Introduction Patients with chronic kidney disease (CKD) are predisposed to heart rhythm disorders, including atrial fibrillation (AF)/atrial flutter, supraventricular tachycardias, ventricular arrhythmias, and sudden cardiac death (SCD). While treatment options, including drug, device, and procedural therapies, are available, their use in the setting of CKD is complex and limited. Patients with CKD and end-stage kidney disease (ESKD) have historically been under-represented or excluded from randomized trials of arrhythmia treatment strategies, 1 although this situation is changing. 2 Cardiovascular society consensus documents have recently identified evidence gaps for treating patients with CKD and heart rhythm disorders. 3–7 To identify key issues relevant to the optimal prevention, management, and treatment of arrhythmias and their complications in patients with kidney disease, Kidney Disease: Improving Global Outcomes (KDIGO) convened an international, multidisciplinary Controversies Conference in Berlin, Germany, titled CKD and Arrhythmias in October 2016. The conference agenda and discussion questions are available on the KDIGO website (http://kdigo.org/conferences/ckd-arrhythmias/; 13 February 2018). Atrial fibrillation and stroke in kidney disease Epidemiology Atrial fibrillation is the most common sustained arrhythmia. 8 Chronic kidney disease affects 10% of adults worldwide, 9 and patients with CKD have an increased burden of AF compared with those without CKD (Supplementary material online, Table S1 ). The prevalence of AF is high: estimates range from 16% to 21% in CKD patients not dependent on dialysis 10–12 and 15% to 40% in patients on dialysis (Supplementary material online, Table S1 ). 13–18 Chronic kidney disease and AF share many risk factors, making it difficult to discern the contributions of individual factors to either condition or associated outcomes (Figure 1 ). For non-dialysis CKD, there seems to be an independent relationship between CKD and the risk of AF, 19–25 although this association has not been well characterized across the spectrum of estimated glomerular filtration rate (eGFR) or proteinuria. 13 , 14 , 26 , 27 In the USA, both incidence and prevalence of AF are increasing among haemodialysis patients, 27 , 28 which could be because of older age of patients, better ascertainment of AF, and improved survival after vascular events. Figure 1 Relationship between chronic kidney disease and atrial fibrillation: shared risk factors and outcomes. Chronic kidney disease and atrial fibrillation share a number of risk factors and conditions that promote their incidence, possibly via systemic processes such as inflammation, oxidative stress, or fibrosis. It is established that chronic kidney disease increases the incidence of atrial fibrillation and there is some evidence to suggest that atrial fibrillation also increases chronic kidney disease progression. When examining the strength of these associations, we acknowledge the potential impact of detection bias in observational studies where more frequent exposure to healthcare likely prompts more clinical findings in this comorbid population. AF, atrial fibrillation; CKD, chronic kidney disease; CVD, cardiovascular disease. Consequences of atrial fibrillation in chronic kidney disease The risk of stroke is elevated in non-dialysis 29–32 and dialysis 29 , 31 , 33 CKD (Supplementary material online, Table S2 ). Separately, both CKD and AF are risk factors for stroke, but it is currently unknown whether the prognostic significance of CKD markers and AF is independent or interdependent. The association between AF and CKD may be bidirectional; AF may predict new-onset low GFR and proteinuria. 21 In CKD, the adjusted risk ratios of stroke with AF have varied considerably across CKD subpopulations, ranging from 4.2 in women in the general population, 34 1.3 in dialysis patients, 33 , 35 and with modestly significant (1.4) 36 and non-significant 37 associations after kidney transplantation. These differences may be due to greater competing risk of death in more advanced CKD, 35 a higher baseline risk of stroke in CKD without AF, or a higher prevalence of unrecognized AF. AF increases the risk of incident CKD and progression to ESKD 21 , 38–40 (Supplementary material online, Table S3 ), and increases risk of death in patients with non-dialysis CKD and those on dialysis. 13 , 35 , 41 , 42 Other outcomes related to AF, including heart failure, SCD, and myocardial infarction (MI), require further research. The contribution of AF as a mediator of stroke in CKD, as well as the stroke subtypes observed, requires further study. The competing risk of death in CKD may reduce the importance of the contribution of AF to stroke, which could mitigate the effectiveness of some stroke prevention strategies. 