- Email: [email protected]
Hemodialyzed type I and type II diabetic patients in the US: Characteristics, glycemic control, and survival
original article
http://www.kidney-international.org & 2006 International Society of Nephrology
see commentary on page 1392
Hemodialyzed type I and type II diabetic patients in the US: Characteristics, glycemic control, and survival ME Williams1, E Lacson Jr2, M Teng2, N Ofsthun2 and JM Lazarus2 1
Renal Unit, Joslin Diabetes Center, One Joslin Place, Boston, Massachusetts, USA and 2Fresenius Medical Care, Lexington, Massachusetts, USA
Diabetes mellitus (DM) constitutes a major end-stage renal disease (ESRD) health problem. Glycemic control is fundamental to the management of diabetes and its complications, and relies on monitoring of hyperglycemia. We therefore performed a primary data analysis of glycemic control and survival on a large national ESRD database. Ninety-five percent of patients with DM had type II diabetes (N ¼ 23 504), and five percent had type I diabetes (N ¼ 1 371). For the combined population, the mean hemoglobin A1c (HgbA1c) was 6.77%, and the mean random blood glucose was 168 mg/dl. Mean HgbA1c values were 47.0% in 35% and 48.5% in 14%. Mean HgbA1c values were below 5% in 11.3% of patients. Type I study patients tended to have higher HgbA1c values. Most patients (75.8%) had three or more random blood glucose determinations within 90 days preceding the HgbA1c measurement. The HgbA1c showed only a weak correlation with mean random glucose values (R2 0.3716; s.e. ¼ 1.36). The survival rates in the subsequent 12-month period ranged from 80 to 85% across different HgbA1c strata. Kaplan–Meier survival curves grouped by HgbA1c levels showed no correlation between HgbA1c and survival at 12 months. More studies are needed to refine recommendations for the role of HgbA1c and glycemic control in this patient population. Kidney International (2006) 70, 1503–1509. doi:10.1038/sj.ki.5001789; published online 30 August 2006 KEYWORDS: diabetes; hemodialysis; hyperglycemia; survival; hemoglobin A1c; ESRD
Correspondence: ME Williams, Renal Unit, Joslin Diabetes Center, One Joslin Place, Boston Massachusetts, USA. E-mail: [email protected] Received 16 March 2006; revised 20 June 2006; accepted 27 June 2006; published online 30 August 2006 Kidney International (2006) 70, 1503–1509
Patients with diabetes mellitus (DM) now account for 45 per cent of the prevalent end-stage renal disease (ESRD) population in the US, and the rate of ESRD caused by diabetes increased 86% between 1993 and 2003.1 Most incident patients with ESRD and diabetes carry multiple chronic comorbid conditions such as ischemic heart disease, congestive heart failure, and peripheral vascular disease when they begin dialysis. Macrovascular complications, including myocardial infarction, stroke, and peripheral vascular disease, are the major causes of morbidity and mortality in patients with ESRD and diabetes. Although only a minority of patients with diabetes achieve recommended goals for glycemic control,2–4 glycemic control is fundamental to the management of diabetes5 and its complications. Poor glycemic control is increasingly considered as an independent cardiovascular risk factor in diabetes.4,6–9 Among patients with diabetes in a large prospective study, high blood glucose concentrations were associated with a greater incidence of cardiovascular disease.10 Recent data suggest that glycosylated hemoglobin levels are related to coronary heart disease risk in persons with type I and II diabetes.11 In fact, chronic hyperglycemia has been hypothesized to contribute to coronary disease in individuals without diabetes as well.12,13 Limited data suggest that poor glycemic control in patients with diabetes on dialysis may be associated with increased morbidity from diabetic vascular complications.14,15 Recent data demonstrate that glycemic control may benefit patients with diabetes who have established coronary disease.16,17 Epidemiologic studies suggest the potential for improved glycemic control to reduce cardiovascular complications in diabetes, hence the benefit of hyperglycemia control on cardiovascular damage in patients with diabetes continues to be studied.18,19 American Diabetes Association (ADA) guidelines indicate that improved glycemic control may lower the risk of myocardial infarction and cardiovascular death.20 Among risk factor reductions for mortality in dialysis patients with diabetes, improved glycemic control is widely recommended.21 However, whether glycemic control affects survival in diabetic ESRD patients remains unclear.22 Monitoring of hyperglycemia is at the center of management of diabetes.23,24 The hemoglobin A1c (HgbA1c) concentration is a marker of hyperglycemia and reflects average blood glucose concentration over 1503
Table 1 | Study patient characteristics with a breakdown into types I and II DM (values in parenthesis are s.d.) DM patients with HgbA1c (drawn between 1 October 2002 and 31 December2002)
Characteristics No. of patients Age* (years) % female Race* White (%) Black (%) Other (%) Vintage** (days) Body surface area* (m2) % of patients on Erythropoietin access type*** Fistula (%) Graft (%) Catheter (%) Blank (%) HD dose (eKt/V) Albumin Hemoglobin Calcium Phosphorus* Creatinine White blood cell count** HgbA1c (%)
1504
Type I DM
Type II DM
24 875 63.7 (12.1) 51.5 — 53.0 36.4 10.6 1051 (951) 1.86 (0.25) 99.0
1371 49.3 (14.4) 49.5 — 60.6 32.5 6.9 1172 (1.075) 1.82 (0.25) 98.7
23 504 64.5 (11.4) 51.6 — 52.6 36.6 10.8 1044 (943) 1.86 (0.25) 99.0
— 30.0 45.7 23.5 0.8 1.41 (0.29) 3.82 (0.38) 11.72 (1.10) 9.29 (0.70) 5.60 (1.44) 8.23 (2.67) 7.77 (2.87) 6.77 (1.44)
— 34.4 40.8 24.4 0.4 1.41 (0.32) 3.83 (0.42) 11.71 (1.17) 9.27 (0.73) 5.89 (1.55) 8.60 (2.61) 7.97 (2.39) 7.49 (1.99)
— 29.7 46.0 23.5 0.8 1.41 (0.28) 3.81 (0.38) 11.72 (1.09) 9.29 (0.70) 5.59 (1.44) 8.20 (2.67) 7.76 (2.89) 6.73 (1.68)
DM, diabetes mellitus; HD, hemodialysis; HgbA1c, hemoglobin A1c. *Po0.0001; **Po0.0016; ***P=0.003: All comparing type I vs type II DM patients only.
Hemoglobin A1c frequency distribution for Q4 2002 20 N = 24 875 patients with diabetes
RESULTS % of patients
15
10
5
5 5– 6 6– 6. 5 6. 5– 7 7– 7. 5 7. 5– 8 8– 8. 5 8. 5– 9 9– 9. 9. 5 5– 1 10 0 –1 0 10 .5 .5 –1 1 >1 1 5.
0
0 5.
Patient characteristics are shown in Table 1. Among the 24 875 eligible patients, a total of 1371 (5.5%) had type I diabetes and 23 504 (94.5%) had type II diabetes. Type I study patients were significantly different from the type II study patients, with younger age (Po0.0001), different racial distribution with more white race (Po0.0001), longer dialysis vintage (Po0.0016), slightly lower body surface area (Po0.0001), and different distribution of vascular accesses (P ¼ 0.003). Type I study patients had higher serum phosphorus levels (Po0.0016). Type I study patients had higher mean HgbA1c values (7.5 vs 6.7%, Po0.0001). The frequency distribution of the HgbA1c results is shown in Figure 1. For all study patients, mean HgbA1c values were 47.0% in 35% and 48.5% in 14%; in 11.3% of patients, mean HgbA1c values were below 5%. Overall, patients with type I DM tended to have higher HgbA1c values, with mean HgbA1c47.0% in 50% and 48.5% in 426% of type I study patients. Of the entire study cohort, 41% were treated with insulin, 20% with oral hypoglycemia agents, 16% with both, and 23% with neither. For those with HgbA1co6%, 37% were on neither insulin nor oral hypoglycemic agents.
All study patients
5.
3 months in diabetic individuals.25 HgbA1c tracks well in individuals over time.26 The ADA recommends using pointof-care glycohemoglobin (HgbA1c) values to track glycemic control, make timely decisions on treatment, and to reduce risks of adverse outcomes by lowering A1c levels to below 7%.27,28 However, less than half of patients with ESRD and diabetes fulfill the recommendations of the ADA to undergo at least four glycohemoglobin tests per year, and up to a quarter receive no testing.1 Monitoring of capillary blood glucose is also recommended, with a frequency and timing dependent on the patient’s medical condition. However, half of chronic kidney patients with diabetes have not had glycosylated hemoglobin testing in the year before dialysis initiation.29 Diabetic monitoring in the ESRD population is also suboptimal, with over 60% not prescribed any diabetic test strips and less than half receiving the four or more HgbA1c tests per year recommended by the ADA1. The apparent variability in HgbA1c testing practices in ESRD may derive from the lack of consensus on HgbA1c testing to reflect accurately glycemic control in ESRD patients with diabetes,30,31 especially those on erythropoietin therapy.32 The relationship between glycosylated hemoglobin and dialysis outcomes in DM has not been adequately characterized. Actual levels of glycemic control are not reported in the United States Renal Data Service Medicare database. The data on the impact of glycemic control on ESRD patient survival are limited to one retrospective15 and one single-center observational study.22 We recently performed a primary data analysis of glycemic control and survival on a large national ESRD database.33 The analysis also provided new information on patient characteristics of the type I and II diabetic subpopulations with ESRD.34 This report summarizes our findings.
