. Today
it is the most common inflammatory joint disease in the
UK, affecting 1.5% of the population [1]. With an ageing
population, and with increasing consumption of alcohol,
sugar, meat and other purine- and protein-rich foods, gout
is likely to remain common in many developed countries
and is likely to increase in the developing world. Since
gout and cardiovascular disease are associated with simi-
lar risk factors and affect similar risk groups (e.g. men
more than women, older rather than younger people),
1
Unit of Health-Care Epidemiology, Nuffield Department of Population
Health, University of Oxford, Oxford, UK.
Submitted 5 March 2013; revised version accepted 23 July 2013.
Correspondence to: Michael J. Goldacre, Unit of Health-Care
Epidemiology, Nuffield Department of Population Health, University of
Oxford, Old Road Campus, Oxford OX3 7LF, UK.
E-mail: michael.goldacre@dph.ox.ac.uk
there is some doubt about whether their co-occurrence
in any individual represents cause and effect or whether it
is coincidence. However, findings from recent experimen-
tal studies and large observational studies are suggestive
of an association between gout and cardiovascular dis-
ease that is independent of other risk factors [2–4].
Novel animal models have demonstrated that increased
levels of uric acid have a pathogenic role in metabolic
syndrome [5], arterial damage and endothelial dysfunction
and raised blood pressure [6, 7]. Most large studies of
gout and vascular disease have been concerned with
the risk of myocardial infarction (MI) in people with gout,
and the association between gout and acute stroke has
been explored less often. Both for heart disease and
stroke, there are few large-scale studies that provide
data specifically on younger adults and on women.
It is important, clinically, to know whether people with
gout have an elevated risk of MI and stroke. We investi-
gated the association between gout requiring hospital

! The Author 2013. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Olena O. Seminog and Michael J. Goldacre
admission and the risk of MI and acute stroke and its
subtypes in men and women, including risks in relatively
young people, in a large population-based cohort study.
Methods
Description of the datasets
We analysed statistical abstracts of medical records, col-
lected for all hospitalized patients and for people who had
day case care, at all National Health Service (NHS) hos-
pitals from 1999 to 2011 in England using record-linked
Hospital Episode Statistics (HES) for England. We also
used a similar dataset, the Oxford Record Linkage Study
(ORLS), spanning the earlier period of 1963–98. We used
both datasets to see if findings in each were similar. Each
dataset includes information on each patient’s basic char-
acteristics, including age, sex, area of residence, dates
of day case or inpatient care, diagnoses recorded in hos-
pital care and coded according to the International
Classification of Diseases (ICD), operations, and personal
identifiers that enable record linkage to be undertaken.
Hospital records for each individual were also linked to
death records using mortality data provided by the
Office for National Statistics. Record linkage was under-
taken by the Oxford Record Linkage team [8], such that all
successive hospital episodes for each individual (and a
record of death, if death occurred) are held in a time-
sequenced file for the individual. The record linkage
items include the NHS number (unique for each individual
in England) and are provided in an anonymous encrypted
format to the Oxford Record Linkage team.
Description of the cohorts
The cohorts of people with gout were constructed by
identifying the first record of hospital care for gout for
each individual in each dataset. Gout cases were defined
based on the ICD codes: we used codes M10, 274, 274
and 288 in, respectively, the 10th, 9th, 8th and 7th revi-
sions of the ICD. The study included only those patients
who had gout as a primary diagnosis, to avoid case mix
bias, i.e. the possibility that gout had been recorded as an
incidental diagnosis in people admitted for other diseases.
The outcomes of interest were acute MI (ICD10 code
I21–I22), acute ischaemic stroke (I61–I62), haemorrhagic
stroke (I63) and stroke without specification of type (I64),
and the corresponding diagnostic codes for these circu-
latory diseases were selected from the earlier revisions of
the ICD. All patients with gout age 520 years admitted for
NHS care between 1999 and 2011 in the English dataset,
and from 1963 to 1998 in the ORLS, were included,
except that we excluded people who had a record of MI
or acute stroke before or at the same time as the first
admission for gout. We did this to distinguish as best as
possible the likely temporal relationship between gout and
subsequent MI or acute stroke.
Analytical design
The rate of MI and acute stroke in patients with gout was
compared to that in a control cohort, which, in accordance
2252
with earlier work by us on disease associations [9], we term
the reference cohort. The results were expressed as a rate
ratio (RR). The reference cohort was constructed of people
with various, mainly minor, medical and surgical conditions
and trauma. This was done to follow the accepted epi-
demiological approach, when using hospital controls, of
selecting a wide range of different conditions (for a list of
the conditions see the legend in Table 1). As with the gout
cohort, in the reference cohort we omitted anyone with MI
or stroke prior to or at the same time as their first day case
or inpatient care for a reference condition. Anyone with
both gout and a reference cohort condition was entered
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023
into the gout cohort and excluded from the reference
cohort. Analysis was undertaken using the same methods
for each disease studied, as described below using the
example of gout and MI. The gout and reference cohorts
were followed to identify subsequent hospital or death re-
cords for MI. We calculated the rates of admission for MI in
the gout and reference cohorts based on person-years. In
all analyses, the datasets were stratified by age (in 5-year
groups), gender, calendar year of first recorded admission,
district of residence in the ORLS (eight districts) or region of
residence in England (nine regions) and quintile of the pa-
tients’ Index of Multiple Deprivation Score (a standard
method of scoring socio-economic status in England, avail-
able in the all-England dataset only). The date of entry into
either the gout or reference cohort was the date of first
admission for gout or the reference condition. The date of
exit was the date of first admission for MI, death, or the
date of the end of the data file, whichever was the earliest.
The RR for MI was calculated based on people hospitalized
for MI or who died from MI.
We used indirect standardization, taking the combined
gout and reference cohort as the standard population. All
calculations were initially done within each stratum. For
instance, the rate of MI in the standard population within
each 5-year age stratum was applied to the number of
people in the same 5-year age stratum in the gout cohort
and separately applied to the number of people in the
same age stratum in the reference cohort. This gave an
expected number of people with MI within each stratum in
the gout cohort and within each stratum in the reference
cohort to compare with the observed number in each stra-
tum in each cohort. Next, the observed and expected
numbers of MI in each stratum were summed to give an
all-ages-combined set of observed (O) and expected (E)
numbers. Finally, the RRs presented in the tables were
calculated using the formula Ogout/Egout divided by Oref/
Eref, where O and E are the observed and expected num-
bers of MI in the gout and reference cohort, respectively.
Where findings are given for a restricted age group (e.g.
<45 years of age), the population denominator as well as
the MI numerator were confined to that age group. We
calculated the 95% CI for the RR and 2 statistics for its
significance as described elsewhere [10]. Continuity cor-
rections were done where there were small observed and
expected numbers. A P-value <0.05 was considered stat-
istically significant. The statistical analysis was carried out
using analytical software designed within the Oxford
www.rheumatology.oxfordjournals.org
 Risk of MI and stroke in patients with gout

