Abstract
The mortality rate associated with chronic kidney disease (CKD) has been steadily increasing, with disturbances in acid–base balance being a significant factor exacerbating the risk of mortality. Our objective was to evaluate whether the albumin-corrected anion gap (ACAG) status, which reflects acid–base balance, could be used as a mortality risk biomarker for the CKD population in the United States. We conducted a cross-sectional study utilizing the National Health and Nutrition Examination Survey data, collected from 1999 to 2018. Kaplan–Meier curves, weighted Cox regression, restricted cubic spline and subgroup analyses were employed to examine the relation of ACAG with all-cause and cardiovascular mortality in CKD participants. The analysis included a total of 6776 participants. We found a positive relation of ACAG to the mortality risk among CKD participants (all-cause mortality: HR = 1.10, 95% CI = 1.08–1.13, P < .01; cardiovascular mortality: HR = 1.09, 95% CI = 1.04–1.14, P < .01). Subgroup analyses revealed significant interactions between smoking and drinking with regard to the relation between ACAG and mortality in CKD patients. Our research indicates that higher levels of ACAG are related to unfavorable outcomes in CKD. Future research should further explore the role of ACAG and acid–base balance in mortality among CKD patients, as well as investigate potential intervention strategies.
Keywords: ACAG, all-cause mortality, cardiovascular mortality, CKD, cross-sectional study, kidney function, NHANES
1. Introduction
Chronic kidney disease (CKD) is characterized by a progressive decline in kidney fsunction. Epidemiological studies indicate that the global prevalence of CKD is approximately 13.4%,[1] with over 25 million adults affected in the United States alone.[2] Despite significant advancements made by clinicians in delaying the progression of CKD and managing its complications, around 120 million individuals worldwide continue to die from this disease each year,[3] with mortality rates on the rise. It is predicted that by the year 2040, CKD will have ascended to become the fifth most prevalent cause of death on a global scale.[4] Notably, cardiovascular disease (CVD) has been established as a significant risk factor for mortality in patients with CKD.[5] This comorbidity presents a substantial socioeconomic burden, as healthcare expenditures for patients with both CKD and CVD are markedly higher than for those with a single disease.[6] Although existing studies have identified biomarkers that reflect the cardiovascular mortality risk in CKD patients,[7] the available evidence remains insufficient. Consequently, some researchers have proposed the establishment of a subspecialty in cardiovascular nephrology to optimize diagnostic and therapeutic strategies.[8]
The anion gap (AG) is an important indicator of unmeasured anion concentration, calculated as the difference between measured serum cations and anions.[9] It plays a crucial role in the assessment and classification of metabolic acidosis[10] and is a readily obtainable, widely applicable clinical parameter with a long history of extensive use spanning several decades.[11] The modified metric, albumin-corrected anion gap (ACAG), significantly enhances the accuracy of results by incorporating the concentration of the primary negatively charged protein – albumin.[12] Current research has shown that elevated ACAG is closely associated with metabolic acidosis and inflammatory states, and it can predict adverse outcomes in various diseases.[13,14] Although a potential association between ACAG and CKD has been suggested,[15] large-scale studies investigating its value in the prediction of mortality from all causes and cardiovascular mortality in CKD populations remain relatively limited. This study utilizes the National Health and Nutrition Examination Survey (NHANES) database to systematically investigate the correlation between ACAG and all-cause mortality in CKD patients, aiming to provide a new biomarker basis for improving clinical outcomes in this population, while also offering scientific references for healthcare providers managing patients with CKD.
2. Materials and methods
2.1. Study design
In this retrospective cohort study, we used published data from the NHANES from 1999 to 2018. This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.[16] Data collection for NHANES participants occurred in specially designed Mobile Examination Centers (MECs) across the United States. NHANES is a continuous, national representative survey program that collects data every 2 years to determine the public’s health and nutrition status, employing a complicated multistage probability sampling design along with specific sample weights tailored to different demographic groups.This program is administered by the National Center for Health Statistics of the Centers for Disease Control and Prevention. The data collection cycle includes self-reported standard biochemical markers, blood cell count metrics, demographic characteristics of participants, and disease-related information. The NHANES study protocol has been approved by the NCHS Institutional Review Board. As this research is based on publicly available de-identified data and has waived the requirement for informed consent, ethical approval and consent are not required. These surveys are carried out in compliance with local legislation and local institutional provisions. For more details, please visit https://wwwn.cdc.gov/nchs/nhanes/Default.aspx.
Participants meeting the criteria for CKD were included in the study. CKD is characterized by an estimated glomerular filtration rate (eGFR) of <60 mL/min/1.73 m2, albuminuria levels of 30 mg/g or higher, or the presence of both criteria concurrently.[2] The formula for eGFR is as follows: eGFR = 175 × standardized serum creatinine (Scr)−1.154 × age−0.203 × 1.212 [if Black] × 0.742 [if female].[17] Albuminuria is calculated using the formula: urinary albumin/urinary creatinine ratio. We first excluded 38,320 participants with missing eGFR and albuminuria data, and based on the criteria for determining CKD, we excluded a further 52,156 participants with non-CKD. From the initial cohort of 10,861 CKD participants, we excluded 21 individuals with missing or incomplete data on the ACAG, 1450 participants without mortality data, and 2614 participants lacking covariates data. Ultimately, the total number of participants included in the study was 6776 (Fig. 1).
Figure 1.
Flow chart of study participants. ACAG = albumin-corrected anion gap, CKD = chronic kidney disease, NHANES = National Health and Nutrition Examination Survey.
