Original Investigation Predialysis and Postdialysis pH and Bicarbonate and Risk of All-Cause and Cardiovascular Mortality in Long-term Hemodialysis Patients Tadashi Yamamoto, PhD,1 Shigeichi Shoji, MD, PhD,1 Tomoyuki Yamakawa, MD, PhD,1 Atsushi Wada, MD, PhD,2 Kazuyuki Suzuki, MD, PhD,2 Kunitoshi Iseki, MD, PhD,2 and Yoshiharu Tsubakihara, MD, PhD2 Background: To date, very few studies have been carried out on the associations of pre- and postdialysis acid-base parameters with mortality in hemodialysis patients. Study Design: An observational study including cross-sectional and 1-year analyses. Setting & Participants: Data from the renal registry of the Japanese Society of Dialysis Therapy (20082009), including 15,132 dialysis patients 16 years or older. Predictor: Predialysis pH , 7.30, 7.30 to 7.34 (reference), 7.35 to 7.39, or $7.40 (1,550, 4,802, 6,023, and 2,757 patients, respectively); predialysis bicarbonate level , 18.0, 18.0 to 21.9 (reference), 22.0 to 25.9, or $26.0 mEq/L (2,724, 7,851, 4,023, and 534 patients, respectively); postdialysis pH , 7.40, 7.40 to 7.44, 7.45 to 7.49 (reference), or $7.50 (2,114, 5,331, 4,975, and 2,712 patients, respectively); and postdialysis bicarbonate level , 24.0, 24.0 to 25.9, 26.0 to 27.9 (reference), or $28.0 mEq/L (5,087, 4,330, 3,451, and 2,264 patients, respectively). Outcomes: All-cause and cardiovascular (CV) mortality during the 1-year follow-up. Measurements: HRs were estimated using unadjusted models and models adjusted for age, sex, dialysis vintage, history of CV disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, serum total cholesterol level, blood hemoglobin level, single-pool Kt/V, and normalized protein catabolic rate. Results: Of 15,132 patients, during follow-up, 1,042 died of all causes, including 408 CV deaths. In the adjusted analysis for all-cause mortality, HRs compared to the reference group were significantly higher in patients with predialysis pH $ 7.40 (HR, 1.36; 95% CI, 1.13-1.65) and postdialysis pH , 7.40 (HR, 1.22; 95% CI, 1.00-1.49). Predialysis pH $ 7.40 was also associated with higher risk of CV mortality (HR, 1.34; 95% CI, 1.01-1.79). No association of pre- or postdialysis bicarbonate level with all-cause and CV mortality was observed. Limitations: Single measurements of acid-base parameters, short duration of follow-up, small number of CV deaths. Conclusions: Predialysis pH $ 7.40 was associated with significantly elevated risk of all-cause and CV mortality. However, pre- and postdialysis bicarbonate levels were not associated with all-cause and CV mortality. Predialysis pH may be the most appropriate reference for accurate correction of metabolic acidosis in dialysis patients. Am J Kidney Dis. -(-):---. ª 2015 by the National Kidney Foundation, Inc. INDEX WORDS: Predialysis pH; postdialysis pH; serum bicarbonate; metabolic acidosis; acid-base status; allcause mortality; cardiovascular mortality; hemodialysis; dialysate bicarbonate; end-stage renal disease (ESRD); renal replacement therapy; Japanese Renal Data Registry (JRDR).
I
t is known for dialysis patients that metabolic acidosis adversely affects both protein and bone metabolism, and accurate correction of metabolic acidosis is an important goal for obtaining a good prognosis. Optimal predialysis bicarbonate levels have been proposed by several entities: $22 mEq/L by the 2000 and 2003 NKF-KDOQI (National Kidney Foundation2Kidney Disease Outcomes Quality Initiative) guidelines,1,2 19 to 22 mEq/L by the DOPPS (Dialysis Outcomes and Practice Patterns Study) in a publication from 2004,3 20 to 22 mEq/L by the 2007 EBPG (European Best Practice Guidelines),4 and $22 mEq/L by the 2007 CARI (Caring for Australasians With Renal Impairment) guidelines.5 However, these suggestions pose 2 difficulties. Am J Kidney Dis. 2015;-(-):---
First, opinions vary about whether the optimal predialysis bicarbonate level should be ,22 or .22 mEq/L. In order to improve protein and bone metabolism, predialysis bicarbonate level should be $22 mEq/L, whereas for avoiding metabolic alkalosis arising from From the 1Kidney Center, Shirasagi Hospital, Osaka; and Committee of Renal Data Registry of the Japanese Society for Dialysis Therapy, Tokyo, Japan. Received November 25, 2014. Accepted in revised form April 5, 2015. Address correspondence to Tadashi Yamamoto, PhD, Kidney Center, Shirasagi Hospital, 7-11-23 Kumata, Osaka 546-0002, Japan. E-mail: [email protected] 2015 by the National Kidney Foundation, Inc. 0272-6386 http://dx.doi.org/10.1053/j.ajkd.2015.04.014
2
1
Yamamoto et al
overcorrection of the acidosis, the value should be ,22 mEq/L. Metabolic alkalosis has been suggested as a possible risk factor for metastatic calcification; therefore, high predialysis bicarbonate level may be associated with an increased risk of cardiovascular (CV) and all-cause mortality.3,6 However, there is still insufficient information on the role of metabolic alkalosis in metastatic calcification. Second, to our knowledge, the only parameter of acid-base status examined to date is predialysis serum bicarbonate level, and the significance of other acidbase parameters such as predialysis pH, postdialysis pH, and postdialysis serum bicarbonate level have not yet been examined. In particular, very few studies to date have been carried out to determine the association of predialysis pH with outcomes. The reason for this is that total carbon dioxide content is easily measured as a marker of bicarbonate level.7,8 In addition, serum bicarbonate measurement is a convenient tool for adjustment of the bicarbonate concentration in dialysate for treating metabolic acidosis. However, to date, there is no consensus regarding which acid-base parameter might be used as the optimal marker to correct metabolic acidosis in dialysis patients. In this observational study, we examined whether predialysis bicarbonate level is the most appropriate parameter for correction of metabolic acidosis and also attempted to clarify the association of other acidbase parameters with the risk of all-cause and CV mortality, using the data registry of the Japanese Society of Dialysis Therapy (JSDT).
METHODS
52,432 patients aged 16 + with baseline blood gas data, receiving hemodialysis 3 times a week
Excluded (n=7,111) Time on dialysis ≤ 12 months (n=6,563) Input errors to the database (n=305) Patients with a central venous catheter (n=243)
45,321 patients with pre- and postdialysis blood gas data Excluded (n=30,189) Patients with incomplete pre- and postdialysis blood gas data
15,132 patients were analyzed
Figure 1. Selection process for study patients.
blood gas analyzer in local laboratories, and bicarbonate level was calculated from pH and PCO2. However, PCO2 data were not available from the database and therefore were computed from pH and bicarbonate values using the Henderson equation.
Primary Exposures We defined pre- and postdialysis pH and serum bicarbonate levels as the primary exposures. To determine factors associated with the primary exposures, analysis of the associations between the primary exposures and potentially relevant factors was performed. We also examined the effects of dialysate bicarbonate concentration (30 and 35 mEq/L) on the primary exposures in 5,659 and 1,913 patients, respectively. Total alkali concentration was 36 mEq/L (including 6 mEq/L of sodium acetate) for the 30mEq/L dialysate and 36 mEq/L (including 1 mEq/L of sodium citrate) for the 35-mEq/L dialysate.