TABLE 1 Age distribution of people with gout in the ORLS and England datasets

ORLS England

Age No. in gout % of No. in reference No. in gout % of No. in reference

group, years cohort (% male) total cohorta cohort (% male) total cohort

20–24 10 (70) 0.3 53 628 228 (82) 0.1 488 260

25–29 16 (81) 0.5 59 194 585 (84) 0.3 545 987
30–34 40 (93) 1.3 65 053 1390 (90) 0.7 603 956
35–39 57 (96) 1.8 59 045 3092 (93) 1.5 650 200
40–44 77 (91) 2.4 46 845 5303 (92) 2.6 568 347
45–49 124 (95) 3.9 39 908 7895 (92) 3.9 480 690

Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023

50–54 169 (94) 5.3 36 186 10 991 (90) 5.4 484 269
55–59 241 (88) 7.6 29 984 14 315 (88) 7.1 518 710
60–64 334 (81) 10.5 27 975 18 641 (85) 9.2 537 313
65–69 406 (80) 12.8 26 764 22 001 (81) 10.9 533 642
70–74 446 (75) 14.2 25 413 27 384 (76) 13.5 542 633
75–79 472 (66) 14.8 22 284 30 890 (70) 15.4 521 526
580 782 (53) 24.6 26 370 59 318 (57) 29.4 679 453
Total 3174 (73) 100 518 649 202 033 (74) 100 7 154 986

a
Conditions used in the reference cohort, with Office of Population, Censuses and Surveys (OPCS) code edition 4 for
operations and ICD10 code for diagnosis (with equivalent codes used for other coding editions): adenoidectomy (OPCS4
E20); appendectomy (H01–H03); dilation and curettage (Q10–Q11); hip replacement (W37–W39); knee replacement
(W40–W42); squint (ICD10 H49–H51); cataract (H25); otitis (H60–H67); upper respiratory tract infections (J00–J06); varicose
veins (I83); haemorrhoids (I84); deflected septum (J34.2); nasal polyp (J33); impacted tooth and other disorders of teeth
(K00–K03); inguinal hernia (K40); in-growing nail, toenail and other diseases of nails (L60); bunion (M20.1); internal derange-
ment of the knee (M23); dislocations, sprains and strains (S03, S13, S23, S33, S43, S53, S63, S73, S83, S93); head injury (S06
excluded from analysis of acute stroke); selected limb fractures (S42, S52, S62, S82, S92); superficial injury and contusion
(S00, S10, S20, S30, S40, S50, S60, S70, S80, S90) and contraceptive management (Z30). Note that in the analysis we
included all people eligible to be in the reference cohort in each stratum and calculated the observed and expected number of
people within each age stratum (see Methods). Then, given that the individual stratum-specific values of observed and
expected cases were equivalent with respect to age, we summed the age-specific values to provide an age-standardized
all-ages set of RRs (see Tables 2–4).

