This document presents standards of care for diabetes from the American Diabetes Association. It discusses:
1) The classification of diabetes into four types: type 1, type 2, other specific types, and gestational diabetes.
2) The diagnostic criteria for diabetes, which can now include an A1C level of ≥6.5% in addition to fasting plasma glucose and oral glucose tolerance test thresholds.
3) Categories of increased risk for diabetes, called prediabetes, which include impaired fasting glucose, impaired glucose tolerance, and an A1C of 5.7-6.4%.
Integrating multiple co-morbidities in guidelinesGuíaSalud
Presentación realizada por Holger Schünemann, profesor y director del Departamento de Epidemiología Clínica y Bioestadísticas en la Universidad McMaster de Hamilton, Canadá, en las Jornadas Científicas "Guías de Práctica Clínica y Pluripatología" de GuíaSalud, Madrid, 21 de febrero de 2013.
Integrating multiple co-morbidities in guidelinesGuíaSalud
Presentación realizada por Holger Schünemann, profesor y director del Departamento de Epidemiología Clínica y Bioestadísticas en la Universidad McMaster de Hamilton, Canadá, en las Jornadas Científicas "Guías de Práctica Clínica y Pluripatología" de GuíaSalud, Madrid, 21 de febrero de 2013.
2. Position Statement
Table 1dADA evidence grading system for clinical practice recommendations PG) remain valid as well (Table 2). Just as
there is less than 100% concordance be-
Level of tween the FPG and 2-h PG tests, there is
evidence Description no perfect concordance between A1C and
either glucose-based test. Analyses of the
A Clear evidence from well-conducted, generalizable RCTs that are adequately National Health and Nutrition Examina-
powered, including: tion Survey (NHANES) data indicate that,
c Evidence from a well-conducted multicenter trial assuming universal screening of the un-
c Evidence from a meta-analysis that incorporated quality ratings in the diagnosed, the A1C cut point of $6.5%
analysis identifies one-third fewer cases of undiag-
Compelling nonexperimental evidence, i.e., “all or none” rule developed by the nosed diabetes than a fasting glucose cut
Centre for Evidence-Based Medicine at the University of Oxford point of $126 mg/dL (7.0 mmol/L) (11),
Supportive evidence from well-conducted RCTs that are adequately powered, and numerous studies have confirmed
including: that at these cut points the 2-h OGTT
c Evidence from a well-conducted trial at one or more institutions value diagnoses more screened people
c Evidence from a meta-analysis that incorporated quality ratings in the analysis with diabetes (12). However, in practice, a
B Supportive evidence from well-conducted cohort studies large portion of the diabetic population re-
c Evidence from a well-conducted prospective cohort study or registry mains unaware of its condition. Thus, the
c Evidence from a well-conducted meta-analysis of cohort studies lower sensitivity of A1C at the designated
Supportive evidence from a well-conducted case-control study cut point may well be offset by the test’s
C Supportive evidence from poorly controlled or uncontrolled studies greater practicality, and wider application
c Evidence from randomized clinical trials with one or more major or three or of a more convenient test (A1C) may actu-
more minor methodological flaws that could invalidate the results ally increase the number of diagnoses made.
c Evidence from observational studies with high potential for bias (such as case As with most diagnostic tests, a test
series with comparison with historical controls) result diagnostic of diabetes should be
c Evidence from case series or case reports repeated to rule out laboratory error,
Conflicting evidence with the weight of evidence supporting the recommendation unless the diagnosis is clear on clinical
E Expert consensus or clinical experience grounds, such as a patient with a hyper-
glycemic crisis or classic symptoms of
hyperglycemia and a random plasma
Federation (IDF), and the European have posited that glycation rates differ by glucose $200 mg/dL. It is preferable
Association for the Study of Diabetes race (with, for example, African Americans that the same test be repeated for confir-
(EASD) recommended the use of the A1C having higher rates of glycation), but this is mation, since there will be a greater likeli-
test to diagnose diabetes, with a threshold controversial. A recent epidemiological hood of concurrence in this case. For
of $6.5% (6), and the ADA adopted this study found that, when matched for FPG, example, if the A1C is 7.0% and a repeat
criterion in 2010 (5). The diagnostic test African Americans (with and without dia- result is 6.8%, the diagnosis of diabetes is
should be performed using a method that betes) indeed had higher A1C than whites, confirmed. However, if two different tests
is certified by the NGSP and standardized but also had higher levels of fructosamine (such as A1C and FPG) are both above the
or traceable to the Diabetes Control and and glycated albumin and lower levels of diagnostic thresholds, the diagnosis of di-
Complications Trial (DCCT) reference as- 1,5 anhydroglucitol, suggesting that their abetes is also confirmed.
say. Although point-of-care (POC) A1C as- glycemic burden (particularly postpran- On the other hand, if two different
says may be NGSP certified, proficiency dially) may be higher (9). Epidemiological tests are available in an individual and the
testing is not mandated for performing studies forming the framework for recom- results are discordant, the test whose result
the test, so use of these assays for diagnostic mending use of the A1C to diagnose diabe- is above the diagnostic cut point should be
purposes could be problematic. tes have all been in adult populations. repeated, and the diagnosis is made based
Epidemiological datasets show a sim- Whether the cut point would be the same on the confirmed test. That is, if a patient
ilar relationship for A1C to the risk of to diagnose children or adolescents with meets the diabetes criterion of the A1C (two
retinopathy as has been shown for the type 2 diabetes is an area of uncertainty results $6.5%) but not the FPG (,126 mg/
corresponding FPG and 2-h PG thresh- (3,10). A1C inaccurately reflects glycemia dL or 7.0 mmol/L), or vice versa, that per-
olds. The A1C has several advantages to with certain anemias and hemoglobinopa- son should be considered to have diabetes.
the FPG and OGTT, including greater thies. For patients with an abnormal hemo- Since there is preanalytical and ana-
convenience (since fasting is not required), globin but normal red cell turnover, such as lytical variability of all the tests, it is also
evidence to suggest greater preanalytical sickle cell trait, an A1C assay without inter- possible that when a test whose result was
stability, and less day-to-day perturbations ference from abnormal hemoglobins should above the diagnostic threshold is re-
during periods of stress and illness. These be used (an updated list is available at www. peated, the second value will be below
advantages must be balanced by greater ngsp.org/interf.asp). For conditions with the diagnostic cut point. This is least
cost, the limited availability of A1C testing abnormal red cell turnover, such as preg- likely for A1C, somewhat more likely for
in certain regions of the developing world, nancy, recent blood loss or transfusion, or FPG, and most likely for the 2-h PG.
and the incomplete correlation between some anemias, the diagnosis of diabetes Barring a laboratory error, such patients
A1C and average glucose in certain indi- must employ glucose criteria exclusively. are likely to have test results near the
viduals. In addition, HbA1c levels may vary The established glucose criteria for margins of the threshold for a diagnosis.