Data Source and Study Participant Characteristics
Outcomes
This study is based on data from 2008 and 2009 recorded in the Japanese Renal Data Registry (JRDR), which is the dialysis registry of the JSDT.9 The study was registered with the University Hospital Medical Information Network Clinical Trials Registry (no. UMIN000009287) and the JRDR (no. JRDR-10001). Eligible patients were those 16 years or older who underwent dialysis 3 times a week and for whom blood gas analysis data were available. The total number of eligible patients was 52,432; of these, 7,111 were excluded on account of dialysis vintage of 12 months or less, presence of a central venous catheter, or input errors in the database. The reason for limiting the analysis to patients with dialysis vintage longer than 1 year is that in our experience, this selects for patients who are clinically stable and have minimal residual kidney function. Of 45,321 remaining patients, those with incomplete preand postdialysis blood gas data were excluded. The final number of patients enrolled was 15,132 (Fig 1). The data collected included demographic information (age, sex, weight gain ratio, dialysis vintage, cause of kidney failure, history of CV disease, and dialysate concentration of bicarbonate and total alkali), clinical data such as serum calcium, inorganic phosphate, calcium-phosphorus product, serum albumin, serum total cholesterol, blood hemoglobin, single-pool Kt/V (Kt/V), body mass index, normalized protein catabolic rate (nPCR), and acid-base parameters such as pre- and postdialysis pH and bicarbonate level. All acid-base parameters were measured on the same day with a
We defined all-cause mortality as the primary outcome and CV mortality as the secondary outcome for this study. Cause of death was assigned in accordance with guidelines of the International Classification of Diseases, Tenth Revision (ICD-10). Cardiovascular mortality was defined as sudden death (ICD-10 code I46) and death from ischemic heart diseases (ICD-10 code I20-I25), heart failure (ICD-10 code I50), and stroke (ICD-10 code I60-I63). Data were obtained from the database of 2009 (maximum followup, 12 months).
2
Statistical Analysis Study patients were divided into quartiles based on baseline characteristics of predialysis pH or bicarbonate level, and quartiles were compared using analysis of variance for continuous variables and c2 test for categorical variables. With regard to acid-base status, it is important to consider dialysate total alkali concentration (sodium bicarbonate 1 sodium acetate or citrate). Dialysate that is available in Japan is characterized by the total alkali concentration being fixed at 35 to 38 mEq/L, whereas bicarbonate concentration varies widely from 25 to 35 mEq/L. Thus, because the effect of the alkalizing agent on acid-base status can only be explained by the dialysate bicarbonate concentration, statistical analysis was performed using dialysate bicarbonate concentration. Multivariable logistic regression analysis was performed to determine factors associated with higher levels of acid-base Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality Table 1. Baseline Characteristics of Study Patients, by Predialysis pH Quartile Overall (N 5 15,132)
pH , 7.30 (n 5 1,550)
pH 7.30-7.34 (n 5 4,802)
pH 7.35-7.39 (n 5 6,023)
pH $ 7.40 (n 5 2,757)
66 (57-74) 6.5 (3.3-11.8) 9,289 (61.4) 3,843 (25.4) 4.7 6 1.7 21.2 6 3.5 47.8 6 13.7 3.7 6 0.4 153 6 35 10.5 6 1.2 1.44 6 0.29 0.88 6 0.17 27.9 6 3.3 34.1 6 2.1
64 (56-72) 7.0 (3.8-11.5) 868 (56.0) 383 (24.7) 5.0 6 1.7 21.3 6 4.0 52.5 6 14.3 3.8 6 0.4 147 6 34 10.7 6 1.3 1.44 6 0.30 0.95 6 0.18 26.9 6 2.8 33.3 6 2.2
65 (57-74) 6.7 (3.4-11.7) 2,984 (62.1) 1,173 (24.4) 4.8 6 1.7 21.3 6 3.5 49.0 6 13.7 3.7 6 0.4 151 6 34 10.6 6 1.2 1.44 6 0.29 0.91 6 0.17 27.5 6 3.1 33.9 6 2.1
66 (58-74) 6.3 (3.2-11.8) 3,743 (62.1) 1,519 (25.2) 4.6 6 1.7 21.3 6 3.4 47.1 6 13.5 3.7 6 0.4 155 6 35 10.4 6 1.2 1.43 6 0.29 0.87 6 0.16 28.0 6 3.4 34.2 6 2.0
67 (58-75) 6.3 (3.3-11.8) 1,694 (61.4) 768 (27.9) 4.4 6 1.7 21.1 6 3.3 44.8 6 13.3 3.7 6 0.5 157 6 36 10.2 6 1.2 1.43 6 0.29 0.83 6 0.17 28.6 6 3.6 34.6 6 2.0
6,292 (41.5) 5,049 (33.4) 1,098 (7.3) 504 (3.3) 2,189 (14.5)
586 (37.8) 570 (36.8) 103 (6.6) 46 (3.0) 245 (15.8)
2,035 (42.4) 1,692 (35.2) 327 (6.8) 139 (2.9) 609 (12.7)
2,531 (42.0) 1,937 (32.1) 444 (7.4) 215 (3.6) 896 (14.9)
1,140 (41.4) 850 (30.8) 224 (8.1) 104 (3.8) 439 (15.9)
Predialysis acid-base parameters pH Bicarbonate, mEq/L PCO2, mm Hg
7.353 6 0.048 20.5 6 2.9 38.1 6 4.7
7.267 6 0.027 17.6 6 2.5 40.0 6 6.0
7.323 6 0.014 19.5 6 2.3 38.8 6 4.5
7.368 6 0.014 21.1 6 2.4 37.9 6 4.3
7.421 6 0.024 22.9 6 2.8 36.4 6 4.4
,0.001 ,0.001 ,0.001
Postdialysis acid-base parameters pH Bicarbonate, mEq/L PCO2, mm Hg
7.444 6 0.047 25.1 6 2.8 37.9 6 4.5
7.398 6 0.046 23.0 6 2.5 38.6 6 5.1
7.429 6 0.041 24.3 6 2.5 38.0 6 4.6
7.453 6 0.039 25.5 6 2.6 37.8 6 4.4
7.477 6 0.044 26.7 6 2.9 37.4 6 4.5
,0.001 ,0.001 ,0.001
Characteristics
Age, y Dialysis vintage, y Male sex History of CVD Weight gain ratio, % BMI, kg/m2 Ca 3 P product, mg2/dL2 Serum albumin, g/dL Serum total cholesterol, mg/dL Hemoglobin, g/dL Single-pool Kt/V nPCR, g/kg/d Dialysate bicarbonate, mEq/L Dialysate total alkali, mEq/L Cause of kidney failure Glomerulonephritis Diabetes Nephrosclerosis Polycystic kidney Other
P
,0.001 ,0.001 ,0.001 ,0.001 ,0.001 0.2 ,0.001 ,0.001 ,0.001 ,0.001 0.3 ,0.001 ,0.001 ,0.001 ,0.001
Note: Values for categorical variables are given as number (percentage); values for continuous variables, as mean 6 standard deviation or median [interquartile range]. P value for comparison between pH quartile groups using analysis of variance or c2 test. Conversion factors for units: serum calcium in mg/dL to mmol/L, 30.2495; serum phosphate in mg/dL to mmol/L, 30.3229; serum total cholesterol in mg/dL to mmol/L, 30.02586. PCO2 was computed from pH and bicarbonate values using the Henderson equation. Abbreviations: BMI, body mass index; Ca, calcium; Ca 3 P, calcium-phosphorus product; CVD, cardiovascular disease; nPCR, normalized protein catabolic rate.
