Recent studies have found low-normal potassium (K) to be associated with increased diabetes risk. We sought to verify these associations in a multi-ethnic US cohort; and to determine if these associations extend to US Hispanics and Asian-Americans.
We analyzed data from Multi-Ethnic Study of Atherosclerosis (MESA) participants who were free-of-diabetes at baseline. We examined cross-sectional associations between measures of K—serum, dietary, and urine—with fasting glucose and HOMA-IR. We examined longitudinal associations between K and diabetes risk over 8 years.
In multivariable models, compared to those with higher serum K (≥4.5mmol/L), those with lower serum K (<4.0mmol/L) had significantly higher fasting glucose [1.3 mg/dL (95%CI 0.2, 2.4), P-value = 0.03]. Incident diabetes developed in 1281 of 5415 at-risk participants. In minimally-adjusted models, we found inverse associations between serum and dietary K and diabetes risk. Compared to those with higher serum K, those with lower serum K had an HR (95% CI) of incident diabetes of 1.23 (1.04, 1.47), P-value = 0.02. However, these associations were attenuated in fully-adjusted models. We found no significant interaction between potassium and ethnicity.
In this multi-ethnic cohort, we found a significant inverse association between serum K and fasting glucose but no significant association with longer-term diabetes risk. This inverse association between potassium and glucose must be studied further to understand the physiology and its potential impact on chronic health.
Citation: Chatterjee R, Zelnick L, Mukamal KJ, Nettleton JA, Kestenbaum BR, Siscovick DS, et al. (2016) Potassium Measures and Their Associations with Glucose and Diabetes Risk: The Multi-Ethnic Study of Atherosclerosis (MESA). PLoS ONE 11(6): e0157252. https://doi.org/10.1371/journal.pone.0157252
Editor: Renate B. Schnabel, University Heart Center, GERMANY
Received: March 9, 2016; Accepted: May 26, 2016; Published: June 9, 2016
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: The MESA dataset was obtained from the MESA coordinating center at the University of Washington and was analyzed at the University of Washington. https://www.mesa-nhlbi.org/. Availability of the MESA dataset is controlled by the MESA coordinating center. However, there is also a publicly available dataset that can be accessed. We would direct readers and reviewers to the publicly available datasets for MESA available from the NIH at: https://biolincc.nhlbi.nih.gov/studies/mesa/.
Funding: This research was supported by grant R01HL096875 from the National Heart, Lung, and Blood Institute, which funded potassium measurements, as well as contracts N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168 and N01-HC-95169 from the National Heart, Lung, and Blood Institute and by grants UL1-TR-000040 and UL1-TR-001079 from NCRR. The authors thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org. Dr. Chatterjee was supported by the NIH/Duke CTSA KL2TR001115-02. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: ARIC, Atherosclerotic Risk in Communities study; BMI, body mass index; CARDIA, Coronary Artery Risk Development in Young Adults; FEK, fractional excretion of potassium; FFQ, food frequency questionnaire; HOMA-IR, homeostasis model assessment of insulin resistance; IRAS, Insulin Resistance and Atherosclerosis Study; K, Potassium; MESA, Multi-Ethnic Study of Atherosclerosis Study
In secondary data analyses of certain US and international cohorts, low serum and low dietary potassium have been associated with higher risk of diabetes.[1–3] Adults in the US generally consume less than the recommended amount of dietary potassium, with one study finding that less than 2% of adults consumed the recommended amount of ≥4700mg/day. There are also known disparities in dietary potassium intake between African Americans and whites, with African Americans consuming lower amounts of dietary potassium compared to whites. In some studies, both serum and dietary potassium (K) have shown stronger associations with diabetes risk in African Americans compared to whites.[2, 5] Low serum K has been associated with increased diabetes risk and increased prevalence of metabolic syndrome in some Asian populations.[3, 6] Experimental studies provide biological plausibility to the association between low K and abnormal glucose metabolism, by showing that thiazide-induced hypokalemia leads to diminished insulin secretion.[7, 8] Genetic mutations in K channels on pancreatic β-cells are associated with increased diabetes risk, also suggesting a potential biological basis for the association between K and glucose metabolism.
