INTRODUCTION
METHODS
Am J Clin Nutr 2011;94:111326. Printed in USA. 2011 American Society for Nutrition
1113
Supplemental Material can be found at:
http://ajcn.nutrition.org/content/suppl/2011/09/20/94.4.1113.
DC1.html
ABSTRACT
Background: The effect of coffee and caffeine on blood pressure
(BP) and cardiovascular disease (CVD) in hypertensive persons is
uncertain.
Objective: The objective was to summarize the evidence on the
acute and longer-term effects of caffeine and coffee intake on BP
and on the association between habitual coffee consumption and
risk of CVD in hypertensive individuals.
Design: A systematic review and meta-analysis of publications
identified in a PubMed and EMBASE search up to 30 April 2011
was undertaken. Data were extracted from controlled trials on the
effect of caffeine or coffee intake on BP change and from cohort
studies on the association between habitual coffee consumption and
CVD.
Results: In 5 trials, the administration of 200300 mg caffeine
produced a mean increase of 8.1 mm Hg (95% CI: 5.7, 10.6 mm
Hg) in systolic BP and of 5.7 mm Hg (95% CI: 4.1, 7.4 mm Hg) in
diastolic BP. The increase in BP was observed in the first hour after
caffeine intake and lasted 3 h. In 3 studies of the longer-term
effect (2 wk) of coffee, no increase in BP was observed after coffee
was compared with a caffeine-free diet or was compared with decaffeinated coffee. Last, 7 cohort studies found no evidence of an
association between habitual coffee consumption and a higher risk
of CVD.
Conclusions: In hypertensive individuals, caffeine intake produces
an acute increase in BP for 3 h. However, current evidence does
not support an association between longer-term coffee consumption
and increased BP or between habitual coffee consumption and an
increased risk of CVD in hypertensive subjects.
Am J Clin Nutr
2011;94:111326.
1114
MESAS ET AL
Literature search
A PubMed (http://www.ncbi.nlm.nih.gov/pubmed) and EMBASE
(http://www.embase.com) search up to 30 April 2011 was conducted
by using the key terms (words in the title or abstract of the
manuscript) coffee, caffeine, hypertension, hypertensive,
hypertensives, high blood pressure, blood pressure, and
mortality and the CVD-related terms ischemic heart disease,
myocardial infarction, angina pectoris, arrhythmia, atrial
fibrillation, stroke, heart failure, coronary heart disease,
coronary bypass, and coronary angioplasty. The complete
PubMed and EMBASE searches are shown elsewhere (see
Supplemental File 1 under Supplemental data in the online
issue). Reference lists in articles retrieved from the electronic
search were also scanned.
Study selection and data extraction
Quality assessment
The clinical trials were assessed to determine whether they
were randomized, whether the intervention was blinded, whether
losses to follow-up were identified, and whether each BP reading
was made at least twice. The cohort studies were assessed to
determine whether the diagnosis of hypertension was based on
self-reports or measurements, whether repeated measurements of
coffee consumption were made during follow-up, whether losses
to follow-up were identified, how morbidity and mortality of
CVD were measured or validated, and the degree of adjustment
for potential confounders, particularly other predictors of CVD
risk and antihypertensive drugs.
Synthesis of the data
In each clinical trial, the net effect of coffee/caffeine on
BP was calculated as the difference in mean SBP and DBP
change between the intervention and control groups. In crossover designs, the effect of coffee/caffeine was calculated as
the difference in BP between the intervention and control
periods.
where SEMi is the SEM in each stratum and ni is the size of each
stratum. In studies without information on SEM or other sources
to calculate the SD (19, 26), the SEM was imputed in a standardized way by using the prognostic method (29) with the following
formula:
.
2
SEMj Rkil SEMi ni 1=2 knj 1=2
where k is the total number of strata.
In the crossover studies, the SD of BP change in each period
(intervention and control) was calculated by using the following
formula:
h
2
SDchange SDbaseline 2 SDfinal
i1=2
2 3 R 3 SDbaseline 3 SDfinal
3
where R is the correlation coefficient (30). Because none of the
studies presented information on individuals to obtain R, the
most conservative estimate was assumed, a minimum correlation
of 0.50. This value was used in a previous meta-analysis of the
effect of coffee on BP (3).