2.2. Measurement of ACAG
The ACAG was calculated using data from NHANES. The AG was determined using the following formula: AG (mmol/L) = [Sodium (mmol/L) + Potassium (mmol/L)] − [Chloride (mmol/L) + Bicarbonate (mmol/L)]. ACAG was assessed using the formula: ACAG (mmol/L) = [4.4 − Albumin (g/dL)] × 2.5 + AG.[18] For detailed information regarding the specific analytical instruments and laboratory methods, please refer to the NHANES Laboratory Procedures Manual. To further investigate the interaction between hyperuricemia and ACAG, participants were divided into 4 groups according to ACAG levels: Q1 (≤14.39), Q2 (14.40–15.85), Q3 (15.86–17.60), and Q4 (≥17.61).
2.3. Measurement of outcomes
In this study, we tracked the survival status of participants by linking them to the National Death Index database. The main outcome was all-cause mortality, while the secondary outcome was cardiovascular mortality, as determined by International Classification of Diseases, Tenth Revision (ICD-10) cause-of-death coding. Specifically, CVD-related mortality events were considered to be mortality resulting from heart disease (including ICD-10 codes I00–I09, I11, I13, I20–I51) and cerebrovascular disease (ICD-10 codes I60–I69).[19]
2.4. Covariates
The confounding variables included in this study were based on prior research.[14,20,21] These variables encompassed demographic information (age, sex, race, marital status, education level, and poverty–income ratio), self-reported lifestyle factors (smoking and alcohol consumption), and clinically determined measurements (body mass index [BMI]) as well as disease status (hypertension, diabetes, CVD, malignancies and cancer). Participants were categorized according to their educational attainment as follows: less than high school, high school or general educational development, and more than high school. Marital status was classified into 3 categories: married and living with partner, widowed and divorced and separated, and never married. Racial categories followed the classification method used by NHANES (Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, other race). Smoking status was determined by the criterion of having smoked at least 100 cigarettes in one’s lifetime. Alcohol consumption was assessed based on having consumed at least 12 alcoholic beverages in the past year. CVD was classified based on 5 self-reported subsets: stroke, congestive heart failure, heart attack, angina and coronary heart disease; participants were categorized as having CVD if they had been diagnosed with one of these conditions by a physician or other medical professional.[22] Disease status was established based on information provided by a physician or other healthcare professional.
2.5. Statistical analysis
Data analysis was performed in accordance with established standards and recommendations for survey weights pertaining to the NHANES data. For the sampling weight calculation for 1999 to 2002, the formula used was: Sample 4-Year MEC Exam Weight (WTMEC4YR) 2/10. For the period from 2003 to 2018, the sampling weight calculation formula was: Sample 2-Year MEC Exam Weight (WTMEC2YR)/2. In alignment with the requirements of complex survey design, all analyses incorporated sampling weights to generate population-level estimates. Weighted means ± standardized deviations were used for continuous variables, whereas categorical variables were presented as frequency counts with adjusted proportions. Comparative analyses utilized both parametric (Student t test) and nonparametric methods (Mann–Whitney U test) for continuous measures, and χ2 tests for categorical comparisons.
The Kaplan–Meier survival curve was employed to illustrate mortality rates and evaluate survival probabilities. The relationships between ACAG and mortality in CKD patients were investigated using a weighted multivariable Cox proportional hazards model. A range of confounding factors was accounted for in the construction of multivariable models. Model 1 served as the baseline crude model devoid of any covariates modification. Model 2 modified sociodemographic characteristics, which included age, race, sex, education level, marital status, and PIR. Model 3 further modified lifestyle factors (smoking and drinking) and BMI. Model 4 included additional adjustments for health conditions such as hypertension, diabetes, CVD, and cancer or malignancies. Furthermore, a weighted restricted cubic spline investigation was undertaken, utilizing the variables adjusted in model 4, to evaluate the potentially relation of ACAG and mortality.
Subgroup analyses were weighted in order to assess the reliability of the correlations of ACAG with mortality among CKD patients. Interaction tests were conducted to evaluate variability in correlations across subgroups. In addition, among the initially identified 10,861 participants with CKD, 2614 individuals had missing covariates data. To access the impact of the missing covariates data on our study results, we employed multiple imputation methods to create 5 imputed datasets, addressing the missing values of all covariates included in model 4. We then performed the Cox proportional hazards regression analysis on the imputed datasets to determine the robustness of the results.
All data were analyzed using the software program Free Statistics (version 2.1.1; Beijing FreeClinical Medical Technology Co., Ltd., Beijing, China). Statistically, this result was considered to be significant at the 0.05 level (2-tailed P-value).
3. Results
3.1. Baseline characteristics
Table 1 displays the baseline characteristics of the research population, which have been stratified according to quartiles of ACAG. The study included 6776 participants with CKD, aged 62 years on average, with women comprising 58.29% of the total participants. Within the cohort, as ACAG levels increased, participants were subject to a higher probability of being older females, smoking, and consuming alcohol, and to having lower educational attainment and PIR. Furthermore, in comparison with the lowermost quartile of ACAG, the uppermost quartile was found to be associated with a greater BMI and a higher prevalence of hypertension, diabetes, cancer and malignancies, CVD, and hyperuricemia.
Table 1.
Weighted demographic characteristics of all participants based on ACAG quartile grouping.