While there has been some exploration of differences in potassium and health effects in African Americans and whites, these explorations have not extended to other ethnicities in the US. Using data from the Multi-ethnic Study of Atherosclerosis (MESA) cohort, a cohort which includes Hispanics and Chinese Americans, we sought to determine the associations between potassium measures with measures of glucose metabolism and diabetes risk.
MESA is a prospective community-based cohort of adults aged 45–84 years of age at baseline. Baseline data were collected between July 2000 and August 2002 with 4 subsequent in-person exams occurring through 2012. Participants were recruited from 6 different communities: Baltimore City and County, MD; Chicago, IL, Forsyth County, NC, Los Angeles County, CA, Northern Manhattan and Bronx, NY. Four ethnicities were represented in this cohort: of the 6814 participants recruited at baseline, 38% were white, 28% were African American, 22% Hispanic, and 12% Asian, predominantly of Chinese descent. Institutional review boards (IRB) at each of the participating institutions for MESA approved the study, and written informed consent was obtained from each participant. Additionally, the IRBs at Duke University and the University of Washington approved the conduct of the data analyses presented in this manuscript.
Participants were excluded from these analyses if they had diabetes mellitus evident at the baseline exam (n = 859). Diabetes was defined as fasting serum glucose ≥126mg/dL and/or use of glucose-lowering medications. We additionally excluded any remaining participants who were missing any of the K exposures (n = 540).
Serum K was measured utilizing indirect potentiometry with an ion-selective electrode. Dietary K intake was derived from a modified Block-style 120-item food frequency questionnaire (FFQ) administered at the baseline exam. This FFQ was initially designed for the Insulin Resistance and Atherosclerosis Study (IRAS) cohort study but was modified to accommodate the diets of the different ethnicities included in the MESA cohort. From this FFQ, individual nutrient intake was quantified using the Nutrition Data System for Research (NDS-R database; Nutrition Coordinating Center, Minneapolis, MN, USA). The coordinating center updated the nutrient variables in June 2015 in order to correct a coding error. This error was found in a subset of participants, depending on participant response to a modifier question, to lead to at least some underreporting of certain food frequency, food group, and nutrient variables. Original data was used to update some of these errors when available, and single imputation methods were used to correct the remaining errors when original data was not available.
Urinary K was measured from a random spot specimen on a Beckman Coulter chemistry analyzer (DXC 600) using indirect potentiometry utilizing a potassium ion- selective electrode in conjunction with a sodium reference electrode. Fractional excretion of K (FEK) was calculated as (urine K/urine Cr)/(serum K/serum Cr). This was presumed to reflect the renal handling of filtered K, as influenced by aldosterone and the delivery of sodium and K to the distal renal tubules.
All exposures were categorized for analyses. Serum K was categorized by clinically relevant cut-points or ≥4.5, 4.0–4.49, and <4.0mmol/L. Dietary K, urinary K, and FEK were categorized into quartiles.
For cross-sectional analyses, outcomes were fasting serum glucose and HOMA-IR, a measure of insulin resistance which was calculated as (fasting glucose*fasting insulin)/405, both measured at the baseline exam. Serum glucose was measured using the glucose oxidase method on the Vitros analyzer (Johnson & Johnson Clinical Diagnostics, Inc, Rochester, NY). Insulin was measured by a radioimmunoassay method using the Linco Human Insulin Specific RIA Kit (Linco Research, Inc., St. Charles, MO).
For longitudinal analyses, incident diabetes mellitus was defined as fasting serum glucose ≥126mg/dL and/or use of glucose-lowering medications at any of the follow-up exams through 2012 for a mean of 8 years of follow-up. Baseline exam was performed between the years 2000–2002; there were 3 additional follow-up exams between baseline exam and the year 2007, approximately 18–24 months apart; and a fifth in-person visit between 2010–2012.
