 |
|
|
The Agency for Healthcare Research and Quality (AHRQ), through its Evidence-Based Practice Centers (EPCs), sponsors the development of evidence reports and technology assessments to assist public- and private-sector organizations in their efforts to improve the quality of health care in the United States. The reports and assessments provide organizations with comprehensive, science-based information on common, costly medical conditions and new health care technologies. The EPCs systematically review the relevant scientific literature on topics assigned to them by AHRQ and conduct additional analyses when appropriate prior to developing their reports and assessments.
To bring the broadest range of experts into the development of evidence reports and health technology assessments, AHRQ encourages the EPCs to form partnerships and enter into collaborations with other medical and research organizations. The EPCs work with these partner organizations to ensure that the evidence reports and technology assessments they produce will become building blocks for health care quality improvement projects throughout the Nation. The reports undergo peer review prior to their release.
AHRQ expects that the EPC evidence reports and technology assessments will inform individual health plans, providers, and purchasers as well as the health care system as a whole by providing important information to help improve health care quality.
We welcome written comments on this evidence report. They may be sent to: Acting Director, Center for Outcomes and Evidence, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850.
Carolyn M. Clancy, M.D.
Director
Agency for Healthcare Research and Quality
Jean Slutsky, P.A., M.S.P.H., Acting Director
Center for Outcomes and Evidnece
Agency for Healthcare Research and Quality
The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services of a particular drug, device, test, treatment, or other clinical service.
Background: While testing for glycemic control and urine albumin are widely recommended for persons with Types 1 and 2 diabetes mellitus, there has not been a systematic assembly of the literature to assess the risk relation between tests assessing long-term glycemic control or tests assessing the presence of microalbuminuria to future cardiovascular, peripheral vascular, renal, and neurological outcomes (all of which represent end-organ effects of long-term diabetes). This report, commissioned by the American Association of Clinical Chemistry, systematically reviews the literature identifying the risk relation between testing for glycemic control and urine albumin with these important clinical outcomes.
Search Strategy: Electronic searches were completed of PubMED, and hand searching of key journals, conference proceedings, and references lists.
Selection Criteria: Articles were included in this evidence report if they were English language reports of prospective studies in which original data was reported. It was required that one of the specific research questions be addressed in non-pregnant adults.
Data Collection and Analysis: Quality analysis and data abstraction were performed by pairs of reviewers. For each study question, evidence tables were developed and qualitative synthesis was completed.
Main Results: The evidence supports a strong, graded relation between the level of glycated hemoglobin and the risk of two major microvascular complications of type 1 and type 2 diabetes, retinopathy and nephropathy. These patterns are observed for various measures of glycated hemoglobin (i.e., HbA1c, HbA1, and total GHb). Cohort studies evaluating coronary artery disease and peripheral arterial disease demonstrated a positive association with glycated hemoglobin exposure in persons with types 1 and 2 diabetes (although risk estimates were much smaller compared to the risk estimates for the microvascular complications). The risk relationship between cerebrovascular disease and glycated hemoglobin is less clear. For studies on urine albumin, reporting of methods for measurement of urine albumin and reporting of cutoffs for microalbuminuria were varied. The evidence supports the association of microalbuminuria at baseline with progression of kidney disease, all-cause death, and cardiovascular morbidity and mortality. These relations appear to be graded with greater levels of urine albumin excretion at baseline predicting greater risk of these outcomes at follow up.
Conclusions: Glycated hemoglobin was strongly associated with an increased risk of microvascular complications as well as macrovascular complications, although the association was weaker for macrovascular disease. Microalbuminuria was associated with progression of chronic kidney disease and the development of cardiovascular morbidity, cardiovascular mortality, and all-cause death. This synthesis is limited by the significant heterogeneity in the measurements of glycated hemoglobin and microalbuminuria as well as heterogeneous measurements for clinical outcomes making comparisons across studies difficult. For research on glycemic control, future work should focus on studying the relation between glycated hemoglobin exposure and the risk of neuropathy and macrovascular complications and determining whether there is a threshold for the effect of glycated hemoglobin on these outcomes. Because of the heterogeneity in the literature on reporting of glycated hemoglobin, future cohort studies and clinical trials should aim to use standardized methods to allow risk comparison across studies. For research on urine albumin, future work should seek to define the optimal and most feasible measures of microalbuminuria and to standardize measurement of microalbuminuria in persons with diabetes. Future research should also characterize the nature of the relation between microalbuminuria and outcomes.
Clinical testing to assess levels of disease control and progression among persons with types 1and 2 diabetes mellitus is widely recommended to clinicians to improve patients' clinical outcomes. Two important foci of recommendations for the followup care of individuals with diabetes include monitoring of glycemic status by measurement of glycated hemoglobin (GHb) and screening for kidney disease with urine albumin to assess overall disease progression and to detect potential progression toward end-organ damage. According to the American Diabetes Association Clinical Practice Recommendations, monitoring of glycemic status is considered a cornerstone of diabetes care and affects how physicians and patients adjust medical therapy as well as behavioral therapy, such as diet and exercise. Screening for urine albumin among persons with diabetes is also widely recommended for the detection and treatment of incipient diabetic nephropathy, and affects the physician's implementation of therapy to slow progression of kidney disease.
Despite widespread recommendations for screening of persons with diabetes for both glycemic control and urine albumin, there has not been a systematic assembly of the literature to assess the risk relation between tests assessing long-term glycemic control or tests assessing the presence of microalbuminuria with cardiovascular, peripheral vascular, renal, and neurological outcomes (all of which represent end-organ effects of long-term diabetes). This report, commissioned by the American Association of Clinical Chemistry, systematically reviews the literature identifying the risk relation between testing for glycemic control or urine albumin and these important clinical outcomes.
The advent of self-monitoring of blood glucose (SMBG) has allowed patients to attain glycemic goals more quickly and has revolutionized the care of individuals with diabetes mellitus; however, it does not provide information regarding glycemic control over an extended period of time. Measurement of glycated hemoglobin, which first began in the 1970's, has become the preferred method of assessing long-term glycemic control. Of the various glycated hemoglobin fractions, HbA1c is the preferred standard for measuring glycemic control over the previous 2–3 months. The American Diabetes Association began to make treatment recommendations based on HbA1c following publication of the results of the Diabetes Control and Complications Trial (DCCT) in the 1990's. The HbA1c has become the gold standard for the therapeutic management of diabetes mellitus in research and in the clinical setting.
While glycated hemoglobin testing gives an assessment of long-term glycemic control, what other prognostic information does it provide in the management of individuals with diabetes mellitus? Several large, randomized clinical trials have demonstrated that intensive glycemic control prevents the development and progression of long-term diabetic microvascular complications. In these studies, glycemic goals were assessed using glycated hemoglobin as the measure of long-term control. Long-term hyperglycemia, as measured by glycated hemoglobin, is clearly related to the development of diabetic microvascular complications; however, its relation to the development of macrovascular complications is less clear. The relation of glycemic control and cardiovascular disease in individuals with diabetes remains controversial, with some studies demonstrating a positive association and others showing no association. Another important issue that is still an area of active investigation in the management of diabetes relates to whether there is a threshold effect of glycated hemoglobin for microvascular and/or macrovascular complications. The issue of a threshold effect of glycated hemoglobin has important implications for where the HbA1c treatment targets are set to prevent diabetic complications.
Screening tests for microalbuminuria are recommended annually for patients with type 1 diabetes of greater than 5 years duration and for all patients with type 2 diabetes from the time of diagnosis. Twenty-four hour collection of urine for quantitative assessment of urinary albumin excretion rate is currently considered the gold standard measurement of microalbuminuria. However, this method is frequently considered to be cumbersome and difficult to carry out in the outpatient clinical setting, and it is subject to timing and collection inaccuracies. Several other methods of testing, which are considered less difficult to perform in outpatient settings, have been studied and have been demonstrated to have varying degrees of correlation with 24 hour urine collection for the detection of urinary albumin. These include random or ‘spot’ testing of a morning urine specimen for urine albumin concentration or albumin to creatinine ratio, overnight or ‘timed’ urine collections for estimation of albumin excretion rates, and dipstick testing. While many experts now support the use of random or first morning tests for urinary albumin to creatinine ratio as a convenient and accurate approach to screening patients, there is currently no clear consensus on standardized testing methods.
In a variety of prospective studies, elevated urinary albumin excretion has been shown to be associated with increased risk of progression of kidney disease toward ESRD as well as increased cardiovascular morbidity, cardiovascular mortality, and total mortality. However, few studies have systematically ascertained the magnitude of increase in both renal and cardiovascular risk associated with microalbuminuria among persons with type 1 and persons with type 2 diabetes. Moreover, although observations of renal and cardiovascular outcomes among persons classified as having microalbuminuria by currently accepted standards have been reported, it remains unclear whether current definitions of microalbuminuria are optimal in terms of predicting renal and cardiovascular outcomes, and it is unclear whether there is a dose-relationship or threshold effect in prediction of outcomes associated with urinary albumin excretion. Knowledge of the pooled magnitude of risk associated with current definitions of microalbuminuria in addition to an improved understanding of the risk relationship of varying levels of baseline albuminuria with cardiovascular and renal outcomes could have important implications in screening and treatment recommendations for persons with diabetes.
This report addresses the following key questions in persons with types 1 and 2 diabetes mellitus:
What is the risk relationship between glycated hemoglobin and the subsequent risk of microvascular diabetic complications (retinopathy, nephropathy, neuropathy)?
What is the risk relationship between glycated hemoglobin and the subsequent risk of macrovascular diabetic complications (coronary heart disease, cerebrovascular disease, peripheral vascular disease)?
What is the risk relationship between microalbuminuria and renal function?
What is the risk relationship between microalbuminuria and cardiovascular disease and death?
The Evidence-based Practice Center (EPC) recruited six technical and community experts to provide input into the definition of the key questions and to review a draft of the report. The EPC also recruited representatives from a range of other stakeholder organizations to serve as peer reviewers of the draft Evidence Report. These stakeholder organizations included organizations of physicians, allied health professionals, and third party health care payers in addition to consumer organizations.
Because of the divergent content of questions related to glycemic control and urine albumin, two separate teams of investigators systematically reviewed literature in these two areas. Investigators from each team executed systematic search strategies pertinent to each set of questions. Thus, search strategies and studies identified were different for questions on glycemic control and urine albumin.
Articles published in English language from 1966, when MEDLINE® began indexing, to April 2002 were accessed through PubMed® using MESH and text words for glycated hemoglobin, diabetes, and individuals diabetic complications, including retinopathy, nephropathy, neuropathy, and cardiovascular disease.
Reviewed studies were restricted to prospective longitudinal cohort studies, non-concurrent prospective cohort studies, and clinical trials that had data on glycated hemoglobin exposure and outcome data on individual microvascular and macrovascular complications during at least 1 year of followup in at least 50 participants with type 1 and type 2 diabetes. Because we were interested in determining the risk relationship between glycated hemoglobin exposure and microvascular and macrovascular complications and in re-examining whether a threshold exists for this relationship, we only included studies that reported prospective, quantitative risk data (i.e. incidence rates, regression coefficients with standard errors reported separately, relative hazards, relative odds, relative risk). Retrospective case-control studies that reported and compared previous glycated hemoglobin values in individuals with a given outcome versus those without a given outcome were excluded. Articles that reported data in graphical form in which specific quantitative values could not be determined were excluded.
Other exclusion criteria included review articles, animal or in vitro studies, non-English language, and studies in which the design was unclear.
Glycated hemoglobin: Data were abstracted on the biochemical method of measurement, whether the method was traceable to the DCCT standard and/or whether the lab was certified by the National Glycohemoglobin Standardization Program (NGSP), and how glycated hemoglobin was reported (i.e. HbA1c, HbA1, total GHb).
Diabetes mellitus: Data were extracted on the type of diabetes that study participants had and the method of diagnosis/confirmation.
Retinopathy: Incident retinopathy was defined as new onset retinopathy, progression of pre-existing retinopathy, cataract extraction, incident macular edema, need for focal or scatter photocoagulation, blindness, or change in visual acuity.
Nephropathy: Incident nephropathy was defined as development of microalbuminuria and progression of nephropathy as progression from micro- to macroalbuminuria or progression to end-stage renal disease requiring renal replacement therapy. Other nephropathy outcomes included change in glomerular filtration rate/creatinine clearance.
Neuropathy: Data on peripheral and autonomic neuropathy were recorded. Peripheral neuropathy was defined as an abnormal neurological exam, subjective symptoms, abnormal biothesiometry, or abnormal nerve conduction study. Autonomic neuropathy was defined as abnormal R-R interval, orthostatic hypotension, or resting tachycardia.
Coronary Artery Disease (CAD): CAD morbidity was defined as non-fatal myocardial infarction, angina, ischemic heart disease, congestive heart failure secondary to ischemic heart disease, coronary artery bypass surgery, and angioplasty. CAD mortality was defined as fatal myocardial infarction or sudden cardiac death.
Cerebrovascular Disease: Cerebrovascular morbidity was defined as non-fatal stroke, transient ischemic attack, or need for carotid endarterectomy. Cerebrovascular mortality was defined as fatal stroke.
Peripheral Arterial Disease (PAD): PAD was defined as claudication, peripheral revascularization procedure (angioplasty, bypass surgery, stenting), gangrene, limb amputation, decreased ankle-brachial index, and decreased arm-toe gradient.
Others: Outcome data were collected on congestive heart failure not related to ischemic heart disease and presence of atherosclerosis (i.e. carotid intimal-medial thickness, abdominal aortic aneurysm).
Articles published in English language from 1966 to April 2002 were identified by searching PubMed® using MESH and text words for diabetes, proteinuria, and cardiovascular or renal outcomes. To obtain additional references not otherwise identified through the electronic search, we searched bibliographies of relevant primary and review articles from our electronic search.
After identification of citations through PubMed®, all abstracts were reviewed for relevance by a single abstractor. All articles to be potentially included in the final review underwent double review by study authors for data abstraction. Differences in opinion were resolved through consensus adjudication. Data abstracted included 1) study design, 2) study location, 3) numbers of study subjects enrolled, 4) study exclusion criteria, 5) type of diabetes studied and numbers of persons with each type of diabetes included in study, 6) measurement and definitions of microalbuminuria, 7) descriptive information about study participants (including age, gender, race, duration of diabetes, glycemic control, baseline diastolic and systolic blood pressure), 8) baseline and followup measures of urinary albumin excretion, and 9) baseline and followup measures of kidney function (e.g. serum creatinine, creatinine clearance, or direct measurement of glomerular filtration rates). For randomized controlled trials featuring medication interventions (e.g. ACE inhibitors vs. other anti-hypertension agents) the type of medication, dose and frequency were also recorded.
Urine albumin: Data were abstracted on the biochemical method of urine measurement, including the timing of urine specimens and the cutoffs used to define different levels of urine albumin exposure at baseline. Microalbuminuria was classified according to the method of ascertainment in each study. Studies reporting only on persons with macroalbuminuria (levels greater than those defined above, unless authors defined microalbuminuria in some other fashion) were excluded from our analysis.
Diabetes mellitus: Data were extracted on study participants' type of diabetes and the reported method of diagnosis/confirmation of diabetes within each study.
Renal Outcomes We divided outcomes reflecting decline in renal function over time into eight categories: 1) change in glomerular filtration rate (GFR) at end of study, 2) rate of change in GFR, 3) change in creatinine clearance at end of study, 4) rate of change in creatinine clearance, 5) change in 1/serum creatinine at end of study, 6) rate of change in 1/serum creatinine), 7) doubling of serum creatinine, or 8) need for renal replacement therapy, including dialysis and transplantation.
Cardiovascular (CVD) Outcomes We defined cardiovascular outcomes as the: 1) incidence of all cause death, 2) incidence of composite CVD deaths 3) incidence of death due to myocardial infarction, 4) incidence of death due to cerebrovascular accident, 5) incidence of CVD morbidity, and 6) incidence of composite CVD morbidity and mortality.
Coronary Artery Disease (CAD): CAD morbidity was defined as non-fatal myocardial infarction, angina, ischemic heart disease, congestive heart failure secondary to ischemic heart disease, coronary artery bypass surgery, and angioplasty. CAD mortality was defined as fatal myocardial infarction or sudden cardiac death.
Cerebrovascular Disease: Cerebrovascular morbidity was defined as non-fatal stroke, transient ischemic attack, or need for carotid endarterectomy. Cerebrovascular mortality was defined as fatal stroke.
