Temporal Trends of Severe Hypoglycemia and Subsequent Mortality in Patients with Advanced Diabetic Kidney Diseases Transitioning to Dialysis

Background: Patients with diabetic kidney disease (DKD) are at higher risk of hypoglycemia than diabetic patients without DKD. We aimed to investigate the temporal trends of severe hypoglycemia in advanced DKD patients transitioning to dialysis and examine risk factors associated with severe hypoglycemia. We also investigated the association of severe hypoglycemia episodes with one-year mortality after initiation of dialysis in patients with advanced DKD. Methods: Using the Taiwan National Health Insurance Research Database, 46,779 advanced DKD patients transitioning to dialysis (Peritoneal dialysis 4216, hemodialysis 42,563) between 1997 and 2011 were enrolled. We calculated the rates of severe hypoglycemia from 5 years before dialysis until 10 years after dialysis. Cox proportional hazard model was used to examine the risk factors of post end stage renal disease (ESRD) one-year hypoglycemia and post ESRD one-year mortality in advanced DKD patients transitioning to dialysis. Results: We found that 11.5% of advanced DKD patients had at least one episode of severe hypoglycemia the year leading up to dialysis initiation. Multivariate analysis revealed hemodialysis compared with peritoneal dialysis, stroke, use of sulfonylurea, glinide, and insulin were associated with higher risk of severe hypoglycemia one year after transitioning to dialysis. Increased frequency of severe hypoglycemia-related hospitalizations was associated with incrementally higher mortality risk one year after transitioning to dialysis (Pre-ESRD hypoglycemia: Hazard ratios: 1.28 (1.18–1.38, p < 0.001), 1.64 (1.49–1.81, p < 0.001) for one, two hypoglycemia-related hospitalizations, respectively; post-ESRD hypoglycemia: HRs of 1.56 (1.40–1.73, p < 0.001), 1.72 (1.39–2.12, p < 0.001) for one, two hypoglycemia-related hospitalizations, respectively (reference group: no hypoglycemia related hospitalization)). Conclusions: Among advanced DKD patients, we observed a progressive elevated risk of hypoglycemia during the critical dialysis transition period. Increased frequency of severe hypoglycemia-related hospitalizations was associated with higher mortality risk one year after transitioning to dialysis. Further study of glycemic management strategies which prevent hypoglycemia during the critical transition period are warranted.

Abstract: Background: Patients with diabetic kidney disease (DKD) are at higher risk of hypoglycemia than diabetic patients without DKD. We aimed to investigate the temporal trends of severe hypoglycemia in advanced DKD patients transitioning to dialysis and examine risk factors associated with severe hypoglycemia. We also investigated the association of severe hypoglycemia episodes with one-year mortality after initiation of dialysis in patients with advanced DKD. Methods: Using the Taiwan National Health Insurance Research Database, 46,779 advanced DKD patients transitioning to dialysis (Peritoneal dialysis 4216, hemodialysis 42,563) between 1997 and 2011 were enrolled. We calculated the rates of severe hypoglycemia from 5 years before dialysis until 10 years after dialysis. Cox proportional hazard model was used to examine the risk factors of post end stage renal disease (ESRD) one-year hypoglycemia and post ESRD one-year mortality in advanced DKD patients transitioning to dialysis. Results: We found that 11.5% of advanced DKD patients had at least one episode of severe hypoglycemia the year leading up to dialysis initiation. Multivariate analysis revealed hemodialysis compared with peritoneal dialysis, stroke, use of sulfonylurea, glinide, and insulin were associated with higher risk of severe hypoglycemia one year after transitioning to dialysis. Increased frequency of severe hypoglycemia-related hospitalizations was associated with incrementally higher mortality risk one year after transitioning to dialysis (Pre-ESRD hypoglycemia: Hazard ratios: 1. 28 (1.18-1.38, p < 0.001), 1.64 (1.49-1.81, p < 0.001) for one, two hypoglycemia-related hospitalizations, respectively; post-ESRD hypoglycemia: HRs of 1.56 (1.40-1.73, p < 0.001), 1.72 (1.39-2.12, p < 0.001) for one, two hypoglycemia-related hospitalizations, respectively (reference group: no hypoglycemia related hospitalization)). Conclusions: Among advanced DKD patients, we observed a progressive elevated risk of hypoglycemia during the critical dialysis transition period. Increased frequency of severe hypoglycemia-related hospitalizations was associated with higher mortality risk one year after transitioning to dialysis. Further study of Figure 1 demonstrates the process used for selecting the participants in the study cohorts. Patients who were newly diagnosed with ESRD (ICD-9-CM code: 585; A code: A350) from 1 January 1997 to 31 December 2011 