Patients with normal renal function have a specific osmolality value, above which thirst is generated and fluid ingested. This “setpoint” results in a relatively stable and reproducible plasma sodium level over time, not only in patients without kidney disease1 but also in patients with advanced renal disease.2 Evidence of this sodium setpoint (SP) is also seen in thrice weekly conventional hemodialysis patients.3–5 However, hemodialysis patients lack the mechanisms to regulate body osmolality and fluid balance. Although previous trials examining the clinical effects of different dialysate sodium concentrations have treated predialysis sodium setpoint as stable, this assumption has not been confirmed in quotidian hemodialysis patients.
Lower predialysis sodium setpoint and higher dialysate sodium concentrations led to important clinical outcomes, such as increased blood pressure and interdialytic weight gain (IDWG),6–13 which may affect cardiovascular and all-cause mortality.14,15 Lower predialysis plasma sodium is independently associated with increased all-cause mortality,16,17 thus a change in sodium setpoint might need ongoing monitoring to minimize IDWG and associated cardiovascular morbidity and mortality.
The objective of this study was to determine whether the SP changed with longer or more frequent exposure to the same dialysate sodium concentrations, when patients transitioned from thrice weekly conventional hemodialysis to dialysis modalities differing in duration and frequency.
Materials and Methods
We performed a retrospective observational design that included all patients in the home hemodialysis program of the Southwestern Ontario Regional Renal Program, from 1998 to December 31, 2011. A total of 87 patients, 23 still current and 64 no longer on home hemodialysis, were included. All patients in our study were on conventional thrice weekly hemodialysis in-center before home hemodialysis; some continued intermittent conventional home hemodialysis (ICHD), whereas others changed hemodialysis modality upon transferring from in-center to home hemodialysis.
The modality of home hemodialysis was defined by the duration of dialysis therapy and the frequency of treatments. Short hours daily hemodialysis (SHD) was defined as a minimum of five treatments per week, with a treatment time of 1.5–4.0 hours. Intermittent conventional home hemodialysis implied a maximum of four treatments per week, with treatment times of 1.5–4 hours. Frequent nocturnal home hemodialysis (FNHD) was a minimum of five treatments per week, with a minimum treatment time of 6.0 hours. Intermittent nocturnal home hemodialysis (INHD) meant a maximum of four treatments per week, with a minimum treatment time of 6.0 hours. Dialysate sodium concentration was not individualized as it was a standard 140 mmol/L for all patients at all times.
Blood Sample Collection
In the 50 days before initiation of home hemodialysis, while the patient is on in-center thrice weekly conventional hemodialysis (ICHD-IC), pre- and postdialysis blood samples are taken every 1–2 weeks. Upon transition to home hemodialysis, pre- and postdialysis blood samples are routinely taken each month. Home patients are trained to take blood from the arterial blood line at the start of dialysis and postdialysis, using a standard slow blood and stop dialysate method. The samples are centrifuged and then stored and refrigerated until delivered to the local laboratory for that patient. All patient blood tests are measured using automated and standardized methods. Of interest to this study were predialysis plasma sodium concentrations. Only outpatient blood tests were used, to assure that the patient was at their baseline health status, so that the plasma sodium concentration would not be confounded by acute illness.
Sodium Concentration Measurement
Plasma sodium concentration was measured using Beckman-Coulter LX20 Pro Chemistry Analyzer with ion selective electrodes before and Roche Modular P Chemistry Analyzer with ion selective electrodes after November 4, 2008. This change was made by the London Health Sciences Center because of the need for higher volume of laboratory testing. Both plasma Na+ concentration methods were regularly calibrated; thus, the measurements were treated as equivalent on data analysis. Dialysate sodium concentration was determined using online conductivity measurements built into the Fresenius H series hemodialysis machine, which was used for all patients. Blood glucose was not measured simultaneous to Na+ concentration; thus, plasma sodium levels were not corrected for glucose. Dialysate Na+ concentration measurement is regularly calibrated to assure stability and accuracy of dialysate Na+ concentrations. Home hemodialysis machines were evaluated and calibrated at least once, and usually twice annually, by the program’s water engineer or one of the trained home hemodialysis nurses.
Blood test results were available from the electronic patient record (PowerChart by Cerner) of London Health Sciences Centre. Age (years), sex, diabetes status, residual renal function (ml/minute/1.73 m2), and months of renal replacement therapy before initiation of home hemodialysis were determined from chart review. Residual renal function was calculated within 3 months of conversion to home hemodialysis, as previously described.18
Weights (kg), dialysis treatment times, and frequency were obtained from archived dialysis treatment run sheets. The average monthly values for these were calculated and entered into the study database. For this analysis, a single value for each patient data point was used; the average of the monthly values was used regardless of time period on hemodialysis modality. There were no duplicate observations for any patient.
Because of concerns regarding the use of a standard dialysate of 140 mmol/L sodium concentration and prompted by the observation of high IDWGs in patients undergoing FNHD, a quality assurance investigation was instituted. All laboratory tests had been taken as per routine care protocols; demographic and dialysis treatment information were available from patient records. Once extracted, all data were de-identified before analysis. No patient had to provide blood samples, answer questionnaires, or do anything specific for this study, which was conducted in accordance with the Declaration of Helsinki. Thus, informed written consent was not obtained from the current patients.
Data were analyzed using the Statistical Package for Social Sciences (SPSS, IBM, Armonk, New York) version 19.0. Patients exposed to different dialysis modalities were compared using two-tailed Student’s t-tests for continuous variables and Fisher’s exact test for categorical variables. Statistical significance was achieved with α < 0.05.
The objective of this study was to determine whether the SP changed with longer and more frequent exposure to the same dialysate sodium concentrations, when patients transitioned from thrice weekly conventional hemodialysis to dialysis modalities differing in duration and frequency (SHD, ICHD, INHD, and FNHD). The “sodium setpoint” was defined as the average predialysis plasma sodium concentration over the time period specified for each of the three end-points. The three end-points were δPP100+Na+, δPP100–150Na+, and slope of predialysis Na+ in 100 days after transition from conventional thrice weekly to home hemodialysis (M100) (Figure 1). They are defined as follows:
- δPP100+Na+ is the difference between Post100+Na+ and Pre-Na+.
- Post100+Na+ is the average predialysis plasma sodium concentration, after 100 days of home hemodialysis, for the life of the patient while still on the same dialysis modality.
- Pre-Na+ is the average of all predialysis plasma sodium values in 50 days before transition to home hemodialysis and while on ICHD.
Each patient’s period of time on home hemodialysis differed after the first 100 days. Thus, δPP100–150Na+ was also calculated as the difference between Post100-150Na+ and Pre-Na+. The Post100-150Na+ is the average predialysis plasma sodium concentration, between 100 and 150 days post-transition to home hemodialysis.
Post100+Na+ and Pre-Na+ were compared, as were Post100-150Na+ and Pre-Na+ in each dialysis modality group, for all patients, and separately for patients with Pre-Na+ values: 1) greater than or equal to, or 2) less than the dialysate sodium concentration of 140 mmol/L. A statistically significant change between Pre and Post-Na+ values implied a change in SP.
A line of best fit was then calculated from the predialysis plasma sodium values versus time plot for each patient over the first 100 days after transitioning to home hemodialysis. The slope of these lines of best fit was measured with its confidence intervals (CIs). The mean M100 values found in different dialysis modality groups were compared overall and again by Pre-Na+/dialysate Na+ relationship. A M100 with 95% CIs that did not cross zero was evidence for a change in SP.
We chose the time period of 100 days because we wanted a minimum of three plasma sodium measurements for each patient to calculate slope of predialysis plasma sodium concentration. Because each home hemodialysis patient undergoes monthly blood work, most patients have a minimum of three predialysis sodium concentrations within 100 days.
Regression models were used to identify an association between the primary outcome δPP100+Na+ and a series of covariates. Specifically, univariate regression analyses were performed using δPP100+Na+, δPP100–150Na+, and M100 as separately evaluated dependent variables. Independent variables evaluated included dialysis frequency and duration, dialysate to (Pre-Na+) gradient (GNa+0), and (GNa+0) times dialysis duration. We evaluated (GNa+0) times dialysis duration, as an independent variable, because we have previously shown that this covariate is predictive for IDWG in a similar patient population.
Multivariate regression was used in an attempt to determine how δPP100+Na+ was associated with dialysis frequency, duration, the dialysate to Pre-Na+ gradient, and the dialysate to Pre-Na+ gradient times dialysis duration. Here, all patients were used regardless of dialysis modality.
A total of 87 patients made up the database, with 31, 13, 30, and 13 from SHD, ICHD, FNHD, and INHD, respectively. There were 29, 13, 28, and 12 patients with sufficient data for δPP100+Na+ and δPP100–150Na+, and 31, 10, 26, and 11 patients with sufficient data for M100 from SHD, ICHD, FNHD, and INHD, respectively. A total of 29 patients had pretransition predialysis SP ≥140 mmol/L, with 12, 3, 12, and 2 from SHD, ICHD, FNHD, and INHD, respectively.
There were no statistically significant differences between dialysis modalities for age, diabetes status, sex, or vintage of renal replacement before initiation of home hemodialysis (Table 1). However, FNHD patients were heavier than ICHD patients (87.5 vs. 68.5 kg, p = 0.008). Residual renal function was higher in ICHD patients than SHD patients (1.94 vs. 0.27 ml/minute/1.73 m2, p < 0.001). While on the assigned dialysis modality, serum albumin did not differ between groups. Predialysis phosphate concentration was lower in FNHD than SHD patients (1.56 vs. 1.76 mmol/L, p = 0.044). Dialysis duration was shorter in SHD patients (142.3 minutes) than ICHD (202.3 minutes, p < 0.001), FNHD (408.6 minutes, p < 0.001), and INHD patients (372.5 minutes, p < 0.001) and shorter in ICHD patients than FNHD (p < 0.001) and INHD patients (p < 0.001). Dialysis frequency was greater in SHD patients (5.7 per week) than ICHD (3.1 per week, p < 0.001), FNHD (5.2 per week, p < 0.001), and INHD patients (3.1 per week, p < 0.001) and greater in FNHD than ICHD (p < 0.001) or INHD patients (p < 0.001). Weekly dialysis duration was lower in ICHD than SHD (633.7 vs. 803.4 minutes, p = 0.001), lower in SHD than INHD (803.4 vs. 1,148.5 minutes, p < 0.001), and lower in INHD than FNHD (1,148.5 vs. 2,128.1 minutes, p < 0.001). Weekly ultrafiltration volume was lower in ICHD and INHD than SHD (6.5 and 6.6 vs. 10.4 L, p < 0.001 and p = 0.004) and lower in ICHD and INHD than FNHD (6.5 and 6.6 vs. 12.0 L, p < 0.001 for both).
Sodium setpoint decreased in FNHD patients when all predialysis sodium concentrations from 100 days post-transition onward were considered (Pre-Na+ > Post100+Na+; 138.5–136.7 mM, p = 0.015; Figure 2).
In both SHD and FNHD patients whose pretransition predialysis sodium (Pre-Na+) was greater than or equal to the dialysate sodium of 140 mM, SP decreased when post-transition predialysis sodium concentrations from 100 days onward were considered (Pre-Na+ > Post100+Na+; SHD, 140.2–138.7 mM, p = 0.019; FNHD, 140.5–137.1 mM, p = 0.001; Figure 3). When predialysis plasma sodium concentrations were restricted to post-transition days 100–150, the SP still decreased in both SHD and FNHD patients (SHD, 140.2–138.6 mM, p = 0.030; FNHD 140.5–138.0 mM, p = 0.008; Figure 3).
There was no difference in any dialysis modality group, between Pre-Na+ and Post100+Na+, or between Pre-Na+ and Post100-150Na+, if the pretransition predialysis sodium (Pre-Na+) was less than the dialysate sodium concentration of 140 mM (Figure 4).
The slope of predialysis plasma sodium in the first 100 days post-transition (M100) was less than zero in all SHD (95% CI, −0.0055 to −0.0318 mM/day) and FNHD (95% CI, −0.0010 to −0.0394 mM/day) patients and in SHD (95% CI, −0.0081 to −0.0351 mM/day) and FNHD (95% CI, −0.0209 to −0.0695 mM/day) patients whose pretransition predialysis sodium (Pre-Na+) was ≥140 mM (Figure 5).
Univariate regression analysis was performed to predict M100 using 73 data sets from 29 SHD, 9 ICHD, 24 FNHD, and 11 INHD patients. Univariate correlation coefficients and p values are shown in Table 2. The strongest predictor of M100 was the dialysate to predialysis plasma sodium gradient (GNa+0) (R2 = 12.65%) although no independent factor reached statistical significance (Table 1).
Univariate regression analysis was performed to predict δPP100+Na+ and δPP100–150Na+ using 82 data sets from 29 SHD, 13 ICHD, 28 FNHD, and 12 INHD patients. Univariate correlation coefficients and p values are shown in Table 2. The covariate of (GNa+0) (dialysis time) had a correlation of 31.8% and 42.0% for δPP100+Na+ and δPP100–150Na+, respectively. However, this was entirely due to the GNa+0 component; elimination of dialysis duration from the covariate improved the correlation coefficient and p value in both δPP100+Na+ (R2 = 31.8–32.8%, p = 0.540–0.030) and δPP100–150Na+ (R2 = 42.0–42.0%, p = 0.859–0.002). Dialysis frequency (R2 = 6.19%, p = 0.060) and dialysis duration (R2 = 2.15%, p = 0.085) trended toward a relationship with δPP100+Na+. A multivariate model was created to investigate the association of δPP100+Na+ with dialysis frequency and GNa+0.
δPP100+Na+ = 0.4765 GNa+0 – 0.3506 (dialysis frequency per week) – 0.2807
R2 = 35.44% (adjusted R2 = 33.8%)
F-statistic = 21.68 (on 2 and 79 degrees of freedom, p<0.001)
δPP100+Na+ = (Post100-Na+) − (Pre-Na+),
GNa+0 = (Dialysis Na+) − (Pre-Na+), adjusted p value < 0.001, R2 = 32.8% in univariate model,
dialysis frequency = dialysis sessions per week;
adjusted p value = 0.077, R2 = 6.2% in univariate model.
The SP is considered to be stable in hemodialysis patients. The results of this study suggest that this is true at least with ICHD. According to model 1, this assumption is reasonable in ICHD patients; a predialysis sodium between 133.0 and 141.0 mmol/L would be associated with δPP100Na+ between −2 and 2 mmol/L. This difference could be attributed to changes in total body water or to laboratory measurement variability. In the initial description of the stability of the predialysis SP by Keen and Gotch,3 89% of patients had an average predialysis plasma sodium from 133.0 to 141.0 mM.
However, there are scenarios in which a SP change may occur on the basis of model 1. Patients whose dialysate sodium is personalized to be equal or less than predialysis sodium may decrease their SP. For example, in a patient dialyzed five times weekly, with a predialysis plasma sodium of 135 mmol/L, whose dialysate sodium is personalized to 132 mmol/L, in an attempt to “desalt,” the associated δPP100+Na+ would be −3.5 mmol/L (model 1), which would bring the predialysis plasma SP down to 131.5 mmol/L, a level associated with increased mortality.16,17
Furthermore, patients dialyzed in units using a “standard dialysate sodium concentration” may increase their predialysis plasma SP. For example, a patient dialyzed three times weekly, with a setpoint of 130 mmol/L, whose dialysate sodium is 140 mmol/L, would have an associated δPP100+Na+ of 3.4 mmol/L (model 1), setting the new predialysis plasma SP to 133.4 mmol/L. These patients would not have been observed in the Keen and Gotch’s description because none of their patients had SPs <131 mmol/L. It is unknown whether the increased IDWG observed in patients with a large dialysate to predialysis plasma sodium gradient is offset by any improvement in mortality by increasing the SP. If so, this may, in part, explain the unexpected results of discovery by Hecking et al. 15 that patients whose predialysis sodium was <137 mmol/L had improved mortality when dialyzing against a higher dialysate sodium concentration, and reduced hospitalization and mortality with higher dialysate sodium concentrations, in units that did not individualize dialysate sodium concentrations.19 Dialysate sodium prescriptions may have changed some of the predialysis sodium concentrations from a low level to a level associated with improved mortality. Prospective trials should evaluate the effect of intentionally increasing predialysis plasma SPs, on cardiovascular and all-cause mortality.
