Strengths and limitations of this study

This is the first study in the general population on the study interest, using a well-designed multicentre survey with a sufficient sample size.

The 24-hour urinary potassium, as the gold standard method, was used to measure dietary potassium intake.

A central laboratory was used for the analysis of serum and urinary potassium.

The cross-sectional design could not establish causality.

The study population was predominantly men and was conducted in China. Cautions should be made to generalise to other populations.

Introduction

Insufficient potassium intake is a well-established risk factor for elevated blood pressure.1 2 Guidelines suggest a daily potassium intake over 90 mmol,3 but the actual potassium consumption worldwide is far from recommendations, despite major efforts from government and health organisations.4 Several dietary strategies to increase potassium intake, such as potassium-enriched salt substitutes,5 oral potassium supplements6 and potassium-enriched diets (dietary approaches to stop hypertension diet),7 have shown blood pressure-lowering effects in trials. However, their public health potential is limited by the fear of hyperkalaemia. Due to the concern about the possible increased risk of hyperkalaemia, these hopeful approaches for the prevention and control of hypertension encountered great resistance in the process of their promotion.

Although dietary potassium intake was postulated to be an important contributor to hyperkalaemia, especially in people with impaired renal function, evidence linking dietary potassium and hyperkalaemia or serum potassium is virtually scarce and inconclusive. The well-known theory8 of potassium homeostasis tells us that individuals with normal renal function would not develop hyperkalaemia from an increase in potassium in their diet. However, this theory was proposed based on preliminary loading trials with limited healthy participants conducted in the late 1980s.9–12 In addition, whether there are preconditions that might influence the association is yet to be elucidated.

The 24-hour urine collection is the gold standard method to quantify dietary potassium.13 Thus, the present study aimed to investigate the association of serum potassium level with potassium intake measured by 24-hour urine. We also explored whether the association differed by health condition status, if any.

Methods

This is a cross-sectional analysis of baseline data from the Diet, ExerCIse and carDiovascular hEalth (DECIDE)-Salt Reduction Strategies for Seniors in Residential Facilities (DECIDE-Salt) study (NCT03290716). The study design of the DECIDE-Salt study has been reported elsewhere.14 Briefly, 48 residential elderly care facilities in northern China were cluster-randomised with a 2×2 factorial design to test the effectiveness and safety of two salt-reduction strategies: salt substitution and progressively restricting salt supply to the facility kitchens. Residents aged 55 years or older with blood pressure measured at the baseline survey were included in the trial, and residents with physician-confirmed hyperkalaemia would be excluded, but none met this exclusion criterion at baseline. The baseline survey was conducted from September 2017 to March 2018 in four regions of northern China, and 1612 participants in total met the eligibility criteria.

For the present study, we further excluded 650 participants who did not have either serum potassium or 24-hour urinary potassium measured at baseline. The 24-hour urinary potassium was used as the measure of daily potassium intake in the present study. Reasons for the failure to collect these data were mainly due to the local cultural belief that blood loss is awful and many participants’ unwillingness to keep urine collectors in their living places. Specifically, 285 had neither drawn blood nor collected 24-hour urine, 61 had not drawn blood, 196 had not collected 24-hour urine and 108 had drawn blood but collected incomplete urine. Criteria for incomplete urine collection include: (1) reported missing urine of over 50 mL, (2) the collection time fell outside the range of 22‒26 hours or (3) the total volume of urine was <250 mL. The rest of the 962 participants who had both serum potassium and complete 24-hour urine potassium measured were included in the present study (figure 1). The study was approved by the Peking University Institutional Review Board, with both group and individual consents obtained.

Figure 1

Participants’ flowchart.

