Serum Potassium and Risk of Cardiovascular Disease: The Framingham Heart Study

Background  Published studies of the association between serum potassium concentration and risk for cardiovascular disease in community-based populations have reported conflicting results. We sought to determine the association between serum potassium concentration and cardiovascular disease risk in the Framingham Heart Study.

Methods  A total of 3151 participants (mean age, 43 years; 48% men) in the Framingham Heart Study who were free of cardiovascular disease and not taking medications affecting potassium homeostasis had serum potassium levels measured (1979-1983). Proportional hazards models were used to determine the association of serum potassium concentration at baseline with the incidence of cardiovascular disease at follow-up.

Results  During mean follow-up of 16 years, 313 cardiovascular disease events occurred, including 46 cardiovascular disease–related deaths. After adjustment for age, serum potassium level was marginally associated with risk of cardiovascular disease (hazard ratio [HR] per 1 mg/dL increment, 1.03; 95% confidence interval [CI], 1.00-1.05; P = .02). However, after further adjustment for multiple confounders, serum potassium level was not significantly associated with cardiovascular disease risk (HR, 1.00; 95% CI, 0.98-1.03). There were no significant associations between serum potassium level and cardiovascular disease–related death in either age- and sex-adjusted models (HR, 1.06; 95% CI, 0.99-1.12) or multivariable-adjusted models (HR, 1.04; 95% CI, 0.97-1.11).

Conclusion  In our community-based sample of individuals free of cardiovascular disease and not taking medications that affect potassium homeostasis, serum potassium level was not associated with risk of cardiovascular disease.

ALTHOUGH SEVERAL observational studies^1-6 have reported an inverse association between dietary potassium intake and risk of cardiovascular disease, less is known about the association of serum potassium levels and cardiovascular risk. Low serum potassium levels may identify individuals at increased risk for cardiovascular events because of insufficient dietary potassium intake^1-6 or activation of the renin-angiotensin-aldosterone system.⁷ In addition, low serum potassium concentration is associated with ventricular ectopy and may identify individuals at risk for death due to ventricular arrhythmias.⁸

Previous community-based observational cohort studies that examined the association between serum potassium levels and cardiovascular disease have reported conflicting results. One study⁹ reported an increased risk of cardiovascular disease associated with high serum potassium levels, whereas another study¹⁰ reported no association between serum potassium levels and cardiovascular disease risk. To clarify the association between serum potassium levels and cardiovascular disease, we determined the association between serum potassium levels and risk of incident cardiovascular disease among participants in the Framingham Offspring Study.

The Framingham Heart Study is a prospective, epidemiological cohort study established in 1948 to evaluate potential risk factors for coronary heart disease. The original cohort consisted of 5209 residents of Framingham, Mass, aged 28 to 62 years at study entry. In 1971, an additional 5124 individuals (offspring of original cohort members and their spouses) were enrolled into the Framingham Offspring Study. The study design and entry criteria for both cohorts have been detailed elsewhere.^11-14

Participants in the present study were members of the offspring cohort who had blood samples taken for the measurement of serum potassium levels at their second examination (1979-1983). Individuals were excluded if they met any of the following criteria: (1) history of cardiovascular disease, (2) treatment with medications that alter potassium homeostasis (eg, β-adrenergic blocking agents, diuretics, and potassium supplements), (3) creatinine concentration of 2.0 mg/dL or greater (≥180 µmol/L), (4) hemolyzed or missing blood specimens, or (5) extreme elevation of serum potassium level (>6.2 mEq/L). All examinations and procedures were approved by the institutional review board of Boston University School of Medicine, and all participants gave informed consent.

