Introduction

According to the paper summarizing the Kidney Disease: Improving Global Outcomes Controversies Conference on potassium homeostasis and management of dyskalemia in kidney disease,¹ hyperkalemia is generally defined based on potassium distribution values in the general population. During the conference, it was also agreed that there is no consensus on the magnitude, duration, and frequency of elevated potassium levels that define chronicity. It is well established that hyperkalemia is more prevalent in patients with low estimated glomerular filtration rate (eGFR).¹

As shown by Nilsson et al,² chronic hyperkalemia is usually detected in patients undergoing more frequent testing, and is often asymptomatic. However, this observation may be biased, as it is confounded by indication for testing.² A potassium level greater than 5.5 mmol/l is also an established risk factor for major adverse cardiovascular events (MACEs),³ as well as a common indication for initiating dialysis.¹

In patients receiving kidney replacement therapy, elevated potassium levels are a common electrolyte disturbance⁴; however, real‑world data on the prevalence of hyperkalemia in dialyzed populations are relatively scarce. Such data can be derived from registries, clinical trials, or other research studies, particularly those involving renin–angiotensin–aldosterone system (RAAS) inhibition.

In the DOPPS (Dialysis Outcomes and Practice Patterns Study) registry, including 55 183 patients from 20 countries, the prevalence of hyperkalemia in hemodialysis (HD) patients was 30%–50%.⁵ In the Chinese PRECEDE‑K study (Hyperkalaemia Prevalence, Recurrence and Treatment in Haemodialysis), hyperkalemia was present in 75.5% of 600 patients, and potassium binders were rarely used in this real‑world population.⁶ Data from Greece showed that the incidence of hyperkalemia (predialysis serum potassium levels ≥5.1 mmol/l) in 149 HD patients was as high as 60.4%, while the prevalence of predialysis serum potassium levels equal to or greater than 5.5 mmol/l reached 42.2%.⁷

Because European data are very limited, the aim of our study was to assess the prevalence of hyperkalemia in a real‑world setting, using a database from the largest nontertiary HD center network in Poland.

Patients and methods

We conducted a retrospective analysis of a fully anonymized database comprising 5879 unselected, consecutive patients undergoing HD, provided by the largest network of nontertiary HD centers in Poland. The database was updated on October 6, 2022, and included averaged data from a complete 4‑week period spanning August 29, 2022 to September 25, 2022. The patients were treated with either HD (96.53%) or hemodiafiltration (HDF; 3.47%). Mean session duration was 240 minutes, and all patients underwent dialysis 3 times per week (Monday–Wednesday–Friday or Tuesday–Thursday–Saturday). Anticoagulation with heparin was administered in all participants. High‑flux synthetic (Helixone) dialyzers were used in 99.9% of the cases, with a surface area of 1.8 m² in 88.4% of the patients and 2 m² in the remaining 11.6%. Only 5 individuals (0.1%) were treated with low‑flux synthetic dialyzers. Mean blood flow rate was 332 ml/min, and mean dialysate flow rate was 410 ml/min. Mean ultrafiltration volume was 2069 ml per session. Among the patients undergoing HDF, the substitution volume exceeded 20 l per session. Dialysate containing bicarbonate and glucose was used in all patients. Additionally, 87.3% of the participants received dialysate with a potassium concentration of 3 mmol/l, 8.74% received a 2‑mmol/l potassium bath, and the remaining 3.96% received a 4‑mmol/l potassium bath. Residual kidney function was present in 54.4% of the study population (reported in a qualitative manner, ie, yes vs no), urine volume was not recorded.

Characteristics of all patients according to hyperkalemia vs normokalemia status are summarized in Table 1. The database included demographic information as well as data on chronic kidney disease etiology, comorbid conditions, classes of antihypertensive medications, predialysis blood pressure and heart rate, prevalence of apparent treatment‑resistant hypertension (defined as uncontrolled blood pressure despite the use of ≥3 antihypertensive medication classes or the use of ≥4 medications regardless of blood pressure value), and HD‑related parameters, including vascular access type (arteriovenous fistula), HDF, and recirculation.

Although data were collected from multiple centers nationwide, all laboratory tests were performed within a single laboratory network operated by Diagnostyka S. A., ensuring a standardized measurement methodology and minimizing potential measurement variability.

All patients undergoing hemodialysis received dietary counseling on nutritional and fluid restrictions from a registered dietitian. Education in this area is a standard component of the initial training provided to every patient starting kidney replacement therapy. However, adherence to these recommendations could not be verified in all participants. No potassium binders were used during the study period.

The Institutional Review Board at the Medical University of Warsaw, Poland, approved this study (AKBE/16/2023). The Board does not require informed consent for retrospective studies based on medical records. The study adhered to the principles of the Helsinki Declaration. Ethical approval was waived by the local ethics committee due to the retrospective nature of the study and because the database was originally developed for health care purposes. Furthermore, all data were fully anonymized before the authors gained access to the database, and no direct contact with the patients occurred at any stage of the study.

