NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Norris SL, Kansagara D, Bougatsos C, et al. Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2008 Jun. (Evidence Syntheses, No. 61.)

Cover of Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force

Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet].

Show details


Scope and Purpose

The objective of this systematic review is to examine the evidence for the potential benefits and harms of screening adults over the age of 20 years for type 2 diabetes mellitus (DM2), and for impaired fasting glucose (IFG) and/or and impaired glucose tolerance (IGT) (prediabetes) in primary care settings in the United States (US). The evidence presented will be used by the US Preventive Services Task Force (USPSTF) to formulate clinical practice recommendations.

Definition of Diabetes

Diabetes mellitus is a group of metabolic diseases characterized by hyperglycemia resulting from defects in insulin secretion, insulin action, or both.1 DM2, previously called non-insulin-dependent diabetes mellitus (NIDDM) or adult-onset diabetes, accounts for 90% to 95% of all diagnosed cases of diabetes. DM2 encompasses individuals who have insulin resistance as well as defective insulin secretion such that insulin levels are insufficient to compensate for the insulin resistance (i.e., a relative, rather than absolute, insulin deficiency).1

There is an intermediate group of persons who do not fulfill the definition of DM2, but who do not have normoglycemia. These persons have IFG [fasting plasma glucose (FPG) levels ≥100 mg/dl (5.6 mmol/l) but <126 mg/dl (7.0 mmol/l)] or IGT [2-h values in the 75-gm oral glucose tolerance test (OGTT) of ≥140 mg/dl (7.8 mmol/l) and <200 mg/dl (11.1 mmol/l)]. Persons with IFG and/or IGT are referred to as having prediabetes. (See Appendix A1 for diabetes definitions, and Appendix A2 for abbreviations referenced in this report.)

Prevalence and Burden of Disease

Diabetes poses a tremendous clinical and public health burden for Americans. Data from the National Health and Examination Survey (NHANES) indicated that 19.3 million Americans (9.3% of the total US population) 20 years of age and older had diabetes in 2002, one third of whom were undiagnosed.2 An additional 26.0% had IFG. The prevalence of diagnosed diabetes rose from 5.1% in 1988-1994 to 6.5% in 1999-2002,2 and is increasing most rapidly among individuals with a body mass index (BMI) of ≥ 35 kg/m.2, 3 The prevalence of diabetes (diagnosed and undiagnosed) rises with age, reaching 21.6% for those aged 65 years of age or more. Other factors may play a role in the increasing diabetes prevalence, including reductions in physical activity, dietary changes, an increase in survival, or more frequent diagnosis.3 African Americans, Hispanic/Latino Americans, American Indians, and some Asian Americans and Native Hawaiians or other Pacific Islanders are at particularly high risk for DM2.4 The prevalence of diagnosed diabetes is twice as high in non-Hispanic blacks and Mexican Americans compared with non-Hispanic whites.2

Diabetes was the sixth leading cause of death listed on US death certificates in 2000, and diabetes is likely to be underreported as a cause of death.4 Overall, the risk for death among people with diabetes is about twice that of people without diabetes. Adults with diabetes have rates of stroke and death from heart disease that are about 2 to 4 times higher than adults without diabetes. Diabetes is the leading cause of new cases of blindness among adults aged 20–74 years and the leading cause of end-stage renal disease, accounting for 44% of new cases. More than 60% of nontraumatic lower-limb amputations occur among people with diabetes.4

The estimated total costs of diabetes in the US in 2002 were $132 billion, of which $92 billion were direct medical costs. Indirect costs such as those due to disability, work absenteeism and premature mortality are estimated at $40 billion.4

Etiology and Natural History of Diabetes

The specific etiologies of DM2 are not known; however, the disease is associated with older age, obesity, family history of diabetes, history of gestational diabetes, impaired glucose metabolism, physical inactivity, and race/ethnicity. Both genetic susceptibility and environmental factors likely contribute to the development of DM2. Insulin resistance and beta-cell dysfunction (i.e., the inability of the pancreas to secrete sufficient insulin in response to glucose levels) are both implicit in the pathogenesis of the disease.5 The process of glycemic dysregulation typically begins long before symptoms develop. It is estimated that, on average, persons with clinically diagnosed diabetes will have lost up to 50% of their beta cell mass by the time of diagnosis.6

The natural history of diabetes and prediabetes may proceed through different pathways, with differing rates of progression from normoglycemia through IFG, IGT, to DM2.7, 8 This progression occurs over many years; by 20 years of follow-up of a normoglycemic cohort, 71% had developed IGT and 39% IFG. Metabolic data also suggest that there are important differences between IFG and IGT, and there is some evidence that IGT may be a stronger predictor of cardiovascular complications than IFG.9, 10 Persons with prediabetes have a 20 to 30% risk for development of DM2 over 5 to 10 years.7, 11 Some persons with IGT can revert to normoglycemia.12 It is unclear if the rate of decline in beta cell function is linear or the same for the progression of prediabetes to diabetes and for undiagnosed DM2 to clinical presentation.13