35 Stroke risk scores The predictive value and calibration of the CHADS2 and CHA2DS2-VASc stroke prediction scores have only been evaluated in dialysis patients, in which performance appears to be similar to their performance in the general population. 16 , 33 , 43 , 44 Inclusion of CKD in risk scores to improve stroke prediction has demonstrated variable results. Adding two points for creatinine clearance 95 Adjusted dose (INR 2–3) 5 mg b.i.d. 150 mg b.i.d. 60 mg QD d 20 mg QD 51–95 Adjusted dose (INR 2–3) 5 mg b.i.d. 150 mg b.i.d. 60 mg QD 20 mg QD 31–50 Adjusted dose (INR 2–3) 5 mg b.i.d.(eCrCl cut-off 25 mL/min) 150 mg b.i.d. or 110 mg b.i.d. e 30 mg QD 15 mg QD INR, international normalized ratio. a Cockcroft-Gault estimated creatinine clearance (eCrCl). b Apixaban dose modification from 5 mg b.i.d. to 2.5 mg b.i.d. if patient has any two of the following: serum creatinine ≥1.5 mg/dL, age ≥80 years, or body weight ≤60 kg. c In the ENGAGE-AF TIMI 48 study, the dose was halved if any of the following: eCrCl of 30–50 mL/min, body weight ≤60 kg, or concomitant use of verapamil or quinidine (potent P-glycoprotein inhibitors). d This dose has not been approved for use by the US Food and Drug Administration in this category of kidney function. e In countries where 110 mg b.i.d. is approved, clinicians may prefer this dose after clinical assessment of thromboembolic vs bleeding risk. This dose has not been approved for use by the US Food and Drug Administration. Although efficacy (prevention of stroke and systemic embolism) may merely be non-inferior to warfarin, the safety profile of DOACs compared to warfarin does appear to be superior. In all pivotal RCTs, DOACs have been associated with a significant reduction (about 50%) in risk of intracranial haemorrhage compared to warfarin. Among patients with eCrCl between 25 and 50 mL/min, treatment with apixaban and edoxaban resulted in significantly fewer major bleeding events compared with warfarin (Figure 2 ). 63 Although these observations do not necessarily indicate the superiority of apixaban and edoxaban relative to other DOACs, it may be helpful to clinicians when treating patients at particularly high-bleeding risk or low time in therapeutic range (TTR) values while receiving warfarin or other vitamin K antagonists (VKAs). Figure 2 Efficacy and safety of direct oral anticoagulants (DOACs) vs. warfarin in the subgroup of patients with moderate chronic kidney disease from randomized controlled trials in atrial fibrillation. Comparison of hazard ratios and 95% confidence intervals for primary efficacy and safety outcomes for 150 and 110 mg dabigatran twice daily, 15 mg rivaroxaban once daily, 5 mg apixaban twice daily, and 30 mg edoxaban once daily. Chronic kidney disease was defined as estimated creatinine clearance of 30 to 49 mL/min or as 25 to 49 mL/min for apixaban. aApixaban 2.5 mg twice daily if patient had any two of the following: age ≥ 80 years, body weight ≤ 60 kg, or serum creatinine ≥1.5 mg/dL. Reproduced from Qamar and Bhatt 63 with permission from the publisher. Chronic kidney disease G4, G5, and G5D In the absence of trial data, the results from observational studies on the efficacy and safety of anticoagulation for stroke prevention in CKD patients with eCrCl 75%) likely contributed to these findings and has been difficult to replicate in other health systems. 68 A large US health care system analysis found that CKD severity is associated with decreased TTR despite similar INR monitoring intensity. 69 These findings suggest that TTR is more likely to be poor in CKD and can mediate the increased stroke and bleeding risk in CKD. 70 VKAs may lead to CKD via repeated subclinical glomerular haemorrhages 71 or through accelerated tissue or vascular calcification. 72 Table 2 Chronic kidney disease categories lacking randomized clinical trial data on the utility of anticoagulation 4 , 63 , 64 eCrCl (mL/min) a Warfarin Apixaban b Dabigatran Edoxaban Rivaroxaban 15–30 Adjusted dose for INR 2–3 could be considered 2.5 mg PO b.i.d. could be considered Unknown (75 mg PO b.i.d.)c,d 30 mg QD e could be considered 15 mg QD could be considered 90% Hepatic metabolism No Serum creatinine may increase, but no dose adjustment is needed Bisoprolol 30% 50% excreted unchanged in urine No Dose may need to be reduced in advanced CKD Metoprolol 12% Hepatic metabolism Yes No dosage reduction needed Carvedilol 99% Mainly biliary and 16% urinary No Specific guidelines for dosage adjustments in renal impairment are not available; it appears no dosage adjustments are needed Labetalol 50% Inactive metabolites excreted in urine (5% unchanged) and bile No Dose reduction recommended in the elderly Verapamil 90% 70% is excreted in the urine and 16% in faeces No Dose reduction by 20–25% if CrCl 35%. Available data seem to suggest that the benefit of ICDs decreases with declining GFRs, in relationship to competing risks of comorbidity and mortality and high risk of complications. 