Results

As there is no universally accepted definition of hyperkalemia, we adopted the potassium level thresholds most widely used to categorize mild, moderate, and severe hyperkalemia: 5, 5.5, and 6 mmol/l, respectively. From this dataset, we identified a total of 3211 patients with hyperkalemia, defined as serum potassium levels equal to or greater than 5 mmol/l. We also assessed the prevalence of moderate and severe hyperkalemia, defined as potassium levels equal to or greater than 5.5 and 6 mmol/l, respectively. The most striking finding was the high prevalence of hyperkalemia. Mild, moderate, and severe hyperkalemia was observed in 54.6%, 30.8%, and 13.5% of the patients, respectively. The individuals with hyperkalemia were older, had longer dialysis vintage, were more likely to receive RAAS inhibitors (RAASis) and less likely to use diuretics, and had higher ultrafiltration volumes during HD, as compared with the normokalemic patients.

The GLM demonstrated a significant and independent association between the presence of hyperkalemia and most covariates included in the model, except for dialysis duration and estimated urea clearance over time (eKt/V). Dialysis vintage, ultrafiltration, and treatment with RAASis were positively associated with hyperkalemia, whereas age, diabetes, and potassium bath concentration showed a negative association (Supplementary material, Figure S1). Clinically meaningful ORs were observed for diabetes (OR, 0.71; 95% CI, 0.63–0.81; P <⁠0.001), RAASi use (OR, 1.45; 95% CI, 1.28–1.64; P <⁠0.001), and potassium bath concentration (OR, 0.81; 95% CI, 0.68–0.97; P <⁠0.05), while the remaining significant ORs were close to 1.

Discussion

Serum potassium balance is influenced by several factors, including dietary potassium intake (quantity and bioavailability), dialysis parameters (dialysate potassium, bicarbonate, and glucose concentrations), residual kidney function, medications (β-blockers, heparin, and renin–angiotensin system inhibitors), and other conditions (acidosis, insulin deficiency, and hyperosmolality).¹ Seasonal variation should also be considered.¹

The rate of hyperkalemia in our real‑world cohort of prevalent HD patients (54.6%) was substantially higher than that reported in the DOPPS registry (30%–50%; data from 20 countries worldwide).⁵ It was also higher than in the NECOSAD (Netherlands Cooperative Study on the Adequacy of Dialysis) study of 1117 incident patients in the Netherlands, in which mild, moderate, and severe hyperkalemia was present in 22%, 15%, and 8% of the participants, respectively, while the overall hyperkalemia prevalence was 45%.⁸

In a United States (US) cohort of 6655 incident patients from a large dialysis organization, hyperkalemia (defined as potassium level >5 mmol/l) was present in 12% of the study population, and it was associated with a decline in residual kidney function.⁹

In a single‑center study from Sri Lanka,¹⁰ 100% of prevalent HD patients had potassium levels exceeding 5 mmol/l, and all were dialyzed twice weekly for 4 hours per session. In the RE‑UTILIZE study, a retrospective observational cohort analysis of 9347 HD patients performed using DOPPS survey data from US patients,¹¹ any predialysis hyperkalemia (potassium level >5 mEq/l) occurred in 74% of the patients within 1 year of the index date, and only 2.8% were prescribed an oral potassium binder. Notably, obesity, recent initiation of hemodialysis, and dialysate potassium bath concentration equal to or greater than 3 mEq/l were associated with lower prevalence of hyperkalemia.

Conversely, in a retrospective single‑center study from Greece, duration of hemodialysis sessions shorter than 4 hours and use of a high‑potassium dialysate bath (4 mmol/l) were independent predictors of hyperkalemia.⁷ In our study, most patients were dialyzed using a 3‑mmol/l potassium bath. The individuals with hyperkalemia were more often dialyzed with a 2‑mmol/l potassium bath and less often with a 4‑mmol/l bath.

In a French prospective multicenter study by Rossignol et al¹² (14 chronic hemodialysis units in the Lorraine region) including 527 patients followed from January 2, 2014 to December 31, 2015, predialysis potassium levels, dialysate potassium concentrations, and concomitant potassium binder prescriptions were analyzed. Mild, moderate, and severe hyperkalemia (potassium levels >5.1, >5.5, and >6 mmol/l, respectively) were observed in 26.4%, 13.8%, and 4.9% of the patients, respectively, with an overall prevalence of 45.1%. Importantly, potassium binders—primarily sodium polystyrene sulfonate—were widely prescribed, with prescription rates increasing from 61% to 78% over time. In that study, 2‑mmol/l and 3‑mmol/l potassium bath concentrations were used at similar rates, whereas in our cohort, the 3‑mmol/l bath was used much more often than the 2‑mmol/l bath (87.3% vs 8.74% of the patients, respectively). Presence of residual diuresis was reported in 54.5% of the French cohort, similarly to our study. None of our patients were treated with potassium binders, which may explain differences between the results of the French study¹² and our findings. To date, real‑world data on hyperkalemia prevalence in the era of newer potassium binders (patiromer and sodium zirconium cyclosilicate) remain limited.

