clinical trial www.kidney-international.org Patients receiving frequent hemodialysis have better health-related quality of life compared to patients receiving conventional hemodialysis OPEN Amit X Garg1, Rita S Suri2, Paul Eggers3, Fredric O Finkelstein4, Tom Greene5, Paul L Kimmel6, Alan S Kliger4, Brett Larive7, Robert M Lindsay1, Andreas Pierratos8, Mark Unruh9 and Glenn M Chertow10; for the Frequent Hemodialysis Network Trial Investigators Division of Nephrology, Department of Medicine, Western University, London, Ontario, Canada; 2Department of Medicine, Section of Nephrology, Centre de Recherche, Centre Hospitalier de l’Université de Montréal (CHUM), Montréal, Québec, Canada; 3National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland, USA; 4Yale School of Medicine and Yale New Haven Health System, New Haven, Connecticut, USA; 5University of Utah Medical Center, Salt Lake City, Utah, USA; 6Division of Kidney Urologic and Hematologic Diseases, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland, USA; 7Department of Quantitative Health Sciences, Cleveland Clinic Foundation, Cleveland, Ohio, USA; Humber River Hospital, University of Toronto, Toronto, Canada; 9Department of Internal Medicine, Division of Nephrology, University of New Mexico, Albuquerque, New Mexico, USA; and 10Department of Medicine, Stanford University School of Medicine, Palo Alto, California, USA Most patients with end-stage kidney disease value their health-related quality of life (HRQoL) and want to know how it will be affected by their dialysis modality We extended the findings of two prior clinical trial reports to estimate the effects of frequent compared to conventional hemodialysis on additional measures of HRQoL The Daily Trial randomly assigned 245 patients to receive frequent (six times per week) or conventional (three times per week) in-center hemodialysis The Nocturnal Trial randomly assigned 87 patients to receive frequent nocturnal (six times per week) or conventional (three times per week) home hemodialysis All patients were on conventional hemodialysis prior to randomization, with an average feeling thermometer score of 70 to 75 (a visual analog scale from to 100 where 100 is perfect health), an average general health scale score of 40 to 47 (a score from to 100 where 100 is perfect health), and an average dialysis session recovery time of to hours Outcomes are reported as the between-treatment group differences in one-year change in HRQoL measures and analyzed using linear mixed effects models After one year in the Daily Trial, patients assigned to frequent in-center hemodialysis reported a higher feeling thermometer score, better general health, and a shorter recovery time after a dialysis session compared to standard thrice-weekly dialysis After one year in the Nocturnal Trial, patients assigned to frequent home hemodialysis also reported a shorter recovery time after a dialysis session, but no statistical Correspondence: Amit Garg, Institute for Clinical Evaluative Sciences Western (ICES Western), Room ELL-108, London Health Sciences Centre, 800 Commissioners Road East, London, Ontario, Canada N6A 5W9 E-mail: amit.garg@lhsc.on.ca Received August 2015; revised 19 October 2016; accepted 27 October 2016 Kidney International (2017) -, -–- difference in their feeling thermometer or general health scores compared to standard home dialysis schedules Thus, patients receiving day or nocturnal hemodialysis on average recovered approximately one hour earlier from a frequent compared to conventional hemodialysis session Patients treated in an in-center dialysis facility reported better HRQoL with frequent compared to conventional hemodialysis Kidney International (2017) j.kint.2016.10.033 -, -–-; http://dx.doi.org/10.1016/ KEYWORDS: clinical trial; daily hemodialysis; health-related quality of life; nocturnal hemodialysis Copyright ª 2017, International Society of Nephrology Published by Elsevier Inc This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) T