research unit using the Statistical Analysis Software pack-
age (SAS, release 9.2, SAS Institute Inc., Cary, NC, USA).
Ethical approval for analysis of the record linkage study
data was obtained from the Central and South Bristol
Multi-Centre Research Ethics Committee (04/Q2006/176).
Results
In the all-England dataset there were 202 033 people with
gout; 74% were men. In the ORLS there were 3174 people
in the gout cohort; 73% were men. The mean age of the
patients was 70.3 years in the all-England dataset and
68.8 years in the ORLS. The mean time of follow-up
from first admission for gout to subsequent MI, stroke
or death was 3.8 years in England and 5.7 years in the
ORLS. Table 1 shows the number of people in each age
group in the gout cohorts, the percentages of men and
the number in each stratum in the reference cohort. The
latter numbers were very large in every age stratum
and were ample to populate all of the model’s strata in
comparisons between the gout and reference cohorts
for each 5-year age group and for the other stratum
variables.
Table 2 shows the overall RRs, RRs within 1 year of
gout and MI or stroke and RRs at >1 year. There was
an approximately 2-fold elevation in the risk of MI and
acute stroke. For example, in the all-England dataset the
overall RR for MI was 1.82 (95% CI 1.78, 1.85), for all
stroke 1.71 (1.68, 1.75), ischaemic stroke 1.68 (1.64,
1.73), haemorrhagic stroke 1.69 (1.61, 1.77) and stroke
without specification of type 2.00 (1.95, 2.06). The findings
in the ORLS were similar (Table 2). In both datasets the
RRs were a little lower, but not much lower, after dis-
counting cases of the diseases that occurred within 1
year of first hospital care for gout (Table 2).
Men and women with gout both had an elevated risk of
MI and stroke (Table 3). Overall, there were small and in-
consistent differences between men and women in the
magnitude of risk. For example, for MI the RR was signifi-
cantly higher for women than men in the all-England data-
set but not in the ORLS dataset (Table 3). The RR for all
stroke, combined, was higher for women than men in the
ORLS but not in the all-England dataset (Table 3). English
nationwide age-specific RRs for MI and stroke for both
genders and for men and women separately are pre-
sented in Table 4. There was a gradient of increasing
risk with decreasing age, especially for stroke. For MI
the RRs in people age 570, 45–69, and 20–44 years
were 1.74 (1.70, 1.78), 2.23 (2.14, 2.33) and 3.64 (2.77,
4.71), respectively. For stroke, the corresponding RRs

www.rheumatology.oxfordjournals.org 2253
2254
TABLE 2 Risk of MI and stroke in patients with gout by time interval

Overall Under 1 year Excluding first year

Diagnosis Dataset O E RRa (95% CI) O E RRa (95% CI) O E RRa (95% CI)
Olena O. Seminog and Michael J. Goldacre

MI England 10 995 6228.2 1.82 (1.78, 1.85) 4019 2019.1 2.11 (2.04, 2.18) 6976 4209.2 1.69 (1.65, 1.73)
ORLS 89 45.7 1.95 (1.57, 2.40) 18 9.2 1.96 (1.16, 3.10) 71 36.4 1.95 (1.52, 2.46)
All stroke combined England 9951 5947.6 1.71 (1.68, 1.75) 3530 1853.9 2.01 (1.94, 2.08) 6421 4093.7 1.60 (1.56, 1.64)
ORLS 300 158.6 1.91 (1.70, 2.14) 79 36.1 2.24 (1.77, 2.81) 221 122.5 1.82 (1.58, 2.07)
Ischaemic stroke England 5391 3281.3 1.68 (1.64, 1.73) 1745 937.7 1.96 (1.87, 2.07) 3646 2343.6 1.58 (1.53, 1.64)
ORLS 63 30.3 2.10 (1.61, 2.70) 22 7.5 3.03 (1.88, 4.66) 41 22.8 1.81 (1.30, 2.47)
Haemorrhagic stroke England 1864 1133.6 1.69 (1.61, 1.77) 659 340.2 2.06 (1.90, 2.23) 1205 793.4 1.54 (1.46, 1.64)
ORLS 29 15.0 1.95 (1.30, 2.81) 4 3.1 1.31 (0.35, 3.40) 25 11.9 2.11 (1.36, 3.13)
Unspecified stroke England 5696 2946.1 2.00 (1.95, 2.06) 2299 992.1 2.51 (1.57, 2.45) 3397 1954 1.78 (1.72, 1.84)
ORLS 251 133.4 1.90 (1.67, 2.15) 70 29.4 2.45 (1.90, 3.11) 181 104.0 1.75 (1.50, 2.03)

O: number of observed events; E: number of expected events. aRR adjusted for sex, age in 5-year intervals, time period in single calendar years in the ORLS and England datasets,
and also adjusted for district of residence in the ORLS and for region of residence and deprivation score associated with patient’s area of residence, in quintiles, in the England
dataset. The numbers in the columns for All stroke can exceed those in the columns for Ischaemic stroke, Haemorrhagic stroke and Unspecified stroke because some people had a
diagnosis of different types of stroke on different occasions. Conditions used in the reference cohort, with OPCS code edition 4 for operations and ICD10 code for diagnosis (with
equivalent codes used for other coding editions): adenoidectomy (OPCS4 E20); appendectomy (H01–H03); dilation and curettage (Q10–Q11); hip replacement (W37–W39); knee
replacement (W40–W42); squint (ICD10 H49–H51); cataract (H25); otitis (H60–H67); upper respiratory tract infections (J00–J06); varicose veins (I83); haemorrhoids (I84); deflected
septum (J34.2); nasal polyp (J33); impacted tooth and other disorders of teeth (K00–K03); inguinal hernia (K40); in-growing nail, toenail and other diseases of nails (L60); bunion (M20.1);
internal derangement of the knee (M23); dislocations, sprains and strains (S03, S13, S23, S33, S43, S53, S63, S73, S83, S93); head injury (S06 excluded from analysis of acute stroke);
selected limb fractures (S42, S52, S62, S82, S92); superficial injury and contusion (S00, S10, S20, S30, S40, S50, S60, S70, S80, S90) and contraceptive management (Z30).