parameters. We defined higher levels of acid-base parameters as follows: predialysis pH $ 7.40, predialysis bicarbonate level $ 26.0 mEq/L, postdialysis pH $ 7.50, and postdialysis bicarbonate level $ 28.0 mEq/L. Odds ratios were estimated with their 95% confidence intervals (CIs). The model included age, sex, dialysis vintage, history of CV disease, diabetes, dialysate bicarbonate concentration, weight gain ratio, body mass index, serum albumin level, hemoglobin level, Kt/V, and nPCR. For the association between each acid-base parameter and outcomes, we conducted Cox proportional hazard analysis for all-cause and CV mortality. Because acid-base parameters show strong multicollinearity,10 the analysis was performed based on 4 models for each acid-base parameter and 2 models for pH and bicarbonate level pre-post differences (difference between pre- and postdialysis values). We also examined a model including the association between narrow categories of acid-base parameters and outcomes by dividing patients into quartiles of predialysis pH (,7.30, 7.30-7.34, 7.35-7.39, or $7.40), predialysis bicarbonate level (,18.0, 18.021.9, 22.0-25.9, or $26.0 mEq/L), postdialysis pH (,7.40, 7.407.44, 7.45-7.49, or $7.50), and postdialysis bicarbonate level (,24.0, 24.0-25.9, 26.0-27.9, or $28.0 mEq/L). All models were adjusted for the full variables (age, sex, dialysis vintage, history of CV disease, diabetes, dialysate bicarbonate concentration, weight gain ratio, body mass index, calcium-phosphorus product, serum Am J Kidney Dis. 2015;-(-):---
albumin level, serum total cholesterol level, hemoglobin level, Kt/ V, and nPCR). Hazard ratios (HRs) for all-cause and CV mortality with their 95% CIs were estimated. We additionally used a cubic spline model to explore the linearity between acid-base parameters and outcomes, and Kaplan-Meier analysis, for all-cause and CV mortality according to 3 categories of predialysis pH (,7.35, 7.357.39, or $7.40) and predialysis bicarbonate level (,22.0, 22.025.9, or $26.0 mEq/L). All statistical analyses were performed using IBM SPSS software (version 17.0; IBM Japan).
RESULTS Baseline Characteristics Baseline characteristics of patients based on predialysis pH are listed in Table 1. Median age was 66 years, median dialysis vintage was 6.5 years, 61.4% were men, 25.4% had a history of CV disease, and 33.4% had diabetes. The most common cause of kidney failure was glomerulonephritis. Patients with a higher predialysis pH had a higher frequency of CV diseases and lower frequency of diabetes and showed 3
Yamamoto et al
lower weight gain ratio, calcium-phosphorus product, hemoglobin, and nPCR values than patients with a lower predialysis pH. Higher predialysis pH was associated with higher dialysate bicarbonate and total alkali levels. Baseline characteristics stratified by predialysis serum bicarbonate level are listed in Table S1 (provided as online supplementary material). The same trends as those for patients categorized according to predialysis pH were observed, except that higher predialysis bicarbonate level was associated with higher frequency of diabetes.
dialysate bicarbonate concentration of 35 versus 30 mEq/L was associated with significantly greater levels of the following acid-base parameters (Table 3): predialysis pH (by 0.018 units), predialysis bicarbonate level (by 1.7 mEq/L), postdialysis pH (by 0.017 units), postdialysis bicarbonate level (by 2.4 mEq/L), and bicarbonate pre-post difference (by 0.6 mEq/L). There was no statistically significant difference in pH pre-post difference between the 2 dialysate concentrations. Outcome Data
Factors Associated With Acid-Base Parameters Results of multivariable logistic regression analysis to determine factors associated with higher levels of acid-base parameters are listed in Table 2. Odds of higher predialysis pH and predialysis bicarbonate level were significantly increased by greater dialysate bicarbonate concentration; however, the odds of elevated levels of these predialysis acid-base parameters were significantly decreased by greater weight gain ratio, serum albumin level, hemoglobin level, and nPCR. The presence of diabetes increased the odds of higher predialysis pH but decreased the odds of higher predialysis bicarbonate level. Odds of higher postdialysis pH and postdialysis bicarbonate level were significantly increased by greater dialysate bicarbonate concentration and Kt/V; however, the odds of elevated levels of these postdialysis acid-base parameters were significantly decreased by greater weight gain ratio and hemoglobin level and absence of CV disease history. Increments in nPCR significantly increased the odds of higher postdialysis pH, but significantly decreased the odds of higher postdialysis bicarbonate level. With regard to the relationship between dialysate bicarbonate concentration and acid-base parameters,
During the 1-year follow-up period, all-cause mortality occurred in 1,042 patients, including heart failure in 211, infection in 186, malignancy in 113, stroke in 109, and other causes in 423. CV mortality occurred in 408 patients, including heart failure in 211, stroke in 109, sudden death in 51, and ischemic heart diseases in 37. Of 15,132 patients, 26 were censored because of kidney transplantation. We analyzed unadjusted and adjusted associations of acid-base parameters with risk of all-cause and CV mortality (Table 4). Unadjusted analysis revealed significant increases in the HR of all-cause mortality with each 0.1-unit increment in predialysis pH and bicarbonate level and significant decreases in HR with each 0.1-unit increment in postdialysis pH and pH prepost difference and each 1-mEq/L increase in bicarbonate level pre-post difference. After full adjustment, the associations remained significant for predialysis pH (HR for 0.1-unit increase, 1.18; 95% CI, 1.02-1.37) and pH pre-post difference (HR for 0.1-unit increase, 0.79; 95% CI, 0.68-0.92). Although predialysis pH and pH and bicarbonate pre-post differences were associated with CV mortality in the unadjusted analysis, the association disappeared after full adjustment.
Table 2. Logistic Regression Analysis to Determine Factors Associated With Acid-Base Parameters Predialysis
pH $ 7.40 (n 5 2,757)
Age, per 10-y older Men, vs women Dialysis vintage, per 1-y longer History of CVD Diabetes Dialysate bicarbonate, per 1-mEq/L greater Weight gain ratio, per 1% greater BMI, per 1-kg/m2 greater Serum albumin, per 0.1-g/dL greater Hemoglobin, per 1-g/dL greater Single-pool Kt/V, per 0.1-unit greater nPCR, per 0.1-g/kg/d greater
0.97 1.07 1.00 1.10 0.80 1.08 0.93 0.99 0.99 0.85 1.03 0.81
(0.94-1.01) (0.98-1.18) (0.99-1.01) (1.00-1.22) (0.73-0.88) (1.07-1.09) (0.91-0.96) (0.98-1.01) (0.97-0.99) (0.82-0.88) (1.01-1.05) (0.79-0.84)
Postdialysis
Bicarbonate $ 26.0 mEq/L (n 5 534)
0.94 0.90 1.00 1.53 1.42 1.11 0.88 0.88 0.97 0.83 1.02 0.83
(0.87-1.02) (0.74-1.09) (0.99-1.01) (1.27-1.84) (1.17-1.29) (1.08-1.13) (0.84-0.93) (0.85-0.91) (0.95-0.99) (0.77-0.89) (0.99-1.06) (0.78-0.88)
pH $ 7.50 (n 5 2,712)
0.97 1.12 1.00 0.90 0.63 1.13 0.95 1.02 1.00 0.84 1.06 1.07
(0.93-1.01) (1.01-1.23) (0.99-1.01) (0.81-0.99) (0.57-0.70) (1.12-1.14) (0.92-0.97) (1.00-1.03) (0.99-1.01) (0.81-0.87) (1.04-1.08) (1.04-1.10)
Bicarbonate $ 28.0 mEq/L (n 5 2,264)
0.96 1.08 0.99 0.83 0.99 1.25 0.94 0.97 1.03 0.81 1.09 0.90
(0.93-1.01) (0.97-1.20) (0.98-1.00) (0.74-0.93) (0.89-1.11) (1.24-1.27) (0.91-0.97) (0.95-0.98) (1.02-1.04) (0.78-0.84) (1.07-1.11) (0.87-0.93)
Note: Values are given as odds ratio (95% confidence interval). Abbreviations: BMI, body mass index; CVD, cardiovascular disease; nPCR, normalized protein catabolic rate. 4
Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality Table 3. Relationship Between Dialysate Bicarbonate Concentrations of 30 and 35 mEq/L and Acid-Base Parameters Dialysate Bicarbonate Concentration
Acid-Base Parameter
30 mEq/La (n 5 5,659)
35 mEq/La (n 5 1,913)
P
Predialysis pH 7.354 6 0.046 7.372 6 0.047 ,0.001 Bicarbonate, mEq/L 20.5 6 2.8 22.2 6 2.8 ,0.001 Postdialysis pH 7.448 6 0.044 7.465 6 0.048 ,0.001 Bicarbonate, mEq/L 25.3 6 2.4 27.7 6 2.4 ,0.001 pH pre-post difference 0.094 6 0.047 0.093 6 0.056 0.3 Bicarbonate pre-post 4.8 6 2.4 5.4 6 2.5 ,0.001 difference, mEq/L Note: Values are given as mean 6 standard deviation. P value for comparison between 2 groups using t test. a Total alkali concentration was 36 mEq/L.