Peripheral Arterial Disease (PAD): PAD was defined as claudication, peripheral revascularization procedure (angioplasty, bypass surgery, stenting), gangrene, limb amputation, decreased ankle-brachial index, and decreased arm-toe gradient.
Others: Outcome data were collected on congestive heart failure not related to ischemic heart disease and presence of atherosclerosis (i.e. carotid intimal-medial thickness, abdominal aortic aneurysm).
Two investigators reviewed titles and abstracts of identified articles and appropriate studies were selected for data abstraction. Articles chosen for abstraction were reviewed by two reviewers to ensure that all relevant data had been obtained and were correct.
What is the risk relationship between glycated hemoglobin and the subsequent risk of microvascular diabetic complications (retinopathy, nephropathy, neuropathy) in individuals with type 1 and type 2 diabetes?
The evidence reported supports a strong, graded relation between glycated hemoglobin exposure and the risk of two major microvascular complications of type 1 and type 2 diabetes, retinopathy and nephropathy. These patterns are observed for various measures of glycated hemoglobin (i.e., HbA1c, HbA1, and total GHb).
The preponderance of the evidence from cohort studies shows a strong relation between glycated hemoglobin and incident retinopathy, incident proliferative retinopathy and macular edema, and progression of retinopathy. Compared to the lowest categories of glycated hemoglobin exposure, the unadjusted relative risks for incident retinopathy were 4 to 7 times greater in the highest categories of glycated hemoglobin exposure for type 1 diabetes and 1.5 to 2.5 times greater in individuals with type 2 diabetes. The unadjusted relative risk for proliferative retinopathy was 6 to 7 times greater in individuals in the highest category of glycated hemoglobin exposure compared to the lowest category for individuals with type 1 diabetes and the unadjusted relative risk was 3 to 13 times greater for individuals with type 2 diabetes, although fewer studies examined this outcome in type 2 diabetes.
This relation between glycated hemoglobin exposure and retinopathy is confirmed in several randomized clinical trials of individuals with type 1 and type 2 diabetes, which show comparable risk reductions in these outcomes in individuals randomized to intensive therapy, where the HbA1c levels were maintained at approximately 7 percent, compared to individuals randomized to conventional therapy, where the mean HbA1c levels were maintained at approximately 9 percent.
Only a few studies address the relation between glycated hemoglobin and the risk of blindness; however, the majority suggest that increased glycated hemoglobin is a risk factor for blindness in individuals with type 1. With the exception of one cohort study and one clinical trial, there are virtually no data on the relation between glycated hemoglobin and risk of blindness in individuals with type 2 diabetes.
There are very few studies examining the relation between glycated hemoglobin and the incidence of cataracts.
The majority of studies evaluating the relation between glycated hemoglobin and the risk of nephropathy have evaluated the risk of developing microalbuminuria. These data show a strong and significant relation between glycated hemoglobin and the risk of microalbuminuria in individuals with type 1 and type 2 diabetes. Compared to individuals in the lowest category of glycated hemoglobin exposure, those in the highest category had a 3 to 9-fold unadjusted increased risk of microalbuminuria for type 1 diabetes and a 1.4 to 8-fold increased risk of microalbuminuria for type 2 diabetes. This is supported by clinical trial data that show significant risk reductions for incident microalbuminuria for individuals randomized to intensive glycemic control, where the mean HbA1c levels were maintained at approximately 7 percent, compared to those randomized the conventional glycemic control, where the mean HbA1c levels were maintained at approximately 9 percent. Among individuals with type 1 diabetes, the unadjusted relative risk reductions were 34 percent to 43 percent, compared to 60 percent to 74 percent for individuals with type 2 diabetes.
Although fewer data exist on the relation between glycated hemoglobin and risk of macroalbuminuria and on the relation between glycated hemoglobin and the risk of nephropathy progression, several cohort studies and clinical trials support a strong and significant positive association in individuals with type 1 and type 2 diabetes.
The only studies identified by our search strategy and inclusion criteria examining the effect of glycated hemoglobin exposure on GFR were cohort studies conducted in individuals with type 1 diabetes. All studies consistently demonstrated that increasing levels of glycated hemoglobin were associated with a decline in GFR. There are no clinical trial data examining the GFR outcomes and there are no data on the relation between glycated hemoglobin and GFR in individuals with type 2 diabetes.
There are very few studies examining the association between glycated hemoglobin and the risk of end-stage renal disease (ESRD).
Among individuals with type 1 diabetes, there appears to be a strong, positive association between glycated hemoglobin and the risk of peripheral neuropathy in both cohort studies and clinical trials; however, the evidence of an association between glycated hemoglobin and peripheral neuropathy in individuals with type 2 diabetes yields conflicting results.
There are fewer data on the association between glycated hemoglobin and the risk of autonomic neuropathy. In individuals with type 1 diabetes, there appears to be a positive association. There are very little data on the relation between glycated hemoglobin and the risk of autonomic neuropathy in individuals with type 2 diabetes.
What is the risk relationship between glycated hemoglobin and macrovascular diabetic complications (coronary artery disease, cerebrovascular disease, and peripheral vascular disease) in individuals with type 1 and type 2 diabetes?
In the cohort studies evaluating cardiovascular outcomes in individuals with diabetes, there was a positive association with glycated hemoglobin exposure; however, the risk estimates are much smaller compared to the risk estimates for the microvascular complications.
The preponderance of the evidence from cohort studies shows a positive association between glycated hemoglobin and risk of fatal and non-fatal coronary artery disease, particularly among individuals with type 2 diabetes. Compared to those in the highest category of glycated hemoglobin exposure, the unadjusted risk of fatal and non-fatal coronary artery disease was 50 percent to 70 percent (relative risk, 1.5 to 1.7) greater than for those in the lowest category of exposure.
There are little data on the relation between CAD and glycated hemoglobin among individuals with type 1 diabetes; however most studies have shown a positive association.
The relation between glycated hemoglobin and the risk of PAD appears to be strong and positive in individuals with type 1 and type 2 diabetes. Compared to those in the lowest category of glycated hemoglobin exposure, the unadjusted risk of PAD was 5 to 6 times greater in individuals in the highest category of exposure among individuals with type 1 diabetes, and the risk was 2 to 4 times greater among individuals with type 2 diabetes.
The risk relationship between cerebrovascular disease and glycated hemoglobin, which has only been examined among individuals with type 2 diabetes, is less clear.
There are very little data on the relation between glycated hemoglobin exposure and congestive heart failure or subclinical atherosclerosis, assessed by carotid intimal-medial thickness, making it difficult to draw any conclusions regarding these outcomes.
Only a few studies have examined the presence of a threshold effect of glycated hemoglobinglycated hemoglobin on the risk of developing diabetic complications (i.e. a level of glycated hemoglobin above which there is a non-constant or exponential increase in risk of complications). The majority of these studies have not found a threshold effect for retinopathy and nephropathy outcomes but have rather demonstrated a continuous risk of complications with increasing glycated hemoglobin levels. There are very few studies that have attempted to examine the presence of a threshold effect of glycated hemoglobin on neuropathy and macrovascular outcomes.
What is the risk relationship between microalbuminuria and renal function?
11 studies reported on this question.
The analyses for this question had important limitations including broad variation in methods of assessing levels of urine albumin excretion as well as substantial heterogeneity in reporting of renal outcomes.
The preponderance of evidence suggests that the presence of microalbuminuria at baseline is associated with progression of chronic kidney disease.
The relation of urine albumin excretion at baseline to progression of chronic kidney disease appears graded; higher levels of urine albumin excretion at baseline are associated with a greater magnitude of decrease in renal function as well as a faster rate of decline in renal function over time.
What is the risk relationship between microalbuminuria, cardiovascular disease, and death?
19 studies reported on cardiovascular morbidity and mortality, and 24 reported on all-cause mortality.
The analyses for this question had important limitations including broad variation in methods of assessing levels of urine albumin excretion as well as few studies focusing on disease specific cardiovascular morbidity or mortality.
The preponderance of evidence from these studies demonstrates an association between microalbuminuria at baseline and increased risk of cardiovascular morbidity, cardiovascular mortality, and all-cause mortality.
The relation of urine albumin excretion at baseline to cardiovascular morbidity, cardiovascular mortality, and all-cause mortality appears graded; greater levels of urine albumin excretion at baseline are independently associated with a greater magnitude of risk of cardiovascular morbidity, cardiovascular mortality, and all-cause mortality over time.
For research on glycemic control, future cohort studies and clinical trials should focus on studying the relation between glycated hemoglobin exposure and the risk of macrovascular complications. There are much fewer data on these outcomes than there are on microvascular outcomes; however, more data are also needed on the relation between glycated hemoglobin and the risk of neuropathy, particularly the risk of peripheral and autonomic neuropathy in individuals with type 2 diabetes.
For research on urine albumin, future work should seek to define the optimal and most feasible tests for measuring microalbuminuria and to standardize measurement of microalbuminuria. Future research should also characterize the nature of the relation (e.g. threshold versus linear) between microalbuminuria and outcomes. In addition, further work is needed to understand whether currently accepted definitions of microalbuminuria are optimal in predicting future renal and cardiovascular outcomes.
Extension of research in both these areas has important future implications for gaining improved understanding of the role of glycemic control in the prevention of the cardiovascular sequelae as well as for future development of guidelines for screening practices among persons with types 1 and 2 diabetes mellitus.
Clinical testing to assess levels of disease control and progression among persons with types 1and 2 diabetes mellitus is widely recommended to clinicians to improve patients' clinical outcomes. (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]) Two important foci of recommendations for the follow-up care of individuals with diabetes include monitoring of glycemic status by measurement of glycated hemoglobin and screening for kidney disease with urine microalbumin to assess overall disease progression and to detect potential progression toward end-organ damage. According to the American Diabetes Association Clinical Practice Recommendations, monitoring of glycemic status is considered a cornerstone of diabetes care and affects how physicians and patients adjust medical therapy as well as behavioral therapy, such as diet and exercise. Screening for urine albumin among persons with diabetes is also widely recommended for the detection and treatment of incipient diabetic nephropathy, which is currently defined as the excretion of small amounts of albumin in the urine (30–200mg albumin per 24 hours) and affects physician's implementation of therapy to slow progression of kidney disease.
Despite widespread recommendations for screening of persons with diabetes for both glycemic control and urine albumin, there has not been a systematic assembly of the literature to assess the risk relation between tests assessing long-term glycemic control or tests assessing the presence of microalbuminuria with cardiovascular, peripheral vascular, renal, and neurological outcomes (all of which represent end-organ effects of long-term diabetes). This report, commissioned by the American Association of Clinical Chemistry, systematically reviews the literature identifying the risk relation between testing for glycemic control or urine albumin with these important clinical outcomes.
Prior to implementation of patient self-monitoring of blood glucose (SMBG), patients tested their glycemic status using urine glucose testing; however, this provided only a rough estimate of glycemic control and provided no information regarding the level of glycemia below the renal threshold of 180 mg/dL (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]). The advent of SMBG has allowed patients to attain glycemic goals more quickly and has revolutionized the care of individuals with diabetes mellitus; however, it does not provide information regarding glycemic control over an extended period of time (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]). Measurement of glycated hemoglobin, which first began in the 1970's, has become the preferred method of assessing long-term glycemic control. Glycated hemoglobin describes a series of stable minor hemoglobin components formed by a post-translational modification in HbA1 in which glucose is nonenzymatically bound to the alpha chain of hemoglobin (Goldstein, et al.1995 [4644]). Rahbar et al. first showed that these minor HbA1 fractions were elevated in individuals with diabetes (Rahbar, et al. 1969; [92177]; Rahbar, et al. 1986 [92178]). HbA1 can be separated into three minor components, HbA1a, HbA1b, and HbA1c (see Appendix C). HbA1c was found to be the specific glycated hemoglobin resulting from post-translational modification of HbA1 by glucose and correlated with various measures of glucose (Goldstein, et al. 1995 [4644]).
Of the various glycated hemoglobin fractions, HbA1c is the preferred gold standard for assessing glycemic control over the previous 2–3 months. Glycated hemoglobin results should be expressed as percent HbA1c or percent HbA1c equivalents (if total GHB is actually measured). The American Diabetes Association began to make treatment recommendations based on HbA1c following publication of the results of the Diabetes Control and Complications Trial in the 1990's; however, at that time, there was a lack of standardization of measurement methodologies, which hindered implementation of the guidelines (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]). As a result, the National Glycohemoglobin Standardization Program (NGSP), partly sponsored by the American Diabetes Association, was started in 1996 http://web.missouri.edu/~diabetes/ngsp.html (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]). A follow-up study of the program reported that in 2000, 90 percent of surveyed laboratories reported glycated hemoglobin results as HbA1c or equivalent and that 78 percent of these used a NGSP-certified method (Little, et al. 2001 [4643]). The current ADA recommendations related to the use of glycated hemoglobin are that (1) all labs use only HbA1c assay methods that have been certified, (2) all labs participate in the College of American Pathologists proficiency testing survey, (3) all results be reported as “percent HbA1c” or “percent HbA1c equivalent” and (4) the goal HbA1c should be <7 percent with a therapeutic action target of >8 percent, using a DCCT traceable method (American Diabetes Association: Clinical Practice Recommendations 2002 [92176]). The HbA1c has become the gold standard for the therapeutic management of diabetes mellitus in research and in the clinical setting. The test results, however, are affected by diseases that shorten the lifespan of the red blood cells (i.e. hemolytic anemias) or diseases that are associated with hemoglobin variants (i.e hemoglobins S, C, and F), resulting in HbA1c values that may not accurately reflect long-term glycemic control (Bry, et al. 2001 [92179]).
While glycated hemoglobin testing gives an assessment of long-term glycemic control, what other prognostic information does it provide in the management of individuals with diabetes mellitus? Pirart first recognized the importance of poor glycemic control in the development of diabetic complications, as evidenced in his 1984 quotation: “The catastrophic loss of the macula will sometimes be prevented by a further destruction by laser of the already dying retina. Haemo- and peritoneal dialysis are a pitiful way leading to kidney transplantation with its disappointing surgical, infectious, and immunological failures. All that could have been avoided or at least much delayed, had hyperglycaemia be better mitigated in the long run and from the very beginning of a diabetic's career.”
Pirart and Lauvaux conducted the first prospective cohort study of the relation between hyperglycemia and microvascular complications in individuals with diabetes mellitus (Pirart 1995 [4642]). The study of 4,400 individuals with diabetes followed for 25 years from the onset of their disease or from a later time demonstrated that retinopathy and nephropathy were significantly related to the duration of diabetes and to sustained hyperglycemia, independent of age, gender, body weight, family history of diabetes, and severity of diabetes (Pirart 1995 [4642]). In this early study, glycemic control was assessed based on multiple daily urine tests to detect glucosuria and blood analyses obtained during clinic visits (Pirart 1995 [4642]).
Since that time, other cohort studies have confirmed Pirart's observations and several large, randomized clinical trials have demonstrated that intensive glycemic control prevents the development and progression of long-term diabetic microvascular complications. In these studies, glycemic goals were assessed using glycated hemoglobin as the measure of long-term control.
Long-term hyperglycemia, as measured by glycated hemoglobin, is clearly related to the development of diabetic microvascular complications; however, its relation to the development of macrovascular complications is less clear. A meta regression analysis performed by Coutinho et al. demonstrated a significant association between fasting, 1-hour, and 2-hour glucose and cardiovascular events and mortality in individuals without diabetes (Coutinho et al. 1999 [4645]). The relation of glycemic control and cardiovascular disease in individuals with diabetes remains controversial, with some studies demonstrating a positive association and others showing no association. Impaired glucose homeostasis may be causal in its relationship to cardiovascular disease or alternatively, both diabetes and cardiovascular disease may arise from a “common soil” of insulin resistance, resulting in the observed association (Donahue, et al. 1992 [4649]). Pooled analyses of glucose intervention trials aimed at improving glycemic control have shown insignificant reductions in cardiovascular mortality (Lawson, et al. 1999 [4646]; Huang, et al. 2001 [4647]).