www.rheumatology.oxfordjournals.org
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023
 Risk of MI and stroke in patients with gout

TABLE 3 Risk of MI and stroke in patients with gout by time interval: men and women

Men Women
a
Diagnosis Dataset O E RR (95% CI) O E RRa (95% CI)

MI England 7831 4677.5 1.73 (1.69, 1.77) 3163 1550.1 2.08 (2.01, 2.16)
ORLS 66 34.4 1.93 (1.49, 2.45) 23 11.3 2.04 (1.29, 3.06)
All stroke combined England 6534 3914.1 1.73 (1.69, 1.78) 3417 2033.1 1.71 (1.65, 1.77)
ORLS 183 104.6 1.77 (1.52, 2.05) 117 54.0 2.18 (1.80, 2.62)
Ischaemic stroke England 3639 2207.5 1.71 (1.65, 1.77) 1752 1073.6 1.65 (1.58, 1.74)
ORLS 38 21.3 1.81 (1.27, 2.49) 63 30.3 2.10 (1.61, 2.70)
Haemorrhagic stroke England 1379 815.6 1.76 (1.66, 1.86) 485 317.9 1.54 (1.41, 1.69)

Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023

ORLS 21 11.1 1.91 (1.18, 2.94) 8 3.9 2.06 (0.89, 4.08)
Unspecified stroke England 3492 1813.1 2.03 (1.96, 2.11) 2204 1132.9 1.99 (1.91, 2.08)
ORLS 152 86.1 1.79 (1.51, 2.10) 251 133.4 1.90 (1.67, 2.15)

O: number of observed events; E: number of expected events. aRR adjusted for sex, age in 5-year intervals, time period in
single calendar years in the ORLS and England datasets, and also adjusted for district of residence in the ORLS and for region
of residence and deprivation score associated with patient’s area of residence, in quintiles, in the England dataset. The
numbers in the columns for All stroke can exceed those in the columns for Ischaemic stroke, Haemorrhagic stroke and
Unspecified stroke because some people had a diagnosis of different types of stroke on different occasions. Conditions used
in the reference cohort, with OPCS code edition 4 for operations and ICD10 code for diagnosis (with equivalent codes used
for other coding editions): adenoidectomy (OPCS4 E20); appendectomy (H01–H03); dilation and curettage (Q10–Q11); hip
replacement (W37–W39); knee replacement (W40–W42); squint (ICD10 H49–H51); cataract (H25); otitis (H60–H67); upper
respiratory tract infections (J00–J06); varicose veins (I83); haemorrhoids (I84); deflected septum (J34.2); nasal polyp (J33);
impacted tooth and other disorders of teeth (K00–K03); inguinal hernia (K40); in-growing nail, toenail and other diseases of
nails (L60); bunion (M20.1); internal derangement of the knee (M23); dislocations, sprains and strains (S03, S13, S23, S33, S43,
S53, S63, S73, S83, S93); head injury (S06 excluded from analysis of acute stroke); selected limb fractures (S42, S52, S62,
S82, S92); superficial injury and contusion (S00, S10, S20, S30, S40, S50, S60, S70, S80, S90) and contraceptive manage-
ment (Z30).

were 1.60 (1.57, 2.64), 2.70 (2.57, 2.84) and 5.39 (3.95,
7.18), respectively (Table 4).
At age 20–44 years, the RR for all stroke combined for
men was 4.94 (3.52, 6.76) compared with 11.89 (4.77,
24.57) for women, but the numbers were small and the
difference was not significant (Table 4). At age 45–69
years the corresponding figures were 2.56 (2.42, 2.70)
and 4.31 (3.79, 4.90), and as judged from the non-over-
lapping CIs, the RR for women was significantly higher
than that for men. By age 570 years, RRs for men and
women were similar, at 1.60 (1.55, 1.64) and 1.64 (1.58,
1.69), respectively. The data for subtypes of stroke and for
MI are also shown (Table 4).
The numbers of events in the ORLS were fairly
small when split into three age subsets (see supplemen-
tary Table S1, available at Rheumatology Online).
Where the numbers (or RRs) were large enough to be
worth consideration, the RRs were similar to those
found in the all-England dataset. The RR for MI in
people age 20–44 years was statistically significantly
high, at 9.73 (1.18, 35.28), based on 2 observed and
0.2 expected cases in the ORLS. In people age 45–69
years, the RR for MI was 2.65 (1.76, 3.83), that for all
stroke combined was 3.37 (1.84, 5.67) and the RRs for
haemorrhagic and unspecified stroke were 9.04 (2.45,
23.27) and 2.67 (1.22, 5.08), respectively. In the ORLS
among people age 570 years, the RR for MI was 1.72
(1.31, 2.22) and that for all stroke combined was 1.70
(1.30, 2.19).
Sensitivity analysis on the ICD codes for gout
In the main analyses above, we selected cases of gout
coded M10 in the ICD10 at three digits. We also did a
sensitivity analysis to determine which of the four-digit
ICD codes constituted the majority of cases. For example,
out of 10 995 cases of gout with the outcome MI, the great
majority, 10 626 cases, were coded as M10.9, Gout, un-
specified. When the exposure cohort was restricted to
M10.9, our estimates for the risk of acute vascular
events did not change: e.g. the RR for MI in patients diag-
nosed with gout of unspecified cause was 1.81 (1.78,
1.85) compared with 1.82 (1.78, 1.85) for the whole
range of gout codes.
Discussion
Principal findings
In both datasets, the HES and ORLS, gout was followed
by an increased risk of MI and acute stroke. The RRs
obtained from the two datasets were similar and each
corroborated the other even though they were years
apart and in different periods of disease management.
This indicates that the findings are robust and not de-
pendent on a particular period of time or on disease def-
inition or treatment regimes that may have changed over
time. RRs were a little higher within the first year after
hospitalization with gout, but they remained elevated
thereafter. Thus the elevated risk was not just a risk