were identified from the Taiwan NHIRD. Some peritoneal dialysis (PD) patients temporarily receive hemodialysis (HD) at the initiation of dialysis; therefore, dialysis modality was defined as the modality at day 90 after the first dialysis session. Because some patients suffering from acute kidney injury lead to temporary dialysis, patients who underwent dialysis for less than 90 days were excluded (n = 10,476). Index date was defined as the date of starting the first dialysis. Patients with DM were defined by at least three outpatient claims or one inpatient claims with ICD-9-CM code = 250. We exclude patients diagnosed DM in less than five years (n = 29,474) because we calculated the 5-year rate of severe hypoglycemia before dialysis. Patients without DM (n = 40,773), diagnosed with DM after dialysis (n = 9292), <20 years of age (n = 88), with missing sex information (n = 6), and who subsequently received kidney transplantation (ICD-9-CM: V42.0) after dialysis (n = 877) were excluded. Finally, 42,563 diabetic patients undergoing HD and 4216 diabetic patients undergoing PD were enrolled in the study cohort. because we calculated the 5-year rate of severe hypoglycemia before dialysis. Patients without DM (n = 40,773), diagnosed with DM after dialysis (n = 9292), <20 years of age (n = 88), with missing sex information (n = 6), and who subsequently received kidney transplantation (ICD-9-CM: V42.0) after dialysis (n = 877) were excluded. Finally, 42,563 diabetic patients undergoing HD and 4216 diabetic patients undergoing PD were enrolled in the study cohort.

Outcome Variables
The outcomes of primary interest were the occurrence of severe hypoglycemia during the course of CKD progression until 10 years after dialysis initiation.
Severe hypoglycemia was defined as patients who visited the emergency department (ED) with a diagnosis of hypoglycemia or a primary discharge diagnosis of the following ICD-9-CM codes: 251.0 (hypoglycemic coma), 251.2 (unspecified hypoglycemia), and 251.8 (diabetic hypoglycemia; hypoglycemic shock) in the inpatient database.
The secondary outcomes were post-ESRD one-year severe hypoglycemia and post-ESRD one-year mortality. Mortality was defined by withdrawal from the NHI program or death in NHIRD or RCIPD. All participants were followed from the index date until the first occurrence of hypoglycemia, death, withdrawal from insurance, or the end of the follow-up period (31 December 2013), whichever came first.

Statistical Analysis
The rates of severe hypoglycemia 5 years before dialysis until 10 years after dialysis were calculated in advanced DKD patients. The rates between dialysis modalities were compared using inverse probability of treatment weighting (IPTW), by calculating the propensity scores based on baseline variables, which were used to balance the differences between the PD and HD patients. Logistic regression was used to calculate the propensity score for each patient by estimating the probability of assignments based on baseline variables, including age, sex, comorbidity, and medication. The balance of covariates at baseline between the two dialysis groups was assessed using absolute standardized mean difference (ASMD), which is not influenced by sample size. An ASMD value of ≤0.1 indicates a negligible difference in potential confounders between the two dialysis groups [13]. Survival analyses (log-rank test in univariate analysis and Cox's proportional hazard model analysis) were conducted to determine the effects of covariates on hypoglycemia. The unadjusted and adjusted hazard ratio (HR and aHR) and its 95% confidence interval (CI) were calculated to demonstrate the relative risk of severe hypoglycemia for the covariates of interest. Survival analysis without propensity score weighting was conducted to evaluate the effects of covariates on hypoglycemia. For one-year mortality after dialysis, survival analysis was also performed to evaluate the effect of covariates, pre-ESRD severe hypoglycemia, and post-ESRD severe hypoglycemia. All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA); a p value of 0.05 was considered statistically significant. Table 1 presents the baseline characteristics of the HD cohort when compared with the PD cohort before and after IPTW. The mean age (years) was 65.83 ± 11.55 in the HD cohort and 60.38 ± 12.89 in the PD cohort. The proportion of patients with stroke, coronary artery disease, and congestive heart failure, as well as the prescription rates of sulfonylurea and insulin, were significantly higher in the HD cohort than in the PD cohort. After propensity score weighting, baseline characteristics were balanced between the two cohorts.  