CHRONIC ANEMIA IS A COMMON finding

Transcription

1 ORIGINAL INVESTIGATIONS Pathogenesis and Treatment of Kidney Disease and Hypertension The Epidemiology of Hemoglobin Levels in Patients With Type 2 Diabetes Merlin C. Thomas, MBChB, PhD, FRACP, Con Tsalamandris, MBBS, Richard J. MacIsaac, MBBS, BSc, PhD, and George Jerums, MD, FRACP Background: Anemia is a common finding in patients with diabetes, for whom it constitutes an additional burden. The aim of this study is to clarify the natural history of anemia in patients with type 2 diabetes and describe factors that predict a decrease in hemoglobin (Hb) levels. Methods: A 5-year prospective cohort study was designed as a follow-up of 503 individuals with type 2 diabetes in a single diabetes clinic. In addition to standard management, a full blood count was obtained at each routine visit. No intervention was undertaken to modify Hb levels. Results: At baseline, 12% of patients had anemia, and an additional 13% developed anemia during follow-up. Overall Hb levels decreased by g/dl/y, suggesting that anemia is the end point of a process that begins more than 10 years previously with the initiation of vascular damage. The greatest decreases in Hb levels were seen in patients with macroalbuminuria, renal impairment, or established macrovascular disease at baseline (all P < 0.01). In patients with microvascular disease, decreasing Hb levels tracked with decreasing glomerular filtration rates (GFRs). Patients with an estimated GFR greater than 90 ml/min/1.73 m 2 (>1.5 ml/s) or normoalbuminuria had stable Hb levels during the 5-year follow-up. In patients with anemia in our cohort who were managed conservatively, Hb levels decreased by g/dl/y. This decrease was associated with HbA 1c levels, but not renal function. Conclusion: This study defines the natural history of Hb levels in patients with type 2 diabetes. Early identification of anemia may be achieved by means of annual or biannual screening in high-risk groups with nephropathy, advanced age, or macrovascular disease. These data are important for developing a rational response to the prevention and management of anemia. Am J Kidney Dis 48: by the National Kidney Foundation, Inc. INDEX WORDS: Anemia; diabetes; diabetic nephropathy; hemoglobin. CHRONIC ANEMIA IS A COMMON finding in patients with diabetes, reflecting the high prevalence of kidney disease in this population. 1,2 However, patients with diabetes are more likely to have anemia than those with other types of chronic kidney disease. 3 Anemia also is found earlier in patients with diabetes than patients with chronic kidney disease from other causes. 4 Renal anemia may occur in individuals with diabetes in the absence of significant renal impairment. 5-7 Anemia has the potential to adversely affect well-being, impair the ability to work, and impact on social and sexual functioning. 8,9 In addition, a decreased hemoglobin (Hb) level identifies patients with diabetes at increased risk for adverse outcomes, including hospitalization and premature mortality, independent of the presence or severity of nephropathy. 10,11 Despite the known association between anemia and adverse outcomes in patients with diabetes, 1,10,11 the natural history of Hb levels in patients with diabetes is poorly understood. Currently, it is not known when a decrease in Hb level first starts, how long anemia takes to develop, and whether, without specific intervention, Hb levels continue to decrease after anemia is identified. Understanding this natural history is pivotal for developing a rational response to anemia prevention. From the Danielle Alberti Memorial Centre for Diabetes Complications, Baker Heart Research Institute; and Endocrine Centre and Department of Medicine, Austin Health and the University of Melbourne, Melbourne, Australia. Received March 3, 2006; accepted in revised form June 13, Originally published online as doi: /j.ajkd on August 29, Support: The costs for measurement of iron storage indices were defrayed by a grant from Jansen-Cilag (manufacturer of epoetin). The costs of erythropoietin levels were defrayed by a grant from Amgen (manufacturers of darbepoetin). Neither party had a role in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the paper for publication. M.C.T. and G.J. were both members of the Jannsen-Cilag Endocrinology advisory panel (2001 to 2002). Potential conflicts of interest: None. Address reprint requests to Merlin C. Thomas, MBChB, PhD, FRACP, Danielle Alberti Memorial Centre for Diabetic Complications, Baker Medical Research Institute, PO Box 6492, Melbourne, VIC 8008, Australia. mthomas@baker.edu.au 2006 by the National Kidney Foundation, Inc /06/ $32.00/0 doi: /j.ajkd American Journal of Kidney Diseases, Vol 48, No 4 (October), 2006: pp

