The Relationship Between Knowledge of Recent HbA_1c Values and Diabetes Care Understanding and Self-Management

The study sample was drawn from 843 adults with type 2 diabetes receiving care in southeast Michigan health care facilities who had been surveyed about their diabetes-related knowledge, attitudes, and service use. Participants were surveyed between May 2001 and October 2002 in a Veterans Affairs (VA) medical center, an academic medical center (AMC), and three inner-city health systems (a total of five sites). The survey protocols received institutional review board approval at all sites, and written informed consent was obtained from all participants. Eligible patients in the VA and AMC samples, identified through electronic medical records, were ≥30 years old, had a prescription for a glucose control medication or supplies or one hospitalization or two outpatient visits with a diabetes-related ICD-9 code, had seen their primary care provider (PCP) in the prior 6 months, and were scheduled to see the same PCP again in the next 6 months.

In all, 562 eligible VA patients and 720 eligible AMC patients were identified and sent surveys by mail. Excluded were 74 patients who reported not having type 2 diabetes, had severe dementia, or were deceased. Of the remaining patients, 663 completed the survey (56% response rate). Participants identified at inner-city health systems from electronic medical records were recruited as part of the “Racial and Ethnic Approaches to Community Health 2010” Detroit Partnership, supported by the Centers for Disease Control and Prevention and described elsewhere (28). Participants had type 2 diabetes and were ≥18 years old, lived in selected zip codes, and received diabetes care for at least 1 year from their PCPs. Patients were excluded if they had a terminal illness that would preclude participation in a self-management intervention. In all, 346 eligible inner-city patients were identified, of which 180 agreed to enroll and completed the baseline survey upon which this study’s analyses are based (52% response rate). In total, 1,554 eligible patients were identified and 843 completed baseline surveys (55% response rate). Respondents at the VA and AMC were more likely to be older, married, nonwhite, and male than nonrespondents. Respondents in the inner-city health systems were more likely to be older and female than nonrespondents.

For the current study, we identified respondents who had recorded HbA_1c values within the 6-month period before taking the survey (81% of all survey respondents). Patients who were nonwhite, using no medications, and receiving care at the inner-city health systems were less likely to have recorded HbA_1c values.

All participants completed surveys that included core questions about their characteristics, diabetes self-management, and quality of diabetes care. The AMC/VA patient survey was a self-administered written English survey and the inner-city survey was conducted in-person in either Spanish or English.

In the survey, respondents were asked, “What has your HbA_1c (lab value for overall sugar control) been in the past 6 months?” Respondents could choose one of six response categories: <7; between 7 and 8; between 8 and 9; between 9 and 10; >10; and don’t know. We classified respondents as knowing their HbA_1c value if their actual test result was within 0.5 percentage points of the lower or upper boundary of the chosen response category. For example, if respondents reported that their HbA_1c was <7, they were grouped as knowing their HbA_1c if their recorded HbA_1c was <7.5. Respondents were coded as not knowing their value if their estimate differed by >0.5% or if they responded, “I don’t know.”

To assess whether respondents had a biomedically accurate assessment of their HbA_1c value, we created a variable comparing the self-evaluation of the level of diabetes control in the past 6 months with the actual HbA_1c test value. On the survey, respondents were asked whether, based on their HbA_1c value in the past 6 months, their diabetes was in “excellent,” “good,” “fair,” or “poor” control. We classified respondents as having an accurate assessment of their HbA_1c value if they evaluated their diabetes control as poor and had HbA_1c values >8.5; reported “fair” and had HbA_1c between 7.5 and 8.5; or reported “good” or “excellent” and had HbA_1c ≤7.5.