Figure 2 shows the temporal trends of severe hypoglycemia in patients with advanced DKD transitioning to dialysis. A gradual increase in rates of severe hypoglycemia occurred as time advanced before the initiation of dialysis. A total of 11.5% of Advanced DKD patients had at least one episode of severe hypoglycemia the year leading up to dialysis initiation. One year after the start of dialysis, the rate of severe hypoglycemia remained high (5.88%), but declined compared with the pre-dialysis period. Two years after dialysis, the hypoglycemic rates were stable at around 4%.  Figure 2 shows the temporal trends of severe hypoglycemia in patients with advanced DKD transitioning to dialysis. A gradual increase in rates of severe hypoglycemia occurred as time advanced before the initiation of dialysis. A total of 11.5% of Advanced DKD patients had at least one episode of severe hypoglycemia the year leading up to dialysis initiation. One year after the start of dialysis, the rate of severe hypoglycemia remained high (5.88%), but declined compared with the pre-dialysis period. Two years after dialysis, the hypoglycemic rates were stable at around 4%.   Figure 3 presents the rates of severe hypoglycemia in the HD cohort compared with the PD cohort. The overall hypoglycemic risk in the HD cohort were not significantly different from those of the PD cohort. However, the risk in the first year after dialysis initiation were significantly higher in the HD cohort compared with the PD cohort before propensity score weighting (HR = 1.45 (1.23-1.7); p < 0.001) and after (aHR = 1.37 (1.16-1.62); p < 0.001) propensity score weighting. Figure 3 presents the rates of severe hypoglycemia in the HD cohort compared with the PD cohort. The overall hypoglycemic risk in the HD cohort were not significantly different from those of the PD cohort. However, the risk in the first year after dialysis initiation were significantly higher in the HD cohort compared with the PD cohort before propensity score weighting (HR = 1.45 (1.23-1.7); p < 0.001) and after (aHR = 1.37 (1.16-1.62); p < 0.001) propensity score weighting.  Table 2 presents the risk factors of severe hypoglycemia in advanced DKD patients one year after transitioning to dialysis. The Cox's proportional hazard model revealed hemodialysis compared with peritoneal dialysis (aHR = 1.29; 1.12-1.30; p < 0.001) and stroke (aHR = 1.16; 1.07-1.25; p < 0.001) were associated with severe hypoglycemia one year after dialysis. Use of antidiabetic agents, especially sulfonylurea (aHR = 1.31; 1.20-1.44; p < 0.001), glinide (aHR =1.20; 1.1-1.30; p < 0.001), and insulin (aHR = 2.63; 2.32-2.98; p < 0.001), predisposed patients to higher risks of severe hypoglycemia one year after dialysis. By contrast, patients with hypertension or hyperlipidemia were associated with lower risks of severe hypoglycemia one year after dialysis. Table 3 presents the risk factors of one-year mortality among DKD patients transitioning to dialysis. Pre-dialysis characteristics, old age (Age > 65), stroke, heart failure, liver cirrhosis, and preor post-ESRD severe hypoglycemia episodes were positively associated with one-year mortality. Frequency of hypoglycemia-related hospitalizations was associated with higher mortality risk compared to no hypoglycemia-related hospitalizations (Pre-ESRD hypoglycemia: 1 versus none, aHR = 1.28,1.18-1.38; p < 0.001; 2 versus none, aHR = 1.64,1.49-1.81; p <0.001; Post-ESRD  Table 2 presents the risk factors of severe hypoglycemia in advanced DKD patients one year after transitioning to dialysis. The Cox's proportional hazard model revealed hemodialysis compared with peritoneal dialysis (aHR = 1.29; 1.12-1.30; p < 0.001) and stroke (aHR = 1.16; 1.07-1.25; p < 0.001) were associated with severe hypoglycemia one year after dialysis. Use of antidiabetic agents, especially sulfonylurea (aHR = 1.31; 1.20-1.44; p < 0.001), glinide (aHR =1.20; 1.1-1.30; p < 0.001), and insulin (aHR = 2.63; 2.32-2.98; p < 0.001), predisposed patients to higher risks of severe hypoglycemia one year after dialysis. By contrast, patients with hypertension or hyperlipidemia were associated with lower risks of severe hypoglycemia one year after dialysis.  Table 3 presents the risk factors of one-year mortality among DKD patients transitioning to dialysis. Pre-dialysis characteristics, old age (Age > 65), stroke, heart failure, liver cirrhosis, and pre-or post-ESRD severe hypoglycemia episodes were positively associated with one-year mortality. Frequency of hypoglycemia-related hospitalizations was associated with higher mortality risk compared to no hypoglycemia-related hospitalizations (Pre-ESRD hypoglycemia: 1 versus none, aHR = 1.28, 1.18-1.38; p < 0.001; 2 versus none, aHR = 1.64, 1.49-1.81; p <0.001; Post-ESRD hypoglycemia; 1 versus none, aHR = 1.56, 1.40-1.73; p < 0.001; 2 versus none, aHR = 1.72, 1.39-2.12; p < 0.001). Hemodialysis compared with peritoneal dialysis was associated with higher one-year mortality (aHR = 1.64, 1.44-1.87; p < 0.001). By contrast, hypertension and hyperlipidemia were negatively associated with one-year mortality.