www.rheumatology.oxfordjournals.org 2255
2256
TABLE 4 Risk of MI and stroke in patients with gout by age group, men and women, England

Men and women Men Women

a a
Diagnosis Age, years O E RR (95% CI) O E RR (95% CI) O E RRa (95% CI)

MI 20–44 59 16.4 3.64 (2.77, 4.71) 55 16.0 3.48 (2.62, 4.55) 4 0.4 10.02 (2.72, 25.76)
45–69 2257 1039.6 2.23 (2.14, 2.33) 1987 978.4 2.10 (2.00, 2.19) 270 61.2 4.50 (3.97, 5.08)
Olena O. Seminog and Michael J. Goldacre

570 8676 5123.6 1.74 (1.70, 1.78) 8676 5123.6 1.74 (1.70, 1.78) 2887 1478.3 2.00 (1.92, 2.07)
All stroke combined 20–44 47 8.8 5.39 (3.95, 7.18) 40 8.2 4.94 (3.52, 6.76) 7 0.6 11.89 (4.77, 24.57)
45–69 1623 624.2 2.70 (2.57, 2.84) 1376 565.8 2.56 (2.42, 2.70) 247 58.3 4.31 (3.79, 4.90)
570 8277 5277.6 1.60 (1.57, 1.64) 5117 3315.6 1.60 (1.55, 1.64) 3160 1961.6 1.64 (1.58, 1.69)
Ischaemic stroke 20–44 21 4.5 4.68 (2.89, 7.18) 18 4.2 4.34 (2.56, 6.90) 3 0.3 9.55 (1.96, 28.02)
45–69 973 377.8 2.68 (2.51, 2.86) 839 344.5 2.57 (2.39, 2.75) 134 33.3 4.10 (3.43, 4.87)
570 4397 2876.7 1.56 (1.51, 1.61) 2782 1843.4 1.56 (1.50, 1.62) 1615 1033.3 1.58 (1.51, 1.67)
Haemorrhagic stroke 20–44 22 3.0 7.49 (4.67, 11.40) 20 2.8 7.24 (4.39, 11.28) 2 0.1 13.42 (1.62, 48.78)
45–69 400 138.9 3.02 (2.72, 3.34) 344 126.9 2.88 (2.57, 3.22) 56 12.0 4.77 (3.59, 6.22)
570 1441 984.1 1.49 (1.41, 1.58) 1015 679.9 1.54 (1.44, 1.64) 426 304.1 1.41 (1.28, 1.56)
Unspecified stroke 20–44 9 2.0 4.53 (2.06, 8.65) 7 1.8 3.85 (1.54, 8.02) 2 0.2 12.57 (1.52, 45.76)
45–69 643 210.1 3.22 (2.96, 3.48) 526 188.2 2.98 (2.72, 3.26) 117 22.0 5.47 (4.51, 6.58)
570 5041 2719.1 1.92 (1.86, 1.97) 2958 1614.5 1.93 (1.86, 2.01) 2083 1104.4 1.93 (1.85, 2.02)

O: number of observed events; E: number of expected events. aRR adjusted for sex, age in 5-year intervals, time period in single calendar years in the ORLS and England datasets,
and also adjusted for district of residence in the ORLS and for region of residence and deprivation score associated with patient’s area of residence, in quintiles, in the England
dataset. The numbers in the columns for All stroke can exceed those in the columns for Ischaemic stroke, Haemorrhagic stroke and Unspecified stroke because some people had a
diagnosis of different types of stroke on different occasions. Conditions used in reference cohort, with OPCS code edition 4 for operations and ICD10 code for diagnosis (with
equivalent codes used for other coding editions): adenoidectomy (OPCS4 E20); appendectomy (H01–H03); dilation and curettage (Q10–Q11); hip replacement (W37–W39); knee
replacement (W40–W42); squint (ICD10 H49–H51); cataract (H25); otitis (H60–H67); upper respiratory tract infections (J00–J06); varicose veins (I83); haemorrhoids (I84); de-
flected septum (J34.2); nasal polyp (J33); impacted tooth and other disorders of teeth (K00–K03); inguinal hernia (K40); in-growing nail, toenail and other diseases of nails (L60);
bunion (M20.1); internal derangement of the knee (M23); dislocations, sprains and strains (S03, S13, S23, S33, S43, S53, S63, S73, S83, S93); head injury (S06 excluded from analysis
of acute stroke); selected limb fractures (S42, S52, S62, S82, S92); superficial injury and contusion (S00, S10, S20, S30, S40, S50, S60, S70, S80, S90) and contraceptive management
(Z30).