To evaluate self-rated understanding of diabetes care, we used the following question from the Diabetes Care Profile: “How well do you understand how to manage your diabetes?” (29,30) Higher values of this measure rated on a 1–5 Likert scale reflected higher levels of self-reported understanding. To assess diabetes care self-efficacy, we used a validated four-item scale (15), with higher scores reflecting higher self-efficacy in managing diabetes. This measure has been associated with glycemic control in prospective studies (15,31). To assess self-care behaviors related to glycemic control, we used respondents’ answers to a validated measure asking on how many of the past 7 days (days 0–7) they performed the following as their doctor had recommended: take diabetes medications, follow a diabetic eating plan, and monitor blood glucose (32,33). Because adherence in one area of diabetes care does not correlate strongly with adherence in others (32,34), we examined each behavior separately.

We reviewed medical records and laboratory data to document respondents’ most recent HbA_1c results taken within 6 months before the survey. If respondents had no documented HbA_1c results in the prior 6 months, we recorded this value as missing.

We included the following patient characteristics in all the multivariate models: age, sex, annual household income (≤$10,000, $10,001–30,000, >$30,000), education (<high school, high school, at least some college), race/ethnicity (non-Latino white, African American, Latino, Asian, Native American, or Middle Eastern), diabetes duration (≤3 or ≥4 years), and hypoglycemic medications they were currently taking (no medications, oral medication only, or insulin ± oral medications). To evaluate thoroughness of provider communication, we assessed the degree to which respondents agreed with the following statement from the well-validated Autonomy Support Scale: “My doctor answers my questions fully and carefully” (with five response categories from “strongly disagree” to “strongly agree”) (15,16). Because responses were positively skewed toward the highest rating, we dichotomized responses between those who “strongly agreed” with the statement versus all other responses. We also included variables for health care site (VA, AMC, or inner-city health system), mean number of outpatient visits in the prior year (continuous), and duration of the relationship with the doctor who takes care of the patient’s diabetes (<6 months, 6 months to 1 year, 1–5 years, >5 years).

We conducted bivariate and multivariate logistic regression analyses to explore patient, provider, and health care system characteristics associated with knowing one’s most recent HbA_1c value. We then used multivariate linear and logistic regression to assess whether knowledge of one’s last HbA_1c was associated with an accurate assessment of one’s level of diabetes control, diabetes care understanding, self-efficacy, and self-management behaviors related to glycemic control. To determine the sensitivity of our findings to the specific cutoff points we used in our constructed variables for knowledge of HbA_1c and assessment of diabetes control, we conducted additional analyses using different cutoff points for these variables. These alternative cutoff points did not significantly change the findings. We also conducted analyses separately for patients who reported some value for their last HbA_1c versus those who reported that they did not know their last HbA_1c value. In these analyses, many of the statistical associations remained significant and the magnitude of effect estimates was similar. Finally, because of the differences in survey methodology and clustering of ethnic groups in different health systems, we conducted all the analyses separately for the AMC/VA sites and the inner-city health sites. Regression diagnostic procedures yielded no evidence of multicolinearity or overly influential outliers in any of the models.

Overall, the sample was socioeconomically and ethnically diverse (Table 1) with 69% non-Latino white, 17% African American, and 8% Latino (primarily Mexican American). For education, 55% had a high school education or less; 71% had annual household incomes of $30,000 or less. In all, 94% were receiving oral medications or insulin ± oral medications. The mean number of outpatient visits in the past year was 4.1 ± 3.0. In total, 38% of respondents received care at the VA, 44% at the AMC, and 18% at the inner-city health systems.

Characteristics of the patients by race/ethnicity are presented in Table 1. A higher percentage of African-American and Latino respondents had incomes <$30,000 than the other two groups and a much higher percentage of Latino respondents had not completed high school. African-American and Latino respondents predominantly received care at the inner-city health systems, with more outpatient visits in the past year and higher rates of insulin use on average. African-American and Latino respondents on average had higher recorded HbA_1c values than the other two ethnic groups, with Latino respondents having the highest HbA_1c values (8.5 ± 2.3). There were no significant differences among ethnic groups in assessments of their diabetes control, diabetes care self-efficacy, or evaluations of providers’ communication. Latino respondents had lower self-reported understanding of their diabetes care and reported following a diabetes eating plan fewer days in the prior 7 days than the other groups. There were no ethnic differences in reported medication taking and blood glucose monitoring over the prior 7 days.