| Characteristic | ACAG | |||||
|---|---|---|---|---|---|---|
| Overall, N = 6776 | Q1 (≤14.39), N = 1694 | Q2 (14.40–15.85), N = 1689 | Q3 (15.86–17.60), N = 1695 | Q4 (≥17.61), N = 1698 | P-value | |
| Age, yr | 62.00 (47.00, 74.00) | 58.00 (44.00, 71.00) | 64.00 (49.00, 75.00) | 62.00 (49.00, 74.00) | 64.00 (48.00, 75.00) | <.001 |
| Sex, n (%) | .69 | |||||
| Male | 3173 (41.71%) | 779 (41.92%) | 818 (43.24%) | 771 (40.93%) | 805 (40.80%) | |
| Female | 3603 (58.29%) | 915 (58.08%) | 871 (56.76%) | 924 (59.07%) | 893 (59.20%) | |
| Race, n (%) | .22 | |||||
| Mexican American | 1000 (6.16%) | 255 (7.32%) | 258 (6.54%) | 221 (5.00%) | 266 (5.79%) | |
| Other Hispanic | 456 (4.47%) | 156 (5.47%) | 113 (4.04%) | 103 (3.75%) | 84 (4.61%) | |
| Non-Hispanic White | 3562 (73.68%) | 818 (70.60%) | 898 (74.10%) | 945 (75.95%) | 901 (74.13%) | |
| Non-Hispanic Black | 1345 (10.44%) | 361 (11.12%) | 311 (9.81%) | 326 (10.26%) | 347 (10.54%) | |
| Other race | 413 (5.25%) | 104 (5.49%) | 109 (5.50%) | 100 (5.05%) | 100 (4.94%) | |
| Education level, n (%) | .13 | |||||
| Less than high school | 2163 (21.48%) | 505 (18.57%) | 529 (21.78%) | 554 (22.83%) | 575 (22.80%) | |
| High school or GED | 1613 (25.25%) | 412 (24.58%) | 405 (25.11%) | 389 (25.34%) | 407 (25.97%) | |
| More than high school | 3000 (53.27%) | 777 (56.85%) | 755 (53.10%) | 752 (51.83%) | 716 (51.23%) | |
| Marital status, n (%) | <.001 | |||||
| Married and living with partner | 3809 (60.64%) | 987 (64.72%) | 967 (61.15%) | 943 (59.92%) | 912 (56.68%) | |
| Widowed and divorced and separated | 2338 (30.10%) | 517 (24.56%) | 583 (30.53%) | 599 (31.12%) | 639 (34.34%) | |
| Never married | 629 (9.26%) | 190 (10.72%) | 139 (8.33%) | 153 (8.96%) | 147 (8.99%) | |
| PIR | 2.56 (1.36, 4.64) | 2.82 (1.45, 4.94) | 2.60 (1.37, 4.74) | 2.49 (1.36, 4.64) | 2.41 (1.25, 4.31) | .051 |
| BMI, kg/m2 | 28.61 (24.80, 33.60) | 27.65 (23.77, 31.99) | 28.17 (24.71, 32.90) | 29.21 (25.50, 33.86) | 29.90 (25.08, 35.70) | <.001 |
| Smoking, n (%) | <.001 | |||||
| No | 3358 (50.44%) | 928 (56.63%) | 816 (48.59%) | 833 (49.69%) | 781 (46.65%) | |
| Yes | 3418 (49.56%) | 766 (43.37%) | 873 (51.41%) | 862 (50.31%) | 917 (53.35%) | |
| Drinking, n (%) | .68 | |||||
| No | 2259 (29.70%) | 591 (31.02%) | 573 (29.72%) | 574 (29.62%) | 521 (28.40%) | |
| Yes | 4517 (70.30%) | 1103 (68.98%) | 1116 (70.28%) | 1121 (70.38%) | 1177 (71.60%) | |
| Hypertension, n (%) | <.001 | |||||
| No | 2695 (44.28%) | 797 (52.99%) | 694 (46.29%) | 624 (41.88%) | 580 (35.79%) | |
| Yes | 4081 (55.72%) | 897 (47.01%) | 995 (53.71%) | 1071 (58.12%) | 1118 (64.21%) | |
| Diabetes, n (%) | <.001 | |||||
| No | 4858 (76.46%) | 1353 (85.62%) | 1260 (79.58%) | 1184 (74.41%) | 1061 (66.07%) | |
| Yes | 1918 (23.54%) | 341 (14.38%) | 429 (20.42%) | 511 (25.59%) | 637 (33.93%) | |
| CVD, n (%) | <.001 | |||||
| No | 5075 (78.66%) | 1354 (84.61%) | 1268 (78.37%) | 1244 (77.29%) | 1209 (74.23%) | |
| Yes | 1701 (21.34%) | 340 (15.39%) | 421 (21.63%) | 451 (22.71%) | 489 (25.77%) | |
| Cancer and malignancy, n (%) | .008 | |||||
| No | 5665 (83.11%) | 1449 (85.56%) | 1431 (84.09%) | 1411 (82.38%) | 1374 (80.37%) | |
| Yes | 1111 (16.89%) | 245 (14.44%) | 258 (15.91%) | 284 (17.62%) | 324 (19.63%) | |
| All-cause mortality, n (%) | <.001 | |||||
| No | 4295 (70.75%) | 1276 (81.53%) | 1099 (71.45%) | 999 (67.52%) | 921 (62.26%) | |
| Yes | 2481 (29.25%) | 418 (18.47%) | 590 (28.55%) | 696 (32.48%) | 777 (37.74%) | |
| Cardiovascular mortality, n (%) | <.001 | |||||
| No | 6032 (91.16%) | 1564 (94.54%) | 1499 (90.97%) | 1488 (89.41%) | 1481 (89.67%) | |
| Yes | 744 (8.84%) | 130 (5.46%) | 190 (9.03%) | 207 (10.59%) | 217 (10.33%) |
ACAG = albumin-corrected anion gap, BMI = body mass index, CVD = cardiovascular disease, GED = general educational development, PIR = poverty–income ratio.