We estimated mortality rates and adjusted HRs from the Cox regression analysis for all-cause mortality (Table 5) and CV mortality (Table 6) based on categories of acid-base parameters. We observed higher HRs of all-cause mortality in patients with predialysis pH $ 7.40 (HR, 1.36; 95% CI, 1.13-1.65)
and postdialysis pH , 7.40 (HR, 1.22; 95% CI, 1.001.49). As for CV mortality, a higher HR was found in patients with predialysis pH $ 7.40 (HR, 1.34; 95% CI, 1.01-1.79). Unadjusted and adjusted models for mortality are summarized in Fig 2 (all-cause mortality) and Fig 3 (CV mortality). A higher HR of allcause and CV mortality was consistently observed in patients with predialysis pH $ 7.40. The cubic spline model showed a nonlinear relationship between acid-base parameters and all-cause (Fig S1) and CV (Fig S2) mortality. Kaplan-Meier analysis for allcause mortality performed for the 3 categories of predialysis pH and bicarbonate level showed significantly higher cumulative mortality in patients with predialysis pH $ 7.4 and with predialysis bicarbonate level $ 26.0 mEq/L (Fig S3). As for CV mortality, significantly higher cumulative mortality was found in patients with predialysis pH $ 7.4 (Fig S4).
DISCUSSION In this study, predialysis pH $ 7.4 was associated with a significantly elevated 1-year risk of all-cause and CV mortality. However, predialysis bicarbonate level and postdialysis pH and bicarbonate level were not associated with all-cause and CV mortality. These findings suggest that predialysis pH may be the most
Table 4. Hazard Ratios for All-Cause and CV Mortality by Predialysis pH and Bicarbonate, Postdialysis pH and Bicarbonate, pH Difference, and Bicarbonate Difference All-Cause Mortality (n 5 1,042)
CV Mortality (n 5 408)
HR (95% CI)
P
HR (95% CI)
1.44 (1.26-1.64) 1.18 (1.02-1.37)
,0.001 0.02
1.38 (1.12-1.69) 1.22 (0.98-1.52)
0.002 0.07
1.06 (1.04-1.09) 1.00 (0.98-1.03)
,0.001 0.8
1.02 (0.98-1.05) 0.97 (0.94-1.01)
0.3 0.9
0.82 (0.72-0.94) 0.95 (0.82-1.09)
0.003 0.4
0.95 (0.77-1.17) 1.14 (0.92-1.42)
0.6 0.2
0.99 (0.97-1.01) 0.98 (0.96-1.01)
0.3 0.2
0.99 (0.95-1.02) 0.99 (0.96-1.03)
0.3 0.9
pH pre-post difference, per 0.1-unit greater Unadjusted Adjusted
0.57 (0.50-0.65) 0.79 (0.68-0.92)
,0.001 0.002
0.68 (0.56-0.84) 0.93 (0.75-1.17)
,0.001 0.5
Bicarbonate pre-post difference, per 1-mEq/L greater Unadjusted Adjusted
0.91 (0.89-0.93) 0.98 (0.95-1.00)
,0.001 0.09
0.96 (0.92-0.99) 1.03 (0.99-1.07)
0.03 0.1
Predialysis pH, per 0.1-unit greater Unadjusted Adjusted Bicarbonate, per 1-mEq/L greater Unadjusted Adjusted Postdialysis pH, per 0.1-unit greater Unadjusted Adjusted Bicarbonate, per 1-mEq/L greater Unadjusted Adjusted
P
Note: Adjusted model included age, sex, dialysis vintage, history of CV disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviations: CI, confidence interval; CV, cardiovascular; HR, hazard ratio. Am J Kidney Dis. 2015;-(-):---
5
Yamamoto et al Table 5. Mortality Rates and Hazard Ratios for All-Cause Mortality by 4 Categories of Predialysis and Postdialysis pH and Bicarbonate
Predialysis pH ,7.30 7.30-7.34 7.35-7.39 $7.40 Bicarbonate ,18.0 mEq/L 18.0-21.9 mEq/L 22.0-25.9 mEq/L $26.0 mEq/L Postdialysis pH ,7.40 7.40-7.44 7.45-7.49 $7.50 Bicarbonate ,24.0 mEq/L 24.0-25.9 mEq/L 26.0-27.9 mEq/L $28.0 mEq/L
Mortality Rate (95% CI)
P
No. of Deaths
Person-Years
HR (95% CI)
94 293 389 266
1,503.0 4,681.3 5,847.6 2,626.8
6.3 6.3 6.7 10.1
(5.0-7.5) (5.6-7.0) (6.0-7.3) (9.0-11.3)
1.07 1.00 0.98 1.36
(0.83-1.37) (reference) (0.83-1.16) (1.13-1.65)
157 504 314 67
2,648.8 7,633.8 3,877.7 498.5
5.9 6.6 8.1 13.5
(5.0-6.8) (6.0-7.2) (7.2-9.0) (10.5-16.5)
1.08 1.00 1.04 1.31
(0.89-1.31) (reference) (0.88-1.21) (0.98-1.77)
185 313 366 178
2,026.7 4,835.0 5,170.6 2,626.5
9.1 6.5 7.1 6.8
(7.7-10.4) (5.8-7.2) (6.4-7.8) (5.8-7.7)
1.22 1.02 1.00 1.16
(1.00-1.49) (0.87-1.21) (reference) (0.95-1.42)
0.05 0.8
361 287 254 140
4,920.5 4,199.3 3,354.1 2,193.8
7.3 6.8 7.6 6.4
(6.6-8.1) (6.1-7.6) (6.7-8.5) (5.4-7.4)
1.05 1.08 1.00 0.96
(0.89-1.24) (0.90-1.30) (reference) (0.76-1.20)
0.5 0.4
0.6 0.8 0.002 0.4 0.6 0.07
0.1
0.6
Note: Patients were divided into quartiles of pre- and postdialysis pH and bicarbonate level as shown. All models were adjusted for age, sex, dialysis vintage, history of cardiovascular disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviations: CI, confidence interval; HR, hazard ratio.