with patients’ race/ethnicity (7,8). Some the diagnosis of diabetes (FPG and 2-h The health care professional might opt to
S12 DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 care.diabetesjournals.org
3. Position Statement
follow the patient closely and repeat the used A1C to predict the progression to Table 3dCategories of increased risk for
testing in 3–6 months. diabetes demonstrated a strong, continu- diabetes (prediabetes)*
The current diagnostic criteria for ous association between A1C and sub- FPG 100 mg/dL (5.6 mmol/L) to 125 mg/dL
diabetes are summarized in Table 2. sequent diabetes. In a systematic review of (6.9 mmol/L) (IFG)
44,203 individuals from 16 cohort stud- OR
C. Categories of increased risk ies with a follow-up interval averaging 5.6 2-h plasma glucose in the 75-g OGTT 140 mg/dL
for diabetes (prediabetes) years (range 2.8–12 years), those with an (7.8 mmol/L) to 199 mg/dL (11.0 mmol/L)
In 1997 and 2003, the Expert Committee A1C between 5.5 and 6.0% had a substan- (IGT)
on Diagnosis and Classification of Diabe- tially increased risk of diabetes with 5-year OR
tes Mellitus (13,14) recognized an inter- incidences ranging from 9 to 25%. An A1C A1C 5.7–6.4%
mediate group of individuals whose range of 6.0–6.5% had a 5-year risk of de-
glucose levels, although not meeting cri- veloping diabetes between 25 to 50% and *For all three tests, risk is continuous, extending be-
low the lower limit of the range and becoming dis-
teria for diabetes, are nevertheless too relative risk (RR) 20 times higher compared proportionately greater at higher ends of the range.
high to be considered normal. These per- with an A1C of 5.0% (15). In a community-
sons were defined as having impaired fast- based study of black and white adults
ing glucose (IFG) (FPG levels 100 mg/dL without diabetes, baseline A1C was a For many illnesses, there is a major dis-
[5.6 mmol/L] to 125 mg/dL [6.9 mmol/L]) stronger predictor of subsequent diabetes tinction between screening and diagnostic
or impaired glucose tolerance (IGT) (2-h and cardiovascular events than was fast- testing. However, for diabetes, the same
values in the OGTT of 140 mg/dL [7.8 ing glucose (16). Other analyses suggest tests would be used for “screening” as for
mmol/L] to 199 mg/dL [11.0 mmol/L]). It that an A1C of 5.7% is associated with diagnosis. Diabetes may be identified any-
should be noted that the World Health diabetes risk similar to that in the high- where along a spectrum of clinical scenar-
Organization (WHO) and a number of risk participants in the Diabetes Prevention ios ranging from a seemingly low-risk
other diabetes organizations define the cut- Program (DPP) (17). individual who happens to have glucose
off for IFG at 110 mg/dL (6.1 mmol/L). Hence, it is reasonable to consider an testing, to a higher-risk individual whom
Individuals with IFG and/or IGT have A1C range of 5.7–6.4% as identifying in- the provider tests because of high suspicion
been referred to as having prediabetes, dividuals with prediabetes. As is the case of diabetes, to the symptomatic patient.
indicating the relatively high risk for the for individuals found to have IFG and The discussion herein is primarily framed
future development of diabetes. IFG and IGT, individuals with an A1C of 5.7–6.4% as testing for diabetes in those without
IGT should not be viewed as clinical should be informed of their increased risk symptoms. The same assays used for test-
entities in their own right but rather risk for diabetes as well as CVD and counseled ing for diabetes will also detect individuals
factors for diabetes as well as cardiovascular about effective strategies to lower their risks with prediabetes.
disease (CVD). IFG and IGT are associated (see Section IV). As with glucose measure-
with obesity (especially abdominal or vis- ments, the continuum of risk is curvilinear, A. Testing for type 2 diabetes and
ceral obesity), dyslipidemia with high tri- so that as A1C rises, the risk of diabetes rises risk of future diabetes in adults
glycerides and/or low HDL cholesterol, and disproportionately (15). Accordingly, inter- Prediabetes and diabetes meet established
hypertension. ventions should be most intensive and criteria for conditions in which early de-
As is the case with the glucose mea- follow-up particularly vigilant for those tection is appropriate. Both conditions are
sures, several prospective studies that with A1Cs above 6.0%, who should be con- common, increasing in prevalence, and
sidered to be at very high risk. impose significant public health burdens.
Table 2dCriteria for the diagnosis of Table 3 summarizes the categories of There is a long presymptomatic phase
diabetes prediabetes. before the diagnosis of type 2 diabetes is
usually made. Relatively simple tests are
A1C $6.5%. The test should be performed in II. TESTING FOR DIABETES IN available to detect preclinical disease. Ad-
a laboratory using a method that is NGSP ASYMPTOMATIC PATIENTS ditionally, the duration of glycemic burden
certified and standardized to the DCCT is a strong predictor of adverse outcomes,
assay.* Recommendations and effective interventions exist to prevent
OR c Testing to detect type 2 diabetes and progression of prediabetes to diabetes (see
FPG $126 mg/dL (7.0 mmol/L). Fasting is prediabetes in asymptomatic people Section IV) and to reduce risk of compli-
defined as no caloric intake for at least 8 h.* should be considered in adults of any cations of diabetes (see Section VI).
OR age who are overweight or obese (BMI Type 2 diabetes is frequently not di-
2-h plasma glucose $200 mg/dL (11.1 mmol/L) $25 kg/m2) and who have one or more agnosed until complications appear, and
during an OGTT. The test should be additional risk factors for diabetes (Table approximately one-fourth of all people
performed as described by the WHO, using 4). In those without these risk factors, with diabetes in the U.S. may be undiag-
a glucose load containing the equivalent of testing should begin at age 45. (B) nosed. The effectiveness of early identifica-
75 g anhydrous glucose dissolved in water.* c If tests are normal, repeat testing at least tion of prediabetes and diabetes through
OR at 3-year intervals is reasonable. (E) mass testing of asymptomatic individuals
In a patient with classic symptoms of c To test for diabetes or prediabetes, the has not been proven definitively, and
hyperglycemia or hyperglycemic crisis, A1C, FPG, or 75-g 2-h OGTT are appro- rigorous trials to provide such proof are
a random plasma glucose $200 mg/dL priate. (B) unlikely to occur. In a large randomized
(11.1 mmol/L). c In those identified with prediabetes, controlled trial (RCT) in Europe, general
*In the absence of unequivocal hyperglycemia, re- identify and, if appropriate, treat other practice patients between the ages of 40–
sult should be confirmed by repeat testing. CVD risk factors. (B) 69 years were screened for diabetes and
care.diabetesjournals.org DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 S13
4. Position Statement
then randomly assigned by practice to 24 kg/m2 in South Asians, 25 kg/m2 in seek, or have access to, appropriate follow-up
routine care of diabetes or intensive treat- Chinese, and 26 kg/m2 in African Ameri- testing and care. Conversely, there may be
ment of multiple risk factors. After 5.3 cans (21). Disparities in screening rates, failure to ensure appropriate repeat testing
years of follow-up, CVD risk factors were not explainable by insurance status, are for individuals who test negative. Commu-
modestly but significantly more improved highlighted by evidence that despite nity screening may also be poorly targeted; i.
with intensive treatment. Incidence of first much higher prevalence of type 2 diabe- e., it may fail to reach the groups most at risk
CVD event and mortality rates were not tes, non-Caucasians in an insured popu- and inappropriately test those at low risk (the
significantly different between groups lation are no more likely than Caucasians worried well) or even those already diag-
(18). This study would seem to add sup- to be screened for diabetes (22). Because nosed.