3.2. Relationship between ACAG quartiles and mortality
Among the participants included in the study, there were 2481 deaths (29.25%), of which 744 individuals (8.84%) died from CVD. Kaplan–Meier curve analysis indicated a significant association between increasing levels of ACAG and mortality risk (P < .001; Fig. 2). A weighted Cox regression model was employed to assess the effect of ACAG on mortality among individuals suffering from CKD, with ACAG considered either as a continuous or categorical variable (Table 2). When ACAG was treated as a continuous variable in the analysis, there was a positive correlation with mortality in the original model (model 1; all-cause mortality: HR = 1.14, 95% CI = 1.11–1.17, P < .001; cardiovascular mortality: HR = 1.13, 95% CI = 1.08–1.18, P < .001). Following the adjustment of confounding factors in model 4, these associations remained significant (all-cause mortality: HR = 1.10, 95% CI = 1.08–1.13, P < .001; cardiovascular mortality: HR = 1.09, 95% CI = 1.04–1.14, P < .001). The categorization of ACAG by quartiles revealed that elevated levels of ACAG (Q2, Q3, Q4) were correlated to an augmented mortality risk among CKD patients in comparison with the reference quartile (Q1). In the model that has undergone full adjustment (model 4), elevated ACAG levels remained significantly associated with mortality in CKD participants. The HRs for all-cause mortality in Q2, Q3, and Q4 were HR = 1.27 (95% CI = 1.10–1.47, P = .001), HR = 1.56 (95% CI = 1.34–1.83, P < .001), and HR = 1.85 (95% CI = 1.57–2.16, P < .001), respectively. The HRs for cardiovascular mortality in Q2, Q3, and Q4 were HR = 1.32 (95% CI = 1.03–1.69, P = .026), HR = 1.69 (95% CI = 1.27–2.25, P < .001), and HR = 1.67 (95% CI = 1.21–2.31, P = .002), respectively. A significant quartile trend was observed across all models (P < .001). The result of the restricted cubic spline study demonstrated a linear correlation linking ACAG with both all-cause and cardiovascular mortality (Fig. 3).
Figure 2.
Kaplan–Meier survival curve for mortality by ACAG quartiles. Kaplan–Meier survival curve for all-cause mortality (A) and CVD mortality (B) by ACAG. ACAG = albumin-corrected anion gap, CVD = cardiovascular disease.
Table 2.
Weighted association between ACAG and mortality.
| Model 1 | Model 2 | Model 3 | Model 4 | |||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | |
| All-cause mortality | ||||||||
| ACAG* | 1.14 (1.11, 1.17) | <.001 | 1.12 (1.09, 1.15) | <.001 | 1.12 (1.09, 1.14) | <.001 | 1.1 (1.08, 1.13) | <.001 |
| ACAG** | ||||||||
| Q1 (≤14.39) | Reference | Reference | Reference | Reference | ||||
| Q2 (14.40–15.85) | 1.59 (1.37, 1.84) | <.001 | 1.31 (1.14, 1.51) | <.001 | 1.3 (1.13, 1.49) | <.001 | 1.27 (1.10, 1.47) | .001 |
| Q3 (15.86–17.60) | 1.88 (1.63, 2.18) | <.001 | 1.64 (1.40, 1.93) | <.001 | 1.63 (1.40, 1.91) | <.001 | 1.56 (1.34, 1.83) | <.001 |
| Q4 (≥17.61) | 2.41 (2.07, 2.81) | <.001 | 2.06 (1.77, 2.41) | <.001 | 2.02 (1.73, 2.36) | <.001 | 1.85 (1.57, 2.16) | <.001 |
| P for trend | <.001 | <.001 | <.001 | <.001 | ||||
| Cardiovascular mortality | ||||||||
| ACAG* | 1.13 (1.08, 1.18) | <.001 | 1.11 (1.06, 1.16) | <.001 | 1.11 (1.06, 1.16) | <.001 | 1.09 (1.04, 1.14) | <.001 |
| ACAG** | ||||||||
| Q1 (≤14.39) | Reference | Reference | Reference | Reference | ||||
| Q2 (14.40–15.85) | 1.7 (1.32, 2.18) | <.001 | 1.39 (1.09, 1.77) | .007 | 1.38 (1.08, 1.76) | .01 | 1.32 (1.03, 1.69) | .026 |
| Q3 (15.86–17.60) | 2.08 (1.58, 2.74) | <.001 | 1.82 (1.37, 2.41) | <.001 | 1.78 (1.34, 2.37) | <.001 | 1.69 (1.27, 2.25) | <.001 |
| Q4 (≥17.61) | 2.26 (1.67, 3.06) | <.001 | 1.93 (1.42, 2.63) | <.001 | 1.86 (1.36, 2.55) | <.001 | 1.67 (1.21, 2.31) | .002 |
| P for trend | <.001 | <.001 | <.001 | <.001 | ||||
Model 1: No covariate was adjusted.
Model 2: Adjust for model 1 + age, sex, race, marital status, education level and PIR.
Model 3: Adjust for model 2 + smoking, drinking and BMI.
Model 4: Adjust for model 3 + hypertension, diabetes, CVD and cancer and malignancy.
ACAG* = ACAG as a continuous variable, ACAG** = ACAG as a categorical variable.
ACAG = albumin-corrected anion gap, BMI = body mass index, CI = confidence interval, CVD = cardiovascular disease, HR = hazard ratio, PIR = poverty–income ratio.
Figure 3.
Weighted relationship between ACAG and mortality. RCS for all-cause mortality (A) and CVD mortality (B) by ACAG. ACAG = albumin-corrected anion gap, CI = confidence interval, CVD = cardiovascular disease, HR = hazard ratio, RCS = restricted cubic spline.
3.3. Subgroup analyses
To assess reliability of the relationship between ACAG and hyperuricemia in specific populations, and to determine possible differences in various subgroups, we undertook a series of subgroup analyses and interaction tests. These were based on sex, race, smoking, drinking, hypertension, diabetes, CVD, as well as cancer. Following adjustment for all potential confounding variables, a positive relation of ACAG to mortality was observed, as depicted in Figure 4. Notably, a statistically substantial interaction was identified between ACAG and all-cause mortality in the smoking and drinking subgroups, while no significant interactions were detected in the subgroups categorized by gender, race, hypertension, diabetes, CVD, or cancer. In the analysis of ACAG concerning cardiovascular mortality, an interaction was also observed in the smoking subgroup, whereas no significant significant interactions were identified in the subgroups classified by gender, race, alcohol consumption, hypertension, diabetes, CVD, or cancer.
Figure 4.