Heterogeneity among studies was quantified by using the I2
statistic (31). To pool the results of trials of the acute effects of
caffeine on BP, fixed-effects models were used when heterogeneity was low or absent (I2 , 20%); when heterogeneity was
greater and it was considered appropriate to pool the results,
random-effects models were used. A sensitivity analysis was
performed by repeating the analyses after the studies that had
the largest effect on the overall result were excluded. In addition, the analyses were stratified by caffeine dose (200, 250, or
.250300 mg), time of abstinence from caffeine before beginning the trial (9, 12, or 48 h), and antihypertensive treatment
(yes or no) to examine the influence of these variables on the
effect of caffeine on BP. Differences in results across strata were
tested by using meta-regression models.
Possible publication bias was explored with Beggs funnel
plots (32) and with Eggers regression asymmetry test (33) in
studies with a sufficient number of observations to allow for
a meaningful analysis. Data were analyzed by using Stata software version 10.0 (34).
Clinical trials of the longer-term effects of coffee consumption
on BP were grouped according to the type of comparison examined: coffee compared with caffeine-free diet, coffee compared with decaffeinated coffee, and comparisons of coffees with
different amounts of CGA and HHQ. Because of the small
number of available studies and their heterogeneity, it was not
considered advisable to pool their results by using meta-analytic
techniques.
RESULTS
FIGURE 1. Flow of publications throughout the study. EMBASE, http://www.embase.com; PubMed, http://www.ncbi.nlm.nih.gov/pubmed.
1115
Eggertsen et al (28),
1993
Rakic et al (22),
1999
R, DB, C, X
R, C, P
Australia
R, DB, C, X
R, DB, C, X
R, DB, C, X
Sweden
Scotland
United States
Vlachopoulos et al
(27), 2003
Longer-term effects of
coffee intake (trials
1 wk duration)
MacDonald et al (20),
1991a,b
United States
DB, C, X
R, DB, C, X
United Kingdom
United States
C, X
Design
United Kingdom
Country
27
23
50
12
24
18
a: 8; b: 7
16
M/F
M/F
2874
50
M/F
M/F
60 6 3
2663
M/F
M/F
Sex
2039
3045
6582
3065
Age
Habitual smokers
Lifestyle/health
status
DBP 90105 mm Hg
Mild-to-moderate
hypertension
BP 140180/90
110 mm Hg and/or
treatment of HT
3 cups/d
34 cups/d
NA
Mild-to-moderate HT
Caffeine
consumers
Yes
Yes
No
Yes
No
(Continued)
BP 140160/90
105 mm Hg
BP 146160/90
99 mm Hg
18 cups/d
50800 mg
caffeine/d
No (antihypertensive
therapy stopped 4
wk before)
BP 160/95 mm Hg
Coffee or tea
drinkers
No
Yes
Antihypertensive
medications
Mild HT
Criteria for
inclusion as
hypertensive
4 cups/d2
Habitual coffee/
caffeine intake
before the trial
Characteristics of participants
Reference
TABLE 1
Clinical trials of coffee/caffeine on blood pressure in individuals with hypertension1
1116
MESAS ET AL
Japan
Japan
Yamaguchi et al
(24), 2008a,b,c,d
Ochiai et al (25),
2009a,b,c,d
R, DB, C, P
R, DB, C, P
R, DB, C, P
Design
21
183
60
3064
3065
2065
Age
M/F
M/F
M/F
Sex
No heavy smokers or heavy
alcohol consumers; no caffeineor coffee-hypersensitive or
allergic persons; no severe liver,
kidney, cerebrovascular, or heart
disease; no endocrine disorders,
metabolic disturbances, diabetes,
or potential pregnancy
No heavy smokers or alcohol
abusers and no history of
medication use for HT,
hyperlipidemia, diabetes
mellitus, stroke, CVD,
or renal or liver dysfunction
Vascular failure and 2 of the
following risk factors related to
hypertension: smoking, waist
circumference, triglycerides,
total cholesterol, lipoprotein,
and blood sugar
Lifestyle/health
status
BP 140159/90
99 mm Hg
BP 140155/90
97 mm Hg
BP 140159/,100
mm Hg
NA
NA
Criteria for
inclusion as
hypertensive
NA
Habitual coffee/
caffeine intake
before the trial
Characteristics of participants
No
No
No
Antihypertensive
medications
a, b, c, and d indicate different protocols in a study and are reported in detail in Figures 2 and 3 and in Table 3. BP, blood pressure; C, controlled; CVD, cardiovascular disease; DB, double-blinded; DBP,
diastolic blood pressure; HT, hypertension; IHD, ischemic heart disease; NA, not available; P, parallel; R, randomized; X, crossover.