Medical histories were taken and physical examinations were performed for each participant at the baseline examination. Systolic and diastolic blood pressures were measured twice in the left arm of each participant. The average of the 2 readings was used for each blood pressure variable. Body mass index was calculated as weight in kilograms divided by the square of height in meters. Diabetes mellitus was defined on the basis of a nonfasting blood glucose level of 200 mg/dL or greater (≥11.1 mmol/L), a fasting blood glucose level of 140 mg/dL or greater (≥7.8 mmol/L), or the use of insulin or an oral hypoglycemic agent.¹⁵ Self-reported smoking history was used to classify participants as nonsmokers or current smokers. Participants who reported smoking at least 1 cigarette daily during the year before the examination were classified as current smokers. Self-reported alcohol intake was used to quantify the number of alcoholic drinks (12 oz of beer, 4 oz of wine, or 1 oz of liquor) consumed weekly.¹⁶ Self-reported consumption of caffeinated beverages was used to quantify the number of caffeinated drinks (8 oz of coffee or 23.2 oz of tea) consumed daily.¹⁶ Serum potassium level was measured using flame-emission spectrophotometry.¹⁷ Serum creatinine values were determined using an autoanalyzer^18,19 on blood specimens drawn at the baseline examination.

The principal outcomes of interest were incident cardiovascular disease, death from cardiovascular disease, and death from all causes. Cardiovascular disease events included the following: angina pectoris, coronary insufficiency, myocardial infarction, stroke, intermittent claudication, and death due to coronary heart disease or stroke. Criteria for these outcomes have been described previously.¹⁴ Cardiovascular disease events without a clearly identifiable time of onset (eg, angina pectoris or intermittent claudication) were considered to have occurred midway between consecutive examinations. A panel of 3 physicians determined the outcome events after reviewing Framingham Heart Study and outside hospital and physician records.

Serum potassium concentration was considered to be a continuous and a categorical variable. Participants were categorized into sex-specific quartiles of serum potassium. For the primary analysis, sex-specific potassium quartiles were pooled, and baseline characteristics were computed for each pooled quartile. Cox proportional hazards models²⁰ were used to calculate age- and sex-adjusted and multivariable-adjusted hazards ratios (HRs) for cardiovascular disease, death due to cardiovascular disease, and death due to all causes. The following variables were included in the multivariable model: age, sex, systolic and diastolic blood pressure, hypertension treatment, diabetes mellitus, cigarette smoking (cigarettes smoked daily), alcohol consumption (drinks per week), caffeine consumption (drinks per day), serum creatinine level, body mass index, and total and high-density lipoprotein (HDL) cholesterol levels. Variables included in the multivariable model were chosen a priori either because they were well-recognized risk factors for cardiovascular disease (eg, diabetes mellitus) or because they were potential determinants of serum potassium levels (eg, cigarette smoking). The highest quartile of serum potassium served as the reference group. Results of the regression models were displayed as HRs and their 95% confidence intervals (CIs). The analyses were repeated after stratifying by sex.

Increased risk of cardiovascular disease or death may be confined to individuals with high or low serum potassium values. In a secondary analysis, participants were classified into those with low (≤4.0 mEq/L), normal (4.1-5.1 mEq/L), and high (≥5.2 mEq/L) serum potassium levels. Age- and sex-adjusted and multivariable-adjusted HRs of cardiovascular disease, death from cardiovascular disease, and death from all causes were calculated in participants with low and high potassium levels compared with those with normal potassium levels (reference group).

A total of 3854 adults attended the second offspring examination. Of these, we excluded 167 with prevalent cardiovascular disease, 371 reporting use of medications that alter serum potassium levels, 5 with serum creatinine levels of 2.0 mg/dL or greater (≥180 µmol/L), 149 with hemolyzed or missing blood specimens, and 11 with extreme elevations of serum potassium concentration (>6.2 mEq/L). Thus, 3151 individuals (1515 men [age, mean ± SD, 44 ± 10 years] and 1636 women [age, mean ± SD, 43 ± 10 years]; 82% of attendees) were eligible for the analysis. The mean ± SD serum potassium level among eligible individuals was 4.7 ± 0.5 mEq/L (men, 4.8 ± 0.4 mEq/L; women, 4.6 ± 0.4 mEq/L). In individuals excluded because of use of medications that alter serum potassium levels, the mean ± SD serum potassium level was 4.4 ± 0.5 mEq/L (men, 4.5 ± 0.5 mEq/L; women, 4.3 ± 0.5 mEq/L).

Participant characteristics at the baseline examination are given in Table 1. Serum potassium concentration was positively associated with age, smoking, caffeine consumption, and total and HDL cholesterol levels. There were no significant associations of serum potassium level with blood pressure, diabetes mellitus, alcohol consumption, body mass index, and serum creatinine level. Serum potassium concentration was negatively associated with hypertension treatment.