Table 6. Mortality Rates and Hazard Ratios for Cardiovascular Mortality by 4 Categories of Predialysis and Postdialysis pH and Bicarbonate
Predialysis pH ,7.30 7.30-7.34 7.35-7.39 $7.40 Bicarbonate ,18.0 mEq/L 18.0-21.9 mEq/L 22.0-25.9 mEq/L $26.0 mEq/L Postdialysis pH ,7.40 7.40-7.44 7.45-7.49 $7.50 Bicarbonate ,24.0 mEq/L 24.0-25.9 mEq/L 26.0-27.9 mEq/L $28.0 mEq/L
Mortality Rate (95% CI)
Hazard Ratio (95% CI)
P
No. of Deaths
Person-Years
42 112 155 99
1,529.7 4,753.0 5,951.4 2,708.5
2.7 2.4 2.6 3.7
(1.9-3.6) (1.9-2.8) (2.2-3.0) (2.9-4.4)
1.21 1.00 1.05 1.34
(0.84-1.75) (reference) (0.81-1.35) (1.01-1.79)
69 209 112 18
2,691.5 7,765.1 3,963.0 522.9
2.6 2.7 2.8 3.4
(2.0-3.2) (2.3-3.1) (2.3-3.3) (1.9-5.0)
1.10 1.00 0.90 0.86
(0.83-1.47) (reference) (0.70-1.14) (0.52-1.43)
62 144 125 77
2,083.2 5,268.2 4,915.8 2,675.4
3.0 2.7 2.5 2.9
(2.2-3.7) (2.3-3.2) (2.1-3.0) (2.2-3.5)
1.00 1.04 1.00 1.29
(0.73-1.36) (0.81-1.34) (reference) (0.96-1.72)
0.9 0.7
141 112 104 51
5,022.5 4,279.2 3,404.0 2,236.8
2.8 2.6 3.1 2.3
(2.4-3.3) (2.1-3.1) (2.5-3.6) (1.7-2.9)
1.02 1.14 1.00 0.91
(0.79-1.32) (0.87-1.51) (reference) (0.65-1.28)
0.8 0.3
0.3 0.7 0.05 0.5 0.3 0.5
0.08
0.5
Note: Patients were divided into quartiles of pre- and postdialysis pH and bicarbonate level as shown. All models were adjusted for age, sex, dialysis vintage, history of cardiovascular disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviation: CI, confidence interval. 6
Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality A Predialysis pH
P-value 0.9 0.6
< 7.30
B Predialysis
Reference
18.0-21.9
7.35-7.39
0.4 0.8
22.0-25.9
I
t is known for dialysis patients that metabolic acidosis adversely affects both protein and bone metabolism, and accurate correction of metabolic acidosis is an important goal for obtaining a good prognosis. Optimal predialysis bicarbonate levels have been proposed by several entities: $22 mEq/L by the 2000 and 2003 NKF-KDOQI (National Kidney Foundation2Kidney Disease Outcomes Quality Initiative) guidelines,1,2 19 to 22 mEq/L by the DOPPS (Dialysis Outcomes and Practice Patterns Study) in a publication from 2004,3 20 to 22 mEq/L by the 2007 EBPG (European Best Practice Guidelines),4 and $22 mEq/L by the 2007 CARI (Caring for Australasians With Renal Impairment) guidelines.5 However, these suggestions pose 2 difficulties. Am J Kidney Dis. 2015;-(-):---
First, opinions vary about whether the optimal predialysis bicarbonate level should be ,22 or .22 mEq/L. In order to improve protein and bone metabolism, predialysis bicarbonate level should be $22 mEq/L, whereas for avoiding metabolic alkalosis arising from From the 1Kidney Center, Shirasagi Hospital, Osaka; and Committee of Renal Data Registry of the Japanese Society for Dialysis Therapy, Tokyo, Japan. Received November 25, 2014. Accepted in revised form April 5, 2015. Address correspondence to Tadashi Yamamoto, PhD, Kidney Center, Shirasagi Hospital, 7-11-23 Kumata, Osaka 546-0002, Japan. E-mail: [email protected] 2015 by the National Kidney Foundation, Inc. 0272-6386 http://dx.doi.org/10.1053/j.ajkd.2015.04.014
2
1
Yamamoto et al
overcorrection of the acidosis, the value should be ,22 mEq/L. Metabolic alkalosis has been suggested as a possible risk factor for metastatic calcification; therefore, high predialysis bicarbonate level may be associated with an increased risk of cardiovascular (CV) and all-cause mortality.3,6 However, there is still insufficient information on the role of metabolic alkalosis in metastatic calcification. Second, to our knowledge, the only parameter of acid-base status examined to date is predialysis serum bicarbonate level, and the significance of other acidbase parameters such as predialysis pH, postdialysis pH, and postdialysis serum bicarbonate level have not yet been examined. In particular, very few studies to date have been carried out to determine the association of predialysis pH with outcomes. The reason for this is that total carbon dioxide content is easily measured as a marker of bicarbonate level.7,8 In addition, serum bicarbonate measurement is a convenient tool for adjustment of the bicarbonate concentration in dialysate for treating metabolic acidosis. However, to date, there is no consensus regarding which acid-base parameter might be used as the optimal marker to correct metabolic acidosis in dialysis patients. In this observational study, we examined whether predialysis bicarbonate level is the most appropriate parameter for correction of metabolic acidosis and also attempted to clarify the association of other acidbase parameters with the risk of all-cause and CV mortality, using the data registry of the Japanese Society of Dialysis Therapy (JSDT).
METHODS
52,432 patients aged 16 + with baseline blood gas data, receiving hemodialysis 3 times a week
Excluded (n=7,111) Time on dialysis ≤ 12 months (n=6,563) Input errors to the database (n=305) Patients with a central venous catheter (n=243)
45,321 patients with pre- and postdialysis blood gas data Excluded (n=30,189) Patients with incomplete pre- and postdialysis blood gas data
15,132 patients were analyzed
Figure 1. Selection process for study patients.
blood gas analyzer in local laboratories, and bicarbonate level was calculated from pH and PCO2. However, PCO2 data were not available from the database and therefore were computed from pH and bicarbonate values using the Henderson equation.
Primary Exposures We defined pre- and postdialysis pH and serum bicarbonate levels as the primary exposures. To determine factors associated with the primary exposures, analysis of the associations between the primary exposures and potentially relevant factors was performed. We also examined the effects of dialysate bicarbonate concentration (30 and 35 mEq/L) on the primary exposures in 5,659 and 1,913 patients, respectively. Total alkali concentration was 36 mEq/L (including 6 mEq/L of sodium acetate) for the 30mEq/L dialysate and 36 mEq/L (including 1 mEq/L of sodium citrate) for the 35-mEq/L dialysate.
Data Source and Study Participant Characteristics
Outcomes
This study is based on data from 2008 and 2009 recorded in the Japanese Renal Data Registry (JRDR), which is the dialysis registry of the JSDT.9 The study was registered with the University Hospital Medical Information Network Clinical Trials Registry (no. UMIN000009287) and the JRDR (no. JRDR-10001). Eligible patients were those 16 years or older who underwent dialysis 3 times a week and for whom blood gas analysis data were available. The total number of eligible patients was 52,432; of these, 7,111 were excluded on account of dialysis vintage of 12 months or less, presence of a central venous catheter, or input errors in the database. The reason for limiting the analysis to patients with dialysis vintage longer than 1 year is that in our experience, this selects for patients who are clinically stable and have minimal residual kidney function. Of 45,321 remaining patients, those with incomplete preand postdialysis blood gas data were excluded. The final number of patients enrolled was 15,132 (Fig 1). The data collected included demographic information (age, sex, weight gain ratio, dialysis vintage, cause of kidney failure, history of CV disease, and dialysate concentration of bicarbonate and total alkali), clinical data such as serum calcium, inorganic phosphate, calcium-phosphorus product, serum albumin, serum total cholesterol, blood hemoglobin, single-pool Kt/V (Kt/V), body mass index, normalized protein catabolic rate (nPCR), and acid-base parameters such as pre- and postdialysis pH and bicarbonate level. All acid-base parameters were measured on the same day with a
We defined all-cause mortality as the primary outcome and CV mortality as the secondary outcome for this study. Cause of death was assigned in accordance with guidelines of the International Classification of Diseases, Tenth Revision (ICD-10). Cardiovascular mortality was defined as sudden death (ICD-10 code I46) and death from ischemic heart diseases (ICD-10 code I20-I25), heart failure (ICD-10 code I50), and stroke (ICD-10 code I60-I63). Data were obtained from the database of 2009 (maximum followup, 12 months).