Covariate data were collected at the baseline exam through a combination of self-administered and interviewer-administered questionnaires as well as in-person measurements. Waist circumference was measured at the umbilicus using a measuring tape and rounding to the nearest centimeter. Body-mass index was calculated using height and weight measurements from a stadiometer and calibrated scale. Physical activity was measured using the MESA Typical Week Physical Activity Survey, which was adapted from the Cross-Cultural Activity Participation Survey. From this survey, physical activity was quantified and categorized as minutes spent on activities of different metabolic equivalents. Systolic blood pressure was calculated as the average of the second and third blood pressure readings with patient in a seated position and with readings taken 5 minutes apart.
Baseline characteristics were evaluated by categories of each exposure of interest. Correlations between exposures were assessed. Multivariable linear regression models were used to test cross-sectional associations with the outcomes of fasting glucose or HOMA-IR. Multivariable discrete Cox regression models were used to test associations with incident diabetes. We included covariates which were thought to potentially affect relationships between K and glucose metabolism based on current understanding of pathophysiology. Model 1 adjusted for age (continuous), sex, race/ethnicity, study site, total energy intake (for dietary analyses only), and log of urinary creatinine (for urine K analyses only). Model 2 additionally adjusted for waist circumference (continuous), body mass index (continuous), smoking, family history of diabetes, education (categories), income (categories), alcohol use (categories), and physical activity (categories), systolic blood pressure, and use of antihypertensive medications (as listed in Table 1). We built parallel models with and without interaction terms for race/ethnicity x K exposure in order to evaluate associations for the full population and for participants of each race/ethnicity, respectively. We also tested for statistical interactions in the full cohort between hypertension status/use of anti-hypertensive medications and K exposure. Wald tests were used to test for heterogeneity. Tests of significance were two-tailed with an α level of 0.05. We performed all analyses using R version 3.2.0.
Table 1 shows baseline characteristics of the 5415 included participants by serum K categories. S4 Appendix Table shows baseline characteristics of the participants who were excluded from our analyses. Of the four ethnicities represented in this cohort, whites had the greatest proportion with a high-normal serum K (≥4.5mmol/L) (52%), while African Americans had the greatest proportions with a low-normal serum K (<4.0mmol/L) (34%). Compared to those with a high-normal serum K, participants with a low-normal serum K had a higher systolic blood pressure and greater use of the blood pressure medications that affect renal K handling. Participants in all 3 categories of serum K had similar mean BMI. Correlations between exposures were found to be fairly low. The correlation between serum K and dietary K was 0.04; the correlation between serum K and urinary K was 0.15; and the correlation between dietary K and urinary K was 0.03.
Fasting glucose and HOMA-IR
In cross-sectional analyses using the baseline data, we did find a significant inverse association between serum K and fasting glucose.(Table 2) In our fully-adjusted model (model 2), compared to those with a high-normal serum K (≥4.5mmol/L), those with a low-normal serum K (<4.0mmol/L) had a higher fasting glucose (95% CI) of 1.3mg/dL (0.2, 2.4), P-value = 0.03. There was no statistical interaction between serum K and ethnic group on this outcome (P = 0.53) There was also no statistical interaction between serum K and hypertension status on this outcome (P = 0.46). The direction of the association was consistent in all ethnic groups, except for the Chinese Americans.(Table 2) In minimally-adjusted models (model 1), we found significant inverse associations between dietary K and urinary FEK with fasting glucose; however, after further adjustment, these associations were no longer significant. (S1 and S2 Appendix Tables) We found no significant associations between random urinary K and fasting glucose.
For all 4 measures of K, in minimally-adjusted models (model 1), we found significant inverse associations between K and HOMA-IR, with lower levels of K associated with higher HOMA-IR. However, none of these associations remained significant in fully-adjusted models.(data not shown)
In longitudinal analyses, of the 5415 participants who were free from diabetes at baseline, 1281 participants developed diabetes during a mean follow-up of 8 years. During this time, there was a higher crude incident rate of diabetes (cases per 1000 person-years) among Hispanics (41.6) compared to Chinese Americans (34.0), African Americans (29.2), and whites (26.0).
