Estimation of blood glucose control in diabetes mellitus

Estimation of blood glucose control in diabetes mellitus

Author:

David K McCulloch, MD

Section Editors:

David M Nathan, MD

Joseph I Wolfsdorf, MB, BCh

Deputy Editor:

Jean E Mulder, MD

Contributor Disclosures

All topics are updated as new evidence becomes available and our peer review process is complete.

Literature review current through: Apr 2018. | This topic last updated: Jul 13, 2017.

INTRODUCTION — The demonstration that the development of microvascular complications in patients with type 1 diabetes can be slowed by treating hyperglycemia has led to increased use of intensive insulin regimens to attain strict glycemic control (figure 1 and figure 2) [1-3]. The efficacy of these regimens requires an accurate method to estimate the degree to which this is achieved. It is helpful to first review some basic concepts before discussing the utility of measurements of glycated hemoglobin (A1C) and serum fructosamine to assess glycemic control.

There are three useful measurements for defining glycemic control:

●The mean blood glucose concentration has often been measured in clinical trials as the mean of values obtained before breakfast, mid-morning, before lunch, mid-afternoon, before dinner, and before sleep each day. Clinically, this can be replaced with or supplemented by the more simple measurement of A1C. In some cases, however, there is a disparity between the A1C values and mean blood glucose values. (See 'Recommendations in type 1 diabetes' below.)

●The degree to which blood glucose concentrations fluctuate within the same day can be formally measured as the mean amplitude of glycemic excursions [4].

●The degree to which blood glucose concentrations fluctuate from day to day can be formally measured as the mean of daily differences [5].

The mean blood glucose has received the greatest attention. However, as will be discussed below, a given mean blood glucose value may be associated with different mean glycemic excursions and mean daily differences (figure 3). The approach to improving blood glucose control may be different depending upon whether an elevated mean blood glucose is associated with large fluctuations or not.

ESTIMATION OF MEAN BLOOD GLUCOSE — It has been known since the 1970s that glucose can attach to many proteins via a nonenzymatic, posttranslational process [6]. This occurs in two stages (figure 4):

●A reversible reaction leads to the formation of an aldimine (or Schiff base)

●This is followed by an Amadori rearrangement to form an irreversible ketoamine

A transient elevation in blood glucose concentration can lead to the formation of a large quantity of aldimines. This reaction reverses if the concentration returns to normal. However, formation of the ketoamine is irreversible because glucose remains permanently attached to the protein until it is metabolized. This sequence is important in understanding and interpreting assays that measure glycated proteins. The large majority of commercially available assays of glycated hemoglobin only measure the stable ketoamine and do not measure the labile (aldimine) fraction. Therefore, they reflect long-term (two to three months) average glycemia, relatively unaffected by recent acute fluctuations in glucose levels.

Glycated hemoglobin — The most widely used clinical test is measurement of glycated hemoglobin (also called A1C, hemoglobin A1C, glycohemoglobin, or HbA1C). Hemoglobin formed in new red blood cells enters the circulation with minimal glucose attached. However, red cells are freely permeable to glucose. As a result, glucose becomes irreversibly attached to hemoglobin at a rate dependent upon the prevailing blood glucose concentration. Approximately 1 percent of erythrocytes are destroyed every day, while an equal number of new ones are formed. Thus, the average amount of A1C changes in a dynamic way and indicates the mean blood glucose concentration over the lifespan of the red cell [7,8]. Although the A1C reflects mean blood glucose over the entire 120-day lifespan of the red blood cell, it correlates best with mean blood glucose over the previous 8 to 12 weeks (calculator 1).

This relationship has been demonstrated in several studies that have calculated average glucose on the basis of frequently measured, usually capillary, glucose levels [7,9-11]. As examples:

●The Diabetes Control and Complications Trial (DCCT) estimated the mean blood glucose concentrations derived from seven measurements a day (before and 90 minutes after each of the three major meals, and before bedtime), performed once every three months and compared the average glucose concentration with A1C values in patients with type 1 diabetes (figure 5) [9]. Non-DCCT investigators noted a relatively strong correlation between the limited glucose sampling and A1C values and translated the A1C values into a comparable average glucose level.