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
5–
original article
HgbA1c (%)
Figure 1 | Frequency distribution of HgbA1C values for all study patients.
Of the 24 744 hemodialysis patients with HgbA1c drawn, 89.6% (N ¼ 22 178) had random blood glucose measurements available from the previous 90 days. For the combined type I and type II population, the mean HgbA1c was 6.77%, and the mean random blood glucose was 168 mg/dl. A scatter plot (not shown) generated by plotting the mean blood Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
95 90 85 80 75 0
1
2 -5.5
4
5 6 7 Months
5.5–6.5
6.5–7.5
8
9 10 11 12
7.5–8.5
>8.5
100
* 30 25
Ref
1.0
**
*
*
*
20
* *
**
15
*
0.5
10 5
N = 24 875 patients with diabetes
85 80 0
1
2
3
-5.5
4 5.5–6.5
5 6 7 Months 6.5–7.5
8
9 10 11 12
7.5–8.5
1
1
>1
.5
–1 .5
–1 0
10
10
9.
5–
5
10
9
9. 9–
5
5– 8.
8
8. 8–
5 7.
5–
7–
7.
6.
5–
7
5
6
6. 6–
5
5– 5.
5. 5–
5.
0
0
-
90
75
HgbA1c (%)
Figure 3 | Impact of varying levels of HgbA1c on all-cause mortality, using Cox proportional hazard models. Data are presented as unadjusted, case-mix adjusted, and case-mix plus laboratory adjusted. Line indicates frequency distribution of HgbA1c values as shown is Figure 1.
>8.5
100
Survival (%)
c
*
0.0
95
Survival (%)
b
3
35
Unadj. CMAdj. CM + Lab Adj. % of patients
% of patients
1.5
100
Hazard ratios of death
Survival (%)
a
95 90 85 80 75 0
1
2 -5.5
3
4 5.5–6.5
5 6 7 Months 6.5–7.5
8
9 10 11 12
7.5–8.5
>8.5
Figure 2 | Survival rates in HgbA1C groups. (a–c) Survival analysis using Kaplan–Meier survival curves for all (a) study patients, and (b) type I, and (c) type II subsets.
glucose vs HgbA1c indicated that the HgbA1c correlated with the mean random blood glucose levels with an R2 value of 0.3716 and a s.e. of 1.36. The 12-month survival rates ranged from 80 to 85% between groups, corresponding to a 15–20% unadjusted 12month mortality rate. Kaplan–Meier survival curves grouped by HgbA1c levels, shown in Figure 2a, did not reach statistical significance. This finding persisted when results from type I and type II DM patients were separately analyzed (Figure 2b and c). Multivariable models indicated that there were no clear patterns between HgbA1c and death risk for all study patients, shown in Figure 3. Despite the higher HgbA1c levels found in type I patients, the models from our subset analyses by DM type were similarly without any pattern (data not shown). For patients with diabetes excluded from the study for having no HgbA1c measured during the baseline period, 16.93% died in the subsequent year, compared to 16.15% who died in the study population. DISCUSSION
The incidence of ESRD attributed to DM in the US continues to increase, linked to escalation in both the general diabetic and diabetic CKD populations. Insulin therapies remain the mainstay of glycemic control in ESRD patients with diabetes, Kidney International (2006) 70, 1503–1509
used in about 60% of patients, up to twice that of non-ESRD Medicare patients.1 However, data on the differences between type I and type II patients with diabetes on dialysis are not available. Furthermore, there is currently limited evidence on the utility of glycohemoglobin (HgbA1c) levels in ESRD as correlated with improvement in outcomes, although such data are available in the non-ESRD population.23,27 This is the first study to characterize a large population of prevalent ESRD patients with diabetes. The majority were type II patients, a proportion similar to the general diabetes population. The smaller population of type I DM patients was younger, more predominantly white, longer on dialysis, had greater elevations of serum phosphorus, and had poorer glycemic control. There were no differences in dialysis adequacy, anemia correction, albumin levels, or serum calcium. Classification of diabetes type in the general population is known to be imperfect.35 For this analysis, classification was based mainly on Medicare reporting from Form 2728 with a few de novo diagnoses of DM (newly added ICD-9 diagnoses) after the initiation of dialysis. The initial analysis revealed that 50% of type I and 34% of type II ESRD patients were above the ADA glycemic goal for HgbA1c of 7%, itself a value well above the normal range for nondiabetics.36 These data are better than or at least no different from the experience in the general population, where only 37% of adult patients with diabetes in the US are achieving the glycemic goal.4 In the general diabetic population, there is no HgbA1c level below which the risk of complications does not continue to fall.27,37 Therefore, it is not unexpected that in a recent survey on improving diabetic ESRD outcomes, nephrologists rated better glycemic control equal in importance to hypertension control, and more important than lipid control.38 The latest United States Renal Data Service report indicates that the cumulative use of insulin and oral agents in ESRD patients with diabetes increased between 2000 and 2003, which suggests a more aggressive treatment approach in diabetes management.1 1505
original article
Nonetheless, in this study, a large percentage of those with the lowest HgbA1c values were on neither insulin nor oral agents. Our data are the first indication that HgbA1c values in a large cohort of hemodialysis patients with diabetes are either similar to or range below that reported for the general diabetic population. HgbA1c correlated with mean random blood glucose levels with an R2 of 0.3716 in this study. However, the s.e. of 1.36 indicated considerable variability not explained by the random blood glucose measurements. Data from the Diabetes Control and Complication Control indicate that mean plasma glucose from multiple determinants of 170g/dl correlated with a HgbA1c of 7%,39 compared with our values of 1.68g/dl and 6.77%. This finding is interesting considering that the random blood glucose measurements in hemodialysis patients may vary considerably depending on their dialysis shift (because the blood specimens are typically drawn pre-dialysis), and on the confounding effect of timing relative to food intake and medications for diabetes. Several factors impacting on ESRD glycemic management are known to exist, including pharmacodynamic effects of uremia and/or the dialysis procedure on insulin and carbohydrate metabolism, pharmacokinetic influences on insulin and oral hypoglycemic agents used to treat hyperglycemia, and potential effects on HbgA1c, used as the basis of glycemic management.40 Glycohemoglobin measurements in this report utilized an immunoturbidimetric assay performed in a single laboratory. Unlike high-pressure liquid chromatography commonly used in routine laboratory determinations of HgbA1c, the immunoturbidimetric assay is not influenced by high urea levels or hemoglobin variants. In the general population of patients with diabetes, the HbA1c provides a glycemic history of the preceding 120 days, the average red blood cell lifespan.23 However, several reports have indicated that erythrocyte survival is reduced in ESRD, which would be expected to lower HgbA1c levels.41 Furthermore, the widespread use of erythropoietin increases the proportion of reticulocytes and ‘younger’ red blood cells in circulation with less glycemic exposure time for glycosylation to occur.32,42 Current Medicare standards call for measurement of HgbA1c every 3 months, or monthly during dosing changes or poor glycemic control, as one of the regular laboratory evaluations considered standard of care for patients with diabetes, and some suggest that more frequent testing than four times annually may be necessary in complex patients.43 However, limited outcome-based evidence is available to support such a recommendation in ESRD. Issues related to poor glycemic control in the ESRD patient with DM would be expected to include acute management problems such as fluid overload, hyperkalemia, and ketoacidosis, and certain chronic manifestations such as refractory gastroparesis and malnutrition.40,44,45 Chronic hyperglycemia has been increasingly hypothesized to contribute to coronary heart disease, the major source of morbidity and mortality, in individuals with diabetes.13 The United Kingdom Prospective Diabetes Study of patients 1506
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
with type II diabetes showed a 16% reduction (P ¼ 0.052) in myocardial infarction with intensive glycemic control after 10 years of follow-up, although the overall results for cardiovascular outcomes were considered negative.46 The European Prospective Investigation of Cancer and Medicine (EPICNorfolk) reported a risk ratio for coronary heart disease of 1.40 for each 1% increase in HgbA1c after adjustment for known cardiovascular risk factors.47 Other epidemiologic cohort studies have also suggested a positive association between HgbA1c levels and coronary risk.11 Multiple coronary factors, traditional and non-traditional, are reported to contribute to risk of cardiovascular outcomes in the diabetic and ESRD populations.48,49 We used an observational epidemiologic design to evaluate the relationship between glycemic control and subsequent mortality risk. No reliable measurements of the cardiovascular risk factors were available for analysis. Previous data on the benefit of glycemic control for patient survival in ESRD are limited.15,22 One caveat: an inherent assumption in evaluating our results is that HgbA1c as measured in the ESRD population, although potentially underestimating the true glycemic state, will nonetheless still rank glycemic control accurately for correlation studies. This assumption is consistent with previous small studies on HgbA1c in ESRD,19,41,50 although the issue remains controversial and the utility of HgbA1c in dialysis patients has not been settled.30,51 The 15–20% mortality rate in the study population is lower than the mortality rates reported by United States Renal Data Service for patients with diabetes.1 This finding was expected because of the study entry criteria that required diabetic patients to have HgbA1c values in the system within the 90-day enrolment period, thus potentially excluding unstable patients (e.g. incident patients or recently hospitalized patients) who were unable to have HgbA1c levels drawn. In addition, incident patients beginning dialysis in 2003 were excluded from the analysis. Kaplan–Meier survival curves did not reveal a trend toward improved survival for lower HgbA1c in this study. Cox multivariable models did not reveal any statistical correlation between HgbA1c and death risk for either casemix or case-mix þ laboratory-adjusted data. There are several potential explanations for this lack of correlation. (1) HgbA1c laboratory values were obtained for only a single quarter. HgbA1c tracks well with fasting glucose in individual patients in the general diabetic population,26 but timevarying glucose levels may relate better than baseline HgbA1c to outcomes.42 However, putative factors such as albumin, hemoglobin, and other variables analyzed using the same methodology show significant association with mortality.52 (2) All-cause mortality follow-up was limited to a 1-year period following the HgbA1c determinations. It is possible that longer term follow-up is required for the HgbA1c effect to become apparent. This assumes that the initial HgbA1c value is representative of long-term glycemic status and remains consistent over time. To our knowledge, this has not been explored in ESRD patients. (3) The study was not Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