DM2 often goes undiagnosed for many years because the hyperglycemia develops gradually and may not produce symptoms.3, 14 However, such patients are at increased risk of developing microvascular and macrovacular complications. The prevalence of advanced microvascular complications such as proliferative retinopathy is relatively low at clinical diagnosis and duration of diabetes and degree of hyperglycemia are associated with increasing risk of these complications.1518 The rate of progression to retinopathy, neuropathy, and microalbuminuria is likely accelerated in those with increased age at diagnosis.19

The epidemiology of macrovascular complications differs from that of microvascular complications: cardiovascular morbidity and mortality are substantially elevated well before diagnosis of diabetes and are also elevated in persons with prediabetes and newly-diagnosed diabetes.2028 A substantial proportion of persons presenting with a new cardiovascular event have undiagnosed diabetes or prediabetes.20, 2933 Though there is good evidence linking chronic hyperglycemia to microvascular complications, the relationship between degree of hyperglycemia and macrovascular complications is less clear. Several recent observational studies and a meta-analysis do suggest a relationship between chronic hyperglycemia and cardiovascular disease and stroke, both in patients with and without known diabetes.3437

Rationale for Screening and Screening Strategies

For screening to be effective in decreasing the complications and mortality from DM2, there must be: 1) a detectable preclinical period; 2) valid and reliable screening tests to detect the disease during that period; and 3) effective treatments for diabetes or related medical conditions during the preclinical phase that reduce morbidity and mortality compared to treatments starting at the time of clinical (symptomatic) diagnosis. Treatments may be different for persons with and without DM2, so that knowledge of diabetes would prompt a change in clinical management, for example, use of a different medication or a different treatment target.

Diabetes has a long preclinical phase, estimated at between 10 and 12 years based on the progression of microvascular complications.38 There are currently valid and reliable tests for screening for DM2. The American Diabetes Association (ADA) recommends a FPG test, repeated in the absence of symptoms.1 The specificity of a single FPG with a cut-point of 126 mg/dl is > 95% and the sensitivity about 50% (lower for older adults), when compared to a 2-hour OGTT.39

As Harris and colleagues described in the prior evidence review for the USPSTF,40 screening is justified if it offers incremental benefits beyond the level of effectiveness of usual care at the time of clinical presentation (see Figure 1). If treatments are started at the time of screening diagnosis, do they reduce the incidence of complications (Line C) below that which would likely occur if treatment commenced with clinical presentation (Line B)? The vertical difference between lines B and C is the reduction in incidence of complications achieved by starting treatment with screening rather than later with clinical diagnosis and treatment. The harms and economic costs of screening and treatment must be small enough so that they do not outweigh the benefits of earlier treatment of screen-detected persons.

Figure 1. The “Delta Question” in Screening for Type 2 Diabetes.


Figure 1. The “Delta Question” in Screening for Type 2 Diabetes. *Reprinted from Harris RP, Lux LJ, Bunton AJ, Sutton SF, Lohr KN, Donahue KP, et al. Screening for Type 2 Diabetes Mellitus. (Prepared by RTI International Evidence-based (more...)

In addition to the necessity for a long preclinical phase, a valid screening test, and effective treatments for screened positive persons, a screening program must be feasible. Feasibility is determined by a number of factors: acceptability of the program to potential screenees; access to health services and appropriate treatment for persons who screen positive; cost-effectiveness; and the yield of cases. We will not address acceptability and access in this report, but will briefly address cost-effectiveness, as described in modeling studies.

Yield is the number of cases detected by a screening program. This includes positive predictive value (the probability that a person actually has the disease given that he or she screens positive) and negative predictive value. Predictive value depends on factors that determine the validity of the test as well as the prevalence of undiagnosed disease in screened populations. As the number of risk factors for DM2 (and thus the prevalence of undiagnosed disease) increases, the yield of screening for DM2 will increase. Screening can be targeted (selective) when directed at individuals with a high prevalence of risk factors; opportunistic when screening persons at provider visits; or universal (mass) screening when an entire population is screened.41

Re-Screening Intervals

Subsidiary Question 1. What are the yields (accuracy and reliability) of different re-screening intervals among persons with an initial normal fasting glucose?

