Serum Potassium Is Positively Associated With Stroke and

Mortality in the Large, Population-Based Malmö Preventive

Project Cohort
Linda S. Johnson, MD, PhD; Nick Mattsson, MD; Ahmad Sajadieh, MD, DMSc;
Per Wollmer, MD, PhD; Martin Söderholm, MD, PhD

Background and Purpose—Low serum potassium is associated with stroke in populations with cardiovascular disease,
hypertension, and diabetes mellitus but has not been studied in a mainly healthy population. We aimed to study the
relation between serum potassium and incident stroke and mortality in the Malmö Preventive Project, a large cohort with
screening in early mid-life and follow-up >25 years.
Methods—Serum potassium measurements and covariates were available in 21 353 individuals (79% men, mean age 44
years). Mean follow-up time was 26.9 years for stroke analyses and 29.3 years for mortality analyses. There were 2061
incident stroke events and 8709 deaths. Cox regression analyses adjusted for multiple stroke risk factors (age, sex, height,
weight, systolic blood pressure, fasting blood glucose, serum sodium, current smoking, prevalent diabetes mellitus,
prevalent coronary artery disease, and treatment for hypertension) were fitted.
Results—There was an independent, linear association between serum potassium, per mmol/L increase, and both stroke
(hazard ratio, 1.33; 95% confidence interval, 1.17–1.52; P<0.0001) and mortality (hazard ratio, 1.20; 95% confidence
interval, 1.13–1.28; P<0.0001). This was significant in subjects both older and younger than the median age (46.5 years),
and there was evidence of an interaction with serum sodium. The association was positive and significant for both
ischemic stroke and intracerebral hemorrhage and in both hypertensive and normotensive subjects.
Conclusions—Serum potassium, measured in early mid-life, was linearly associated with both incidence of ischemic stroke
and intracerebral hemorrhage and all-cause mortality. An interaction with serum sodium implies that factors related to
electrolyte balance and incident hypertension may be mediating factors.   (Stroke. 2017;48:2973-2978. DOI: 10.1161/
STROKEAHA.117.018148.)
Downloaded from http://ahajournals.org by on May 25, 2021

Key Words: epidemiology ◼ mortality ◼ population ◼ potassium ◼ risk factor ◼ sodium ◼ stroke

L ow, as well as high, serum potassium (s-potassium) is

associated with increased mortality in hypertensive sub-
jects.1 Green et al2 have reported that low s-potassium is asso-
ciated with increased risk of stroke among elderly users of
diuretics in the Cardiovascular Health Study, and Smith et al3
have found low s-potassium to be associated with ischemic
and hemorrhagic stroke in patients with treated hypertension.
Furthermore, diets rich in fruit and vegetables, and thereby
potassium, are inversely associated with stroke risk4,5 and
stroke mortality.6 The authors of a recent meta-analysis of
potassium intake and stroke risk found a partly blood pres-
sure–dependent inverse association between potassium intake
and stroke risk,7 and high urinary sodium and low urinary
potassium excretion have both been reported to be associ-
ated with increased risk of stroke in patients with established
cardiovascular disease or diabetes mellitus.8 High potassium
intake has also been associated with a reduced risk of stroke in
hypertensive, but not normotensive, women.9 In light of this, it
seems plausible to assume that there could be a link between
low potassium intake, low s-potassium, and later stroke.
S-Potassium is regulated via aldosterone secretion, which
is closely related to dietary sodium intake, and may also be
affected by renin–angiotensin–aldosterone system activation.
High dietary sodium leads to reduced aldosterone release,
which results in reduced excretion of potassium. It is possi-
ble, therefore, that the association between dietary potassium
intake, s-potassium, and stroke may differ between hyperten-
sive and normotensive subjects and, perhaps, also between
younger and older subjects. Furthermore, s-potassium within
the normal range has been shown to have a positive association
with all-cause mortality in the National Health and Nutrition
Examination Study, Multi-Ethnic Study of Atherosclerosis,
Cardiovascular Health Study, and Atherosclerosis Risk in
Communities Study cohorts,10–13 and the effect of the com-
peting risk of death on the relationship between s-potassium
and stroke has not been fully addressed. To our best knowl-
edge, the association between s-potassium and incident stroke
has not been studied in a largely nonhypertensive, healthy

Received May 23, 2017; final revision received August 19, 2017; accepted August 23, 2017.
From the Department of Clinical Sciences Malmö, Lund University, and Skåne University Hospital, Sweden (L.S.J., M.S.); Department of Cardiology,
Copenhagen University Hospital of Bispebjerg, Denmark (N.M., A.S.); and Department of Translational Medicine, Lund University, and Skåne University
Hospital, Malmö, Sweden (P.W.).
Correspondence to Linda S. Johnson, MD, PhD, Inga-Marie Nilssons gata 49, 20502 Malmö, Sweden. E-mail linda.johnson@med.lu.se
© 2017 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.117.018148