2 538 Study Participants METHODS A 5-year prospective cohort study was designed as a follow-up of 622 individuals defined in a clinical audit of all ambulant patients with type 2 diabetes in a single tertiarycare specialist diabetes clinic. 4 Characteristics of these patients were described previously. 4 It should be noted that no patient had an estimated glomerular filtration rate (egfr) less than 30 ml/min/1.73 m 2 ( 0.5 ml/s) or was under the care of a nephrologist at the time of the initial survey. The majority of patients were referred to this diabetes clinic by general practitioners requiring assistance with surveillance and management of the long-term complications of diabetes. Approximately 20% of patients were referred from other sources, including specialty units within the hospital. At the time of our initial audit, correction of anemia with erythropoietin was not indicated under Australian prescribing guidelines in the absence of severe renal impairment. Moreover, the high prevalence of pure red cell aplasia in our center meant that intervention carried additional risks. 12 The potential negative effects of iron on metabolic control also were being realized. 13,14 Consequently, no specific intervention was undertaken in managing anemia or iron status, and transfusion was reserved for specific patients with moderate to severe anemia (Hb 9.0 g/dl [ 90 g/l]) and significant symptoms. However, at each routine clinical visit ( every 6 months), all patients had a full blood count added to standard blood and urine tests to closely track Hb levels. Otherwise, all patients received standard care as provided in the clinic. Data collected as part of this study were obtained with the informed consent of participants and approval of Austin Clinical Ethics Committee and met National Health and Medical Research Council ethics guidelines in accordance with the Declaration of Helsinki. Exclusion Criteria Patients with fewer than 5 full blood counts performed during 5 years and those identified at baseline as having an Hb level less than 11 g/dl ( 110 g/l) and hematinic deficiency were excluded from the study. Results obtained outside the outpatient setting (eg, patients in an emergency situation or hospitalized) also were excluded. Patients administered transfusions or drugs that reproducibly modify Hb levels (eg, iron infusion, perhexeline, immunosuppression, chemotherapy, oral anticoagulation, heparin, and lowmolecular-weight heparin) were eliminated from the analysis. No patient included in the study was administered erythropoietin. In patients who died, were lost to follow-up during the study period, or started on renal replacement therapy, the most recent (prior) routine results were included in the analysis providing that at least 5 full blood counts had been performed. THOMAS ET AL Determination of Variables At baseline, clinical data, including anthropometric measurements, age, sex, body mass index, duration of diabetes, length of follow-up, drug treatment, and presence or absence of macrovascular and microvascular complications, were obtained from patient records. 1 Standard indices were collected from baseline and subsequent routine clinical visits, including urinary albumin excretion from a 24-hour urine collection, serum urea and creatinine, C-reactive protein, fasting lipid profile, fasting glucose, HbA 1c, full blood count, serum iron availability (transferrin saturation), iron stores (ferritin), and erythropoietin concentrations. Patients were surveyed on all aspects of cardiac risk, including their previous cardiac history (myocardial infarction, angina, heart failure, coronary angiography, previous angioplasty, and/or bypass surgery) and other macrovascular complications (stroke, transient ischemic attack, and peripheral vascular disease). Outcome Measures Anemia is defined according to European Best Practice Guidelines as an Hb level less than 13.5 g/dl ( 135 g/l) in men younger than 70 years, less than 12.0 g/dl ( 120 g/l) in adult male patients older than 70 years, and less than 11.5 g/dl ( 115 g/l) in women of any age. 15 Mean change in Hb levels during the 5-year study (expressed as grams per deciliter per year) was calculated by using the linear regression of all Hb levels obtained in each individual patient plotted against time. Mean change in Hb level was handled both as a continuous variable and recoded as a binary outcome for categorical analysis. Individuals experiencing a decrease in Hb levels greater than 0.2 g/dl/y ( 1 g/dl/5 y) were defined arbitrarily as having a significant decrease in Hb level and compared with individuals who experienced either lesser change, mean increase in Hb levels, or no change. A serum ferritin level less than 100 ng/ml ( 225 pmol/l) and transferrin saturation less than 20% were used to denote inadequate iron stores and reduced iron availability, respectively. egfr was estimated by using the 6-variable equation proposed by the Modification of Diet in Renal Disease Study Group 16 and simultaneously measured by using isotopic methods in a subgroup of patients, as previously documented. 4 Baseline albumin excretion rate (AER) was derived from the 3 most recent 24-hour urinary albumin measurements and categorized according to International Diabetes Federation guidelines. Macroalbuminuria is defined as 2 of 3 AER measurements greater than 200 g/min. Microalbuminuria is defined as 2 of 3 AER measurements greater than 20 g/min. Normoalbuminuria is defined as 2 of 3 AER measurements less than 20 g/min. Statistical Methods Continuous data are expressed as mean SEM, except for nonparametric parameters, which are expressed as their geometric mean. Differences in continuous variables were compared by using Student t-test (2 groups), and differences in categorical variables were compared by using chi-square analysis. Pearson correlation was used to analyze univariate associations between continuous variables. Multivariate analysis used reverse stepwise regression to model independent predictors of a significant decrease in Hb level.

