Delayed post‐dialysis recovery times are associated with relative changes in intracellular and extracellular fluid ratios between different body compartments

Hemodialysis patient groups have advocated reducing dialysis fatigue and symptoms. We investigated whether compartmental fluid shifts were associated with peri‐dialytic fatigue and symptoms.


| INTRODUCTION
Although hemodialysis is a routine outpatient lifesustaining treatment for kidney failure patients, many patients report a reduced quality of life [1], with fatigue being one of the most frequently reported symptoms which impacts on their quality of life [2,3].Even though fatigue may be reported on non-dialysis days, many patients feel tired and suffer with post-dialysis fatigue [4].In addition, patients may suffer with a range of peridialytic symptoms, including muscle cramps, chest pain, dizziness, headache, and nausea [5].
Consensus meetings of dialysis stakeholders, clinicians, patients, and their carers identified symptom relief as a key priority for further research to guide changes in clinical practice to reduce the prevalence and severity of peri-dialytic symptoms [2,6].As such, several centres have developed specific symptom questionnaires to record peri-dialytic symptoms, in particular targeting post-dialysis fatigue [7,8].Despite this, relatively little is known about the key factors which induce post-dialysis fatigue and cause peri-dialytic symptoms [4].
There appears to be little or no consensus on the drivers to the cause of post-dialysis fatigue, with some studies reporting an association with patient factors such as female gender, age, frailty, comorbidity, and depression [9][10][11][12]; and dialysis factors including higher ultrafiltration rates, lower dialysate sodium, intra-dialytic hypotension, dialysis modality, session duration, and frequency [11,[13][14][15][16][17], whereas other observational reports failed to confirm these associations [8,[18][19][20][21].As most of these studies either reported on a single session or asked patients to reflect on a period of dialysis treatment, it is unclear as to which specific factors associated with a dialysis session may contribute to post-dialysis fatigue.Thus, to determine which factors, associated with the dialysis prescription and treatment, affect recovery times, we reviewed self-reported post-dialysis fatigue times in a series of patients who had reported both a short recovery time of <1 h with one dialysis session and delayed recovery, ≥1 h with another session.

| PATIENTS AND METHODS
Following United Kingdom National Health Service (NHS) guidelines requiring patient feedback on treatment, all adult patients attending for their routine midweek outpatient dialysis treatment under the care of a university hospital, were asked to self-report the time to recover from dialysis, using a previously reported visual analogue scale as less than 1 h and ≥1 h [5].Patients who had recently been admitted to hospital as an acute emergency, or those with a recent infection were excluded.They were also requested to self-report the frequency of dialysis-associated symptoms, including fatigue, feeling cold, cramps, dizziness, headache, nausea, abdominal pain, back ache, pruritus, and shortness of breath [5].Psychological distress was assessed using the distress thermometer visual analogue scale [22].Medical history and frailty assessments, determined with the Canadian geriatric frailty score [23], were obtained from computerized medical records, along with the UK Stoke-Davies comorbidity score [24].
Pre-and post-midweek blood samples were taken for standard analyses on the days that patients completed the two visual analogue scales.The dialysis prescriptions and dialysis session details were retrospectively obtained from hospital computerized records, including episodes of intra-dialytic hypotension requiring nursing interventions (symptomatic) or session records documenting those patients that felt unwell post-dialysis.Patients were dialysed using either a Fresenius 4000H, 5008 dialysis machines (Fresenius MC, Bad Homburg, Germany) or BBraun DialogueR + (BBraun, Melsungen, Germany) with a polysulfone dialyzer (Fresenius MC, Bad Homburg, Germany) either in hemodialysis or post-dilution hemodiafiltration mode, and anticoagulated with a bolus of low molecular weight heparin (Tinzaparin, Leo Laboratories, Princes Risborough, UK).All dialysates used ultrapure quality dialysis water with a final dialysate concentration of bicarbonate 32 mmol/L, acetate 3.0 mmol/ L, magnesium 0.5 mmol/L, and glucose 5.5 mmol/L, with clinicians individualizing the prescription of dialysate sodium, potassium, calcium, and temperature.Dialysis conductivity modules were regularly calibrated and checked [25].We used constant ultrafiltration profiles and ultrafiltration rates were calculated as the difference in pre-and post-dialysis weights divided by sessional time and adjusted for patient weight.
Blood pressure was measured in the sitting position immediately prior to and at the end of the dialysis treatment using integrated automatic oscillometric devices.Multifrequency bioimpedance measurements (InBody 720, Seoul, South Korea), were made pre and post the corresponding mid-week dialysis session, using a standardized protocol, to determine intracellular water (ICW), extracellular water (ECW), and body composition [26,27].Bioimpedance devices were regularly serviced and calibrated.