port for early treatment of screen-detected age is a major risk factor for diabetes, test-
diabetes, as risk factor control was excel- ing of those without other risk factors
lent even in the routine treatment arm should begin no later than age 45 years. B. Screening for type 2 diabetes
and both groups had lower event rates The A1C, FPG, or the 2-h OGTT are in children
than predicted. The absence of a control appropriate for testing. It should be noted Recommendations
unscreened arm limits the ability to defi- that the tests do not necessarily detect c Testing to detect type 2 diabetes and
nitely prove that screening impacts out- diabetes in the same individuals. The prediabetes should be considered in chil-
comes. Mathematical modeling studies efficacy of interventions for primary pre- dren and adolescents who are overweight
suggest that screening independent of vention of type 2 diabetes (23–29) has and who have two or more additional
risk factors beginning at age 30 years primarily been demonstrated among in- risk factors for diabetes (Table 5). (E)
or age 45 years is highly cost-effective dividuals with IGT, not for individuals
(,$11,000 per quality-adjusted life- with isolated IFG or for individuals with The incidence of type 2 diabetes in
year gained) (19). specific A1C levels. adolescents has increased dramatically in
Recommendations for testing for di- The appropriate interval between the last decade, especially in minority
abetes in asymptomatic, undiagnosed tests is not known (30). The rationale populations (31), although the disease
adults are listed in Table 4. Testing should for the 3-year interval is that false nega- remains rare in the general pediatric pop-
be considered in adults of any age with tives will be repeated before substantial ulation (32). Consistent with recom-
BMI $25 kg/m2 and one or more of the time elapses, and there is little likelihood mendations for adults, children and
known risk factors for diabetes. In addi- that an individual will develop significant youth at increased risk for the presence
tion to the listed risk factors, certain med- complications of diabetes within 3 years or the development of type 2 diabetes
ications, such as glucocorticoids and of a negative test result. In the modeling should be tested within the health care
antipsychotics (20), are known to in- study, repeat screening every 3 or 5 years setting (33). The recommendations of
crease the risk of type 2 diabetes. There was cost-effective (19). the ADA consensus statement “Type 2
is compelling evidence that lower BMI cut Because of the need for follow-up and Diabetes in Children and Adolescents,”
points suggest diabetes risk in some racial discussion of abnormal results, testing with some modifications, are summa-
and ethnic groups. In a large multiethnic should be carried out within the health rized in Table 5.
cohort study, for an equivalent incidence care setting. Community screening outside
rate of diabetes conferred by a BMI of 30 a health care setting is not recommended
kg/m2 in whites, the BMI cutoff value was because people with positive tests may not C. Screening for type 1 diabetes
Recommendations
c Consider referring relatives of those
Table 4dCriteria for testing for diabetes in asymptomatic adult individuals with type 1 diabetes for antibody test-
ing for risk assessment in the setting
1. Testing should be considered in all adults who are overweight (BMI $25 kg/m2*)
of a clinical research study. (E)
and have additional risk factors:
c physical inactivity
Generally, people with type 1 diabetes
c first-degree relative with diabetes
present with acute symptoms of diabetes
c high-risk race/ethnicity (e.g., African American, Latino, Native American, Asian
and markedly elevated blood glucose
American, Pacific Islander)
levels, and some cases are diagnosed with
c women who delivered a baby weighing .9 lb or were diagnosed with GDM
life-threatening ketoacidosis. Evidence
c hypertension ($140/90 mmHg or on therapy for hypertension)
from several studies suggests that mea-
c HDL cholesterol level ,35 mg/dL (0.90 mmol/L) and/or a triglyceride
surement of islet autoantibodies in rela-
level .250 mg/dL (2.82 mmol/L)
tives of those with type 1 diabetes
c women with polycystic ovary syndrome
identifies individuals who are at risk for
c A1C $5.7%, IGT, or IFG on previous testing
developing type 1 diabetes. Such testing,
c other clinical conditions associated with insulin resistance (e.g., severe obesity,
coupled with education about symptoms
acanthosis nigricans)
of diabetes and follow-up in an observa-
c history of CVD
tional clinical study, may allow earlier
2. In the absence of the above criteria, testing for diabetes should begin at age 45 years.
identification of onset of type 1 diabetes
3. If results are normal, testing should be repeated at least at 3-year intervals, with
and lessen presentation with ketoacidosis
consideration of more frequent testing depending on initial results (e.g., those with
at time of diagnosis. This testing may be
prediabetes should be tested yearly) and risk status.
appropriate in those who have relatives
*At-risk BMI may be lower in some ethnic groups. with type 1 diabetes, in the context of
S14 DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 care.diabetesjournals.org
5. Position Statement
Table 5dTesting for type 2 diabetes in asymptomatic children* screen women with risk factors for type
Criteria 2 diabetes (Table 4) for diabetes at their
c Overweight (BMI .85th percentile for age and sex, weight for height .85th percentile, or initial prenatal visit, using standard diag-
weight .120% of ideal for height) nostic criteria (Table 2). Women with di-
Plus any two of the following risk factors: abetes found at this visit should receive
c Family history of type 2 diabetes in first- or second-degree relative a diagnosis of overt, not gestational,
c Race/ethnicity (Native American, African American, Latino, Asian American, Pacific diabetes.
Islander) GDM carries risks for the mother and
c Signs of insulin resistance or conditions associated with insulin resistance (acanthosis neonate. The Hyperglycemia and Ad-
nigricans, hypertension, dyslipidemia, polycystic ovary syndrome, or small-for-gestational- verse Pregnancy Outcome (HAPO) study
age birth weight) (37), a large-scale (;25,000 pregnant
c Maternal history of diabetes or GDM during the child’s gestation women) multinational epidemiological
Age of initiation: age 10 years or at onset of puberty, if puberty occurs at a younger age study, demonstrated that risk of adverse
Frequency: every 3 years maternal, fetal, and neonatal outcomes
continuously increased as a function of
*Persons aged 18 years and younger. maternal glycemia at 24–28 weeks, even
within ranges previously considered nor-
mal for pregnancy. For most complica-
clinical research studies (see, for example, GDM at 24–28 weeks of gestation, tions, there was no threshold for risk.
http://www.diabetestrialnet.org). However, using a 75-g 2-h OGTT and the di- These results have led to careful recon-
widespread clinical testing of asymptomatic agnostic cut points in Table 6. (B) sideration of the diagnostic criteria for
low-risk individuals cannot currently be c Screen women with GDM for persistent GDM. After deliberations in 2008–
recommended, as it would identify very diabetes at 6–12 weeks postpartum, 2009, the International Association of
few individuals in the general population using the OGTT and nonpregnancy Diabetes and Pregnancy Study Groups
who are at risk. Individuals who screen diagnostic criteria. (E) (IADPSG), an international consensus
positive should be counseled about their c Women with a history of GDM should group with representatives from multiple
risk of developing diabetes and symptoms have lifelong screening for the de- obstetrical and diabetes organizations,
of diabetes, followed closely to prevent de- velopment of diabetes or prediabetes at including ADA, developed revised rec-
velopment of diabetic ketoacidosis, and least every 3 years. (B) ommendations for diagnosing GDM.
informed about clinical trials. Clinical c Women with a history of GDM found The group recommended that all women
studies are being conducted to test various to have prediabetes should receive not known to have prior diabetes
methods of preventing type 1 diabetes in lifestyle interventions or metformin to undergo a 75-g OGTT at 24–28 weeks
those with evidence of autoimmunity. prevent diabetes. (A) of gestation. Additionally, the group de-
Some interventions have demonstrated veloped diagnostic cut points for the fast-
modest efficacy in slowing b-cell loss early For many years, GDM was defined as ing, 1-h, and 2-h plasma glucose
in type 1 diabetes (34,35), and further re- any degree of glucose intolerance with measurements that conveyed an odds
search is needed to determine whether onset or first recognition during preg- ratio for adverse outcomes of at least
they may be effective in preventing type nancy (13), whether or not the condition 1.75 compared with women with the
1 diabetes. persisted after pregnancy, and not ex- mean glucose levels in the HAPO study.
cluding the possibility that unrecognized Current screening and diagnostic strate-
glucose intolerance may have antedated gies, based on the IADPSG statement
III. DETECTION AND or begun concomitantly with the preg- (38), are outlined in Table 6.
DIAGNOSIS OF GDM nancy. This definition facilitated a uniform These new criteria will significantly
strategy for detection and classification of increase the prevalence of GDM, primar-
Recommendations GDM, but its limitations were recognized ily because only one abnormal value, not
c Screen for undiagnosed type 2 diabetes for many years. As the ongoing epidemic two, is sufficient to make the diagnosis.
at the first prenatal visit in those with of obesity and diabetes has led to more The ADA recognizes the anticipated sig-
risk factors, using standard diagnostic type 2 diabetes in women of childbearing nificant increase in the incidence of GDM
criteria. (B) age, the number of pregnant women with diagnosed by these criteria and is sensitive
c In pregnant women not previously undiagnosed type 2 diabetes has increased to concerns about the “medicalization” of
known to have diabetes, screen for (36). Because of this, it is reasonable to pregnancies previously categorized as nor-
mal. These diagnostic criteria changes are
being made in the context of worrisome
Table 6dScreening for and diagnosis of GDM worldwide increases in obesity and diabe-
tes rates, with the intent of optimizing ges-
Perform a 75-g OGTT, with plasma glucose measurement fasting and at 1 and 2 h, at 24–28
tational outcomes for women and their
weeks of gestation in women not previously diagnosed with overt diabetes.
babies.
The OGTT should be performed in the morning after an overnight fast of at least 8 h.
Admittedly, there are few data from
The diagnosis of GDM is made when any of the following plasma glucose values are exceeded:
randomized clinical trials regarding ther-
c Fasting: $92 mg/dL (5.1 mmol/L)
apeutic interventions in women who will
c 1 h: $180 mg/dL (10.0 mmol/L)
now be diagnosed with GDM based on
c 2 h: $153 mg/dL (8.5 mmol/L)
only one blood glucose value above the
care.diabetesjournals.org DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 S15
6. Position Statement
specified cut points (in contrast to the older IV. PREVENTION/DELAY A1C of 5.7–6.4%, IGT, or IFG should be
criteria that stipulated at least two abnor- OF TYPE 2 DIABETES counseled on lifestyle changes with goals
mal values). However, there is emerging similar to those of the DPP (7% weight
observational and retrospective evidence Recommendations loss and moderate physical activity of at
that women diagnosed with the new c Patients with IGT (A), IFG (E), or an A1C least 150 min/week). Regarding drug
criteria (even if they would not have of 5.7–6.4% (E) should be referred to an therapy for diabetes prevention, metfor-
been diagnosed with older criteria) have effective ongoing support program tar- min has a strong evidence base and dem-
increased rates of poor pregnancy out- geting weight loss of 7% of body weight onstrated long-term safety (53). For other
comes similar to those of women with and increasing physical activity to at least drugs, issues of cost, side effects, and lack
GDM by prior criteria (39,40). Expected 150 min/week of moderate activity such of persistence of effect in some studies
benefits to these pregnancies and offspring as walking. (54) require consideration. Metformin
are inferred from intervention trials that c Follow-up counseling appears to be was less effective than lifestyle modifica-
focused on women with more mild hyper- important for success. (B) tion in the DPP and DPPOS, but may be
glycemia than identified using older GDM c Based on the cost-effectiveness of diabetes cost-saving over a 10-year period (51). It
diagnostic criteria and that found modest prevention, such programs should be was as effective as lifestyle modification
benefits (41,42). The frequency of follow- covered by third-party payers. (B) in participants with a BMI of at least 35
up and blood glucose monitoring for these c Metformin therapy for prevention of kg/m2, but not significantly better than
women is not yet clear, but likely to be less type 2 diabetes may be considered in placebo than those over age 60 years
intensive than for women diagnosed by the those with IGT (A), IFG (E), or an A1C of (23). In women in the DPP with a history
older criteria. It is important to note that 5.7–6.4% (E), especially for those with of GDM, metformin and intensive lifestyle
80–90% of women in both of the mild BMI .35 kg/m2, aged ,60 years, and modification led to an equivalent 50% re-
GDM studies (whose glucose values over- women with prior GDM. (A) duction in the risk of diabetes (55). Met-
lapped with the thresholds recommended c At least annual monitoring for the de- formin therefore might reasonably be
herein) could be managed with lifestyle velopment of diabetes in those with recommended for very high-risk individ-
therapy alone. prediabetes is suggested. (E) uals (those with a history of GDM, the
The American College of Obstetri- c Screening for and treatment of modifi- very obese, and/or those with more severe
cians and Gynecologists announced in able risk factors for CVD is suggested. (B) or progressive hyperglycemia).
2011 that they continue to recommend People with prediabetes often have
use of prior diagnostic criteria for GDM RCTs have shown that individuals at other cardiovascular risk factors, such as
(43). Several other countries have high risk for developing type 2 diabetes obesity, hypertension, and dyslipidemia.
adopted the new criteria, and a report (those with IFG, IGT, or both) can signif- Assessing and treating these risk factors is
from the WHO on this topic is pending icantly decrease the rate of onset of diabetes an important aspect of reducing cardio-
at the time of publication of these stand- with particular interventions (23–29). metabolic risk. In the DPP and DPPOS,
ards. The National Institutes of Health is These include intensive lifestyle modifica- cardiovascular event rates have been very
planning to hold a consensus develop- tion programs that have been shown to be low, perhaps due to appropriate manage-
ment conference on this topic in 2013. very effective (;58% reduction after 3 ment of cardiovascular risk factors in all
Because some cases of GDM may years) and use of the pharmacological arms of the study (56).
represent pre-existing undiagnosed type agents metformin, a-glucosidase inhibi-
2 diabetes, women with a history of tors, orlistat, and thiazolidinediones, each V. DIABETES CARE
GDM should be screened for diabetes of which has been shown to decrease inci-
6–12 weeks postpartum, using nonpreg- dent diabetes to various degrees. Follow-up A. Initial evaluation
nant OGTT criteria. Because of their pre- of all three large studies of lifestyle interven- A complete medical evaluation should be
partum treatment for hyperglycemia, use tion has shown sustained reduction in the performed to classify the diabetes, detect
of the A1C for diagnosis of persistent di- rate of conversion to type 2 diabetes, with the presence of diabetes complications,
abetes at the postpartum visit is not rec- 43% reduction at 20 years in the Da Qing review previous treatment and risk factor
ommended (44). Women with a history study (47), 43% reduction at 7 years in the control in patients with established diabe-
of GDM have a greatly increased subse- Finnish Diabetes Prevention Study (DPS) tes, assist in formulating a management
quent risk for diabetes (45) and should (48), and 34% reduction at 10 years in plan, and provide a basis for continuing
be followed up with subsequent screen- the U.S. Diabetes Prevention Program care. Laboratory tests appropriate to the
ing for the development of diabetes or Outcomes Study (DPPOS) (49). A cost- evaluation of each patient’s medical condi-
prediabetes, as outlined in Section II. effectiveness model suggested that lifestyle tion should be performed. A focus on the
Lifestyle interventions or metformin interventions as delivered in the DPP are components of comprehensive care (Table
should be offered to women with a his- cost-effective (50), and actual cost data 7) will assist the health care team to ensure
tory of GDM who develop prediabetes, from the DPP and DPPOS confirm that life- optimal management of the patient with
as discussed in Section IV. In the pro- style interventions are highly cost-effective diabetes.