3 HEMOGLOBIN AND DIABETES 539 RESULTS Baseline Patient Characteristics At baseline, 622 patients with type 2 diabetes were in current long-term follow-up in our center. During the follow-up period, 71 patients died (or were lost to follow-up) before 5 routine clinic visits had been made. Fourteen patients with iron deficiency and an Hb level less than 11 g/l ( 110 g/l) were excluded from analysis and treated with iron. In addition, an additional 34 patients were excluded because of secondary influences on Hb levels, including 8 patients with known malignancy, 6 patients receiving immunosuppressive therapy, 10 patients with documented gastrointestinal bleeding, and 10 patients on long-term anticoagulation therapy. Nineteen patients required symptomatic transfusion, all of whom met exclusion criteria. This left 503 patients with type 2 diabetes in whom a change in Hb levels during the 5-year follow-up was analyzed. Baseline characteristics of this cohort were reported previously. 1 Briefly, the study included 283 men and 220 women with a mean age of 65 1 years (range, 29 to 86 years). Less than 16% of women were younger than 55 years. More than 95% of patients were of Caucasian descent, and the study included no African- American patient. Median duration of diabetes was 11 years, with a median of 6 years of follow-up previously spent in the diabetes clinic at our center, attending an average of every 6 months. More than half the patients were obese (body mass index 30 kg/m 2 ). Mean HbA 1c level was 8.0% 0.1% ( ). In this cohort, 40% of patients with type 2 diabetes also were administered insulin. Mean Hb levels at baseline were g/dl (140 1 g/l) in men and g/dl (132 1 g/l) in women. At baseline, 64 patients (12.7%) had anemia defined according to the Best Practice Guidelines. 15 As previously published, 1,4 independent predictors of anemia at baseline in this cohort were moderate to severe renal impairment, elevated urinary AER, and decreased iron availability (multivariate P 0.01), with these 3 parameters explaining more than 40% of the variability in prevalence of anemia. 4 Although sex was an important determinant of raw Hb levels, it did not predict anemia when using a sex-specific definition. Development of Anemia During Follow-Up During the 5-year follow-up, an additional 63 individuals (12.5%) experienced a decrease in Hb levels into the range denoting the presence of anemia. Most patients could have been predicted by a low-normal baseline Hb level, reflecting the lead time required for the development of anemia. Over and above this bias, baseline renal function, AER, and macrovascular disease at baseline were all independently associated with the subsequent development of anemia (multivariate P 0.01). Change in Hb Levels During 5 Years of Follow-Up For the 503 patients in our cohort, Hb levels remained modestly stable without intervention A Incidence (%) B Incidence (%) 35% 30% 25% 20% 15% 10% 5% 0% 40% 35% 30% 25% 20% 15% 10% 5% 0% 11.6 g/l < Change in Hb from baseline (g/dl/year) 14.0g/dL Change in Hb from baseline (g/dl/year) Fig 1. Change in Hb levels during 5 years of follow-up in ambulatory individuals with type 2 diabetes (A) with and (B) without anemia at baseline. Grey bars, individuals with a decrease in Hb levels of more than 0.2 g/dl/y (>1 g/dl/5 y); dotted line, mean baseline Hb level. To convert Hb in g/dl to g/l, multiply by 10.