Serum K was inversely associated with diabetes risk, but this association was attenuated in fully-adjusted models.(Table 3) Compared to those with a higher serum K (≥4.5mmol/L), after adjustment for age, sex, race, and study site, for those participants with a low-normal serum K (<4.0mmol/L), there was a significantly increased risk of diabetes, with an HR (95% CI) of 1.23 (1.04, 1.47), P-value = 0.02; however this association was no longer significant with further adjustment. There was no statistical interaction between serum K and ethnic group on this association (P = 0.25, model 2). There was also no statistical interaction between serum K and hypertension status In this association (P = 0.53) This pattern of association was similar for all ethnic groups except for Chinese Americans.
Dietary K was also inversely associated with diabetes risk, but this association, again, was attenuated in fully-adjusted models. For each standard deviation (SD) decrease in dietary K, after adjustment for age, sex, race, and study site, and total energy intake, there was a significantly increased risk of diabetes, with an HR (95% CI) of 1.11 (1.00, 1.24), P-value = 0.04.(Table 4) We found no significant associations between random spot urinary K measures and diabetes risk. However, in our minimally-adjusted model, we did find an inverse association between FEK and diabetes risk.(S3 Appendix Table) Compared to those in the highest quartile of FEK, in models adjusted for age, sex, race, and study site only, those in the lowest quartile (<8%), had a higher risk of diabetes (HR) (95% CI) of 1.19 (1.01, 1.39), P-value = 0.03; however, this association was attenuated in model 2. There were no significant interactions between dietary K, urinary K or FEK and ethnic group on these associations for any of the models evaluated. Patterns of association between urinary K or FEK with diabetes risk were similar for all 4 ethnic groups.
In these analyses of a multi-ethnic US cohort, we found a significant inverse association between serum K with fasting glucose. We also found crude inverse associations between K measures, insulin resistance, and diabetes risk; however, these associations were not significant with multivariable-adjustment of confounders. We found no statistical interaction between measures of K and ethnic group on the outcomes of either fasting glucose or diabetes risk. Some of the effect estimates of these associations, particularly involving serum K, appeared to differ by ethnicity in direction and/or magnitude, but these differences were not statistically significant. These differences were most notable among the Chinese Americans, a population in whom these associations have not been well-studied before.
The finding of a significant inverse association between serum K and fasting glucose is consistent with other cross-sectional studies which evaluated associations between K and measures of glucose metabolism. In analyses of the Cardiovascular Health Study (CHS), compared to those with higher levels, those with lower serum and dietary K had reduced insulin sensitivity; this study did not evaluate the association between K and fasting glucose. A cross-sectional study of a German cohort found that participants with a low-normal serum K had a higher prevalence of prediabetes among the participants with hypertension.
The finding of no significant associations between measures of K and diabetes risk differs from some studies which did show a significant inverse association between serum or dietary K and diabetes risk.[1–3] In the analyses of the Atherosclerotic Risk in Communities (ARIC) and Coronary Artery Risk Development in Young Adults (CARDIA) cohorts, significant associations were found between serum K and dietary K intake with diabetes risk, respectively.[1, 2] In the ARIC cohort, compared with those with a high-normal serum potassium level (5.0–5.5mmol/L), adults with serum K <4.0mmol/L, 4.0 -< 4.5mmol/L, and 4.5 -< than 5.0mmol/L had an adjusted HR (95% confidence interval [CI]) of incident DM of 1.64 (95% CI, 1.29–2.08), 1.64 (95% CI, 1.34–2.01), and 1.39 (95% CI, 1.14–1.71), respectively. In the CARDIA cohort, there was a statistical interaction between dietary K intake and race (black/white). The association between dietary K intake and diabetes risk was not significant for whites; however, the association between dietary K intake, based in diet histories, and diabetes risk was non-linear but significant for blacks. In the CARDIA cohort, compared with black participants in the highest quintile of dietary K intake, and with adjustment for dietary factors and diabetes risk factors, those in the lowest to higher dietary K intake quintile had adjusted HRs (95% CI) of incident diabetes of 1.72 (0.87, 3.42), 1.99 (1.04, 3.78), 2.10 (1.14, 3.87), and 1.46 (0.79, 2.68) respectively. In a sub-cohort of CARDIA, a significant inverse association was also found between 24-hour urinary measures and diabetes risk. Compared to both ARIC and CARDIA cohorts, the participants of the MESA cohort were, on average, older in age. The MESA cohort has the strength of having a diverse population, but the number of participants in each specific minority ethnic group is smaller than the number of African Americans in either ARIC or CARDIA.[1, 2] In the analyses of the ARIC and CARDIA cohorts, respectively, over 2700 and 2300 African Americans were included in the analyses, while in these analyses of the MESA cohort, there were half this number of African Americans and even fewer Hispanics and Chinese Americans. The smaller population sizes could have limited the detection of a significant association between K measures and diabetes risk in the higher risk populations of MESA. Further study is needed to determine the potential differences in these cohorts that could account for these differing results.