spective Nurses’ Health Study II, risk of (51). Group delivery of the DPP interven-
subsequent diabetes after a history of tion in community settings has the poten- B. Management
GDM was significantly lower in women tial to be significantly less expensive while People with diabetes should receive med-
who followed healthy eating patterns. still achieving similar weight loss (52). ical care from a team that may include
Adjusting for BMI moderately, but not Based on the results of clinical trials physicians, nurse practitioners, physician’s
completely, attenuated this association and the known risks of progression of assistants, nurses, dietitians, pharmacists,
(46). prediabetes to diabetes, persons with an and mental health professionals with
S16 DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 care.diabetesjournals.org
7. Position Statement
Table 7dComponents of the comprehensive diabetes evaluation and conditions, physical activity, eating
Medical history patterns, social situation and cultural fac-
c Age and characteristics of onset of diabetes (e.g., DKA, asymptomatic laboratory finding) tors, and presence of complications of di-
c Eating patterns, physical activity habits, nutritional status, and weight history; growth and abetes or other medical conditions.
development in children and adolescents
c Diabetes education history C. Glycemic control
c Review of previous treatment regimens and response to therapy (A1C records)
c Current treatment of diabetes, including medications, medication adherence and barriers
1. Assessment of glycemic control
Two primary techniques are available for
thereto, meal plan, physical activity patterns, and readiness for behavior change
c Results of glucose monitoring and patient’s use of data
health providers and patients to assess the
c DKA frequency, severity, and cause
effectiveness of the management plan on
c Hypoglycemic episodes
glycemic control: patient self-monitoring
c Hypoglycemia awareness
of blood glucose (SMBG) or interstitial
c Any severe hypoglycemia: frequency and cause
glucose, and A1C.
c History of diabetes-related complications
c Microvascular: retinopathy, nephropathy, neuropathy (sensory, including history of foot a. Glucose monitoring
lesions; autonomic, including sexual dysfunction and gastroparesis) Recommendations
c Macrovascular: CHD, cerebrovascular disease, and PAD c Patients on multiple-dose insulin (MDI)
c Other: psychosocial problems*, dental disease* or insulin pump therapy should do
Physical examination SMBG at least prior to meals and snacks,
c Height, weight, BMI occasionally postprandially, at bedtime,
c Blood pressure determination, including orthostatic measurements when indicated prior to exercise, when they suspect low
c Fundoscopic examination* blood glucose, after treating low blood
c Thyroid palpation glucose until they are normoglycemic,
c Skin examination (for acanthosis nigricans and insulin injection sites) and prior to critical tasks such as driv-
c Comprehensive foot examination ing. (B)
c Inspection c When prescribed as part of a broader
c Palpation of dorsalis pedis and posterior tibial pulses educational context, SMBG results may
c Presence/absence of patellar and Achilles reflexes be helpful to guide treatment decisions
c Determination of proprioception, vibration, and monofilament sensation and/or patient self-management for
Laboratory evaluation patients using less frequent insulin in-
c A1C, if results not available within past 2–3 months jections or noninsulin therapies. (E)
If not performed/available within past year c When prescribing SMBG, ensure that
c Fasting lipid profile, including total, LDL and HDL cholesterol and triglycerides patients receive ongoing instruction
c Liver function tests and regular evaluation of SMBG tech-
c Test for urine albumin excretion with spot urine albumin-to-creatinine ratio nique and SMBG results, as well as their
c Serum creatinine and calculated GFR ability to use SMBG data to adjust ther-
c TSH in type 1 diabetes, dyslipidemia or women over age 50 years apy. (E)
Referrals c Continuous glucose monitoring (CGM)
c Eye care professional for annual dilated eye exam in conjunction with intensive insulin
c Family planning for women of reproductive age regimens can be a useful tool to lower
c Registered dietitian for MNT A1C in selected adults (aged $25 years)
c DSME with type 1 diabetes. (A)
c Dentist for comprehensive periodontal examination c Although the evidence for A1C lower-
c Mental health professional, if needed ing is less strong in children, teens, and
younger adults, CGM may be helpful
*See appropriate referrals for these categories. in these groups. Success correlates with
adherence to ongoing use of the device.
(C)
c CGM may be a supplemental tool to
expertise and a special interest in diabetes. of problem-solving skills in the various SMBG in those with hypoglycemia
It is essential in this collaborative and in- aspects of diabetes management. Imple- unawareness and/or frequent hypogly-
tegrated team approach that individuals mentation of the management plan re- cemic episodes. (E)
with diabetes assume an active role in their quires that the goals and treatment plan
care. are individualized and take patient pref- Major clinical trials of insulin-treated
The management plan should be erences into account. The management patients that demonstrated the benefits of
formulated as a collaborative therapeutic plan should recognize diabetes self- intensive glycemic control on diabetes
alliance among the patient and family, management education (DSME) and on- complications have included SMBG as
the physician, and other members of the going diabetes support as an integral part of multifactorial interventions, sug-
health care team. A variety of strategies component of care. In developing the gesting that SMBG is a component of
and techniques should be used to provide plan, consideration should be given to effective therapy. SMBG allows patients
adequate education and development the patient’s age, school or work schedule to evaluate their individual response to
care.diabetesjournals.org DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 S17
8. Position Statement
therapy and assess whether glycemic tar- one study of insulin-naïve patients with and for the burgeoning work on “artificial
gets are being achieved. Results of SMBG suboptimal initial glycemic control, use pancreas” systems.
can be useful in preventing hypoglycemia of structured SMBG (a paper tool to col-
and adjusting medications (particularly lect and interpret 7-point SMBG profiles b. A1C
prandial insulin doses), medical nutrition over 3 days at least quarterly) reduced Recommendations
therapy (MNT), and physical activity. A1C by 0.3% more than in an active con- c Perform the A1C test at least two
The frequency and timing of SMBG trol group (65). Patients should be taught times a year in patients who are meet-
should be dictated by the particular needs how to use SMBG data to adjust food in- ing treatment goals (and who have
and goals of the patient. SMBG is espe- take, exercise, or pharmacological therapy stable glycemic control). (E)
cially important for patients treated with to achieve specific goals, and the ongoing c Perform the A1C test quarterly in pa-
insulin to monitor for and prevent asymp- need for and frequency of SMBG should be tients whose therapy has changed or
tomatic hypoglycemia and hyperglyce- re-evaluated at each routine visit. who are not meeting glycemic goals. (E)
mia. Most patients with type 1 diabetes Real-time CGM through the measure- c Use of POC testing for A1C provides
and others on intensive insulin regimens ment of interstitial glucose (which corre- the opportunity for more timely treat-
(MDI or insulin pump therapy) should do lates well with plasma glucose) is available. ment changes. (E)
SMBG at least prior to meals and snacks, These sensors require calibration with
occasionally postprandially, at bedtime, SMBG, and the latter are still recommended Because A1C is thought to reflect aver-
prior to exercise, when they suspect low for making acute treatment decisions. age glycemia over several months (63) and
blood glucose, after treating low blood CGM devices have alarms for hypo- and has strong predictive value for diabetes
glucose until they are normoglycemic, hyperglycemic excursions. A 26-week ran- complications (71,72), A1C testing
and prior to critical tasks such as driving. domized trial of 322 type 1 diabetic pa- should be performed routinely in all pa-
For many patients, this will require test- tients showed that adults aged $25 years tients with diabetes, at initial assessment
ing 6–8 times daily, although individual using intensive insulin therapy and CGM and then as part of continuing care. Mea-
needs may be greater. Although there are experienced a 0.5% reduction in A1C surement approximately every 3 months
few rigorous studies, a database study of (from ;7.6 to 7.1%) compared with usual determines whether patient’s glycemic tar-
almost 27,000 children and adolescents intensive insulin therapy with SMBG (66). gets have been reached and maintained.
with type 1 diabetes showed that, after Sensor use in children, teens, and adults to For any individual patient, the frequency
adjustment for multiple confounders, in- age 24 years did not result in significant of A1C testing should be dependent on
creased daily frequency of SMBG was sig- A1C lowering, and there was no significant the clinical situation, the treatment regimen
nificantly associated with lower A1C difference in hypoglycemia in any group. used, and the judgment of the clinician.