4 540 THOMAS ET AL Table 1. Baseline Characteristics of Patients With Type 2 Diabetes Without Anemia at Baseline Stratified According to Subsequent Decrease in Hb Levels Greater Than 0.2 g/dl/y During 5 Years of Follow-Up No Decrease (n 334) Decrease (n 105) Mean change in Hb (g/dl/y) Baseline Hb (g/dl) Men (%) Duration of diabetes (y) Age (y) Obesity (body mass index 30 kg/m 2 ) (%) Blood pressure (mm Hg) 143/80 144/80 Antihypertensive therapy (%) Blockade of the renin-angiotensin system (%) Lipid-lowering therapy (%) Insulin (%) egfr (ml/min/1.73 m 2 ) AER ( g/min) Retinopathy (%) Cardiovascular disease (%) Cerebrovascular disease (%) Peripheral vascular disease (%) 9 23 Any macrovascular disease (%) HbA 1c (%) C-Reactive protein (IU/mL) Ferritin (ng/ml) Transferrin saturation (%) Erythropoietin (IU/mL) NOTE. To convert Hb in g/dl to g/l, multiply by 10; GFR in ml/min to ml/s, multiply by ; ferritin in ng/ml to pmol/l, multiply by Univariate P Multivariate P Nonlinear data expressed as its geometric mean. during 5 years of follow-up, with a small, but significant, average decrease of g/dl/y (median, 0.06 g/dl/y; skewness, 0.39; kurtosis, 0.6). A histogram of the distribution of Hb levels changes during the 5-year follow-up period in patients with and without anemia is shown in Fig 1. Mean change in Hb levels was similar regardless of baseline Hb level or sex. Overall, independent predictors of a decrease in Hb levels during 5 years of follow-up were baseline AER and egfr and the presence of macrovascular disease at baseline (Table 1). The rate of decrease in Hb levels in patients developing anemia ( g/dl/y) was not different from that observed in patients who did not develop anemia after adjusting for differences in prevalence of diabetic complications. One hundred twenty-four individuals (24%) had a mean decrease in Hb levels of more than 0.2 g/dl/y (1.0 g/dl/5 y), arbitrarily defined as a significant decrease in Hb level. These patients were more likely to be older than 65 years with decreased egfr and increased AER and have macrovascular disease at baseline than patients experiencing lesser or no change in Hb levels (Table 1). After adjusting for these variables, a significant decrease in Hb level was not associated with sex, duration of diabetes, baseline Hb level, C-reactive protein level, lipid level, blood pressure control, serum erythropoietin level, iron availability or stores, use of agents that block the renin-angiotensin system, or thiazolidinediones (Table 1). Macrovascular Disease and Hb Levels Mean change in Hb levels in patients with overt macrovascular disease at baseline was more than twice that observed in patients without established cardiovascular complications ( 0.11 versus 0.03 g/dl/y), with nearly twice as many individuals experiencing a decrease in Hb levels greater than 0.2 g/dl/y than those free of symp-

5 HEMOGLOBIN AND DIABETES 541 tomatic macrovascular disease at baseline (33% versus 18%; P 0.01). This was not explained by increased levels of systemic inflammation in individuals with macrovascular disease. In addition, this association was not modified by drug treatments, including use of aspirin, diuretics, or angiotensin-converting enzyme inhibitors. However, the increased prevalence of chronic kidney disease at baseline in patients known to have macrovascular disease partly explains this risk. Nonetheless, even after adjusting for egfr or AER at baseline, patients with established macro- A B Mean change in Hb (g/l/year) Mean change in Hb (g/l/year) >200 > <60 # # # Fig 3. Mean change in Hb levels during the 5-year follow-up stratified according to (A) AER and (B) baseline GFR. P < 0.05 versus AER less than 10 g/min. #P < 0.05 versus GFR greater than 120 ml/min/1.73 m 2. To convert Hb in g/dl to g/l, multiply by 10; GFR in ml/min to ml/s, multiply by vascular disease were more likely to have a decrease in Hb levels during 5 years of followup, and the size of this decrease was larger on average than in patients free of macrovascular complications (Fig 2A and B). Fig 2. A significant decrease in Hb levels is seen more commonly in patients with macrovascular disease (grey bars) than individuals without overt macrovascular disease (white) after adjusting for (A) egfr and (B) urinary AER. (C) GFR and urinary AER have additive effects on Hb levels. To convert Hb in g/dl to g/l, multiply by 10; GFR in ml/min to ml/s, multiply by Albuminuria, Renal Function, and Hb Levels Markers of renal injury were associated independently with a decrease in Hb levels. Individuals with macroalbuminuria were nearly 3 times more likely to have a decrease in Hb levels of more than 0.2 g/dl/y compared with those with an AER less than 10 g/min, and patients with microalbuminuria had an intermediate incidence (Fig 3A). This effect was independent of baseline renal function (Fig 2C). Baseline egfr was associated independently with decreased Hb level and increased risk for Hb level decrease. However, mean rate of decrease in Hb levels was approximately similar (Fig 3B) after egfr decreased to less than 90 ml/min/1.73 m 2 ( 1.5 ml/s) in patients with type 2 diabetes, consistent with the steady decrease in mean Hb levels observed in crosssectional studies. 17 Moreover, Hb levels also tracked with changes in egfr, such that the risk for a decrease in Hb levels associated with an

6 542 Mean change in Hb (g/dl/year) Age at baseline (years) < >70 Fig 4. Change in Hb levels during the study period stratified according to age and sex (male, grey; female, white). Sex difference, P < To convert Hb in g/dl to g/l, multiply by 10. egfr less than 90 ml/min/1.73 m 2 ( 1.5 ml/s) was largely attributable to patients whose GFR decreased during the study period. Overall, one third of patients whose GFR decreased by more than 3 ml/min/y during the study period also experienced a significant decrease in Hb levels. Conversely, individuals not experiencing a decrease in egfr had no net decrease in Hb levels regardless of baseline egfr. Similarly, patients with an egfr greater than 120 ml/ min/1.73 m 2 ( 2 ml/s) or AER less than 10 g/min had stable Hb levels during the 5-year study follow-up. Considering that most patients with diabetes and anemia have chronic kidney disease 1 and the rate of Hb level decrease in those with established kidney disease at baseline was approximately 0.1 g/dl/y, it is estimated that it would take more than 10 years for most patients to experience a decrease in Hb levels from levels seen in patients without chronic kidney disease into the anemic range. THOMAS ET AL Other Predictors of Hb Level Decrease Decrease in Hb levels also correlated with age at baseline and age at onset of diabetes on univariate analysis (Table 1). Men and women older than 60 years were twice as likely as those younger than 60 years to experience a significant decrease in Hb levels (P 0.01). This was offset in part by a postmenopausal increase in Hb level observed in women in our study (Fig 4). Even when men were analyzed separately, a decrease in Hb levels correlated with chronological ageing (P 0.02). However, this association was eliminated after adjusting for macrovascular disease and renal impairment, which were more common in elderly patients. Although systemic inflammation may be associated with anemia in some patients, C-reactive protein levels were not associated independently with rate of decrease in Hb levels in this population. Change in Hb Levels During 5 Years of Follow-Up in Patients With Anemia at Baseline For the 64 patients with baseline anemia in our cohort who were managed conservatively, Hb levels decreased only slightly during 5 years of follow-up, with a mean decrease of g/dl/y (median, 0.05 g/dl/y; skewness, 0.39; kurtosis, 0.6). A histogram of the distribution of Hb level changes in anemic patients during the 5-year follow-up is shown in Fig 1A. Twenty-two individuals (34%) had a decrease in Hb levels greater than 0.2 g/dl/y (1.0 g/dl/5 y). Characteristics of these individuals are listed in Table 2. Unlike patients without anemia at baseline, AER, egfr, and the presence of macrovascular complications were not associated