Weighted subgroup analysis for the association between ACAG and mortality. Subgroup analysis for all-cause mortality (A) and CVD mortality (B) by ACAG. ACAG = albumin-corrected anion gap, CI = confidence interval, CVD = cardiovascular disease, HR = hazard ratio.
3.4. Sensitivity analyses
To examine the reliability and stability of the results, we performed 5 multiple imputations for the missing values of the covariates. The newly generated datasets were analyzed separately, and their combined effects were calculated. In the sensitivity analysis, the correlation of the ACAG with mortality risk was maintained (P < .05; Table 3).
Table 3.
Sensitivity analyses.
| Model 1 | Model 2 | Model 3 | Model 4 | |||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | |
| All-cause mortality | ||||||||
| ACAG* | 1.11 (1.1, 1.12) | <.001 | 1.1 (1.09, 1.1) | <.001 | 1.1 (1.09, 1.1) | <.001 | 1.09 (1.08, 1.1) | <.001 |
| ACAG** | ||||||||
| Q1 (≤14.39) | Reference | Reference | Reference | Reference | ||||
| Q2 (14.40–15.85) | 1.45 (1.38, 1.52) | <.001 | 1.29 (1.23, 1.35) | <.001 | 1.28 (1.22, 1.34) | <.001 | 1.27 (1.21, 1.33) | <.001 |
| Q3 (15.86–17.60) | 1.78 (1.7, 1.86) | <.001 | 1.56 (1.49, 1.63) | <.001 | 1.56 (1.49, 1.63) | <.001 | 1.51 (1.44, 1.59) | <.001 |
| Q4 (≥17.61) | 2.17 (2.07, 2.27) | <.001 | 1.91 (1.83, 2) | <.001 | 1.9 (1.82, 1.99) | <.001 | 1.82 (1.74, 1.91) | <.001 |
| P for trend | 1.28 (1.26, 1.3) | <.001 | 1.24 (1.22, 1.25) | <.001 | 1.23 (1.22, 1.25) | <.001 | 1.22 (1.2, 1.23) | <.001 |
| Cardiovascular mortality | ||||||||
| ACAG* | 1.1 (1.09, 1.11) | <.001 | 1.09 (1.07, 1.1) | <.001 | 1.08 (1.07, 1.1) | <.001 | 1.08 (1.07, 1.09) | <.001 |
| ACAG** | ||||||||
| Q1 (≤14.39) | Reference | Reference | Reference | Reference | ||||
| Q2 (14.40–15.85) | 1.57 (1.44, 1.71) | <.001 | 1.39 (1.27, 1.52) | <.001 | 1.38 (1.27, 1.51) | <.001 | 1.35 (1.24, 1.48) | <.001 |
| Q3 (15.86–17.60) | 1.83 (1.68, 1.99) | <.001 | 1.59 (1.46, 1.72) | <.001 | 1.57 (1.45, 1.71) | <.001 | 1.5 (1.37, 1.63) | <.001 |
| Q4 (≥17.61) | 2.07 (1.9, 2.26) | <.001 | 1.81 (1.66, 1.97) | <.001 | 1.78 (1.64, 1.94) | <.001 | 1.72 (1.58, 1.87) | <.001 |
| P for trend | 1.25 (1.22, 1.28) | <.001 | 1.2 (1.17, 1.23) | <.001 | 1.2 (1.17, 1.23) | <.001 | 1.18 (1.15, 1.21) | <.001 |
Model 1: No covariate was adjusted.
Model 2: Adjust for model 1 + age, sex, race, marital status, education level and PIR.
Model 3: Adjust for model 2 + smoking, drinking and BMI.
Model 4: Adjust for model 3 + hypertension, diabetes, CVD and cancer and malignancy.
ACAG = albumin-corrected anion gap, BMI = body mass index, CI = confidence interval, CVD = cardiovascular disease, HR = hazard ratio.
4. Discussion
This study utilized comprehensive and representative data from the NHANES, covering the United States population from 1999 to 2018. The findings indicate a positive correlation between ACAG and the risks of all-cause mortality and cardiovascular mortality in patients with CKD. Proportional hazards regression analysis demonstrated that every unit increase in ACAG is related to a 10% and 9% increase in the all-cause mortality and cardiovascular mortality risk, respectively, among CKD patients. Additionally, patients in the highest quartile (Q4) of ACAG exhibited an 85% and 67% increase in the risks of all-cause mortality and cardiovascular mortality when in comparison to those in the lowest quartile (Q1). Subgroup analyses were conducted to ascertain the validity of these results. These results highlight the significance of ACAG as a potential biomarker for all-cause and cardiovascular mortality in CKD patients, indicating that elevated ACAG levels may be associated with an increased risk of adverse outcomes in this demographic. Therefore, monitoring ACAG levels is crucial for the management of patients with CKD.
As the central regulator of acid–base balance in the body, renal impairment adversely affects both acid excretion and bicarbonate reabsorption capabilities. When the rate of nonvolatile acid production exceeds the compensatory capacity of the kidneys, metabolic acidosis may ensue.[23] ACAG has emerged as a critical biochemical marker for assessing metabolic acidosis, and recent studies have revealed its significant association with adverse outcomes in kidney disease. A cohort study involving 708 acute kidney injury patients undergoing continuous renal replacement therapy demonstrated that patients with elevated ACAG levels were more prone to developing hypoalbuminemia and metabolic acidosis, leading to a significantly increased mortality rate in this subgroup.[24] Further corroborating evidence from a cohort study based on NHANES III data indicated that ACAG effectively reflects serum acid–base parameters even in patients without metabolic acidosis. In particular, elevated ACAG levels in adults with moderate CKD were shown to be independently correlated with an elevated risk of progression to end-stage renal disease.[25] Notably, a retrospective analysis of a randomized controlled trial revealed that patients with high AG levels experienced significantly higher all-cause mortality, likely due to indications of diminished glomerular filtration and tubular reabsorption functions compared to those in the low AG group.[15] This finding aligns with the observed increased mortality rates in CKD patients with high ACAG levels in our study, underscoring the potential of ACAG to identify high-risk individuals and promote early intervention. Despite the well-established pathological link between CKD and CVD, the potential of ACAG to predict cardiovascular mortality risk in CKD patients remains unclear.[26] This study effectively addresses this gap by analyzing a nationally representative, multiethnic cohort of CKD patients from the NHANES database, thus controlling for biases arising from population heterogeneity. Our findings confirm that ACAG serves as an independent risk indicator for assessing cardiovascular mortality risk in patients with CKD.
