HbA1c in the diagnosis of type 2 diabetes
Mapping review
|Updated
In this rapid review, we have summarised and quality assessed two systematic reviews.
Key message
In this rapid review, we have summarized and quality assessed two systematic reviews. They are evidence support for the World Health Organization (WHO)’s recommendations from 2011 on use of HbA1c as diagnostic criteria for diabetes, and the recommendation from 2012 issued by National Institute for Health and Care Excellence (NICE) on use of HbA1c in the risk assessment for preventing diabetes type 2 or detecting so-called “prediabetes”.
Both systematic reviews have been assessed to be of high quality, which means they are likely not to be affected by weaknesses related to methodological issues. We have not looked into how WHO and NICE actually have used results from the systematic reviews in their guidelines, nor have we examined whether the recommendation can be transferred into Norwegian settings.
Regarding use of HbA1c as diagnostic criteria for diabetes type 2, conclusions from the two systematic reviews are the following:
- HbA1c with a cut-off of 6.5 % has diagnostic accuracy for detecting microvascular complications similar to oral glucose tolerance testing and testing of fasting plasma glucose, but there are uncertainties related to the quality of the evidence from the included studies.
- HbA1c with a cut-off of 5.8 % for Europeans and 6.0 % for Asians has a relatively good diagnostic accuracy for identifying “prediabetes” or increased risks of developing diabetes type 2.
- The diagnostic accuracy using HbA1c is similar to the diagnostic accuracy using fasting plasma glucose for identifying “prediabetes”.
- No conclusions can be drawn regarding use of HbA1c alone to diagnose “prediabetes”, however uncertainties are reduced when risk factors are determined first (age, blood pressure, waist size etc), followed by measurement of HbA1c-level (or fasting glucose or oral glucose tolerance test).
Summary
Background
The national guidelines for diabetes have recently been updated. This includes the recommended diagnostic criteria, which now have been revised to also include the use of HbA1c.
During the process of updating the guidelines, the Norwegian Knowledge Centre for the Health Services (NOKC) was commissioned by Tore Julsrud Berg at the Norwegian Directorate of Health to assess the use of HbA1c as diagnostic criteria for diabetes type 2.
HbA1c (glycated haemoglobin A1c) is a test that reflects the mean glucose level in the blood during the last 6 to 8 weeks before the blood sample is taken. By using HbA1c, problems related to the daily variations in glucose levels are avoided. In addition, fasting before taking the blood sample is not necessary. These are the main arguments for introducing HbA1c as diagnostic criteria for diabetes.
Objective
The purpose of this rapid review was to summarize and quality assess two systematic reviews that are evidence support for recommendations on the use of HbA1c as diagnostic criteria for diabetes.
Method
In March 2012, we performed a hand search for diabetes guidelines, and identified two systematic reviews which were evidence support for the World Health Organization (WHO)’s recommendations from 2011 on use of HbA1c as diagnostic criteria for diabetes, and the recommendation from 2012 issued by National Institute for Health and Care Excellence (NICE) on use of HbA1c in the risk assessment for preventing diabetes type 2 in the United Kingdom. These two systematic reviews are summarized and quality assessed using the checklist for systematic reviews developed by NOKC.
Results
The systematic review carried out for WHO
This systematic review was performed in 2011 by the Boden Institute of Obesity, Nutrition and Exercise, The University of Sydney in Australia, as evidence support for the new WHO guidelines for diabetes. According to the criteria for methodological quality of systematic reviews set by NOKC, this systematic review was of high quality. The research question to be answered was: “How does HbA1c perform in the diagnosis of type 2 diabetes based on the detection and prediction of microvascular complications (retinopathy)?”. Searches were carried out in Medline, Embase, PubMed, Cinahl, PsycInfo and the Cochrane Library for cohort studies evaluating relations between HbA1c levels and prevalence and/or incidence of microvascular complications, including retinopathy which affects sight due to damages in the retina.