here is a growing effort to make health care and health policies more patient-centered This requires greater knowledge of how a treatment affects patientreported outcomes, where information comes directly from a patient without interpretation of their treatment response by a clinician.1 Most patients with end-stage kidney disease place enormous value on their health-related quality of life (HRQoL), even over survival, and want to know how their HRQoL will be affected by their choice of dialysis modality.2 More than million people worldwide receive conventional hemodialysis sessions per week to sustain life.3 More frequent hemodialysis (5 or sessions per week) results in greater weekly solute and fluid removal, and may be associated with better HRQoL (as observed in small studies4,5) We previously reported the primary results of parallel 12-month follow-up randomized controlled trials of frequent hemodialysis (Frequent Hemodialysis Network [FHN] Daily and Nocturnal Trials).6,7 One of coprimary outcomes reported in the primary results of these trials was the baseline clinical trial AX Garg et al.: Frequent hemodialysis network randomized trials to 12-month change in a commonly used patient-reported physical health survey score (the Physical Health Composite from the RAND Short-Form 36-item survey8) After year in the Daily Trial, patients assigned to frequent versus conventional in-center hemodialysis reported a better score.6 After year in the Nocturnal Trial, we could not reliably determine whether patients assigned to frequent nocturnal versus conventional hemodialysis truly differed in this score, as the trial did not meet its recruitment target, resulting in estimates with wide confidence intervals.7 Several additional patientreported measures of general well-being were collected in the FHN trials to better assess the multidimensional concept of HRQoL These measures have not been reported elsewhere In this article, we investigated whether frequent hemodialysis compared with conventional hemodialysis affects measures of HRQoL collected in the FHN trials reports, the baseline characteristics of patients assigned to frequent versus conventional hemodialysis were similar in each trial The baseline and follow-up HRQoL measures for patients assigned to frequent versus conventional hemodialysis in each trial are presented in Table The values of the baseline HRQoL measures were similar in patients assigned to frequent compared with conventional hemodialysis in both trials, with the exception of slight differences in the health utilities index and general health scale scores in the Nocturnal Trial The completeness of each HRQoL measure in follow-up is also presented in Table Each measure was completed by more than 80% of eligible patients, after patients who died or received a kidney transplant in follow-up were excluded from consideration (14 patients in the Daily Trial and patients in the Nocturnal Trial, and 24 patients in the Daily Trial and patients in the Nocturnal Trial, respectively) RESULTS a ENROLLMENT Patient selection from both trials is presented in Figure 1a and 1b Baseline characteristics of patients in each trial are summarized in Tables and As confirmed in previous Daily Trial Feeling thermometer The baseline mean (SD) score was 74 (19) in patients assigned to frequent hemodialysis and Enrolled (n = 378) ALLOCATION Allocated to 3x-per-week (n = 120) Allocated to 6x-per-week (n = 125) FOLLOW-UP Deaths (n = 9) Kidney transplants (n = 13) Missing at baseline / end-of-study: Time until resumed activity (n = / 11) Feeling Thermometer (n = / 17) General Health (n = / 5) HUI3 (n = / 6) Deaths (n = 5) Kidney transplants (n = 11) * Missing at baseline / end-of-study: Time until resumed activity (n = / 5) Feeling Thermometer (n = / 13) General Health (n = / 3) HUI3 (n = / 3) ANALYSIS Randomized (n = 245) Excluded (n = 133) 6x-per-week dialysis not feasible (n = 38) MRI not obtained (n = 18) Baseline