●A much smaller study captured average glucose by using continuous glucose monitoring (CGM) over three months in 25 participants (type 1 and type 2 diabetics and nondiabetic subjects) and compared the calculated average glucose levels with the A1C at three months [11].

●A large international study (A1C-Derived Average Glucose or ADAG study) calculated average glucose levels in 507 subjects (268 type 1, 159 type 2, and 80 nondiabetic), using a combination of CGM and self-monitoring of blood glucose testing, and established a reliable regression equation that can be used to translate A1C results into an estimated average glucose value [12]. (See 'Assay' below.)

Studies suggest that A1C values may also be helpful in the diagnosis of impaired glucose tolerance or overt diabetes mellitus, being simpler to perform and to repeat than the oral glucose tolerance test. (See "Clinical presentation and diagnosis of diabetes mellitus in adults".)

Assay — In the past, the results of the DCCT could not be extrapolated widely, because of differences in methodology and a lack of standardization among laboratories [13]. The National Glycohemoglobin Standardization Program (NGSP) has standardized more than 99 percent of the assays used in the United States to the DCCT standard [14]. A strict quality control program has improved precision and accuracy of assays in the United States and many international assays.

In addition, a new reference method has been established that will provide for even more reliable worldwide standardization of all A1C assays [15]. With this new reference system, A1C results will be reported globally in SI (Systeme International) units (mmol/mol) and derived NGSP units (the same values as reported currently as percent of total hemoglobin) using a master equation. An estimated average glucose (eAG), calculated from the A1C result and based upon the results of the ADAG study described above, will be included in the report (calculator 1).

The eAG (mg/dL or mmol/L) is a more relevant term for patients who self-monitor blood glucose. If there are differences in the eAG derived from the A1C result and the meter-calculated average glucose, further exploration is required. As an example, if the eAG is higher than the patient's meter-calculated average glucose, it is possible that fingerstick testing is not being performed at times when the blood glucose is highest, eg, after meals. If the eAG is lower than the meter average, a patient may be having undetected periods of low blood glucose, most often nocturnal hypoglycemia. In such cases, the timing of fingerstick blood glucose monitoring requires adjustment.

Sources of error — Although the international standardization of the A1C assay has decreased potential technical errors in interpreting A1C results, there are other biological and patient-specific factors that may cause misleading results [16]:

●A1C values are influenced by red cell survival. Thus, falsely high values in relation to a mean blood glucose values can be obtained when red cell turnover is low, resulting in a disproportionate number of older red cells. This problem can occur in patients with iron, vitamin B12, or folate deficiency anemia.

●On the other hand, rapid red cell turnover leads to a greater proportion of younger red cells and falsely low A1C values. Examples include patients with hemolysis or anemia and those treated for iron, vitamin B12, or folate deficiency, and patients treated with erythropoietin [17-20].

●Depending upon the methodology, the values may be falsely high in patients with abnormal hemoglobins (such as hemoglobin F [HbF]) or low with hemoglobin S (HbS) [21]. However, many methods for measuring A1C are no longer affected by the most common hemoglobin variants. The NGSP website contains current information about substances that interfere with A1C test results. (See "Interactive diabetes case 8: Discordant values for A1C and home blood glucose values".)

●A1C values may be falsely elevated or decreased in those with chronic kidney disease. False elevations may be due in part to analytical interference from carbamylated hemoglobin formed in the presence of elevated concentrations of urea, leading to false elevations in the A1C level with some assays. False decreases in measured A1C may occur with hemodialysis and altered red cell turnover, especially in the setting of erythropoietin treatment. (See "Management of hyperglycemia in patients with type 2 diabetes and pre-dialysis chronic kidney disease or end-stage renal disease", section on 'Monitoring glycemic control'.)