During 16 years of follow-up, 313 cardiovascular disease events occurred (217 in men and 96 in women). Of these, 214 were coronary heart disease events. The age- and sex-adjusted cumulative incidence of cardiovascular disease stratified by serum potassium quartile is displayed in Figure 1. Of 50 cardiovascular disease events occurring between 1.5 and 2.5 years of follow-up, 39 did not have a clearly identifiable time of onset (24 angina pectoris, 10 intermittent claudication, and 5 unrecognized myocardial infarction) and were considered to have occurred midway between the baseline examination and the first follow-up examination, accounting for the sudden increase in the incidence of cardiovascular disease at 2 years of follow-up (Figure 1). The results of proportional hazards models comparing serum potassium quartile and cardiovascular disease risk are given in Table 2 (top). After adjusting for age and sex, there was no significant association between serum potassium quartile and risk of cardiovascular disease. Further adjustment for multiple confounders did not materially alter the results.

Considered to be a continuous variable, potassium concentration was marginally associated with risk of cardiovascular disease after adjustment for age (HR, 1.03; 95% CI, 1.00-1.05). However, this association was no longer statistically significant after further adjustment for systolic and diastolic blood pressure, hypertension treatment, diabetes mellitus, cigarette smoking, alcohol intake, caffeine consumption, serum creatinine level, body mass index, and total and HDL cholesterol levels (HR, 1.00; 95% CI, 0.98-1.03).

During follow-up, there were 46 cardiovascular disease–related deaths (34 in men and 12 in women). The age- and sex-adjusted cumulative incidence of death due to cardiovascular disease stratified by serum potassium quartile is displayed in Figure 2. The results of proportional hazards models testing the association between serum potassium quartile and death due to cardiovascular disease are given in Table 2 (middle). After adjusting for age and sex, there was no significant association between serum potassium quartile and risk of death due to cardiovascular disease. Further adjustment for multiple potential confounders did not materially alter the results.

Considered to be a continuous variable, potassium concentration was not significantly associated with risk of death due to cardiovascular disease after adjustment for age and sex (HR, 1.06; 95% CI, 0.99-1.12) or after further adjustment for systolic and diastolic blood pressure, hypertension treatment, diabetes mellitus, cigarette smoking, alcohol intake, caffeine consumption, serum creatinine level, body mass index, and total and HDL cholesterol levels (HR, 1.04; 95% CI, 0.97-1.11).

During follow-up there were 214 deaths from all causes (131 in men and 83 in women). The age- and sex-adjusted cumulative incidence of death from all causes stratified by serum potassium quartile is displayed in Figure 3. The results of proportional hazards models testing the association between serum potassium quartile and death due to all causes are given in Table 2 (bottom). After adjusting for age and sex, there was no significant association between serum potassium quartile and all-cause mortality. Further adjustment for multiple potential confounders did not materially alter the results.

Considered to be a continuous variable, potassium concentration was not significantly associated with all-cause mortality after adjustment for age and sex (HR, 1.01; 95% CI, 0.98-1.04) or after further adjustment for systolic and diastolic blood pressure, hypertension treatment, diabetes mellitus, cigarette smoking, alcohol intake, caffeine consumption, serum creatinine level, body mass index, and total and HDL cholesterol levels (HR, 0.99; 95% CI, 0.96-1.02).

Of 3151 participants, 175 had serum potassium levels of 4.0 mEq/L or less (low potassium concentration) and 540 had serum potassium levels of 5.2 mEq/L or greater (high potassium concentration). Neither low nor high potassium levels were significantly associated with risk of cardiovascular disease, death due to cardiovascular disease, or death due to all causes compared with individuals with normal potassium levels after adjustment for potential confounders (Table 3).

In our community-based sample of 3151 individuals free of cardiovascular disease, serum potassium concentration was not associated with risk of cardiovascular disease, death due to cardiovascular disease, or death from all causes.