designed to adjust for known coronary heart disease risk factors, although multiple risk factors (left ventricular hypertrophy, anemia, vascular calcification) are present chronically in the diabetic ESRD population. Previous studies assessing glycemic control as a modifiable cardiovascular risk factor have had similar limitations in adjusting for known CAD risk factors.13 In the general population of patients with diabetes, for example, it is understood that lowering HbgA1c to below the 7% target may not be adequate to remove the cardiovascular risk associated with hyperglycemia. Patients with diabetes whose A1c levels are between 5 and 6.9% are still at 2.5 more likely to die of cardiovascular disease than those below 5%.47 (4) Other metabolic abnormalities including those specific to ESRD accompany the diabetic ESRD condition and may exert significantly more powerful effects on mortal risk than that of poor glycemic control. The recently reported failure of statin treatment to impact on outcomes in ESRD patients with diabetes supports the multifactorial nature of cardiovascular disease in this population.53 As a corollary to no. 4 above, it is also possible that the microvascular and macrovascular disease is so far advanced in ESRD patients that improved glycemic control may no longer be as dominant an influence on mortal risk. A recent attempt by the Department of Veterans Affairs (VA) to develop evidence-based clinical practice guidelines differed from that promulgated by the ADA.54 They found the evidence for improving outcomes that support tight glycemic control (goal of HgbA1co7%) lacking for patients older than 65 years or with pre-existing severe microvascular damage (e.g. proliferative retinopathy, renal insufficiency, neuropathy) who were excluded from the definitive trials. Furthermore, the evidence base best supports stringent glycemic control for patients with extended life expectancy.55 The VA panel recommended that HgbA1c target be o8% for patients with pre-existing microvascular comorbidity and life expectancy of 5–15 years and a target of HgbA1co9% for patients with end-stage vascular disease and advanced age with life expectancy o5 years,54 with a goal of limiting the incidence of severe hyperglycemia while concomitantly avoiding the risk of hypoglycemia. A recent study found that the incidence of hypoglycemia in ESRD patients with diabetes on insulin increased when they initiated dialysis therapy, when compared to the pre-dialysis period.56 Insulin requirements dropped by 29% associated with declining residual kidney function. Apart from the changes in biologic half-life of insulin and oral hypoglycemic agents in ESRD, factors such as decreased appetite, inconsistent food intake, decreased vision, declining manual dexterity, and/or subclinical decline in cognitive function may place dialysis patients who are attempting tight glucose control at risk for hypoglycemia. HgbA1c is the gold standard for evaluating risk of complications in patients with diabetes. Further research will be required to explore the value of HgbA1c or other glycemic markers and mortality in diabetic ESRD. Single Kidney International (2006) 70, 1503–1509
measurements may not adequately reflect glycemic control for previous periods, and provide no data on glycemic control during the actual period of mortality assessment. In the UKPDS, for example, the updated mean HgbA1c more strongly related to increased risk of myocardial infarction than did the baseline HgbA1c.46 Other studies have linked cardiovascular mortality with 2-h post-prandial glucose levels.57 Finally, the pathological effects of chronic hyperglycemia on vascular tissues may occur through pathways that do not correlate with current markers of glycemic exposure. It is possible that the correlation of HgbA1c levels with more immediate outcomes such as quality of life, hospitalization, and/or specific crises such as the occurrence of severe hypoglycemia or hyperglycemia may be reasonable targets for future research. CONCLUSION
The study found that only 72% of ESRD patients with diabetes were tested with HgbA1c within a 3-month period, in a national database of patients on dialysis. The average HgbA1c and the proportion of patients that did not meet the ADA guideline targets were either lower than or similar to the general population. The study found no correlation between HgbA1c levels and subsequent 12-month mortality risk in ESRD patients, either singly or when adjusted for case mix and laboratory variables. Clinical practice recommendations by the ADA acknowledge differences in the risks and benefits of glycemic targets for individual patients.5 It is prudent to individualize the HgbA1c target in ESRD patients, perhaps becoming less stringent based on age, comorbidity, life expectancy, and the presence of risk factors for the occurrence of hypoglycemia. More studies are needed in CKD and ESRD patients in order to refine evidence-based recommendations for the appropriate application of tests (such as HgbA1c) that are currently used to define glycemic control. Similarly, more studies are needed to determine evidence-based treatment goals for glycemic control in this patient population. MATERIALS AND METHODS Patients were drawn from 76.178 ESRD patients treated with hemodialysis at Fresenius Medical Care, North America between 1 October and 31 December 2002. Of 38.701 patients with DM (50.8%), 13.827 (35.7%) were excluded for having no HgbA1C measured during the 3-month baseline period. A total of 28.159 patients had HgbA1C tests done, including 3.284 non-diabetics. Among the patients with DM, 24.875 were still being treated by hemodialysis and in the database as of 1 January 2003 (i.e. not transferred or discharged). These 24,875 hemodialysis patients with diabetes and with at least one HgbA1c value drawn from 1 October to 31 December 2002 and surviving into 1 January 2003 formed the study cohort. The diagnosis of DM was based on either diabetic cause of ESRD or on reporting of diabetes as a comorbidity diagnosis before 1 January 2003. Seventy-five percent of the 24 875 patients with diabetes had diabetes as the cause of their kidney disease based on the Medicare 2728 form. The investigators had no influence over the 1507
original article
frequency of HgbA1c determinations, which was decided by the patient’s nephrologist. All HgbA1c values were measured using the Roche Cobar Integra 800 whole-blood immunoturbidimetric assay (standardized according to the Diabetes Complications Control Trial/National Glycohemoglobin Standardization Program) performed by a single laboratory (Spectra Laboratories, Rockleigh, NJ, USA). The normal HgbA1c range by the assay was 4.5–5.7%. Patient demographic and case-mix variables were collected as of 31 December 2002. These included diabetes type (type I or II), age, gender, race, dialysis vintage, body surface area (from height and weight), vascular access type (native fistula, graft, or catheter), and the presence or absence of ethropoietin use within the prior 180 days. All available laboratory results for the 1 October–31 December 2002 for HgbA1c, eKt/V, hemoglobin, albumin, creatinine, calcium, and phosphorus were collected, and the 3-month average value was used for each patient in the analyses. In addition, all available random blood glucose measurements on patients with HgbA1c obtained in the last quarter of 2002, drawn within the previous 90 days from the date of the HgbA1c, were analyzed. Most patients (75.8%) had three or more determinations. All-cause mortality data as the final outcome were collected for the 12-month follow-up period from 1 January to 31 December 2003. The patient variables were grouped into three categories: all study patients; and type I and type II DM subsets. Data for continuous variables are presented as mean7s.d. Demographic and clinical characteristics were compared between groups using w2 tests for categorical variables and unpaired t-tests for continuous variables. Kaplan–Meier curves were constructed for survival analysis, and Cox proportional hazard models were used to compare the impact on mortality by varying levels of HgbA1c. The multivariable models are presented as case mix (age, gender, etc) and case mix þ laboratory-adjusted results. The HgbA1c reference range used for all analyses was 6.5–7.0%, consistent with the ADA recommendation to keep HgbA1c below 7%. All statistical analyses were performed using SAS software version 9.2 (Cary, NC, USA). REFERENCES 1. System USRD: USRDS 2005 Annual Data Report. Atlas of End-Stage Renal Disease in the United States. National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases: Bethesda, MD 2005. 2. Koro CE, Bowlin SJ, Bourgeois N et al. Glycemic control from 1988 to 2000 among US adults diagnosed with type 2 diabetes: a preliminary report. Diabetes Care 2004; 27: 17–20. 3. Liebl A, Mata M, Eschwege E. Evaluation of risk factors for development of complications in Type II diabetes in Europe. Diabetologia 2002; 45: S23–28. 4. Saydah SH, Fradkin J, Cowie CC. Poor control of risk factors for vascular disease among adults with previously diagnosed diabetes. JAMA 2004; 291: 335–342. 5. American Diabetes Association. Tests of glycemia in diabetes (position statement). Diabetes Care 2004; 27: S91–S93. 6. Wei M, Gaskill SP, Haffner SM et al. Effects of diabetes and level of glycemia on all-cause and cardiovascular mortality. The San Antonio Heart Study.. Diabetes Care 1998; 21: 1167–1172. 7. Hanefeld M, Fischer S, Julius U et al. Risk factors for myocardial infarction and death in newly detected NIDDM: the Diabetes Intervention Study, 11-year follow-up. Diabetologia 1996; 39: 1577–1583. 8. Kuusisto J, Mykkanen L, Pyorala K et al. NIDDM and its metabolic control predict coronary heart disease in elderly subjects. Diabetes 1994; 43: 960–967. 9. Andersson DK, Svardsudd K. Long-term glycemic control relates to mortality in type II diabetes. Diabetes Care 1995; 18: 1534–1543. 10. Engelgau MM, Geiss LS, Saaddine JB et al. The evolving diabetes burden in the United States. Ann Intern Med 2004; 140: 945–950.