2
1 cup = 240 mL.
Japan
Country
Chikama et al
(23), 2008a,b,c,d
Reference
TABLE 1 (Continued )
1117
1118
MESAS ET AL
FIGURE 2. Meta-analysis of the acute effects of caffeine on SBP in hypertensive individuals, by time after caffeine intake. *The net change in SBP is the
difference in SBP between the intervention period and the control period. yWeights are from fixed-effects models. In Potters study, a indicates the protocol
with a long caffeine abstention period before the intervention, and b indicates the protocol with a short caffeine abstention period before the intervention. NA,
not available; SBP, systolic blood pressure.
1119
(2325). The cutoff points to define hypertension were heterogeneous, and the oldest study (20) considered only the DBP
value. The study subjects continued with antihypertensive
medication in 2 studies (22, 28), and they abstained from coffee
and caffeine during 1 or 2 wk beforehand in 4 studies (22, 24,
28).
The longer-term effects of coffee intake on BP are shown in
Table 3. Two of the studies (20, 28) presented BP data only at
the end of the intervention and control periods. Thus, in the
study by MacDonald et al (20), baseline BP was assumed to be
the same as BP at the end of the period with the normal diet. In
the study by Eggertsen et al (28), the effect of caffeine on BP
was estimated by comparing only the final values of BP between
the treatment branches. The duration of the intervention varied
from 2 to 12 wk. No increase in BP was observed when coffee
intake was compared with the caffeine-free diet and when coffee
intake was compared with the decaffeinated coffee.
Three studies examined the effect of drinking coffee containing different amounts of HHQ and/or CGA on BP. Chikama
et al (23) compared the effect of coffee with reduced HHQ with
that of normal coffee. In comparison with normal coffee, the
reduced-HHQ coffee resulted in a decrease in BP after 412 wk
of consumption. In the second study, Yamaguchi et al (24)
studied the effect of coffee without HHQ and with different
FIGURE 3. Meta-analysis of the acute effects of caffeine on DBP in hypertensive individuals, by time after caffeine intake. *The net change in DBP is the
difference in DBP between the intervention period and the control period. yWeights are from random-effects models. In Potters study, a indicates the protocol
with a long caffeine abstention period before the intervention, and b indicates the protocol with a short caffeine abstention period before the intervention. DBP,
diastolic blood pressure; NA, not available.
1120
MESAS ET AL
TABLE 2
Meta-analysis of the acute effects of caffeine on blood pressure in hypertensive individuals, stratified by caffeine intake,
caffeine abstinence before the start of the trial, and use of antihypertensive medication
Systolic blood pressure
Amount of caffeine
200 mg (26)
250 mg (21, 27)
.250300 mg (18, 19)
Caffeine abstinence before the trial
9 h (26)
12 h (18, 19, 21, 27)
48 h (21)
Use of antihypertensive medication
No (18, 19, 21)
Yes (26, 27)
Net change1
95% CI
mm Hg
mm Hg
9.33
9.56
7.40
(2.69, 15.98)
(4.30, 14.83)
(4.32, 10.48)
9.33
7.68
12.99
7.49
9.93
P value2
Net change1
95% CI
P value2
mm Hg
mm Hg
0.958
0.605
4.50
3.11
8.82
(1.06, 7.94)
(0.35, 5.87)
(6.21, 11.43)
0.537
0.050
(2.69, 15.98)
(4.96, 10.41)
(1.18, 24.80)
0.653
0.597
4.50
6.19
3.53
(1.06, 7.94)
(3.87, 8.51)
(22.00, 9.06)
0.353
0.770
(4.60, 10.37)
(5.16, 14.69)
0.391
6.21
4.51
(3.77, 8.65)
(1.82, 7.16)
0.240
DISCUSSION
The net change in blood pressure is the difference in blood pressure between the intervention and the control periods.
Fixed-effects models were used for systolic blood pressure and random-effects models for diastolic blood pressure metaanalysis.