2
Statistical Analysis Study patients were divided into quartiles based on baseline characteristics of predialysis pH or bicarbonate level, and quartiles were compared using analysis of variance for continuous variables and c2 test for categorical variables. With regard to acid-base status, it is important to consider dialysate total alkali concentration (sodium bicarbonate 1 sodium acetate or citrate). Dialysate that is available in Japan is characterized by the total alkali concentration being fixed at 35 to 38 mEq/L, whereas bicarbonate concentration varies widely from 25 to 35 mEq/L. Thus, because the effect of the alkalizing agent on acid-base status can only be explained by the dialysate bicarbonate concentration, statistical analysis was performed using dialysate bicarbonate concentration. Multivariable logistic regression analysis was performed to determine factors associated with higher levels of acid-base Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality Table 1. Baseline Characteristics of Study Patients, by Predialysis pH Quartile Overall (N 5 15,132)
pH , 7.30 (n 5 1,550)
pH 7.30-7.34 (n 5 4,802)
pH 7.35-7.39 (n 5 6,023)
pH $ 7.40 (n 5 2,757)
66 (57-74) 6.5 (3.3-11.8) 9,289 (61.4) 3,843 (25.4) 4.7 6 1.7 21.2 6 3.5 47.8 6 13.7 3.7 6 0.4 153 6 35 10.5 6 1.2 1.44 6 0.29 0.88 6 0.17 27.9 6 3.3 34.1 6 2.1
64 (56-72) 7.0 (3.8-11.5) 868 (56.0) 383 (24.7) 5.0 6 1.7 21.3 6 4.0 52.5 6 14.3 3.8 6 0.4 147 6 34 10.7 6 1.3 1.44 6 0.30 0.95 6 0.18 26.9 6 2.8 33.3 6 2.2
65 (57-74) 6.7 (3.4-11.7) 2,984 (62.1) 1,173 (24.4) 4.8 6 1.7 21.3 6 3.5 49.0 6 13.7 3.7 6 0.4 151 6 34 10.6 6 1.2 1.44 6 0.29 0.91 6 0.17 27.5 6 3.1 33.9 6 2.1
66 (58-74) 6.3 (3.2-11.8) 3,743 (62.1) 1,519 (25.2) 4.6 6 1.7 21.3 6 3.4 47.1 6 13.5 3.7 6 0.4 155 6 35 10.4 6 1.2 1.43 6 0.29 0.87 6 0.16 28.0 6 3.4 34.2 6 2.0
67 (58-75) 6.3 (3.3-11.8) 1,694 (61.4) 768 (27.9) 4.4 6 1.7 21.1 6 3.3 44.8 6 13.3 3.7 6 0.5 157 6 36 10.2 6 1.2 1.43 6 0.29 0.83 6 0.17 28.6 6 3.6 34.6 6 2.0
6,292 (41.5) 5,049 (33.4) 1,098 (7.3) 504 (3.3) 2,189 (14.5)
586 (37.8) 570 (36.8) 103 (6.6) 46 (3.0) 245 (15.8)
2,035 (42.4) 1,692 (35.2) 327 (6.8) 139 (2.9) 609 (12.7)
2,531 (42.0) 1,937 (32.1) 444 (7.4) 215 (3.6) 896 (14.9)
1,140 (41.4) 850 (30.8) 224 (8.1) 104 (3.8) 439 (15.9)
Predialysis acid-base parameters pH Bicarbonate, mEq/L PCO2, mm Hg
7.353 6 0.048 20.5 6 2.9 38.1 6 4.7
7.267 6 0.027 17.6 6 2.5 40.0 6 6.0
7.323 6 0.014 19.5 6 2.3 38.8 6 4.5
7.368 6 0.014 21.1 6 2.4 37.9 6 4.3
7.421 6 0.024 22.9 6 2.8 36.4 6 4.4
,0.001 ,0.001 ,0.001
Postdialysis acid-base parameters pH Bicarbonate, mEq/L PCO2, mm Hg
7.444 6 0.047 25.1 6 2.8 37.9 6 4.5
7.398 6 0.046 23.0 6 2.5 38.6 6 5.1
7.429 6 0.041 24.3 6 2.5 38.0 6 4.6
7.453 6 0.039 25.5 6 2.6 37.8 6 4.4
7.477 6 0.044 26.7 6 2.9 37.4 6 4.5
,0.001 ,0.001 ,0.001
Characteristics
Age, y Dialysis vintage, y Male sex History of CVD Weight gain ratio, % BMI, kg/m2 Ca 3 P product, mg2/dL2 Serum albumin, g/dL Serum total cholesterol, mg/dL Hemoglobin, g/dL Single-pool Kt/V nPCR, g/kg/d Dialysate bicarbonate, mEq/L Dialysate total alkali, mEq/L Cause of kidney failure Glomerulonephritis Diabetes Nephrosclerosis Polycystic kidney Other
P
,0.001 ,0.001 ,0.001 ,0.001 ,0.001 0.2 ,0.001 ,0.001 ,0.001 ,0.001 0.3 ,0.001 ,0.001 ,0.001 ,0.001
Note: Values for categorical variables are given as number (percentage); values for continuous variables, as mean 6 standard deviation or median [interquartile range]. P value for comparison between pH quartile groups using analysis of variance or c2 test. Conversion factors for units: serum calcium in mg/dL to mmol/L, 30.2495; serum phosphate in mg/dL to mmol/L, 30.3229; serum total cholesterol in mg/dL to mmol/L, 30.02586. PCO2 was computed from pH and bicarbonate values using the Henderson equation. Abbreviations: BMI, body mass index; Ca, calcium; Ca 3 P, calcium-phosphorus product; CVD, cardiovascular disease; nPCR, normalized protein catabolic rate.
parameters. We defined higher levels of acid-base parameters as follows: predialysis pH $ 7.40, predialysis bicarbonate level $ 26.0 mEq/L, postdialysis pH $ 7.50, and postdialysis bicarbonate level $ 28.0 mEq/L. Odds ratios were estimated with their 95% confidence intervals (CIs). The model included age, sex, dialysis vintage, history of CV disease, diabetes, dialysate bicarbonate concentration, weight gain ratio, body mass index, serum albumin level, hemoglobin level, Kt/V, and nPCR. For the association between each acid-base parameter and outcomes, we conducted Cox proportional hazard analysis for all-cause and CV mortality. Because acid-base parameters show strong multicollinearity,10 the analysis was performed based on 4 models for each acid-base parameter and 2 models for pH and bicarbonate level pre-post differences (difference between pre- and postdialysis values). We also examined a model including the association between narrow categories of acid-base parameters and outcomes by dividing patients into quartiles of predialysis pH (,7.30, 7.30-7.34, 7.35-7.39, or $7.40), predialysis bicarbonate level (,18.0, 18.021.9, 22.0-25.9, or $26.0 mEq/L), postdialysis pH (,7.40, 7.407.44, 7.45-7.49, or $7.50), and postdialysis bicarbonate level (,24.0, 24.0-25.9, 26.0-27.9, or $28.0 mEq/L). All models were adjusted for the full variables (age, sex, dialysis vintage, history of CV disease, diabetes, dialysate bicarbonate concentration, weight gain ratio, body mass index, calcium-phosphorus product, serum Am J Kidney Dis. 2015;-(-):---
albumin level, serum total cholesterol level, hemoglobin level, Kt/ V, and nPCR). Hazard ratios (HRs) for all-cause and CV mortality with their 95% CIs were estimated. We additionally used a cubic spline model to explore the linearity between acid-base parameters and outcomes, and Kaplan-Meier analysis, for all-cause and CV mortality according to 3 categories of predialysis pH (,7.35, 7.357.39, or $7.40) and predialysis bicarbonate level (,22.0, 22.025.9, or $26.0 mEq/L). All statistical analyses were performed using IBM SPSS software (version 17.0; IBM Japan).