The systematic review included 9 studies in total. Seven studies of moderate to high quality assessed HbA1c, FPG and OGTT cut-off levels in relation to the prevalence of retinopathy. Here, optimal HbA1c cut-offs ranged from 6.0 to 7.6 % with sensitivities and specificities spanning from 68 to 87 % and 77 to 100 % respectively, whereas cut-offs for FPG ranged from 6.4 to 8.5 mmol/L with sensitivities and specificities spanning from 81 to 87 % and 77 to 100 %, and cut-offs for OGTT ranged from 9.8 to 14.4 mmol/L with sensitivities and specificities spanning from 83 to 90 % and 76 to 100 % respectively. Two studies, one of moderate and one of high quality according to The Australian National Health and Medical Research Council (NHMRC) criteria for diagnostic studies, assessed HbA1c and FPG cut-off levels in relation to the incidence of retinopathy, but only the one of high quality reported sensitivity and specificity. Here, optimal HbA1c cut-offs ranged from 6.0 to 7.6 % with sensitivities and specificities spanning from 68 to 87 % and 77 to 100 % respectively, whereas cut-offs for FPG ranged from 6.4 to 8.5 mmol/L with sensitivities and specificities spanning from 81 to 87 % and 77 to 100 %, and cut-offs for OGTT ranged from 9.8 to 14.4 mmol/L with sensitivities and specificities spanning from 83 to 90 % and 76 to 100 % respectively. The quality of the evidence for sensitivity and specificity estimates was low to moderate, according to GRADE (Grading of Recommendations Assessment, Development and Evaluation), which means that further research is likely to affect the estimates, thus our confidence in the results.
The systematic review carried out for NICE
This systematic review was performed in 2011 at the University of Sheffield, School of Health and Related Research (ScHARR) in the United Kingdom as evidence support for the NICE guidelines for diabetes. According to NOKC’ criteria for quality assessment of systematic reviews, the systematic review was of high quality. In terms of use of HbA1c as diagnostic criteria, the aim was to answer the following research question: “What is the evidence on the effect of HbA1c testing for identifying prediabetes”? Literature searches were performed in Medline, Embase, British Nursing Index and Archive, Cinahl, PsycInfo, Science Citation Index, EPPI Centre Databases and Cochrane Library.
The systematic review included totally 18 studies that assessed various strategies using blood glucose indicators, including HbA1c. Four of the studies assessed HbA1c alone, where optimal cut-off levels were 5.8 % in Europeans and 6.0 % in Asians, both being relatively sensitive and specific (≥ 61 %). Eight studies had assessed HbA1c in relation to fasting glucose, while six studies assessed multi-step strategies and at least two different diagnostic approaches. According to the QUADAS (Quality Assessment of Diagnostic Studies) criteria, one study was of low quality, one was of high quality, whereas the other 16 studies were of moderate quality. There were large variations between the studies in terms of sensitivity and specificity however uncertainties were reduced when multiple risk factors were assessed in combination with blood samples. Strength of evidence regarding sensitivity and specificity estimates was not assessed in this systematic review.
Discussion
Although the quality of the systematic reviews is high, we have not looked into how WHO and NICE actually have used results from the systematic reviews in their guidelines, nor have we examined whether the recommendation can be transferred into Norwegian settings.
Conclusion
For the detection of microvascular complications (retinopati) HbA1c with a cut-off level of 6.5 % has a diagnostic accuracy similar to the oral glucose tolerance and the fasting plasma glucose test. There is however uncertainty related to the level of evidence.
In terms of identifying “prediabetes” or increased risks of developing diabetes type 2, HbA1c with a cut-off of 5.8 % for Europeans and 6.0 % for Asians had a relatively good diagnostic accuracy. Diagnostic accuracy using HbA1c is similar to that of fasting plasma glucose for detecting “prediabetes”. However no conclusions were drawn regarding the use of HbA1c alone for diagnosing “prediabetes”, but it was pinpointed that the uncertainty was reduced when risk factors were assessed first (age, blood pressure, waist size etc.), followed by measurement of the HbA1c level (or fasting plasma glucose or oral glucose tolerance test).
Long term prospective studies are needed for all major ethnic groups to be able to determine more precisely the levels of HbA1c and plasma glucose that are predictive for microvascular og macrorovascular complications.