129 , 161 Studies with subcutaneous defibrillators, which do not have transvenous hardware, are needed since this approach might be associated with fewer and less severe complications, such as infection. 162 Wearable cardioverter defibrillators may provide protection for a limited high-risk period. 145 Further assessment of pacing devices for bradyarrhythmias (including leadless pacemakers) is needed. 146 Potassium homeostasis and handling in chronic kidney disease and dialysis Electrolyte abnormalities and risk for cardiovascular or arrhythmic events Although definitive evidence for causality is lacking, both hyperkalaemia and hypokalaemia have been associated with higher risk of all-cause and cardiovascular mortality in patients with ESKD. In patients on haemodialysis, when pre-dialysis serum potassium values (i.e. potassium values on blood drawn at the start of the haemodialysis procedure, in keeping with clinical practice) rise or fall away from 5 mEq/L, the risk for sudden cardiac arrest increases. 147 Among incident haemodialysis patients, higher mortality and hospitalization rates have been documented to occur immediately after the 2-day interdialytic interval. 163 , 164 A contributing factor may be larger fluid accumulation followed by excessive ultrafiltration and abrupt fluctuations in serum potassium concentrations (Supplementary material online, Figure S1 ). 165 In contrast, hypokalaemia is more common in patients on peritoneal dialysis, and hypokalaemia has been associated with increased risk of all-cause, cardiovascular, and infectious mortality in this subgroup of patients. 166 Treatment options for improving potassium homeostasis Treatments for hyperkalaemia include dietary restriction, correction of acidosis, increasing distal sodium load, and loop diuretics, and in the case of hypokalaemia, potassium-sparing diuretics and potassium supplements could be used. 167 It may be possible to reduce the dose or stop drugs that interfere with potassium homeostasis, such as nonsteroidal anti-inflammatory drugs, sulfamethoxazole-trimethoprim, calcineurin inhibitors, and non-selective beta blockers. Pharmacologic treatments for managing hyperkalaemia include the cation-exchange resin kayaexalate, 168 calcium-resin resonium, 169 the potassium-binding polymer patiromer, 170 and the potassium trap ZS-9. 167 Beyond the treatment of hyperkalaemia, these agents might also enable more patients with concomitant CKD to be started on or maintained on guideline-recommended renin–angiotensin–aldosterone system (RAAS) inhibitors, and this possibility is currently being investigated. 167 In addition to reducing serum potassium, patiromer has been shown to reduce serum aldosterone levels in patients with CKD and hyperkalaemia taking RAAS inhibitors. 171 Other important questions regarding potassium binders relate to their safety and efficacy in post-kidney transplant patients, patients with Type IV renal tubular acidosis, or patients taking calcineurin inhibitors. Data from three clinical trials have indicated that dual RAAS blockade therapy increases the risk of hyperkalaemia in patients with CKD. 172–174 Meta-analysis data have indicated that mineralocorticoids can mediate hyperkalaemia in patients undergoing dialysis, but large trials are needed to better evaluate this process and its clinical significance. 175 In patients with Type 2 diabetes, a sodium-glucose cotransporter 2 (SGLT2) inhibitor has been associated with small mean changes in serum electrolytes and less hyperkalaemia compared to placebo, especially in patients taking anti-hypertensives that interfere with potassium excretion. 176 Dialysate and dialysis parameters For patients undergoing haemodialysis, both the potassium concentration in the dialysate and the schedule of haemodialysis treatments affect the risk of sudden death (Figure 6 ). Potential confounding factors, such as nutrition, treatment compliance, and comorbidities, have not been thoroughly evaluated. It is also not clear whether or how much central venous pressure, hypervolemia, and pulmonary hypertension predispose patients to arrhythmic events. Three studies have indicated that a low potassium dialysate concentration ( 5 mEq/L, the risks associated with low potassium dialysates have not been statistically significant. In Dialysis Outcomes and Practice Patterns Study (DOPPS), mortality rates were similar in patients prescribed 2 and 3 mEq/L dialysate. 179 Rapid correction of acidaemia, low serum or dialysate calcium, and high ultrafiltration rates may contribute to the arrhythmogenic potential of low potassium dialysate. 