Another important issue that is still an area of active investigation in the management of diabetes relates to whether there is a threshold effect of glycated hemoglobin for microvascular and/or macrovascular complications—that is, is there a glycated hemoglobin level above which the risk of complications rises exponentially. While the majority of studies have not found a threshold effect, a few studies suggest a threshold level between 8 percent and 9 percent for retinopathy (Klein, et al. 1988 [92172]; Janka, et al. 1989 [92173]; Danne, et al. 1994 [92174]) and nephropathy (Krolewski, et al. 1995 [92175]). The issue of a threshold effect of glycated hemoglobin has important implications for where the HbA1c treatment targets are set to prevent diabetic complications. At present, the ADA recommends an HbA1c target of ≤7 percent for the prevention of complications; however, the American College of Endocrinology (ACE) recently altered its recommendations such that a HbA1c target of ≤6.5 percent is recommended for prevention of diabetic complications (ACE Consensus Development Conference on Guidelines for Glycemic Control ACE, 2001 [92180]) and this new target has been adopted by the American Association of Clinical Endocrinologists (American Association of Clinical Endocrinologists Medical Guidelines for Management of Diabetes Mellitus (ACCE), 2002 [92181]). Also, if there is a threshold effect of HbA1c for complications, it is important to determine whether it is the same for both microvascular and macrovascular complications.
Great interest in screening and monitoring urinary albumin excretion in patients with diabetes lies not only in the fact that diabetes mellitus is the most common cause of incident ESRD (resulting in annual costs to Medicare of over $15 billion) but also in the fact that microalbuminuria is associated with increased cardiovascular morbidity, cardiovascular mortality, and total mortality in patients with diabetes mellitus in select studies. (Dinneen, et al. 1997 [6]; Borch-Johnsen, et al. 1987 [5]; U.S. Renal Data System 2002 [20]; Donahue, et al. 1992 [4])
Screening tests for microalbuminuria are recommended annually for patients with type 1 diabetes of greater than 5 years duration and for all patients with type 2 diabetes from the time of diagnosis. (American Diabetes Association 2002 [2]) Twenty-four hour collection of urine for quantitative assessment of urinary albumin excretion rate is currently considered the gold standard measurement of microalbuminuria. However, this method is frequently considered to be cumbersome and difficult to carry out in the outpatient clinical setting, and it is subject to timing and collection inaccuracies. (American Diabetes Association 2002 [2]; Remuzzi, et al. 2002 [7]; Bakker 1999 [8]) Several other methods of testing, which are considered less difficult to perform in outpatient settings, have been studied and have been demonstrated to have varying degrees of correlation with 24-hour urine collection for the detection of urinary albumin. These include random or ‘spot’ testing of a morning urine specimen for urine albumin concentration or albumin to creatinine ratio, overnight or ‘timed’ urine collections for estimation of albumin excretion rates, and dipstick testing. (American Diabetes Association 2002 [2]; Leong, et al. 1998 [16]; Mogensen, et al. 1997 [15]; Piehlmeier, et al. 1998 [14]; Tiu, et al. 1993 [13]; Eshoj, et al. 1987 [12]; Nathan, et al. 1987 [11]; Schwab, et al. 1992 [10]; Assadi 2002 [9]) While many experts now support the use of random or first morning tests for urinary albumin to creatinine ratio as a convenient and accurate approach to screening patients, there is currently no clear consensus on standardized testing methods. (American Diabetes Association 2002 [2]; Nathan, et al. 1987 [11])
In a variety of prospective studies, elevated urinary albumin excretion has been shown to be associated with increased risk of progression of kidney disease toward ESRD as well as increased cardiovascular morbidity, cardiovascular mortality, and total mortality. Results of a recent systematic review of the association of microalbuminuria with subsequent cardiovascular morbidity and mortality among persons with non-insulin dependent diabetes mellitus demonstrate a 2–3 fold increase in risk of these outcomes associated with the presence of microalbuminuria at baseline. (Dinneen, et al. 1997 [6]) However, few studies have systematically ascertained the magnitude of increase in both renal and cardiovascular risk associated with microalbuminuria among both persons with both type 1 and type 2 diabetes.
Moreover, although observations of renal and cardiovascular outcomes among persons classified as having microalbuminuria by currently accepted standards have been reported, it remains unclear whether current definitions of microalbuminuria are optimal in terms of predicting renal and cardiovascular outcomes and if there is a dose-relationship or threshold effect in prediction of outcomes associated with urinary albumin excretion. Knowledge of the pooled magnitude of risk associated with current definitions of microalbuminuria in addition to an improved understanding of the risk relationship of varying levels of baseline microalbuminuria with cardiovascular and renal outcomes could have important implications in screening and treatment recommendations for persons with diabetes.
With these issues taken into account, we asked the following key questions concerning both testing for glycemic control and urine albumin for persons with both type 1 and type 2 diabetes mellitus:
What is the risk relationship between glycated hemoglobin and the subsequent risk of microvascular diabetic complications (retinopathy, nephropathy, neuropathy) in individuals with type 1 and type 2 diabetes?
What is the risk relationship between glycated hemoglobin and the subsequent risk of macrovascular diabetic complications (coronary heart disease, cerebrovascular disease, peripheral vascular disease) in individuals with type 1 and type 2 diabetes?
What is the risk relationship between microalbuminuria and renal function?
What is the risk relationship between microalbuminuria, cardiovascular disease, and death?
The Evidence-based Practice Center (EPC) recruited six technical and community experts to provide input into the definition of the key questions and to review a draft of the report. The EPC also recruited representatives from a range of other stakeholder organizations to serve as peer reviewers of the draft Evidence Report. These stakeholder organizations included organizations of physicians, allied health professionals, and third party health care payers in addition to consumer organizations.
Because of the divergent content of questions related to glycemic control and urine albumin, two separate teams of investigators systematically reviewed literature in these two areas. Investigators from each team executed systematic search strategies pertinent to each set of questions. Thus, search strategies and studies identified were different for questions on glycemic control and urine albumin. This chapter details methodology used to systematically review literature pertaining to each set of questions.
Articles published in English language from 1966 to April 2002 were identified by searching MEDLINE® accessed through PubMed® using MESH and text words for glycohemoglobin, diabetes, and individuals diabetic complications, including retinopathy, nephropathy, neuropathy, and cardiovascular disease. To identify studies that measured glycated hemoglobin exposure, the MESH terms “hemoglobin A, glycosylated” and “hyperglycemia” and the following text terms were used: “A1c,” “HbA1c,” “HbA1,” “GHb,” “glycohemoglobin,” “glycohaemoglobin,” “glycosylated,” “hemoglobin,” “glycosylaed,” “haemoglobin,” “glycemia,” “glycaemia,” “hyperglycemia,” “hyperglycaemia.” The MeSH term “diabetes mellitus” and the text term “diabetes” were used to identify studies of persons with diabetes mellitus. To identify studies focusing on diabetic retinopathy, we used the MESH term “diabetic retinopathy” and the text terms “retinopathy,” “eye-disease,” “retinal-disease,” “eye,” and “eyes.” To identify studies reporting nephropathy outcomes, we used the MESH terms “diabetic nephropathies,” “albuminuria,” “glomerular filtration rate,” “kidney disease,” and “kidney failure” and the text words “nephropathy” and “albuminuria.” To identify studies reporting neuropathy outcomes, we used the MESH term “diabetic neuropathies” and the text terms “neuropathy” and “neurological.” To identify studies reporting cardiovascular outcomes, we used the MESH term “cardiovascular diseases” and the following text terms: “angina,” “cardiovascular artery disease,” “cardiovascular disease,” “cardiovascular heart disease,” “congestive heart disease,” “ischemic heart disease,” “ischaemic heart disease,” “myocardial infarction,” “peripheral artery disease,” “stroke,” and “transient ischemic attack.” To obtain additional references not otherwise identified through the electronic search, we searched bibliographies of relevant primary and review articles from our electronic search.
We restricted our review to prospective longitudinal cohort studies, non-concurrent prospective cohort studies, and clinical trials that had data on glycated hemoglobin exposure and outcome data on individual microvascular and macrovascular complications during at least 1 year of follow-up in at least 50 participants with type 1 or type 2 diabetes. Because we were interested in determining the risk relationship between glycated hemoglobin exposure and microvascular and macrovascular complications and in re-examining whether a threshold exists for this relation, we only included studies that reported prospective, quantitative risk data (i.e. incidence rates (IR), regression coefficients with standard errors reported separately; relative hazards (RH), relative odds (OR), relative risk (RR)). Retrospective case-control studies that reported and compared previous glycated hemoglobin values in individuals with a given outcome versus those without a given outcome were excluded. Articles that reported data in graphical form in which specific quantitative values could not be determined were also excluded. If outcome risks were reported by glycated hemoglobin level, the levels had to be quantitatively defined for a study to be included. Similarly, if the outcome risk was reported for each unit change in glycated hemoglobin, the mean glycated hemoglobin of the study population had to be reported for a study to be included so that there was adequate exposure summary data.
Other exclusion criteria included review articles, animal or in vitro studies, non-English language, and studies in which the design was unclear.
Glycosylated hemoglobin: Data were abstracted on the biochemical method of measurement, whether the method was traceable to the DCCT standard and/or whether the lab was certified by the NGSP, and how glycosylated hemoglobin was reported (i.e. HbA1c, HbA1, total GHB)(See Appendix C). Summary data on glycated hemoglobin exposure was recorded as reported in each study—mean or median overall glycated hemoglobin and/or categories of glycated hemoglobin exposure.
Diabetes mellitus: Data were extracted on the type of diabetes that study participants had and the method of diagnosis/confirmation.
Retinopathy: Incident retinopathy was defined as new onset retinopathy, progression of pre-existing retinopathy, cataract extraction, incident macular edema, need for focal or scatter photocoagulation, blindness, or change in visual acuity. When reported, the method of assessment of retinopathy was recorded (i.e. fundus photography, direct ophthalmoscopy) as well as the grading scale for retinopathy (i.e. Early Treatment of Diabetic Retinopathy Study [ETDRS] scale; see Appendix C).
Nephropathy: Incident nephropathy was defined as development of microalbuminuria and progression of nephropathy as progression from micro- to macroalbuminuria or progression to end-stage renal disease requiring renal replacement therapy. Other nephropathy outcomes included change in glomerular filtration rate/creatinine clearance. Data were recorded on how albumin excretion was defined (i.e. albumin excretion rate, albumin:creatinine ratio, dipstick) and the biochemical methods of measuring urinary albumin and creatinine.
Neuropathy: Data on peripheral and autonomic neuropathy were recorded. Peripheral neuropathy was defined as an abnormal neurological exam, subjective symptoms, abnormal biothesiometry, or abnormal nerve conduction study. Autonomic neuropathy was defined as abnormal R-R interval, orthostatic hypotension, or resting tachycardia.
Coronary Artery Disease (CAD): CAD morbidity was defined as non-fatal myocardial infarction, angina, ischemic heart disease, congestive heart failure secondary to ischemic heart disease, coronary artery bypass surgery, and angioplasty. When reported, confirmatory ICD 9 codes were recorded. CAD mortality was defined as fatal myocardial infarction (ICD 9 code recorded if reported) and sudden cardiac death. If the death was confirmed by national or state death certificate registry, these data were recorded.
Cerebrovascular Disease: Cerebrovascular morbidity was defined as non-fatal stroke, transient ischemic attack, or need for carotid endarterectomy. Cerebrovascular mortality was defined as fatal stroke. If the death was confirmed by national or state death certificate registry, these data were recorded.
Peripheral Arterial Disease (PAD): PAD was defined as claudication, peripheral revascularization procedure (angioplasty, bypass surgery, stenting), gangrene, limb amputation, decreased ankle-brachial index, and decreased arm-toe gradient.
Others: Outcome data were collected on congestive heart failure (CHF) not related to ischemic heart disease and presence of atherosclerosis (i.e. carotid intimal-medial thickness (IMT), abdominal aortic aneurysm).
Data were recorded on other variables including treatment (if indicated), demographic characteristics, metabolic characteristics, study design, study location, source of study participants, and exclusion criteria.
Two investigators reviewed titles and abstracts of identified articles and appropriate studies were selected for data abstraction. Three investigators extracted data from each article onto standardized data collection forms. Each abstracted article was reviewed by a second reviewer to ensure that all relevant data had been obtained and was correct. All three reviewers are board-certified internists and one reviewer is an endocrinologist cross-trained in epidemiology and specializing in diabetology.
Articles published in English language from 1966 to April 2002 were identified by searching MEDLINE® accessed through PubMed® using MESH and text words for diabetes, proteinuria, and cardiovascular or renal outcomes. The MESH term “diabetes mellitus” and the text term “diabetes” was used to identify studies of persons with diabetes mellitus. To identify studies focusing on urine albumin, we used the MESH terms “proteinuria,” “kidney diseases,” and “albuminuria,” and the following text terms: “albumin excretion”, “albuminuria,” “microalbuminuria,” “nephropathy,” and “proteinuria.” To identify studies reporting on renal, cardiovascular outcomes, and death, we used the MESH terms “glomerular filtration rate,” “kidney failure,” and “peripheral vascular diseases” and the following text words: “angina,” coronary artery disease,” “cardiovascular disease,” “cardiovascular heart disease,” “congestive heart disease,” “death,” “ischemic heart disease,” “mortality,” “myocardial infarction,” “peripheral artery disease,” “stroke,” and “transient ischemic attack”. To obtain additional references not otherwise identified through the electronic search, we searched bibliographies of relevant primary and review articles from our electronic search.
After identification of citations through PubMed®, all abstracts were reviewed for relevance by a single abstractor. All articles to be potentially included in the final review underwent double review by study authors for data abstraction. Differences in opinion were resolved through consensus adjudication. Data abstracted included 1) study design, 2) study location, 3) numbers of study subjects enrolled, 4) study exclusion criteria, 5) type of diabetes studied and numbers of persons with each type of diabetes included in study, 6) measurement and definitions of levels of urine albumin exposure, 7) descriptive information about study participants (including age, gender, race, duration of diabetes, glycemic control, baseline diastolic and systolic blood pressure), 8) baseline and followup measures of urinary albumin excretion, and 9) baseline and followup measures of kidney function (e.g. serum creatinine, creatinine clearance, or direct measurement of glomerular filtration rates). For randomized controlled trials featuring medication interventions (e.g. ACE inhibitors vs. other anti-hypertension agents) the type of medication, dose and frequency were also recorded.
Urine albumin: Data were abstracted on the biochemical method of measurement of urine measurement, including the timing of urine specimens and the cutoffs used to define different levels of urine albumin exposure at baseline. Unless otherwise specified by study authors, we considered ‘microalbuminuria’ to be present among study subjects if urine albumin excretion was classified as being in the range of 30–200 mg/24 hours in 24 hour urine specimens, 20–199μg/min in 24 hour and other timed urine specimens, or 30–299μg albumin per mg creatinine in random or spot urine specimens. For studies assessing multiple cutoff levels for urine albumin exposure, persons with the least level of exposure (less than microalbuminuria) were considered to have ‘normoalbuminuria’, while persons with greatest level of exposure were considered to have ‘macroalbuminuria’. In cases where additional cutoffs for urine albumin level were made between non-detectable levels of urine albumin (or no albuminuria) and levels considered to be consistent with ‘microalbuminuria’ such cutoffs were deemed to represent ‘intermediate’ levels of exposure to urine albumin. Use of dipstick methods for ascertainment of microalbuminuria was also recorded. Studies reporting only on persons with macroalbuminuria (levels greater than those defined above, unless authors defined microalbuminuria in some other fashion) were excluded from our analysis.
Diabetes mellitus: Data were extracted on study participants' type of diabetes and the reported method of diagnosis/confirmation of diabetes within each study. To help organize this diverse literature, persons reported as having type 1, insulin dependent, or younger onset diabetes were classified as having type 1 diabetes in our analysis. Persons reported as having type 2, non-insulin dependent or, older onset diabetes were classified as having type 2 diabetes in our analysis.
We divided outcomes reflecting decline in renal function over time into eight categories: 1) change in glomerular filtration rate (GFR) at end of study, 2) rate of change in GFR, 3) change in creatinine clearance at end of study, 4) rate of change in creatinine clearance, 5) change in 1/serum creatinine at end of study, 6) rate of change in 1/serum creatinine), 7) doubling of serum creatinine, or 8) need for renal replacement therapy, including dialysis and transplantation.
We defined cardiovascular outcomes as the: 1) incidence of all cause death, 2) incidence of composite CVD deaths, 3) incidence of death due to myocardial infarction, 4) incidence of death due to cerebrovascular accident, 5) incidence of CVD morbidity, and 6) incidence of composite CVD morbidity and mortality.