Overall, 66% of respondents reported that they did not know their last HbA_1c value, and 25% of respondents accurately reported their most recent HbA_1c value. The majority of those who reported an HbA_1c value were relatively accurate (Table 2). A higher percentage of those who reported lower HbA_1c values were accurate compared with patients who reported higher HbA_1c levels. For example, 76% of the 67 respondents who reported that their last HbA_1c was <7 had documented HbA_1c values in that range, whereas 40% of the 25 respondents who reported that their last HbA_1c was >10 had documented HbA_1c values >10.

Table 3 shows the bivariate and adjusted odds of accurately knowing one’s most recent HbA_1c values. In the bivariate analyses, significantly lower percentages of Latinos (8%) accurately reported their last HbA_1c than respondents of other ethnicities. Lower percentages of respondents with less than a high school education, income ≤$10,000, and who received care at the VA or inner-city health systems knew their HbA_1c values than other groups (Table 3). Moreover, 30% of those who strongly agreed that their diabetes doctor fully answered their questions knew their HbA_1c, compared with 21% of those who did not strongly agree (P < 0.01).

In multivariate logistic regression models (Table 3), years of formal education and high evaluations of thoroughness of provider communication were associated with knowledge of recent HbA_1c values. Receiving care at the VA was associated with lower odds of knowing one’s recent HbA_1c value. In additional analyses that included language preference (Spanish vs. English), there was no association with knowledge of HbA_1c. In the separate multivariate analyses for the AMC/VA populations and inner-city populations, we found the same pattern of associations in the two groups, except that higher income remained significantly associated with HbA_1c knowledge in the AMC/VA subsample.

In multivariate analyses, respondents who knew their last HbA_1c value had higher odds of reporting a biomedically accurate level of diabetes control (odds ratio 1.59; 95% CI 1.05–2.42). In all, 56% of respondents who knew their HbA_1c gave accurate assessments of their diabetes control compared with 45% of those who did not know their HbA_1c (P < 0.001). Similarly, knowing one’s HbA_1c was associated with higher scores on the measure of patients’ reported diabetes care understanding (β 0.17, P < 0.001). Those who knew their HbA_1c had mean diabetes understanding scores of 3.80 ± 0.89 compared with mean scores of 3.38 ± 0.93 among those who did not know their last HbA_1c value (P < 0.001). However, knowledge of HbA_1c was not associated with better diabetes care self-efficacy or any of the three specific diabetes self-care domains that we assessed (β coefficients −0.032 to −0.006; P values 0.50–0.90). This same pattern of associations was also found when we analyzed the AMC/VA and inner-city subsamples separately.

Regular testing of HbA_1c values is now the principal way to measure and track glycemic control in diabetes. Because of its importance as a marker of disease control, it makes sense that patient knowledge of recent and target HbA_1c values might be a useful precondition for involvement in diabetes management. Accordingly, in recent years there has been an increased focus on encouraging patients to be aware of and discuss these values with their clinicians (21–23).

Few respondents in our study, however, knew their most recent HbA_1c value. These low rates are similar to those of earlier studies (24–26). For example, in a 2002 study, 24% of those who reported having an HbA_1c test in the past year reported an actual test value, and those self-reported values correlated poorly with the medical record HbA_1c (26). The correlates of HbA_1c knowledge in our study corresponded to factors associated with key aspects of diabetes care understanding and self-care in prior studies (34–37). Although Latino respondents were less than half as likely to know their last HbA_1c than patients of other ethnicities, this was largely explained by their lower educational levels. In multivariate models, more years of formal education was the only sociodemographic characteristic associated with higher odds of knowing one’s recent HbA_1c values. The independent association between a high evaluation of provider thoroughness of communication with HbA_1c knowledge is consistent with research on the importance of effective provider communication for improved patient understanding and chronic disease self-care (38–42). Further research should elucidate why VA respondents had significantly lower odds than patients at the other sites of knowing their last HbA_1c.