| Ethics
This audit of clinical service development complied with the UK NHS guidelines for clinical audit and service development and was registered with the hospital and met Health Research Authority guidelines.In keeping with UK guidelines, all patient data was anonymized prior to analysis (https://www.hra.nhs.uk).

| Statistical analysis
Data are presented as mean ± standard deviation, median (interquartile range), or as percentage.Standard statistical tests were used to analyze data; D'Agostino-Pearson normality test, Chi square test (χ 2 ), paired t-test for normally distributed data, and Wilcoxon rank sum pair test for nonparametric data, with appropriate corrections made for multiple testing, by Bonferoni or Tukey.Univariate analysis was by Spearman rho, and logistical regression models were created to investigate variables associated wth delayed recovery and also comparing those between dialysis sessions with a recovery time of <1 h and ≥1 h, and included all variables with a p < 0.1.Variables were then excuded in a step backward analysis, unless they improved the model fit.The models were checked for collineraity and variable inflation factor.Statistical analysis used Prism 10 (Graph Pad, San Diego, CA, USA) and Social and Political Sciences statistical package (SPSS 28.0; IBM, Armonk, NY, USA).Statistical significance was taken as p < 0.05.

| RESULTS
Patients were surveyed 6 months apart and a total of 645 patients returned their peri-dialytic recovery and symptom visual analogue scores following a single dialysis session in 2017 and 2018, and 124 patients returned two visual analogue scores, one reporting a very short recovery time of <1 h, and one with a recovery time of ≥1 h (62.1% between 1 and 4 h, 16.1% between 4 and 8 h, 4% between 8 and 12 h, and 17.7% >12 h).The mean age of the cohort was 66.0 ± 14.8 years, 66.1% male, 52.4% diabetic, and 12.1% current smokers.Past medical history included hypertension 93.7%, myocardial infarction 11.3%, coronary artery by-pass surgery 9.7%, coronary artery stenting 7.3%, valvular heart disease 8.1%, peripheral vascular disease 7.9%, stroke 11.9%, transient ischemic attack 3.1%, cancer 4%, and cirrhosis 2.4%.Nine (7.3%) patients were prescribed opiates/gabapentin for pain, 6.5% antidepressants or antipsychotics, and 4.8% sleeping medications.Three patients changed medications between the different dialysis sessions.Two patients were prescribed oral vasopressors.
When we analyzed delayed recovery time sessions, then patients reporting a recovery time of >12 h reported more post-dialysis fatigue (p = 0.005), cramps (p = 0.003), headache (p = 0.007), and dizziness (p = 0.027), than those with a recovery time of 1-4 h.
We then compared patient demographics and medications at the time of the dialysis sessions with a recovery time of <1 h and ≥1 h and are described in Table 1.Apart from a higher distress thermometer score, there were no other differences in pre-dialysis patient demographics, medications, or blood tests including hemoglobin, albumin, C-reactive protein, or N-terminal brain natriuretic peptide.Similarly, there were no statistically significant differences in dialysis prescriptions, in terms of weight loss and ultrafiltration rates, serum to dialysate sodium gradient, nursing interventions, and patient clinical state postdialysis (Table 2).Treatment with hemodiafiltration was the standard of care; however, some patients had been recruited into the H4RT study and randomized to hemodialysis [28], and more of the short recovery time sessions were with hemodiafiltration rather than high-flux hemodialysis mode (χ 2 = 4.6, p = 0.032).There were no differences in patient age, dialysis vintage or session time, and ultrafiltration rate between hemodialysis and hemodiafiltration sessions (Appendix A, Table A1).
Reviewing pre-and post-bioimpedance measurements, then there appeared to be no differences in either total body ICW or ECW, although the median ratio of ECW/ICW post dialysis was slightly but not statistically lower (Table 3).We then compared the change in segmental ECW/TBW ratios following dialysis and the relative change between body segments (Figure 1).The relative ratio of ECW/TBW was lower in those treatments associated with longer peri-dialytic recovery times when comparing the left arm with other body compartments.
When other patient self-reported peri-dialytic symptoms were reviewed, then fatigue was the only individual symptom to be reported to be significantly greater with the dialysis session associated with longer recovery time, although the total peri-dialytic score was much higher with the delayed recovery dialysis session (Table 4).On univariate analysis, delayed recovery time was associated with the following individual self-reported peri-dialytic symptoms: fatigue (rho 0.22, p = 0.005), dizziness (rho 0.21, p = 0.007), headache (rho 0.19, p = 0.002), shortness of breath (rho 0.13, p = 0.025), and cramps (rho 0.13, p = 0.04) and the distress thermometer score (rho 0.19, p = 0.006).
A multivariable logistic regression step backward model was generated, and a delayed recovery time was associated with greater distress thermometer and dizziness scores and reduced with hemodiafiltration mode (Table 5).In a separate model to investigate changes in body compartments then the patients with shorter recovery times had a greater relative change in ECW/TBW between the arm and trunk, odds ratio 0.44, 95% confidence limits 0.19-0.98,p = 0.045.