The kidneys and heart are closely linked in both physiological regulation and pathological processes, with CKD significantly associated with adverse cardiovascular outcomes. Current research attributes the increased cardiovascular risk in CKD patients primarily to endothelial dysfunction,[27] disturbances in bone mineral metabolism,[28] and a state of chronic inflammation.[29] A systematic review highlighted that the risk of cardiovascular mortality in CKD patients escalates exponentially with declining renal function.[30] This phenomenon was corroborated by the 2020 United States Renal Data System report, which indicated a cardiovascular event rate as high as 75.3% in patients with stages 4 to 5 CKD.[31] In addition to traditional risk factors such as hypertension and diabetes, metabolic acidosis – a common pathological alteration in CKD – may exacerbate cardiovascular damage through mechanisms such as inducing myocardial injury[17] and accelerating the progression of atherosclerosis.[14] Although the specific pathways through which ACAG influences cardiovascular outcomes in CKD patients have not been fully elucidated, existing evidence suggests 5 potential pathways of action. First, in response to acid load, CKD patients release compensatory hormones such as angiotensin II and aldosterone to promote acid excretion.[20,32] However, this compensatory mechanism may lead to progressive renal function deterioration and albuminuria, while exacerbating complications such as anemia and bone mineral metabolism abnormalities,[33,34] ultimately increasing cardiovascular mortality risk through multi-organ interactions.[35] Second, beyond indicating acid–base imbalance, ACAG can serve as a biomarker of inflammatory activity.[36,37] The chronic inflammatory state mediated by ACAG not only disrupts immune system function but also creates a vicious cycle through promoting endothelial injury and the progression of atherosclerosis.[14] Third, electrolyte disturbances induced by metabolic acidosis not only accelerate renal damage but may also directly worsen clinical outcomes by interfering with cardiac electrophysiological stability.[38] Fourth, elevated ACAG may reflect an increased dietary acid load, a factor that has been shown to synergistically enhance risks of comorbidities which include, but are not limited to, hypertension and diabetes.[1] Lastly, acidosis may directly jeopardize the cardiovascular system through multiple pathways, including impairing myocardial contractility, inducing life-threatening arrhythmias, and reducing the efficacy of vasoactive medications.[39]
This investigation discloses a substantial interaction effect between ACAG and unfavorable outcomes in the subgroup of CKD patients who smoke and consume alcohol. A meta-analysis involving 65,064 incident CKD cases confirmed that smoking not only independently predicts cardiovascular risk but also exacerbates CKD progression through pathological mechanisms such as chronic endothelial injury, induction of inflammatory responses, and promotion of glomerulosclerosis.[40] Regarding the relationship between alcohol consumption and CKD, existing studies present conflicting conclusions: some evidence suggests that moderate alcohol consumption may reduce the kidney disease risk,[41] while other cohort studies have found no significant association.[42] However, it is noteworthy that reducing alcohol intake has been shown to effectively improve patient outcomes.[43,44] These findings indicate that CKD patients should strictly abstain from smoking and alcohol consumption and adopt healthy behavioral patterns, which holds significant clinical implications for improving disease prognosis.
This study has several significant advantages. First, it identifies the ACAG as an important indicator reflecting the risk of mortality in patients with CKD. Secondly, we employed proportional hazards regression and curve-fitting analysis to ensure the reliability of our findings, thereby further supporting the validity of our conclusions. This was done to minimize the effect of confounding factors. Finally, we utilized a population cohort constructed from the NHANES database, which, considering the sample design and weighting, encompasses diverse samples representing multiple races and age groups. This significantly enhances the generalizability of our results to the adult population in the United States. However, it is essential to acknowledge that this study has certain limitations. First, self-reported variables, such as CVD and cancer, may introduce potential recall bias. Second, because of the cross-sectional character of the investigation, no conclusions could be drawn regarding the causal relationships between ACAG and mortality in CKD patients. Third, the applicability of findings from the NHANES dataset to populations in other regions may be limited due to variations in genetic and environmental factors across different countries and areas. Future research should aim to validate the predictive role of ACAG on mortality in CKD patients through multicenter prospective cohort studies, and explore intervention studies based on dynamic monitoring to elucidate its clinical translational potential.
5. Conclusion
This study confirms that elevated levels of ACAG are significantly and independently associated with all-cause and cardiovascular mortality among CKD patients. It is suggested by these findings that ACAG may serve as an effective biomarker for risk assessment in CKD patients. We recommend incorporating ACAG measurement into the routine risk assessment framework for individuals with CKD. Clinicians should investigate and address potential underlying causes in patients with elevated ACAG levels, while also managing concomitant cardiovascular risk factors more cautiously. Furthermore, we advise CKD patients to refrain from smoking and to consume alcohol in moderation, as our study observed significant interactions between these behaviors and ACAG levels concerning mortality risk. Integrating ACAG into risk stratification not only enhances prognostic accuracy but also facilitates timely and personalized interventions. Future prospective studies are warranted in various clinical settings to validate these associations and elucidate the causal relationships between dynamic changes in ACAG and outcomes, while also assessing the impact of ACAG-based interventions on improving survival rates.
Acknowledgments
We express our gratitude to the participants and staff of the NHANES. We also appreciate the efforts of all the reviewers who took part in the review process.
