Introduction
The incidence of diabetes mellitus (DM)-associated kidney disease is increasing rapidly and DM has become the leading cause of end-stage renal disease requiring hemodialysis (HD) treatment [1,2]. DM accounts for nearly 50% of patients on incident dialysis, and patients on HD with DM show the highest rates of all-cause and cardiovascular (CV) mortality [3–5]. Uncontrolled hyperglycemia accelerates end-organ damage, and good glycemic control reduces the risk of complications and mortality associated with DM [6–8]. Nevertheless, patients on HD are prone to hypoglycemic events, and tight blood glucose control increases the risk of hypoglycemia, leading to an increased risk of mortality in patients with DM [9,10].
Fasting blood glucose (FBG) level is a representative marker of glucose control status, and it is widely used along with glycated hemoglobin (HbA1c) to assess the risk of clinically important complications [11,12]. Higher FBG levels are reliable predictors of premature death in the general population, and several studies have suggested that FBG target achievement is the main therapeutic goal for successful DM management [13–16]. A recent study further highlighted the significance of FBG levels in managing DM among HD patients [17]. Nevertheless, determining the optimal FBG target to minimize mortality risk has proven challenging, and the current guidelines do not provide any clear information on the target range of FBG levels in HD patients with DM [18–20]. Furthermore, there is a lack of analysis regarding the association between FBG levels and the risk of cause-specific deaths.
It is crucial to determine the levels of FBG that decrease the exceedingly high mortality risk in HD patients with DM. The target FBG level was evaluated to minimize the risk of all-cause mortality and we also tried to assess the association between FBG level and risk of cause-specific death to find out which causes of death are most susceptible to FBG level.
Methods
Data sources
We conducted a nationwide, population-based cohort study using the National Health Insurance Database (NHID) [21]. The NHID collects all medical fees incurred in Korea and accumulates complete data on claimed medical services. The National Health Insurance Service (NHIS) obligatorily requested specific codes (V001) to provide additional insurance coverage for all patients on HD. Therefore, research data on patients undergoing HD can be obtained from the NHID using specific insurance codes.
The NHIS also operates the National Health Screening Examination Program, which provides demographic, anthropometric, and laboratory data. Registered centers for National Health Screening throughout the country are quality-controlled and regulated by government organizations. We combined this information with that of maintenance patients on HD identified from the NHID.
Study population and design
We included adult (aged ≥19 years) patients on HD with DM from January 1, 2002, to December 31, 2018. Maintenance patients on HD were defined as those who had dialysis-specific insurance codes more than twice and medical expense claims for HD therapy for more than 90 days. The presence of DM was defined as at least one claim per year under the International Classification of Diseases, 10th Revision (ICD-10) codes (E11–14), and at least more than 30 days of claim per year for the prescription of antidiabetic medication. We identified 98,802 patients with DM on HD. Among them, 26,162 who underwent the National Health Screening Examination were included in this study. All patients were followed up from the initial health screening to the date of death, the date of the last checkup, or until December 31, 2018. Participant follow-up was censored at the time of kidney transplantation, transfer to peritoneal dialysis, or loss of follow-up.
The Institutional Review Board of Korea University Guro Hospital approved this study (No. 1607-187-779) and the use of the NHIS database was approved (No. 2018-1-148). The review board waived the requirement for informed consent and the study protocol complied with the Declaration of Helsinki.
Data collection
Baseline characteristics included age, sex, body mass index (BMI), duration of HD, comorbidities, systolic and diastolic blood pressure, hemoglobin levels, and lipid profiles. BMI was calculated as weight in kilograms divided by height in meters squared. Duration of HD refers to the period of time during which a person has received HD therapy prior to their initial health screening. During the health examination, blood pressure was measured at least twice using mercury or an automatic sphygmomanometer after a minimum rest period of 5 minutes with the individual in a sitting position. Laboratory findings were measured after 8 hours of fasting. The comorbidity burden was evaluated using the Charlson Comorbidity Index (CCI) score. Health Insurance Review and Assessment Service (HIRA) provided ‘the healthcare big data analysis guide: comorbidity analysis’, and we calculated the CCI score under the guideline [22]. CCI score weighted values to 17 medical conditions classified under the ICD-10 system. These weights varied from 1 to 6 for each item [23–26].