1508
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
11.
12.
13.
14.
15.
16.
17.
18. 19.
20. 21. 22.
23. 24.
25.
26.
27. 28. 29.
30.
31. 32.
33.
34. 35. 36. 37.
38.
Selvin E, Marinopoulos S, Berkenblit G et al. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med 2004; 141: 421–431. Menon V, Greene T, Pereira AA et al. Glycosylated hemoglobin and mortality in patients with nondiabetic chronic kidney disease. J Am Soc Nephrol 2005; 16: 3411–3417. Selvin E, Coresh J, Golden SH et al. Glycemic control and coronary heart disease risk in persons with and without diabetes: the atherosclerosis risk in communities study. Arch Intern Med 2005; 165: 1910–1916. Manske CL. Hyperglycemia and intensive glycemic control in diabetic patients with chronic renal disease. Am J Kidney Dis 1998; 32: S157–S171. Tzamaloukas AH, Murata GH, Zager PG et al. The relationship between glycemic control and morbidity and mortality for diabetics on dialysis. Asaio J 1993; 39: 880–885. Malmberg K. Prospective randomised study of intensive insulin treatment on long term survival after acute myocardial infarction in patients with diabetes mellitus. DIGAMI (Diabetes Mellitus, Insulin Glucose Infusion in Acute Myocardial Infarction) Study Group. BMJ 1997; 314: 1512–1515. Malmberg K, Norhammar A, Wedel H et al. Glycometabolic state at admission: important risk marker of mortality in conventionally treated patients with diabetes mellitus and acute myocardial infarction: longterm results from the Diabetes and Insulin–Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study. Circulation 1999; 99: 2626–2632. Ceriello A, Hanefeld M, Leiter L et al. Postprandial glucose regulation and diabetic complications. Arch Intern Med 2004; 164: 2090–2095. Joy MS, Cefalu WT, Hogan SL et al. Long-term glycemic control measurements in diabetic patients receiving hemodialysis. Am J Kidney Dis 2002; 39: 297–307. American Diabetes Association. Standards of medical care in diabetes (position statement). Diabetes Care 2004; 27: S15–S35. Akmal M. Hemodialysis in diabetic patients. Am J Kidney Dis 2001; 38: S195–199. Morioka T, Emoto M, Tabata T et al. Glycemic control is a predictor of survival for diabetic patients on hemodialysis. Diabetes Care 2001; 24: 909–913. Goldstein DE, Little RR, Lorenz RA et al. Tests of glycemia in diabetes. Diabetes Care 2004; 27: 1761–1773. Larsen ML, Horder M, Mogensen EF. Effect of long-term monitoring of glycosylated hemoglobin levels in insulin-dependent diabetes mellitus. N Engl J Med 1990; 323: 1021–1025. Dunn PJ, Cole RA, Soeldner JS et al. Reproducibility of hemoglobin AIc and sensitivity to various degrees of glucose intolerance. Ann Intern Med 1979; 91: 390–396. Meigs JB, Nathan DM, Cupples LA et al. Tracking of glycated hemoglobin in the original cohort of the Framingham Heart Study. J Clin Epidemiol 1996; 49: 411–417. American Diabetes Association. Standards of medical care in diabetes (position statement). Diabetes Care 2005; 28: S4–S36. American Diabetes Association. Clinical practice recommendations. Diabetes Care 2006; 29: S3. Collins AJ, Li S, Gilbertson DT et al. Chronic kidney disease and cardiovascular disease in the Medicare population. Kidney Int Suppl 2003: S24–S31. Morgan L, Marenah CB, Jeffcoate WJ et al. Glycated proteins as indices of glycaemic control in diabetic patients with chronic renal failure. Diabet Med 1996; 13: 514–519. Stettler C, Mueller B, Diem P. What you always wanted to know about HbA1c. Schweiz Med Wochenschr 2000; 130: 993–1005. Nakao T, Matsumoto H, Okada T et al. Influence of erythropoietin treatment on hemoglobin A1c levels in patients with chronic renal failure on hemodialysis. Intern Med 1998; 37: 826–830. Williams ME LE, Lazarus JM et al. Glycohemoglobin levels and the subsequent impact on mortality risk in diabetic hemodialysis (HD) patients. J Am Soc Nephrol 2005; 16: 723A–724A. Williams ME LE, Lazarus JM et al. Differences between Type 1 and Type 2 DM patients on hemodialysis. J Am Soc Nephrol 2005; 16: 723A. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 2003; 26(Suppl 1): S5–S20. Feinglos MN, Bethel MA. Emerging care for type 2 diabetes: using insulin to reach lower glycemic goals. Cleveland Clin J Med 2005; 72: 791–799. The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial. Diabetes 1996; 45: 1289–1298. Flauto RP, Leon JB, Sehgal AR. The provision and outcomes of diabetic care of hemodialysis patients. Am J Kidney Dis 2003; 41: 125–131.
Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
39.
40. 41.
42.
43. 44. 45.
46.
47.
48. 49.
Rohlfing CL, Wiedmeyer HM, Little RR et al. Defining the relationship between plasma glucose and HbA(1c): analysis of glucose profiles and HbA(1c) in the Diabetes Control and Complications Trial. Diabetes Care 2002; 25: 275–278. Tuohy KWM. Management of the diabetic ESRD Patient: Dialysis and transplantation. Elsevier: Philadelphia, 2005. Little RR, Tennill AL, Rohlfing C et al. Can glycohemoglobin be used to assess glycemic control in patients with chronic renal failure? Clin Chem 2002; 48: 784–786. Uehlinger DE, Gotch FA, Sheiner LB. A pharmacodynamic model of erythropoietin therapy for uremic anemia. Clin Pharmacol Ther 1992; 51: 76–89. www.cmms.hhs.gov. US Department of Health and Human Services. Mak RH. Impact of end-stage renal disease and dialysis on glycemic control. Semin Dial 2000; 13: 4–8. Ifudu O, Dulin AL, Friedman EA. Interdialytic weight gain correlates with glycosylated hemoglobin in diabetic hemodialysis patients. Am J Kidney Dis 1994; 23: 686–691. Stratton IM, Adler AI, Neil HA et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405–412. Khaw KT, Wareham N, Luben R et al. Glycated haemoglobin, diabetes, and mortality in men in Norfolk cohort of european prospective investigation of cancer and nutrition (EPIC-Norfolk). BMJ 2001; 322: 15–18. Sarnak MJ, Levey AS. Cardiovascular disease and chronic renal disease: a new paradigm. Am J Kidney Dis 2000; 35: S117–131. Block GA, Klassen PS, Lazarus JM et al. Mineral metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc Nephrol 2004; 15: 2208–2218.
Kidney International (2006) 70, 1503–1509
50.
51.
52.
53.
54.
55.
56.
57.
Ichikawa H, Nagake Y, Takahashi M et al. What is the best index of glycemic control in patients with diabetes mellitus on hemodialysis? Nippon Jinzo Gakkai Shi 1996; 38: 305–308. Ansari A, Thomas S, Goldsmith D. Assessing glycemic control in patients with diabetes and end-stage renal failure. Am J Kidney Dis 2003; 41: 523–531. Lowrie EG, Lew NL. Death risk in hemodialysis patients: the predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 1990; 15: 458–482. Wanner C, Krane V, Marz W et al. Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med 2005; 353: 238–248. Pogach LM, Brietzke SA, Cowan Jr CL et al. Development of evidence-based clinical practice guidelines for diabetes: the Department of Veterans Affairs/Department of Defense guidelines initiative. Diabetes Care 2004; 27(Suppl 2): B82–B89. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998; 352: 854–865. Loipl J, Schmekal B, Biesenbach G. Long-term impact of chronic hemodialysis on glycemic control and serum lipids in insulin-treated type 2-diabetic patients. Renal Fail 2005; 27: 305–308. Glucose tolerance and mortality: comparison of WHO and American Diabetes Association diagnostic criteria. The DECODE study group. European Diabetes Epidemiology Group. Diabetes Epidemiology: Collaborative analysis Of Diagnostic criteria in Europe. Lancet 1999; 354: 617–621.