CAD morbidity was defined as non-fatal myocardial infarction, angina, ischemic heart disease, congestive heart failure secondary to ischemic heart disease, coronary artery bypass surgery, and angioplasty. When reported, confirmatory ICD 9 codes were recorded. CAD mortality was defined as fatal myocardial infarction (ICD 9 code recorded if reported) and sudden cardiac death. If the death was confirmed by national or state death certificate registry, that data was recorded. Cerebrovascular morbidity was defined as non-fatal stroke, transient ischemic attack, or need for carotid endarterectomy. Cerebrovascular morbidity was defined as fatal stroke. If the death was confirmed by national or state death certificate registry, that data was recorded. PAD was defined as claudication, peripheral revascularization procedure (angioplasty, bypass surgery, stenting), gangrene, limb amputation, decreased ankle-brachial index, and decreased arm-toe gradient. Outcome data were also collected on congestive heart failure not related to ischemic heart disease and presence of atherosclerosis (i.e. carotid intimal-medial thickness, abdominal aortic aneurysm).
Two investigators reviewed titles and abstracts of identified articles and appropriate studies were selected for data abstraction. Articles chosen for abstraction were reviewed by two reviewers to ensure that all relevant data had been obtained and was correct.
All reviewed articles for both questions were rated for the quality of reporting on several aspects of study external and internal validity, including 1)description of inclusion and exclusion criteria for study subjects, 2) description of study subjects' baseline characteristics, 3) description of study non-enrollees, 4)description of the study intervention, 5) length of study followup, 6)study subject attrition, 7)statistical analysis, and (for RCTs) 8) quality of randomization and 9) blinding. Based on work by Chalmers, et al, a quality score ranging from 0–100 was developed, in which ratings of reporting of external and internal validity received 70 percent of the total quality score, and aspects of statistical analysis of data received 30 percent of the total quality score. (Chalmers, et al. 1981 [565].
For studies in which the exposure data for glycated hemoglobin and microalbuminuria and baseline characteristics were given by the clinical outcomes but the risk analyses were conducted on the entire cohort, a weighted average for the whole group was calculated according to the following formulae:
Weighted mean calculations:
Our computerized MEDLINE® search identified 3336 potentially relevant abstracts. Of these 3336, abstracts were eliminated for the following reasons: no diabetic participants (n= 55), no information relevant to our study questions (n=1669), no human data (n=7), fewer than 50 participants (n=403), no original data (i.e. review articles; n=135), no prospective data (i.e. cross-sectional studies; n= 472), only special diabetic subgroups studied (i.e. diabetes in pregnancy, transplant patients; n=256), data only given in graphical form (n=3), glycated hemoglobin was a confounder and not an exposure variable (n=2), less than 1 year of followup (n=3), and study design unclear (n=1). This left 330 articles for review. Of these, articles were eliminated for the following reasons: no diabetic participants (n=1), outcomes not analyzed specifically in diabetic individuals (n= 5), no information relevant to our study question (n=63), data only given in graphical form (n=8), glycated hemoglobin was a confounder and not an exposure variable (n=6), fewer than 50 participants (n=15), no original data (i.e., review articles; n=2), no prospective data (n=25), no risk data reported (n=57), glycated hemoglobin exposure not clearly defined (n=13), less than 1 year of followup (n=3), study design unclear (n=1), outcomes not clearly defined (n=5), diabetes type not specified (n= 12), and study represented subgroup analysis of main study results already abstracted (n=2). Data were abstracted from the remaining 112 articles.
Of the remaining 112 studies, 79 were cohort studies, 24 were cohort studies from previous clinical trial populations, and 9 were randomized clinical trials. For summary purposes, the cohort studies from clinical trials are grouped together with the other cohort studies. The studies were conducted in diverse populations and included 18 studies in Asia, 44 studies in Europe, 1 study in Israel, 44 studies in North America, 4 studies in New Zealand, and 1 study in South America. The majority of studies were conducted in a single city (n= 62) or a single country (n=37). Seven studies were conducted in more than one country. The majority of participants were recruited from a hospital based specialty clinic (n=18), an unspecified hospital based clinic (n= 14), an outpatient specialty (n= 15) or general clinic (n=13), or from a previous or concurrent clinical trial (n=14).
Among the cohort studies, 51 were conducted in individuals with type 1 diabetes, 40 were conducted in individuals with type 2 diabetes, and 12 included individuals with both type 1 and type 2 diabetes. Among the clinical trials, 6 were conducted in individuals with type 1 diabetes and 3 were conducted in individuals with type 2 diabetes. For the summary purposes, individuals described as IDDM or younger/juvenile onset diabetes were considered to have type 1 diabetes and individuals described as NIDDM or older onset diabetes were considered to have type 2 diabetes. Some studies contained individuals with both type 1 and type 2 diabetes.
Glycated hemoglobin data was reported as HbA1c in 60, as HbA1 in 34, and as total GHb in 18. All randomized controlled trials reported glycated hemoglobin exposure data as HbA1c. The glycated hemoglobin assay method was DCCT traceable in 16 studies and was not DCCT traceable or not stated in the remainder of the studies. A variety of biochemical measurement methods were used; however, the most common method was HPLC.
Incidence of Retinopathy
In one study, the glycated hemoglobin exposure variable was total GHb. In this study, there was a 40 percent increased risk of incident retinopathy for each 1 percent increase in total GHb in a population with an average total GHb of 12.5 percent (Klein, et al. 1988 [2832]).
In three studies, the glycated hemoglobin exposure variable was total GHb and these studies included 610 to 1780 individuals followed for 4 to 10 years (Klein, et al. 1995 [2437]; Klein, et al. 1989 [2787]; Klein, et al. 1988 [2832]). The odds of macular edema increased with increasing quartiles of total GHb, with a fourfold increased risk in individuals with a total GHb≥12.1 percent compared to those with a GHb≤9.4 percent in one study (OR=4.0; 95 percent CI: 2.2 to 7.1) (Klein, et al. 1995 [2437]) and a nine fold increased risk in individuals with a GHb≥14.2 percent compared to those with a GHb≤10.8 in another study (OR=8.9; 95 percent CI: 2.7 to 29.4) (Klein, et al. 1989 [2787]). In two studies, there was approximately 50 percent increased risk of macular edema and proliferative retinopathy for each 1 percent increase in HbA1 following multivariate adjustmen (Klein, et al. 1995 [2437]; Klein, et al. 1988 [2832]).
Three cohort studies used total GHb as the exposure variable and included 652 to 1780 individuals followed for 4 to 10 years (Klein, et al. 1995 [2437]; Klein, et al. 1989 [2787]; Klein, et al. 1988 [2832]). The risk of proliferative retinopathy/macular edema increased with increasing GHb and individuals in the highest quartile were nearly 14 times more likely to develop macular edema compared to those in the lowest quartile in one study (GHb≥10.9 percent vs. GHb≤8.1 percent; OR=13.7; 95 percent CI: 6.8 to 27.8) (Klein, et al. 1995 [2437]) and 12 times more likely to develop macular edema in another study (GHb≥12.7 percent vs. GHb≤9.2 percent; OR=12.5; 95 percent CI: 3.0 to 51.8) (Klein, et al. 1989 [2787]). Among individuals using insulin, the risk of proliferative retinopathy increased 30 percent for each 1 percent increase in GHb (OR=1.3; 95 percent CI: 1.0 to 1.6) (Klein, et al. 1988 [2832]).
One study examined the relation between HbA1 and progression of retinopathy (Klein, et al. 1994 [2458]). In this study, the cumulative incidence of retinopathy progression increased with increasing quartiles of HbA1 in individuals using insulin and in those not using insulin. Among insulin users, the risk of retinopathy progression was twofold higher in those in the highest quartile of HbA1 (≥11.6 percent) compared to those in the lowest quartile (≤8.8 percent) (RR=2.1; 95 percent CI: 1.6 to 2.8) and there was a 40 percent increased risk for each 1 percent change in HbA1 (Klein, et al. 1994 [2458]). Among non-users of insulin, the risk of retinopathy progression was fourfold higher in those in the highest quartile of HbA1 (≥10.1 percent) compared to those in the lowest quartile (≤7.6 percent) (RR=4.3; 95 percent CI: 3.0 to 6.2) and there was an 80 percent increased risk for each 1 percent increase in HbA1 (Klein, et al. 1994 [2458]).
One study examined the relation between total GHb and retinopathy progression over 4 years of followup (Klein, et al. 1988 [2832]). The risk was increased by 30 percent for each 1 percent increase in GHb in individuals using insulin whose average GHb was 11.8 percent and was increased by 40 percent for each 1 percent increase in GHb in individuals not using insulin whose average GHb was 10.2 percent (Klein, et al. 1988 [2832]).
There were five cohort studies examining incidence of blindness and changes in visual acuity in individuals with type 1 diabetes. In one study of 3674 individuals over 10 years of followup, the incidence of blindness was 0.5 percent among normoalbuminuric individuals with a mean HbA1c of 8.1 percent, a mean age of 26.1 years, and average diabetes duration of 8.4 years, and increased to 5.8 percent among macroalbuminuric individuals with a mean HbA1c of 8.4 percent, a mean age of 33.5 years, and an average diabetes duration of 19.7 years (Muhlhauser, et al. 2000 [185]).
Two cohort studies evaluated the relation between HbA1 and incidence of blindness and included 3570 and 634 individuals over 10 and 14 years of followup, respectively (Muhlhauser, et al. 2000 [178]; Moss, et al. 1998 [2145]). Incidence of blindness ranged from 0.6 percent to 2.1 percent (HbA1c≤7 vs. HbA1c≥9) in one study with a mean age of 27.5 years and an average diabetes duration of 22 years (Muhlhauser, et al. 2000 [178]) and from 5.6 percent to 22 percent (HbA1c≤10.8 vs. HbA1c≥14.1 percent) in another study with a mean age of 26.8 years and an average diabetes duration of 12.6 years (Moss, et al. 1998 [2145]). In one of these studies, following multivariate adjustment, the risk of blindness over 14 years increased 46 percent for each 1 percent increase in HbA1 in a population with a mean HbA1 of 10.6 percent (OR=1.46; 95 percent CI: 1.28 to 1.66) (Moss, et al. 1998 [2145]).
One study examined the relation between total GHb and risk of doubling of the visual angle over 10 years and found that the cumulative incidence increased with increasing GHb. In this study, there was a 28 percent increased risk of decline in visual acuity for each 1 percent increase in GHb (OR=1.28; 95 percent CI: 1.15 to 1.42) (Moss, et al. 1994 [2482]).
One cohort study examined the relation between total GHb and the risk of doubling of visual angle in individuals with type 2 diabetes. The cumulative incidence of worsening visual acuity increased with increasing quartiles of GHb over 10 years of followup (Moss, et al. 1994 [2482]).
One cohort study examined the relation between glycated hemoglobin and the risk of cataract extraction in individuals with type 2 diabetes (Stratton, et al. 2000 [4087]). In this study, the RRR for cataract extraction was 9 percent to 19 percent for each 1 percent decrease in HbA1c, depending on multivariate adjustment, in a population with a mean age of 53 years (Stratton, et al. 2000 [4087]). There were no cohort studies identified by our search strategy examining the relation between glycated hemoglobin and the risk of cataract extraction in individuals with type 1 diabetes.
There were two clinical trials conducted in individuals with type 2 diabetes that examined the relation between glycemic control and incident retinopathy, which included 106 and102 individuals over 4 and 6 years of followup, respectively (Ohkubo, et al. 1995 [888]). In one study, the cumulative incidence of retinopathy was lower in the intensively treated group, with a mean HbA1c of 8.4 percent, than in the standard treatment groups, with a mean HbA1c of 8.8 percent (Gaede, et al. 1999 [416]). The participants in this study had a mean age of approximately 55 years and median diabetes duration of 5.5 to 6 years (Gaede, et al. 1999 [416]). In another study, the incidence of retinopathy was lower in the intensively controlled group, which maintained a HbA1c of approximately 7.1 percent (mean age 50.5 years, average diabetes duration approximately 10 years), compared to the conventional group, which maintained a HbA1c of approximately 9.4 percent (mean age 48 years, average diabetes duration approximately 6.4 years)(7.7 percent vs. 32 percent) (Ohkubo, et al. 1995 [888]).
There were 4 clinical trials evaluating the relation between intensive glycemic control and progression of retinopathy in individuals with type 1 diabetes. These studies included 96 to 1441 individuals who were followed for 4 to 7 years (Gaede, et al. 1999 [416]; Reichard, et al. 1991 [1266]; The effect of intensive diabetes treatment on the progression of diabetic retinopathy in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial 1995 [2438]; The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group1993 [4638]). The RRRs for intensive vs. conventional glycemic control ranged from 39 to 79 percent (Gaede, et al. 1999 [416]; The effect of intensive diabetes treatment on the progression of diabetic retinopathy in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial 1995 [2438]; The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group1993 [4638]). The HbA1cs for these studies ranged from approximately 7 to 8 percent in the intensive groups and from 8.7 to 9 percent in the conventional groups (Gaede, et al. 1999 [416]; The effect of intensive diabetes treatment on the progression of diabetic retinopathy in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial 1995 [2438]; The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group 1993 [4638]). In one study, there was an approximately twofold increased risk of progression of retinopathy for a 1 percent increase in HbA1c (Reichard, et al. 1991 [1266]).
Three clinical trials evaluated the relation between tight glycemic control and progression of retinopathy in type 2 diabetes. These studies included 110 to 160 individuals followed over 4 to 8 years of followup, respectively (Shichiri, et al. 2000 [247]; Gaede, et al. 1999 [416]; Ohkubo, et al. 1995 [888]). In one study, the RRR for intensive therapy vs. conventional therapy was 68 percent for the primary prevention cohort and 57 percent for the secondary prevention cohort (Shichiri, et al. 2000 [247]). In another study, the ROR for progression of retinopathy in intensive group vs. the standard group was 55 percent (Gaede, et al. 1999 [416]). The incidences of retinopathy progression in the conventional groups ranged from 38–56 percent compared to lower incidence rates in the intensive therapy groups of 13.4–24 percent (Shichiri, et al. 2000 [247]) (Ohkubo, et al. 1995 [888]).
There was one clinical trial of the effect of glycemic control on incident blindness in individuals with type 1 diabetes which included 89 individuals followed over 7.5 years (Reichard, et al. 1993 [1069]). The incidence rates ranged from 14 percent in the intensive therapy groups compared to 35 percent in the conventional therapy group (Reichard, et al. 1993 [1069]). Following multivariate adjustment, the risk of blindness increased twofold for each 1 percent increase in HbA1c (OR=2.2; 95 percent CI: 1.2 to 3.9) (Reichard, et al. 1993 [1069]).
There were two clinical trials evaluating the relation between HbA1c and incidence of blindness in individuals with type 2 diabetes. These studies included 160 and 4209 individuals followed for 4 and 10 years, respectively (Gaede, et al. 1999 [416]; Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]). There was a non-significant RRR for blindness in the intensive group, with a median HbA1c over followup of 7 percent, compared to the conventional therapy group, where the median HbA1c was 7.9 percent (RR=0.84; 95 percent CI: 0.51, 1.40) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group1998 [2113]). In the other study, the incidence of blindness over 4 years was 1.3 percent in the intensive group versus 9 percent in the conventional therapy group (Gaede, et al. 1999 [416]).
There was one clinical trial evaluating the relation between tight glycemic control and the risk of cataract extraction, which showed a non-significant reduction in the risk of cataract extraction in the intensively controlled group compared the intensively controlled group (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
Three studies examined the relation between HbA1 and incidence of microalbuminuria. These studies included 107 to 943 individuals followed for 3 to 4 years (Scott, et al. 2001 [25]; Izumi, et al. 1995 [871]; Klein, et al. 1991 [2691]). The incidence of microalbuminuria increased with increasing HbA1 also, from 8.4 percent to 24.8 percent over 4 years of followup in one study (HbA1≤10.8 percent vs. HbA1≥14.2 percent) (Klein, et al. 1991 [2691]). However, in another study, the incidence of microalbuminuria decreased with increasing HbA1, from 13.6 percent to 5.9 percent over 3.2 years of followup (HbA1≤9.9 percent vs. HbA1≥12.5 percent) (Izumi, et al. 1995 [871]). In two studies, the RR of microalbuminuria ranged from 0.98 to 6.3 with increasing quartiles of HbA1 (Scott, et al. 2001 [25]; Klein, et al. 1991 [2691]). In one of these studies, the mean age was 28.7 years and the average diabetes duration was 13 years (Scott, et al. 2001 [25]).