| DISCUSSION
Fatigue is the most commonly self-reported symptom by kidney dialysis patients [4].Although there is an overlap between self-reported fatigue and depression [3], many patients additionally experience a feeling of tiredness following a dialysis session.This additional peri-dialytic fatigue is most likely due to changes in cerebral oxygenation and hypoperfusion during the dialysis session [29], and reports of cerebral cytotoxicity due to edema [30].By using patients as their own controls, we wished to determine which processes of the dialysis treatment were most likely to cause this fatigue, so we compared the dialysis sessions of patients who reported both a relatively quick post-dialysis recovery time of less than 1 hour with another of their dialysis sessions, when then had reported a longer recovery time.We found no difference in predialysis clinical status, medications, blood pressure, volume status, or pre-dialysis laboratory investigations, in keeping with several previous reports [4,[18][19][20].Similarly, there were no statistically significant differences in the prescribed dialysis prescription between the two sessions, in terms of dialysate composition, dialyzer, peridialytic changes in blood pressure, or ultrafiltration rates, supporting earlier studies [4,[18][19][20][21].However, longer recovery times were reported more frequently with high flux hemodialysis sessions compared to those with hemodiafiltration.One previous study reported no difference in recovery time with hemodiafiltration, although onethird of patients in this study reported a zero time to recover [29].Whereas other studies in both adults and children reported shorter recovery times with hemodiafiltration [14,30].Both observational and randomized controlled trials have noted that intra-dialytic hypotensive episodes are less frequent with hemodiafiltration [31], and this may be due to differences in sodium fluxes between convective and diffusion-based dialysis modes, with higher plasma sodium concentration reported with convection [32].
In addition, there was a relatively larger fall in both total body ICW and ECW in those sessions reporting a longer recovery time.To explore this further we compared the change in ECW/TBW ratio between different body segments.If fluid was being removed equally between segments, then the expected ratio would be around 1.0, whereas we found that the ratio was higher when comparing changes in the left arm and some of the other body segments for those dialysis sessions with the longer recovery times.This ratio became progressively higher moving from the contralateral arm to trunk and then the right leg, suggesting a differential change in ECW between body compartments [33].This relative difference in the change in ECW could potentially lead to a more pronounced effect in terms of reducing cerebral oxygenation in this older cohort of patients [34], and so possibly induce greater cytotoxic cerebral edema [35], consequently increasing the sensation of post-dialysis fatigue.This greater fall in ECW would support earlier reports that patients who start dialysis closer to their target weight, with lower ultrafiltration rates are more vulnerable to delayed post-dialysis recovery and fatigue [4,21], and equally excessive ultrafiltration rates could also reduce ECW and increase intercompartmental disequilibrium [11,18].Although not statistically different, the median ultrafiltration rate adjusted for body weight, and percentage body weight change were slightly greater for those sessions with longer recovery times which would support a greater compartmental disequilibrium leading to patients reporting a longer recovery time.
Hemodiafiltration by potentially inducing greater thermal energy losses, along with changes in sodium flux, may improve vascular tone and by reducing the risk of intra-dialytic hypotension, lead to shorter posttreatment recovery times [31].Another study observed that intradialytic exercise reduced post-dialysis fatigue [36].During conventional dialysis, blood flow to muscle is typically reduced, but studies have observed that intradialytic exercise leads to greater dialytic urea clearance due to better muscle blood supply, and as such this may lead to a better balance in fluid removal from the ECW and ICW compartments.Studies have suggested that differences in sodium flux with convective therapies better maintains plasma osmolality and transfer of fluid between compartments, which may partially