2
Significance of the difference between the categories of the stratification variables, obtained from meta-regression
analyses.
60
60
60
Chikama et al (23)b
Chikama et al (23)c
Chikama et al (23)d
183
183
Yamaguchi et al (24)b
Yamaguchi et al (24)c
183
60
2 wk
23
2 wk
2 wk
2 wk
No
No
No
No
No
2 wk
27
50
No
50
Coffee/decaffeinated
coffee
MacDonald et al
(20)b
Eggertsen et al (28)
Coffee/caffeine-free
diet
MacDonald
et al (20)a
Rakic et al (22)
Reference
34
3
3
cups/d
Instant
Instant
Instant
Instant
Type and
preparation
7581
7581
7581
Instant HHQ-free
coffee with zero
dose of CGA
Instant HHQ-free
coffee and low
dose of CGA
Instant HHQ-free
coffee with
middle dose
of CGA
77 (active group),
Instant coffee with
75 (control group)
reduced HHQ
77 (active group),
Instant coffee with
75 (control group)
reduced HHQ
77 (active group),
Instant coffee with
75 (control group)
reduced HHQ
77 (active group),
Instant coffee with
75 (control group)
reduced HHQ
NA
NA
300
NA
mg/d
Caffeine
doses
12
10
wk
142.8
145.0
143.1
144.7
144.7
144.7
144.7
131.6
134.0
135.4
134.0
mm Hg
Duration
of the
intervention Baseline
86.3
72.3
86.3
80.3
90.5
90.5
90.5
90.5
90.6
91.2
90.6
mm Hg
mm Hg
Baseline
Net change3
(95% CI)
SBP
Coffee/caffeine
No. of
abstinence
subjects before treatment Coffee2
TABLE 3
Longer-term effects (1 wk) of coffee consumption on BP in hypertensive individuals1
Conclusions
Coffee consumption
did not increase BP
Coffee consumption
did not increase BP
(Continued)
HHQ-free coffee
consumption
decreased BP
22.2 (22.8, 21.6) HHQ-free coffee
consumption
decreased BP
21.8 (22.5, 21.1) HHQ-free coffee
consumption
decreased BP
0 (20.7, 0.7)
Coffee consumption
did not increase BP
2.7 (26.6, 12.0) Coffee consumption
increased BP
mm Hg
Net change3
(95% CI)
DBP
1121
21
21
21
Ochiai et al (25)b
Ochiai et al (25)c
Ochiai et al (25)d
No
No
No
No
2 wk
Instant HHQ-free
coffee with high
dose of CGA
Type and
preparation
79 (active group),
Instant coffee with
59 (control group)
reduced HHQ
and elevated
CGA content
79 (active group),
Instant coffee with
59 (control group)
reduced HHQ
and elevated
CGA content
79 (active group),
Instant coffee with
59 (control group)
reduced HHQ
and elevated
CGA content
79 (active group),
Instant coffee with
59 (control group)
reduced HHQ
and elevated
CGA content
7581
Caffeine
doses
147.3
147.3
147.3
147.3
144.1
Duration
of the
intervention Baseline
Baseline
91.2
88.4
88.4
88.4
88.4
Net change3
(95% CI)
23.3 (25.6, 21.1)
SBP
Conclusions
Net change3
(95% CI)
DBP
1
In MacDonald et als study, a and b indicate protocols with different comparisons in the control period. In Yamaguchi et als study, a, b, c, and d indicate protocols with different interventions. In Chikama
et als and Ochiai et als studies, a, b, c and d indicate protocols with different durations. BP, blood pressure; CGA, chlorogenic acid; DBP, diastolic blood pressure; HHQ, hydroxyhydroquinone; NA, not
available; SBP, systolic blood pressure.
2
1 cup = 240 mL or 8 ounces.
3
The net change in BP is the difference in BP changes between the treated group and the control group for parallel trials and the difference in BP between the intervention period and the control period for
crossover trials.
21
183
Coffee/caffeine
No. of
abstinence
subjects before treatment Coffee2
Yamaguchi et al (24)
Reference
TABLE 3 (Continued )
1122
MESAS ET AL
United States, 25 y
Finland, 13.6 y
Hakim
(37), 1998
Greenberg
(38), 2007
Larsson
(40), 2008
Greenberg
(39), 2008
Stage 1: BP
140159/9099 mm
Hg (n = 512); stage
2: BP 160/100 mm
Hg (n = 290)
BP 140/90 mm Hg
(n = 499)
BP 160/100 mm
Hg (n = 302)
M, 5069 y
BP 140/90 mm Hg
(n = 26,556)
(number of
hypertensive
individuals not
reported)
M/F, 3286 y
(n = 6594)
M, 5568 y
(n = 8006)
United States, 4 y
Population
data
Martin
(36), 1988
Country, duration
of follow-up
Coffee consumption:
0 and 1 cups/d
Coffee consumption:
,2, 23, 45, 67,
and 8 cups/d
Caffeinated beverage
consumption: ,1.5
and 1.5 cups/d
Coffee consumption: 0,
48, 1216, and
20 oz/d
Caffeinated beverage
consumption: 0,
.02, .24, or
.4 cups/d
Exposure/categories2
Confounding factors
Cerebral infarction
(n = 1729 cases in
those with SBP 140
mm Hg and 1455
in those with DBP
90 mm Hg)
Thromboembolic stroke
(n = 76)
Outcome (n)
Reference
TABLE 4
Cohort studies of habitual coffee consumption and CVD in hypertensive individuals1
Caffeinated beverage
consumption not
associated with
increased
cerebrovascular or
other CVD mortality
RR of total mortality
for categories of
consumption: 1.0, 0.82
(0.65, 1.03), 0.82 (0.62,
1.82), and 0.90 (0.63, 1.28);
RR of cerebrovascular
mortality for categories
of consumption: 1.0, 0.73
(0.37, 1.46), 0.61 (0.26,
1.44), and 1.30 (0.56, 3.04);
RR of other CVD mortality
for categories of consumption:
1.0, 0.93 (0.66, 1.30), 0.81
(0.53, 1.23), and 0.80
(0.46, 1.39)
RR for consumption of 20
oz/d: 2.1 (1.2, 3.7), P-trend =
0.006 (information for other
categories not available)
(Continued)
RR for consumption of
1.5 cups/d in stage
1 hypertensive subjects:
0.62 (0.39, 0.99); RR for
consumption of 1.5 cups/d
in stage 2 hypertensive
subjects: 0.81 (0.47, 1.41)
RR for categories of
Coffee consumption
consumption in those with
was associated with
SBP 140 mm Hg: 1.0,
lower risk of
0.92 (0.77, 1.10), 0.90
cerebral infarction
(0.76, 1.07), 0.79 (0.65,
0.95), and 0.76 (0.63, 0.93),
P-trend = 0.001; RR categories
of consumption in those with
DBP 90 mm Hg: 1.0, 0.88
(0.72, 1.06), 0.88 (0.73, 1.06),
0.75 (0.61, 0.92), and 0.70
(0.57, 0.87), P-trend , 0.001
Coffee consumption
RR of CHD mortality for
was not associated
consumption of 1 cup/d:
with increased risk
0.87 (0.44, 1.72), P-trend =
of CHD mortality
0.48; RR of heart valve
or heart valve disease
disease for consumption
of 1 cup/d: 1.72 (0.41,
7.25), P-trend = 0.04
Coffee consumption
associated with
increased risk of
thromboembolic
stroke
Caffeinated beverage
consumption not
associated with
increased heart
disease mortality
Conclusions
Multivariate-adjusted results
1123
1124
Cerebral infarction
(n = 482)
F, 4983 y
Self-reported history of Coffee consumption:
(n = 34,670)
hypertension (number
,1, 12, 34, and
of hypertensive
5 cups/d
individuals not
reported)
Sweden, 10.4 y
Larsson
(41), 2011
BP, blood pressure; CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; SBP, systolic blood pressure.
1 cup = 240 mL or 8 ounces.
Coffee consumption
not associated with
increased risk of
cerebral infarction
Coffee consumption
not associated with
increased risk
of stroke
Outcome (n)
Exposure/categories2
BP criteria for inclusion
as hypertensive
Population
data
Country, duration
of follow-up
Reference
TABLE 4 (Continued )
individuals. Finally, longitudinal studies in normotensive individuals (42), and the studies reviewed in this work in hypertensive
individuals, suggest that habitual coffee consumption is not associated with a higher risk of cardiovascular events.