Two studies evaluated the relation between total GHb and risk of microalbuminuria (Powrie, et al. 1994 [929]; Klein, et al. 1995 [853]). In one study, individuals with total GHb above 12 were nearly 10-times as likely to develop microalbuminuria during the 9.6 years of followup than individuals below this cut-point in a population with a mean age of 31.3 years and an average diabetes duration of 16 years (Powrie, et al. 1994 [929]). In another study, the risk of microalbuminuria was over threefold greater in individuals in the highest quartile of GHb (≥12.1 percent) compared to those in the lowest quartile (≤9.4 percent)(RR=3.45; 95 percent CI: 2.30 to 5.17) (Klein, et al. 1995 [853]).
There were 8 cohort studies that evaluated the relation between glycated hemoglobin and development of microalbuminuria in individuals with type 2 diabetes. Five of these studies measured HbA1c as a measure of glycated hemoglobin exposure and included 133 to 963 individuals followed for 2 to 10 years (Villar, et al. 1999 [325]; Molyneaux, et al. 1998 [424]; Yokoyama, et al. 1998 [503]; Ravid, et al. 1998 [531]; Niskanen, et al. 1996 [749]). In one study, the RRR was estimated to be 9.9 percent for a 10 percent reduction in HbA1c among a group of individuals with a mean change in HbA1c of 9.5 percent, a mean age of 54.9 years, and an average diabetes duration of 3.8 years (Molyneaux, et al. 1998 [424]). In two studies, the incidence of microalbuminuria increased with increasing HbA1c, from 0 percent to 27 percent over 6.8 years in one study (HbA1c≤6.4 percent vs. HbA1c≥10.5 percent) and from 15 percent to 54.9 percent over 7.8 years in another study (HbA1c≤8.9 percent vs. HbA1c≥9 percent) (Yokoyama, et al. 1998 [503]; Ravid, et al. 1998 [531]). In one study, following multivariate adjustment, the risk of microalbuminuria increased by 56 to 63 percent for each 1 percent increase in HbA1c in a population with a mean HbA1c of 8.4 percent, a mean age of 27 years, and an average diabetes duration of 4.4 years (Yokoyama, et al. 1998 [503]). In another study, individuals with a HbA1c above 9.0 percent were over 8-times as likely to develop microalbuminuria during 7.8 years of followup compared to individuals with a HbA1c below 9 percent in a population with a mean age of 47.7 years and an average diabetes duration of 1.92 years (Ravid, et al. 1998 [531]).
One cohort study evaluated the relation between HbA1 and microalbuminuria. This study showed a much higher cumulative incidence rate among individuals with a HbA1>9.5 percent (58.3 percent over 5 years of followup) compared to those with a HbA1c below this cut-point (18.2 percent) (Haneda, et al. 1992 [1172]). The mean age of this population was approximately 57.2 years and the average diabetes duration was approximately 10 years.
In two cohort studies, the glycated hemoglobin exposure variable was total GHb. Among individuals using insulin and those not using insulin in both studies, there was an increase in cumulative incidence of microalbuminuria with increasing GHb quartiles (Klein, et al. 1995 [853]; Klein et al. 1993 [1103]). Among insulin users, there was a non-significant 33 percent to 43 percent increased risk of microalbuminuria in individuals in the highest quartile of GHb (≥11.6 percent) compared to those in the lowest quartile GHb (≤8.8 percent) (RR=1.43; 95 percent CI: 0.85 to 2.42) (Klein, et al. 1995 [853]) and (RR=1.33; 95 percent CI: 0.7 to 2.52) for individuals with GHB>13.5 percent vs. individuals with GH<10.1 percent (Klein et al. 1993 [1103]). Similarly, among non-insulin users, there was an 83 percent to 87 percent increased risk of microalbuminuria in individuals in the highest quartile of GHb (≥10.1 percent) compared to those in the lowest quartile (≤7.6 percent)(RR=1.87; 95 percent CI: 1.12 to 3.13) (Klein, et al. 1995 [853]) and (RR=1.83; 95 percent CI: 0.87 to 3.85) for individuals with GHB>11.7 percent vs. individuals with GHB<8.5 percent (Klein et al. 1993 [1103]).
Seven cohort studies examined the relation between glycosylated hemoglobin and risk of macroalbuminuria among individuals with type 1 diabetes. In two studies, the glycated hemoglobin exposure variable was HbA1c (Retinopathy and nephropathy in patients with type 1 diabetes four years after a trial of intensive therapy. The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Research Group 2000 [292]; The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial 1996 [718]). These studies included 1208 to 1273 individuals followed for 4 to 6.5 years. In one study, there was a significant risk reduction in macroalbuminuria in individuals with a mean HbA1 of 7.3 percent compared to those with a mean HbA1 of 9.8 percent in a population with a mean age of 27 years and an average diabetes duration of approximately 12 years (RRR=86 percent; 95 percent CI: 60 to 95 percent) (Retinopathy and nephropathy in patients with type 1 diabetes four years after a trial of intensive therapy. The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Research Group 2000 [292]). In the other study, the RRR for macroalbuminuria for a 10 percent change in HbA1c was 36 percent (95 percent CI: 26 to 45 percent) for the whole study population, as well as the conventional and intensive therapy groups (The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial 1996 [718]).
In four studies, the glycated hemoglobin exposure variable was HbA1. These studies included 148 to 340 individuals followed for 2 to 8 years (Warram, et al. 2000 [304]; Predictors of the development of microalbuminuria in patients with Type 1 diabetes mellitus: a seven-year prospective study. The Microalbuminuria Collaborative Study Group 1999 [310]; Villar, et al. 1999 [325]; Jones, et al. 1998 [482]). In one study, the incidence and RR of macroalbuminuria increased with increasing quartiles of HbA1, such that individuals in the higher quartile (HbA1≥10 percent) were over 5-times more likely to develop macroalbuminuria compared to individuals with HbA1<7.9 in a population with a mean age of 30 years and an average diabetes duration of approximately 17.5 years (RR=5.5; 95 percent CI: 1.6 to 18.8) (Warram, et al. 2000 [304]). In the other study, the risk of macroalbuminuria increased by 32 percent for each 1 percent increase in HbA1 in a population with a mean HbA1 of approximately 9.2 percent, mean age of approximately 29 years, and average diabetes duration of approximately 15 years (Predictors of the development of microalbuminuria in patients with Type 1 diabetes mellitus: a seven-year prospective study. The Microalbuminuria Collaborative Study Group 1999 [310]).
In one cohort study, glycated hemoglobin exposure was assessed by measuring total GHb. The risk of macroalbuminuria over 4 years of followup was threefold greater in individuals with GHb≥11 percent compared to individuals below this cut-point (Lloyd, et al. 1996 [777]).
Four cohort studies examined the relation between glycated hemoglobin and incident macroalbuminuria in individuals with type 2 diabetes (Villar, et al. 1999 [325]; Gall, et al. 1997 [654]; Park, et al. 1998 [534]; Klein, et al. 1996 [809]). In one study, in which HbA1c was used to measure glycated hemoglobin exposure, the adjusted RR of developing macroalbuminuria over 6 years of followup was 20 percent for each 1 percent increase in HbA1c for a population with an average HbA1c of 7.79 percent, a mean age of approximately 55 years, and an average diabetes duration of approximately 5.5 years (RR=1.2; 95 percent CI: 1.0 to 1.4) (Gall, et al. 1997 [654]).
In two studies the measure of glycated hemoglobin exposure was HbA1. In one study, the risk of developing macroalbuminuria over 5.5 years of followup was 13 percent for each 1 percent increase in HbA1 for a population with an average HbA1 of 11.25 percent, a mean age of approximately 62 years, and an average diabetes duration of approximately 14 years (RH=1.13; 95 percent CI: 1.05 to 1.21) (Park, et al. 1998 [534]). In the other study, there was no relation between HbA1 and the risk of macroalbuminuria in insulin users (mean age 65.2 years; average diabetes duration 15 years) or in non-users (mean age 68 years; average diabetes duration 8.8 years) (Klein, et al. 1996 [809]).
Only two cohort studies examined the progression of nephropathy in individuals with type 1 diabetes. In one study, the glycated hemoglobin exposure variable was HbA1c (Villar, et al. 1999 [325]). In this study, the risk of progression was twofold higher for each 1 percent increase in HbA1c in a population with a mean age of 35 years and average diabetes duration of 19 years (OR=2.08; 95 percent CI: 1.34 to 3.21) (Villar, et al. 1999 [325]). In the other study, in which the glycated hemoglobin exposure variable was HbA1, the risk of nephropathy progression increased by 65 percent for each 1 percent increase in HbA1 over 6 years of followup in a population with a mean HbA1 of 7.8 percent, a mean age of 26.4 years, and an average diabetes duration of 8.4 years (Sawicki, et al. 2000 [289]).
Three cohort studies examined the progression of nephropathy in individuals with type 2 diabetes and all studies used HbA1c as a measure of glycated hemoglobin exposure. These studies included 67 and 258 individuals followed for 2 and 10 years, respectively (Nosadini, et al. 2000 [245]; Oue, et al. 1999 [315]; Villar, et al. 1999 [325]). In one study the risk of nephropathy progression was 3-times greater in individuals with the higher HbA1c (>9.01 percent) compared to those with the lowest HbA1c (<6.7 percent) in a population with a mean age of approximately 58 years, and an average diabetes duration of approximately 12 years (Nosadini, et al. 2000 [245]). In another study, the risk of nephropathy progression increased 7-fold for each 1 percent increase in HbA1c over 10 years of followup in individuals with a mean HbA1c of approximately 8.2 percent, mean age of approximately 56 years, and an average diabetes duration of 11 years (OR=7.30; 95 percent CI: 1.29 to 41.4); however, this relation was attenuated following multivariate adjustment (OR=1.65; 95 percent CI: 1.09 to 2.50) (Oue, et al. 1999 [315]). The risk of nephropathy progression increased 77 percent for each 1 percent increase in HbA1c in a population with an average HbA1c of 7.5 percent, a mean age of 60 years, and an average diabetes duration of 14 years (Villar, et al. 1999 [325]).
Three cohort studies examined the relation between glycated hemoglobin and the risk of ESRD in individuals with type 1 diabetes. In one study, in which the glycated hemoglobin exposure variable was HbA1c, the incidence of ESRD increased from 0.5 percent among normoalbuminuric individuals with a mean HbA1c of 8.1 percent, mean age of 26.1 years, and average diabetes duration of 8.4 years to 37.2 percent in macroalbuminuric individuals with a mean HbA1c of 8.4 percent, mean age of 33.5 years, and average diabetes duration of 19.7 years during 10 years of followup (Muhlhauser, et al. 2000 [185]). In two studies, the glycated hemoglobin exposure variable was HbA1 (Muhlhauser, et al. 2000 [178]; Alaveras, et al. 1997 [681]). In one study, the incidence of ESRD increased from 2.4 percent among individuals with a HbA1<7 percent to 5.9 percent among individuals with HbA1c>9 percent in a population with a mean age of 27.5 years and an average diabetes duration of 10.5 years (Muhlhauser, et al. 2000 [178]).
Only one cohort study examined the relation between glycated hemoglobin and ESRD among individuals with type 2 diabetes and found that there was no relation between HbA1c level and development of ESRD over 8 years of followup in a population with a mean age of 57 years (Yokoyama, et al. 1997 [653]).
Two clinical trials examined the association between HbA1c and incidence of microalbuminuria in individuals randomized to intensive vs. conventional glycemic control. These studies included 89 to 1441 individuals over approximately 7 years of followup (Reichard, et al. 1993 [1069]); The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group 1993 [4638]). In one study, the RRR was 34 percent in for intensive insulin therapy vs. conventional insulin therapy in the primary prevention cohort with a mean age of 26.5 years and was 43 percent for the secondary prevention cohort with a mean age of 27 years (The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group 1993 [4638]). In another study, for the whole study population, the odds of microalbuminuria was 6-fold for each 1 percent increase in HbA1c (OR=5.9; 95 percent CI: 1.9 to 16.8) (Reichard, et al. 1993 [1069]). The population had a mean age of approximately 32 years for those with standard treatment and 30 years for those with intensive treatment and average diabetes duration of 16 years for those with standard treatment and 18 years for those with intensive treatment (Reichard, et al. 1993 [1069]).
One clinical trial examined the relation between intensive glycemic control and the incidence of macroalbuminuria in individuals with type 1 diabetes. This study included 1441 individuals followed for 6.5 years. The study reported a RRR for macroalbuminuria for the intensive vs. conventional glycemic control was 44 percent (95 percent CI: -124 to 86 percent) for the primary prevention cohort and 56 percent (95 percent CI: 18 to 76 percent) for the secondary prevention cohort (The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group 1993 [4638]).
Three clinical trials examined the association between intensive glycemic control and incidence of macroalbuminuria in individuals with type 2 diabetes. These studies included 110 to 4209 individuals followed for 6 to 10 years. The incidence rates of macroalbuminuria were lower in the intensive compared to the conventional therapy groups (Shichiri, et al. 2000 [247]; Ohkubo, et al. 1995 [888]). The RRR ranged from 44 to 100 percent (Shichiri, et al. 2000 [247]; Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]), although in one study, the event rates were very low (Shichiri, et al. 2000 [247]).
There was one clinical trial examining the association between intensive glycemic control and the risk of nephropathy progression. For the whole study population, the odds of nephropathy progression was 2-fold for each 1 percent increase in HbA1c in a population with a mean age of approximately 30 years and average diabetes duration of 17.1 years (OR=2.3; 95 percent CI: 1.3 to 3.4) (Reichard, et al. 1991 [1266]).
Similarly, there was only one clinical trial examining the association between intensive glycemic control and the risk of nephropathy progression in individuals with type 2 diabetes. This study showed a lower incidence of nephropathy progression in individuals randomized to intensive glycemic control (9.6 percent per 100 person years), where the HbA1c at the end of followup was 7.1 percent, compared to those randomized the conventional glycemic control (30 percent per 100 person years), where the HbA1c at the end of followup was 9.4 percent (Ohkubo, et al. 1995 [888]).
One study examined the relation between total glycated hemoglobin and the risk of neuropathy progression. During 3 years of followup in 407 individuals with a mean age of 31.4 years and an average diabetes duration of 6.5 years, the risk of progression to definite distal symmetric polyneuropathy was six fold greater in individuals with GHb>13.5 percent compared to those with GHb<10.1 percent (RR=6.4; 95 percent CI: 2.4 to 16.8), and the risk of progression of neuropathy was nearly threefold greater in those with GHb>13.5 percent compared to those with GHb<10.1 percent (RR=2.8; 95 percent CI: 1.6 to 4.8) (Christen, et al. 1999 [1999]). In multivariate analyses, there was a 25 percent increased risk of progression to definite distal symmetric polyneuropathy (RR=1.25; 95 percent CI: 1.12 to 1.39) and a 17 percent increased risk of progression of neuropathy (RR=1.17; 95 percent CI: 1.08 to 1.26) for each 1 percent increase in total GHb.
There were only three cohort studies that examined the relation between glycated hemoglobin and neuropathy outcomes in individuals with type 2 diabetes. These two studies, in which the glycated hemoglobin exposure variable was HbA1, included 231 and 1298 individuals over 5 and 10 years of followup, respectively (Sands, et al. 1997 [2275]; Klein, et al. 1996 [809]). In one study, there was no difference in the risk of definite distal symmetric neuropathy or development of an abnormal neurological exam based on HbA1 level (Sands, et al. 1997 [2275]). In another study, the incidence rates of sensory loss and diminished temperature sensation were similar in older onset diabetic individuals who were taking insulin (mean age 65.2 years; average diabetes duration 15 years) and in those who were not taking insulin (mean age 68 years; average diabetes duration 8.8 years) (Klein, et al. 1996 [809]).
One study examined the relation between total GHb and the risk of neuropathy. This study included 149 individuals followed for 6 years and in multivariate analyses, showed a significant positive association between GHb and neuropathy impairment score in a population with an average diabetes duration of 17.1 years (Dyck, et al. 1999 [2022]).
The two cohort studies identified by our search examining the relation between glycated hemoglobin and autonomic neuropathy have been conducted in individuals with type 1 diabetes. These two studies included 83 and 373 individuals over 10 and 5 years of followup, respectively (Makimattila, et al. 2000 [184]; Stella, et al. 2000 [224]). One study showed an inverse association between HbA1 and total power, a measure of heart rate variability, in a population with a mean HbA1 of 9.3 percent, a mean age of 22 years, and an average diabetes duration of 11 years (Makimattila, et al. 2000 [184]). In another study, there was a positive association between HbA1 and the incidence of cardiac autonomic neuropathy in a population with a mean age of 28.7 years and an average diabetes duration of 20.2 years (Stella, et al. 2000 [224]). There was a 67 percent increased risk of cardiac autonomic neuropathy in individuals with HbA1>10 percent compared to those with a HbA1 below this cut-point (RR=1.67; 95 percent CI: 1.16 to 2.41) (Stella, et al. 2000 [224]). In multivariate analyses, there was a 50 percent increased risk of cardiac autonomic neuropathy for each 1 percent increase in HbA1 (RR=1.50; 95 percent CI: 1.21 to 1.86) (Stella, et al. 2000 [224]). We found no studies examining the relation between glycated hemoglobin and autonomic neuropathy in individuals with type 2 diabetes using our search strategy.