147 , 180 In a study of 50 patients undergoing thrice-weekly dialysis, risk of SCD and significant arrhythmias was greater during the 72-h vs. 48-h breaks. There were no analyses specifically related to potassium levels in these studies. 146 Whether shortening the interval between haemodialysis sessions could result in clinically significant reductions in sudden cardiac arrest and its relationship to potassium levels is not clear and warrants further study. Dialysate concentrations of bicarbonate, calcium, magnesium, and glutamic acid also are likely to be relevant to risk for arrhythmic events. It is possible that personalizing dialysis parameters for individual patients could reduce risk of SCD, but this is untested and would be logistically complicated to implement. Fluid control during dialysis Ultrafiltration rates higher than 10 mL/h/kg have been associated with a higher likelihood of intradialytic hypotension and risk of mortality. 181 Haemodynamic stress during dialysis induces cardiac stunning, which over time may progress to the development of regional fixed systolic dysfunction, consistent with underlying myocardial hibernation and fibrosis. 182 A retrospective analysis has indicated that greater interdialytic weight gain is associated with an increased risk of cardiovascular morbid events 183 ; therefore, strategies that mitigate interdialytic weight gain warrant investigation. Conclusion People with CKD have an increased burden from AF relative to those without CKD, and an elevated risk of stroke. For preventing stroke in patients with eCrCl 30–50 mL/min, DOACs are non-inferior to warfarin and have a more favourable safety profile. For CKD G5D patients with AF, there are insufficient clinical efficacy and safety data to routinely recommend VKA treatment for preventing stroke. Evidence from older randomized trials indicates that pharmacological rhythm and rate control strategies are equivalent in terms of their efficacy on risks of heart failure, stroke, and survival. However, catheter ablation, which is superior to antiarrhythmic drug therapy for freedom from AF recurrence, has comparable safety in CKD and non-CKD. The role of AF ablation may continue to evolve, particularly among other co-morbid conditions such as heart failure. Regardless of whether a rhythm or rate strategy is pursued, anticoagulation should also be prescribed unless otherwise contraindicated based on stroke risk. The risk for SCD is increased in patients with CKD, and for those with ESKD on dialysis, several factors that increase risk have been identified. Studies are needed to identify risk factors for SCD in CKD non-dialysis patients. For preventing SCD in ESKD, primary prevention ICD therapy is indicated in patients with LVEF ≤ 35%, although data on its benefits in these patients are not encouraging. Data regarding secondary prevention ICD therapy indicate some benefits, but further studies are needed to assess long-term risk–benefit ratios in these patients. Available data seem to suggest that the benefit of ICDs decreases with declining GFR. For patients undergoing haemodialysis, both the potassium concentration in the dialysate and the schedule of haemodialysis treatments affect the risk of sudden death. Whether shortening the interval between haemodialysis sessions could result in clinically significant reductions in sudden cardiac arrest is not yet clear and warrants further study. It is possible that personalizing dialysis parameters for individual patients could reduce risk of SCD, but this is untested and would be logistically complicated to implement. Recent guidelines include considerable practical and scientific detail on management of these arrhythmias in CKD. 3–7 , 85 , 184 However, there remain substantial evidence gaps, which will require clinical trials, and when not possible, robust observational data. We have outlined research recommendations in the hopes that future investigations can better advance the evidence base in this area (Table 6 ). A multidisciplinary approach is vital for understanding the mechanisms of arrhythmias in CKD as well as for evaluating therapies and improving clinical care. Nephrologists and cardiologists should initiate and continue partnerships in designing and conducting clinical trials as well as treating individual patients with CKD and AF. Table 6 Arrhythmias and chronic kidney disease: current knowledge gaps and future research recommendations Should AF be a required secondary endpoint in future cardiovascular clinical trials among CKD patients? This will enable future studies to examine the contribution of AF to various outcomes (e.g. cognitive impairment). Can we improve upon risk assessment in patients with CKD/CKD G5D by examining unique risk factors