GFR > 3ml/min (n = 27) Adherence to 6x-perweek dialysis judged unlikely (n = 13) Other reasons (n = 37) Change in time until activity (n = 86) Change in Feeling Thermometer (n = 81) Change in General Health (n = 93) Change in HUI3 (n = 92) Change in time until activity (n = 102) Change in Feeling Thermometer (n = 96) Change in General Health (n = 106) Change in HUI3 (n = 106) *Two patients received transplants late during follow-up and were included in the analyses for 12-month changes for the HUI3, General Health subscore and the Health Thermometer One of the two transplanted patients also contributed to the analysis for 12-month changes in postdialysis recovery time Figure | Flow diagram for Daily Trial and Nocturnal Trial (a) Flow diagram for Daily Trial showing the number of patients enrolled and assigned to each study arm (intervention/control), and number of patients who completed the baseline and 12-month HRQoL measures, including reasons for dropout (continued) Kidney International (2017) -, -–- clinical trial AX Garg et al.: Frequent hemodialysis network randomized trials ENROLLMENT b Enrolled (n = 118) ANALYSIS FOLLOW-UP ALLOCATION Randomized (n = 87) Excluded ((n n = 31) Unwilling/unable to receive home dialysis(n dialysis( n = 9) Home unsuitable for dialysis (n (n = 7) MRI not obtained (n (n = 3) Baseline GFR > 10ml/min (n = 1) Other reasons ((n n = 11) Allocated to 3x-per-week (n = 42) Allocated to 6x-per-week (n = 45) Deaths (n (n = 1) Kidney transplants (n (n = 2) Missing at baseline / end-of-study: Time until resumed activity (n (n = / 2) Feeling Thermometer (n (n = / 3) General Health (n (n = / 0) HUI3 (n (n = / 0) Deaths ((n n = 2) Kidney transplants ((n n = 3) * Missing at baseline / end-of-study: Time until resumed activity ((n n = / 3) Feeling Thermometer ((n n = / 4) General Health (n (n = / 1) HUI3 (n (n = / 1) Change in time until activity (n (n = 36) Change in Feeling Thermometer (n (n = 36) Change in General Health (n (n = 39) Change in HUI3 (n (n = 39) Change in time until activity (n (n = 37) Change in Feeling Thermometer (n (n = 36) Change in General Health (n (n = 39) Change in HUI3 (n (n = 39) *One patients received a transplant late during follow-up and was included in the analyses for 12-month changes for the HUI3, General Health subscore and postdialysis recovery time Figure | (Continued) (b) Flow diagram for Nocturnal Trial showing the number of patients enrolled and assigned to each study arm (intervention/control), and number of patients who completed the baseline and 12-month HRQoL measures, including reasons for dropout 71 (17) in patients assigned to conventional hemodialysis Patients assigned to frequent hemodialysis demonstrated a significant improvement in their feeling thermometer score over the first months of follow-up, whereas the score remained similar in patients assigned to conventional hemodialysis (change of 5.6 vs À0.5 from baseline) These trends persisted to 12 months after randomization (5.8 vs À0.6 from baseline) There was a significant 6.4-point between-treatment group difference in the change in score over year (95% confidence interval [CI] 1.8–11.1) General health scale The baseline mean (SD) score was 47 (22) in patients assigned to frequent hemodialysis and 44 (22) in patients assigned to conventional hemodialysis Patients assigned to frequent hemodialysis demonstrated a significant improvement in their general health scale score over the first months of follow-up, whereas the score remained similar in patients assigned to conventional hemodialysis (change of 6.5 vs À1.6 from baseline) These trends persisted to 12 months after randomization (6.3 vs À3.4 from baseline) There was a significant 9.7-point between-treatment group difference in the change in score over year (95% CI 4.7–14.7) Time to recover from a dialysis session Between baseline and 12 months of follow-up, in patients assigned to frequent Kidney International (2017) -, -–- hemodialysis the median recovery