(20.2% per additional test per day, level- Importantly, the greatest predictor of A1C Some patients with stable glycemia well
ing off at five tests per day) and with fewer lowering in this study for all age-groups within target may do well with testing
acute complications (57). The optimal was frequency of sensor use, which was only twice per year, while unstable or
frequency of SMBG for patients on non- lower in younger age-groups. In a smaller highly intensively managed patients (e.g.,
intensive regimens, such as those with RCT of 129 adults and children with base- pregnant type 1 diabetic women) may be
type 2 diabetes on basal insulin, is not line A1C ,7.0%, outcomes combining tested more frequently than every 3
known, although a number of studies A1C and hypoglycemia favored the group months. The availability of the A1C result
have used fasting SMBG for patient or pro- utilizing CGM, suggesting that CGM is also at the time that the patient is seen (POC
vider titration of the basal insulin dose. beneficial for individuals with type 1 dia- testing) has been reported in small studies
The evidence base for SMBG for patients betes who have already achieved excellent to result in increased intensification of ther-
with type 2 diabetes on noninsulin therapy control (67). apy and improvement in glycemic control
is somewhat mixed. Several randomized A trial comparing CGM plus insulin (73,74). However, two recent systematic
trials have called into question the clinical pump to SMBG plus multiple injections reviews and meta-analyses found no signif-
utility and cost-effectiveness of routine of insulin in adults and children with type icant difference in A1C between POC and
SMBG in non–insulin-treated patients 1 diabetes showed significantly greater laboratory A1C usage (75,76).
(58–60). A recent meta-analysis suggested improvements in A1C with “sensor- The A1C test is subject to certain
that SMBG reduced A1C by 0.25% at 6 augmented pump” therapy (68,69), but limitations. Conditions that affect eryth-
months (61), while a Cochrane review con- this trial did not isolate the effect of CGM rocyte turnover (hemolysis, blood loss)
cluded that the overall effect of SMBG in itself. Overall, meta-analyses suggest that and hemoglobin variants must be consid-
such patients is small up to 6 months after compared with SMBG, CGM lowers A1C ered, particularly when the A1C result
initiation and subsides after 12 months (62). by ;0.26% (70). Altogether, these data does not correlate with the patient’s clin-
Because the accuracy of SMBG is suggest that, in appropriately selected pa- ical situation (63). In addition, A1C does
instrument and user dependent (63), it tients who are motivated to wear it most of not provide a measure of glycemic vari-
is important to evaluate each patient’s the time, CGM reduces A1C. The technol- ability or hypoglycemia. For patients
monitoring technique, both initially and ogy may be particularly useful in those with prone to glycemic variability (especially
at regular intervals thereafter. Optimal hypoglycemia unawareness and/or fre- type 1 diabetic patients or type 2 diabetic
use of SMBG requires proper review and quent episodes of hypoglycemia, although patients with severe insulin deficiency),
interpretation of the data, both by the pa- studies as yet have not shown significant glycemic control is best judged by the
tient and provider. Among patients who reductions in severe hypoglycemia (70). combination of results of self-monitoring
checked their blood glucose at least once CGM forms the underpinning for the de- and the A1C. The A1C may also serve as a
daily, many reported taking no action velopment of pumps that suspend insulin check on the accuracy of the patient’s me-
when results were high or low (64). In delivery when hypoglycemia is developing ter (or the patient’s reported SMBG
S18 DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 care.diabetesjournals.org
9. Position Statement
results) and the adequacy of the SMBG there are significant differences in how microvascular (retinopathy and nephro-
testing schedule. A1C relates to average glucose in children pathy) and neuropathic complications.
Table 8 contains the correlation be- or in African American patients is an area Follow-up of the DCCT cohorts in the Ep-
tween A1C levels and mean plasma glu- for further study. For the time being, the idemiology of Diabetes Interventions and
cose levels based on data from the question has not led to different recom- Complications (EDIC) study (80,81) dem-
international A1C-Derived Average Glu- mendations about testing A1C or to dif- onstrated persistence of these microvascu-
cose (ADAG) trial utilizing frequent ferent interpretations of the clinical lar benefits in previously intensively treated
SMBG and CGM in 507 adults (83% Cau- meaning of given levels of A1C in those subjects, even though their glycemic con-
casian) with type 1, type 2, and no diabe- populations. trol approximated that of previous stan-
tes (77). The ADA and the American For patients in whom A1C/eAG and dard arm subjects during follow-up.
Association for Clinical Chemistry have measured blood glucose appear discrep- The Kumamoto Study (82) and UK
determined that the correlation (r 5 ant, clinicians should consider the possi- Prospective Diabetes Study (UKPDS)
0.92) is strong enough to justify reporting bilities of hemoglobinopathy or altered (83,84) confirmed that intensive glycemic
both an A1C result and an estimated av- red cell turnover, and the options of more control was associated with significantly
erage glucose (eAG) result when a clini- frequent and/or different timing of SMBG decreased rates of microvascular and neu-
cian orders the A1C test. The table in pre- or use of CGM. Other measures of chronic ropathic complications in patients with
2009 versions of the “Standards of Medi- glycemia such as fructosamine are avail- type 2 diabetes. Long-term follow-up of
cal Care in Diabetes” describing the cor- able, but their linkage to average glucose the UKPDS cohorts showed persistence of
relation between A1C and mean glucose and their prognostic significance are not the effect of early glycemic control on
was derived from relatively sparse data as clear as is the case for A1C. most microvascular complications (85).
(one 7-point profile over 1 day per A1C Subsequent trials in patients with
reading) in the primarily Caucasian type 1 2. Glycemic goals in adults more long-standing type 2 diabetes, de-
diabetic participants in the DCCT (78). Recommendations signed primarily to look at the role of
Clinicians should note that the numbers c Lowering A1C to below or around 7% intensive glycemic control on cardiovas-
in the table are now different, as they are has been shown to reduce microvas- cular outcomes, also confirmed a benefit,
based on ;2,800 readings per A1C in the cular complications of diabetes and if although more modest, on onset or pro-
ADAG trial. implemented soon after the diagnosis gression of microvascular complications.