RESULTS Baseline Characteristics Baseline characteristics of patients based on predialysis pH are listed in Table 1. Median age was 66 years, median dialysis vintage was 6.5 years, 61.4% were men, 25.4% had a history of CV disease, and 33.4% had diabetes. The most common cause of kidney failure was glomerulonephritis. Patients with a higher predialysis pH had a higher frequency of CV diseases and lower frequency of diabetes and showed 3
Yamamoto et al
lower weight gain ratio, calcium-phosphorus product, hemoglobin, and nPCR values than patients with a lower predialysis pH. Higher predialysis pH was associated with higher dialysate bicarbonate and total alkali levels. Baseline characteristics stratified by predialysis serum bicarbonate level are listed in Table S1 (provided as online supplementary material). The same trends as those for patients categorized according to predialysis pH were observed, except that higher predialysis bicarbonate level was associated with higher frequency of diabetes.
dialysate bicarbonate concentration of 35 versus 30 mEq/L was associated with significantly greater levels of the following acid-base parameters (Table 3): predialysis pH (by 0.018 units), predialysis bicarbonate level (by 1.7 mEq/L), postdialysis pH (by 0.017 units), postdialysis bicarbonate level (by 2.4 mEq/L), and bicarbonate pre-post difference (by 0.6 mEq/L). There was no statistically significant difference in pH pre-post difference between the 2 dialysate concentrations. Outcome Data
Factors Associated With Acid-Base Parameters Results of multivariable logistic regression analysis to determine factors associated with higher levels of acid-base parameters are listed in Table 2. Odds of higher predialysis pH and predialysis bicarbonate level were significantly increased by greater dialysate bicarbonate concentration; however, the odds of elevated levels of these predialysis acid-base parameters were significantly decreased by greater weight gain ratio, serum albumin level, hemoglobin level, and nPCR. The presence of diabetes increased the odds of higher predialysis pH but decreased the odds of higher predialysis bicarbonate level. Odds of higher postdialysis pH and postdialysis bicarbonate level were significantly increased by greater dialysate bicarbonate concentration and Kt/V; however, the odds of elevated levels of these postdialysis acid-base parameters were significantly decreased by greater weight gain ratio and hemoglobin level and absence of CV disease history. Increments in nPCR significantly increased the odds of higher postdialysis pH, but significantly decreased the odds of higher postdialysis bicarbonate level. With regard to the relationship between dialysate bicarbonate concentration and acid-base parameters,
During the 1-year follow-up period, all-cause mortality occurred in 1,042 patients, including heart failure in 211, infection in 186, malignancy in 113, stroke in 109, and other causes in 423. CV mortality occurred in 408 patients, including heart failure in 211, stroke in 109, sudden death in 51, and ischemic heart diseases in 37. Of 15,132 patients, 26 were censored because of kidney transplantation. We analyzed unadjusted and adjusted associations of acid-base parameters with risk of all-cause and CV mortality (Table 4). Unadjusted analysis revealed significant increases in the HR of all-cause mortality with each 0.1-unit increment in predialysis pH and bicarbonate level and significant decreases in HR with each 0.1-unit increment in postdialysis pH and pH prepost difference and each 1-mEq/L increase in bicarbonate level pre-post difference. After full adjustment, the associations remained significant for predialysis pH (HR for 0.1-unit increase, 1.18; 95% CI, 1.02-1.37) and pH pre-post difference (HR for 0.1-unit increase, 0.79; 95% CI, 0.68-0.92). Although predialysis pH and pH and bicarbonate pre-post differences were associated with CV mortality in the unadjusted analysis, the association disappeared after full adjustment.
Table 2. Logistic Regression Analysis to Determine Factors Associated With Acid-Base Parameters Predialysis
pH $ 7.40 (n 5 2,757)
Age, per 10-y older Men, vs women Dialysis vintage, per 1-y longer History of CVD Diabetes Dialysate bicarbonate, per 1-mEq/L greater Weight gain ratio, per 1% greater BMI, per 1-kg/m2 greater Serum albumin, per 0.1-g/dL greater Hemoglobin, per 1-g/dL greater Single-pool Kt/V, per 0.1-unit greater nPCR, per 0.1-g/kg/d greater
0.97 1.07 1.00 1.10 0.80 1.08 0.93 0.99 0.99 0.85 1.03 0.81
(0.94-1.01) (0.98-1.18) (0.99-1.01) (1.00-1.22) (0.73-0.88) (1.07-1.09) (0.91-0.96) (0.98-1.01) (0.97-0.99) (0.82-0.88) (1.01-1.05) (0.79-0.84)
Postdialysis
Bicarbonate $ 26.0 mEq/L (n 5 534)
0.94 0.90 1.00 1.53 1.42 1.11 0.88 0.88 0.97 0.83 1.02 0.83
(0.87-1.02) (0.74-1.09) (0.99-1.01) (1.27-1.84) (1.17-1.29) (1.08-1.13) (0.84-0.93) (0.85-0.91) (0.95-0.99) (0.77-0.89) (0.99-1.06) (0.78-0.88)
pH $ 7.50 (n 5 2,712)
0.97 1.12 1.00 0.90 0.63 1.13 0.95 1.02 1.00 0.84 1.06 1.07
(0.93-1.01) (1.01-1.23) (0.99-1.01) (0.81-0.99) (0.57-0.70) (1.12-1.14) (0.92-0.97) (1.00-1.03) (0.99-1.01) (0.81-0.87) (1.04-1.08) (1.04-1.10)
Bicarbonate $ 28.0 mEq/L (n 5 2,264)
0.96 1.08 0.99 0.83 0.99 1.25 0.94 0.97 1.03 0.81 1.09 0.90
(0.93-1.01) (0.97-1.20) (0.98-1.00) (0.74-0.93) (0.89-1.11) (1.24-1.27) (0.91-0.97) (0.95-0.98) (1.02-1.04) (0.78-0.84) (1.07-1.11) (0.87-0.93)
Note: Values are given as odds ratio (95% confidence interval). Abbreviations: BMI, body mass index; CVD, cardiovascular disease; nPCR, normalized protein catabolic rate. 4
Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality Table 3. Relationship Between Dialysate Bicarbonate Concentrations of 30 and 35 mEq/L and Acid-Base Parameters Dialysate Bicarbonate Concentration
Acid-Base Parameter
30 mEq/La (n 5 5,659)
35 mEq/La (n 5 1,913)
P
Predialysis pH 7.354 6 0.046 7.372 6 0.047 ,0.001 Bicarbonate, mEq/L 20.5 6 2.8 22.2 6 2.8 ,0.001 Postdialysis pH 7.448 6 0.044 7.465 6 0.048 ,0.001 Bicarbonate, mEq/L 25.3 6 2.4 27.7 6 2.4 ,0.001 pH pre-post difference 0.094 6 0.047 0.093 6 0.056 0.3 Bicarbonate pre-post 4.8 6 2.4 5.4 6 2.5 ,0.001 difference, mEq/L Note: Values are given as mean 6 standard deviation. P value for comparison between 2 groups using t test. a Total alkali concentration was 36 mEq/L.
We estimated mortality rates and adjusted HRs from the Cox regression analysis for all-cause mortality (Table 5) and CV mortality (Table 6) based on categories of acid-base parameters. We observed higher HRs of all-cause mortality in patients with predialysis pH $ 7.40 (HR, 1.36; 95% CI, 1.13-1.65)
and postdialysis pH , 7.40 (HR, 1.22; 95% CI, 1.001.49). As for CV mortality, a higher HR was found in patients with predialysis pH $ 7.40 (HR, 1.34; 95% CI, 1.01-1.79). Unadjusted and adjusted models for mortality are summarized in Fig 2 (all-cause mortality) and Fig 3 (CV mortality). A higher HR of allcause and CV mortality was consistently observed in patients with predialysis pH $ 7.40. The cubic spline model showed a nonlinear relationship between acid-base parameters and all-cause (Fig S1) and CV (Fig S2) mortality. Kaplan-Meier analysis for allcause mortality performed for the 3 categories of predialysis pH and bicarbonate level showed significantly higher cumulative mortality in patients with predialysis pH $ 7.4 and with predialysis bicarbonate level $ 26.0 mEq/L (Fig S3). As for CV mortality, significantly higher cumulative mortality was found in patients with predialysis pH $ 7.4 (Fig S4).