explain the shorter recovery times reported with hemodiafiltration mode [32].We noted that generally overall those sessions were associated with a greater recovery time, patients also reported more peri-dialytic symptoms, and greater psychological distress with higher distress thermometer scores, in keeping with previous reports [12,37].Whether these changes to the brain increase self-reported distress thermometer scores, or whether greater previous ischemic damage to the brain [34], predisposes to greater vulnerability of the brain to the stresses of dialysis treatments and changes in compartmental fluid volumes remains to be determined.A multivariable logistic model demonstrated that the symptom of dizziness was the most important peri-dialytic symptom associated with delayed post-dialysis recovery, along with higher distress thermometer scores, whereas hemodiafiltration was less likely to be associated with delayed recovery compared to hemodialysis.
As with any observational study, we can only report associations and not causality.We chose a simple cutoff of less than 1 h to determine shorter and longer peri-dialytic recovery times, whereas other studies have used different metrics.In our multiethnic dialysis population, we found using a Likeart scale was a more reliable method of recording peri-dialytic symptoms [5].Our centers used a low dialysate temperature as standard of care, with a constant ultrafiltration profile and dialysate sodium, and most dialysis sessions were with post-dilutional hemodiafiltration treatments, and as such our results may not be equally applicable to other dialysis centers with different clinical practices.Similarly we cannot comment on whether peri-dialytic fatigue would be affected by different ultrafiltration profiles or dialysate electrolyte composition.Although we report pre-and post-bioimpedance measurements, our dialysis machines were not equipped with relative blood volume monitoring (RBVM) to determine whether there may have been differences in RBVM between sessions.
Collaborative workshops between relevant clinicians, health care providers, and kidney dialysis patients have highlighted the need to improve the quality of life for kidney dialysis patients.To improve patient experiences of dialysis treatments it is important to understand the causes of delayed post-dialysis recovery times and peridialytic symptoms.We used patients as their own controls, comparing dialysis sessions when patients reported a short with a longer recovery time.Our study shows that bioimpedance may potentially be useful to determine changes in fluid status between body compartments by measuring changes in ECW/TBW ratios, which are not noted by standard clinical monitoring, and were associated with delayed patient recovery post-dialysis.This technology has the potential to be developed in association with artificial intelligence systems to develop fuzzy logic systems in the future that can potentially help improve the experience of patients when attending for their dialysis session, as patients reported more peridialytic symptoms during sessions associated with a longer dialysis recovery time.

F I G U R E 1
Change in the ratio of extracellular water (ECW) to total body water (TBW) comparing pre-and post-dialysis session measurements in different body compartments compared to those in the left arm with the right arm, trunk, right leg and left leg in patients with shorter (<1 h) and delayed (≥1 h) peri-dialytic recovery time.*p < 0.05; **p < 0.01.T A B L E 4 Patient self-reported peri-dialytic symptom scores.Patients divided according to recovery times into those with short (<1 h) and longer (≥1 h) recovery.Data expressed as median and interquartile range.**<0.01 vs. short recovery time.
Patient demographics, medical history, and predialysis blood results.
T A B L E 1 Bioimpedance measurements pre-and post-dialysis.
T A B L E 3Note: Patients divided according to recovery times into those with short (<1 h) and longer (≥1 h) recovery.Data expressed as median and interquartile range.No statistically significant differences.Abbreviations: ECW, extracellular water; ICW, intracellular water.T A B L E 2 Patient dialysis session records and dialysis prescriptions.Note: Patients divided according to recovery times into those with short (<1 h) and longer (≥1 h) recovery.Data expressed as integer, percentage, mean with standard deviation, or median and interquartile range.*p < 0.05 vs. short recovery time.
T A B L E 5Note: