Introduction: The optimal dialysate bicarbonate concentration (DBIC) for hemodialysis (HD) remains controversial. Herein, we analyzed the effect of dialysate bicarbonate levels on mortality in HD patients. Methods: Patients undergoing maintenance HD were recruited from the HD unit of the Daping Hospital. Patients were categorized into quartiles according to their DBIC level (quartile 1: <31.25 mmol/L, n = 77; quartile 2: 31.25–32.31 mmol/L, n = 76; quartile 3: 32.31–33.6 mmol/L; n = 81; quartile 4: ≥33.6 mmol/L, n = 79). Demographic and clinical data were collected. Survival curves were estimated using the Kaplan-Meier method. A Cox proportional hazards regression model was used to estimate the association between DBIC and all-cause mortality. Results: We included 313 patients undergoing maintenance HD with a mean DBIC of 32.16 ± 1.59 mmol/L (range, 27.20–34.72 mmol/L). The patients in quartile 4 were more likely to have higher pre- and post-HD serum bicarbonate concentrations than those in other quartiles. The mortality rate was lowest in quartile 2 (10.53%). The survival time was significantly lower in the quartile 4 group than in the other quartiles (p = 0. 008, log-rank test). After full adjustment, the hazard ratio (per 3 mmol/L higher DBIC) for all-cause mortality was 4.29 (95% confidence interval, 2.11–8.47) in all patients, whereas no significant association was observed between DBIC and initial hospitalization. Conclusions: Our data indicate that DBIC is positively associated with all-cause mortality. A DBIC concentration of 31–32 mmol/L may benefit patient outcomes. This study provides an evidence-based medical basis for optimal dialysis prescription in the future.

Hemodialysis (HD) is a common renal replacement therapy for patients with end-stage renal disease (ESRD). According to the 2021 annual report on blood purification, the corresponding number of patients undergoing HD in China was approximately 749,000 [1]. Metabolic acidosis is one of the main causes of death in patients undergoing maintenance HD (MHD) due to the progressive loss of renal function and intermittency of HD [2]. The acid-base balance is a major target of HD therapy. In this sense, most patients undergoing HD are currently treated with a bicarbonate-buffered dialysate. Nevertheless, a better balance of the acid-base state is needed to avoid post-HD alkalosis and prevent acidosis at the beginning of the next HD session.

One of the most important factors affecting serum bicarbonate levels before and after dialysis is dialysate bicarbonate concentration (DBIC). The blood purification process pertaining to the standard operating procedure [3] in China suggests that the pre-HD serum bicarbonate concentration (pre-HD SBIC) to be controlled at 22–26 mmol/L; however, no specific recommendations regarding DBIC were indicated. Contrastingly, the 2,000 National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKF-KDOQI) guidelines recommend a high DBIC level of 38 mmol/L to achieve pre-dialysis serum bicarbonate levels of 22 mmol/L [4]. Nevertheless, the Dialysis Outcomes and Practice Patterns Study (DOPPS) found that a high DBIC may contribute to adverse outcomes, most likely through the development of post-HD metabolic alkalosis, but not in Asian countries [5]. Consequently, the optimal DBIC for HD remains unclear. Therefore, we designed a retrospective observational cohort study to evaluate the association between DBIC, hospitalization, and all-cause mortality among patients undergoing MHD.

Study Populations

We retrospectively analyzed the records of adult patients who received MHD at the HD center of Daping Hospital between January 1, 2015, and December 31, 2020. The inclusion criteria were as follows: (1) age equal to or greater than 18 years; (2) HD at a frequency of three times per week, 4 h per session; and (3) dialysis for more than 3 months. The exclusion criteria were as follows: (1) combination therapy with peritoneal dialysis and (2) incomplete bicarbonate concentration data (i.e., participants who missed one or more of the follow-up measurements, including those of DBIC, pre-HD SBIC, or post-HD SBIC).

A total of 987 patients undergoing HD were enrolled in this study. After screening, 405 patients were excluded for dialysis vintage at less than 3 months, 165 were excluded for not meeting the dialysis frequency, 54 were excluded for incomplete bicarbonate concentration data, and 50 were excluded for combination therapy with peritoneal dialysis. Thus, 313 patients undergoing HD were included in the analysis. Patients were enrolled in the Chinese National Renal Data System. This study was approved by the Medical Ethics Committee of the Daping Hospital (approval number: YYLS2022-110), and informed consent was obtained from all patients.

Dialysate Bicarbonate Concentration

According to the blood purification of standard operating procedure in China, the target range of pre-HD SBIC was 22–26 mmol/L; pre-HD SBIC was performed monthly, and post-HD SBIC was performed every 6 months. DBIC was promptly adjusted according to the pre-HD SBIC of patients; the specific adjustment protocol is shown in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000531267). Initial HD treatment data within the first week of each month were collected from the enrolled patients. DBIC data were collected from the HD treatment data, and the average value was considered as the final DBIC data for each patient. Patients were grouped according to DBIC quartiles and divided into four groups (quartile 1: <31 mmol/L, n = 77; quartile 2: 31.25–32.31 mmol/L, n = 76; quartile 3: 32.31–33.6 mmol/L, n = 81; quartile 4: ≥33.6 mmol/L, n = 79) according to the DBIC quartile.

Data Collection

Demographic and clinical data were obtained from the medical records of patients at enrolment. Baseline demographic data included age, sex, height, weight, body mass index, causes of ESRD, and comorbidities. The first HD treatment data within the first week of each month were collected from the enrolled patients and included pre-HD systolic blood pressure, pre-HD diastolic blood pressure, blood flow rate, dialysate flow rate and dialysate calcium, dialysate sodium, and dialysate potassium levels. The average value was taken as the final value of each patient. The types of vascular access and dialyzer membranes of patients were recorded based on the last HD treatment data. Blood samples were collected before and after midweek dialysis, and pre-HD and post-HD SBIC were examined before and after the same dialysis. Clinical data, including serum hemoglobin, albumin, calcium, phosphorus, intact parathyroid hormone, and serum potassium levels and single-pool Kt/V (spKt/V), were obtained every 6 months until the end of follow-up. spKt/V was calculated using the Daugirdas formula [6].

Endpoint

The primary endpoint was the all-cause mortality rate, and the secondary endpoint was the first hospitalization for any cause during the follow-up period. All patients were followed-up until death or end of the study; those lost to follow-up, kidney transplantation, or transfer to peritoneal dialysis were defined as censors.

Statistics Analysis

Continuous variables were presented as mean ± standard deviation or median (25th, 75th) appropriately, while categorical variables were expressed as percentages. Differences in the mean or median between groups were evaluated using a one-way analysis of variance or nonparametric tests. Categorical data were compared using the χ2 test. Survival curves were estimated using the Kaplan-Meier method and compared using the log-rank test among the four groups. A Cox proportional hazards regression model was used to estimate the association between DBIC, all-cause mortality, and first hospitalization. Factors that were verified using the univariate Cox regression analysis were entered into the multivariate Cox regression analysis. Correlations were calculated using Spearman analysis. A p value <0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 22 (SPSS Inc., Chicago, IL, USA).

Patient Characteristics

A total of 313 patients (191 female, 122 male) with a mean age of 61.01 ± 14.48 years were included. Patients underwent HD for a median of 67.98 months. The etiology of ESRD was as follows: glomerulonephritis (55.27%), diabetes (23.64%), and others (21.09%). The mean DBIC was 32.16 ± 1.59 mmol/L (range, 27.20–34.72 mmol/L). For patients in quartiles 1, 2, 3, and 4, the mean DBIC was 29.91 ± 0.95, 31.81 ± 0.28, 32.83 ± 0.35, and 33.98 ± 0.17 mmol/L, respectively.

Patient demographics and baseline laboratory variables across the four DBIC categories are shown in Table 1. Patients in quartile 4 were more likely to have higher pre- and post-HD serum bicarbonate levels than those in the other quartiles. Comparison showed that the differences in dialysis vintage, blood flow rate, dialysate sodium levels, dialysate potassium levels, intact parathyroid hormone, and triglyceride levels among groups were significant. However, there were no significant differences in age, sex, body mass index, causes of ESRD, comorbid conditions, serum albumin, hemoglobin, serum calcium, serum phosphorus, serum potassium, post-HD blood pressure, arteriovenous fistula use, spKt/V, dialysate flow rate, dialysate calcium levels, and dialysate potassium levels among groups.

Table 1.

Baseline characteristics in HD patients based on DBIC

CharacteristicAll patientsDBIC, mmol/Lp value
quartile 1quartile 2quartile 3quartile 4
(<31.25)(31.25–32.31)(32.31–33.6)(≥33.6)
Patients, n 313 77 76 81 79  
Age (years) 61.01±14.48 62.57±13.94 57.17±16.18 62.07±12.96 62.10±14.7 0.067 
Male (sex), n (%) 191 (61.02) 44 (57.14) 46 (60.52) 51 (62.96) 50 (63.29) 0.399 
Dialysis vintage, months 67.98±37.85 45.56±24.65 63.84±39.12 86.07±46.03 55.28±32.63 <0.001 
Body mass index, kg/m2 21.94±3.18 21.37±3.11 21.56±2.87 22.52±2.98 22.27±3.61 0.066 
Causes of ESRD, n (%) 
 Glomerulonephritis 173 (55.27) 34 (44.16) 52 (68.42) 49 (60.49) 38 (48.10) 0.631 
 Diabetes mellitus 74 (23.64) 22 (28.57) 15 (19.74) 16 (19.75) 21 (26.58) 
 Others/unknown 66 (21.09) 21 (27.27) 9 (11.84) 16 (19.75) 20 (25.32) 
Comorbid conditions, n (%) 
 Cardiovascular disease 43 (13.74) 11 (14.29) 11 (14.47) 11 (13.58) 10 (12.66) 0.106 
 Hypertension 65 (20.77) 18 (23.38) 14 (18.42) 14 (17.28) 19 (24.05) 
 Diabetes 86 (27.48) 27 (35.06) 17 (22.37) 19 (23.46) 23 (29.11) 
 Cancer (non-skin) 12 (3.83) 2 (2.60) 3 (3.95) 4 (4.94) 3 (3.80) 
 Fracture 16 (5.11%) 1 (1.30%) 6 (7.89%) 3 (3.70%) 6 (7.59%) 
Hemodialysis parameters 
 Pre-HD SBP, mm Hg 145.40±13.82 147.83±16.79 145.65±13.09 143.50±13.22 144.74±11.66 0.249 
 Pre-HD DBP, mm Hg 77.71±9.84 75.01±10.22 78.68±8.39 78.55±11.57 78.53±8.43 0.053 
 Arteriovenous fistula use, n (%) 300 (95.85) 72 (93.50) 73 (96.05) 79 (97.53) 76 (96.20) 0.562 
 Single-pool Kt/V 1.46±0.25 1.46±0.24 1.46±0.22 1.43±0.24 1.46±0.29 0.825 
 The DBIC for the first treatment, mmol/L 32.41±2.49 29.84±2.51 32.17±2.43 33.63±1.38 33.89±0.86 <0.001* 
 Pre-HD SBIC, mmol/L 21.07±2.18 20.45±1.78 21.05±1.95 21.03±2.12 21.73±2.61 0.003* 
 Pre-HD SBIC range, mmol/L 10.4–27.90 14.9–23.73 10.4–25 16.84–27.90 15.6–27.4  
 Post-HD SBIC, mmol/L 23.89±1.93 23.21±1.55 23.76±1.48 23.62±1.93 24.94±2.21 <0.001* 
 Blood flow rate, mL/min 216.12±15.92 211.13±1.74 217.05±13.83 219.35±14.76 216.79±16.74 0.01 
 Dialysate flow rate, mL/min 498.54±11.01 498.55±10.41 499.31±5.98 499.90±0.86 496.38±19.37 0.37 
 High-flux dialyzer, n (%) 278 (88.8) 65 (84.42) 72 (94.74) 73 (90.12) 68 (86.08) 0.178 
 Dialysate calcium, mmol/L 1.46±0.07 1.47±0.05 1.46±0.06 1.46±0.07 1.46±0.08 0.087 
 Dialysate sodium, mmol/L 139.38±1.53 139.95±1.12 139.18±1.64 139.23±1.42 139.18±1.74 0.008 
 Dialysate potassium, mmol/L 2.01 (2.00, 2.28) 2.06 (2.00, 2.42) 2.03 (2.00, 2.21) 2.00 (2.00, 2.67) 2.00 (2.00, 2.29) 0.019 
Laboratory variables 
 Serum albumin, g/L 36.86±3.54 36.51±3.49 37.25±3.70 37.48±2.45 36.19±4.05 0.074 
 Hemoglobin, g/L 107.30±11.32 106.75±12.17 106.2±9.68 109.86±11.45 106.44±11.57 0.124 
 Serum calcium, mmol/L 2.20±0.18 2.18±0.17 2.22±0.19 2.21±0.19 2.20±0.16 0.56 
 Serum phosphorus, mmol/L 1.70±0.41 1.67±0.39 1.74±0.44 1.72±0.35 1.68±0.45 0.666 
 Intact parathyroid hormone, pg/mL 363.53±217.02 299.56±142.09 378.09±240.89 387.76±203.26 387.03±255.20 0.041 
 Serum potassium, mmol/L 4.74±0.51 4.73±0.50 4.80±0.45 4.68±0.44 4.74±0.62 0.454 
 White blood cell, 109/L 6.02 (5.12, 7.22) 6.05 (4.99, 6.92) 5.86 (5.07, 7.20) 6.08 (5.32, 7.44) 6.02 (5.13, 7.24) 0.951 
 Triglyceride, mmol/L 1.83±1.03 1.59±0.88 1.59±0.88 1.73±0.93 1.92±0.99 0.023* 
 Cholesterol, mmol/L 3.91±0.94 3.98±1.27 3.97±1.27 3.86±0.85 3.91±0.80 0.875 
CharacteristicAll patientsDBIC, mmol/Lp value
quartile 1quartile 2quartile 3quartile 4
(<31.25)(31.25–32.31)(32.31–33.6)(≥33.6)
Patients, n 313 77 76 81 79  
Age (years) 61.01±14.48 62.57±13.94 57.17±16.18 62.07±12.96 62.10±14.7 0.067 
Male (sex), n (%) 191 (61.02) 44 (57.14) 46 (60.52) 51 (62.96) 50 (63.29) 0.399 
Dialysis vintage, months 67.98±37.85 45.56±24.65 63.84±39.12 86.07±46.03 55.28±32.63 <0.001 
Body mass index, kg/m2 21.94±3.18 21.37±3.11 21.56±2.87 22.52±2.98 22.27±3.61 0.066 
Causes of ESRD, n (%) 
 Glomerulonephritis 173 (55.27) 34 (44.16) 52 (68.42) 49 (60.49) 38 (48.10) 0.631 
 Diabetes mellitus 74 (23.64) 22 (28.57) 15 (19.74) 16 (19.75) 21 (26.58) 
 Others/unknown 66 (21.09) 21 (27.27) 9 (11.84) 16 (19.75) 20 (25.32) 
Comorbid conditions, n (%) 
 Cardiovascular disease 43 (13.74) 11 (14.29) 11 (14.47) 11 (13.58) 10 (12.66) 0.106 
 Hypertension 65 (20.77) 18 (23.38) 14 (18.42) 14 (17.28) 19 (24.05) 
 Diabetes 86 (27.48) 27 (35.06) 17 (22.37) 19 (23.46) 23 (29.11) 
 Cancer (non-skin) 12 (3.83) 2 (2.60) 3 (3.95) 4 (4.94) 3 (3.80) 
 Fracture 16 (5.11%) 1 (1.30%) 6 (7.89%) 3 (3.70%) 6 (7.59%) 
Hemodialysis parameters 
 Pre-HD SBP, mm Hg 145.40±13.82 147.83±16.79 145.65±13.09 143.50±13.22 144.74±11.66 0.249 
 Pre-HD DBP, mm Hg 77.71±9.84 75.01±10.22 78.68±8.39 78.55±11.57 78.53±8.43 0.053 
 Arteriovenous fistula use, n (%) 300 (95.85) 72 (93.50) 73 (96.05) 79 (97.53) 76 (96.20) 0.562 
 Single-pool Kt/V 1.46±0.25 1.46±0.24 1.46±0.22 1.43±0.24 1.46±0.29 0.825 
 The DBIC for the first treatment, mmol/L 32.41±2.49 29.84±2.51 32.17±2.43 33.63±1.38 33.89±0.86 <0.001* 
 Pre-HD SBIC, mmol/L 21.07±2.18 20.45±1.78 21.05±1.95 21.03±2.12 21.73±2.61 0.003* 
 Pre-HD SBIC range, mmol/L 10.4–27.90 14.9–23.73 10.4–25 16.84–27.90 15.6–27.4  
 Post-HD SBIC, mmol/L 23.89±1.93 23.21±1.55 23.76±1.48 23.62±1.93 24.94±2.21 <0.001* 
 Blood flow rate, mL/min 216.12±15.92 211.13±1.74 217.05±13.83 219.35±14.76 216.79±16.74 0.01 
 Dialysate flow rate, mL/min 498.54±11.01 498.55±10.41 499.31±5.98 499.90±0.86 496.38±19.37 0.37 
 High-flux dialyzer, n (%) 278 (88.8) 65 (84.42) 72 (94.74) 73 (90.12) 68 (86.08) 0.178 
 Dialysate calcium, mmol/L 1.46±0.07 1.47±0.05 1.46±0.06 1.46±0.07 1.46±0.08 0.087 
 Dialysate sodium, mmol/L 139.38±1.53 139.95±1.12 139.18±1.64 139.23±1.42 139.18±1.74 0.008 
 Dialysate potassium, mmol/L 2.01 (2.00, 2.28) 2.06 (2.00, 2.42) 2.03 (2.00, 2.21) 2.00 (2.00, 2.67) 2.00 (2.00, 2.29) 0.019 
Laboratory variables 
 Serum albumin, g/L 36.86±3.54 36.51±3.49 37.25±3.70 37.48±2.45 36.19±4.05 0.074 
 Hemoglobin, g/L 107.30±11.32 106.75±12.17 106.2±9.68 109.86±11.45 106.44±11.57 0.124 
 Serum calcium, mmol/L 2.20±0.18 2.18±0.17 2.22±0.19 2.21±0.19 2.20±0.16 0.56 
 Serum phosphorus, mmol/L 1.70±0.41 1.67±0.39 1.74±0.44 1.72±0.35 1.68±0.45 0.666 
 Intact parathyroid hormone, pg/mL 363.53±217.02 299.56±142.09 378.09±240.89 387.76±203.26 387.03±255.20 0.041 
 Serum potassium, mmol/L 4.74±0.51 4.73±0.50 4.80±0.45 4.68±0.44 4.74±0.62 0.454 
 White blood cell, 109/L 6.02 (5.12, 7.22) 6.05 (4.99, 6.92) 5.86 (5.07, 7.20) 6.08 (5.32, 7.44) 6.02 (5.13, 7.24) 0.951 
 Triglyceride, mmol/L 1.83±1.03 1.59±0.88 1.59±0.88 1.73±0.93 1.92±0.99 0.023* 
 Cholesterol, mmol/L 3.91±0.94 3.98±1.27 3.97±1.27 3.86±0.85 3.91±0.80 0.875 

HD, hemodialysis; SBP, systolic blood pressure; DBP, diastolic blood pressure; SBIC, serum bicarbonate concentration; DBIC, dialysate bicarbonate concentration.

*versus Quartile 4, p < 0.05.

Relationship between Pre-HD SBIC, Post-HD SBIC, and All-Cause Mortality

Both SBIC before (r = 0.17, p = 0.002) and after (r = 0.25, p < 0.001) the HD session were positively correlated with DBIC levels. However, a negative relationship was observed between DBIC and all-cause mortality (r = −0.15, p = 0.009). We found a weak negative correlation between DBIC and post-HD potassium (r = −0.17, p = 0.002), and there was no significant correlation between DBIC and calcium. Correlations are described in online supplementary Table 2. Patients were categorized into five groups based on the difference between post- and pre-HD serum bicarbonate levels. A U-shaped relationship existed between the difference and mortality. Moreover, when the difference between pre- and post-HD bicarbonate levels was ≥4 mmol/L, mortality rates decreased markedly; however, a comparison among groups revealed no significant differences (p = 0.31) (online suppl. Table 3).

Associations between DBIC and All-Cause Mortality

The median follow-up period was 65 (range, 8–71) months. A total of 56 patients (17.89%) died during the follow-up period, and the absolute mortality rate was 2.98 deaths per 100 person-years. Compared to that in other quartiles, the mortality rate was the lowest in quartile 2 (10.53%) and the highest in quartile 4 (39.29%), demonstrating a significant difference between groups (p = 0.03). The leading causes of death were cardiovascular disease (62.50%), cancer (12.50%), and infectious diseases (7.14%). There were no significant differences in the causes of death among the four groups (p = 0.11) (Table 2).

Table 2.

Causes of deaths in different DBIC group

DiseasesAll patientsDBIC, mmol/Lp value
quartile 1 (<31.25)quartile 2 (31.25–32.31)quartile 3 (32.31–33.6)quartile 4 (≥33.6)
Cardiovascular diseases 35 10 13 0.112 
Infectious diseases 
Cancer (non-skin) 
Others 
Total 56 11 15 22 0.031 
DiseasesAll patientsDBIC, mmol/Lp value
quartile 1 (<31.25)quartile 2 (31.25–32.31)quartile 3 (32.31–33.6)quartile 4 (≥33.6)
Cardiovascular diseases 35 10 13 0.112 
Infectious diseases 
Cancer (non-skin) 
Others 
Total 56 11 15 22 0.031 

DBIC, dialysate bicarbonate concentration.

Figure 1 shows the Kaplan-Meier plot of patient survival and mortality according to DBIC. Compared to that in patients in other quartiles, the survival time was significantly low in patients in quartile 4 (p = 0.008, log-rank test).

Fig. 1.

Kaplan-Meier survival curve for mortality.

Fig. 1.

Kaplan-Meier survival curve for mortality.

Close modal

Cox Regression Analysis of All-Cause Mortality-Related Clinical Features

In the unadjusted Cox regression model, relative to that in quartile 2, the hazard ratio (HR) for all-cause mortality in quartile 4 was 3.3 (95% confidence interval [CI], 1.47–7.42; p = 0. 004). All-cause mortality was positively correlated with age, dialysis vintage, dialysate potassium, pre-dialysis systolic blood pressure, dialyzer membrane, blood flow rate, hemoglobin levels, and serum albumin levels (Table 3). A higher DBIC was significantly associated with a higher all-cause mortality rate (HR per 3 mmol/L higher DBIC, 2.21; 95% CI, 1.35–3.61; p = 0.002). After full adjustment, the HR (per 3 mmol/L increase in DBIC) for all-cause mortality was 4.29 (95% CI, 2.11–8.47; p ≤ 0.001) for all patients (Fig. 2). Age was not a risk factor for patient mortality (HR per 3 mmol/L higher DBIC, 1.03; 95% CI, 0.99–1.06; p = 0.126). In the Cox multivariate analysis, the following comorbidities were not risk factors for patient mortality: diabetes mellitus (HR per 3 mmol/L higher DBIC, 1.25; 95% CI, 0.59–2.69; p = 0.561), coronary heart disease (HR per 3 mmol/L higher DBIC, 0.46; 95% CI, 0.17–1.24; p = 0.126), and hypertension (HR per 3 mmol/L higher DBIC, 0.52; 95% CI, 0.22–1.23; p = 0.137).

Table 3.

Hazard ratios for clinical factors associated with all-cause mortality

HR95% CIp value
Quartile 1 (<31.25) 1.380 0.555 3.431 0.488 
Quartile 2 (31.25–32.31) Reference 
Quartile 3 (32.31–33.6) 1.837 0.779 4.333 0.165 
Quartile 4 (≥33.6) 3.300 1.469 7.415 0.004 
Serum albumin (g/L) 0.861 0.814 0.911 <0.001 
Hemoglobin (g/L) 0.986 0.964 1.009 0.240 
Pre-HD DBP (mm Hg) 0.970 0.943 0.997 0.031 
Single-pool Kt/V 2.656 1.001 7.046 0.050 
Post-HD SBIC (mmol/L) 1.149 1.002 1.318 0.047 
Blood flow rate (mL/min) 0.977 0.962 0.992 0.003 
High-flux dialyzer (%) 0.303 0.168 0.549 <0.001 
Dialysate potassium (mmol/L) 2.458 1.173 5.151 0.017 
Dialysis vintage (months) 0.987 0.979 0.995 0.002 
Age (years) 1.027 1.008 1.047 0.006 
HR95% CIp value
Quartile 1 (<31.25) 1.380 0.555 3.431 0.488 
Quartile 2 (31.25–32.31) Reference 
Quartile 3 (32.31–33.6) 1.837 0.779 4.333 0.165 
Quartile 4 (≥33.6) 3.300 1.469 7.415 0.004 
Serum albumin (g/L) 0.861 0.814 0.911 <0.001 
Hemoglobin (g/L) 0.986 0.964 1.009 0.240 
Pre-HD DBP (mm Hg) 0.970 0.943 0.997 0.031 
Single-pool Kt/V 2.656 1.001 7.046 0.050 
Post-HD SBIC (mmol/L) 1.149 1.002 1.318 0.047 
Blood flow rate (mL/min) 0.977 0.962 0.992 0.003 
High-flux dialyzer (%) 0.303 0.168 0.549 <0.001 
Dialysate potassium (mmol/L) 2.458 1.173 5.151 0.017 
Dialysis vintage (months) 0.987 0.979 0.995 0.002 
Age (years) 1.027 1.008 1.047 0.006 

HD, hemodialysis; DBP, diastolic blood pressure; SBIC, serum bicarbonate concentration; HR, hazard ratio.

Fig. 2.

Multivariate Cox regression analysis. Model 1: multivariate model including age, sex, body mass index, dialysis vintage, diabetes mellitus, cardiovascular disease, and hypertension; model 2: multivariate model including model 1 variables, as well as pre-dialysis serum bicarbonate, post-dialysis serum bicarbonate, dialysate calcium, dialysate potassium, dialysate sodium, pre-dialysis systolic blood pressure, pre-dialysis diastolic blood pressure, pre-dialysis heart rate, dialyzer, hemodialysis access, blood flow rate, dialysate flow rate, and single-pool Kt/V; model 3: multivariate model including model 2 variables, as well as serum albumin, hemoglobin, serum Calcium, serum phosphorus, intact parathyroid hormone, pre-dialysis serum potassium, pre-dialysis alkaline phosphatase, white blood cell, triglyceride, and cholesterol levels.

Fig. 2.

Multivariate Cox regression analysis. Model 1: multivariate model including age, sex, body mass index, dialysis vintage, diabetes mellitus, cardiovascular disease, and hypertension; model 2: multivariate model including model 1 variables, as well as pre-dialysis serum bicarbonate, post-dialysis serum bicarbonate, dialysate calcium, dialysate potassium, dialysate sodium, pre-dialysis systolic blood pressure, pre-dialysis diastolic blood pressure, pre-dialysis heart rate, dialyzer, hemodialysis access, blood flow rate, dialysate flow rate, and single-pool Kt/V; model 3: multivariate model including model 2 variables, as well as serum albumin, hemoglobin, serum Calcium, serum phosphorus, intact parathyroid hormone, pre-dialysis serum potassium, pre-dialysis alkaline phosphatase, white blood cell, triglyceride, and cholesterol levels.

Close modal

Relationship of the First Hospitalization and DBIC

A total of 215 patients (68.69%) had their first hospitalization record. However, no significant association was observed between DBIC and first hospitalization after adjusting for full (HR per 3 mmol/L higher bicarbonate concentration, 1.04; 95% CI, 0.79–1.35; p = 0.790) (Fig. 2). The number of cardiovascular hospitalizations did not increase as DBIC increased.

In this retrospective cohort analysis of DBIC and mortality in patients undergoing MHD, the absolute mortality rate was 2.98 deaths per 100 person-years, and the mortality was lowest in quartile 2 (DBIC range, 31.25–32.31 mmol/L). After full adjustment, the HR (per 3 mmol/L higher DBIC) for all-cause mortality was 4.29 (95% CI, 2.11–8.47) in all patients. No significant association was found between DBIC and the first hospitalization. These findings are of great importance for the further optimization of dialysis prescriptions.

DBIC prescriptions vary significantly worldwide, usually fluctuating between 30 and 40 mmol/L [5]. Compared to those in Western countries, the DBICs in Asian countries are relatively low (≤30 mmol/L in a Japanese study [7] and 30.9 ± 1.6 mmol/L in a Korean study [8]). In our study, the DBIC of almost all patients was adjusted according to the monthly pre-HD bicarbonate levels. Moreover, concentration fluctuated from 26 to 36 mmol/L, with an average concentration (32.16 ± 1.59 mmol/L) closer to that in Asian countries, including Japan and South Korea, and lower than that in Western countries, especially the USA (37.0 ± 2.6 mmol/L). These findings indicate that the choice of DBIC in each region may be related to distinct ethnic patterns and dietary protein intake. A study [9] suggested that reducing DBIC from 35 mmol/L to 32 mmol/L significantly and safely reduced pre- and post-HD TCO2, avoiding acidosis overcorrection. Furthermore, sodium, pre-HD potassium, calcium, and phosphorous levels did not change significantly.