1509
http://www.kidney-international.org & 2006 International Society of Nephrology
see commentary on page 1392
Hemodialyzed type I and type II diabetic patients in the US: Characteristics, glycemic control, and survival ME Williams1, E Lacson Jr2, M Teng2, N Ofsthun2 and JM Lazarus2 1
Renal Unit, Joslin Diabetes Center, One Joslin Place, Boston, Massachusetts, USA and 2Fresenius Medical Care, Lexington, Massachusetts, USA
Diabetes mellitus (DM) constitutes a major end-stage renal disease (ESRD) health problem. Glycemic control is fundamental to the management of diabetes and its complications, and relies on monitoring of hyperglycemia. We therefore performed a primary data analysis of glycemic control and survival on a large national ESRD database. Ninety-five percent of patients with DM had type II diabetes (N ¼ 23 504), and five percent had type I diabetes (N ¼ 1 371). For the combined population, the mean hemoglobin A1c (HgbA1c) was 6.77%, and the mean random blood glucose was 168 mg/dl. Mean HgbA1c values were 47.0% in 35% and 48.5% in 14%. Mean HgbA1c values were below 5% in 11.3% of patients. Type I study patients tended to have higher HgbA1c values. Most patients (75.8%) had three or more random blood glucose determinations within 90 days preceding the HgbA1c measurement. The HgbA1c showed only a weak correlation with mean random glucose values (R2 0.3716; s.e. ¼ 1.36). The survival rates in the subsequent 12-month period ranged from 80 to 85% across different HgbA1c strata. Kaplan–Meier survival curves grouped by HgbA1c levels showed no correlation between HgbA1c and survival at 12 months. More studies are needed to refine recommendations for the role of HgbA1c and glycemic control in this patient population. Kidney International (2006) 70, 1503–1509. doi:10.1038/sj.ki.5001789; published online 30 August 2006 KEYWORDS: diabetes; hemodialysis; hyperglycemia; survival; hemoglobin A1c; ESRD
Correspondence: ME Williams, Renal Unit, Joslin Diabetes Center, One Joslin Place, Boston Massachusetts, USA. E-mail: [email protected] Received 16 March 2006; revised 20 June 2006; accepted 27 June 2006; published online 30 August 2006 Kidney International (2006) 70, 1503–1509
Patients with diabetes mellitus (DM) now account for 45 per cent of the prevalent end-stage renal disease (ESRD) population in the US, and the rate of ESRD caused by diabetes increased 86% between 1993 and 2003.1 Most incident patients with ESRD and diabetes carry multiple chronic comorbid conditions such as ischemic heart disease, congestive heart failure, and peripheral vascular disease when they begin dialysis. Macrovascular complications, including myocardial infarction, stroke, and peripheral vascular disease, are the major causes of morbidity and mortality in patients with ESRD and diabetes. Although only a minority of patients with diabetes achieve recommended goals for glycemic control,2–4 glycemic control is fundamental to the management of diabetes5 and its complications. Poor glycemic control is increasingly considered as an independent cardiovascular risk factor in diabetes.4,6–9 Among patients with diabetes in a large prospective study, high blood glucose concentrations were associated with a greater incidence of cardiovascular disease.10 Recent data suggest that glycosylated hemoglobin levels are related to coronary heart disease risk in persons with type I and II diabetes.11 In fact, chronic hyperglycemia has been hypothesized to contribute to coronary disease in individuals without diabetes as well.12,13 Limited data suggest that poor glycemic control in patients with diabetes on dialysis may be associated with increased morbidity from diabetic vascular complications.14,15 Recent data demonstrate that glycemic control may benefit patients with diabetes who have established coronary disease.16,17 Epidemiologic studies suggest the potential for improved glycemic control to reduce cardiovascular complications in diabetes, hence the benefit of hyperglycemia control on cardiovascular damage in patients with diabetes continues to be studied.18,19 American Diabetes Association (ADA) guidelines indicate that improved glycemic control may lower the risk of myocardial infarction and cardiovascular death.20 Among risk factor reductions for mortality in dialysis patients with diabetes, improved glycemic control is widely recommended.21 However, whether glycemic control affects survival in diabetic ESRD patients remains unclear.22 Monitoring of hyperglycemia is at the center of management of diabetes.23,24 The hemoglobin A1c (HgbA1c) concentration is a marker of hyperglycemia and reflects average blood glucose concentration over 1503
Table 1 | Study patient characteristics with a breakdown into types I and II DM (values in parenthesis are s.d.) DM patients with HgbA1c (drawn between 1 October 2002 and 31 December2002)
Characteristics No. of patients Age* (years) % female Race* White (%) Black (%) Other (%) Vintage** (days) Body surface area* (m2) % of patients on Erythropoietin access type*** Fistula (%) Graft (%) Catheter (%) Blank (%) HD dose (eKt/V) Albumin Hemoglobin Calcium Phosphorus* Creatinine White blood cell count** HgbA1c (%)
1504
Type I DM
Type II DM
24 875 63.7 (12.1) 51.5 — 53.0 36.4 10.6 1051 (951) 1.86 (0.25) 99.0
1371 49.3 (14.4) 49.5 — 60.6 32.5 6.9 1172 (1.075) 1.82 (0.25) 98.7
23 504 64.5 (11.4) 51.6 — 52.6 36.6 10.8 1044 (943) 1.86 (0.25) 99.0
— 30.0 45.7 23.5 0.8 1.41 (0.29) 3.82 (0.38) 11.72 (1.10) 9.29 (0.70) 5.60 (1.44) 8.23 (2.67) 7.77 (2.87) 6.77 (1.44)
— 34.4 40.8 24.4 0.4 1.41 (0.32) 3.83 (0.42) 11.71 (1.17) 9.27 (0.73) 5.89 (1.55) 8.60 (2.61) 7.97 (2.39) 7.49 (1.99)
— 29.7 46.0 23.5 0.8 1.41 (0.28) 3.81 (0.38) 11.72 (1.09) 9.29 (0.70) 5.59 (1.44) 8.20 (2.67) 7.76 (2.89) 6.73 (1.68)
DM, diabetes mellitus; HD, hemodialysis; HgbA1c, hemoglobin A1c. *Po0.0001; **Po0.0016; ***P=0.003: All comparing type I vs type II DM patients only.
Hemoglobin A1c frequency distribution for Q4 2002 20 N = 24 875 patients with diabetes
RESULTS % of patients
15
10
5
5 5– 6 6– 6. 5 6. 5– 7 7– 7. 5 7. 5– 8 8– 8. 5 8. 5– 9 9– 9. 9. 5 5– 1 10 0 –1 0 10 .5 .5 –1 1 >1 1 5.
0
0 5.
Patient characteristics are shown in Table 1. Among the 24 875 eligible patients, a total of 1371 (5.5%) had type I diabetes and 23 504 (94.5%) had type II diabetes. Type I study patients were significantly different from the type II study patients, with younger age (Po0.0001), different racial distribution with more white race (Po0.0001), longer dialysis vintage (Po0.0016), slightly lower body surface area (Po0.0001), and different distribution of vascular accesses (P ¼ 0.003). Type I study patients had higher serum phosphorus levels (Po0.0016). Type I study patients had higher mean HgbA1c values (7.5 vs 6.7%, Po0.0001). The frequency distribution of the HgbA1c results is shown in Figure 1. For all study patients, mean HgbA1c values were 47.0% in 35% and 48.5% in 14%; in 11.3% of patients, mean HgbA1c values were below 5%. Overall, patients with type I DM tended to have higher HgbA1c values, with mean HgbA1c47.0% in 50% and 48.5% in 426% of type I study patients. Of the entire study cohort, 41% were treated with insulin, 20% with oral hypoglycemia agents, 16% with both, and 23% with neither. For those with HgbA1co6%, 37% were on neither insulin nor oral hypoglycemic agents.
All study patients
5.
3 months in diabetic individuals.25 HgbA1c tracks well in individuals over time.26 The ADA recommends using pointof-care glycohemoglobin (HgbA1c) values to track glycemic control, make timely decisions on treatment, and to reduce risks of adverse outcomes by lowering A1c levels to below 7%.27,28 However, less than half of patients with ESRD and diabetes fulfill the recommendations of the ADA to undergo at least four glycohemoglobin tests per year, and up to a quarter receive no testing.1 Monitoring of capillary blood glucose is also recommended, with a frequency and timing dependent on the patient’s medical condition. However, half of chronic kidney patients with diabetes have not had glycosylated hemoglobin testing in the year before dialysis initiation.29 Diabetic monitoring in the ESRD population is also suboptimal, with over 60% not prescribed any diabetic test strips and less than half receiving the four or more HgbA1c tests per year recommended by the ADA1. The apparent variability in HgbA1c testing practices in ESRD may derive from the lack of consensus on HgbA1c testing to reflect accurately glycemic control in ESRD patients with diabetes,30,31 especially those on erythropoietin therapy.32 The relationship between glycosylated hemoglobin and dialysis outcomes in DM has not been adequately characterized. Actual levels of glycemic control are not reported in the United States Renal Data Service Medicare database. The data on the impact of glycemic control on ESRD patient survival are limited to one retrospective15 and one single-center observational study.22 We recently performed a primary data analysis of glycemic control and survival on a large national ESRD database.33 The analysis also provided new information on patient characteristics of the type I and II diabetic subpopulations with ESRD.34 This report summarizes our findings.