with an additional decrease in Hb levels. However, patients with an additional decrease in Hb levels tended to have a lower HbA 1c level at baseline and were less likely to be obese (both P 0.01). Neither insulin use nor modality of oral hypoglycemic therapy explained this association. The effect of obesity in these patients was independent of serum albumin level, transferrin saturation, or other nutrition indices. Furthermore, iron availability or stores at baseline were not associated with change in Hb levels in patients with anemia (Table 2). During the 5-year follow-up without intervention, only 5 patients with anemia experienced a decrease in Hb levels to less than 10 g/dl (6%). These individuals were mostly women (4 women, 1 man) with a lower baseline Hb level (11.0 g/dl [110 g/l] versus 11.6 g/dl [116 g/l]). In no patient did Hb levels decrease to less than 9.0 g/dl ( 90 g/l) in the absence of identifiable secondary causes (eg, blood loss, malignancy, and sepsis). Nine patients with anemia (14%) spontaneously experienced an increase in Hb levels by more than 1 g/dl ( 10 g/l) during the 5-year follow-up. Compared with individuals experiencing no change in Hb levels or a de-

7 HEMOGLOBIN AND DIABETES 543 crease, these patients tended to have milder anemia at baseline (11.4 g/dl [114 g/l] versus 12.0 g/dl [120 g/l]; P 0.02) and were less likely to have macrovascular disease (33% versus 63%; P 0.05). DISCUSSION Chronic anemia is common in patients with diabetes. By the end of the study period, 1 of 4 individuals with type 2 diabetes in our cohort (13% at the beginning and 12% during follow-up) experienced anemia, although severe renal impairment was absent at baseline. An insidious decrease in Hb levels was observed early in the clinical course of chronic kidney disease in patients with type 2 diabetes, suggesting that anemia is the end point of a process that begins with the initiation of microvascular damage. The rate of decrease was fastest in individuals with established and progressive renal injury and macrovascular disease at baseline. These are the same risk factors previously associated with the presence of anemia in cross-sectional surveys, 1,4 but this is the first description of Hb level change and its modifiers in a prospective setting. Evidence of renal injury at baseline was the strongest risk factor for a decreasing Hb level during 5 years of follow-up in patients with type 2 diabetes. Half the patients with macroalbuminuria experienced a significant decrease in Hb levels compared with less than 10% of individuals with an AER less than 10 g/min or those with normal renal function. In addition, individuals experiencing an additional decrease in renal function during the study were more likely to have a decrease in Hb level regardless of baseline GFR. These data again point to the key role of the renal microvasculature in the development and progression of anemia in patients with diabetes. Anemia is associated closely with both renal and cardiovascular disease as part of the cardio- Table 2. Baseline Characteristics of Patients With Type 2 Diabetes and Anemia at Baseline Stratified According to Subsequent Decrease in Hb Levels Greater Than 2 g/l/y During 5 Years of Follow-Up No Decrease (n 42) Decrease (n 22) Mean change in Hb (g/dl/y) Baseline Hb (g/dl) Men (%) Duration of diabetes (y) Age at diagnosis (y) Obesity (body mass index 30 kg/m 2 ) (%) Serum albumin (g/dl) Blood pressure (mm Hg) 148/78 152/80 Antihypertensive therapy (%) Blockade of the renin-angiotensin system (%) Lipid-lowering therapy (%) Insulin (%) Thiazolidinedione therapy (%) egfr (ml/min/1.73 m 2 ) AER ( g/min) Cardiovascular disease (%) Cerebrovascular disease (%) Peripheral vascular disease (%) Any macrovascular disease (%) HbA 1c (%) C-Reactive protein (IU/mL) Ferritin (ng/ml) a Transferrin saturation (%) Erythropoietin (IU/mL) NOTE. To convert Hb and albumin in g/dl to g/l, multiply by 10; GFR in ml/min to ml/s, multiply by ; ferritin in ng/ml to pmol/l, multiply by Univariate P Nonlinear data expressed as its geometric mean.