www.rheumatology.oxfordjournals.org
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023

associated closely in time with a first hospital admission
for gout. The majority of patients with gout were elderly
people and the overall elevation of risk is largely attribut-
able to them. However, the RRs for MI and stroke were
higher in the younger age groups, those age 20–44 and
45–69 years, and they were higher in women than men. It
is worth emphasizing, nonetheless, that both gout itself
and MI and stroke in people with gout are very uncommon
in young women. For example, there were just 274 women
age <70 years (and only four age <45 years) with gout
and MI in the whole of England (population approximately
55 million) in the 13 years of the study.
Comparisons with other studies
Our estimates of elevated risks of MI and stroke in people
with gout in England are consistent with findings else-
where [3, 11–13]. Findings from a study in Rotterdam sug-
gested that a high level of uric acid was a strong risk
factor for MI and stroke [11]. Our findings, the first in a
large population-based study of gout and circulatory dis-
ease in England, were very similar to those reported in this
Dutch population: e.g. the risk of ischaemic stroke was
1.77 (1.10, 2.83) in the Dutch study and 1.68 (1.61, 1.77)
in ours and the risk of MI was 1.87 (1.12, 3.13) in the Dutch
study and 1.82 (1.78, 1.85) in ours (Table 2). The magni-
tude of the risk of MI as quantified in our study is almost
the same as the hazard ratio of 1.84 (1.51, 2.24) in a study
by Kuo et al. in Taiwan [4].
Three papers have been published from the
Framingham Heart Study on the relationship between
gout, uric acid and acute vascular events. One investi-
gated whether the patients with gout had an increased
risk of cardiovascular disease and concluded that gout
was a risk factor for acute MI [14]. Two others investigated
the association between high levels of serum uric acid and
cardiovascular morbidity and mortality and reported an
excess of cardiovascular disease [15, 16].
A recent study published from Canada found that gout
independently increases the risk of coronary heart disease
and mortality, but the study was limited to men and the
authors did not investigate associations with acute stroke
[3]. Generally, data on gout and circulatory disease in
women are sparse. Consistent with our findings, another
large population-based study in Canada found that
women with gout have an increased risk of MI, although
the Canadian study only included women age 565 years
[13]. We cannot explain why young women with gout have
a particularly high risk of severe vascular events, but our
findings are consistent with a recent study by Mangge
et al. [17], who reported that increased serum uric acid
was a good predictor of increased cardiovascular risk in
young people. A study by Kuo et al. [4] also reported a
high risk of MI following gout in younger patients. It is
worth noting that, although risks of MI and stroke were
higher in young women than in others, gout (and MI and
stroke) was exceedingly uncommon in young women (see
Results).
It is hard to know whether the occurrence of gout in
hospital admissions is in line with the expected
www.rheumatology.oxfordjournals.org
Risk of MI and stroke in patients with gout
prevalence of the disease. In the HES as a whole, 2.8%
of all patients had an admission for gout on at least one
occasion (202 033 out of 7 154 986 people), which ap-
proximates the prevalence of gout in England and Wales
reported by Elliot et al. [1] of 1.5% overall, 2.1% in men
age 65–74 years and 2.6% in men age >75 years. It is also
worth noting that the mean age of our study population
was 70 years old and that 73% were males.
Possible mechanisms
The increased risk of acute stroke and MI in people with
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023
gout is likely mediated through gout itself and/or high
serum levels of uric acid. There are several mechanisms
that might account for these associations, including in-
flammation, endothelial dysfunction and worsening of co-
existing diseases. A comprehensive review of this subject
was done by Johnson et al. [18]. Recent experimental
studies have produced evidence suggesting that uric
acid is not an inert molecule and might contribute to the
development of cardiovascular disease [6, 19, 20].
Increased risk of stroke and MI might result from several
pathophysiological effects associated with increased
serum uric acid, including inflammation [21, 22], endothe-
lial dysfunction [23], impaired nitric oxide production [24,
25], increased blood pressure [6, 7] and a possible in-
crease in risk of atherosclerosis [26–28]. In addition to
the suggested role of uric acid in the development of ath-
erosclerosis, persistent systemic and vascular inflamma-
tory processes have been shown to contribute to the
development of atherosclerotic changes and to stimulate
prothrombotic activity that might lead to arterial occlusion
or rupture [21, 29]. There are models suggesting that uric
acid is able to enter vascular smooth muscle cells [30] and
stimulate their proliferation [22, 31, 32]. Additionally,
the association between hyperuricaemia and acute vas-
cular events might be linked through insulin resistance [33]
or caused by a combination of the factors discussed
above.
Strengths and weakness
The methods used in this study have both strengths and
weaknesses. The weaknesses include an inability, through
lack of data in the routine NHS datasets, to control for a
number of potentially confounding factors, including
smoking, blood pressure, blood cholesterol, alcohol
intake, body mass index and treatment received, includ-
ing the use of diuretics, urate lowering agents, analgesics
and anti-inflammatory therapies. However, the
Framingham Study and several large, more recent studies