Discussion
The present study demonstrated that the risk of severe hypoglycemia gradually increased in patients with advanced DKD transitioning to dialysis. We found that 11.5% of advanced DKD patients had at least one episode of severe hypoglycemia the year leading up to dialysis initiation. Hemodialysis compared with peritoneal dialysis, stroke, use of insulin, sulfonylurea, and glinide were associated with higher risk of severe hypoglycemia in the critical transition period. We observed a dose response relationship between increased frequency of severe hypoglycemia-related hospitalizations and incrementally higher one-year mortality, regardless if its occurrence was before or after dialysis.

Trajectory of Severe Hypoglycemia over the Dialysis Transition Period
Our study describes the trajectory of severe hypoglycemia over the dialysis transition period and demonstrated a gradual increase in severe hypoglycemia as time advanced before dialysis initiation. This may be explained by the differences in glucose metabolism among DKD patients and DM patients without kidney disease. Renal clearance of insulin and degradation of insulin in peripheral tissues are decreased in patients with CKD [14,15]. With the progressive decline in renal mass, decreased renal gluconeogenesis frequently occurs in these patients [14]. Moreover, most antidiabetic drugs have a prolonged half-life in patients with CKD because these drugs are eliminated through renal excretion. In secondary analysis of the ACCORD study, patients with mild to moderate CKD (stage 1-3) were associated with nearly a two-fold higher risk of hypoglycemia compared with those of normal renal function [16]. In another large population-based study involving 243,222 veterans, the incidence of hypoglycemia, which was defined by blood glucose level <70 mg/dL, was higher in patients with CKD versus without CKD [6]. However, the above studies mainly focus on mild to moderate CKD and did not demonstrate the different hypoglycemia risk as CKD progression. Our study gives an insight into the heightened risk of severe hypoglycemia in patients with advanced DKD during the dialysis transition period (one year before and after dialysis). Patients with advanced DKD may present with symptoms of uremia, such as decreased oral intake, nausea, and vomiting, leading to malnutrition and increased hypoglycemia [17]. The constellations of the aforementioned factors predispose patients with advanced DKD to higher risks of hypoglycemia. Thus, a spontaneous decrease in hemoglobin A1c levels in the absence of any treatment in the period of advanced CKD transitioning to dialysis may be considered a "hypoglycemia alarm", rather than true improvement by the primary care physician. Without medication adjustments, patients may encounter hypoglycemia leading to hospitalizations.
This large population-based study used propensity score weighting to compare risk of severe hypoglycemia between HD and PD patients. This method reduced the unbalance inherent to retrospective cohort studies. In this study, similar risks of developing severe hypoglycemia were observed between the HD and PD patients, except high risk in the HD cohort than in the PD cohort during the first year of dialysis. Several studies have reported higher risks of mortality and stroke during the first year following the start of HD [18][19][20]. Pathophysiological changes such as rapid toxin and fluid removal, infection related to central venous catheter, inflammation related to repeated needle puncture, and rapid decline of residual renal function are more pronounced after HD initiation. Whether the dialysis treatment itself, the progression of comorbid diseases, or the rapid decline of residual renal function accounted for the increased risk of severe hypoglycemia during the first year of HD is uncertain and warrants further investigation. Nevertheless, our study provides insight into the risk of severe hypoglycemia, which required medical attention during dialysis transition period, especially when a patient chooses hemodialysis as modality.