Racial variation — Several studies have shown that A1C concentrations are higher in some ethnic groups (African American, Hispanic, Asian) than in white persons with similar plasma glucose concentrations [22-26]. The differences are evident across the spectrum of glycemia (from normal glucose tolerance to prediabetes to diabetes) and are independent of assay method. In one cross-sectional study, A1C was 0.13 to 0.47 percentage points higher in black than in white persons, with the difference increasing as glucose intolerance worsened [27]. Whether the higher A1C observed in black persons is due to worse glycemic control or racial variation in the glycation of hemoglobin is uncertain. Retrospective studies suggest that the presence of sickle cell trait does not account for the racial variation in African Americans [28,29]. In one study, A1C was actually lower at any fasting glucose value in African Americans with than those without sickle cell trait [30].

All of these studies estimated mean glucose levels on the basis of very limited measurements, usually a single fasting glucose level or oral glucose tolerance test. This leaves open the possibility that if measured appropriately, ie, with frequent enough glucose measurements over time, mean glucose levels might be different between different racial groups and the putative different A1C levels between races reflect different mean glucose levels. This latter scenario has been supported by a cross-sectional study in which differences in A1C between races was paralleled by other, independent measures of chronic glycemia [25]. This finding supports differences in glucose levels as the reason for different levels of A1C in different racial/ethnic groups.

In a prospective, 12-week study comparing A1C with mean glucose values measured by CGM in black and white persons with type 1 diabetes, both average CGM glucose (191 versus 180 mg/dL [10.6 versus 10 mmol/L]) and A1C (9.1 versus 8.3 percent) were higher in black than white individuals [31]. Although the correlation between mean glucose and A1C did not differ significantly by race, the mean A1C in black compared with white individuals was 0.4 percentage points higher for any given mean glucose concentration. The racial variation explained only a proportion of the difference in mean A1C levels between the two groups, with higher mean glucose values likely accounting for the rest. There was no racial difference in the relationship between other measures of chronic glycemia (eg, glycated albumin, fructosamine) and mean glucose concentration, suggesting an erythrocyte-specific mechanism for the variation.  

If differences in A1C between races do exist, the differences appear to be small and have not been shown to significantly modify the association between A1C and cardiovascular outcomes [32], retinopathy [33-35], or nephropathy [34]. In a cross-sectional study using the National Health and Nutrition Examination Survey (NHANES 2005 to 2008), the prevalence of retinopathy actually increased at a lower A1C level in blacks compared with whites (5.5 to 5.9 versus 6.0 to 6.4 percent) [33].

If a difference does exist, potential mechanisms include genetic differences in intracellular glycation, glucose transport across the erythrocyte membrane, or in erythrocyte survival [36,37].

Fructosamine — Many proteins other than hemoglobin also undergo nonenzymatic glycation [38], leading to the formation of advanced glycosylation end products, which may play a direct role in the development of diabetic microvascular complications (figure 4) [39] (see "Glycemic control and vascular complications in type 1 diabetes mellitus", section on 'Pathogenesis'). The serum concentration of some of these proteins can also be used to estimate glycemic control. The term fructosamine has been applied to the ketoamines formed in this process.

Several methods are available for measuring serum fructosamine [38]. Some of the assays are cheaper and easier to perform than the A1C assay. There is generally a good correlation between serum fructosamine and A1C values [40-42]. There are, however, several potential problems with the use of serum fructosamine measurements:

●The within-subject variation for serum fructosamine is higher than that for A1C; as a result, serum fructosamine concentrations must change more before a significant change can be said to have occurred [43].

●The turnover of serum albumin is more rapid than that of hemoglobin (28 versus 120 days). Thus, serum fructosamine values reflect mean blood glucose values over a much shorter period of time (one to two weeks).

●Serum fructosamine values must be adjusted if the serum albumin concentration is abnormal [44]. Furthermore, falsely low values in relation to mean blood glucose values will occur with rapid albumin turnover as occurs in patients with protein-losing enteropathy or the nephrotic syndrome.

These limitations plus the lack of necessity of following changes in mean blood glucose concentrations every one to two weeks means that A1C is usually preferable for estimating mean blood glucose concentrations [45].