DISCUSSION In this study, predialysis pH $ 7.4 was associated with a significantly elevated 1-year risk of all-cause and CV mortality. However, predialysis bicarbonate level and postdialysis pH and bicarbonate level were not associated with all-cause and CV mortality. These findings suggest that predialysis pH may be the most
Table 4. Hazard Ratios for All-Cause and CV Mortality by Predialysis pH and Bicarbonate, Postdialysis pH and Bicarbonate, pH Difference, and Bicarbonate Difference All-Cause Mortality (n 5 1,042)
CV Mortality (n 5 408)
HR (95% CI)
P
HR (95% CI)
1.44 (1.26-1.64) 1.18 (1.02-1.37)
,0.001 0.02
1.38 (1.12-1.69) 1.22 (0.98-1.52)
0.002 0.07
1.06 (1.04-1.09) 1.00 (0.98-1.03)
,0.001 0.8
1.02 (0.98-1.05) 0.97 (0.94-1.01)
0.3 0.9
0.82 (0.72-0.94) 0.95 (0.82-1.09)
0.003 0.4
0.95 (0.77-1.17) 1.14 (0.92-1.42)
0.6 0.2
0.99 (0.97-1.01) 0.98 (0.96-1.01)
0.3 0.2
0.99 (0.95-1.02) 0.99 (0.96-1.03)
0.3 0.9
pH pre-post difference, per 0.1-unit greater Unadjusted Adjusted
0.57 (0.50-0.65) 0.79 (0.68-0.92)
,0.001 0.002
0.68 (0.56-0.84) 0.93 (0.75-1.17)
,0.001 0.5
Bicarbonate pre-post difference, per 1-mEq/L greater Unadjusted Adjusted
0.91 (0.89-0.93) 0.98 (0.95-1.00)
,0.001 0.09
0.96 (0.92-0.99) 1.03 (0.99-1.07)
0.03 0.1
Predialysis pH, per 0.1-unit greater Unadjusted Adjusted Bicarbonate, per 1-mEq/L greater Unadjusted Adjusted Postdialysis pH, per 0.1-unit greater Unadjusted Adjusted Bicarbonate, per 1-mEq/L greater Unadjusted Adjusted
P
Note: Adjusted model included age, sex, dialysis vintage, history of CV disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviations: CI, confidence interval; CV, cardiovascular; HR, hazard ratio. Am J Kidney Dis. 2015;-(-):---
5
Yamamoto et al Table 5. Mortality Rates and Hazard Ratios for All-Cause Mortality by 4 Categories of Predialysis and Postdialysis pH and Bicarbonate
Predialysis pH ,7.30 7.30-7.34 7.35-7.39 $7.40 Bicarbonate ,18.0 mEq/L 18.0-21.9 mEq/L 22.0-25.9 mEq/L $26.0 mEq/L Postdialysis pH ,7.40 7.40-7.44 7.45-7.49 $7.50 Bicarbonate ,24.0 mEq/L 24.0-25.9 mEq/L 26.0-27.9 mEq/L $28.0 mEq/L
Mortality Rate (95% CI)
P
No. of Deaths
Person-Years
HR (95% CI)
94 293 389 266
1,503.0 4,681.3 5,847.6 2,626.8
6.3 6.3 6.7 10.1
(5.0-7.5) (5.6-7.0) (6.0-7.3) (9.0-11.3)
1.07 1.00 0.98 1.36
(0.83-1.37) (reference) (0.83-1.16) (1.13-1.65)
157 504 314 67
2,648.8 7,633.8 3,877.7 498.5
5.9 6.6 8.1 13.5
(5.0-6.8) (6.0-7.2) (7.2-9.0) (10.5-16.5)
1.08 1.00 1.04 1.31
(0.89-1.31) (reference) (0.88-1.21) (0.98-1.77)
185 313 366 178
2,026.7 4,835.0 5,170.6 2,626.5
9.1 6.5 7.1 6.8
(7.7-10.4) (5.8-7.2) (6.4-7.8) (5.8-7.7)
1.22 1.02 1.00 1.16
(1.00-1.49) (0.87-1.21) (reference) (0.95-1.42)
0.05 0.8
361 287 254 140
4,920.5 4,199.3 3,354.1 2,193.8
7.3 6.8 7.6 6.4
(6.6-8.1) (6.1-7.6) (6.7-8.5) (5.4-7.4)
1.05 1.08 1.00 0.96
(0.89-1.24) (0.90-1.30) (reference) (0.76-1.20)
0.5 0.4
0.6 0.8 0.002 0.4 0.6 0.07
0.1
0.6
Note: Patients were divided into quartiles of pre- and postdialysis pH and bicarbonate level as shown. All models were adjusted for age, sex, dialysis vintage, history of cardiovascular disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviations: CI, confidence interval; HR, hazard ratio.
Table 6. Mortality Rates and Hazard Ratios for Cardiovascular Mortality by 4 Categories of Predialysis and Postdialysis pH and Bicarbonate
Predialysis pH ,7.30 7.30-7.34 7.35-7.39 $7.40 Bicarbonate ,18.0 mEq/L 18.0-21.9 mEq/L 22.0-25.9 mEq/L $26.0 mEq/L Postdialysis pH ,7.40 7.40-7.44 7.45-7.49 $7.50 Bicarbonate ,24.0 mEq/L 24.0-25.9 mEq/L 26.0-27.9 mEq/L $28.0 mEq/L
Mortality Rate (95% CI)
Hazard Ratio (95% CI)
P
No. of Deaths
Person-Years
42 112 155 99
1,529.7 4,753.0 5,951.4 2,708.5
2.7 2.4 2.6 3.7
(1.9-3.6) (1.9-2.8) (2.2-3.0) (2.9-4.4)
1.21 1.00 1.05 1.34
(0.84-1.75) (reference) (0.81-1.35) (1.01-1.79)
69 209 112 18
2,691.5 7,765.1 3,963.0 522.9
2.6 2.7 2.8 3.4
(2.0-3.2) (2.3-3.1) (2.3-3.3) (1.9-5.0)
1.10 1.00 0.90 0.86
(0.83-1.47) (reference) (0.70-1.14) (0.52-1.43)
62 144 125 77
2,083.2 5,268.2 4,915.8 2,675.4
3.0 2.7 2.5 2.9
(2.2-3.7) (2.3-3.2) (2.1-3.0) (2.2-3.5)
1.00 1.04 1.00 1.29
(0.73-1.36) (0.81-1.34) (reference) (0.96-1.72)
0.9 0.7
141 112 104 51
5,022.5 4,279.2 3,404.0 2,236.8
2.8 2.6 3.1 2.3
(2.4-3.3) (2.1-3.1) (2.5-3.6) (1.7-2.9)
1.02 1.14 1.00 0.91
(0.79-1.32) (0.87-1.51) (reference) (0.65-1.28)
0.8 0.3
0.3 0.7 0.05 0.5 0.3 0.5
0.08
0.5
Note: Patients were divided into quartiles of pre- and postdialysis pH and bicarbonate level as shown. All models were adjusted for age, sex, dialysis vintage, history of cardiovascular disease, diabetes, weight gain ratio, body mass index, calcium-phosphorus product, serum albumin level, total cholesterol level, hemoglobin level, Kt/V, and normalized protein catabolic rate. Abbreviation: CI, confidence interval. 6
Am J Kidney Dis. 2015;-(-):---
Acid-Base Parameter and Mortality A Predialysis pH
P-value 0.9 0.6
< 7.30
B Predialysis
Reference
18.0-21.9
7.35-7.39
0.4 0.8
22.0-25.9