In the Chinese PRECEDE‑K study, despite a high prevalence of hyperkalemia (75.5%), only 5.7% of the patients received potassium binders (resins [4.4%] or sodium zirconium cyclosilicate [1.3%]).⁶ In a Japanese study involving 2846 prevalent HD patients, hyperkalemia—defined using interdialytic interval–specific criteria—was observed in 31.9% of the participants. RAASis were used in 59.9% of the hyperkalemic patients, while potassium binders (sodium or calcium polystyrene sulfonate) were used in 21.2%.¹³ All patients used a 2‑mmol/l potassium dialysate. The prevalence of hyperkalemia increased with dialysis vintage (P <⁠0.001).

In Brazil, an online voluntary survey assessing hyperkalemia prevalence (defined as serum potassium level ≥6 mEq/l) found a national prevalence of 16.1% in 2019, ranging from 12.1% in the North to 18.7% in the Northeast.¹⁴ These rates are consistent with our results and the findings from Greece.⁷ However, the prevalence of severe hyperkalemia was lower in Spain (7.8%)¹⁵ and in a large US cohort from 2007 (4.5%).¹⁶ In contrast, in a tertiary care center in central Sri Lanka, hyperkalemia (potassium level >5 mmol/l) was observed in all analyzed patients (n = 114), likely due to a lower dialysis dose (twice weekly, 4 h per session) than the standard thrice‑weekly regimen.¹⁰ Differences across countries and regions may be explained by variations in diet, dialysis prescriptions, and potassium binder use.

In our study, hyperkalemia was more common among the RAASi users than the nonusers.¹⁷ This finding is consistent with a recent analysis of 4764 dialysis patients from the Sicilian Registry of Nephrology, Dialysis, and Transplantation (2018–2020),¹⁸ which demonstrated a direct association between RAASi use and serum potassium levels. Hyperkalemia at baseline was present in 39% of the patients, 42% of whom were receiving RAASis. The authors also reported significant associations between hyperkalemia, diabetes, albumin and phosphate levels, chronic obstructive pulmonary disease, dialytic age, acetylsalicylic acid use, and diuresis in a multivariate model.¹⁸ No data on potassium binder use were provided. Importantly, no differences in mortality were observed when analyses were stratified by potassium level (cutoff, 5.1 mmol/l).

More recently, Almalki et al¹⁹ conducted a retrospective cohort study of 905 adult RAASi users in outpatient clinics (not on dialysis), reporting hyperkalemia (potassium levels ≥5.1 mmol/l) in 32.8% of the patients; moderate and severe hyperkalemia was present in 19% and 6.4%, respectively. Hyperkalemia led to RAASi discontinuation in 6.2% and dose reduction in 4.5% of the patients, and to dialysis initiation in 0.2%. Risk factors included age of 75 years or above, diabetes, congestive heart failure, and reduced eGFR.

In our previous analysis of the same database,¹⁷ we examined a subgroup of 2117 patients receiving at least 3 antihypertensive medications (including diuretics), and compared RAASi users (n = 1009) with nonusers (n = 1008). Serum potassium levels were slightly but markedly higher in the RAASi users (mean [SD], 5.3 [0.8] mmol/l vs 5.1 [0.7] mmol/l; P <⁠0.01). Similarly, in a recent Chinese study by Liao et al,²⁰ the prevalence of hyperkalemia among 499 HD patients (1812 potassium measurements) was 55.1%, 27.7%, and 11.6% for the thresholds of 5, 5.5, and 6 mmol/l, respectively. Factors associated with higher potassium levels included sex, dialysis duration, Kt/V, creatinine, and treatment with angiotensin‑converting enzyme inhibitors or angiotensin receptor blockers. Elevated potassium levels were also associated with increased risks of MACEs and hospitalization. No data on potassium binder use were reported. It remains to be determined whether newer potassium binders reduce hyperkalemia prevalence and improve outcomes, including mortality, MACEs, and urgent dialysis initiation. In the Visualize‑HD study of 50 983 patients undergoing HD from 231 HD centers in China, the prevalence of hyperkalemia (potassium level >5 mmol/l) was 40.84%.²¹ In addition, 91.5% of the patients were dialyzed using a 2‑mmol/l potassium bath. Only 7.7% of the participants used potassium binders.

The lack of data on residual kidney function is the major limitation of available studies.^{6,10,11,13,14,20,21} Even in the DIALIZE‑Outcomes trial,²² which evaluated the effect of sodium zirconium cyclosilicate, a novel potassium binder, on arrhythmia‑related cardiovascular outcomes in 2690 hemodialyzed patients with predialysis hyperkalemia (serum potassium >5.5 mmol/l), no data on residual diuresis were provided.