This study extends prior studies by examining associations of K with glucose and diabetes among Hispanics and Chinese Americans. Among Hispanics, inverse associations between serum K with fasting glucose and diabetes risk were similar to those among white and black MESA participants. In contrast, among Chinese Americans, these associations were reversed; however, these associations were also not significant and sample size was small. These potential variations in associations by ethnic group deserve further study.
In multivariate models, urinary K was not significantly associated with fasting glucose or incident diabetes. However, in minimally-adjusted models, there was an inverse association between urine K with HOMA-IR. Random spot urine specimens are easy to collect during study visits, but their use for study measures likely needs further study for validation. 24-hour urinary potassium measurements have been found to be a fairly accurate reflection of dietary potassium intake; however, 24-hour urine samples are very difficult to collect and have their own measurement problems.[15, 16] In the CARDIA study, three consecutive 24-hour urine specimens were collected in a sub-cohort of the participants. In this sub-cohort, those in the lowest quintile of 24-hour urinary K had a significantly increased diabetes risk compared to those participants in the highest quintile of 24-hour urinary K (HR 2.45; 95% CI 1.08, 5.59; p for trend = 0.04). Recently in an international cohort, 24-hour urinary K levels were estimated from morning fasting midstream urine samples.[17, 18] In this study, investigators found a significant association between lower urinary K levels and increased systolic blood pressure an increase in a composite outcome of all-cause mortality and major cardiovascular events. This study did not evaluate the associations of urinary K with diabetes risk or measures of glucose metabolism. In our analyses of MESA participants, random samples were used for measurement of urinary K. Given the diurnal variation in urinary K excretion, the use of spot urine samples may need to be standardized to yield the best estimate of 24-hour urinary K measures.[19, 20]
In theory, FEK should reflect the renal handling of filtered potassium, as influenced by aldosterone as well as dietary K and sodium intake as well as serum K. In minimally-adjusted models only, our analyses did reveal inverse associations between FEK with fasting glucose, HOMA-IR, and diabetes risk, which may reflect these influences. In future studies, it may be important to assess the individual effect of these influences, such as measures of aldosterone, directly on measures of glucose, insulin, and diabetes risk, to help determine if and how FEK can be used in future studies.
There are several limitations of these analyses that should be mentioned. Limitations of this study include the measurement of serum and urine K at a single time point. Serum K and indeed all biological analytes have inherent variability that can lead to misclassification of exposures. While serum K is tightly regulated with homeostatic mechanisms within a fairly tight range, there is within-person variation that is significant with within-person coefficients of variation of about 5%.[21, 22] While this within-person variability is reasonably low, the within-population range is relatively small, and so persons could easily fluctuate between categories. Another limitation of this study is in the definition of the diabetes outcome which is limited by the dataset to use of fasting glucose and use of medications; post-challenge glucose and hemoglobin A1c measures were not available. Compared to the participants included in our analyses, the baseline characteristics of the participants who were excluded from our analyses differed somewhat with regards to demographics. We found no significant differences in K measures or fasting glucose. Other differences were modest and would be unlikely to bias the conclusions drawn from our analyses.