have demonstrated that after controlling for these factors
the association between gout and cardiovascular dis-
eases remained significant [2, 4, 14]. We did not have
any data on serum uric acid levels. We did not have infor-
mation on the time from the onset of gout to the time of
hospitalization with circulatory disease.
The data available from the HES and ORLS include only
records of hospitalized patients, including those managed
as day case care, and deaths. Although the great majority
of people with MI and stroke would be admitted to
2257
Olena O. Seminog and Michael J. Goldacre
hospital, and their hospital records would appear in our
datasets, some patients with gout are managed wholly
within primary care. This means that our findings may
only be relevant to people whose gout was serious
enough to warrant hospital day case or inpatient care
and thus our RRs may only apply to those with severe
gout.
Important strengths include the large size of the study; it
included two datasets that are almost independent of one
another and that corroborated each other; the national
dataset, by definition, provides findings that are relevant
to the entire population of England and the study had the
statistical power to provide findings on patients subdi-
vided by age group and sex.
We had to rely on the accuracy of coding of gout, MI
and stroke. Diagnostic coding in routine hospital statistics
in England is generally considered to be sufficiently robust
enough to be used for research [34]. Studies have con-
firmed that the HES is a valid and reliable source of infor-
mation on severe vascular events, including MI, stroke
and subtypes of stroke [35, 36]. The fact that our findings
were similar to those in studies of other design and in
other countries suggests that the data have face validity.
The ICD coding for gout, MI and stroke is straightforward
and, in the setting of hospitalized cases, miscoding
of these conditions is unlikely. However, due to current
privacy regulations, we were precluded from sampling
medical case notes to test the reliability of diagnostic
coding.
Conclusion
Findings of the excess risk of MI and acute stroke in
people with gout are important for clinical practice. The
results of this study indicate that consideration should be
given to prevention of acute stroke and MI in patients with
gout. Doctors caring for people with gout may wish to
consider whether it is appropriate to counsel individual
patients on possible risks of cardiovascular events and
on how to recognize and promptly seek medical help if
they develop any early symptoms of acute MI or stroke.
Additionally, our findings suggest a need for clinical trials
to investigate effective pharmacological and behavioural
interventions aimed at decreasing the risks of acute car-
diovascular events in people with gout.
Rheumatology key messages
. Gout was associated with an increased risk of MI
and stroke.
. Gout in women and young people is rare, but their
risk of MI and stroke is high.
Acknowledgements
David Yeates wrote the software package used for the
analysis. Over many years the linked data files were built
by Leicester Gill and Matt Davidson, Unit of Health-Care
Epidemiology, University of Oxford.
2258
Funding: The Unit of Health-Care Epidemiology is funded
by the English National Institute for Health Research (grant
number RNC/035/02) to analyse the linked data. This
study was funded from the general grant received by the
Unit of Health-Care Epidemiology. The views expressed in
this article do not necessarily reflect those of the funding
body.
Disclosure statement: The authors have declared no con-
flicts of interest.
Supplementary data
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023
Supplementary data are available at Rheumatology
Online.
References
1 Elliot AJ, Cross KW, Fleming DM. Seasonality and trends
in the incidence and prevalence of gout in England and
Wales 1994–2007. Ann Rheum Dis 2009;68:1728–33.
2 Krishnan E, Baker JF, Furst DE, Schumacher HR. Gout
and the risk of acute myocardial infarction. Arthritis Rheum
2006;54:2688–96.
3 Choi HK, Curhan G. Independent impact of gout on mor-
tality and risk for coronary heart disease. Circulation 2007;
116:894–900.
4 Kuo CF, Yu KH, See LC et al. Risk of myocardial infarction
among patients with gout: a nationwide population-based
study. Rheumatology 2013;52:111–7.
5 Nakagawa T, Hu H, Zharikov S et al. A causal role for uric
acid in fructose-induced metabolic syndrome. Am J
Physiol Renal Physiol 2006;290:F625–31.
6 Mazzali M, Hughes J, Kim YG et al. Elevated uric acid
increases blood pressure in the rat by a novel crystal-
independent mechanism. Hypertension 2001;38:1101–6.
7 Feig DI, Nakagawa T, Karumanchi SA et al. Hypothesis:
uric acid, nephron number, and the pathogenesis of
essential hypertension. Kidney Int 2004;66:281–7.
8 Gill L, Goldacre M. English national record linkage of
hospital episode statistics and death registration records.
Report to the Department of Health. Oxford: Unit of
Health-Care Epidemiology, Oxford University, 2003.
http://nchod.uhce.ox.ac.uk/NCHOD%20Oxford%20E5%
20Report%201st%20Feb_VerAM2.pdf (10 July 2013, date
last accessed).
9 Seminog OO, Goldacre MJ. Risk of pneumonia and
pneumococcal disease in people with severe mental
illness: English record linkage studies. Thorax 2013;68:
171–6.
10 Breslow N, Day N. Statistical Methods in Cancer
Research. Vol. II. The Design and Analysis of Cohort
Studies. Lyon: International Agency for Research on
Cancer, 1987:93–5.
11 Bos MJ, Koudstaal PJ, Hofman A et al. Uric acid is a risk
factor for myocardial infarction and stroke: the Rotterdam
study. Stroke 2006;37:1503–7.
12 Lin YH, Hsu HL, Huang YC et al. Gouty arthritis in
acute cerebrovascular disease. Cerebrovasc Dis 2009;28:
391–6.
www.rheumatology.oxfordjournals.org