Determining the pathophysiology of a change in plasma SP is not the objective of this study and will need to be established prospectively. Stability in blood glucose, lipid, and paraprotein concentrations needs to be initially assumed. Later, if a patient has a predialysis plasma sodium concentration greater than the dialysate sodium concentration, one could hypothesize that the postdialysis plasma sodium concentration would decrease toward the dialysate sodium concentration because sodium loss would occur relative to the isosmotic ultrafiltration, leaving the plasma with relative sodium to water loss. It is possible that equilibration back to SP homeostasis requires an interdialytic interval longer than patients on quotidian, but not intermittent hemodialysis modalities. This hypothesis would need to be evaluated prospectively. However, this would explain why adding (dialysis time) to (δGNa+0) did not improve (δGNa+0) prediction of δPP100Na+ (Table 2), because dialysis frequency is a much greater determinant of interdialytic interval duration. For example, doubling a patient’s dialysis duration from 4 to 8 hours (at dialysis frequency three times a week) only marginally decreases interdialytic time interval from 39.0 to 36.0 hours, whereas doubling a patient’s dialysis frequency from 3 to 6 weekly sessions (at dialysis duration 4 hours a session) significantly decreases interdialytic time interval from 39.0 to 20.6 hours. Indeed, a patient with a predialysis plasma SP of 140 mmol/L, dialyzing 7 days weekly with a dialysate sodium of 140 mmol/L, would decrease their predialysis plasma SP to 137 mmol/L; this is a surprising and unexpected finding, the etiology of which will need to be elucidated with prospective investigations.
Finally, quotidian dialysis therapies appear from these results to be associated with an increased chance of decreasing the SP when the initial predialysis plasma SP is equal to or greater than the dialysate sodium (Figure 3). For example, a patient on 6 nights a week hemodialysis, with an initial predialysis plasma sodium of 143 mmol/L, and a dialysate sodium concentration of 140 mmol/L will have a δPP100+Na+ of −3.8, bringing the predialysis plasma SP to 140.2 mmol/L. Targeting the dialysate sodium concentration to below the predialysis SP could lead to repeated drops in the predialysis sodium with every change in dialysate concentration. This may be undesirable from an outcome perspective.
Statistically significant changes in predialysis plasma SP were observed with 150 days after transition from thrice weekly conventional to home hemodialysis (Figure 3). It is thus unlikely that any decreases in predialysis SP related to patients developing comorbidities associated with lower plasma sodium concentrations such as heart or liver disease. Indeed, the strongest associations with change in plasma SP were iatrogenic, specifically the choices of dialysis frequency and the dialysate sodium to predialysis plasma sodium gradient (Table 2). Furthermore, only outpatient blood tests were considered, so acute illness or comorbid illness is unlikely to be a confounding factor.
In light of numerous studies that suggest personalizing dialysate sodium concentrations can decrease IDWG,6–13 these data give reason for caution. If dialysate sodium is intentionally decreased to the predialysis plasma sodium concentration, the IDWG may fall, but any benefit in morbidity and mortality may be offset by a decrease in SP.
There are limitations to this study. First, all data in this study are retrospective and measurements did not occur at exact time intervals in all patients. Thus, it remains unclear whether any change in SP is a continuous process or whether any change is upon initiation of dialysis and complete after a short interval of time. However, the predialysis plasma SP change was completed within 150 days in our study, suggesting that patients reach a new “steady state” in which the effects of dialysis frequency and dialysate to predialysis plasma sodium gradient offset each other. Second, data points used were aggregates of variable numbers of dialysis and laboratory values occurring between variable time periods. This may explain why model 1 only provides 35% explanation for the change in PP100Na+. Third, there were baseline differences between dialysis modality groups, such as residual renal function and patient weight, which may be confounders. The study also has strengths, in that numerous predialysis plasma sodium values are available and that modalities differing in frequency and duration were used with this home hemodialysis population. While the sample size of patients was small (n = 87), the findings were statistically significant and likely of clinical importance.
Further studies are indicated in quotidian hemodialysis patients that will vary prospectively the dialysate sodium to establish the effect of dialysate sodium and SP on cardiovascular morbidity and all-cause mortality.
In hemodialysis patients, the predialysis plasma sodium setpoint is dynamic and correlated to the dialysate sodium concentration and dialysis frequency. Nephrologists should consider how the selected dialysate sodium concentration affects the dialysate to predialysis plasma sodium concentration gradient and should also continue to monitor predialysis plasma sodium concentrations. Prospective trials are needed to establish when the benefits of a decrease in IDWG are offset by a decrease in SP and how dialysate sodium concentration should be targeted to minimize cardiovascular and all-cause mortality.