Two clinical trials in individuals with type 2 diabetes has risk data on the relation between HbA1c and peripheral neuropathy. These studies included 160 and 4209 individuals followed for 4 and 10 years, respectively (Gaede, et al. 1999 [416]; Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]). While one study did not demonstrate a significant reduction in the odds of peripheral neuropathy progression in individuals randomized to intensive therapy (OR=0.89; 95 percent CI: 0.43 to 1.85) (Gaede, et al. 1999 [416]), the other study showed that individuals randomized to the intensive therapy group, where the median HbA1c was 7 percent and the mean age was 53.2 years, had a 30 percent reduction in the risk of neuropathy compared to individuals randomized to the conventional therapy group, where the median HbA1c was 7.9 percent and the mean age was 53.4 years (RR=0.60; 95 percent CI: 0.39 to 0.94) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
Only one clinical trial reported data on autonomic neuropathy and this study, conducted in individuals with type 2 diabetes, demonstrated a significant reduction in the risk of autonomic neuropathy progression (OR=0.31; 95 percent CI: 0.12 to 0.78) (Gaede, et al. 1999 [416]). There are no clinical trial data on autonomic neuropathy outcomes in individuals with type 1 diabetes identified by our search strategy.
There were no studies identified by our search strategy in individuals with type 1 diabetes that examined the relation between glycated hemoglobin and coronary artery disease morbidity. There were 2 cohort studies that examined the association between glycated hemoglobin and risk of CAD morbidity in individuals with type 2 diabetes. These studies included 5063 and 1059 individuals followed for 10 and 7 years, respectively (Adler, et al. 1999 [4124]; Lehto, et al. 1997 [4240]). In one study of newly diagnosed diabetics in which the mean age of the participants was 53 years, there was a 60 to 70 percent increased risk of angina among individuals with HbA1c>7.7 percent compared to their counterparts with HbA1c<6.3 percent, following multivariate adjustment, including cardiovascular risk factors such as age, blood pressure, smoking and lipids. (Adler, et al. 1999 [4124]). In another study, in which the mean age of the population was approximately 58 years and the average diabetes duration was approximately 8 years, there was a 30 percent increased risk developing non-fatal MI over 7 years of followup in individuals with HbA1c>10.7 percent compared to individuals with HbA1c below this cut-point (RH=1.3; 95 percent CI: 1.0 to 1.7) (Lehto, et al. 1997 [4240]). Following multivariate adjustment, which included cardiovascular risk factors such as age, prevalent MI, and lipids, this association was no longer significant (Lehto, et al. 1997 [4240]).
One cohort study examined the association between glycemic control and non-fatal stroke in individuals with type 2 diabetes. This study, in which the mean age of the participants was 61.2 years, showed an inconsistent association between the incidence of stroke and quartiles of HbA1c over 4 years of followup (Fu, et al. 1993 [4403]). This relation has not been examined in individuals with type 1 diabetes.
Peripheral arterial disease (PAD): Limb amputation, ulcer formation
There were 7 cohort studies that evaluated the association between non-fatal PAD and glycohemoglobin (Muhlhauser, et al. 2000 [185]; Olson, et al. 2002 [7]; Muhlhauser, et al. 2000 [178]; Lloyd, et al. 1996 [777]; Moss, et al. 1999 [2047]; Moss, et al. 1996 [2326]; Moss, et al. 1992 [2636]). In one study, in which the glycated hemoglobin exposure variable was HbA1c, there was a positive association between HbA1c and the risk of limb amputations in which the mean age of the population ranged from 26.1 to 33.5 years and the average diabetes duration ranged from 8.4 years to 19.7 years (Muhlhauser, et al. 2000 [185]). In two studies, the glycated hemoglobin exposure variable was HbA1 (Olson, et al. 2002 [7]; Muhlhauser, et al. 2000 [178]). Both demonstrated a positive association between HbA1 and the risk of PAD over 10 years of followup. In one study, there was a 53 percent increased risk of PAD for each 1.84 unit change in HbA1 in a population with an average HbA1 of approximately 10.6 percent, a mean age of approximately 28 years, and an average diabetes duration of approximately 20 years, following adjustment for diabetes duration, hypertension, log AER, and heart rate (Olson, et al. 2002 [7]). In another study, the cumulative incidence of limb amputation over 10 years of followup was 1.1 percent for individuals with HbA1<7 percent compared to 2.7 percent for individuals with HbA1>9 percent in a population with a mean age of 27.5 years and an average diabetes duration of 10.5 years (Muhlhauser, et al. 2000 [178]).
Four cohort studies examined the association between glycated hemoglobin and the risk of non-fatal PAD in individuals with type 2 diabetes (Lehto, et al. 1996 [2330]; Moss, et al. 1992 [2636]; Moss, et al. 1999 [2047]; Moss, et al. 1996 [2326]). In the one study that used HbA1 as the exposure variable, individuals with HbA1>10.8 percent were over 2-times as likely to have a limb amputation over 7 years of followup compared to individuals with a HbA1 below that cut-point (Lehto, et al. 1996 [2330]).
We found two studies examining the association between glycated hemoglobin the risk of congestive heart failure (CHF) in individuals with type 2 diabetes. While one study, in which the mean age was 53 years, showed a significant RRR in the incidence of CHF for each 1 percent decline in HbA1c during 10 years of followup following adjustment for gender, race, blood pressure, smoking, age at diabetes diagnosis, lipids, and albuminuria (Stratton, et al. 2000 [4087]), the other study, in which the mean age was 63 years and the average diabetes duration was 4.5 years, demonstrated no relation between HbA1c and risk of CHF over 2.5 years of followup in population with a mean HbA1c of 7.7 percent (Nichols, et al. 2001 [4025]). There are no cohort study data identified by our search strategy on the relation between glycated hemoglobin and the risk of CHF among individuals with type 1 diabetes.
Carotid intimal-medial thickness (IMT)
Fatal Coronary Artery Disease (CAD): Fatal MI
There was only one cohort study that examined the risk of CAD mortality in relation to glycated hemoglobin in individuals with type 1 diabetes (Lehto, et al. 1999 [397]). In this study in which the mean age of the population was approximately 56 years and the average diabetes duration ranged from 13 to 22 years, individuals with an HbA1≥10.4 percent were over 6-times as likely to suffer a fatal CAD event compared to individuals with an HbA1 below this cut-point (RH=6.8; 95 percent CI: 2.5 to 19) (Lehto, et al. 1999 [397]). This relation persisted following multivariate adjustment for cardiovascular risk factors, including age, prevalent MI, weight, blood pressure, smoking, alcohol, and HDL (RH=5.4; 95 percent CI: 1.4 to 20.4) (Lehto, et al. 1999 [397]).
In one study, the glycated hemoglobin exposure variable was total GHb (Florkowski, et al. 1998 [519]). In this study, which included 6 years of followup in a population with a mean age of 62 years and an average diabetes duration of approximately 10 year, the risk of fatal CAD was increased by 90 percent in individuals free of CAD at baseline following multivariate adjustment for multiple cardiovascular risk factors including smoking, lipids, hypertension, CAD, and albuminuria (RH=1.90; 95 percent CI: 1.04 to 3.47); however, in this same study, there was not a significant relation between total GHb and fatal CAD in individuals who had a prior history of CAD (RH=0.81; 95 percent CI: 0.42 to 1.58) (Florkowski, et al. 1998 [519]).
One study in individuals with type 2 diabetes examined the relation between total GHb and the risk of fatal stroke during 10 years of followup in a population with a mean age of 66.6 years and an average diabetes duration of 11.6 year and showed a 17 percent increased risk of stroke for each 1 percent increase in total GHb (RH=1.17; 95 percent CI: 1.05 to 1.30) (Moss, et al. 1994 [4338]).
Fatal MI or Fatal stroke
Coronary artery disease: Fatal and non-fatal MI
Two cohort studies examined the association between glycated hemoglobin and fatal and non-fatal CAD in individuals with type 1 diabetes. One study, in which the mean age was approximately 56 years and the average diabetes duration ranged from approximately 13 to 22 years, found a nearly 3-fold increase in risk of fatal and non-fatal MI in individuals with a HbA1≥10.4 percent compared to individuals below this cut-point (RH=3.4; 95 percent CI: 1.6 to 7.4), which persisted following multivariate adjustment for age, gender, diabetes duration, baseline CVD, weight, blood pressure, smoking, alcohol, and HDL (RH=2.8; 95 percent CI: 1.2 to 6.9) (Lehto, et al. 1999 [397]). In contrast, the other study found no significant relation between total GHb and risk of fatal and non-fatal CAD (Lloyd, et al. 1996 [777]).
Three cohort studies have examined the relation between HbA1c and risk of fatal and non-fatal MI in individuals with type 2 diabetes. These studies included 2693 to 5063 individuals followed for 10 years (Stratton, et al. 2000 [4087]; Adler, et al. 1999 [4124]; Turner, et al. 1998 [4217]). In one study, in which the mean age of the study population was 53 years, there was a 14 percent risk reduction in fatal or non-fatal MI for each 1 percent reduction in HbA1c, following multivariate adjustment for gender, race, blood pressure, smoking, age at diagnosis, lipids, and albuminuria (RRR=14; 95 percent CI: 8 to 21) (Stratton, et al. 2000 [4087]). In two other studies, there was an approximately 50 percent increased risk of fatal and non-fatal MI in individuals with a HbA1c>7.5–7.7 percent compared to those with HbA1c<6.1–6.3 percent following multivariate adjustment for multiple cardiovascular risk factors (Adler, et al. 1999 [4124]; Turner, et al. 1998 [4217]).
Composite cerebrovascular disease: Fatal and non-fatal stroke
One cohort study has examined the association between total GHb and the risk of fatal and non-fatal stroke in individuals with type 1 diabetes. This study showed an 18 percent increased risk of stroke over 10 years for each 1 percent increase in total GHb in a population with a mean total GHb of 12.6 percent, a mean age of 29.1 years, and an average diabetes duration of 14.6 years (Moss, et al. 1994 [4338]).
Three cohort studies have examined the association between glycated hemoglobin and fatal and non-fatal stroke. In two of these studies, which included 3642 and 5063 individuals followed for 10 years, the glycated hemoglobin exposure variable was HbA1c (Stratton, et al. 2000 [4087]; Adler, et al. 1999 [4124]). In one study, there was a 12 percent increased risk of stroke for each 1 percent increase in HbA1c in a population with a mean HbA1c of 7.1 percent and a mean age of 53 years, following adjustment for gender, race, blood pressure, smoking, age at diagnosis, lipids, and albuminuria (Stratton, et al. 2000 [4087]); however, the other study, in which the mean age of the participants was also 53 years, found no relation between HbA1c and the risk of stroke following multivariate analyses (Adler, et al. 1999 [4124]). In the third study, in which the glycated hemoglobin exposure variable was HbA1, there was a 60 percent increased risk of stroke in individuals with a HbA1>10.8 percent compared to those with a HbA1 below this cut-point in individuals with a mean age of approximately 59 years and an average diabetes duration of approximately 8 years (OR=1.6; 95 percent CI: 1.1 to 2.4) (Lehto, et al. 1996 [4313]).
Compared to individuals randomized to conventional glycemic control, where the median HbA1c over 10 years of followup was 7.9 percent, those randomized to intensive glycemic control, where the median HbA1c was 7.0 percent, had non-significant risk reductions in non-fatal MI (RR=0.79; 95 percent CI: 0.58 to 1.09), fatal MI (RR=0.94; 95 percent CI: 0.68 to 1.30), and sudden cardiac death (RR=0.54; 95 percent CI: 0.24 to 1.21) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
There was no relation between randomization to intensive vs. conventional glycemic control and the risk of non-fatal stroke (RR=1.07; 95 percent CI: 0.68 to 1.69) and fatal stroke (RR=1.17; 95 percent CI: 0.54 to 2.54) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
There was a non-significant reduction in the incidence of limb amputation in the intensive therapy group compared to the conventional therapy group (RR=0.61; 95 percent CI=0.28 to 1.33) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
Similarly, there was a non-significant reduction in the incidence of CHF in the intensive therapy group compared to the conventional therapy group (RR=0.91; 95 percent CI: 0.54 to 1.52) (Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group 1998 [2113]).
Of 4383 potentially relevant abstracts and articles identified through our search, articles were eliminated for the following reasons: 1) no information relevant to the questions asked (n=3613), 2) under 100 persons enrolled (n=276), 3) no prospective data included (n=183), 4) no original data (i.e. editorials, reviews) (n=117), 5) albumin exposure not defined adequately or studies not reporting outcomes relative to the presence or absence of study participants at baseline (n=75), 6) no study participants with diabetes or study did not analyze outcomes for study participants with diabetes (n=33), 7) no populations of interest studied (i.e. pregnant women, transplant studies, persons already on dialysis, persons with pre-existing CVD, persons with pre-existing amputations) (n=31), 8) data not extractable (i.e. presented only in graphical form) (n=9), 9) no human data (i.e. in vitro studies, animal studies) (n=7), and 10) less than 3 months total followup (n=3). To help ensure that important eligible articles were not overlooked, we scanned bibliographies and relevant reviews. This process did not uncover additional references not already identified using the above strategy.
Of the remaining 36 articles, 4 reported on randomized controlled trials (RCTs), 31 reported on prospective cohort studies, and one reported on a post-hoc cohort analysis of a study initially designed as a RCT (from this point on, referred to as “RCT analyzed as a cohort study.” Most articles reported on studies including persons with only type 2 diabetes (n=25), while 8 articles reported on studies which included persons with only type 2 diabetes, 2 articles reported on studies including persons with both type 1 and type 2 diabetes, and 1 article did not specify study participants' type of diabetes. Reported study locations were varied, with the majority (21) of studies performed in Europe and most studies (23) being performed in a single city or country.
Of three methods of reported urine collection, 15 studies articles collection of random urine specimens, 12 reported collection of 24 hour urine specimens, and 9 reported collection of timed or ‘overnight’ specimens (referred to as ‘timed’ in tables and report). Articles also varied in their reporting of the types of measurement used in assessing microalbuminuria; 30 studies reported albumin excretion rates, 5 reported albumin to creatinine ratios, and one reported dipstick positive results.
Quantification and definitions of microalbuminuria varied among reported studies. In studies reporting albumin excretion rates from 24-hour urine specimens (n=12), the lower and upper values for defining microalbuminuria ranged from 10.1–50.5 mg/24hrs and 35 to 300mg/24hrs, respectively. (Stehouwer, et al. 2002 [6834]; Niskanen, et al. 1993 [1452]; Lebovitz, et al. 1994 [1366]; Gall, et al. 1995 [1146]; Deckert, et al. 1996 [1074]; Rossing, et al. 1996 [991]; Araki, et al. 1997 [930]; Gaede, et al. 1999 [613]; Nosadini, et al. 2000 [376]; Rachmani, et al. 2000 [336]; Muhlhauser, et al. 2000 [293]; Muhlhauser, et al. 2000 [283]) One study of this group gave no upper limit in defining microalbuminuria. (Niskanen, et al. 1993 [1452]) In studies reporting albumin excretion obtained from timed urine collections (n=9), the lower and upper values for defining microalbuminuria ranged from 15.0 to 30 μg/min and 149.9 to 499 μg/min, respectively. (Damsgaard, et al. 1992 [1622]; Mattock, et al. 1992 [1603]; Damsgaard, et al. 1993 [1429]; MacLeod, et al. 1995 [1218]; Crepaldi, et al. 1995 [1158]; Captopril reduces the risk of nephropathy in IDDM patients with microalbuminuria. The Microalbuminuria Captopril Study Group1996 [1061]; Mattock, et al. 1998 [659]; Hanninen, et al. 1999 [580]) One study reporting albumin excretion obtained from timed urine collection reported the lower value for microalbuminuria as 30mg/24 hrs; this study gave no upper value for defining microalbuminuria. (Niskanen, et al. 1996 [1067]) In studies reporting albumin excretion obtained from random urine specimens (n= 9), the lower and upper values for defining microalbuminuria ranged from 20 to 150 mg/l and 200 to 300mg/l, respectively. (Torffvit, et al. 1993 [1508]; Neil, et al. 1993 [1454]; Agardh, et al. 1996 [1075]; Beilin, et al. 1996 [1009]; Miettinen, et al. 1996 [977]; Valmadrid, et al. 2000 [411]; Biderman, et al. 2000 [392]; de Grauw, et al. 2001 [216]) One study of this group gave no upper value for defining microalbuminuria. (Biderman, et al. 2000 [392])
Among studies reporting microalbuminuria as albumin to creatinine ratio (ACR) (n=5), ACR was quantified in different ways; one study quantified ACR in grams albumin per gram creatinine (g/g), three studies quantified ACR in milligrams albumin per millimol creatinine (mg/mmol), and one quantified ACR in grams albumin per mol creatinine (g/mol). (Hoy, et al. 2001 [160]; Mlacak, et al. 1999 [465]; Nelson, et al. 1996 [967]; Chan, et al. 1995 [1195]; Gerstein, et al. 2001 [136]) (See Acronyms for abbreviations).
