Use of A1C for Screening and Diagnosis of Type 2 Diabetes in Three Rural Health Care Systems Free

This study evaluated trends in A1C testing used to screen for and diagnose type 2 diabetes in three rural health care systems in the upper Midwest after the 2010 change in ADA guidelines. Additionally, it surveyed health care providers to assess their knowledge and attitudes pertaining to screening for and diagnosing type 2 diabetes.

A1C test results used for analysis in this retrospective review of electronic medical record (EMR) charts were obtained from individuals receiving an A1C test in 2009 and 2010 (before and after, respectively, the 2010 change in the ADA recommendations) in three health care systems in the Red River Valley of North Dakota and Minnesota. De-identified population data used included the date of testing; the International Classification of Diseases Ninth Revision, diagnosis codes; and A1C results. Data were available for 2009 and 2010 from two of the institutions but were limited to the last three quarters of 2010 from the third institution because of a change in EMR systems.

Participants were excluded if they were < 18 years of age.

Data were categorized based on diagnosis code for the presence or absence of diabetes, abnormal glucose, or prediabetes. Data were cross-tabulated by institution and diagnosis code using SPSS (version 10) statistical software (IBM, Amonk, N.Y.). Additionally, the data were analyzed for A1C utilization trends.

Data within the different diagnosis codes (presence or absence of diabetes, abnormal glucose, or prediabetes) were further analyzed by year using the ADA guidelines for diagnosis of diabetes (A1C ≥ 6.5%), prediabetes (A1C 5.7–6.4%), or normal glycemia (A1C ≤ 5.6%).

Additionally, a 13-item survey was sent electronically in February 2011 to health care providers in the area. All providers affiliated with the three health care systems were provided an opportunity to complete the survey. The survey assessed awareness and knowledge of the ADA guidelines, methods currently used to diagnose diabetes, and reported barriers to adoption of A1C as a diagnostic tool. Data were analyzed using summary statistics.

Figure 1 provides a map of the clinic and hospital locations by institution. The data are derived from any individual ≥ 18 years of age who had an A1C test in 2009 or 2010. The A1C data are population-based and include all A1C tests (n = 104,047) performed in 2009 and 2010 in Institution 1 and Institution 2 and during the final three quarters of 2010 in Institution 3. The number of tests is not indicative of the number of patients tested because individuals being treated for diabetes may receive more than one A1C test per year.

Glycemic status was categorized by diagnosis code and then analyzed for trends in A1C usage comparing 2009 to 2010 (Table 1). Institution 1 had the highest number of tests performed (total n = 74,976). For this institution, the relative increase in the total number of tests between 2009 (n = 35,948) and 2010 (n = 39,028) was 9%. There were increases of 26 and 41%, respectively, in the number of tests performed without a diagnosis code of diabetes and with an abnormal glucose or prediabetes diagnosis code (Table 1).

A steady increase was observed in the number of A1C test results without a diagnosis code for diabetes after the ADA guideline change in January 2010 (Figure 2). The A1C test results from individuals with diabetes were included in the population data and analyzed concurrently to provide a comparator for any trends that might have been present in a patient population that is routinely followed for treatment with A1C testing. There was also a more subtle increase in the number of tests performed in the group with a diagnosis of abnormal glucose or prediabetes (Figure 2).

Institution 2 ran 10,135 tests in 2009 and 12,267 in 2010 (total n = 22,402) with a relative increase of 21%. The data showed 21, 74, and 19% increases, respectively, in testing for A1C without a diagnosis code for diabetes, with a diagnosis code for abnormal glucose or prediabetes, and with a diagnosis code for diabetes (Table 1).

A steady increase in the number of tests for individuals without a diagnosis of diabetes was observed starting in the second quarter of 2010 after the ADA guideline change (Figure 3). Although test results from the abnormal glucose or prediabetes group also demonstrated an increase in A1C testing, the rise was more gradual (Figure 3).

Institution 3 only had data available from the last three quarters of 2010 because of adoption of a new EMR. A total of 6,751 tests were performed. The data demonstrated a marked increase in the number of tests performed on individuals without a diagnosis code of diabetes and those with a diagnosis code of abnormal glucose or prediabetes in the fourth quarter of 2010 (Figure 4). There was no known event that prompted this increase, and it was not possible to compare it to data from 2009 for trends.