2973
 2974  Stroke  November 2017
population, and the effect of serum sodium (s-sodium) on the
association has not been assessed.
The Malmö Preventive Project (MPP) is a large prospective
cohort with a long follow-up time. The mean age at screen-
ing is comparatively young (44.8 years), a large proportion of
the population is normotensive, and only 4% were using anti-
hypertensive medication at baseline. We aimed to model the
association between s-potassium and incident stroke, as well
as all-cause mortality in the MPP cohort, including assess-
ment of interaction by other stroke risk factors, as well as
s-sodium. Another aim was to assess the association between
s-potassium and stroke subtypes. We also wished to determine
whether the competing risk of death influenced the association
between s-potassium and stroke.
Materials and Methods
Study Population
The MPP cohort consists of 22 444 men and 10 902 women who
participated in a health examination intended to detect high-risk
individuals and provide preventive intervention. The screening
activities included physical examination, blood samples, and an
extensive self-administered questionnaire. Subjects were recruited
through invitation of prespecified full birth-year groups from the
city of Malmö in southern Sweden, age range 21 to 61 years. The
attendance rate was over 70%.14 Men were screened mostly during
the years 1974 to 1982 and women during 1982 to 1992, resulting
in different lengths of follow-up, and the screening activities per-
formed varied somewhat over different calendar years. S-Potassium
was measured mainly in the years 1974 to 1980 and 1989 to 1992,
resulting in a higher proportion of males in the present study
(80.1%) compared with the full MPP cohort (67.3%). Interventions,
Downloaded from http://ahajournals.org by on May 25, 2021
in terms of lifestyle advice and drug therapy available at the time,
were offered to around 25% of the screened population, but these
measures had no effect on incident cardiovascular mortality or mor-
bidity in general.15 S-Potassium was available for 17 243 men and
4912 women. From these subjects, we excluded cases of prevalent
stroke at baseline (n=15), missing data for systolic blood pressure
(n=17), body mass index (n=4), s-creatinine (n=60), fasting blood
glucose (n=99), current smoking status (n=618), s-sodium (n=4),
or s-cholesterol (n=27). The final study population (n=21 326) con-
sisted of 17 076 men and 4250 women.
Data Collection
Height (m) and weight (kg) were measured in light indoor cloth-
ing, without shoes, using a fixed stadiometer and balance beam
scale, respectively. Body mass index was calculated as weight (kg)/
height (m)2. Blood pressure (mm Hg) was measured twice after 10
minutes of supine rest using a sphygmomanometer with a modifiable
cuff. Blood samples were drawn after overnight fast and analyzed
using standard laboratory procedures at the Department of Clinical
Chemistry at the Malmö University Hospital. Fasting blood glucose
was analyzed using the glucose-oxidase method (1974–1977) or the
hexokinase-oxidase method (1977–1992). These methods give sim-
ilar results and were used without a conversion factor. Individuals
who reported that they were current smokers were classified as such.
Alcohol risk use was classified as ≥2 positive answers to a modi-
fication of the Michigan Alcohol Screening Test (Mm-Mast)16; data
were available for 19 794 individuals. Sedentary lifestyle was defined
as a positive answer to the question, “Are you mostly sedentary in
your spare time?”, and these data were available for 19 951 partici-
pants. Use of antihypertensive medication was defined as a positive
answer to the question, “Do you use medication for high blood pres-
sure?”. Prevalent coronary event was defined using International
Classification of Diseases code 410* in the ninth revision or I21 in the
tenth revision. Prevalent diabetes mellitus was defined as a baseline
fasting blood glucose ≥6.1 mmol/L or a prior diagnosis of diabetes
mellitus retrieved from multiple local and national registers. Diabetes
mellitus ascertainment has been described in detail elsewhere.17
The study complies with the declaration of Helsinki and has been
approved by the regional ethics review board. Subjects gave oral con-
sent, and the need for written consent was waived by the ethics review
board.15
Endpoint Retrieval
Cases were retrieved through linkage with the stroke register of
Malmö18 and the Swedish National Hospital Discharge Register
(SNHDR; International Classification of Diseases codes 430, 431, 434,
and 436 for the ninth version and I60, I61, I63, and I64 for the tenth
version), of which the latter is administered by The Swedish National
Board of Health and Welfare. The stroke register of Malmö was started
in 1989 and has been used to study the incidence of stroke in Malmö.
A specialized research nurse has systematically searched for cases
of stroke in both inpatient and outpatient departments at the Malmö
University Hospital, which is the only hospital in the city. All diagno-
ses have been validated by review of medical records and sometimes
also patient interviews. Stroke is defined in accordance with the World
Health Organization.18 Ischemic stroke was diagnosed when brain
imaging or autopsy showed an infarction corresponding to the clini-
cal neurological deficit or excluded hemorrhage and nonvascular dis-
ease. Intracerebral hemorrhage (ICH) was considered when imaging or
autopsy showed intraparenchymal blood in the brain, and subarachnoid
hemorrhage (SAH) diagnosis was based on imaging, autopsy, or lum-
bar puncture. If neither imaging nor autopsy was performed, the stroke
was classified as unspecified (International Classification of Diseases-
Ninth Revision, 436; International Classification of Diseases-Tenth
Revision, I64). Further details of case finding and definitions of stroke
types have been described previously.18–20 Seventy-five percent of all
stroke cases in the present study were retrieved from the stroke regis-
ter of Malmö. Stroke events that occurred before 1989, or in hospitals
outside of Malmö, were identified in the SNHDR, wherein diagnoses
are made by physicians in routine care and approved by board-certified
specialists, and which has been shown to have high coverage and over-
all validity.21 The end of follow-up for stroke analyses was December
31, 2010, resulting in a mean follow-up time (SD) of 26.9 (7.9) years.
The Swedish Cause of Death register was used to assess mortality rate.
End of follow-up for mortality analyses was December 31, 2013. Mean
follow-up (SD) in mortality analyses was 29.4 (8.5) years.
Statistics
Hazard ratios for incident stroke and mortality were estimated with
Cox proportional hazards regression. Two prespecified statistical
models were used based on plausible confounders of any associa-
tion between s-potassium and stroke. The prespecified approach was
chosen to avoid overfitting the model, while at the same time includ-
ing all relevant possible confounders. Model 1 adjusts for age and
sex, and model 2 additionally adjusts for height, weight, systolic
blood pressure, fasting blood glucose (transformed by the natural
logarithm), s-sodium, s-creatinine, s-cholesterol, current smok-
ing, prevalent diabetes mellitus, prevalent coronary artery disease,
and treatment for hypertension. A third model adjusting for alcohol
risk use, self-reported history of angina, and sedentary lifestyle was
tested, but these variables did not substantially affect results and were
excluded from the final model. The proportional hazards assumption
was assessed with –log-log plots and was found to be valid.
All covariates were tested for interaction with serum potassium.
Stratified data were reported for s-sodium (P for interaction =0.045
in stroke analyses and 0.003 in mortality analyses). The P value for
all other interaction parameters was >0.10.
Results
Mean age (SD) at first stroke event was 68.2 (8.3) years, and
mean age at death was 70.4 (9.8) years. Selected baseline
characteristics are reported in Table 1. Briefly, 80.1% of the
study population were male, and the mean age (SD) was 44.6
 Johnson et al   Serum Potassium, Stroke, and Mortality    2975