Data collection and definitions

All eligible participants were invited for a health examination at baseline, including a questionnaire interview, a physical examination, a fasting blood sample for serum potassium and creatinine measurement and a 24-hour urine sample collected to measure urinary potassium excretion. Both serum and urinary electrolytes were measured with the ion-selective electrode method,15 and serum creatinine was measured using a Roche enzymatic assay.16

Statistical analysis

We divided our study population into the following three groups based on renal function and health conditions, according to our previous study17 and the literature review18: high, moderate and low risk of hyperkalaemia. High-risk group was defined as those with at least one health condition and impaired renal function (estimated glomerular filtration rate (eGFR) <90 mL/min×1.73 m2); moderate-risk group was defined as those with at least one health condition and renal function normal (eGFR ≥90 mL/min×1.73 m2) or without health condition but had eGFR <90 mL/min×1.73 m2, and low-risk group was defined as those without health conditions and with normal renal function (eGFR ≥90 mL/min×1.73m2). Health conditions included the following: hypertension (systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg, or on antihypertension medications within 2 weeks); self-reported renal disease, diabetes mellitus, coronary heart disease, stroke, cancer and chronic obstructive pulmonary disease and being bedridden.

The scatter plot of serum potassium level by 24-hour urinary potassium level was drawn and the Pearson correlation coefficient was calculated to describe the correlation between the two. The linear mixed model with adjustment for age, sex, region and accounting for cluster effect was used to test the statistical significance of the association, in the entire population and separately, among all individuals, among individuals with certain health conditions and among individuals with low, moderate and high risk of hyperkalaemia. The sensitivity analysis was performed by additionally adjusting for medications that might elevate serum potassium.

All analyses were done using SAS V.9.4 and R V.4.0.2. A p-value of <0.05 was considered for statistical significance.

Patient and public involvement

None.

ResultsCharacteristics of the study population

The study population included 962 participants (figure 1) and had a mean age of 69.1 years, with 86.8% of men 17.3% classified as low risk, 48.4% as moderate risk and 34.3% as high risk of hyperkalaemia. Serum potassium was normally distributed with a mean of 4.44±0.48 mmol/L (table 1). Compared with those not included in the analysis, our study participants were younger, more likely to be men and less educated, more likely to smoke and drink, more likely to have hypertension and cardiometabolic disease and more likely to use antihypertensive medications, including those that may elevate serum potassium (online supplemental table S1).

Table 1

Characteristics of study population

Association of serum potassium level with 24-hour urinary potassium excretion

Serum potassium was associated with urinary potassium weakly but significantly among all participants (correlation coefficient=0.12, p<0.001; adjusted β coefficient=0.0040/L, 95% CI 0.0018 to 0.0062; p=0.001) (figure 2). The results were similar when medications that may elevate serum potassium were additionally adjusted (adjusted β coefficient=0.0037 /L, 95% CI 0.0014 to 0.0059; p=0.001) (table 2).

Figure 2

Association of serum potassium level with 24-hour urinary potassium excretion in the subgroup population. The forest plot showed the correlation and association of serum potassium and 24-hour urinary potassium in subgroups. Adjusted β coefficient was obtained from a linear mixed model adjusted for age, sex, region and accounting for facilities as cluster effect in each subgroup. Other health conditions include chronic obstructive pulmonary disease, cancer and bedridden.

Table 2

Association of serum potassium level with 24-hour urinary potassium excretion in overall population

The association differed by risk subgroups (figure 2). Point estimates of correlation and adjusted β coefficient were larger among those with health conditions and abnormal renal function. When categorising the participants using hyperkalaemia risk, urinary potassium was not associated with serum potassium among apparently healthy participants (correlation coefficient=0.03, p=0.721; adjusted β coefficient=0.0018/L, 95% CI −0.0017 to 0.0053; p=0.311), weakly associated among participants with moderate risk (correlation coefficient=0.14, p=0.002; adjusted β coefficient=0.0040/L, 95% CI 0.0007 to 0.0073; p=0.017) and among participants with high risk (correlation coefficient=0.18, p<0.001; adjusted β coefficient=0.0078/L, 95% CI 0.0027 to 0.0129; p=0.003) (figure 3).

Figure 3

Scatter plot with regression lines representing the correlations between 24-hour urinary potassium excretion with serum potassium in (A) whole study population (n=962), (B) low risk (n=166), (C) moderate risk (n=466) and (D) high risk (n=330). Shaded areas represent the 95% CIs. Adjusted β coefficient was obtained from a linear mixed model adjusted for age, sex, region and accounting for facilities as cluster effect.