Figure 5 presents the data for utilization of A1C for the management of individuals with diabetes. Utilization at Institution 1 remained relatively stable, with the exception of the second quarter of 2009. Institution 2, however, showed an increase in the number of tests after the second quarter of 2010. On investigation, the increase at Institution 2 corresponds to a best-practice initiative that began around the same time period. The initiative may also have contributed to an increase in testing for individuals without a diagnosis of diabetes and for those with an abnormal glucose or prediabetes diagnosis code. Institution 3 shows a marked increase in A1C utilization in established diabetic individuals in the fourth quarter of 2010. Although not substantiated, Institution 3 instituted a new EMR system in the first quarter of 2010 that may have facilitated screening and following patients for diabetes.

Analysis was also performed on A1C test results for patients without a diagnosis of diabetes or abnormal glucose or prediabetes as defined by the ADA guidelines for use of A1C for screening and diagnosis. Data were cross-tabulated by diagnosis code, year, and institution (Table 2). For A1C test results that did not have a diagnosis code for diabetes, the relative percentages of A1C test results within the ADA guideline definitions did not show a substantial difference between 2009 and 2010 for either Institution 1 or Institution 2. However, data from Institution 1 fell primarily in the diabetic range (~ 46%), whereas data from Institution 2 fell primarily into the increased risk for future diabetes (prediabetes) range (~ 48%). For A1C test results that had an abnormal glucose or prediabetes diagnosis code, the relative percentages for Institution 1 remained similar for 2009 and 2010, with the majority of the A1C test results falling in the increased risk for future diabetes range (~ 58%). Institution 2, however, showed some shift between the ADA guideline definitions primarily in the ≤ 5.6% range (26% in 2009 compared to 18% in 2010) and the ≥ 6.5% range (15% in 2009 compared to 24% in 2010). The majority of the A1C test results for Institution 2 also fell into the increased risk for future diabetes range (~ 58%). The patterns for Institution 3 were substantially different from those of the other two institutions, with the majority of data in the A1C ≤ 5.6% range (~ 41%) and the 5.7–6.4% range (~ 42%).

A 13-item provider survey was used to identify awareness of the ADA guideline change, diabetes screening methods currently employed, changes in provider use of A1C for all conditions after the guideline change, use of A1C testing before the guideline change, and potential resistance to use of A1C. The response rate to the survey was ~ 8% (62 of 750 providers), and the results of the survey are summarized in Table 3.

The survey results indicated that 90% of respondents were aware of the ADA guideline change, and 61% had actually used A1C to screen for or diagnose diabetes before January 2010. Forty-four percent of respondents who had not previously used A1C before the guideline change began using it to screen for or diagnose diabetes after January 2010. The primary method reported to screen for and diagnose diabetes and to screen for prediabetes continued to be the FPG test, at 75 and 92%, respectively.

A1C was reported as the most common secondary method for screening for and diagnosing diabetes and prediabetes. Other methods such as OGTT or casual glucose > 200 mg/dl with symptoms of diabetes were used less frequently.

The survey also indicated that only 25% of respondents had attended a continuing medical education (CME) session related to the ADA guideline change, whereas 76% of those who had not attended a CME event thought doing so would be beneficial.

When asked about the barriers to using A1C testing for screening for or diagnosing diabetes, respondents cited the cost of testing, concerns about the validity of the testing method, and reimbursement as possible issues.

The results of this study are limited by several factors. The data reported included all A1C testing regardless of its purpose within the institutions, but specific demographic information about the patient population was not provided. Additionally, the institutions and geographical area of the study may not be representative of other health care institutions.

Institution 2 had a best-practice initiative regarding diabetes screening, which may have contributed to the large increase in A1C screening for type 2 diabetes and monitoring of diabetic patients. Similarly, Institution 3 initiated a new EMR system in 2010, and it is not certain whether that contributed to an increase in A1C screening for type 2 diabetes.

This study provides data demonstrating that, although A1C testing has increased, use of A1C testing has not shown the exponential increase that was anticipated by some individuals.^9,10  Additional studies are needed to identify how A1C is being used in patient population subsets.

The use of A1C testing increased in nondiabetic and prediabetic populations in the three health care systems in this study. An increase in A1C use for screening and diagnosis of type 2 diabetes was identified. Institution 2 had a best-practice initiative regarding diabetes screening that may have contributed to its large increase in A1C screening for type 2 diabetes and monitoring of diabetic patients. Institution 3 initiated a new EMR system in 2010, and it is not certain whether that contributed to an increase in A1C screening for type 2 diabetes at that site.

FPG remained the primary test of choice to diagnose type 2 diabetes; this was true even though survey responses indicated a high awareness of the 2010 ADA recommendation regarding use of A1C for screening and diagnosis of type 2 diabetes. The results of this study and responses to its accompanying survey indicate a need for continuing efforts to increase awareness of A1C as a diagnostic as well as management tool among health care providers and an interest among providers in CME opportunities on this topic.