Table 1.  Baseline Characteristics formula. The association was statistically significant and simi-
lar in each tertile (data not shown).
All Men Women
N (%) 21 326 17 076 (80.1) 4250 (19.9)
S-Potassium and Stroke Subtypes
Age, y 44.6 (6.5) 44.1 (5.4) 46.6 (9.4) We also analyzed serum potassium in relation to stroke sub-
Height, cm 174 (8) 177(7) 164 (6) types. S-Potassium, per mmol/L, was independently associ-
ated with both ischemic stroke (HR, 1.30; 95% CI, 1.12–1.50;
Weight, kg 74.8 (12.7) 77.4 (11.5) 64.2 (11.5)
P=0.001) and hemorrhagic (ICH and SAH) stroke (HR, 1.42;
Body mass index, kg/m 2
24.5 (3.5) 24.7 (3.3) 23.9 (4.2) 95% CI, 1.03–1.96; P=0.035) after model 2 adjustment. The
Systolic blood pressure, mm Hg 126 (15) 127 (15) 122 (16) proportion of stroke types was similar in validated and nonval-
Serum potassium, mmol/L 4.1 (0.3) 4.1 (0.3) 4.1 (0.3) idated stroke events, but the proportion of unspecified stroke
was higher (12% versus 2%) in nonvalidated stroke events.
Serum sodium, mmol/L 141 (2.6) 141(2.6) 140 (2.4)
Results were largely unchanged when only validated ischemic
Fasting blood glucose, mmol/L* 4.9 (0.8) 4.9 (0.7) 4.8 (0.6) and hemorrhagic stroke events were included. We also ana-
Serum cholesterol, mmol/L 5.7 (1.1) 5.6 (1.4) 5.6 (1.5) lyzed ICH and SAH separately. The association was similar
for ICH (HR, 1.49; 95% CI, 1.02–2.16; P=0.04) but weakened
Serum creatinine, μmol/L 90 (18) 93 (18) 77 (12)
for SAH (HR, 1.19; 95% CI, 0.61–2.30; P=0.61), compared
Current smokers, % 48.7 50.4 41.8 with the whole group of hemorrhagic stroke, both in model 2.
Alcohol risk use, % 29.8 35.0 11.1
Sedentary lifestyle, % 55.3 55.3 55.6 Normotensive and Hypertensive Subjects
Furthermore, we also tested the association between s-potas-
Prevalent coronary event, % 0.3 0.4 0.2
sium and stroke and mortality in subgroups of hypertensive
Prevalent diabetes mellitus, % 3.3 3.4 3.1 and normotensive subjects. Hypertension was defined as either
Prevalent heart failure, % 0.00 0.01 0.00 a systolic blood pressure ≥140 mm Hg, diastolic blood pressure
Use of antihypertensive drugs, % 4.3 3.8 6.4 ≥90 mm Hg, or a positive response to the question, “Are you
using medication for hypertension?”. The definition resulted
Use of diuretics, % 2.1 1.2 4.5
in identification of 8699 hypertensive subjects (40.8%). After
All values are mean (SD) unless stated otherwise. model 2 adjustment, there was a significant association between
*Presented as median (interquartile range).
s-potassium, per mmol/L increase, and incident stroke among
Downloaded from http://ahajournals.org by on May 25, 2021

normotensive subjects (HR, 1.28; 95% CI, 1.05–1.57; P=0.01),

(6.5) years. Mean (SD) systolic blood pressure was 126 (15)
mm Hg, and <5% of the population were using antihyperten-
sive drugs. The range of s-potassium was 2.3 to 8.4 mmol/
L (quartile 1, 2.3–3.8 mmol/ L; quartile 2, 3.9–4.0 mmol/L;
quartile 3, 4.1–4.2 mmol/L; and quartile 4, 4.3–8.4 mmol/L).
S-Potassium, Stroke Incidence, and Mortality
A total of 2061 stroke events occurred, of which 1620 were
ischemic strokes, 240 were ICH, 89 were SAH, and 112
were unspecified. There were significant positive associa-
tions between s-potassium and incident stroke and mortality
(Table 2); both P for trend across quartiles of s-potassium were
<0.0001. There was no evidence of a quadratic trend in the
full population (P=0.46 for stroke and 0.44 for mortality). The
association between s-potassium and stroke and mortality was
also tested in subgroups split at the median age of 46.5 years,
using adjustment for model 2 covariates. Mean age (SD) was
39.7 (5.1) years in the younger group and 49.5 (3.2) in the
older group. The associations, per mmol/L increase, were pos-
itive and statistically significant for both stroke (hazard ratio
[HR], 1.50; 95% confidence interval [CI], 1.18–1.92; P=0.001
in the younger subgroup and HR, 1.28; 95% CI, 1.09–1.49;
P=0.002 in the older subgroup) and death (HR, 1.32; 95%
CI, 1.16–1.49; P<0.0001 in the younger subgroup and HR,
1.16; 95% CI, 1.07–1.25; P<0.0001 in the older subgroup) in
both age groups. We also analyzed the association between
s-potassium and incident stroke across tertiles of estimated
glomerular filtration rate, as estimated by the Cockcroft–Gault
as well as among hypertensive subjects (HR, 1.36; 95% CI,
1.14–1.61; P=0.001). S-Potassium, per mmol/L increase,
was likewise associated with mortality among normotensive
(HR, 1.22; 95% CI, 1.12–1.34; P<0.0001) and hypertensive
subjects alike (HR, 1.17; 95% CI, 1.07–1.28; P=0.001) after
model 2 adjustment. To rule out confounding by the use of
antihypertensive drugs or diuretics on the association between
s-potassium and incident stroke, we also performed a subanal-
ysis wherein all subjects who reported use of either diuretics
or antihypertensive medications were excluded. Results were
largely unchanged (data not shown).
Competing Risks Analyses
Because stroke and mortality were both associated with
s-potassium, we considered it relevant to test the associa-
tion between s-potassium and stroke independently of the
competing risk of death, using competing risks regression
as described by Fine and Gray.22 S-potassium was associated
with stroke, independently of the competing risk of death;
subhazard ratio 1.27 (95% CI, 1.12–1.43; P<0.0001) after
model 2 adjustment.
Interaction With S-Sodium
Table 3 presents the association between s-potassium and
stroke and mortality by quartiles of s-sodium. The association
was not significant for either endpoint in the top quartile of
s-sodium and not significant for stroke in the bottom quartile
of s-sodium.
 2976  Stroke  November 2017

Table 2.  Cox Regression Analyses for Serum Potassium and Incident Stroke and Mortality,
Stratified on Screening Year
Model 1* Model 2†
HR 95% CI P Value HR 95% CI P Value
All stroke events
 Per mmol/L increase 1.34 1.18–1.52 <0.0001 1.33 1.17–1.52 <0.0001
 S-Potassium Q1 1 1
 S-Potassium Q2 1.03 0.89–1.19 0.70 1.06 0.92–1.22 0.43
 S-Potassium Q3 1.14 0.99–1.32 0.06 1.16 1.01–1.34 0.04
 S-Potassium Q4 1.35 1.17–1.54 <0.0001 1.34 1.17–1.53 <0.0001
Ischemic stroke
 Per mmol/L increase 1.34 1.16–1.54 <0.0001 1.31 1.14–1.51 <0.0001
 S-Potassium Q1 1 1
 S-Potassium Q2 1.00 0.86–1.17 0.69 1.03 0.88–1.20 0.76
 S-Potassium Q3 1.09 0.93–1.27 0.28 1.09 0.94–1.28 0.26
 S-Potassium Q4 1.33 1.15–1.55 <0.0001 1.30 1.12–1.52 <0.0001
Hemorrhagic stroke
 Per mmol/L increase 1.34 0.97–1.85 0.08 1.42 1.03–1.96 0.04
 S-Potassium Q1 1 1
 S-Potassium Q2 1.20 0.81–1.68 0.34 1.28 0.88–1.85 0.19
 S-Potassium Q3 1.45 0.99–2.01 0.04 1.57 1.09–2.25 0.01
 S-Potassium Q4 1.43 0.98–1.97 0.05 1.54 1.04–2.17 0.02
Mortality
Downloaded from http://ahajournals.org by on May 25, 2021

 Per mmol/L increase 1.31 1.23–1.39 <0.0001 1.20 1.12–1.28 <0.0001

 S-Potassium Q1 1 1
 S-Potassium Q2 1.07 1.00–1.14 0.07 1.04 0.97–1.12 0.24
 S-Potassium Q3 1.17 1.09–1.25 <0.0001 1.10 1.02–1.17 0.009
 S-Potassium Q4 1.29 1.21–1.38 <0.0001 1.16 1.08–1.24 <0.0001
Includes 21 326 individuals, 2061 stroke events, and 8697 deaths. One-thousand six-hundred and twenty stroke events
were classified as ischemic and 329 were classified as hemorrhagic strokes (intracerebral hemorrhage or subarachnoid
hemorrhage). Quartiles of s-potassium: quartile 1, 2.3–3.8 mmol/ L; quartile 2, 3.9–4.0 mmol/L; quartile 3, 4.1–4.2
mmol/L; and quartile 4, 4.3–8.4 mmol/L. CI indicates confidence interval; HR, hazard ratio; and Q, quartile.
*Adjusted for age and sex.
†Adjusted for model 1 + height, body mass index, systolic blood pressure, fasting blood glucose, serum sodium,
current smoking, serum creatinine, serum cholesterol, prevalent diabetes mellitus, prevalent coronary artery disease, and
treatment for hypertension.

Discussion cardiovascular disease, hypertension, and diabetes mellitus and

This is to our best knowledge the first study of S-potassium and
stroke that has been conducted in a relatively young population
mostly free of cardiovascular disease and hypertension. We found
s-potassium, including in the normal range, to be linearly associ-
ated with increased incidence of stroke, after extensive adjust-
ment for potential confounders.1–3,23 Authors of previous studies
have reported that hyperkalemia is associated with increased
risk of death.10–13 The present study confirms this and also shows
that the positive association between s-potassium and stroke is
independent of the competing risk of death. Compared with pre-
vious studies of s-potassium and stroke, the MPP is screened
at a younger age and followed for a longer time. The result is
a population with a comparatively low baseline prevalence of
a low prevalence of hypertensive treatment. One can speculate
that this constitutes the reason why the results of the present
study differ from previous studies of the subject. We were also
able to assess the association between s-potassium and stroke
subtypes. S-Potassium was independently associated with both
ischemic and hemorrhagic stroke. When the analysis of hemor-
rhagic strokes was split into ICH and SAH, the association was
significant for ICH. The higher proportion of unspecified stroke
events in the nonvalidated sample than the validated sample
would likely bias the results toward null rather than exaggerate
any associations between s-potassium and stroke subtypes.
The results of the present study are not obviously consistent
with other studies in which high intake of dietary potassium
 Johnson et al   Serum Potassium, Stroke, and Mortality    2977

Table 3.  Cox Regression Analyses for Incident Stroke and include a significant proportion of dehydrated subjects in whom
Mortality Stratified on Screening Year, per mmol/L Increase in high potassium simply constitutes a marker of dehydration. The
Serum Potassium and Presented by Quartile of Serum Sodium bottom quartile of s-sodium could be composed of individuals
Model 1* Model 2† with a low sodium consumption, in whom high sodium excre-
tion has not induced potassium conservation. In these subjects,
HR 95% CI P Value HR 95% CI P Value
the s-potassium levels could be more closely related to dietary
Stroke potassium, which has beneficial effects on stroke risk.
 Q1‡ 1.41 1.01–1.95 0.04 1.24 0.89–1.73 0.21 We confirm previous reports of a positive association
 Q2§ 1.72 1.36–2.18 <0.0001 1.76 1.37–2.25 <0.0001
between s-potassium and mortality.10–13 The present study has
also permitted stratification on hypertension status, and results
 Q3‖ 1.28 1.00–1.65 0.05 1.30 1.01–1.68 0.04
were largely independent of this. The low mean age of the
 Q4¶ 1.17 0.93–1.47 0.22 1.13 0.90–1.42 0.31 cohort implies that the duration of any hypertension is prob-
Mortality able to be rather low, and differences in the association between
s-potassium and stroke or mortality among hypertensives with a
 Q1‡ 1.70 1.44–2.00 <0.0001 1.45 1.23–1.72 <0.0001
longer duration of elevated blood pressure cannot be ruled out.
 Q2§ 1.39 1.23–1.57 <0.0001 1.22 1.07–1.40 0.003 The findings of the present study have clinical relevance
 Q3‖ 1.25 1.11–1.42 <0.0001 1.20 1.05–1.36 0.005 because they imply that s-potassium should not be used as a
 Q4¶ 1.21 1.08–1.34 0.001 1.08 0.97–1.20 0.18 predictive marker for stroke or mortality without consideration
of the source population. Furthermore, the present study high-
CI indicates confidence interval; and HR, hazard ratio.
*Adjusted for age and sex.
lights the need for further studies on the subject of mechanisms
†Adjusted for model 1 + height, body mass index, systolic blood pressure, linking s-potassium to stroke incidence in diverse populations.
fasting blood glucose, current smoking, serum creatinine, serum cholesterol,
prevalent diabetes mellitus, prevalent coronary artery disease, and treatment
Strengths and Limitations
for hypertension.
‡Includes 3785 individuals, 364 stroke events, and 1463 deaths.
The MPP cohort is large and population based, with a high
§Includes 5774 individuals, 530 stroke events, and 2130 deaths. attendance rate of invited subjects. Data regarding ethnic-
‖Includes 6017 individuals, 559 stroke events, and 2403 deaths. ity were not available. At the time of screening, Malmö was
¶Includes 5750 individuals, 608 stroke events, and 2701 deaths. an ethnically homogenous white population, and results from
the present study may not be generalizable to other ethnicities.
has been associated with reduced risk of stroke.5–9 However, the S-Potassium was not measured all screening years, but because
Downloaded from http://ahajournals.org by on May 25, 2021