We identified only one study which directly examined re-screening intervals,42 in addition to several modeling studies.4345 A fair-quality, longitudinal cohort study42 followed annual fasting serum glucose levels in healthy, community-based volunteers over 65 years of age for up to 18 years (n = 299) (see Appendix Table B1). Of subjects without diabetes at baseline, 1.3% developed DM2 over the follow-up period. Fasting glucose decreased over time in most participants, and in 16% of subjects the rate of decrease was significant (p<0.05); in only 3% was the rate of increase significant. None of the subjects over the age of 75 years at baseline (n=68) developed diabetes or had a significantly positive slope. The authors concluded that it is not necessary to screen non-obese persons (excluding minorities) over 65 years of age who have a baseline fasting glucose of less than 100 mg/dl, and it is not necessary to screen persons over age 75 years every 3 years. This study involved a group of healthy and health-conscious Caucasian participants, and is not likely to be applicable to broader populations. In addition, half of the original cohort was lost to follow-up.

Several modeling studies have examined screening intervals. In a Markov model, Chen and colleagues43 found that the number of quality-adjusted life-years (QALYs) gained was similar with screening intervals of 2 and 5 years, but the 5-year screening interval was more cost-effective (incremental cost per QALY $10,531 compared with $17,833) due to the higher costs of screening more frequently. A simulation of alternative DM2 screening intervals (1, 3, and 5 years) and random glucose cut-off levels (100, 130, and 160 mg/dl) for the US population aged 45 to 74 years44 found that screening every 3 years with a random glucose cut-off of 130mg/dl provided optimal yield and minimized false-positive test results and screening costs.

For groups in whom DM2 screening is recommended, the frequency with which that screening should occur is unclear. Screening frequency is dependent on the rate of rise of blood glucose over time, and data are sparse on this progression and how it may vary across the age spectrum, between sexes, and among different races or ethnic groups. Screening interval could be contingent on the results of the first screen, as suggested by Waugh and colleagues.13 The ADA recommends screening every 3 years if the test is normal46 based on expert opinion and the rationale that false negative results will be repeated before substantial time has elapsed.

A1c Screening Test Subsidiary Question 2. What is the yield (accuracy, reliability, and prevalence) of screening for type 2 diabetes with A1c?

The OGTT diabetes screening tool has been in use for many years and has served as a gold standard for diabetes diagnosis in a number of large epidemiological studies, but it is cumbersome to perform and is no longer recommended for routine clinical use by groups such as the ADA.2 FPG is a commonly performed screening test, but the stipulation of fasting introduces possible barriers to use in clinical settings. Moreover, FPG may not reliably identify those with post-prandial hyperglycemia.9, 4749 Therefore, there has been significant interest in evaluating A1c as a potential screening tool,5065 (see Appendix Table B2) as A1c correlates with glucose intolerance as defined by OGTT results, does not require fasting, and is relatively easy to perform in the primary care setting. A1c levels predict microvascular complications in persons with DM2 and may also predict macrovascular complications in those with and without diabetes across a range of A1c values.1518, 36, 37, 66 In the past, the utility of A1c as a screening tool was limited in part by its relatively poor reproducibility and the lack of standardization across labs. More recently, there has been widespread adoption of standardized A1c measurements, as newer techniques for measurement are generally highly reproducible across a wide range of A1c values, though inter-individual biologic variability is present.6769

A fair-quality systematic review in 1996 found that an A1c cutoff of 6.4% was 66% sensitive, 98% specific, and was associated with a positive predictive value of 63% in a population with a diabetes prevalence of 6%.61 Increasing the cutoff to 7% increased the positive predictive value to 90%. The authors argued that an A1c cutoff of 7% was reasonable since it was associated with low false positive rates and because values higher than this would generally prompt consideration of pharmacologic treatment, while the clinical approach to lower values would focus mainly on lifestyle modification. Because this review is older, the included studies do suffer from the potential for variability from lack of standardization of A1c assay methodology across studies.

A recent good-quality systematic review examined studies through 2004 that compared the operating characteristics of A1c and FPG in detecting diabetes and prediabetes as defined by OGTT results according to World Health Organization (WHO) criteria.51 The review found that FPG and A1c were similarly effective in detecting diabetes, but both had low sensitivity (about 50%) for detection of IGT. Though there were a variety of different cutpoints examined, many studies found that the optimum Diabetes Control and Complications Trial (DCCT) -aligned A1c cut-point was ≥ 6.1 – 6.2%, with corresponding sensitivities 43–81% and specificities 79–99%. We identified 9 studies published since, or excluded from, this review examining the utility of A1c as a screening test for DM2 with results also suggesting moderate sensitivity and high specificity of A1c values in a comparable borderline range.50, 52, 5456, 58, 6365 A1c values in the high-normal range (5.6 – 6.0%) appear to predict a higher incidence of future diabetes,54, 60 and values in this range seem to be the most cost-effective for diagnosing diabetes (though a lower cutpoint of 5.0% would be most efficient for diagnosing both prediabetes and diabetes).70 Several studies underscored the improved sensitivity of A1c in detecting abnormal glucose tolerance in high-risk ethnic groups.50, 55, 64

In summary, A1c is a convenient and potentially clinically meaningful screening test with sensitivity and specificities similar to, or better than, FPG at cutpoints in the high-normal/borderline range. Technical issues with the test may limit its current application as a screening test, though widespread standardization efforts are underway.