Risk factor analyses of one-year severe hypoglycemia after transitioning to dialysis revealed that hemodialysis compared with peritoneal dialysis and stroke were associated with higher risk of severe hypoglycemia, whereas hypertension and hyperlipidemia were associated with lower risk of severe hypoglycemia in advanced DKD patients transitioning to dialysis. Borzi et al. reported that hypoglycemia was more frequent among patients with cognitive dysfunction [21]. Stroke may lead to an increased risk of hypoglycemia because of impaired cognitive function, difficulty in ambulation, and poor nutrition. With regard to the various antidiabetic agents used in advanced DKD patients, sulfonylurea, glinide, and insulin were found to be associated with higher risks of hypoglycemia in the present study. Previous studies have demonstrated that insulin requirements decrease by 25% when the GFR falls below 50 mL/min/1.73 m 2 , and by a further 50% when it falls below 10 mL/min/1.73 m 2 [15,22,23]. Renal clearance of insulin and insulin catabolism in peripheral tissues is also decreased in patients with advanced DKD [14]. The glucose-lowering effect of insulin persists for a long duration and increases the risk of symptomatic hypoglycemia. After the initiation of dialysis, increased insulin resistance in the uremic milieu improves and further decreases insulin requirements [24]. The metabolism of sulfonylurea, which belongs to a class of long-acting antidiabetic agents, is prolonged in patients with advanced DKD. Previously, two studies have demonstrated that patients with CKD experienced severe hypoglycemia, which required ED visits or hospitalization, especially in patients who received insulin and sulfonylurea treatments [25,26]. Thus, monitoring the trend of the glucose level and dosing adjustments of sulfonylurea, glinide, and insulin to prevent hypoglycemia are particularly crucial and may decrease the mortality risks in advanced DKD patients transitioning to dialysis.

One-Year Mortality after Dialysis
High mortality risks were observed particularly in the period of non-dialysis dependent (NDD) advanced DKD transitioning to dialysis dependent ESRD [9,10]. Our study gave an insight into the long-term sequelae of severe hypoglycemia on mortality risks of advanced DKD transitioning to dialysis, and demonstrated increased frequency of severe hypoglycemia-related hospitalizations were associated with higher risks of one-year mortality. In a secondary analysis of the ACCORD trial, intensive glycemic control was associated with a 31% higher risk all cause mortality compared with standard therapy in patients with CKD, but not in patients without CKD [16]. Though the relationship between mortality and hypoglycemia is not direct examined, the increased mortality may partially be explained by frequent hypoglycemia in patients with CKD during intensive glycemic control. Hypoglycemia-related hospitalizations and CKD were independently associated with higher mortality in 8767 diabetic patients from the Hong Kong Diabetes Registry [27]. Hypoglycemia was reported to be associated with increased one-day mortality in a large population-based study in Veterans [6]. Hypoglycemia is linked to increased short-term mortality, possibly related to encephalopathy, seizure, and coma. The adrenergic surge, in the setting of hypoglycemia, aggravated pre-existing underlying disorders, including cardiac arrhythmia, vasoconstriction, and sudden cardiac death [8]. Regarding long term sequelae, recurrent hypoglycemia events were reported to decrease autonomic system responses and further aggravate cardiovascular risk [28]. Furthermore, frequent severe hypoglycemia may reflect underlying comorbidities including uremia, protein energy malnutrition, bedridden status, which might contribute to higher mortality during the critical dialysis transition period. Our study demonstrated that severe hypoglycemia in the dialysis transition period was not only a sign of poor sugar control but was also associated with death. Further studies are warranted to elicit the mechanism by which severe hypoglycemia contributes to higher mortality in DKD patients in the transition period.
This study had several limitations. First, the NHIRD is based on administrative data, which may be subject to data recording and coding errors. Second, the renal metabolism of antidiabetic agents differs at various stages of CKD. Unfortunately, laboratory data such as creatinine, glucose, and glycated hemoglobin levels were not available in the Taiwan NHIRD. Third, we were unable to access the severity of comorbidities (blood pressure, heart function, and lipid level), which could contribute to the development of hypoglycemia. Unavailability of laboratory data and concomitant disorders severity may limit the generalizability and usefulness of our study. Fourth, only patients with severe hypoglycemia, which requiring an ED visit and hospitalization, were considered in this study to avoid the double counting of hypoglycemic events and improve the accuracy of the outcome. Thus, the reported hypoglycemic rate might be underestimated. Finally, although we selected numerous variables for the propensity score method, residual confounding by unmeasured factors cannot be ruled out.

Conclusions
Progressive heightened risk of severe hypoglycemia was observed in advanced DKD patients transitioning to dialysis. Severe hypoglycemia episodes were associated with an incremental higher risk of mortality in the transition period. Use of anti-diabetic agents, especially sulfonylurea, glinide, and insulin, warrants close glucose level monitoring and dose adjustments to prevent hypoglycemia. Physicians must pay more attention to severe hypoglycemia, which might prove to be a concern in glycemic management in advanced DKD patients transitioning to dialysis. Further study of glycemic management strategies which prevent severe hypoglycemia during the critical transition period are warranted.