for stroke (e.g. proteinuria) and bleeding (e.g. proteinuria, platelet dysfunction, vascular access, dialysis anticoagulation)? Based on a review of large observational studies, can we ascertain the combinations of risk factors that predict competing SCD vs. non-SCD and cardiac vs. non-cardiac death endpoints in patients with CKD/CKD G5D? Are there modifiable risk factors (e.g. long chain omega-3 fatty acids) or pharmacological therapies for SCD worth investigating? What is the incidence and prognostic significance of syncope in dialysis patients (on conventional or novel modalities) and transient hypotension, hypovolemia, and bradycardia during and outside dialysis sessions? Is there a role for biomarkers (e.g. troponins, BNP) and markers of autonomic dysregulation and sympathetic overactivity in predicting cardiac death and SCD? Is there prognostic significance in incidentally detected arrhythmias? Among patients on dialysis, can we use modern imaging techniques (e.g. cardiac magnetic resonance imaging with T1 mapping and speckle tracking imaging echocardiography both during haemodialysis and on a non-dialysis day), long-term ECG monitoring, and emerging biomarkers to ascertain predisposing factors to SCD? Since patients with CKD G5D have consistently lower time in TTR values (despite comparable intensity of monitoring) that may contribute to higher risk of bleeding, what is the evidence regarding the role of TTR in decision-making and transitioning to DOAC therapy with suboptimal TTR? Estimates of kidney function using eGFR and eCrCl are not equivalent and can lead to important dose discrepancies with DOACs. Both the conference participants and ESC advocate the use of eGFR (over eCrCl) in future trials because of established superiority in estimating kidney function and to reconcile the measure used in pragmatic clinical practice. For adoption of this measure in future trials however, we recognize that there would be need for upfront endorsement of eGFR as the preferred measure for estimating kidney function by regulatory agencies. Should serial measurements of kidney function be considered to determine if anticoagulation (e.g. DOACs) is associated with changes in kidney function? Does heparin use during haemodialysis alter the risk–benefit ratio when used with concomitant oral anticoagulation? Are there clinical efficacy or safety data evaluating whether the use of erythropoietin therapy influences stroke reduction with anticoagulant therapy? Is there utility in employing left atrial appendage occluder devices in patients with CKD G5D who are already at high risk of bleeding and endovascular infections? What is the role of DOACs among kidney transplant patients? Do specific drug–drug interactions favour certain agents over others? Is ICD therapy efficacious in the primary and secondary prevention of SCD in ESKD? If so, what are the risk–benefit ratios? Utility of leadless pacemakers? Additional studies examining transvenous, subcutaneous, and wearable defibrillators are needed in CKD patients with EF >35% since they account for 90% of ESKD patients. What are the long-term outcomes of rate vs. rhythm control in CKD or dialysis patients? What should guide the selection of rate vs. rhythm control in this patient population? For the former, what is the optimal rate control and what are the preferred rate-controlling agents? Utility of transvenous vs. leadless permanent pacemaker following AV node ablation? For rhythm control, what is benefit–risk ratio for ablation vs. antiarrhythmic drugs? What is the ideal ablation approach? For antiarrhythmic drugs, are there comparative trials to provide information on safety, pharmacokinetics and efficacy on various agents (especially amiodarone)? Is there a long-term need for oral anticoagulation in patients with successful rhythm control? Does personalizing dialysis prescription (e.g. electrolyte dialysate, close monitoring of potassium levels or volume management) reduce the risk for SCD? Do changes in other electrolytes associated with arrhythmic predisposition in haemodialysis patients (such as magnesium) affect clinical outcomes? AF, atrial fibrillation; AV, atrioventricular; BNP, B-type natriuretic peptide; CKD, chronic kidney disease; DOAC, direct oral anticoagulant; ECG, electrocardiogram; eCrCl, estimated creatinine clearance; EF, ejection fraction; eGFR, estimated glomerular filtration rate; ESC, European Society of Cardiology; ESKD, end-stage kidney disease; G5D, CKD stage G5 patients on dialysis therapy; ICD, implantable cardioverter-defibrillator; SCD, sudden cardiac death; TTR, time in therapeutic range. Supplementary material Supplementary material is available at European Heart Journal online. Supplementary Material Supplementary Data Click here for additional data file.