time (10th, 90th percentile) decreased from 150 minutes (5, 1440) to 60 minutes (0, 180), whereas in patients assigned to conventional hemodialysis the median recovery time remained similar from 120 minutes (0, 480) to 180 minutes (15, 1440) (Figure 2) The mean between-treatment group difference in the change in recovery time over year was À84 minutes (95% CI À89 to À80; P < 0.0001) A recovery time of 60 minutes or more at baseline was observed in 80% of patients assigned to frequent hemodialysis and 79% assigned to conventional hemodialysis Corresponding numbers year after randomization were 43% and 63%, respectively Health utilities index There was no consistent statistical difference between frequent and conventional hemodialysis in the change in the health utilities index over and 12 months of follow-up, or its attributes (Appendix S1) Nocturnal trial Feeling thermometer The baseline mean (SD) score was 74 (16) in patients assigned to frequent hemodialysis and 75 (13) in patients assigned to conventional hemodialysis Patients assigned to frequent hemodialysis demonstrated an improvement in their feeling thermometer score over the first months of follow-up while the score remained similar in clinical trial AX Garg et al.: Frequent hemodialysis network randomized trials Table | Baseline characteristics of the Daily Trial Table | Baseline characteristics in the Nocturnal Trial n (%) Mean ± SD Median [25th, 75th percentiles] Characteristics Age (yr) Male Black race Education (completed high school or less) Primary language English Duration of end-stage renal disease (yr) 5 Diabetes Charlson Comorbidity Index Predialysis serum albumin (g/dl) Predialysis hemoglobin (mg/dl) Dialysis Std Kt/Vurea Antidepressant Opioid Physical Health Composite (RAND short form-36) Mental Health Composite (RAND short form-36) Beck Depression Inventory score All patients (N [ 245) times/wk (n [ 120) times/wk (n [ 125) 50.4 Ỉ 13.9 50 [42, 59] 151 (61.6) 102 (41.6) 109 (45.1) 52.0 Ỉ 14.1 52 [43, 60] 73 (60.8) 53 (44.2) 53 (44.5) 48.9 Ỉ 13.6 47 [41, 58] 78 (62.4) 49 (39.2) 56 (45.5) 196 (80.0) 101 (84.2) 95 (76.0) 72 (29.4) 76 (31.0) 97 (39.6) 100 (40.8) 1.82 Ỉ 1.95 [0, 3] 3.94 Ỉ 0.42 35 (29.2) 42 (35.0) 43 (35.8) 50 (41.7) 1.88 Ỉ 2.03 [0, 3] 3.94 Ỉ 0.46 37 (29.6) 34 (27.2) 54 (43.2) 50 (40.0) 1.76 Ỉ 1.89 [0, 3] 3.94 Ỉ 0.37 11.9 Ỉ 1.26 12.0 Ỉ 1.24 11.9 Ỉ 1.28 2.52 Ỉ 0.35 35 (14.3) 42 (17.1) 38.1 Ỉ 10.5 2.53 Ỉ 0.39 15 (12.5) 19 (15.8) 38.0 Ỉ 9.7 2.50 Ỉ 0.31 20 (16.0) 23 (18.4) 38.1 Ỉ 11.2 45.1 Ỉ 11.8 46.0 Ỉ 10.3 44.3 Ỉ 13.0 12.5 Æ 9.0 12.4 Æ 9.5 12.6 Æ 8.6 Std, standard patients assigned to conventional hemodialysis (change of 3.7 vs À0.4 from baseline) This change persisted to 12 months after randomization (4.0 vs À4.1 from baseline) However, the between-treatment group difference in the change in score by year was not statistically significant (8.0, 95% CI À0.5 to 16.1) General health scale The baseline mean (SD) score was 40 (18) in patients assigned to frequent hemodialysis and 45 (20) in patients assigned to conventional hemodialysis Patients assigned to frequent hemodialysis demonstrated a greater improvement in their general health scale by 12 months than patients assigned to conventional hemodialysis (change of 8.0 vs 1.5 from baseline) However, the between-treatment group difference in the change in score by year was not statistically significant (6.6, 95% CI À1.5 to 16.1) Time to recover from a dialysis session Between baseline and 12 months of follow-up, in patients assigned to frequent hemodialysis, the median recovery time (10th, 90th percentile) decreased from 180 minutes (0, 1440) to 30 minutes (0,180), whereas in patients assigned to conventional hemodialysis, the median recovery time changed from 180 minutes (0, 780) to 120 minutes (0, 1440) The between-treatment group difference in the change in score by year was –60 n (%) Mean ± SD Median [25th, 75th percentiles] Characteristics Age (yr) All patients (N [ 87) times/wk (n [ 42) times/wk (n [ 45) 52.8 Ỉ 13.6 54 [43, 69] 57 (65.5) 23 (26.4) 34 (39.5) 54.0 Ỉ 12.9 54 [45, 62] 28 (66.7) 11 (26.2) 14 (33.3) 51.7 Ỉ 14.4 53 [42, 64] 29 (64.4) 12 (26.7) 20 (45.5) 36 (85.7) 41 (91.1) 30 (71.4) (11.9) (16.7) 18 (42.9) 1.88 Ỉ 1.93 [0, 3] 3.92 Ỉ 0.51 28 (62.2) (17.8) (20.0) 19 (42.2) 1.58 Ỉ 1.57 [0, 2] 3.90 Æ 0.48 11.9 Æ 1.09 11.6 Æ 1.12 2.34 Æ 0.34 10 (23.8) (16.7) 38.2 Ỉ 8.3 2.35 Ỉ 0.28 12 (26.7) 11 (24.4) 37.5 Ỉ 9.6 45.9 Ỉ 12.6 45.6 Ỉ 10.5 12.2 Ỉ 9.2 11.2 Ỉ 8.1 Male Black race Education (completed high school or less) Primary language English 77 (88.5) End-stage renal disease vintage (yr) 5 16 (18.4) Diabetes 37 (42.5) Charlson Comorbidity Index 1.72 Ỉ 1.75 [0, 3] Predialysis serum albumin 3.91 Ỉ 0.49 (g/dl) Predialysis hemoglobin 11.8 Æ 1.11 (mg/dl) Dialysis Std Kt/Vurea 2.34 Æ 0.31 Antidepressant 22 (25.3) Opioid 18 (20.7) Physical Health Composite 37.8 Ỉ 8.97 (RAND SF36) Mental Health Composite 45.8 Ỉ 11.5 (RAND SF36) 11.7 Ỉ 8.60 Beck Depression Inventory score Std, standard minutes (95% CI À68 to À53) A recovery time of 60 minutes or more at baseline was observed in 58% of patients assigned to frequent hemodialysis and 63% assigned to conventional hemodialysis Corresponding numbers year after randomization were 24% and 59%, respectively Health utilities index There was no consistent statistical difference between frequent and conventional hemodialysis in the change in the health utilities index over and 12 months of follow-up, or its attributes (Appendix S1) Additional analyses We performed a post hoc subgroup analysis of the Daily Trial and confirmed the observed 12-month effects of frequent hemodialysis on our HRQoL measures did not significantly differ in patients who had a daily urine volume less than or greater than 100 ml at the time of randomization (Appendix S2) We performed a post hoc subgroup analysis of the Daily Trial and found the observed benefit of frequent hemodialysis on our measures of HRQoL did not significantly differ in patients who reported a shorter or longer recovery time after a dialysis session at the time of randomization (Appendix S3) The smaller number of patients in the Nocturnal Trial precluded meaningful subgroup analyses We performed a post hoc analysis Kidney International (2017) -, -–- Adjusted means and treatment effectsb (± SE or with 95% confidence intervals) Observed dataa [10th, 50th, and 90th percentiles] Mean ± SD Month Trial Trt Baseline Month Month 12 Change from baseline Feeling thermometer Daily 3x 6x 70.5 Ỉ 17.1 73.8 Ỉ 18.8 71.9 Ỉ 19.1 79.3 Ỉ 15.2 71.3 Ỉ 20.8 79.0 Ỉ 14.1 0.5 ặ 1.7 ỵ5.6 ặ 1.6 Noct 3x 6x 75.3 Æ 13.3 74.5 Æ 16.5 74.3 Æ 18.4 77.5 Æ 16.1 70.6 Ỉ 20.2 78.1 Ỉ 17.9 À0.4 Ỉ 2.9 þ3.7 Ỉ 2.8 Daily 3x 6x 0.53 Ỉ 0.38 0.57 Æ 0.39 0.60 Æ 0.35 0.62 Æ 0.36 0.63 Æ 0.36 0.60 ặ 0.37 ỵ0.07 ặ 0.03 ỵ0.06 ặ 0.03 Noct 3x 6x 0.64 Ỉ 0.34 0.53 Ỉ 0.32 0.70 Æ 0.32 0.59 Æ 0.33 0.61 Æ 0.35 0.48 Æ 0.38 ỵ0.09 ặ 0.05 ỵ0.01 ặ 0.05 Daily 3x 6x 43.7 Ỉ 21.5 46.8 Ỉ 21.9 42.5 Ỉ 20.6 52.6 Æ 21.2 41.3 Æ 21.5 52.4 Æ 22.7 À1.61 Æ 1.66 6.51 Ỉ 1.58 Noct 3x 6x 45.0 Ỉ 20.4 40.3 Ỉ 18.4 47.8 Ỉ 21.1 45.0 Ỉ 23.7 46.5 Æ 18.5 51.8 Æ 24.5 2.45 Æ 2.99 1.38 Æ 2.95 Daily 3x [0, 120, 480] — [15, 180, 1440] — 6x [5, 150, 1440] — [0, 60, 180] — 3x [0, 180, 780] — [0, 120, 1440] — 6x [0, 180, 1440] — [0, 30, 180] — HUI3 score General health scale Time to recovery (min) Noct ỵ6.1 (1.9 to 10.4) P ẳ 0.0051 ỵ4.2 (3.0 to 11.3) P ¼ 0.25 À0.01 (À0.09 to 0.06) P ¼ 0.75 À0.08 (0.22 to 0.05) P ẳ 0.23 ỵ8.12 (3.84 to 12.40) P ¼ 0.0002 À1.06 (À9.07 to 6.94) P ¼ 0.79 Change from baseline 0.6 ặ 1.9 ỵ5.8 ặ 1.7 4.1 ặ 3.3 ỵ4.0 ặ 3.3 ỵ0.09 ặ 0.03 þ0.04 Ỉ 0.03 À0.01 Ỉ 0.05 À0.08 Ỉ 0.05 À3.35 ặ 1.95 ỵ6.34 ặ 1.83 ỵ1.51 ặ 3.04 ỵ8.07 ặ 3.00 ỵ46 (69 to 161) 79 (83 to 74) ỵ33 (À109 to 176) À60 (À68 to À53) Treatment comparison (6x vs 3x) ỵ6.4 (1.8 to 11.1) P ẳ 0.0071 ỵ8.0 (À0.5 to 16.6) P ¼ 0.066 À0.06 (À0.14 to 0.02) P ¼ 0.17 À0.07 (À0.21 to 0.08) P ¼ 0.36 þ9.69 (4.68 to 14.70) P ¼ 0.0002 þ6.56 (À1.54 to 14.66) P ¼ 0.11 À84 (À89 to À80) P