association between potassium intake and s-potassium is weak24
and influenced by sodium intake. High dietary sodium leads
to plasma volume expansion and reduced aldosterone release,
which results in reduced potassium excretion. It has been argued
that the kidney has evolved in the context of a high potassium,
low sodium diet and is more effective at excreting excess potas-
sium and conserving sodium than the opposite.25 The western
diet is a high-sodium, low-potassium diet.26 Because of reduced
aldosterone release, a high dietary sodium consumption would
result in higher s-potassium. The association of s-potassium and
incident stroke could, therefore, be because of s-potassium serv-
ing as a marker of sodium consumption rather than any direct
effect of s-potassium on stroke risk. Dietary intake of sodium is
associated with increased blood pressure,27,28 and intermediate
hypertension is, thus, a possible explanation for the association
with s-potassium and stroke. The finding that s-potassium is
independently associated with ICH strengthens this argument.
Hypertension is the most important risk factor for ICH,18 and
increased blood pressure caused by dietary sodium is a plau-
sible explanation for the link between s-potassium and ICH in
this study, although there may also be other mechanisms, which
should be addressed in future studies.
We found an interaction with s-sodium and s-potassium
and the risk of stroke and mortality, a subject that, to our best
knowledge, has not been studied previously. The association
between s-potassium and stroke was not present in either the
top or bottom quartile of s-sodium. The reasons behind this
interaction cannot be deduced from this observational study,
but one might speculate that the top quartile of s-sodium may
no individual selections of which subjects were screened were
made, this should not have introduced any bias. The study
sample is roughly two thirds male, but there was no evidence
of interaction between s-potassium and sex, and considering
that >4000 women were included, we consider the results to be
valid in both sexes. The study population is larger than previous
studies of the subject, and the follow-up time, via national reg-
istries, is long. Because of the fact that all residents of Sweden
have a unique personal identification number, there is no loss
to follow-up at the time of linkage with registers. These are all
strengths. Most stroke cases were retrieved from the local stroke
register of Malmö, which has been used for studies of stroke
incidence and has a high detection rate of both hospitalized and
nonhospitalized nonfatal and fatal stroke.18,20 All diagnoses have
been validated by review of hospital records. Using this local
stroke register as the main source of case finding is also a major
strength. To identify cases of stroke occurring before 1989, or
in hospitals outside of Malmö, the national hospital discharge
register, SNHDR, was used (25% of all stroke in the present
study). Review of medical records also of these cases would
have been preferable. However, studies have shown that the
detection rate is high and validity is reasonable for stroke in the
SNHDR.21 The proportion of stroke types was similar in cases
from the local stroke register and from the SNHDR, although
the proportion of unspecified stroke was somewhat higher in the
SNHDR, as expected. A sensitivity analysis that included only
validated stroke events showed largely unchanged results for
the association between s-potassium and stroke. Stroke cases
could potentially have been missed in the case of fatal cases
 2978  Stroke  November 2017
occurring out of hospital. However, review of autopsy records
from patients dying suddenly outside of hospital has shown that
fatal stroke cases occurring outside of hospital accounted for
<1% of all cases of stroke in the stroke register of Malmö.20
We have adjusted for many potential confounders and
found results to be largely independent of these. However,
residual confounding is always a potential limitation and
cannot be entirely ruled out. Some factors associated with
s-potassium, s-sodium, and intermediate causes, such as
hypertension, were not available for adjustment, most nota-
bly s-aldosterone measurements. Further studies that include
measures of renin–angiotensin–aldosterone system activa-
tion would be interesting. Dietary sodium and potassium
intake, renin–angiotensin–aldosterone system activation, and
s-sodium and s-potassium are inter-related, and there is a risk
of bias in studying such factors. Individuals with poor health
may choose to consume less sodium, introducing reverse cau-
sality. However, the duration of follow-up was long, and the
population was mainly free of cardiovascular disease, which
is a strength, in that such bias should be minimized. Only one
measurement of s-potassium was available, and intraindivid-
ual variation is not accounted for. This is a limitation, but any
bias introduced by this would have been toward null.
Conclusions
S-Potassium, including in the normal range, is positively and
linearly associated with incident stroke and death among sub-
jects screened in early mid-life and followed for >25 years.
The risk of stroke is independent of the competing risk of
Downloaded from http://ahajournals.org by on May 25, 2021
death.
Sources of Funding
Drs Johnson and Söderholm are supported by governmental funding
within the Swedish National Health Services. Dr Söderholm is sup-
ported by funds from the Swedish Stroke foundation. Dr Wollmer is
supported by the Swedish Heart and Lung Foundation.
Disclosures
Dr Wollmer’s wife works for GE Healthcare. We do not think that
this association has influenced results in any way. The other authors
report no conflicts.
References