IFG, IGT, and Incidence of Diabetes Subsidiary Question 3. Does beginning treatment for IFG or IGT early as a result of screening decrease the incidence of diabetes compared with initiating treatment after clinical diagnosis?

This question was systematically reviewed and incorporated into Key Question 3 in the Results Section of this report.

Recommendations of Other Groups

Many public and private groups internationally have made recommendations on screening for DM2 (Table 1). The ADA recommends that testing be considered in all adults at age 45 years and above, particularly those with BMI ≥ 25 kg/m2; and if testing is normal, it should be repeated at 3-y intervals.46 Testing should be also considered in younger adults or carried out more frequently among persons with risk factors for DM2. The ADA states that these recommendations are based on expert consensus or clinical experience.1 The American Academy of Family Physicians follows the recommendations of the USPSTF.71 The Australian Evidence-based Guideline recommends screening each year for people with IGT or IFG, and every 3 years for people with high risk and a negative screening test.72 The United Kingdom Position Statement recommends targeted case finding.73 The WHO does not recommend screening.74

Previous USPSTF Recommendations

In 2003 the USPSTF made two recommendations regarding screening for DM2:75


The USPSTF concludes that the evidence is insufficient to recommend for or against routinely screening asymptomatic adults for type 2 diabetes, impaired glucose tolerance, or impaired fasting glucose. I recommendation.

The USPSTF found good evidence that available screening tests can accurately detect type 2 diabetes during an early, asymptomatic phase. The USPSTF also found good evidence that intensive glycemic control in patients with clinically detected (not screening detected) diabetes can reduce the progression of microvascular disease. However, the benefits of tight glycemic control on microvascular clinical outcomes take years to become apparent. It has not been demonstrated that beginning diabetes control early as a result of screening provides an incremental benefit compared with initiating treatment after clinical diagnosis. Existing studies have not shown that tight glycemic control significantly reduces macrovascular complications, including myocardial infarction and stroke. The USPSTF found poor evidence to assess possible harms of screening. As a result, the USPSTF could not determine the balance of benefits and harms of routine screening for type 2 diabetes.


The USPSTF recommends screening for type 2 diabetes in adults with hypertension or hyperlipidemia. B recommendation.

The USPSTF found good evidence that, in adults who have hypertension and clinically detected diabetes, lowering blood pressure below conventional target blood pressure values reduces the incidence of cardiovascular events and cardiovascular mortality; this evidence is considered fair when extrapolated to cases of diabetes detected by screening. Among patients with hyperlipidemia, there is good evidence that detecting diabetes substantially improves estimates of individual risk for coronary heart disease, which is an integral part of decisions about lipid-lowering therapy.

Update Key and Subsidiary Questions

This report examines five Key Questions and three subsidiary questions, which were updated and revised from the prior report:40, 76

Update Key Question 1. Is there direct evidence that systematic screening for type 2 diabetes, IFG, or IGT among asymptomatic adults over the age of 20 years at high-risk for diabetes complications improves health outcomes? Does it improve health outcomes for asymptomatic individuals at average-risk for diabetes complications?

Update Key Question 2. Does beginning treatment of type 2 diabetes in adults early as a result of screening provide an incremental benefit in health outcomes compared with initiating treatment after clinical diagnosis?

Update Key Question 3. Does beginning treatment for IFG and/or IGT in adults early as a result of screening provide an incremental benefit in final health outcomes compared with initiating treatment after clinical diagnosis of type 2 diabetes?

Update Key Question 4. What adverse effects result from screening an adult for type 2 diabetes or IFG/IGT?

Update Key Question 5. What adverse effects result from treating an adult with type 2 diabetes, IFG, or IGT detected by screening?

Subsidiary Question 1. What are the yields (accuracy and reliability) of different re-screening intervals among persons with an initial normal fasting glucose?

Subsidiary Question 2. What is the yield [accuracy, reliability, and prevalence] of screening for type 2 diabetes with A1c?

Subsidiary Question 3. Does beginning treatment for IFG or IGT early as a result of screening decrease the incidence of diabetes compared with initiating treatment after clinical diagnosis?

PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...