In the ADAG trial, there were no sig- of diabetes is associated with long-term The Veterans Affairs Diabetes Trial
nificant differences among racial and eth- reduction in macrovascular disease. (VADT) showed significant reductions
nic groups in the regression lines between Therefore, a reasonable A1C goal for in albuminuria with intensive (achieved
A1C and mean glucose, although there many nonpregnant adults is ,7%. (B) median A1C 6.9%) compared with stan-
was a trend toward a difference between c Providers might reasonably suggest dard glycemic control, but no difference
African/African American participants more stringent A1C goals (such as in retinopathy and neuropathy (86,87).
and Caucasian ones. A small study com- ,6.5%) for selected individual pa- The Action in Diabetes and Vascular Dis-
paring A1C to CGM data in type 1 di- tients, if this can be achieved without ease: Preterax and Diamicron MR Con-
abetic children found a highly statistically significant hypoglycemia or other ad- trolled Evaluation (ADVANCE) study of
significant correlation between A1C and verse effects of treatment. Appropriate intensive versus standard glycemic con-
mean blood glucose, although the corre- patients might include those with short trol in type 2 diabetes found a statistically
lation (r 5 0.7) was significantly lower duration of diabetes, long life expec- significant reduction in albuminuria, but
than in the ADAG trial (79). Whether tancy, and no significant CVD. (C) not in neuropathy or retinopathy, with an
c Less stringent A1C goals (such as A1C target of ,6.5% (achieved median
Table 8dCorrelation of A1C with average
,8%) may be appropriate for patients A1C 6.3%) compared with standard ther-
glucose
with a history of severe hypoglycemia, apy achieving a median A1C of 7.0% (88).
limited life expectancy, advanced mi- Analyses from the Action to Control Car-
crovascular or macrovascular complica- diovascular Risk in Diabetes (ACCORD)
Mean plasma glucose tions, extensive comorbid conditions, trial have shown lower rates of onset or
A1C (%) mg/dL mmol/L and those with long-standing diabetes in progression of early-stage microvascular
whom the general goal is difficult to at- complications in the intensive glycemic
6 126 7.0 tain despite DSME, appropriate glucose control arm compared with the standard
7 154 8.6 monitoring, and effective doses of mul- arm (89,90).
8 183 10.2 tiple glucose-lowering agents including Epidemiological analyses of the DCCT
9 212 11.8 insulin. (B) and UKPDS (71,72) demonstrate a curvi-
10 240 13.4 linear relationship between A1C and mi-
11 269 14.9 Hyperglycemia defines diabetes, and crovascular complications. Such analyses
12 298 16.5 glycemic control is fundamental to the suggest that, on a population level, the
These estimates are based on ADAG data of ;2,700 management of diabetes. The DCCT greatest number of complications will be
glucose measurements over 3 months per A1C (71), a prospective RCT of intensive ver- averted by taking patients from very poor
measurement in 507 adults with type 1, type 2, and sus standard glycemic control in patients control to fair or good control. These anal-
no diabetes. The correlation between A1C and av- with relatively recently diagnosed type 1 yses also suggest that further lowering of
erage glucose was 0.92 (ref. 77). A calculator for
converting A1C results into eAG, in either mg/dL or
diabetes, showed definitively that A1C from 7 to 6% is associated with further
mmol/L, is available at http://professional.diabetes improved glycemic control is associ- reduction in the risk of microvascular
.org/eAG. ated with significantly decreased rates of complications, albeit the absolute risk
care.diabetesjournals.org DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 S19
10. Position Statement
reductions become much smaller. Given All three of these trials were conducted in was associated with excess mortality in
the substantially increased risk of hypogly- participants with more long-standing di- either arm, but the association was stron-
cemia (particularly in those with type 1 di- abetes (mean duration 8–11 years) and ger in those randomized to the standard
abetes, but also in the recent type 2 diabetes either known CVD or multiple cardiovas- glycemic control arm (97). Unlike the
trials), the concerning mortality findings in cular risk factors. Details of these three case with the DCCT trial, where lower
the ACCORD trial (91), and the relatively studies are reviewed extensively in an achieved A1C levels were related to sig-
much greater effort required to achieve ADA position statement (94). nificantly increased rates of severe hypo-
near-normoglycemia, the risks of lower gly- The ACCORD study enrolled partici- glycemia, in ACCORD every 1% decline
cemic targets may outweigh the potential pants with either known CVD or two or in A1C from baseline to 4 months into the
benefits on microvascular complications more major cardiovascular risk factors trial was associated with a significant de-
on a population level. However, selected and randomized them to intensive glyce- crease in the rate of severe hypoglycemia
individual patients, especially those with mic control (goal A1C ,6%) or standard in both arms (96).
little comorbidity and long life expectancy glycemic control (goal A1C 7–8%). The The primary outcome of ADVANCE
(who may reap the benefits of further low- glycemic control comparison was halted was a combination of microvascular
ering of glycemia below 7%), may, based early due to the finding of an increased events (nephropathy and retinopathy)
on provider judgment and patient prefer- rate of mortality in the intensive arm com- and major adverse cardiovascular events
ences, adopt more intensive glycemic tar- pared with the standard arm (1.41% vs. (MI, stroke, and cardiovascular death).
gets (e.g., an A1C target ,6.5%) as long as 1.14% per year; HR 1.22; 95% CI 1.01– Intensive glycemic control (to a goal A1C
significant hypoglycemia does not become 1.46), with a similar increase in cardiovas- ,6.5% vs. treatment to local standards)
a barrier. cular deaths. This increase in mortality in significantly reduced the primary end
CVD, a more common cause of death the intensive glycemic control arm was point. However, this was due to a significant
in populations with diabetes than micro- seen in all prespecified patient subgroups. reduction in the microvascular outcome, pri-
vascular complications, is less clearly The primary outcome of ACCORD (non- marily development of macroalbuminuria,
impacted by levels of hyperglycemia or fatal MI, nonfatal stroke, or cardiovascu- with no significant reduction in the macro-
the intensity of glycemic control. In the lar death) was nonsignificantly lower in vascular outcome. There was no difference
DCCT, there was a trend toward lower the intensive glycemic control group in overall or cardiovascular mortality be-
risk of CVD events with intensive control, due to a reduction in nonfatal MI, both tween the intensive compared with the
and in 9-year post-DCCT follow-up of the when the glycemic control comparison standard glycemic control arms (88).
EDIC cohort participants previously ran- was halted and all participants transi- The VADT randomized participants
domized to the intensive arm had a sig- tioned to the standard glycemic control with type 2 diabetes uncontrolled on
nificant 57% reduction in the risk of intervention (91), and at completion of insulin or maximal-dose oral agents (me-
nonfatal myocardial infarction (MI), stroke, the planned follow-up (95). dian entry A1C 9.4%) to a strategy of
or CVD death compared with those pre- Exploratory analyses of the mortality intensive glycemic control (goal A1C
viously in the standard arm (92). The ben- findings of ACCORD (evaluating vari- ,6.0%) or standard glycemic control,
efit of intensive glycemic control in this ables including weight gain, use of any with a planned A1C separation of at least
type 1 diabetic cohort has recently been specific drug or drug combination, and 1.5%. The primary outcome of the VADT
shown to persist for several decades (93). hypoglycemia) were reportedly unable to was a composite of CVD events. The cu-
In type 2 diabetes, there is evidence identify a clear explanation for the excess mulative primary outcome was nonsig-
that more intensive treatment of glycemia mortality in the intensive arm (91). The nificantly lower in the intensive arm
in newly diagnosed patients may reduce ACCORD investigators subsequently (86). An ancillary study of the VADT
long-term CVD rates. During the UKPDS published additional epidemiological demonstrated that intensive glycemic
trial, there was a 16% reduction in car- analyses showing no increase in mortality control significantly reduced the primary
diovascular events (combined fatal or in the intensive arm participants who CVD outcome in individuals with less
nonfatal MI and sudden death) in the achieved A1C levels below 7% nor in atherosclerosis at baseline (assessed by
intensive glycemic control arm that did those who lowered their A1C quickly af- coronary calcium) but not in persons
not reach statistical significance (P 5 ter trial enrollment. In fact, although there with more extensive baseline atheroscle-
0.052), and there was no suggestion of was no A1C level at which intensive arm rosis (98). A post hoc analysis showed a
benefit on other CVD outcomes such as participants had significantly lower mor- complex relationship between duration
stroke. However, after 10 years of follow- tality than standard arm participants, the of diabetes before glycemic intensifica-
up, those originally randomized to inten- highest risk for mortality was observed in tion and mortality: mortality in the inten-
sive glycemic control had significant intensive arm participants with the high- sive vs. standard glycemic control arm
long-term reductions in MI (15% with est A1C levels (96). was inversely related to duration of dia-
sulfonylurea or insulin as initial pharma- The role of hypoglycemia in the ex- betes at the time of study enrollment.