Given that this is a prospective cohort study, a post-hoc power calculation is generally not appropriate. However, given the relatively large number of events of incident diabetes and the large population, while modest associations, particularly within each ethnic sub-group, could be missed, it is unlikely that large associations would be missed. While there were no significant associations found between measures of K and diabetes risk in this cohort, there have been significant associations found in other cohorts. There could be other factors that mediate associations between K and glucose metabolism that we were unable to determine from this dataset, including the presence of genetic susceptibility. Further study may be warranted to take into account genetic susceptibility particularly with regards to the presence of mutations within genes of K-channels such as KCNJ11.
In conclusion, these analyses of the MESA cohort identified significant although small inverse associations between serum K and fasting glucose. There was no significant association between K and diabetes risk. While we found no statistical interaction between K and ethnicity, we did find differences in these associations that varied in magnitude and direction by ethnic group. Given the potential importance of K measures on not only diabetes risk, but also cardiovascular health, further study is needed to describe, more precisely, the relationship of potassium with glucose metabolism and potential effects on longer-term health outcomes. Also further study is needed to determine if these associations are similar in different ethnic groups.
S1 Appendix Table. Associations Between Dietary Potassium and Fasting Glucose, by Ethnicity (N = 5415).
S2 Appendix Table. Associations Between Fractional Excretion (%) (FEK) of Potassium and Fasting Glucose, by Ethnicity.
S3 Appendix Table. Associations Between Fractional Excretion (%) of Potassium (FEK) and Incident Diabetes, by Ethnicity.
Conceived and designed the experiments: RC DE IHdB. Performed the experiments: LZ. Analyzed the data: RC LZ DE IHdB JAN. Wrote the paper: RC LZ DE IHdB LPS KJM JAN BRK DSS JHI RT ANH.
- 1. Chatterjee R, Yeh H, Shafi T, Selvin E, Anderson C, Pankow JS, et al. Serum and dietary potassium and risk of incident type 2 diabetes: The Atherosclerosis Risk in Communities (ARIC) Study. Arch Intern Med. 2010;170(19):1745–51. pmid:20975023
- 2. Chatterjee R, Colangelo LA, Yeh HC, Anderson CA, Daviglus ML, Liu K, et al. Potassium intake and risk of incident type 2 diabetes mellitus: the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Diabetologia. 2012;55(5):1295–303. Epub 2012/02/11. pmid:22322920.
- 3. Heianza Y, Hara S, Arase Y, Saito K, Totsuka K, Tsuji H, et al. Low serum potassium levels and risk of type 2 diabetes: the Toranomon Hospital Health Management Center Study 1 (TOPICS 1). Diabetologia. 2011;54(4):762–6. Epub 2011/01/08. pmid:21212932.
- 4. Cogswell ME, Zhang Z, Carriquiry AL, Gunn JP, Kuklina EV, Saydah SH, et al. Sodium and potassium intakes among US adults: NHANES 2003–2008. Am J Clin Nutr. 2012;96(3):647–57. Epub 2012/08/03. ajcn.112.034413 [pii]. pmid:22854410; PubMed Central PMCID: PMC3417219.
- 5. Chatterjee R, Yeh HC, Shafi T, Anderson C, Pankow JS, Miller ER, et al. Serum potassium and the racial disparity in diabetes risk: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Clin Nutr. 2011;93(5):1087–91. Epub 2011/03/04. [pii]. pmid:21367942.
- 6. Lee H, Lee J, Hwang SS, Kim S, Chin HJ, Han JS, et al. Potassium intake and the prevalence of metabolic syndrome: the Korean National Health and Nutrition Examination Survey 2008–2010. PLoS One. 2013;8(1):e55106. Epub 2013/02/02. PONE-D-12-32218 [pii]. pmid:23372822; PubMed Central PMCID: PMC3555937.
- 7. Rowe JW, Tobin JD, Rosa RM, Andres R. Effect of experimental potassium deficiency on glucose and insulin metabolism. Metabolism. 1980;29(6):498–502. Epub 1980/06/01. pmid:6991855.