Author contributions
Conceptualization: Haoyang Yin, Song Gu.
Data curation: Haoyang Yin.
Investigation: Haoyang Yin.
Methodology: Haoyang Yin.
Software: Haoyang Yin.
Validation: Song Gu.
Visualization: Song Gu.
Writing – original draft: Haoyang Yin.
Writing – review & editing: Song Gu.
Abbreviations:
- ACAG
- albumin-corrected anion gap
- AG
- anion gap
- BMI
- body mass index
- CKD
- chronic kidney disease
- CVD
- cardiovascular disease
- eGFR
- estimated glomerular filtration rate
- ICD-10
- International Classification of Diseases, Tenth Revision
- MEC
- Mobile Examination Centers
- NHANES
- National Health and Nutrition Examination Survey
The data collection for NHANES was approved by the Research Ethics Review Board (ERB) of the National Center for Health Statistics (NCHS), and written informed consent was obtained from all participants. The NHANES datasets are publicly available and fully de-identified; therefore, no separate ethical approval was required for this secondary analysis. According to the Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects (China, February 18, 2023, Article 32, Items 1 and 2), research that utilizes only publicly available and anonymized data is exempt from obtaining additional approval from an Institutional Review Board (IRB).
The authors have no funding and conflicts of interest to disclose.
The datasets generated for this study can be found in the CDC National Center for Health Statistics NHANES database at https://wwwn.cdc.gov/nchs/nhanes/Default.aspx.
How to cite this article: Yin H, Gu S. Association between albumin-corrected anion gap status with all-cause and cardiovascular mortality in individuals with chronic kidney disease: A national retrospective cohort study. Medicine 2026;105:7(e47539).
References
- [1].Jha V, Garcia-Garcia G, Iseki K, et al. Chronic kidney disease: global dimension and perspectives. Lancet. 2013;382:260–72. [DOI] [PubMed] [Google Scholar]
- [2].GBD Chronic Kidney Disease Collaboration. Global, regional, and national burden of chronic kidney disease, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2020;395:709–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Kalantar-Zadeh K, Jafar TH, Nitsch D, Neuen BL, Perkovic V. Chronic kidney disease. Lancet. 2021;398:786–802. [DOI] [PubMed] [Google Scholar]
- [4].Foreman KJ, Marquez N, Dolgert A, et al. Forecasting life expectancy, years of life lost, and all-cause and cause-specific mortality for 250 causes of death: reference and alternative scenarios for 2016-40 for 195 countries and territories. Lancet. 2018;392:2052–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Zhu W, Liu Q, Zhang L, Jiao C, Xie H. Associations of combined physical activity and dietary quality with all-cause and cardiovascular disease mortality among US adults with chronic kidney disease. Ren Fail. 2024;46:2437120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Nichols GA, Ustyugova A, Déruaz-Luyet A, O’Keeffe-Rosetti M, Brodovicz KG. Health care costs by type of expenditure across eGFR stages among patients with and without diabetes, cardiovascular disease, and heart failure. J Am Soc Nephrol. 2020;31:1594–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Marques GL, Hayashi S, Bjällmark A, et al. Osteoprotegerin is a marker of cardiovascular mortality in patients with chronic kidney disease stages 3-5. Sci Rep. 2021;11:2473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Díez J, Ortiz A. The need for a cardionephrology subspecialty. Clin Kidney J. 2021;14:1491–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Wang J, Yan X, Wang H, Tang L. Predictive value of anion gap for acute kidney injury progression and outcomes. Med Sci Monit. 2025;31:e947405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Funes S, de Morais HA. A quick reference on high anion gap metabolic acidosis. Vet Clin North Am Small Anim Pract. 2017;47:205–7. [DOI] [PubMed] [Google Scholar]
- [11].Zhang Y, Ma H, Wang R, et al. Correlation between baseline anion gap and early acute kidney injury in patients with acute pancreatitis in the intensive care unit: a single-center retrospective cohort study. PLoS One. 2025;20:e0315386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Chen Y, You MY, Chu L. Association of albumin-corrected anion gap with severe consciousness disorders and outcomes in ischemic stroke: a retrospective MIMIC analysis. Sci Rep. 2024;14:26006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Wang Y, Zhong L, Min J, Lu J, Zhang J, Su J. Albumin corrected anion gap and clinical outcomes in elderly patients with acute kidney injury caused or accompanied by sepsis: a MIMIC-IV retrospective study. Eur J Med Res. 2025;30:11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Wang M, Yang S, Deng J, et al. Unveiling the hidden risks: albumin-corrected anion gap as a superior marker for cardiovascular mortality in type 2 diabetes: insights from a nationally prospective cohort study. Front Endocrinol (Lausanne). 2024;15:1461047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Lee SW, Kim S, Na KY, et al. Serum anion gap predicts all-cause mortality in patients with advanced chronic kidney disease: a retrospective analysis of a randomized controlled study. PLoS One. 2016;11:e0156381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Int J Surg. 2014;12:1495–9.25046131 [Google Scholar]