Discussion

To our knowledge, this is so far the largest population study with both serum potassium and potassium intake (measured by 24-hour urinary excretion) measured. The study results showed that serum potassium increased with potassium intake measured by 24-hour urinary potassium in the entire study population, but the association was extremely weak; each increment of 10 mmol in urinary potassium (equivalent to about 13 mmol potassium intake) was associated with an increase of serum potassium concentration by only 0.04 mmol/L. Furthermore, the association differed by individual’s renal function and health status; it existed only among individuals with impaired renal function or other health conditions but not among apparently healthy individuals. Our findings support the potassium homeostasis theory that under the condition of good health, the human body can maintain serum potassium stable and not affected by the amount of potassium intake. In that case, there is no need to worry at all about the risk of developing hyperkalaemia when potassium intake is increased. Even with the conditions of impaired renal function or other diseases, the possibility of developing hyperkalaemia from the increase of potassium intake within the daily life plausible range should be very small and could probably be offset by significant health gains in reducing the risk of hypokalaemia, cardiovascular disease and total death.19 20

The association between serum potassium and dietary potassium in the general population has not been well-studied ever since the theory of potassium homeostasis was propounded. So far, no large-scale observational study has analysed the association between dietary potassium and serum potassium in the general population. Most of the intervention studies were small trials. Back in the 1940s, two small studies,21 22 including 10 dialysis patients and seven patients with or without renal insufficiency, respectively, observed an increase in serum potassium after oral potassium loading. A meta-analysis23 of small-scale studies on the effects of oral potassium supplementation revealed a slight elevation of 0.14 mmol/L in serum potassium concentrations with a moderate dose of 45 mmol/day on average. Based on the assumption that 75% of the ingested potassium would be eliminated by the kidney,24 this corresponds to a 0.0041 mmol/L increment in serum potassium per 1 mmol/day increase in 24-hour urinary potassium excretion, which is in agreement with our results.

In addition, we found the association differed by subgroups defined by renal function and health conditions. The association existed only among individuals with health conditions or impaired renal function but not among apparently healthy individuals. This finding suggested that health conditions, particularly impaired kidney function, are also modifiers of the association of serum potassium with potassium intake, in addition to being a risk factor for hyperkalaemia as reported previously.25 Previous guidelines,26 27 systematic reviews,28 29 randomised controlled trials30 and observational studies31 32 have provided limited and inconclusive evidence for the association between potassium intake and serum potassium in population with chronic kidney disease (CKD), not to mention the evidence for renal function act as a modifier. However, it is not hard to interpret the role of renal function. In healthy subjects, the kidney eliminates the surplus potassium intake through urinary excretion, thereby preserving a constant serum potassium concentration.33 However, when kidney function is impaired, its capability to excrete the extra potassium would be lowered and serum potassium would be increased, but in a non-linear manner. A similar non-linearity association was observed in a recent observational study among patients with CKD,34 which found the positive associations of urinary and serum potassium to be more pronounced, with the mean difference increasing from 0.08 to 0.16 per 10 mmol increment in urinary potassium, as CKD stage progresses from G3 to G5. The aetiology may be more complicated for other health conditions. So far, it is unknown why health conditions other than kidney disease could alter the association of serum potassium with potassium intake. We hypothesise that some health conditions might alter the potassium regulation indirectly or directly, via the internal (eg, intracellular buffer) and external (eg, renal excretion) balance of potassium.35 In the present study, we observed a stronger association in participants with coronary artery disease, compared with those without. These patients may use multiple agents, including RAS blockade and β-blockers, and medications like ACEI/ARB, aldosterone antagonists and β-blockers impair aldosterone signalling and change potassium translocation.36 We also observed strong associations among participants with chronic obstructive pulmonary disease (COPD), bedridden or cancer. Health conditions like COPD may experience altered acid-base status, for example, respiratory acidosis, thus resulting in transcellular shifts of potassium from intracellular room to extracellular fluid.37 Here we also proposed a ‘set point’ theory for potassium homeostasis. Suppose we simplify the balance of potassium as the equilibrium of dietary potassium, blood potassium and urinary potassium. Imagine the blood potassium as a water tank, the dietary potassium as an inflow and the urinary potassium as an outflow. Under normal conditions, the inflow and outflow are balanced under the initial set point, maintaining a stable blood potassium level. However, the set point may change due to certain alterations in the internal environment such as health conditions like coronary artery disease. Just as a series of systemic changes occur after the set point of the hypothalamic temperature centre changes during fever, the systemic blood potassium balance system also makes corresponding adjustments. Similarly to the systemic changes after the alteration of set point in the hypothalamic thermoregulatory centre during fever,38 the potassium homeostasis system may undergo adaptions to set point changes. For instance, after the set point elevates, the same amount of dietary potassium requires less potassium excretion to attain a higher blood potassium. At this point, the association of blood potassium becomes more pronounced with dietary or urinary potassium. Further investigation is needed to elucidate the possible mechanism of the findings.