The exposure of interest was FBG, which was analyzed after categorization into six groups (<80, 80–100, 100–125, 125–150, 150–180, and ≥180 mg/dL). FBG levels were collected at the baseline health examination visit from the general health examination database. All-cause mortality and cause-specific mortality based on FBG categories were estimated. In the subgroup analyses, we defined the 80–100 mg/dL category as the reference to granularly investigate mortality risk associated with low (<80 mg/dL) and high (≥100 mg/dL) FBG categories.
Study outcomes
The study outcome was the association between FBG level and both all-cause mortality and cause-specific mortality. We categorized specific causes of death into three categories, CV death, infection-related death, and malignancy-related death. CV mortality was defined as deaths from diseases of the circulatory system (I00–I99). We defined infection-related deaths as deaths from certain infectious and parasitic diseases (A00–B99) and various types of pneumonia originating from viruses or bacteria (J10–J18). Malignancy-related deaths were identified by ICD-10 codes C00–C97. We also classified the study participants into subgroups to assess mortality risk based on age, sex, CCI, and BMI [27,28]. The NHIS database includes nationwide mortality information through issued death certificates. A computerized search of death certificate data from the Statistics Korea was performed for each mortality case.
Statistical analysis
Continuous variables with a normal distribution are expressed as mean ± standard deviation and compared using a one-way analysis of variance. Categorical variables were compared using the chi-square tests, when appropriate. Associations between FBG levels and both all-cause mortality and cause-specific mortality risk were determined using Cox proportional hazards models. The following variables were included in the Cox regression adjustment: age, sex, dialysis duration, systolic blood pressure, diastolic blood pressure, CCI, BMI, hemoglobin levels, and total cholesterol levels. The variable corresponding to the criteria used for dividing subgroups was not employed as the adjustment factor in the subgroup analysis. All-cause mortality according to FBG categories in the HD population was estimated with the Kaplan-Meier method. Statistical significance was set at p < 0.05. Statistical analyses were performed using the SAS Enterprise Guide version 7.1 (SAS Institute).
Results
Baseline characteristics
Table 1 shows the baseline clinical characteristics, demographics, and laboratory results of 26,162 patients. The mean age of the participants was 58.2 years, and 64.6% were male. The number of patients was 2,504 (11.2%), 7,051 (30.9%), 6,386 (24.1%), 3,675 (13.3%), 2,663 (8.7%), and 3,883 (11.9%) in the FBG <80, 80–100, 100–125, 125–150, 150–180, and ≥180 mg/dL category, respectively. Patients with higher FBG levels were more likely to have higher CCI scores and shorter HD duration. Patients in the higher FBG categories had lower hemoglobin and higher total cholesterol levels.
All-cause mortality risk based on fasting blood glucose categories
During a follow-up of 6.5 ± 3.8 years, we observed 9,295 deaths. The Kaplan-Meier survival curves for the six FBG categories showed a significant difference in survival probability (p < 0.001). Patients with FBG 80–100 mg/dL demonstrated the highest survival rate, whereas the lowest survival rate was observed in patients with FBG ≥180 mg/dL (Fig. 1).
Table 2 shows the significant association between FBG levels and all-cause mortality in the univariate Cox regression models. Compared to patients with FBG 80–100 mg/dL, a greater FBG level was associated with a higher risk of all-cause mortality in each of the four elevated FBG categories. The association between higher FBG levels and higher risk of all-cause mortality remained significant after multiple adjustments. Compared with the FBG 80–100 mg/dL, adjusted hazard ratios (HRs) were 1.10 (95% confidence interval [CI], 1.04–1.17; p = 0.002), 1.21 (95% CI, 1.13–1.29; p < 0.001), 1.36 (95% CI, 1.26–1.46; p < 0.001), and 1.61 (95% CI, 1.51–1.72; p < 0.001) for the FBG categories of 100–125, 125–150, 150–180, and ≥180 mg/dL, respectively. Patients with FBG <80 mg/dL also had 1.14-fold (95% CI, 1.05–1.23; p = 0.001) increased mortality risk.