Although it is well recognized that extremely high and extremely low serum potassium levels can be lethal,²¹ less is known about the prognostic value of serum potassium at the ranges commonly encountered in individuals in the community. Serum potassium level is inversely related to risk of ventricular arrhythmias⁸ and may therefore be related to risk of sudden cardiac death. Also, serum potassium level is inversely related to activation of the renin-angiotensin-aldosterone system.⁷ Hypertensive individuals with high renin profiles and low serum potassium levels may be at increased risk for acute myocardial infarction.⁷

Published studies that have examined the association between serum potassium level and risk of cardiovascular disease in community-based samples have reported conflicting results. In 7262 men enrolled in the British Regional Heart Study followed for 11.5 years, Wannamethee et al¹⁰ reported no association between serum potassium level and risk of death due to cardiovascular disease. However, in 2836 men and women enrolled in the first National Health and Nutrition Examination Survey Epidemiological Follow-up Survey followed for 15.9 years, Fang et al⁹ reported that risk of death due to cardiovascular disease was increased in individuals with high serum potassium levels (≥4.5 mEq/L) compared with those with normal serum potassium levels (3.8-4.4 mEq/L) after adjustment for multiple confounders (HR, 1.54; 95% CI, 1.03-2.31).

Our study was similar to the aforementioned studies with respect to sample size, time period, and follow-up. In addition, the meticulous assessment of outcome and risk factors in the Framingham Offspring Study allowed us to examine the association between serum potassium concentration and multiple end points (cardiovascular disease, death due to cardiovascular disease, and death due to all causes) and to account for important confounders in men and women. In our study, we found no association between serum potassium level and risk of either cardiovascular disease or death due to cardiovascular disease, consistent with the study by Wannamethee et al.¹⁰ We excluded individuals with serum creatinine levels of 2.0 mg/dL or greater (≥180 µmol/L) and those taking diuretics. Although not specifically excluded, few individuals in the study by Wannamethee et al¹⁰ had renal dysfunction or were taking diuretics. Subgroup analysis of the study by Fang et al⁹ showed that the increased risk of death from cardiovascular disease was limited to individuals with abnormal renal function or those taking diuretics. In individuals with normal renal function and those not taking diuretics, risk of death from cardiovascular disease was not significantly associated with serum potassium concentration. Thus, among individuals with normal renal function who were not taking diuretics, serum potassium level was not associated with increased risk of cardiovascular disease.

Although serum potassium concentration is inversely related to risk of ventricular arrhythmias,⁸ we found no association between serum potassium level and cardiovascular mortality. Thus, in patients without cardiovascular disease, low serum potassium levels may identify individuals at increased risk for benign ventricular rhythms but not individuals at increased risk for fatal ventricular rhythms.

Our study was conducted in a largely white population and therefore may have limited generalizability to other ethnic populations. Also, because serum potassium concentration was measured at a single point in time, we could not account for changes in serum potassium levels over time that may modify the association between serum potassium level and cardiovascular disease. In addition, relatively few individuals in our study sample died of cardiovascular disease during follow-up. Therefore, we had limited statistical power to detect a modest association between serum potassium level and death due to cardiovascular disease. Further studies in larger sample sizes are warranted to clarify the association between serum potassium concentration and death due to cardiovascular disease.

Because of the relatively narrow range of serum potassium values, even minor errors in sample preparation or measurement of potassium could result in nondifferential misclassification and would decrease the likelihood of observing a significant association. We attempted to minimize such errors by excluding individuals with either hemolyzed specimens or markedly elevated values of potassium. Furthermore, 2 lines of evidence support the rank-order validity of the serum potassium categories used for our analysis. First, we observed statistically significant positive associations between serum potassium quartile and age, sex, and smoking, consistent with published results.^9,10 Second, Tsuji et al⁸ reported an inverse association between serum potassium level and ventricular arrhythmias using serum potassium values determined at the second offspring examination.

In conclusion, in our community-based sample of individuals free of cardiovascular disease and not taking medications that alter potassium homeostasis, serum potassium level was not associated with risk of cardiovascular disease.

Accepted for publication September 13, 2001.

This work was supported in part by grant N01-HC-38038 from the National Heart, Lung, and Blood Institute's Framingham Heart Study, National Institutes of Health, and in part by research career award 1K24 HL04334 from the National Heart, Lung, and Blood Institute (Dr Vasan).

Corresponding author and reprints: Daniel Levy, MD, Framingham Heart Study, 5 Thurber St, Framingham, MA 01702.