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
5–
original article
HgbA1c (%)
Figure 1 | Frequency distribution of HgbA1C values for all study patients.
Of the 24 744 hemodialysis patients with HgbA1c drawn, 89.6% (N ¼ 22 178) had random blood glucose measurements available from the previous 90 days. For the combined type I and type II population, the mean HgbA1c was 6.77%, and the mean random blood glucose was 168 mg/dl. A scatter plot (not shown) generated by plotting the mean blood Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
95 90 85 80 75 0
1
2 -5.5
4
5 6 7 Months
5.5–6.5
6.5–7.5
8
9 10 11 12
7.5–8.5
>8.5
100
* 30 25
Ref
1.0
**
*
*
*
20
* *
**
15
*
0.5
10 5
N = 24 875 patients with diabetes
85 80 0
1
2
3
-5.5
4 5.5–6.5
5 6 7 Months 6.5–7.5
8
9 10 11 12
7.5–8.5
1
1
>1
.5
–1 .5
–1 0
10
10
9.
5–
5
10
9
9. 9–
5
5– 8.
8
8. 8–
5 7.
5–
7–
7.
6.
5–
7
5
6
6. 6–
5
5– 5.
5. 5–
5.
0
0
-
90
75
HgbA1c (%)
Figure 3 | Impact of varying levels of HgbA1c on all-cause mortality, using Cox proportional hazard models. Data are presented as unadjusted, case-mix adjusted, and case-mix plus laboratory adjusted. Line indicates frequency distribution of HgbA1c values as shown is Figure 1.
>8.5
100
Survival (%)
c
*
0.0
95
Survival (%)
b
3
35
Unadj. CMAdj. CM + Lab Adj. % of patients
% of patients
1.5
100
Hazard ratios of death
Survival (%)
a
95 90 85 80 75 0
1
2 -5.5
3
4 5.5–6.5
5 6 7 Months 6.5–7.5
8
9 10 11 12
7.5–8.5
>8.5
Figure 2 | Survival rates in HgbA1C groups. (a–c) Survival analysis using Kaplan–Meier survival curves for all (a) study patients, and (b) type I, and (c) type II subsets.
glucose vs HgbA1c indicated that the HgbA1c correlated with the mean random blood glucose levels with an R2 value of 0.3716 and a s.e. of 1.36. The 12-month survival rates ranged from 80 to 85% between groups, corresponding to a 15–20% unadjusted 12month mortality rate. Kaplan–Meier survival curves grouped by HgbA1c levels, shown in Figure 2a, did not reach statistical significance. This finding persisted when results from type I and type II DM patients were separately analyzed (Figure 2b and c). Multivariable models indicated that there were no clear patterns between HgbA1c and death risk for all study patients, shown in Figure 3. Despite the higher HgbA1c levels found in type I patients, the models from our subset analyses by DM type were similarly without any pattern (data not shown). For patients with diabetes excluded from the study for having no HgbA1c measured during the baseline period, 16.93% died in the subsequent year, compared to 16.15% who died in the study population. DISCUSSION
The incidence of ESRD attributed to DM in the US continues to increase, linked to escalation in both the general diabetic and diabetic CKD populations. Insulin therapies remain the mainstay of glycemic control in ESRD patients with diabetes, Kidney International (2006) 70, 1503–1509
used in about 60% of patients, up to twice that of non-ESRD Medicare patients.1 However, data on the differences between type I and type II patients with diabetes on dialysis are not available. Furthermore, there is currently limited evidence on the utility of glycohemoglobin (HgbA1c) levels in ESRD as correlated with improvement in outcomes, although such data are available in the non-ESRD population.23,27 This is the first study to characterize a large population of prevalent ESRD patients with diabetes. The majority were type II patients, a proportion similar to the general diabetes population. The smaller population of type I DM patients was younger, more predominantly white, longer on dialysis, had greater elevations of serum phosphorus, and had poorer glycemic control. There were no differences in dialysis adequacy, anemia correction, albumin levels, or serum calcium. Classification of diabetes type in the general population is known to be imperfect.35 For this analysis, classification was based mainly on Medicare reporting from Form 2728 with a few de novo diagnoses of DM (newly added ICD-9 diagnoses) after the initiation of dialysis. The initial analysis revealed that 50% of type I and 34% of type II ESRD patients were above the ADA glycemic goal for HgbA1c of 7%, itself a value well above the normal range for nondiabetics.36 These data are better than or at least no different from the experience in the general population, where only 37% of adult patients with diabetes in the US are achieving the glycemic goal.4 In the general diabetic population, there is no HgbA1c level below which the risk of complications does not continue to fall.27,37 Therefore, it is not unexpected that in a recent survey on improving diabetic ESRD outcomes, nephrologists rated better glycemic control equal in importance to hypertension control, and more important than lipid control.38 The latest United States Renal Data Service report indicates that the cumulative use of insulin and oral agents in ESRD patients with diabetes increased between 2000 and 2003, which suggests a more aggressive treatment approach in diabetes management.1 1505
original article
Nonetheless, in this study, a large percentage of those with the lowest HgbA1c values were on neither insulin nor oral agents. Our data are the first indication that HgbA1c values in a large cohort of hemodialysis patients with diabetes are either similar to or range below that reported for the general diabetic population. HgbA1c correlated with mean random blood glucose levels with an R2 of 0.3716 in this study. However, the s.e. of 1.36 indicated considerable variability not explained by the random blood glucose measurements. Data from the Diabetes Control and Complication Control indicate that mean plasma glucose from multiple determinants of 170g/dl correlated with a HgbA1c of 7%,39 compared with our values of 1.68g/dl and 6.77%. This finding is interesting considering that the random blood glucose measurements in hemodialysis patients may vary considerably depending on their dialysis shift (because the blood specimens are typically drawn pre-dialysis), and on the confounding effect of timing relative to food intake and medications for diabetes. Several factors impacting on ESRD glycemic management are known to exist, including pharmacodynamic effects of uremia and/or the dialysis procedure on insulin and carbohydrate metabolism, pharmacokinetic influences on insulin and oral hypoglycemic agents used to treat hyperglycemia, and potential effects on HbgA1c, used as the basis of glycemic management.40 Glycohemoglobin measurements in this report utilized an immunoturbidimetric assay performed in a single laboratory. Unlike high-pressure liquid chromatography commonly used in routine laboratory determinations of HgbA1c, the immunoturbidimetric assay is not influenced by high urea levels or hemoglobin variants. In the general population of patients with diabetes, the HbA1c provides a glycemic history of the preceding 120 days, the average red blood cell lifespan.23 However, several reports have indicated that erythrocyte survival is reduced in ESRD, which would be expected to lower HgbA1c levels.41 Furthermore, the widespread use of erythropoietin increases the proportion of reticulocytes and ‘younger’ red blood cells in circulation with less glycemic exposure time for glycosylation to occur.32,42 Current Medicare standards call for measurement of HgbA1c every 3 months, or monthly during dosing changes or poor glycemic control, as one of the regular laboratory evaluations considered standard of care for patients with diabetes, and some suggest that more frequent testing than four times annually may be necessary in complex patients.43 However, limited outcome-based evidence is available to support such a recommendation in ESRD. Issues related to poor glycemic control in the ESRD patient with DM would be expected to include acute management problems such as fluid overload, hyperkalemia, and ketoacidosis, and certain chronic manifestations such as refractory gastroparesis and malnutrition.40,44,45 Chronic hyperglycemia has been increasingly hypothesized to contribute to coronary heart disease, the major source of morbidity and mortality, in individuals with diabetes.13 The United Kingdom Prospective Diabetes Study of patients 1506
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
with type II diabetes showed a 16% reduction (P ¼ 0.052) in myocardial infarction with intensive glycemic control after 10 years of follow-up, although the overall results for cardiovascular outcomes were considered negative.46 The European Prospective Investigation of Cancer and Medicine (EPICNorfolk) reported a risk ratio for coronary heart disease of 1.40 for each 1% increase in HgbA1c after adjustment for known cardiovascular risk factors.47 Other epidemiologic cohort studies have also suggested a positive association between HgbA1c levels and coronary risk.11 Multiple coronary factors, traditional and non-traditional, are reported to contribute to risk of cardiovascular outcomes in the diabetic and ESRD populations.48,49 We used an observational epidemiologic design to evaluate the relationship between glycemic control and subsequent mortality risk. No reliable measurements of the cardiovascular risk factors were available for analysis. Previous data on the benefit of glycemic control for patient survival in ESRD are limited.15,22 One caveat: an inherent assumption in evaluating our results is that HgbA1c as measured in the ESRD population, although potentially underestimating the true glycemic state, will nonetheless still rank glycemic control accurately for correlation studies. This assumption is consistent with previous small studies on HgbA1c in ESRD,19,41,50 although the issue remains controversial and the utility of HgbA1c in dialysis patients has not been settled.30,51 The 15–20% mortality rate in the study population is lower than the mortality rates reported by United States Renal Data Service for patients with diabetes.1 This finding was expected because of the study entry criteria that required diabetic patients to have HgbA1c values in the system within the 90-day enrolment period, thus potentially excluding unstable patients (e.g. incident patients or recently hospitalized patients) who were unable to have HgbA1c levels drawn. In addition, incident patients beginning dialysis in 2003 were excluded from the analysis. Kaplan–Meier survival curves did not reveal a trend toward improved survival for lower HgbA1c in this study. Cox multivariable models did not reveal any statistical correlation between HgbA1c and death risk for either casemix or case-mix þ laboratory-adjusted data. There are several potential explanations for this lack of correlation. (1) HgbA1c laboratory values were obtained for only a single quarter. HgbA1c tracks well with fasting glucose in individual patients in the general diabetic population,26 but timevarying glucose levels may relate better than baseline HgbA1c to outcomes.42 However, putative factors such as albumin, hemoglobin, and other variables analyzed using the same methodology show significant association with mortality.52 (2) All-cause mortality follow-up was limited to a 1-year period following the HgbA1c determinations. It is possible that longer term follow-up is required for the HgbA1c effect to become apparent. This assumes that the initial HgbA1c value is representative of long-term glycemic status and remains consistent over time. To our knowledge, this has not been explored in ESRD patients. (3) The study was not Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