1. Krogager ML, Eggers-Kaas L, Aasbjerg K, Mortensen RN, Køber L,
Gislason G, et al. Short-term mortality risk of serum potassium lev-
els in acute heart failure following myocardial infarction. Eur Heart
J Cardiovasc Pharmacother. 2015;1:245–251. doi: 10.1093/ehjcvp/
pvv026.
2. Green DM, Ropper AH, Kronmal RA, Psaty BM, Burke GL;
Cardiovascular Health Study. Serum potassium level and dietary potas-
sium intake as risk factors for stroke. Neurology. 2002;59:314–320.
3. Smith NL, Lemaitre RN, Heckbert SR, Kaplan RC, Tirschwell DL,
Longstreth WT, et al. Serum potassium and stroke risk among treated
hypertensive adults. Am J Hypertens. 2003;16:806–813.
4. Iso H, Stampfer MJ, Manson JE, Rexrode K, Hennekens CH, Colditz
GA, et al. Prospective study of calcium, potassium, and magnesium
intake and risk of stroke in women. Stroke. 1999;30:1772–1779.
5. Gillman MW, Cupples LA, Gagnon D, Posner BM, Ellison RC, Castelli
WP, et al. Protective effect of fruits and vegetables on development of
stroke in men. JAMA. 1995;273:1113–1117.
6. Khaw KT, Barrett-Connor E. Dietary potassium and stroke-associated
mortality. A 12-year prospective population study. N Engl J Med.
1987;316:235–240. doi: 10.1056/NEJM198701293160502.
7. Larsson SC, Orsini N, Wolk A. Dietary potassium intake and risk of
stroke: a dose-response meta-analysis of prospective studies. Stroke.
2011;42:2746–2750. doi: 10.1161/STROKEAHA.111.622142.
8. O’Donnell MJ, Yusuf S, Mente A, Gao P, Mann JF, Teo K, et al. Urinary
sodium and potassium excretion and risk of cardiovascular events.
JAMA. 2011;306:2229–2238. doi: 10.1001/jama.2011.1729.
9. Larsson SC, Virtamo J, Wolk A. Potassium, calcium, and magnesium
intakes and risk of stroke in women. Am J Epidemiol. 2011;174:35–43.
doi: 10.1093/aje/kwr051.
10. Fang J, Madhavan S, Cohen H, Alderman MH. Serum potassium and
cardiovascular mortality. J Gen Intern Med. 2000;15:885–890.
11. Hughes-Austin JM, Rifkin DE, Beben T, Katz R, Sarnak MJ, Deo R, et
al. The relation of serum potassium concentration with cardiovascular
events and mortality in community-living individuals. Clin J Am Soc
Nephrol. 2017;12:245–252. doi: 10.2215/CJN.06290616.
12. Chen Y, Chang AR, McAdams DeMarco MA, Inker LA, Matsushita
K, Ballew SH, et al. Serum potassium, mortality, and kidney outcomes
in the atherosclerosis risk in communities study. Mayo Clin Proc.
2016;91:1403–1412.
13. Loprinzi PD, Hall ME. Effect of serum potassium on all-cause mortality
in the general us population. Mayo Clin Proc. 2017;92:320.
14. Nilsson PM, Nilsson JA, Berglund G. Population-attributable risk of
coronary heart disease risk factors during long-term follow-up: the
Malmö Preventive Project. J Intern Med. 2006;260:134–141. doi:
10.1111/j.1365-2796.2006.01671.x.
15. Berglund G, Nilsson P, Eriksson KF, Nilsson JA, Hedblad B, Kristenson
H, et al. Long-term outcome of the Malmö preventive project: mortality
and cardiovascular morbidity. J Intern Med. 2000;247:19–29.
16. Kristenson H, Trell E. Indicators of alcohol consumption: comparisons
between a questionnaire (Mm-MAST), interviews and serum gamma-
glutamyl transferase (GGT) in a health survey of middle-aged males. Br
J Addict. 1982;77:297–304.
17. Enhörning S, Sjögren M, Hedblad B, Nilsson PM, Struck J, Melander
O. Genetic vasopressin 1b receptor variance in overweight and diabetes
mellitus. Eur J Endocrinol. 2016;174:69–75. doi: 10.1530/EJE-15-0781.
18. Zia E, Hedblad B, Pessah-Rasmussen H, Berglund G, Janzon L,
Engström G. Blood pressure in relation to the incidence of cerebral
infarction and intracerebral hemorrhage. Hypertensive hemorrhage:
debated nomenclature is still relevant. Stroke. 2007;38:2681–2685. doi:
10.1161/STROKEAHA.106.479725.
19. Söderholm M, Zia E, Hedblad B, Engström G. Lung function as a risk
factor for subarachnoid hemorrhage: a prospective cohort study. Stroke.
2012;43:2598–2603. doi: 10.1161/STROKEAHA.112.658427.
20. Pessah-Rasmussen H, Engström G, Jerntorp I, Janzon L. Increasing
stroke incidence and decreasing case fatality, 1989-1998: a study from
the stroke register in Malmö, Sweden. Stroke. 2003;34:913–918. doi:
10.1161/01.STR.0000063365.10841.43.
21. Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL,
Reuterwall C, et al. External review and validation of the Swedish
national inpatient register. BMC Public Health. 2011;11:450. doi:
10.1186/1471-2458-11-450.
22. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of
a competing risk. J Am Stat Assoc. 1999;94:496–509.
23. Alderman MH, Piller LB, Ford CE, Probstfield JL, Oparil S, Cushman
WC, et al; Antihypertensive and Lipid-Lowering Treatment to Prevent
Heart Attack Trial Collaborative Research Group. Clinical significance
of incident hypokalemia and hyperkalemia in treated hypertensive
patients in the antihypertensive and lipid-lowering treatment to pre-
vent heart attack trial. Hypertension. 2012;59:926–933. doi: 10.1161/
HYPERTENSIONAHA.111.180554.
24. Cappuccio FP, Buchanan LA, Ji C, Siani A, Miller MA. Systematic
review and meta-analysis of randomised controlled trials on the effects of
potassium supplements on serum potassium and creatinine. BMJ Open.
2016;6:e011716. doi: 10.1136/bmjopen-2016-011716.
25. He FJ, MacGregor GA. Beneficial effects of potassium on human health.
Physiol Plant. 2008;133:725–735.
26. Aaron KJ, Sanders PW. Role of dietary salt and potassium intake in
cardiovascular health and disease: a review of the evidence. Mayo Clin
Proc. 2013;88:987–995.
27. Stamler J. The INTERSALT Study: background, methods, findings, and
implications. Am J Clin Nutr. 1997;65(2 suppl):626S–642S.
28. Takase H, Sugiura T, Kimura G, Ohte N, Dohi Y. Dietary sodium con-
sumption predicts future blood pressure and incident hypertension
in the Japanese normotensive general population. J Am Heart Assoc.
2015;4:e001959. doi: 10.1161/JAHA.115.001959.