In bivariate and multivariate analyses, knowing one’s last HbA_1c was associated both with accurately assessing one’s level of diabetes control and with reporting better diabetes care understanding. Respondents who accurately reported their HbA_1c values had 60% higher odds of correctly assessing their level of biomedical control of their diabetes than those who did not. These findings reinforce the importance of sharing with patients clear and specific information on their disease status and markers, such as HbA_1c, blood pressure, and lipid values (43). An estimated 50–80% of adults with diabetes have significant diabetes-specific knowledge and skill deficits (7,26,44). Potentially effective strategies to complement verbal communication to patients on actual and target disease values include providing clear graphical representations of these values (21,22,45) and encouraging patients to record and track laboratory and other measurements in diabetes logbooks or “passports” (46–48). Such strategies may be especially important to convey clinical information to patients with low health literacy and little formal education (49,50).

Knowledge of recent HbA_1c test findings was not associated with either diabetes care self-efficacy or with better short-term self-management practices related to glycemic control. These findings reinforce that factors beyond knowledge of disease-specific information are necessary to heighten patients’ self-confidence in their diabetes management and to improve diabetes self-management. A growing number of studies suggest that addressing patients’ own perceptions of barriers to self-care and tapping into patients’ values, motivations, and goals are more effective in improving metabolic control than seeking exclusively to increase knowledge about diabetes care (6,7,34,51–54). Greater patient knowledge alone does not correlate with improved glycemic control, and simply providing information more clearly is not enough to motivate patients. To enhance patients’ diabetes care self-efficacy and self-management, providers need to promote patients’ capacity to define the problems they are facing, make informed decisions about their diabetes management, and set realistic goals and strategies to meet those goals (16,17,22).

Although our study’s findings are consistent with prior work in this area, the study has limitations. First, we can only show associations, not causality, in this cross-sectional study. Second, these analyses only included respondents for whom we had recorded HbA_1c values and who had received regular medical care in the prior year. Third, patients only were asked whether they knew their actual HbA_1c values, not whether they knew what their target level should be. Knowledge of both actual and target values may be important for enabling patients to monitor their progress toward achieving diabetes control. In addition, the multiple-choice format of the HbA_1c may have cued some respondents to what their HbA_1c value was. Fourth, the response rate was relatively low (55%). Those most intrinsically interested in the subject matter or with some other motivation (e.g., possibly increased need) may have been disproportionately likely to participate, reducing the generalizability of our findings. Finally, we did not have uniform eligibility criteria and selection processes for all respondents, and survey administration differed between the inner-city and other populations (in-person vs. written). Moreover, because the ethnic groups were largely clustered in different health systems, even though health site was included as a variable in all of the models, we could not adequately adjust for health site differences.

In conclusion, a minority of diabetic patients in our study knew their most recent HbA_1c value. Respondents who knew their HbA_1c values reported significantly better diabetes care understanding and assessment of their biomedical level of glycemic control than those who did not. These findings support the importance of providers actively discussing HbA_1c test results with patients and ensuring that patients understand the meaning of their HbA_1c level. Knowledge of HbA_1c alone, however, was not associated with better diabetes care self-efficacy and self-management behaviors. As with other areas of diabetes knowledge, knowledge of one’s last HbA_1c value appears to be useful but not sufficient for translating increased understanding of diabetes care into the increased confidence and motivation necessary to improve patients’ diabetes self-management. Strategies to provide information must be combined with other behavioral strategies to motivate and help patients effectively manage their diabetes.

This work was supported by the Department of Veterans Affairs Health Services Research & Development Service, the Robert Wood Johnson Foundation Clinical Scholars Program, and the Centers for Disease Control and Prevention, REACH 2010 Grant U50CCU522189. M.H. and S.V. are VA Health Services Research & Development Service Career Development awardees, and J.D.P. is a VA Health Services Research & Development Service Career Scientist.