designed to adjust for known coronary heart disease risk factors, although multiple risk factors (left ventricular hypertrophy, anemia, vascular calcification) are present chronically in the diabetic ESRD population. Previous studies assessing glycemic control as a modifiable cardiovascular risk factor have had similar limitations in adjusting for known CAD risk factors.13 In the general population of patients with diabetes, for example, it is understood that lowering HbgA1c to below the 7% target may not be adequate to remove the cardiovascular risk associated with hyperglycemia. Patients with diabetes whose A1c levels are between 5 and 6.9% are still at 2.5 more likely to die of cardiovascular disease than those below 5%.47 (4) Other metabolic abnormalities including those specific to ESRD accompany the diabetic ESRD condition and may exert significantly more powerful effects on mortal risk than that of poor glycemic control. The recently reported failure of statin treatment to impact on outcomes in ESRD patients with diabetes supports the multifactorial nature of cardiovascular disease in this population.53 As a corollary to no. 4 above, it is also possible that the microvascular and macrovascular disease is so far advanced in ESRD patients that improved glycemic control may no longer be as dominant an influence on mortal risk. A recent attempt by the Department of Veterans Affairs (VA) to develop evidence-based clinical practice guidelines differed from that promulgated by the ADA.54 They found the evidence for improving outcomes that support tight glycemic control (goal of HgbA1co7%) lacking for patients older than 65 years or with pre-existing severe microvascular damage (e.g. proliferative retinopathy, renal insufficiency, neuropathy) who were excluded from the definitive trials. Furthermore, the evidence base best supports stringent glycemic control for patients with extended life expectancy.55 The VA panel recommended that HgbA1c target be o8% for patients with pre-existing microvascular comorbidity and life expectancy of 5–15 years and a target of HgbA1co9% for patients with end-stage vascular disease and advanced age with life expectancy o5 years,54 with a goal of limiting the incidence of severe hyperglycemia while concomitantly avoiding the risk of hypoglycemia. A recent study found that the incidence of hypoglycemia in ESRD patients with diabetes on insulin increased when they initiated dialysis therapy, when compared to the pre-dialysis period.56 Insulin requirements dropped by 29% associated with declining residual kidney function. Apart from the changes in biologic half-life of insulin and oral hypoglycemic agents in ESRD, factors such as decreased appetite, inconsistent food intake, decreased vision, declining manual dexterity, and/or subclinical decline in cognitive function may place dialysis patients who are attempting tight glucose control at risk for hypoglycemia. HgbA1c is the gold standard for evaluating risk of complications in patients with diabetes. Further research will be required to explore the value of HgbA1c or other glycemic markers and mortality in diabetic ESRD. Single Kidney International (2006) 70, 1503–1509
measurements may not adequately reflect glycemic control for previous periods, and provide no data on glycemic control during the actual period of mortality assessment. In the UKPDS, for example, the updated mean HgbA1c more strongly related to increased risk of myocardial infarction than did the baseline HgbA1c.46 Other studies have linked cardiovascular mortality with 2-h post-prandial glucose levels.57 Finally, the pathological effects of chronic hyperglycemia on vascular tissues may occur through pathways that do not correlate with current markers of glycemic exposure. It is possible that the correlation of HgbA1c levels with more immediate outcomes such as quality of life, hospitalization, and/or specific crises such as the occurrence of severe hypoglycemia or hyperglycemia may be reasonable targets for future research. CONCLUSION
The study found that only 72% of ESRD patients with diabetes were tested with HgbA1c within a 3-month period, in a national database of patients on dialysis. The average HgbA1c and the proportion of patients that did not meet the ADA guideline targets were either lower than or similar to the general population. The study found no correlation between HgbA1c levels and subsequent 12-month mortality risk in ESRD patients, either singly or when adjusted for case mix and laboratory variables. Clinical practice recommendations by the ADA acknowledge differences in the risks and benefits of glycemic targets for individual patients.5 It is prudent to individualize the HgbA1c target in ESRD patients, perhaps becoming less stringent based on age, comorbidity, life expectancy, and the presence of risk factors for the occurrence of hypoglycemia. More studies are needed in CKD and ESRD patients in order to refine evidence-based recommendations for the appropriate application of tests (such as HgbA1c) that are currently used to define glycemic control. Similarly, more studies are needed to determine evidence-based treatment goals for glycemic control in this patient population. MATERIALS AND METHODS Patients were drawn from 76.178 ESRD patients treated with hemodialysis at Fresenius Medical Care, North America between 1 October and 31 December 2002. Of 38.701 patients with DM (50.8%), 13.827 (35.7%) were excluded for having no HgbA1C measured during the 3-month baseline period. A total of 28.159 patients had HgbA1C tests done, including 3.284 non-diabetics. Among the patients with DM, 24.875 were still being treated by hemodialysis and in the database as of 1 January 2003 (i.e. not transferred or discharged). These 24,875 hemodialysis patients with diabetes and with at least one HgbA1c value drawn from 1 October to 31 December 2002 and surviving into 1 January 2003 formed the study cohort. The diagnosis of DM was based on either diabetic cause of ESRD or on reporting of diabetes as a comorbidity diagnosis before 1 January 2003. Seventy-five percent of the 24 875 patients with diabetes had diabetes as the cause of their kidney disease based on the Medicare 2728 form. The investigators had no influence over the 1507
original article
frequency of HgbA1c determinations, which was decided by the patient’s nephrologist. All HgbA1c values were measured using the Roche Cobar Integra 800 whole-blood immunoturbidimetric assay (standardized according to the Diabetes Complications Control Trial/National Glycohemoglobin Standardization Program) performed by a single laboratory (Spectra Laboratories, Rockleigh, NJ, USA). The normal HgbA1c range by the assay was 4.5–5.7%. Patient demographic and case-mix variables were collected as of 31 December 2002. These included diabetes type (type I or II), age, gender, race, dialysis vintage, body surface area (from height and weight), vascular access type (native fistula, graft, or catheter), and the presence or absence of ethropoietin use within the prior 180 days. All available laboratory results for the 1 October–31 December 2002 for HgbA1c, eKt/V, hemoglobin, albumin, creatinine, calcium, and phosphorus were collected, and the 3-month average value was used for each patient in the analyses. In addition, all available random blood glucose measurements on patients with HgbA1c obtained in the last quarter of 2002, drawn within the previous 90 days from the date of the HgbA1c, were analyzed. Most patients (75.8%) had three or more determinations. All-cause mortality data as the final outcome were collected for the 12-month follow-up period from 1 January to 31 December 2003. The patient variables were grouped into three categories: all study patients; and type I and type II DM subsets. Data for continuous variables are presented as mean7s.d. Demographic and clinical characteristics were compared between groups using w2 tests for categorical variables and unpaired t-tests for continuous variables. Kaplan–Meier curves were constructed for survival analysis, and Cox proportional hazard models were used to compare the impact on mortality by varying levels of HgbA1c. The multivariable models are presented as case mix (age, gender, etc) and case mix þ laboratory-adjusted results. The HgbA1c reference range used for all analyses was 6.5–7.0%, consistent with the ADA recommendation to keep HgbA1c below 7%. All statistical analyses were performed using SAS software version 9.2 (Cary, NC, USA). REFERENCES 1. System USRD: USRDS 2005 Annual Data Report. Atlas of End-Stage Renal Disease in the United States. National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases: Bethesda, MD 2005. 2. Koro CE, Bowlin SJ, Bourgeois N et al. Glycemic control from 1988 to 2000 among US adults diagnosed with type 2 diabetes: a preliminary report. Diabetes Care 2004; 27: 17–20. 3. Liebl A, Mata M, Eschwege E. Evaluation of risk factors for development of complications in Type II diabetes in Europe. Diabetologia 2002; 45: S23–28. 4. Saydah SH, Fradkin J, Cowie CC. Poor control of risk factors for vascular disease among adults with previously diagnosed diabetes. JAMA 2004; 291: 335–342. 5. American Diabetes Association. Tests of glycemia in diabetes (position statement). Diabetes Care 2004; 27: S91–S93. 6. Wei M, Gaskill SP, Haffner SM et al. Effects of diabetes and level of glycemia on all-cause and cardiovascular mortality. The San Antonio Heart Study.. Diabetes Care 1998; 21: 1167–1172. 7. Hanefeld M, Fischer S, Julius U et al. Risk factors for myocardial infarction and death in newly detected NIDDM: the Diabetes Intervention Study, 11-year follow-up. Diabetologia 1996; 39: 1577–1583. 8. Kuusisto J, Mykkanen L, Pyorala K et al. NIDDM and its metabolic control predict coronary heart disease in elderly subjects. Diabetes 1994; 43: 960–967. 9. Andersson DK, Svardsudd K. Long-term glycemic control relates to mortality in type II diabetes. Diabetes Care 1995; 18: 1534–1543. 10. Engelgau MM, Geiss LS, Saaddine JB et al. The evolving diabetes burden in the United States. Ann Intern Med 2004; 140: 945–950.