Dr. Stephanie Dixon was helpful in her guidance on statistical methods. This research was made possible by a grant from the Program Of Experimental Medicine (POEM) at Western University, Canada.
1. Zerbe RL, Miller JZ, Robertson GL. The reproducibility and heritability of individual differences in osmoregulatory function in normal human subjects. J Lab Clin Med. 1991;117:51–59
2. Argent NB, Burrell LM, Goodship TH, Wilkinson R, Baylis PH. Osmoregulation of thirst and vasopressin release in severe chronic renal failure. Kidney Int. 1991;39:295–300
3. Keen ML, Gotch FA. The association of the sodium “setpoint” to interdialytic weight gain
and blood pressure in hemodialysis
patients. Int J Artif Organs. 2007;30:971–979
4. Basile C, Libutti P, Lisi P, et al. Sodium setpoint
and gradient in bicarbonate hemodialysis
. J Nephrol. 2012
5. Peixoto AJ, Gowda N, Parikh CR, Santos SF. Long-term stability of serum sodium in hemodialysis
patients. Blood Purif. 2010;29:264–267
6. Krautzig S, Janssen U, Koch KM, Granolleras C, Shaldon S. Dietary salt restriction and reduction of dialysate sodium to control hypertension in maintenance haemodialysis patients. Nephrol Dial Transplant. 1998;13:552–553
7. Farmer CKT, Donohoe P, Dallyn PE, Cox J, Koingswood JC, Goldsmith DJ. Low sodium hemodialysis
without fluid removal improves blood pressure control in hemodialysis
patients. Nephrology. 2000;5:237–241
8. Ferraboli R, Manuel C, Abensur H, Elias R, Luders C. Reduction of sodium dialysate for hypertensive HD patients: Analysis of beneficial and adverse effects. J Am Soc Nephrol. 2002;13:211a
9. de Paula FM, Peixoto AJ, Pinto LV, Dorigo D, Patricio PJ, Santos SF. Clinical consequences of an individualized dialysate sodium prescription in hemodialysis
patients. Kidney Int. 2004;66:1232–1238
10. Lambie SH, Taal MW, Fluck RJ, McIntyre CW. Online conductivity monitoring: Validation and usefulness in a clinical trial of reduced dialysate conductivity. ASAIO J. 2005;51:70–76
11. Thein H, Haloob I, Marshall MR. Associations of a facility level decrease in dialysate sodium concentration with blood pressure and interdialytic weight gain
. Nephrol Dial Transplant. 2007;22:2630–2639
12. Sayarlioglu H, Erkoc R, Tuncer M, et al. Effects of low sodium dialysate in chronic hemodialysis
patients: An echocardiographic study. Ren Fail. 2007;29:143–146
13. Hecking M, Kainz A, Hörl WH, Herkner H, Sunder-Plassmann G. Sodium setpoint
and sodium gradient: Influence on plasma sodium change and weight gain. Am J Nephrol. 2011;33:39–48
14. Foley RN, Herzog CA, Collins AJUnited States Renal Data System. . Blood pressure and long-term mortality in United States hemodialysis
patients: USRDS Waves 3 and 4 Study. Kidney Int. 2002;62:1784–1790
15. Hecking M, Karaboyas A, Saran R, et al. Predialysis serum sodium level, dialysate sodium, and mortality in maintenance hemodialysis
patients: The Dialysis Outcomes and Practice Patterns Study (DOPPS). Am J Kidney Dis. 2012;59:238–248
16. Mc Causland FR, Brunelli SM, Waikar SS. Dialysate sodium, serum sodium and mortality in maintenance hemodialysis
. Nephrol Dial Transplant. 2012;27:1613–1618
17. Thijssen S, Usvyat L, Kotanko P. Prediction of mortality in the first two years of hemodialysis
: Results from a validation study. Blood Purif. 2012;33:165–170
18. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461–470
19. Hecking M, Karaboyas A, Saran R, et al. Dialysate sodium concentration and the association with interdialytic weight gain
, hospitalization, and mortality. Clin J Am Soc Nephrol. 2012;7:92–100