- [17].Kawasoe S, Kubozono T, Salim AA, et al. Development of a risk prediction score and equation for chronic kidney disease: a retrospective cohort study. Sci Rep. 2023;13:5001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Bai N, Ying T, Li D, Liu A. Relationship between albumin-corrected anion gap and non-alcoholic fatty liver disease: a cross-sectional analysis of NHANES 2017-2018. Front Med (Lausanne). 2025;12:1518540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Zhao E, Xie H, Gao Y, Wen X, Dong B, Zhang C. Association between high sensitivity cardiac troponin and all-cause and cardiovascular mortality in adults at risk of non-alcoholic fatty liver disease: a cohort study. Glob Heart. 2025;20:40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Jiang H, Lan X, Zhou L, Xie X. Association between albumin-corrected anion gap and kidney function in individuals with hypertension - NHANES 2009-2016 cycle. Ren Fail. 2024;46:2416719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Eule CJ, Hu J, Davies KD, Jani A, Pshak T, Lam ET. Clinical features and mutational frequency of renal cell carcinoma from patients undergoing kidney transplant evaluation. Front Oncol. 2025;15:1526545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Lai G, Zhao Y, Yang C, et al. Association of naples prognostic score with cardiovascular disease risk and its longitudinal prognostic impact on mortality in cardiovascular disease patients: evidence from NHANES. Nutr Metab Cardiovasc Dis. 2025;35:103840. [DOI] [PubMed] [Google Scholar]
- [23].Greenberg KI, Lecker SH. Metabolic acidosis. Adv Kidney Dis Health. 2025;32:61–8. [DOI] [PubMed] [Google Scholar]
- [24].Zhong L, Xie B, Ji XW, Yang XH. The association between albumin corrected anion gap and ICU mortality in acute kidney injury patients requiring continuous renal replacement therapy. Intern Emerg Med. 2022;17:2315–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Giri M, Puri A, Huang L, Guo S. Association between albumin corrected anion gap and in-hospital mortality in critically ill patients with chronic obstructive pulmonary disease. Ther Adv Respir Dis. 2025;19:17534666251315352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004;351:1296–305. [DOI] [PubMed] [Google Scholar]
- [27].Xu X, Zhang Y, Chen Y, et al. The effect of short-term remote ischemic preconditioning on endothelial function of patients with chronic kidney disease: a randomized pilot study. Nephrology (Carlton). 2024;29:344–53. [DOI] [PubMed] [Google Scholar]
- [28].Shanahan CM. Mechanisms of vascular calcification in CKD-evidence for premature ageing? Nat Rev Nephrol. 2013;9:661–70. [DOI] [PubMed] [Google Scholar]
- [29].Adu-Amankwaah J, Adzika GK, Adekunle AO, et al. ADAM17, a key player of cardiac inflammation and fibrosis in heart failure development during chronic catecholamine stress. Front Cell Dev Biol. 2021;9:732952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Tonelli M, Wiebe N, Culleton B, et al. Chronic kidney disease and mortality risk: a systematic review. J Am Soc Nephrol. 2006;17:2034–47. [DOI] [PubMed] [Google Scholar]
- [31].Matsushita K, Ballew SH, Wang AY-M, Kalyesubula R, Schaeffner E, Agarwal R. Epidemiology and risk of cardiovascular disease in populations with chronic kidney disease. Nat Rev Nephrol. 2022;18:696–707. [DOI] [PubMed] [Google Scholar]
- [32].Ravikumar NPG, Pao AC, Raphael KL. Acid-mediated kidney injury across the spectrum of metabolic acidosis. Adv Chronic Kidney Dis. 2022;29:406–15. [DOI] [PubMed] [Google Scholar]
- [33].Ishigami J, Grams ME, Naik RP, et al. Hemoglobin, albuminuria, and kidney function in cardiovascular risk: the ARIC (Atherosclerosis Risk in Communities) study. J Am Heart Assoc. 2018;7:e00720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Kidney Disease: Improving Global Outcomes (KDIGO) CKD-MBD Work Group. KDIGO clinical practice guideline for the diagnosis, evaluation, prevention, and treatment of Chronic Kidney Disease-Mineral and Bone Disorder (CKD-MBD). Kidney Int Suppl. 2009;76:S1–130. [DOI] [PubMed] [Google Scholar]
- [35].Li N, Wang Z, Yang X, et al. Development and validation of a cardiovascular disease risk prediction model for patients with non-dialysis-dependent chronic kidney diseases based on the nomogram. Kidney Blood Press Res. 2023;48:7–17. [DOI] [PubMed] [Google Scholar]
- [36].Zhao B, Li Y, Lang X, et al. Increased serum albumin corrected anion gap levels are associated with increased incidence of new-onset HF and poor prognosis in patients with acute myocardial infarction. Clin Chim Acta. 2023;544:117354. [DOI] [PubMed] [Google Scholar]
- [37].Zhang X, Hu X, Qian L, et al. The association between nutritional-inflammatory status and chronic kidney disease prognosis: a population-based study. Ren Fail. 2025;47:2471016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [38].Wu Y, Kong XJ, Ji YY, et al. Serum electrolyte concentrations and risk of atrial fibrillation: an observational and mendelian randomization study. BMC Genomics. 2024;25:280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [39].Wang Y, Tao Y, Yuan M, et al. Relationship between the albumin-corrected anion gap and short-term prognosis among patients with cardiogenic shock: a retrospective analysis of the MIMIC-IV and eICU databases. BMJ Open. 2024;14:e081597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [40].Xia J, Wang L, Ma Z, et al. Cigarette smoking and chronic kidney disease in the general population: a systematic review and meta-analysis of prospective cohort studies. Nephrol Dial Transplant. 2017;32:475–87. [DOI] [PubMed] [Google Scholar]
- [41].Chen IC, Tsai WC, Hsu LY, et al. Association between alcohol consumption and chronic kidney disease: a population-based survey. Clin Exp Nephrol. 2024;28:1121–33. [DOI] [PubMed] [Google Scholar]
- [42].Menon V, Katz R, Mukamal K, et al. Alcohol consumption and kidney function decline in the elderly: alcohol and kidney disease. Nephrol Dial Transplant. 2010;25:3301–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [43].Clay JM, Callaghan RC, Sherk A, Naimi TS, Stockwell T, Asbridge M. Alcohol consumption and mortality among Canadian drinkers: a national population-based survival analysis (2000-2017). Drug Alcohol Rev. 2025;44:434–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [44].Sarich P, Canfell K, Egger S, et al. Alcohol consumption, drinking patterns and cause-specific mortality in an Australian cohort of 181,607 participants aged 45 years and over. Public Health. 2025;239:230–41. [DOI] [PubMed] [Google Scholar]