Cause-specific mortality risk based on fasting blood glucose categories
We conducted the cause-specific mortality risk using Cox regression models. Compared to patients with FBG levels between 80 and 100, the patients in the other FBG categories had a higher risk of CV death (Table 3). The adjusted HR of CV death was 1.20 (95% CI, 1.00–1.43; p = 0.047), 1.21 (95% CI, 1.06–1.39; p = 0.01), 1.24 (95% CI, 1.06–1.46; p = 0.01), 1.31 (95% CI, 1.09–1.56; p = 0.003), and 1.30 (95% CI, 1.11–1.52; p = 0.001) for patients with FBG <80, 100–125, 125–150, 150–180, and ≥180 mg/dL, respectively. However, the risk of infection-related death did not significantly increase in the patients with most FBG categories, while patients with FBG levels ≥180 mg/dL had a higher risk. Patients with different FBG levels did not show any statistical significance in the risk of malignancy-related death.
Subgroup analyses based on the predefined criteria
We divided the study participants into subgroups to assess the all-cause mortality risk based on specific parameters (Table 4). Compared to patients with FBG 80–100 mg/dL, those with FBG ≥100 mg/dL were mostly associated with a higher risk of all-cause mortality. FBG <80 mg/dL was associated with a higher risk of mortality in patients aged <65 years, male patients, and those with a CCI score ≥6. Significant interaction was observed between FBG level and CCI score (p for interaction = 0.04). Patients with a CCI score ≥6 had a higher risk of mortality when their FBG was <80 mg/dL (adjusted HR, 1.16; 95% CI, 1.05–1.28; p = 0.004) compared to those with a CCI score <6 (adjusted HR, 1.10; 95% CI, 0.97–1.25; p = 0.15). Additionally, patients with a BMI ≥23 kg/m2 exhibited an increased risk of mortality for FBG <80 mg/dL (adjusted HR, 1.12; 95% CI, 1.00–1.24; p = 0.05). Nonetheless, individuals with a BMI <23 kg/m2 showed a further increased mortality risk with FBG <80 mg/dL (adjusted HR, 1.15; 95% CI, 1.03–1.28; p = 0.01). The interaction between FBG level and BMI was statistically significant (p for interaction = 0.02).
Discussion
In this nationwide population-based cohort study, FBG levels significantly affected the risk of all-cause mortality in patients with DM undergoing HD. We found that patients with FBG levels of 80–100 mg/dL had the lowest mortality rate and that the mortality risk increased as the FBG level increased. FBG levels of <80 mg/dL also increased the risk of all-cause mortality, and the association between FBG levels and mortality risk differed significantly between patient-specific conditions. These findings indicate that FBG is one of the contributing factors to clinical outcomes in patients with DM undergoing HD and that the intensity of glycemic control should be individualized based on the patient setting.
Several guidelines recommend that HD patients exercise greater caution regarding low glycemic status to reduce the risk of adverse clinical events. They also suggest a less strict HbA1c target approach [20,29]. However, these guidelines do not specify a clear target for FBG levels in patients with renal impairment. Consequently, HD patients often maintain higher FBG levels in adherence to these recommendations [11,20]. Our study findings indicate that diabetic HD patients with FBG levels between 80 and 100 mg/dL had the lowest mortality rate, suggesting that maintaining FBG levels within this range provides the greatest benefit. These findings offer valuable information to establish strategies for glycemic control in HD patients with DM.