cotherapy, 33% with metformin as initial cess mortality findings was also complex. Those with diabetes duration less than
pharmacotherapy) and in all-cause mor- Severe hypoglycemia was significantly 15 years had a mortality benefit in the in-
tality (13% and 27%, respectively) (85). more likely in participants randomized tensive arm, while those with duration of
Three more recent large trials to the intensive glycemic control arm. 20 years or more had higher mortality in
(ACCORD, ADVANCE, and VADT) sug- However, excess mortality in the inten- the intensive arm (99).
gested no significant reduction in CVD sive versus standard arms was only sig- The evidence for a cardiovascular ben-
outcomes with intensive glycemic control nificant for participants with no severe efit of intensive glycemic control primarily
in participants who had more advanced hypoglycemia, and not for those with one rests on long-term follow-up of study
type 2 diabetes than UKPDS participants. or more episodes. Severe hypoglycemia cohorts treated early in the course of type
S20 DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 care.diabetesjournals.org
11. Position Statement
1 and type 2 diabetes and subset analyses of The issue of pre- versus postprandial c preprandial: #95 mg/dL (5.3 mmol/L),
ACCORD, ADVANCE, and VADT. A SMBG targets is complex (102). Elevated and either:
group-level meta-analysis of the latter three postchallenge (2-h OGTT) glucose values c 1-h postmeal: #140 mg/dL (7.8 mmol/L)
trials suggests that glucose lowering has a have been associated with increased car- or
modest (9%) but statistically significant diovascular risk independent of FPG in c 2-h postmeal: #120 mg/dL (6.7 mmol/L)
reduction in major CVD outcomes, pri- some epidemiological studies. In diabetic
marily nonfatal MI, with no significant subjects, some surrogate measures of For women with pre-existing type 1
effect on mortality. However, heterogeneity vascular pathology, such as endothelial or type 2 diabetes who become pregnant, a
of the mortality effects across studies was dysfunction, are negatively affected by recent consensus statement (106) recom-
noted, precluding firm summary measures postprandial hyperglycemia (103). It is mended the following as optimal glycemic
of the mortality effects. A prespecified sub- clear that postprandial hyperglycemia, goals, if they can be achieved without ex-
group analysis suggested that major CVD like preprandial hyperglycemia, contrib- cessive hypoglycemia:
outcome reduction occurred in patients utes to elevated A1C levels, with its rela-
without known CVD at baseline (HR 0.84, tive contribution being higher at A1C c premeal, bedtime, and overnight glu-
95% CI 0.74–0.94) (100). Conversely, the levels that are closer to 7%. However, cose 60–99 mg/dL (3.3–5.4 mmol/L)
mortality findings in ACCORD and sub- outcome studies have clearly shown c peak postprandial glucose 100–129
group analyses of the VADT suggest that A1C to be the primary predictor of com- mg/dL (5.4–7.1 mmol/L)
the potential risks of intensive glycemic plications, and landmark glycemic con- c A1C ,6.0%
control may outweigh its benefits in some trol trials such as the DCCT and UKPDS
patients, such as those with very long du- relied overwhelmingly on preprandial
ration of diabetes, known history of severe SMBG. Additionally, an RCT in patients D. Pharmacological and overall
hypoglycemia, advanced atherosclerosis, with known CVD found no CVD benefit approaches to treatment
and advanced age/frailty. Certainly, provid- of insulin regimens targeting postpran-
ers should be vigilant in preventing severe dial glucose compared with those 1. Insulin therapy for type 1 diabetes
hypoglycemia in patients with advanced targeting preprandial glucose (104). A Recommendations
disease and should not aggressively at- reasonable recommendation for post- c Most people with type 1 diabetes should
tempt to achieve near-normal A1C levels prandial testing and targets is that for in- be treated with MDI injections (three
in patients in whom such a target cannot dividuals who have premeal glucose to four injections per day of basal and
be safely and reasonably easily achieved. values within target but have A1C values prandial insulin) or continuous sub-
Severe or frequent hypoglycemia is an ab- above target, monitoring postprandial cutaneous insulin infusion (CSII). (A)
solute indication for the modification of plasma glucose (PPG) 1–2 h after the start c Most people with type 1 diabetes
treatment regimens, including setting of the meal and treatment aimed at reduc- should be educated in how to match
higher glycemic goals. Many factors, in- ing PPG values to ,180 mg/dL may help prandial insulin dose to carbohydrate
cluding patient preferences, should be lower A1C. intake, premeal blood glucose, and
taken into account when developing a pa- Glycemic goals for children are pro- anticipated activity. (E)
tient’s individualized goals (101). vided in Section VIII.A.1.a. As regards goals c Most people with type 1 diabetes
Recommended glycemic goals for for glycemic control for women with should use insulin analogs to reduce
many nonpregnant adults are shown in GDM, recommendations from the Fifth hypoglycemia risk. (A)
Table 9. The recommendations are based International Workshop-Conference on c Consider screening those with type 1
on those for A1C values, with listed blood Gestational Diabetes Mellitus (105) were diabetes for other autoimmune dis-
glucose levels that appear to correlate to target maternal capillary glucose con- eases (thyroid, vitamin B12 deficiency,
with achievement of an A1C of ,7%. centrations of: celiac) as appropriate. (B)
The DCCT clearly showed that inten-
Table 9dSummary of glycemic recommendations for many nonpregnant adults with diabetes sive insulin therapy (three or more in-
jections per day of insulin, CSII, or insulin
A1C ,7.0%* pump therapy) was a key part of im-
Preprandial capillary plasma glucose 70–130 mg/dL* (3.9–7.2 mmol/L) proved glycemia and better outcomes
Peak postprandial capillary plasma glucose† ,180 mg/dL* (,10.0 mmol/L) (71,92). At the time of the study, therapy
c *Goals should be individualized based on:
was carried out with short- and intermedi-
c duration of diabetes
ate-acting human insulins. Despite better
c age/life expectancy
microvascular outcomes, intensive insulin
c comorbid conditions
therapy was associated with a high rate in
c known CVD or advanced microvascular complications
severe hypoglycemia (62 episodes per 100
c hypoglycemia unawareness
patient-years of therapy). Since the time of
c individual patient considerations
the DCCT, a number of rapid-acting and
c More or less stringent glycemic goals may be appropriate
long-acting insulin analogs have been de-
for individual patients veloped. These analogs are associated with
c Postprandial glucose may be targeted if A1C goals are
less hypoglycemia with equal A1C lower-
not met despite reaching preprandial glucose goals ing in type 1 diabetes (107,108).
†Postprandial glucose measurements should be made 1–2 h after the beginning of the meal, generally peak Recommended therapy for type 1 di-
levels in patients with diabetes. abetes consists of the following components:
care.diabetesjournals.org DIABETES CARE, VOLUME 36, SUPPLEMENT 1, JANUARY 2013 S21