13 De Vera MA, Rahman MM, Bhole V et al. Independent
impact of gout on the risk of acute myocardial infarction
among elderly women: a population-based study. Ann
Rheum Dis 2010;69:1162–4.
14 Abbott RD, Brand FN, Kannel WB et al. Gout and coronary
heart disease: the Framingham Study. J Clin Epidemiol
1988;41:237–42.
15 Brand F, McGee D, Kannel W et al. Hyperuricemia as a
risk factor of coronary heart disease: the Framingham
Study. Am J Epidemiol 1985;121:11–8.
16 Culleton BF, Larson MG, Kannel WB et al. Serum
uric acid and risk for cardiovascular disease and death:
the Framingham Heart Study. Ann Intern Med 1999;131:
7–13.
17 Mangge H, Zelzer S, Puerstner P et al. Uric acid best
predicts metabolically unhealthy obesity with increased
cardiovascular risk in youth and adults. Obesity (Silver
Spring) 2013;21:E71–7.
18 Johnson RJ, Kang DH, Feig D et al. Is there a pathogenetic
role for uric acid in hypertension and cardiovascular and
renal disease? Hypertension 2003;41:1183–90.
19 Sánchez-Lozada LG, Tapia E, Avila-Casado C et al.
Mild hyperuricemia induces glomerular hypertension in
normal rats. Am J Physiol Renal Physiol 2002;283:
F1105–10.
20 Sánchez-Lozada LG, Tapia E, Jiménez A et al. Fructose-
induced metabolic syndrome is associated with glomeru-
lar hypertension and renal microvascular damage in rats.
Am J Physiol Renal Physiol 2007;292:F423–9.
21 Libby P, Ridker PM, Maseri A. Inflammation and athero-
sclerosis. Circulation 2002;105:1135–43.
22 Kanellis J, Watanabe S, Li JH et al. Uric acid stimulates
monocyte chemoattractant protein-1 production in vas-
cular smooth muscle cells via mitogen-activated protein
kinase and cyclooxygenase-2. Hypertension 2003;41:
1287–93.
23 Kanellis J, Kang DH. Uric acid as a mediator of endothelial
dysfunction, inflammation, and vascular disease. Semin
Nephrol 2005;25:39–42.
24 Kang DH, Park SK, Lee IK et al. Uric acid-induced
C-reactive protein expression: implication on cell prolifer-
ation and nitric oxide production of human vascular cells.
J Am Soc Nephrol 2005;16:3553–62.
www.rheumatology.oxfordjournals.org
Risk of MI and stroke in patients with gout
25 Gersch C, Palii SP, Kim KM et al. Inactivation of nitric
oxide by uric acid. Nucleosides Nucleotides Nucleic Acids
2008;27:967–78.
26 Neogi T, Ellison RC, Hunts S et al. Serum uric acid is
associated with carotid plaques: the National Heart, Lung,
and Blood Institute Family Heart Study. J Rheumatol 2009;
36:378–84.
27 Erdogan D, Gullu H, Caliskan M et al. Relationship of
serum uric acid to measures of endothelial function and
atherosclerosis in healthy adults. Int J Clin Pract 2005;59:
1276–82.
28 Gür M, S¸ahin DY, Elbasan Z et al. Uric acid and high
Downloaded from https://academic.oup.com/rheumatology/article/52/12/2251/1802190 by guest on 02 December 2023
sensitive C-reactive protein are associated with subclin-
ical thoracic aortic atherosclerosis. J Cardiol 2012;61:
144–8.
29 Tiong AY, Brieger D. Inflammation and coronary artery
disease. Am Heart J 2005;150:11–8.
30 Kang DH, Han L, Ouyang X et al. Uric acid causes vascular
smooth muscle cell proliferation by entering cells via a
functional urate transporter. Am J Nephrol 2005;25:
425–33.
31 Rao GN, Corson MA, Berk BC. Uric acid stimulates
vascular smooth muscle cell proliferation by increasing
platelet-derived growth factor A-chain expression.
J Biol Chem 1991;266:8604–8.
32 Corry DB, Eslami P, Yamamoto K et al. Uric acid
stimulates vascular smooth muscle cell proliferation and
oxidative stress via the vascular renin-angiotensin system.
J Hypertens 2008;26:269–75.
33 Wortmann RL. Gout and hyperuricemia. Curr Opin
Rheumatol 2002;14:281–6.
34 Burns E, Rigby E, Mamidanna R et al. Systematic review
of discharge coding accuracy. J Public Health 2012;34:
138–48.
35 Wright FL, Green J, Canoy D et al. Vascular disease in
women: comparison of diagnoses in hospital episode
statistics and general practice records in England. BMC
Med Res Methodol 2012;12:161.
36 Britton A, Milne B, Butler T et al. Validating self-reported
strokes in a longitudinal UK cohort study (Whitehall II):
extracting information from hospital medical records
versus the Hospital Episode Statistics database. BMC
Med Res Methodol 2012;12:83.
2259