WITHIN-DAY GLUCOSE VARIATIONS — The range over which blood glucose concentrations vary within a day has been used to define one type of "unstable diabetes" [4]. A useful approximation of these fluctuations in patients who do blood glucose monitoring can be obtained by measuring the blood glucose before and 90 minutes after meals for several days (see "Self-monitoring of blood glucose in management of adults with diabetes mellitus"). While unstable within-day glucose variation may increase the risk of hypoglycemia, it does not appear to be an additional factor in the development of microvascular complications over and above the average blood glucose level as measured by A1C [46].

1,5-anhydroglucitol — Measurement of serum 1,5-anhydroglucitol (1,5-AG), a naturally occurring dietary polyol, is another method that provides information on daily glycemic variations [47-49]. During euglycemia, 1,5-AG is filtered and is completely reabsorbed by the kidneys; as a result, serum concentrations remain stable. However, renal reabsorption of 1,5-AG is competitively inhibited by glucose. Within 24 hours of a rise in serum glucose to >180 mg/dL (10 mmol/L, the renal threshold for glucose excretion), serum 1,5-AG concentrations fall as urinary losses increase [50,51]. Thus, serum 1,5-AG measurements reflect blood glucose values over the past 24 hours, whereas A1C and fructosamine measurements reflect values over two to three months and one to two weeks, respectively.

In addition, serum 1,5-AG may reflect postprandial glycemic excursions better than A1C [49,52]. However, there are no data to suggest that complementary measurement of 1,5-AG (GlycoMark assay) improves glycemic control or complications of diabetes more than measurement of A1C alone. Thus, we do not typically measure 1,5-AG concentrations. In patients who have high A1C values despite excellent fasting and preprandial blood glucose levels, we prefer to test blood glucose one to two hours after meals and adjust medication accordingly.

Optimal therapy of postprandial hyperglycemia (figure 3) requires changes in either the nutrition prescription or the insulin regimen. As an example, a large meal containing a lot of quickly absorbed carbohydrate that is low in soluble fiber will cause substantial postprandial hyperglycemia. Changing the overall distribution of carbohydrate (eating somewhat less with each meal and more with between meal snacks) and increasing the intake of soluble fiber should dampen the glycemic excursions after meals. (See "Nutritional considerations in type 1 diabetes mellitus" and "Nutritional considerations in type 2 diabetes mellitus".)

Altering the insulin regimen and using an adjustable premeal algorithm also may be helpful. Large postprandial rises in blood glucose levels can be minimized by increasing the time between the injection of regular insulin (which begins to act within 20 to 30 minutes after subcutaneous injection) and the beginning of the meal. The availability of monomeric rapidly absorbed insulin analogues, such as insulin lispro, which have a much more rapid onset of action and larger peak effect than regular insulin, reduces the lag time between injection and onset of action to approximately 15 minutes [53]. (See "General principles of insulin therapy in diabetes mellitus" and "Management of blood glucose in adults with type 1 diabetes mellitus".)

DAY-TO-DAY GLUCOSE VARIATIONS — The extent to which blood glucose concentrations vary at the same time each day is another useful measure of overall glycemic control [5]. For patients who test their own blood glucose at the same times every day, this variability can be evaluated simply by scanning down columns of blood glucose measurements over several days. The usual explanation for large daily fluctuations is an erratic lifestyle in terms of eating or exercise habits. These problems should be corrected before increasing the insulin dose. (See "Cases illustrating problems with insulin therapy for diabetes mellitus", section on 'Lack of control due to diet'.)

The use of continuous glucose monitoring (CGM) devices can make the identification of within-day and between-day variation easier. (See "Self-monitoring of blood glucose in management of adults with diabetes mellitus".)