- 8. Helderman JH, Elahi D, Andersen DK, Raizes GS, Tobin JD, Shocken D, et al. Prevention of the glucose intolerance of thiazide diuretics by maintenance of body potassium. Diabetes 1983;Feb;32(2):106–11. pmid:6337892
- 9. Nettleton JA, Rock CL, Wang Y, Jenny NS, Jacobs DR. Associations between dietary macronutrient intake and plasma lipids demonstrate criterion performance of the Multi-Ethnic Study of Atherosclerosis (MESA) food-frequency questionnaire. The British journal of nutrition. 2009;102(8):1220–7. pmid:19454126; PubMed Central PMCID: PMC3035638.
- 10. Ainsworth BE, Irwin ML, Addy CL, Whitt MC, Stolarczyk LM. Moderate physical activity patterns of minority women: the Cross-Cultural Activity Participation Study. Journal of women's health & gender-based medicine. 1999;8(6):805–13. pmid:10495261.
- 11. Team RC. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/. 2015.
- 12. Chatterjee R, Biggs ML, de Boer IH, Brancati FL, Svetkey LP, Barzilay J, et al. Potassium and glucose measures in older adults: the cardiovascular health study. The journals of gerontology Series A, Biological sciences and medical sciences. 2015;70(2):255–61. pmid:24895271.
- 13. Meisinger C, Stockl D, Ruckert IM, Doring A, Thorand B, Heier M, et al. Serum potassium is associated with prediabetes and newly diagnosed diabetes in hypertensive adults from the general population: The KORA F4-Study. Diabetologia. 2012. Epub 2012/11/28. pmid:23183943.
- 14. Panel on Dietary Reference Intakes for Electrolytes and Water. Chapter 5. Potassium Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate. 2004;Washington DC, The National Academies Press:186–268.
- 15. Tasevska N, Runswick SA, Bingham SA. Urinary potassium is as reliable as urinary nitrogen for use as a recovery biomarker in dietary studies of free living individuals. J Nutr. 2006;136(5):1334–40. Epub 2006/04/15. 136/5/1334 [pii]. pmid:16614426.
- 16. Bingham SA, Day NE. Using biochemical markers to assess the validity of prospective dietary assessment methods and the effect of energy adjustment. Am J Clin Nutr. 1997;65(4 Suppl):1130S–7S. Epub 1997/04/01. pmid:9094909.
- 17. Mente A, O'Donnell MJ, Rangarajan S, McQueen MJ, Poirier P, Wielgosz A, et al. Association of urinary sodium and potassium excretion with blood pressure. N Engl J Med. 2014;371(7):601–11. pmid:25119606.
- 18. O'Donnell M, Mente A, Rangarajan S, McQueen MJ, Wang X, Liu L, et al. Urinary sodium and potassium excretion, mortality, and cardiovascular events. N Engl J Med. 2014;371(7):612–23. pmid:25119607.
- 19. Kawasaki T, Itoh K, Uezono K, Sasaki H. A simple method for estimating 24 h urinary sodium and potassium excretion from second morning voiding urine specimen in adults. Clin Exp Pharmacol Physiol. 1993;20(1):7–14. Epub 1993/01/01. pmid:8432042.
- 20. Tanaka T, Okamura T, Miura K, Kadowaki T, Ueshima H, Nakagawa H, et al. A simple method to estimate populational 24-h urinary sodium and potassium excretion using a casual urine specimen. J Hum Hypertens. 2002;16(2):97–103. Epub 2002/02/19. pmid:11850766.
- 21. Sebastian-Gambaro MA, Liron-Hernandez FJ, Fuentes-Arderiu X. Intra- and inter-individual biological variability data bank. European journal of clinical chemistry and clinical biochemistry: journal of the Forum of European Clinical Chemistry Societies. 1997;35(11):845–52. pmid:9426342.
- 22. Lacher DA, Hughes JP, Carroll MD. Biological variation of laboratory analytes based on the 1999–2002 National Health and Nutrition Examination Survey. National health statistics reports. 2010;(21):1–7. pmid:20540274.