8 544 renal-anemia syndrome. 18 In our study, patients with established macrovascular disease at baseline were more likely to have anemia at baseline and more likely to develop anemia during 5 years of follow-up. This association appeared to be synergistic with renal damage in contributing to decreasing Hb levels (Fig 2). This finding is similar to observations in patients with diabetes with chronic kidney disease, for whom anemia appears to act as a mortality multiplier. 19 Interactions within cardiorenal-anemia syndrome are clearly complex because anemia may be considered both the cause and effect of cardiac and renal disease. Cardiac disease may contribute to a decreasing Hb level through hemodilution and the release of antierythropoietic cytokines and inflammatory mediators associated with macrovascular disease, 18 although C-reactive protein level did not correlate with Hb level. Drug therapy also may confound the interaction, although Hb level change did not correlate with aspirin use in our study. Furthermore, patients administered oral anticoagulants were specifically excluded from this study. Anemia also may contribute to cardiac dysfunction in patients with diabetes 20 and cardiovascular disease. 21 The association between macrovascular disease and Hb level in our study also may reflect the role of anemia in exacerbating ischemic symptoms. No specific intervention was undertaken for correction of anemia in our patients. Individuals with identified causes of anemia were excluded from the study. Even so, this left a large group of patients (13% of our cohort) with Hb levels less than the normal range for their age and sex. Their anemia was managed conservatively, consistent with practice guidelines. Moreover, at the time of initiation of this study, significant risks were associated with erythropoietin-based therapies in patients with chronic kidney disease. 12 Consequently, we undertook to monitor Hb levels at every clinical visit, but not correct anemia with erythropoietin or related analogues, although most patients with anemia had functional erythropoietin deficiency. 6 Although there was a small, but significant, decrease in Hb levels of 0.09 g/dl/y (0.4 g/dl/5 y), most patients with anemia (91%) did not experience a decrease in Hb levels to less than 10 g/dl ( 100 g/l) with standard supportive therapy. Although renal impairment, increased albuminuria, and macrovascular disease THOMAS ET AL were strong predictors of a decreasing Hb level in individuals without anemia, these factors were already highly prevalent in individuals with anemia at baseline and not predictive of a subsequent Hb level decrease in patients with anemia. However, HbA 1c level was lower in individuals experiencing a decrease in Hb levels despite similar fasting blood glucose levels. It is possible that this reflects intrinsic differences in red blood cell turnover 22 that contribute to both a decreased HbA 1c level and decreasing Hb level in the presence of impaired renal compensation. 6 On the basis of this study, it is possible to suggest that individuals with an Hb level less than 11 g/dl ( 110 g/l) probably had Hb levels significantly less than the normal range for an average of 5 years or longer. This study shows that regular monitoring is able to identify individuals with decreasing Hb levels before the onset of anemia. In particular, in individuals with elevated albuminuria, established renal impairment, macrovascular disease, or advanced age, for whom a decrease in Hb levels occurred in at least 1 in 3 individuals, regular screening of Hb levels on an annual or biannual basis would facilitate the early detection of individuals with anemia. There seems to be no additional utility in performing a routine full blood count more often (or on every visit, as performed in this study) unless specifically indicated because mean rate of change in Hb levels is on average slow. Moreover, a decrease in Hb levels of more than 1 g/dl/y was not observed in this study in the absence of a secondary cause, such as blood loss or major illness. In individuals with normal renal function or normoalbuminuria, the development of anemia was uncommon, meaning that screening these patients would have limited utility. The appropriate clinical response to anemia in patients with diabetes remains controversial. Correction of anemia improves quality of life, neurological function, and exercise tolerance. 23 Treatment with erythropoietin in patients with diabetes with heart failure also may decrease the length of hospitalization and improve functional class. 24 However, there also is a potential for deleterious effects from erythropoietin and related analogues in patients with diabetes, such as hypertension and increased peripheral vascular resistance. The benefits arising from correction of anemia in