1508
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
11.
12.
13.
14.
15.
16.
17.
18. 19.
20. 21. 22.
23. 24.
25.
26.
27. 28. 29.
30.
31. 32.
33.
34. 35. 36. 37.
38.
Selvin E, Marinopoulos S, Berkenblit G et al. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med 2004; 141: 421–431. Menon V, Greene T, Pereira AA et al. Glycosylated hemoglobin and mortality in patients with nondiabetic chronic kidney disease. J Am Soc Nephrol 2005; 16: 3411–3417. Selvin E, Coresh J, Golden SH et al. Glycemic control and coronary heart disease risk in persons with and without diabetes: the atherosclerosis risk in communities study. Arch Intern Med 2005; 165: 1910–1916. Manske CL. Hyperglycemia and intensive glycemic control in diabetic patients with chronic renal disease. Am J Kidney Dis 1998; 32: S157–S171. Tzamaloukas AH, Murata GH, Zager PG et al. The relationship between glycemic control and morbidity and mortality for diabetics on dialysis. Asaio J 1993; 39: 880–885. Malmberg K. Prospective randomised study of intensive insulin treatment on long term survival after acute myocardial infarction in patients with diabetes mellitus. DIGAMI (Diabetes Mellitus, Insulin Glucose Infusion in Acute Myocardial Infarction) Study Group. BMJ 1997; 314: 1512–1515. Malmberg K, Norhammar A, Wedel H et al. Glycometabolic state at admission: important risk marker of mortality in conventionally treated patients with diabetes mellitus and acute myocardial infarction: longterm results from the Diabetes and Insulin–Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study. Circulation 1999; 99: 2626–2632. Ceriello A, Hanefeld M, Leiter L et al. Postprandial glucose regulation and diabetic complications. Arch Intern Med 2004; 164: 2090–2095. Joy MS, Cefalu WT, Hogan SL et al. Long-term glycemic control measurements in diabetic patients receiving hemodialysis. Am J Kidney Dis 2002; 39: 297–307. American Diabetes Association. Standards of medical care in diabetes (position statement). Diabetes Care 2004; 27: S15–S35. Akmal M. Hemodialysis in diabetic patients. Am J Kidney Dis 2001; 38: S195–199. Morioka T, Emoto M, Tabata T et al. Glycemic control is a predictor of survival for diabetic patients on hemodialysis. Diabetes Care 2001; 24: 909–913. Goldstein DE, Little RR, Lorenz RA et al. Tests of glycemia in diabetes. Diabetes Care 2004; 27: 1761–1773. Larsen ML, Horder M, Mogensen EF. Effect of long-term monitoring of glycosylated hemoglobin levels in insulin-dependent diabetes mellitus. N Engl J Med 1990; 323: 1021–1025. Dunn PJ, Cole RA, Soeldner JS et al. Reproducibility of hemoglobin AIc and sensitivity to various degrees of glucose intolerance. Ann Intern Med 1979; 91: 390–396. Meigs JB, Nathan DM, Cupples LA et al. Tracking of glycated hemoglobin in the original cohort of the Framingham Heart Study. J Clin Epidemiol 1996; 49: 411–417. American Diabetes Association. Standards of medical care in diabetes (position statement). Diabetes Care 2005; 28: S4–S36. American Diabetes Association. Clinical practice recommendations. Diabetes Care 2006; 29: S3. Collins AJ, Li S, Gilbertson DT et al. Chronic kidney disease and cardiovascular disease in the Medicare population. Kidney Int Suppl 2003: S24–S31. Morgan L, Marenah CB, Jeffcoate WJ et al. Glycated proteins as indices of glycaemic control in diabetic patients with chronic renal failure. Diabet Med 1996; 13: 514–519. Stettler C, Mueller B, Diem P. What you always wanted to know about HbA1c. Schweiz Med Wochenschr 2000; 130: 993–1005. Nakao T, Matsumoto H, Okada T et al. Influence of erythropoietin treatment on hemoglobin A1c levels in patients with chronic renal failure on hemodialysis. Intern Med 1998; 37: 826–830. Williams ME LE, Lazarus JM et al. Glycohemoglobin levels and the subsequent impact on mortality risk in diabetic hemodialysis (HD) patients. J Am Soc Nephrol 2005; 16: 723A–724A. Williams ME LE, Lazarus JM et al. Differences between Type 1 and Type 2 DM patients on hemodialysis. J Am Soc Nephrol 2005; 16: 723A. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 2003; 26(Suppl 1): S5–S20. Feinglos MN, Bethel MA. Emerging care for type 2 diabetes: using insulin to reach lower glycemic goals. Cleveland Clin J Med 2005; 72: 791–799. The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial. Diabetes 1996; 45: 1289–1298. Flauto RP, Leon JB, Sehgal AR. The provision and outcomes of diabetic care of hemodialysis patients. Am J Kidney Dis 2003; 41: 125–131.
Kidney International (2006) 70, 1503–1509
original article
ME Williams et al.: Hemodialyzed type I and type II diabetic patients
39.
40. 41.
42.
43. 44. 45.
46.
47.
48. 49.
Rohlfing CL, Wiedmeyer HM, Little RR et al. Defining the relationship between plasma glucose and HbA(1c): analysis of glucose profiles and HbA(1c) in the Diabetes Control and Complications Trial. Diabetes Care 2002; 25: 275–278. Tuohy KWM. Management of the diabetic ESRD Patient: Dialysis and transplantation. Elsevier: Philadelphia, 2005. Little RR, Tennill AL, Rohlfing C et al. Can glycohemoglobin be used to assess glycemic control in patients with chronic renal failure? Clin Chem 2002; 48: 784–786. Uehlinger DE, Gotch FA, Sheiner LB. A pharmacodynamic model of erythropoietin therapy for uremic anemia. Clin Pharmacol Ther 1992; 51: 76–89. www.cmms.hhs.gov. US Department of Health and Human Services. Mak RH. Impact of end-stage renal disease and dialysis on glycemic control. Semin Dial 2000; 13: 4–8. Ifudu O, Dulin AL, Friedman EA. Interdialytic weight gain correlates with glycosylated hemoglobin in diabetic hemodialysis patients. Am J Kidney Dis 1994; 23: 686–691. Stratton IM, Adler AI, Neil HA et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405–412. Khaw KT, Wareham N, Luben R et al. Glycated haemoglobin, diabetes, and mortality in men in Norfolk cohort of european prospective investigation of cancer and nutrition (EPIC-Norfolk). BMJ 2001; 322: 15–18. Sarnak MJ, Levey AS. Cardiovascular disease and chronic renal disease: a new paradigm. Am J Kidney Dis 2000; 35: S117–131. Block GA, Klassen PS, Lazarus JM et al. Mineral metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc Nephrol 2004; 15: 2208–2218.
Kidney International (2006) 70, 1503–1509
50.
51.
52.
53.
54.
55.
56.
57.
Ichikawa H, Nagake Y, Takahashi M et al. What is the best index of glycemic control in patients with diabetes mellitus on hemodialysis? Nippon Jinzo Gakkai Shi 1996; 38: 305–308. Ansari A, Thomas S, Goldsmith D. Assessing glycemic control in patients with diabetes and end-stage renal failure. Am J Kidney Dis 2003; 41: 523–531. Lowrie EG, Lew NL. Death risk in hemodialysis patients: the predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 1990; 15: 458–482. Wanner C, Krane V, Marz W et al. Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med 2005; 353: 238–248. Pogach LM, Brietzke SA, Cowan Jr CL et al. Development of evidence-based clinical practice guidelines for diabetes: the Department of Veterans Affairs/Department of Defense guidelines initiative. Diabetes Care 2004; 27(Suppl 2): B82–B89. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998; 352: 854–865. Loipl J, Schmekal B, Biesenbach G. Long-term impact of chronic hemodialysis on glycemic control and serum lipids in insulin-treated type 2-diabetic patients. Renal Fail 2005; 27: 305–308. Glucose tolerance and mortality: comparison of WHO and American Diabetes Association diagnostic criteria. The DECODE study group. European Diabetes Epidemiology Group. Diabetes Epidemiology: Collaborative analysis Of Diagnostic criteria in Europe. Lancet 1999; 354: 617–621.
1509