Several previous studies have analyzed the risk of all-cause mortality based on HbA1c levels in patients undergoing HD. The results consistently demonstrated that the risk of all-cause mortality significantly increased when the HbA1c level was ≥8.0% to 8.5% [8,30,31]. However, HD patients have issues that need to be considered when tracking their glycemic status using HbA1c alone. There was no significant relationship between HbA1c levels <8.0% to 8.5% and the risk of mortality, and the benefits of glycemic control were not fully evident for HbA1c levels <8.0% [32]. In this study, we demonstrated that FBG levels significantly affected the risk of all-cause mortality, and even mild increases in FBG levels were associated with higher mortality risk [33–36]. These findings indicate that FBG could be an additional marker of glycemic control in combination with HbA1c and that it would be useful to support the limitation of HbA1c, particularly when it is less than 8%.
Several studies have consistently reported that lower HbA1c levels are associated with a higher risk of mortality in patients with DM on HD [30,37]. However, HbA1c has limitations in setting lower glycemic control targets for patients with HD. The shortened lifespan of red blood cells, the use of erythropoietin-stimulating agents, and alterations in iron metabolism can cause HbA1c tests to underestimate blood glucose levels [38–40]. Therefore, HbA1c levels that are not excessively low may not be associated with an increased risk of mortality, and our previous study found no significant association between HbA1c <6.5% and increased risk of mortality [36]. In this study, we demonstrated a J-shaped association between FBG levels and the risk of all-cause mortality, with a significant increase in mortality risk at FBG <80 mg/dL. These findings suggest that FBG is valuable in determining the lower limit of glycemic control independent of HbA1c-related confounding factors.
The analysis of cause-specific mortality revealed a J-shaped curve in the risk of CV deaths with lower or higher FBG levels. However, FBG did not exhibit a linear increase in the risk of infection or malignancy-related deaths as it increased. This suggests that FBG significantly increases the risk of death due to CV causes compared to other causes, and setting an FBG target of 80–100 mg/dL could reduce the risk of CV mortality [13]. The strong association between the risk of CV death and FBG aligns with the results of our previous study that utilized HbA1c, indicating that glycemic status has the most substantial impact on increasing CV risk [36].
In HD patients, all-cause mortality risk is closely associated with BMI, and the risk is significantly higher in nonobese patients or patients with a lower BMI [41]. In addition, a high CCI score was a substantial risk factor for mortality [42]. In the present studies, we found that mortality risk was more pronounced in patients with CCI score ≥6 or BMI <23 kg/m2 for FBG <80 mg than those who did not have, and that there was a significant interaction. These findings suggest that the association between FBG levels <80 mg/dL and mortality risk is stronger in these patients. We suggest that more effort and close monitoring of FBG levels should be provided to susceptible patients to prevent FBG levels from dropping below 80 mg/dL. Additionally, we found that FBG levels ≥100 mg/dL were associated with the increased risk of all-cause mortality in all subgroup analyses. These findings suggest that higher FBG levels should be controlled regardless of age, sex, comorbidity, and BMI, and that the benefits of hyperglycemia management could be obtained across different patient characteristics.
This cohort study had some limitations. First, we did not evaluate hypoglycemic events during follow-up. A strict target FBG level might increase the risk of hypoglycemia, and data on hypoglycemic risk would provide useful information about the effect of FBG levels. Secondly, we included patients undergoing HD who underwent health screening, and healthy individuals were more likely to be enrolled in this study. Therefore, selection bias could not be eliminated. Third, antihyperglycaemic medications and insulin were not evaluated, and the type and dose of antihyperglycaemic agents were not reflected in this analysis.
In conclusion, FBG levels are significantly associated with the risk of all-cause mortality in HD patients with DM. A J-shaped association was observed between mortality risk and FBG level, and glycemic control within FBG 80–100 mg/dL in this population improved and decreased the survival rate. These results provide valuable information regarding appropriate FBG target levels and suggest that FBG targets could be lower than those in current clinical practice.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