Though significant, the association of dietary and serum potassium is too weak to oppose a dietary supplement among unhealthy patients. Our findings among the high hyperkalaemia risk population were in line with results from a recent randomised feeding trial24 in 29 patients with stage 3 CKD, in which the mean serum potassium increased by 0.21 mmol/L in participants with a higher potassium diet of 100 mmol/day, compared with participants with potassium diet of 40 mmol/day. Whether this small increment of serum potassium would outweigh the benefit of potassium supplementation requires comprehensive evaluation. The promotion of dietary strategies to increase potassium intake should not be impeded just by the fear of the very small possibility of hyperkalaemia in the general population.

The present study has strengths and limitations. As far as we know, it is the first study in the general population that explores the association between serum potassium and urinary potassium. We used 24-hour urinary potassium as the measure of potassium intake, which is considered the gold standard method for measuring potassium intake. The central laboratory measurements of both serum and urinary potassium with an independent quality control programme eliminated the possible biases introduced by the between-centre variability. And the large sample ensured the robustness of our findings.

There are also some limitations. First, our study was a cross-sectional study and hence could not establish causality. Second, our study had only one 24-hour urinary potassium measurement, which may suffer from a large intraindividual variation of the measurement. Considering the observed very small correlation coefficients, relatively larger sample size and similar results in previous studies, we believe the bias introduced by the single measurement should be minimal. Third, there were only a small number of participants with severe renal dysfunction (39 cases with eGFR <60 mL/min×1.73 m2), which precluded us from any reliable analysis among this group with a particularly high risk of hyperkalemia. Fourth, the potassium intake in our study population was relatively low. Generalisation to other populations should be made with caution, particularly to those with higher potassium intake. Fifth, the majority of the study population was male, reflecting the typical sex distribution in elderly care facilities in China. However, the interaction between urinary potassium and sex was not significant for the association with serum potassium in the overall population or in any of the hyperkalaemia risk groups (all p for interaction >0.21). This may indicate that the results of the present study are generalisable regardless of the sex distribution of the study population.

Conclusions

A weak association of dietary potassium intake with serum potassium level existed only among individuals with impaired renal function or other health conditions but not among apparently healthy individuals. The results imply that increasing dietary potassium intake may slightly increase the risk of hyperkalaemia but may also decrease the risk of hypokalaemia in unhealthy individuals, both of which bear important health concerns.

Data availability statement

Data are available upon reasonable request. Data will be disclosed only on request and approval of the proposal by the study review committee. Deidentified participant data and a data dictionary will be made available following approval.

Ethics statementsPatient consent for publication

Consent obtained directly from patient(s).

Ethics approval

This study involves human participants and was approved by Peking University Institutional Review Board. Participants gave informed consent to participate in the study before taking part.

Acknowledgments

The authors thank all facility residents for their participation and cooperation. We also like to thank all investigators, study team members, facility managers and staff and administrative agencies for their proactive participation in the study.