9 HEMOGLOBIN AND DIABETES 545 recent large clinical studies have been equivocal (eg, Cardiovascular Risk Reduction by Early Anemia Treatment with Epoetin Beta) or possibly associated with increased risk for cardiovascular morbidity and mortality (eg, the Correction of Hemoglobin and Outcomes in Renal Insufficiency Study). The true balance of risks and benefits in patients with diabetes hopefully will be answered by ongoing large randomized controlled studies. 25 In summary, anemia is common in patients with type 2 diabetes, in whom it is associated with adverse outcomes. This study defines the natural history of Hb levels in patients with type 2 diabetes, showing that a decrease in Hb levels is insidious and occurs predominantly in individuals with chronic kidney disease, advanced age, or macrovascular disease. Consequently, early identification of anemia may be achieved by annual or biannual screening in these high-risk groups. These data are important for developing a rational response to the prevention and early management of anemia in individuals with diabetes. REFERENCES 1. Thomas MC, MacIsaac RJ, Tsalamandris C, et al: The burden of anaemia in type 2 diabetes and the role of nephropathy: A cross-sectional audit. Nephrol Dial Transplant 19: , Thomas MC, MacIsaac RJ, Tsalamandris C, et al: Anemia in patients with type 1 diabetes. J Clin Endocrinol Metab 89: , Astor BC, Muntner P, Levin A, Eustace JA, Coresh J: Association of kidney function with anemia: The Third National Health and Nutrition Examination Survey ( ). Arch Intern Med 162: , Thomas MC, MacIsaac RJ, Tsalamandris C, Power D, Jerums G: Unrecognized anemia in patients with diabetes: A cross-sectional survey. Diabetes Care 26: , Craig KJ, Williams JD, Riley SG, et al: Anemia and diabetes in the absence of nephropathy. Diabetes Care 28: , Thomas MC, Cooper ME, Tsalamandris C, MacIsaac R, Jerums G: Anemia with impaired erythropoietin response in diabetic patients. Arch Intern Med 165: , Al-Khoury S, Afzali B, Shah N, Covic A, Thomas S, Goldsmith DJ: Anaemia in diabetic patients with chronic kidney disease Prevalence and predictors. Diabetologia 49: , Lundin AP: Quality of life: Subjective and objective improvements with recombinant human erythropoietin therapy. Semin Nephrol 9:S22-S29, 1989 (suppl 1) 9. Canadian Erythropoietin Study Group: Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving haemodialysis. BMJ 300: , Keane WF, Lyle PA: Recent advances in management of type 2 diabetes and nephropathy: Lessons from the RENAAL Study. Am J Kidney Dis 41:S22-S25, 2003 (suppl 1) 11. Rossing K, Christensen PK, Hovind P, Tarnow L, Rossing P, Parving HH: Progression of nephropathy in type 2 diabetic patients. Kidney Int 66: , Evens AM, Bennett CL, Luminari S: Epoetin-induced pure red-cell aplasia (PRCA): Preliminary results from the research on adverse drug events and reports (RADAR) group. Best Pract Res Clin Haematol 18: , Thomas MC, MacIsaac RJ, Tsalamandris C, Jerums G: Elevated iron indices in patients with diabetes. Diabet Med 21: , Fernandez-Real JM, Lopez-Bermejo A, Ricart W: Cross-talk between iron metabolism and diabetes. Diabetes 51: , Locatelli F, Aljama P, Barany P, et al: Revised European Best Practice Guidelines for the management of anaemia in patients with chronic renal failure. Nephrol Dial Transplant 19:S1-S47, 2004 (suppl 2) 16. Levey AS, Coresh J, Balk E, et al: National Kidney Foundation practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Ann Intern Med 139: , Thomas MCTC, MacIssaac R, Jerums G: Anemia in diabetes; An emerging complication of microvascular disease. Curr Diabetes Rev 1: , Silverberg D, Wexler D, Blum M, Wollman Y, Iaina A: The cardio-renal anaemia syndrome: Does it exist? Nephrol Dial Transplant 18:S7-S12, 2003 (suppl 8) 19. Collins AJ, Ma JZ, Ebben J: Impact of hematocrit on morbidity and mortality. Semin Nephrol 20: , Srivastava PM, Thomas MC, Calafiore P, Macisaac RJ, Jerums G, Burrell LM: Diastolic dysfunction is associated with anaemia in patients with type II diabetes. Clin Sci (Lond) 110: , Astor BC, Coresh J, Heiss G, Pettitt D, Sarnak MJ: Kidney function and anemia as risk factors for coronary heart disease and mortality: The Atherosclerosis Risk in Communities (ARIC) Study. Am Heart J 151: , Jones RL, Peterson CM: Hematologic alterations in diabetes mellitus. Am J Med 70: , Cody J, Daly C, Campbell M, et al: Recombinant human erythropoietin for chronic renal failure anaemia in pre-dialysis patients. Cochrane Database Syst Rev 4:CD003266, Silverberg DS, Wexler D, Blum M, et al: The effect of correction of anaemia in diabetics and non-diabetics with severe resistant congestive heart failure and chronic renal failure by subcutaneous erythropoietin and intravenous iron. Nephrol Dial Transplant 18: , Thomas MC, Cooper ME, Rossing K, Parving HH: Anaemia in diabetes: Is there a rationale to TREAT? Diabetologia 49: , 2006