Among cohort studies, the following renal outcomes (number of articles reporting outcome) were reported: 1) change in GFR at study followup (n=1), 2) rate of change of GFR during study (n=2), 3) rate of change in 1/serum creatinine during study (n=1), 4) change in creatinine clearance at study followup (n=1), and 5) incidence of ESRD, need for renal replacement therapy, or kidney transplantation (n=3).
Change in GFR at Study Followup
Type 2 diabetes
Rate of Change in GFR During Study
Type 2 diabetes
Rate of Change in 1/Serum Creatinine During Study
Type 2 diabetes
Rate of Change in Creatinine Clearance During Study
Type 1 diabetes
Incidence of ESRD, Need for Renal Replacement Therapy, or Kidney Transplantation
Type 1 diabetes
Articles reporting on two prospective cohort studies demonstrate a greater risk of of ESRD, need for renal replacement therapy, or kidney transplantation among persons with greater levels of baseline UAE.
Among articles reporting on RCTs, the following renal outcomes (number of studies reporting outcome) were reported: 1) change in GFR at study followup (n=1), 2) rate of change of GFR during study (n=1), 3) change in creatinine clearance at study followup (n=1), and 4) rate of change in creatinine clearance during study.
Change in GFR at Study Followup
Type 2 diabetes
Rate of Change in GFR During Study
Type 2 diabetes
Change in Creatinine Clearance at Study Followup
Type 2 diabetes
Rate of Change in Creatinine Clearance During Study
Type 1 diabetes
Among all articles reporting on cohort studies in which total mortality was assessed (n=22), four reported on persons with type 1 diabetes, 16 reported on persons with type 2 diabetes, one reported on persons with both types 1 and 2 diabetes, and one article did not specify the type of diabetes.
Type 1 diabetes
Type 2 diabetes
Types 1 and 2 Diabetes, or diabetes type not specified
Types 1 and 2 diabetes
Type 2 diabetes
Type 1 diabetes
Type 2 diabetes
Type 2 diabetes
Type 1 diabetes
Type 2 diabetes
Type 1 diabetes
Type 2 diabetes
Type 1 diabetes
Type 2 diabetes
Types 1 and 2 diabetes
Type 2 diabetes
Type 2 diabetes
Types 1 and 2 diabetes
Among all evidence reporting on the relation between glycemic control and cardiovascular, neurologic, opthamologic, and renal outcomes, there is a strong, graded relation between glycated hemoglobin exposure and the risk of two major microvascular complications of type 1 and type 2 diabetes, retinopathy and nephropathy. These patterns are observed for various measures of glycated hemoglobin (i.e., HbA1c, HbA1, and total GHb). The relation between glycated hemoglobin and macrovascular complications is not as strong as that seen for the microvascular complications, although there are fewer studies examining these outcomes. In general, however, there appears to be a positive relation between glycated hemoglobin and the risk of CAD and PAD, particularly among individuals with type 2 diabetes.
Similarly, among all evidence reporting on the relation between microalbuminuria and cardiovascular and renal outcomes, there appears to be a strong relation between baseline urinary albumin excretion and progression of kidney disease, development of cardiovascular morbidity and mortality, and incidence of all cause death. In addition, there appears to be a graded influence of the degree of urine albumin excretion at baseline and the magnitude of future risk of renal and cardiovascular outcomes.
The preponderance of the evidence from cohort studies shows a strong relation between glycated hemoglobin and incident retinopathy, incident proliferative retinopathy and macular edema, and progression of retinopathy. This is confirmed in several randomized clinical trials of individuals with type 1 and type 2 diabetes, which show comparable risk reductions in these outcomes in individuals randomized to intensive therapy, where the HbA1c's were maintained at approximately 7 percent, compared to individuals randomized to conventional therapy, where the mean HbA1c's were maintained at approximately 9 percent. Only a few cohort studies address the relation between glycated hemoglobin and the risk of blindness; however, the majority suggest that increased glycated hemoglobin is a risk factor for blindness in individuals with type 1 diabetes. With the exception of one clinical trial, there are virtually no data on the relation between glycated hemoglobin and risk of blindness in individuals with type 2 diabetes.
The majority of studies evaluating the relation between glycated hemoglobin and the risk of nephropathy have evaluated the risk of developing microalbuminuria. These data show a strong and significant relation between glycated hemoglobin and the risk of microalbuminuria in individuals with type 1 and type 2 diabetes. This is supported by clinical trial data that show significant risk reductions for incident microalbuminuria for individuals randomized to intensive glycemic control, where the mean HbA1c's were maintained at approximately 7 percent, compared to those randomized the conventional glycemic control, where the mean HbA1c's were maintained at approximately 9 percent. While there are less data on the relation between glycated hemoglobin and risk of macroalbuminuria and on the risk of nephropathy progression, several cohort studies and clinical trials support a strong and significant positive association in type 1 and type 2 diabetes. Based on our search strategy and exclusion criteria, the only studies examining the effect of glycated hemoglobin exposure on GFR were cohort studies conducted in individuals with type 1 diabetes. All studies consistently demonstrated that increasing levels of glycated hemoglobin were association with a decline in GFR. There are no clinical trial data examining the GFR outcomes in individuals with type 1 or type 2 diabetes and there are no data on the relation between glycated hemoglobin and GFR in individuals with type 2 diabetes. There are very few studies examining the association between glycated hemoglobin and risk of ESRD.
The preponderance of the evidence examining the risk relationship between glycated hemoglobin and neuropathy has shown a strong, positive association between glycated hemoglobin and the risk of peripheral neuropathy in individuals with type 1 diabetes. This has been demonstrated in several cohort studies and confirmed in clinical trials showing a significant risk reduction in the development of peripheral neuropathy in individuals randomized to intensive vs. conventional glycemic control. The association is less clear in individuals with type 2 diabetes because there have only been a few studies in this population and they have yielded conflicting results, with some studies showing a positive association and others showing no association.
There are little data on the relation between glycated hemoglobin and the risk of autonomic neuropathy. The few studies in individuals with type 1 diabetes have all shown a positive association; however, there was only one study addressing this relationship in individuals with type 2 diabetes.
In the cohort studies evaluating cardiovascular outcomes in individuals with diabetes, there was a positive association with glycated hemoglobin exposure; however, the risk estimates are much smaller compared to the risk estimates for the microvascular complications. The preponderance of the evidence from cohort studies shows a positive association between glycated hemoglobin and risk of fatal and non-fatal CAD, particularly among individuals with type 2 diabetes. There are little data on the relation between CAD and glycated hemoglobin among individuals with type 1 diabetes; however most studies have shown a positive association. The relation between glycated hemoglobin and the risk of PAD appears to be strong and positive in individuals with type 1 and type 2 diabetes. The risk relationship between cerebrovascular disease and glycated hemoglobin, which has only been examined among individuals with type 2 diabetes, is less clear. In one clinical trial, there was a non-significant reduction in the risk of cardiovascular disease in individuals with type 2 diabetes. There are very little data on the relation between glycated hemoglobin exposure and CHF or subclinical atherosclerosis, assessed by carotid IMT, making it difficult to draw any conclusions regarding these outcomes.
Only a few studies have specifically examined the presence of a threshold effect of glycated hemoglobin on the risk of developing diabetic complications. The majority of these have not found a threshold effect for retinopathy and nephropathy outcomes. There are very little data examining the presence of a threshold effect of glycated hemoglobin on neuropathy and macrovascular outcomes.
Our literature review has several limitations. First, the majority of the studies did not use glycated hemoglobin assay methods that were DCCT traceable, indicating that the majority of these measurements were not standardized. Second, there was great heterogeneity in the glycated hemoglobin measurement techniques used, especially for HbA1 and total GHb measurements, making comparisons across studies difficult. Third, our review is subject to publication bias, in that the search was more likely to identify articles that reported a positive association between glycated hemoglobin and outcomes. Fourth, certain outcomes, such as risk of blindness, progression of nephropathy, change in GFR, and certain macrovascular outcomes, were only reported in a few studies, making it difficult to draw definitive conclusions regarding the risk relationships. Finally, our data are limited to examine the risk relation between glycated hemoglobin and microvascular and macrovascular complications below HbA1c of 7 percent because there were very few studies that had exposure data below this cut-point. Therefore, we are still unable, in this report, to address whether there is added benefit to lowering the HbA1c level below the current threshold of 7 percent. Finally, our report focuses on the prognostic value of glycated hemoglobin in predicting clinical outcomes but it does not address the effects of using this test in actual clinical management of patients with diabetes."
The preponderance of the evidence from both cohort studies and RCTs indicates a strong relation between the presence of microalbuminuria at baseline and renal outcomes among persons with types 1 and 2 diabetes. In addition, there appears to be a graded relation between the total quantity of urine albumin at baseline and progression of renal disease such that persons with macroalbuminuria have faster rates of progression of chronic kidney disease and a greater decrement in kidney function over follow up when compared to persons with lesser degrees of urinary albumin excretion. When employed in RCTs, ACE inhibitor therapy appears to slow rates of progression toward ESRD.
It remains unclear whether currently accepted thresholds for microalbuminuria are ideal in terms of assisting with prediction of risk for renal outcomes. Of all articles assessing the relation between urine albumin and renal outcomes, none reported results pertaining to intermediate cut-off thresholds of urinary albumin (i.e. detectable levels of urine albumin that are too low to qualify as microalbuminuria by current standards or by standards defined by study investigators). Thus, we were unable to assess whether assignment of other thresholds cut-offs could be helpful in predicting increased risk of kidney disease.
The preponderance of the evidence from both cohort studies and RCTs demonstrates an independent association between microalbuminuria at baseline and all-cause mortality for persons with types 1 and 2 diabetes. In some cases, greater than two to three fold increases in risk of all-cause mortality were identified for persons with microalbuminuria at baseline when compared to their counterparts with normoalbuminuria after adjustment for potential confounding factors. In addition, there appears to be a graded relation between the total quantity of urine albumin at baseline and total mortality such that greater levels of urine albumin at follow up are related to greater risk of all-cause mortality.
It remains unclear whether currently accepted thresholds for microalbuminuria are ideal in terms of assisting with prediction of risk of all-cause mortality. However, in five studies analyzing intermediate levels of urine albumin excretion, there appeared to be a graded influence on the degree of urine albumin present at baseline and the incidence of all-cause mortality at follow up. These data suggest that prediction of increased risk of future risk of death may be possible at levels of urinary albumin excretion that are lower than thresholds currently used to define microalbuminuria.
The preponderance of the evidence from cohort studies and RCTs demonstrates an independent association between microalbuminuria at baseline and CVD events at follow up for persons with types 1 and 2 diabetes. Both disease specific (myocardial infarction and cerebrovascular accidents) and composite CVD death outcomes were increased among persons with microalbuminuria when compared to their counterparts with normoalbuminuria. Persons with microalbuminuria also had increased risk of composite measures of CVD mortality and morbidity in most studies. In some cases, greater than two to three fold increases in risk of CVD morbidity outcomes were identified for persons with microalbuminuria at baseline when compared to their counterparts with normoalbuminuria after adjustment for potential confounders. Similarly, in some studies, two to seven-fold increases in risk of CVD mortality was reported for persons with microalbuminuria at baseline versus their counterparts with normoalbuminuria at after adjustment for potential confounders. In addition, there appears to be a graded relation between levels of urinary albumin excretion at baseline such that greater levels of baseline urinary albumin excretion are related to greater risk of all cardiovascular outcomes.
It remains unclear whether currently accepted thresholds for microalbuminuria are ideal in terms of assisting with prediction of risk of cardiovascular morbidity and mortality. However, in four studies analyzing intermediate levels of urine albumin excretion, there appears to be a graded influence on the degree of urine albumin present at baseline and the incidence of cardiovascular morbidity and mortality at follow up. These data suggest that prediction of increased risk of future risk of death may be possible at levels of urinary albumin excretion that are lower than thresholds currently used to define microalbuminuria.
Future cohort studies and clinical trials should focus on studying the relation between glycated hemoglobin exposure and the risk of macrovascular complications, particularly among individuals with type 1 diabetes. In addition, more studies are needed that examine the relation between glycated hemoglobin and the risks of stroke and CHF, as there are little data on these outcomes in individuals with type 1 or type 2 diabetes. There are much fewer data on these outcomes than there are on microvascular outcomes. This has important implications for the role of glycemic control in the prevention of the cardiovascular sequelae in individuals with type 1 and type 2 diabetes. In addition, more studies are needed to evaluate the presence of a threshold effect of glycated hemoglobin on the risk of macrovascular complications.
Among the microvascular complications, there are fewer data on the risk relationship between glycated hemoglobin and peripheral neuropathy in individuals with type 2 diabetes compared to those with type 1 diabetes and there are very little data on the relationship of glycated hemoglobin to autonomic neuropathy in either type 1 or type 2 diabetes. More studies are also needed to determine whether there is a threshold effect of glycated hemoglobin on the risk of neuropathy outcomes.
The majority of the studies reviewed for this report were conducted in homogeneous populations that did not report the race of the cohort. Because many of these studies were conducted in Europe, we assume the race to be Caucasian. Future studies should examine the risk relation between glycated hemoglobin and risk of complications in ethnic minority populations (i.e. African-Americans, Hispanic-Americans) who have a disproportionately high incidence of type 2 diabetes. Knowing which complications are more common in which ethnic groups and the role of glycemic control in the development of these complications can help to target areas of interventions for these groups.
There is great heterogeneity in the literature on the reporting of glycated hemoglobin (i.e. HbA1c, HbA1, total GHb) in relation to microvascular and macrovascular complications, and in the biochemical measurement methods used. Future cohort studies and clinical trials should aim to use NGSP certified methods of measuring glycated hemoglobin and report results as percent HbA1c or percent HbA1c equivalents to allow risk comparisons across studies.
Finally, this report focused on the risk relation between glycated hemoglobin and the risk of microvascular and macrovascular outcomes among individuals with diabetes; however, there is emerging evidence suggesting that more subtle forms of glucose dysregulation that predate the onset of diabetes may contribute to vascular disease (de Vegt, et al. 1999 [92182]; Park, et al. 1996 [92183]; Khaw, et al. 2001 [92184]; Singer, et al. 1992 [92185]; Vitelli, et al. 1997 [92186]). Future studies are needed to examine the relation between glycated hemoglobin and risk of complications, particularly macrovascular complications, in individuals without diabetes.
There is great need for future research which seeks to define the optimal and most feasible tests for standardized measurement of microalbuminuria. Our review revealed that published definitions of microalbuminuria (as determined by reported lower and upper limits for this designation), were similar but quite varied. Consistent adherence to accepted guidelines for the definition of microalbuminuria will help with ascertaining more precisely the magnitude of risk of renal and cardiovascular risk afforded by microalbuminuria.
Similarly, there is great need for studies which assess the relation between urinary albumin excretion and progression of renal disease in a standard fashion. Particularly important is the standardized ascertainment of renal disease progression, which we found to be summarized in several manners. Use of standardized definitions for progression of renal disease will help with the ascertainment of the magnitude of risk of renal disease progression afforded by the presence of microalbuminuria at baseline.