RECOMMENDATIONS IN TYPE 1 DIABETES — We measure A1C every three months in patients with type 1 diabetes. The goal must be correlated with the normal range for the assay in the particular laboratory, although the National Glycohemoglobin Standardization Program (NGSP) has largely standardized the values for much of the world. For patients who do blood glucose monitoring, both the patient's technique and the meter's reliability should be checked periodically. It is then useful to compare an average of the 50 or so most recent blood glucose readings with the A1C value obtained at that visit (figure 5):

●If the A1C value is higher than expected, it is possible that the patient is falsifying his or her blood glucose results or has made an effort to improve glycemic control in the two weeks before the appointment. Another explanation is that the blood glucose concentrations are much higher at times between blood tests (such as between meals). Finally, some of the factors discussed above, which can falsely elevate the A1C, should be excluded (such as abnormal hemoglobins, untreated iron deficiency, or renal failure). (See 'Sources of error' above.)

●If the A1C value is lower than expected, it is possible that the blood glucose is low for long periods of time when the patient is not testing (such as undetected nocturnal hypoglycemia). Other possibilities include reduced red cell survival (such as hemolysis) or conditions in which a disproportionate number of red cells are young (as with chronic bleeding, venisection, treatment of anemia). (See 'Sources of error' above.)

If the patient is interested in improving glycemic control by intensifying his or her diabetes regimen, then attention must be paid to the blood glucose patterns to detect problems of within-day and between-day variability. These fluctuations should be corrected before increasing the insulin regimen.

The Diabetes Control and Complications Trial (DCCT) found an inverse relationship between the A1C value and the incidence of developing diabetic retinopathy in patients with type 1 diabetes (figure 2). The risk of retinopathy was near its lowest level in patients with A1C values of approximately 7 to 7.5 percent. There is a continued relative risk reduction of retinopathy with A1C levels less than 7.5 percent, similar in magnitude to the reduction at higher A1C levels; however, the absolute risk of retinopathy is low (figure 6). Thus, increasing the intensity of glycemic control should only be considered if recurrent severe hypoglycemia is not occurring. In general, we aim for an A1C value of 7 percent or less. This approach is consistent with that of the American Diabetes Association (ADA), which recommends a target of less than 7 percent for most patients. (See "Glycemic control and vascular complications in type 1 diabetes mellitus".)

RECOMMENDATIONS IN TYPE 2 DIABETES — Although similar general considerations apply to patients with type 2 diabetes, there is less variability in blood glucose concentrations in this disorder. As a result, the fasting blood glucose concentration correlates fairly well with the A1C value and can be used with the A1C to estimate glycemic control [54].

Some authors have argued that nonfasting blood glucose measurements are a better marker of blood glucose control than fasting values [9,55]. However, the effectiveness of self-monitoring of blood glucose in terms of improving glycemic control in patients with type 2 diabetes is less clear than for type 1 diabetes. Blood glucose self-monitoring should not be regarded as an intervention that in its own right will be associated with improved glycemic control. Rather, it is a tool that gives invaluable information with which motivated and well-educated patients can modify their behavior and improve their glycemic control (and A1C values) safely. (See "Self-monitoring of blood glucose in management of adults with diabetes mellitus", section on 'Type 2 diabetes'.)

Improving glycemic control is often more difficult in patients with type 2 diabetes because of obesity and insulin resistance (see "Initial management of blood glucose in adults with type 2 diabetes mellitus" and "Management of persistent hyperglycemia in type 2 diabetes mellitus"). The frequency of blood glucose monitoring is usually less than for type 1 diabetes but can vary considerably depending upon the stage of the disease, the targets being set, and the treatments being used. (See "Case illustrating blood glucose monitoring in type 2 diabetes".)

The optimal level of glycemic control to prevent microvascular complications in patients with type 2 diabetes is not clear. The United Kingdom Prospective Diabetes Study (UKPDS) showed that intensive therapy (mean A1C value of 7 percent) with oral hypoglycemic drugs or insulin resulted in a decreased risk of microvascular complications compared with conventional therapy with diet (mean A1C value of 7.9 percent) (figure 7) [56]. Subanalysis of the UKPDS showed no evidence of a threshold effect of A1C; a 1 percent reduction in A1C was associated with a 35 percent reduction in microvascular endpoints. Finally, long-term follow-up of UKPDS revealed a benefit of intensive therapy compared with conventional therapy, with A1C levels of 7 versus 7.9 percent during the trial, on cardiovascular events [57].