Our work makes some original methodologic contributions to
the study of the acute effect of coffee and caffeine on BP. First,
changes in BP after caffeine intake, from the beginning to the end
of each study, were considered independently for each period of
observation. This allowed evaluation of the temporal course of
the effect of caffeine on BP in the first hours after its administration. Second, we found a suggestion of a dose-response relation between acute caffeine intake and DBP. Furthermore, the
increase in BP was independent of the use of antihypertensive
treatment.
Given the small number and the heterogeneity of the studies, it
was not possible to calculate an overall estimator of the longerterm effect of coffee on BP in hypertensive individuals. However,
no study found a significant increase in BP; moreover, modifying
the concentrations of some components of coffeespecifically
decreasing HHQ and increasing CGAproduced reductions in
BP.
Coffee is a drink with a very complex chemical composition.
Although the cardiometabolic effects of caffeine are well
documented (1, 4345), the effects of other substances in coffee
are not yet well known. Recent studies suggest that some components of coffee, such as CGA and other phenolic compounds,
magnesium, and trigonelline, improve glucose metabolism and
reduce inflammation and endothelial dysfunction (4648). In the
long term, these effects would be seen in habitual coffee drinkers
who have developed tolerance to caffeine. Thus, the harmful
cardiovascular effects of caffeine would be offset by the beneficial effects of these other components. This may help explain
the lack of association between coffee and long-term CVD risk in
large cohort studies, in both normotensive (12, 40, 4951) and
hypertensive (13, 38, 39) persons.
The results of this review have clinical implications for the
control of hypertensive patients. On the one hand, the acute
increase in BP may temporarily increase the risk of a cardiovascular event (10, 11). This is consistent with the increased risk
of coronary disease and stroke in the hours after coffee consumption. Thus, hypertensive patients with uncontrolled BP
should avoid consuming large doses of caffeine. Furthermore, the
consumption of coffee or other caffeinated substances in the
hours before measuring BP may elevate it and give the erroneous
impression that BP is poorly controlled. However, we found no
evidence to justify avoidance of habitual coffee consumption in
well-controlled hypertensive patients; therefore, in these patients,
medical recommendations should prioritize modification of other
lifestyles, such as quitting smoking, controlling weight, and
increasing physical activity.
In conclusion, caffeine raises BP for 3 h after ingestion
in hypertensive persons. However, the available evidence does
not support the assertion that coffee consumption for 2 wk increases BP or that habitual coffee consumption raises the risk of
CVD. Nonetheless, studies of the effect of prolonged coffee
consumption (.2 wk) on BP are needed (Table 3). It is also
necessary to investigate directly the influence of coffee or caffeine on the degree of BP control in hypertensive individuals
and its possible variation with type of antihypertensive medication (eg, renin-angiotensin-aldosterone inhibitors, calcium
Confounding factors
Multivariate-adjusted results
Conclusions
MESAS ET AL
REFERENCES
1125
1126
MESAS ET AL
46. Abel ED. Myocardial insulin resistance and cardiac complications of diabetes. Curr Drug Targets Immune Endocr Metabol Disord 2005;5:21926.
47. Lopez-Garcia E, van Dam RM, Qi L, Hu FB. Coffee consumption and
markers of inflammation and endothelial dysfunction in healthy and
diabetic women. Am J Clin Nutr 2006;84:88893.
48. Gomez-Ruiz JA, Leake DS, Ames JM. In vitro antioxidant activity of coffee
compounds and their metabolites. J Agric Food Chem 2007;55:69629.
49. Andersen LF, Jacobs DR Jr, Carlsen MH, Blomhoff R. Consumption of
coffee is associated with reduced risk of death attributed to inflammatory and cardiovascular diseases in the Iowa Womens Health Study.
Am J Clin Nutr 2006;83:103946.
50. Sofi F, Conti AA, Gori AM, Eliana Luisi ML, Casini A, Abbate R,
Gensini GF. Coffee consumption and risk of coronary heart disease:
a meta-analysis. Nutr Metab Cardiovasc Dis 2007;17:20923.
51. Bidel S, Hu G, Qiao Q, Jousilahti P, Antikainen R, Tuomilehto
J. Coffee consumption and risk of total and cardiovascular mortality
among patients with type 2 diabetes. Diabetologia 2006;49:261826.