While our review identifies what appears to be a graded relation between urinary albumin excretion at baseline and renal, cardiovascular outcomes, and death at follow up, it is not known whether levels of detectable urinary albumin excretion which are lower than currently accepted definitions for microalbuminuria might improve prediction of these outcomes. Future research is needed to characterize the nature of the relation between microalbuminuria and these outcomes and to assess whether lower levels of detectable urinary albumin excretion should be used to help predict progression toward important clinical outcomes in persons with types 1 and 2 diabetes. Such research should seek to identify whether a threshold level of urinary albumin excretion exists at which detection of urinary albumin might yield optimal prediction of renal and cardiovascular outcomes, or whether the relation between urinary albumin excretion and development of outcomes is purely linear, indicating that use of more sensitive tests for microalbuminuria early in disease might provide more optimal prediction of future outcomes. Further, it is not known whether current testing is accurate enough to reliably detect low levels of urinary albumin excretion which are associated with poor outcomes at follow up. This is particularly important now that many experts favor spot urine collections for more reliable ascertainment of urinary albumin excretion versus timed urine collections, which are often deemed difficult to collect from patients in a standardized manner. (National Kidney Foundation 2002 [3]), (American Diabetes Association 2002 [2])) Future work is needed to understand which testing strategies most reliably and efficiently predict the risk of future outcomes early in the disease process. Research to determine the optimal test for microalbuminuria will need to take into account how the results might be used to guide clinical decisions and the effects of specific interventions to prevent outcomes associated with microalbuminuria.
Finally, while this review focuses on albuminuria among persons with known diabetes, very little is known about the prognostic significance of albuminuria in persons with lower, albeit still abnormally high, levels of serum glucose (e.g. persons with glucose intolerance). Future work is needed to determine whether urinary albumin excretion in persons with lower levels of serum glucose remains an independent predictor of renal, cardiovascular and mortality outcomes. Improvement in all of the aforementioned areas will likely have important implications for the future development of guidelines for screening practices among persons with types 1 and 2 diabetes.
| ~ | Data not provided |
| * | Level representing microalbuminuria |
| ACR | Albumin to creatinine ratio |
| CAD | Coronary artery disease |
| CHF | Congestive heart failure |
| DCCT | Diabetes Control and Complications Trial |
| DM | Diabetes mellitus |
| ETDRS | Early Treatment of Diabetic Retinopathy Scale |
| ESRD | End-stage renal disease |
| F/U | Follow up |
| GFR | Glomerular filtration rate |
| GHB | Total glycosylated hemoglobin |
| IMT | Intimal-medial thickness |
| IR | Incidence rate |
| n | Number of participants enrolled in for a given level of urinary albumin excretion |
| NGSP | National Glycohemoglobin Standardization Program |
| OR | Odds ratio |
| PAD | Peripheral arterial disease |
| RCT | Randomized controlled trial |
| RH | Relative hazard |
| ROR | Relative odds reduction |
| RR | Relative risk |
| RRR | Relative risk reduction |
| SD | Standard deviation |
| SEM | Standard error of the mean |
| SMBG | Self monitoring blood glucose |
| UAC | Urine albumin concentration |
| UAE | Urinary albumin excretion |
Abbreviations for urine albumin excretion (UAE)
Albumin excretion rate (AER)
mg/min=milligrams albumin excreted per minute
mg/24hrs=milligrams urinary albumin excreted per 24 hours
Urine albumin concentration (UAC)
mg/ml=milligrams albumin per milliliter urine
mg/l= milligrams albumin per liter urine
Albumin to creatinine ratio (ACR)
mg/g=milligrams urinary albumin excreted per gram urinary creatinine excreted
mg/mmol=milligrams urinary albumin excreted per millimol urinary creatinine excreted
mg/mol=milligrams urinary albumin excreted per mol urinary creatinine excreted
g/mol=grams urinary albumin excreted per mol urinary creatinine excreted
At-large reviewers with content expertise
Thomas Kickler, MD*
Professor of Oncology, Pathology, and Medicine
Director, Hematology Lab, Pathology
Randie R. Little, Ph.D., Network Coordinator
National Glycohemoglobin Standardization Program (NGSP)
Katherine R. Tuttle, MD
Clinical Associate Professor of Medicine
University of Washington School of Medicine
The Heart Institute of Spokane
NIDDK Organizations
NIDDK—National Kidney Disease Education Program
Thomas H. Hostetter, MD
Director, National Kidney Disease Education Program
Senior Scientific Advisor
National Institutes of Health
NIDDK—National Diabetes Education Program
Saul N. Malozowski
Senior Advisor for Clinical Trials and Diabetes Translation
Division of Diabetes, Endocrinology, and Metabolic Diseases
Diabetes Organizations
American Association of Diabetes Educators
Gary Arsham, MD, PhD
Arsham Consultants, Inc.
Pathology/Laboratory Organizations
American Association of Clinical Chemistry
William E. Winter, MD
Professor of Pathology and Pediatrics
University of Florida
Organizations representing minority interests
Indian Health Service
Charlton Wilson, MD
IHS National Diabetes Program
Private third-party payer or public payer representative
Blue Cross Blue Shield (BC/BS) Association
Timothy Ranney, MD
Blue Cross & Blue Shield of Nebraska
Veterans Administration Health Care System
Leonard M. Pogach, MD
National Diabetes Director
New Jersey Veterans Administration Health Care System
Glycohemoglobin
Sherita Hill Golden, MD, MHS
Spyridon S. Marinopoulos, MD, MBA
Gail Berkenblit, MD, PhD
Microalbuminuria
Ebony Boulware, MD, MPH
Geetangali Chander, MD, MPH
Michael Paasche-Orlow, MD
Feedback Form for Question Refinement
Evidence Report on the Use of Glycohemoglobin and Microalbuminuria in Diagnosis and Monitoring of Diabetes Mellitus Johns Hopkins Evidence-based Practice Center
Name: <<NAME>>
Below is a non-prioritized list of research questions to be considered for inclusion in the evidence report. For each question, please circle your response for the following:
Rank the question to indicate what you feel is the priority/clinical importance of the question.
Indicate your sense of the ability to answer the question, yes (there is appropriate evidence available to address this question) or no (there is no appropriate evidence to address this question)
Indicate the clarity of the wording of the questions, yes (question clear and complete) or no (question is unclear or incomplete). If no, please indicate in space provided the problem with the wording and/or suggestion for alternative wording.
Please provide comments on each question.
Please fax back completed form no later than 5 p.m. Friday January 4, 2002
(Attention: Dr. Sherita Golden).
| Glycohemoglobin Questions | Ranking of question | Answerable? | Clear wording? | ||||||
|---|---|---|---|---|---|---|---|---|---|
| What is the relationship between HbA1c and microvascular diabetic complications (retinopathy, nephropathy, neuropathy)? | 1 | 2 | 3 | 4 | 5 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | |||||||||
| What is the relationship between HbA1c and macrovascular diabetic complications (CVD, peripheral vascular disease)? | 1 | 2 | 3 | 4 | 5 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | |||||||||
| What is the shape of the relationship between HbA1c and diabetic complications in various ranges of HbA1c, particularly at levels of HbA1c below 7%? | 1 | 2 | 3 | 4 | 5 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | |||||||||
| What is the effect of diabetic treatment on the relationship between HbA1c and microvascular / macrovascular complications? | 1 | 2 | 3 | 4 | 5 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | |||||||||
| Do differences in the methodologies of measuring glycohemoglobin warrant standardization of methodology? | 1 | 2 | 3 | 4 | 5 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | |||||||||
| Microalbuminuria Questions | Ranking of question | Answerable? | Clear wording? | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Do various tests for measuring microalbuminuria (i.e. dipstick testing for microalbuminuria and albumin:creatinine ratio (both random and timed) correlate? | 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
Are the various tests for microalbuminuria predictive of
| 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
What is the risk relationship between microalbuminuria and
| 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
| Do the risk relationships between proteinuria and clinical complications allow selection of an optimal cut point for defining microalbuminuria? | 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
How does treatment of microalbuminuria affect the relationship between microalbuminuria and the risk of
| 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
| How often should microalbuminuria be monitored to confirm a diagnosis? | 1 | 2 | 3 | 4 | 5 | 6 | Y | N | Y | N |
| Comments. For example, if question is not clear, please describe problem or suggest alternative wording. | ||||||||||
Are there other important, answerable questions we have not included that should be addressed in the evidence report? Please write question(s) below.
Thank you for your guidance in the essential task of refining the questions for the evidence report.
Please contact Tejal Rami or Dr. Sherita Golden.
Please fax back completed form no later than 5 p.m. Friday January 4, 2002.
(Attention: Dr. Sherita Golden).
GHb (glycated hemoglobin, HbA1c, HbA1) describes a series of stable minor hemoglobin components formed slowly and nonenzymatically from hemoglobin and glucose. This alteration in HbA is a post-translational modification [4644].
Rahbar et al. demonstrated elevation in the minor HbA fractions in patients with diabetes using gel electrophoresis (Rahbar, et al. 1969; [92177]; Rahbar, et al. 1986 [92178]).
Historical perspective:
HbA1 can be separated into three minor components that have a more negative charge:
HbA1a
HbA1b
HbA1c
HbA1c was the component resulting from post-translational modification of HbA by glucose at the N-terminus of the β chain. HbA1c has been correlated with:
fasting plasma glucose
glucose peak during glucose tolerance test
area under the curve of glucose tolerance test
mean glucose levels over preceding few weeks
[4644].
Glycohemoglobin nomenclature:
| Terminology | |
|---|---|
| HbA | The major form of hemoglobin; a native, unmodified tetramer consisting of two α- and β-chains |
| GHb | General term for glucose bound nonenzymatically to hemoglobin with a ketoamine structure |
| HbA1 | GHb species that are more negatively charged forms of HbA detected by cation-exchange chromatographic and electrophoretic methods, which include HbA1a, HbA1b, and HbA1c, “fast” hemoglobins |
| HbA1c | Specific GHb that is an adduct of glucose attached to the β-chain terminal valine residue |
| Total GHb | Term used to describe all GHb species as measured by affinity chromatographic methods |
Glycated hemoglobin measurement methods available in US as of 2002
| Common interferences | ||||||
|---|---|---|---|---|---|---|
| Glycated Hb Measured | NGSP Certified | HbS | HbC | Elevated HbF | HbE | |
| Ion Exchange Chromatography | ||||||
 Bio-Rad DiaStat | HbA1c | |||||
| Bio-Rad Variant | HbA1c | Yes | Yes | No | - | No |
| Total GHb | No | No | - | - | ||
| Bio-Rad Variant II | HbA1c | Yes | No | No | No (HbF≤24%) | Yes |
| Tosoh A1c 2.2 Plus | HbA1c | Yes | No | No | - | Yes |
| Tosoh G7 Auto | HbA1c | Yes | ||||
| Other: Minicolumns | No | |||||
| Boronate Affinity | ||||||
| Abbott IMx | HbA1c | No | Yes | - | - | |
| Total GHb | ||||||
| Abbott Vision | Total GHb | |||||
| Primus HPLC | HbA1c | Yes | No | No | - | No |
| Axis Shield | Yes | No | No | - | - | |
| NycoCard (Primus) | ||||||
| Provalis Glycosal | Yes | |||||
| Bio-Rad MicroMat II | No | Yes | - | - | ||
| Cholestech | ||||||
| Other: Minicolumns | ||||||
| Electrophoresis | ||||||
| Beckman Diatrac | HbA1c | |||||
| Immunoassay | ||||||
| Bayer DCA 2000 | HbA1c | Yes | No | No | - | No |
| Beckman Synchron | HbA1c | Yes | No | No | - | - |
| Dade Behring Dimension | HbA1c | Yes | ||||
| Roche Cobas Integra | HbA1c | Yes | Yes | Yes | - | - |
| Roche Unimate | Yes | Yes | Yes | - | - | |
| Roche Tina-quant II | Yes | No | No | No (HbF≤30%) | No | |
| Metrika A1c Now | Yes | |||||
Adapted from National Glycohemoglobin Standardization Program Website(www.ngsp.org).
College of American Pathologists (CAP) Survey Data (updated 5/02)
The American Diabetes Association (ADA) recommends that laboratories use only GHB assay methods that have been NGSP certified and report results as “%HbA1c” or “%HbA1c equivalents”. The ADA also recommends that all laboratories performing GHB testing participate in the College of American Pathologists (CAP) fresh sample proficiency testing survey (see ADA Recommendations section on this website for more details). The following methods were included in the recent 2002 CAP GH2 survey:
Methods reporting HbA1c or equivalent:
Bayer DCA 2000
Beckman Synchron Syst
Bio-Rad Diastat
Bio-Rad Variant A1c
Bio-Rad Variant II A1c
Dade Behring Dimension
Primus (affinity)
Roche Cobas Integra
Roche/Hitachi (Tina Quant II)
Tosoh A1c 2.2 Plus
Tosoh G7 Auto HPLC
Abbott IMX
Beckman Diatrac
Methods reporting total GHb:
Abbott IMX
Bio-Rad Variant
Helena Glyco-Tek
Primus
Adapted from NGSP Website (www.ngsp.org).
For each eye, the maximum grade in any of the seven standard photographic fields was determined for each of the lesions and used in defining the “retinopathy levels” as follows:
(From: Klein et al, Ophthalmology, 1998 [2108])
| Level 10: | No retinopathy |
| Level 21: | Microaneurysms (MAs) only or retinal hemorrhages (H) or soft exudates in the absence of MAs. |
| Level 31: | Microaneurysms and one or more of the following: venous loops 31 μm or greater; questionable soft exudate, intraretinal microvascular abnormalities (IRMA), or venous beading; and retinal H. |
| Level 37: | Microaneurysms and one or more of the following: hard exudate and soft exudate. |
| Level 43: | Microaneurysms and one or more of the following: H/MAs equaling or exceeding those in Standard Photo (SP) I in four or five fields; H/MAs equaling or exceeding those in SP 2A in one field; and IRMA in one to three fields. |
| Level 47: | Microaneurysms and one or more of the following: both IRMA and H/MA characteristics from level 43; IRMA in four or five fields; H/MAs equaling or exceeding those in SP 2A in two or three fields; and venous beading in one field. |
| Level 53: | Microaneurysms and one or more of the following: any two or three characteristics from level 47; H/Mas equaling or exceeding those in SP 2A in four or five fields; IRMA equaling or exceeding those in SP 8A; venous beading in two or more fields. |
| Level 60: | Fibrous proliferations only. |
| Level 61: | No evidence of level 60 or 65 but scars of photocoagulation either in “scatter” or confluent patches, presumably directed at new vessels. |
| Level 65: | Proliferative diabetic retinopathy (PDR) less than Diabetic Retinopathy Study high-risk characteristics (DRS-HRC). Lesions as follows: new vessels elsewhere (NVE): new vessels on or within 1 disc diameter (NVD) of the disc graded less than SP 10A: or preretinal (PRH) or vitreous hemorrhage (VH) less than 1 disc area (DA). |
| Level 71: | Diabetic Retinopathy Study high-risk characteristics (DRS-HRC). Lesions as follows: VH and/or PRH equaling or exceeding 1 DA: NVE equaling or exceeding half DA with VH and/or PRH: NVD less than SP 10A with VH or PRH or both: and NVD equaling or exceeding SP 10A. |
| Level 75: | Advanced PDR, lesions as follows: NVD equaling or exceeding SP 10A with VH or PRH or both. |
| Level 85: | End-stage PDR, lesions as follows: macula obscured by VH or PRH or both; retinal detachment at center of macula: phthisis bulbi; and enucleation secondary to complications of diabetic retinopathy. |
The team was lead by endocrinologist/epidemiologist Sherita Golden and internist/epidemiologist Ebony Boulware.Further information about the Johns Hopkins Evidence-Based Practice Center can be found at:
http://www.hopkinsmedicine.org/epc/default.htm
Gail Berkenblit, MD, PhD (Research Associate)
L. Ebony Boulware, MD, MPH (Project Co-Director)
Frederick L. Brancati, MD, MHS (Co-Investigator)
Geetanjali Chander, MD, MPH (Research Associate)
Sherita Hill Golden, MD, MHS (Project Co-Director)
Spyridon S. Marinopoulos, MD, MBA (Research Associate)
Michael Paasche-Orlow, MD, MPH (Research Associate)
Neil Powe, MD, MPH, MBA (Task Leader/Co-Investigator)
Tejal Rami, MPH (Project Coordinator/Search Director)
Free Full text in PMC]Free Full text in PMC]
[PubMed]Free Full text in PMC]
[PubMed]Free Full text in PMC]
[PubMed]Free Full text in PMC]Free Full text in PMC]
[PubMed]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]
[PubMed]Free Full text in PMC]
[PubMed]Free Full text in PMC]Free Full text in PMC]
[PubMed]Free Full text in PMC]
[PubMed]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]