Several trials [58,59] have examined the effects of even lower A1C levels than achieved during UKPDS on cardiovascular outcomes. The randomly assigned intensive groups achieved lower A1C levels than the control groups (approximately 6.5 compared with 7.5 percent) and microvascular outcomes, which were secondary outcomes, were reduced [60-62]. However, these trials had either no effect on cardiovascular disease [59] or were associated with higher overall and cardiovascular mortality [58]. Thus, the trials that achieved A1C levels of approximately 6.5 percent had lower microvascular disease but either no better or worse cardiovascular effects. These trials are reviewed in detail separately. (See "Glycemic control and vascular complications in type 2 diabetes mellitus", section on 'Intensive therapy'.)

On balance and taking the results of the UKPDS and the more recent large clinical trials into account, we recommend that the target A1C value should be 7 percent or lower for most patients. The goal should be set somewhat higher for older patients and those with comorbid conditions or a limited life expectancy in whom the risk of hypoglycemia may outweigh the potential benefit. (See "Glycemic control and vascular complications in type 2 diabetes mellitus" and "Initial management of blood glucose in adults with type 2 diabetes mellitus" and "Management of persistent hyperglycemia in type 2 diabetes mellitus".)

PATIENT FEEDBACK — In adults, immediate patient feedback at clinician visits about A1C values may improve glycemic control [63,64]. These studies used rapid A1C assays, but it may also be feasible to measure A1C several days before routine office visits so that the results are available at the time of the visit. In a randomized trial of immediate (fingerstick point-of-care testing at office visits) or conventional (venipuncture with follow-up days after visit) feedback in children under 18 years of age, immediate feedback resulted in improvement in A1C after three months, which was not sustained at 6 or 12 months [65]. There were no significant differences between the groups with regard to frequency of treatment changes to insulin dosing, dietary, exercise, or blood glucose monitoring recommendations. Point-of-care testing was less painful and, therefore, more acceptable to children, and immediate feedback resulted in more efficient patient-clinician communication.

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5th to 6th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10th to 12th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topics (see "Patient education: Type 1 diabetes (The Basics)" and "Patient education: Type 2 diabetes (The Basics)" and "Patient education: Hemoglobin A1C tests (The Basics)")

●Beyond the Basics topics (see "Patient education: Diabetes mellitus type 1: Overview (Beyond the Basics)" and "Patient education: Diabetes mellitus type 2: Overview (Beyond the Basics)" and "Patient education: Self-monitoring of blood glucose in diabetes mellitus (Beyond the Basics)")

SUMMARY

●The most widely used clinical test to estimate blood glucose control is measurement of glycated hemoglobin (also called A1C, hemoglobin A1C, glycohemoglobin, or HbA1C). A1C reflects mean blood glucose over the entire 120-day lifespan of the red blood cell, but it correlates best with mean blood glucose over the previous 8 to 12 weeks. (See 'Glycated hemoglobin' above.)

●As part of an effort to provide worldwide standardization of all A1C assays, A1C results will be reported globally in International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) units (mmol/mol) and derived National Glycohemoglobin Standardization Program (NGSP) units (the same values as reported currently as percent of total hemoglobin) using a master equation. An estimated average glucose (eAG), calculated from the A1C result, will be included in the report (calculator 1). The eAG is a more relevant term for patients who self-monitor blood glucose. (See 'Glycated hemoglobin' above.)

●Although the international standardization of the A1C assay has decreased potential technical errors in interpreting A1C results, there are other biological and patient-specific factors that may cause misleading results. (See 'Sources of error' above.)

●The extent to which blood glucose concentrations vary at the same time each day is another useful measure of overall glycemic control. For patients who test their own blood glucose at the same times every day, this variability can be evaluated simply by scanning down columns of blood glucose measurements over several days (see 'Day-to-day glucose variations' above). The use of continuous glucose monitoring (CGM) devices can make the identification of within-day and between-day variation easier. (See "Self-monitoring of blood glucose in management of adults with diabetes mellitus", section on 'Continuous glucose monitoring'.)