In 2013, Tentori et al. [5] analyzed the DOPPS database and found a positive association between DBIC and mortality (9% higher mortality per 4 mmol/L increase in DBIC), suggesting that a high DBIC may contribute to the development of post-HD metabolic alkalosis. Nevertheless, that study did not estimate the data on post-HD bicarbonate, and data from Asian countries were lacking. Several small-scale studies have suggested that a higher DBIC may lead to adverse outcomes. However, the underlying mechanisms have not been elucidated [10, 16]. Our results are similar to those reported in the study of Tentori et al. [5] after full adjustment, wherein a higher DBIC was significantly associated with higher all-cause mortality (HR per 3 mmol/L higher DBIC, 4.29; 95% CI, 2.11–8.47; p ≤ 0.001). In our study, the mean pre- and post-HD bicarbonate levels were 21.07 ± 2.18 and 23.89 ± 1.93 mmol/L, respectively. Importantly, even group 4, which had the highest mortality rate, did not meet the significant post-HD alkalosis criteria, with pre- and post-HD bicarbonate levels of 21.73 ± 2.61 and 24.94 ± 21 mmol/L, respectively. In the Cox multivariate analysis, the results for DBIC and all-cause mortality did not change when bicarbonate concentration was corrected after dialysis, which may be related to the low incidence of post-dialysis alkalosis caused by our individualized DBIC prescriptions. Additional research is required to elucidate the exact mechanism underlying this effect. Combined with the results of our center, we believe that a DBIC concentration of 31–32 mmol/L may be beneficial for patient outcomes, demonstrating the lowest mortality rate and offering a range where albumin and hemoglobin levels can be controlled within normal limits.

Furthermore, our results showed that, when the difference between pre- and post-HD bicarbonate levels was ≥4 mmol/L, the mortality rate was positively associated with that difference, even though no significant difference was found. We speculated that this might be because the included patients were from single centers, and the number of cases was relatively small.

As previously reported, during the first 2 h of dialysis, serum bicarbonate levels tend to rapidly increase, though they gradually stabilize thereafter [17]. At the end of the dialysis session, serum bicarbonate levels are usually 4–7 mmol/L lower than the DBIC [18]. Although data on blood bicarbonate levels in patients after 2 h of dialysis are lacking, we speculate that, in the absence of alkalosis, whether there is a rapid change in serum bicarbonate levels during the first 2 h of dialysis treatment that may affect patient prognosis needs to be further clarified in multicenter, prospective randomized controlled studies.

This study has some potential limitations. (1) Although the dialysate prescription for bicarbonate was individualized, it was based on empirical adjustment. Therefore, no rigorous formula was used for calculation. (2) We did not have an accurate assessment of the nutritional status, dietary protein intake, and calorie intake of patients. (3) Our study lacked data on the arterial blood gas analysis of patients. We did not assess the effect of respiratory function on the experimental results, and patients with chronic lung disease were excluded. Moreover, we did not consider the physical activity or residual renal function in detail. (4) Although the timing of post-HD blood sampling in most patients was at midweek, a very small number of patients underwent blood sampling after a long interval (i.e., at the beginning of the week). The delay in the collection of a small number of samples may have also affected the accuracy of results, all of which may have influenced our findings to some extent. (5) This study used single-center data, meaning that it included a relatively small number of patients, which may have resulted in patient selection bias.

Despite significant improvements in HD technology, the most appropriate DBIC in patients undergoing MHD remains controversial. According to our study, a higher DBIC was significantly associated with higher all-cause mortality after full adjustment (HR per 3 mmol/L higher DBIC, 4.321; 95% CI, 2.185–8.547); however, no significant association was observed between DBIC and first hospitalization. A DBIC concentration of 31–32 mmol/L may benefit patient outcomes. A prospective randomized multicenter study further validates these findings and determines the most appropriate DBIC for patients undergoing MHD is warranted.

The authors would like to thank Professor Wangliang Jia of the Daping Hospital for providing statistical support.

This study was reviewed and approved by the Medical Ethics Committee of the Daping Hospital (approval number: YYLS2022-110). Our study was retrospective; therefore, we did not require written informed consent from patients. Moreover, written informed consent from participants was not required in accordance with local and national guidelines.

All authors declare that they have no competing interests.

This work was supported by grants from the National Natural Science Foundation of China (82270768); Chongqing technology innovation project (2022YSZX-JCX0007CSTB); and Key laboratory open projects (SKLKF202201).

Jingfang Wan, Kehong Chen, and Yani He conceived of and designed the research. Jia Chen, Jun Liu, and Lili Fu collected data. Jingfang Wan, Weiwei Zhang, and Weidong Wang analyzed and interpreted data. Jingfang Wan and Jing Lin wrote the manuscript. Jingfang Wan and Yang Xiang prepared the figures. All authors contributed to discussions and reviewed the manuscript.

All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.

1.
The 2021 annual report of blood purification [EB/OL]
. http://www.cnrds.net/TxLogin.
2.
Abramowitz
MK
.
Bicarbonate balance and prescription in ESRD
.
J Am Soc Nephrol
.
2017 Mar
28
3
726
34
.
3.
Chen
X
.
The blood purification of standard operating procedure (SOP)
2021
. [EB/OL].
4.
Clinical practice guidelines for nutrition in chronic renal failure. K/DOQI, National Kidney Foundation
.
Am J Kidney Dis
.
2000 Jun
35
6 Suppl 2
S17
04
.
5.
Tentori
F
,
Karaboyas
A
,
Robinson
BM
,
Morgenstern
H
,
Zhang
J
,
Sen
A
.
Association of dialysate bicarbonate concentration with mortality in the dialysis outcomes and practice patterns study (DOPPS)
.
Am J Kidney Dis
.
2013 Oct
62
4
738
46
.
6.
Daugirdas
JT
.
Second generation logarithmic estimates of single-pool variable volume Kt/V: an analysis of error
.
J Am Soc Nephrol
.
1993 Nov
4
5
1205
13
.
7.
Yamamoto
T
,
Shoji
S
,
Yamakawa
T
,
Wada
A
,
Suzuki
K
,
Iseki
K
.
Predialysis and postdialysis pH and bicarbonate and risk of all-cause and cardiovascular mortality in long-term hemodialysis patients
.
Am J Kidney Dis
.
2015 Sep
66
3
469
78
.
8.
Chang
KY
,
Kim
HW
,
Kim
WJ
,
Kim
YK
,
Kim
SH
,
Song
HC
.
The impact of high serum bicarbonate levels on mortality in hemodialysis patients
.
Korean J Intern Med
.
2017 Jan
32
1
109
16
.
9.
Montagud-Marrahi
E
,
Broseta
J
,
Rodriguez-Espinosa
D
,
Lidia
R
,
Hermida-Lama
E
,
Xipell
M
.
Optimization of dialysate bicarbonate in patients treated with online haemodiafiltration
.
Clin Kidney J
.
2021
;
14
(
3
):
1004
13
.
10.
Di Iorio
B
,
Torraca
S
,
Piscopo
C
,
Sirico
ML
,
Di Micco
L
,
Pota
A
.
Dialysate bath and QTc interval in patients on chronic maintenance hemodialysis: pilot study of single dialysis effects
.
J Nephrol
.
2012 Sep–Oct
25
5
653
60
.
11.
Sethi
D
,
Curtis
JR
,
Topham
DL
,
Gower
PE
.
Acute metabolic alkalosis during haemodialysis
.
Nephron
.
1989
;
51
(
1
):
119
20
.
12.
Kaye
M
,
Somerville
PJ
,
Lowe
G
,
Ketis
M
,
Schneider
W
.
Hypocalcemic tetany and metabolic alkalosis in a dialysis patient: an unusual event
.
Am J Kidney Dis
.
1997 Sep
30
3
440
4
.
13.
Gabutti
L
,
Ferrari
N
,
Giudici
G
,
Mombelli
G
,
Marone
C
.
Unexpected haemodynamic instability associated with standard bicarbonate haemodialysis
.
Nephrol Dial Transplant
.
2003 Nov
18
11
2369
76
.
14.
Gabutti
L
,
Ross
V
,
Duchini
F
,
Mombelli
G
,
Marone
C
.
Does bicarbonate transfer have relevant hemodynamic consequences in standard hemodialysis
.
Blood Purif
.
2005
;
23
(
5
):
365
72
.
15.
Harris
DC
,
Yuill
E
,
Chesher
DW
.
Correcting acidosis in hemodialysis: effect on phosphate clearance and calcification risk
.
J Am Soc Nephrol
.
1995 Dec
6
6
1607
12
.
16.
Locatelli
F
,
La Milia
V
,
Violo
L
,
Del Vecchio
L
,
Di Filippo
S
.
Optimizing haemodialysate composition
.
Clin Kidney J
.
2015 Oct
8
5
580
9
.
17.
Gennari
FJ
.
Acid-base balance in dialysis patients
.
Semin Dial
.
2000 Jul–Aug
13
4
235
9
.
18.
Basile
C
,
Libutti
P
,
Di Turo
AL
,
Vernaglione
L
,
Casucci
F
,
Losurdo
N
.
Effect of dialysate calcium concentrations on parathyroid hormone and calcium balance during a single dialysis session using bicarbonate hemodialysis: a crossover clinical trial
.
Am J Kidney Dis
.
2012 Jan
59
1
92
101
.

Additional information

Jingfang Wan and Jing Lin contributed equally to this work and are co-first authors.