Literacy and health outcomes: a cross-sectional study in 1002 adults with diabetes

Navigating the US health care system requires sophisticated reading, writing, and numeracy skills that go beyond the health literacy abilities of many Americans [1, 2]. Adults with limited health literacy face challenges interpreting and analyzing health information, becoming knowledgeable about their specific health conditions and health risks, and understanding prescriptions and other treatment recommendations [3‐9]. Evidence is beginning to accumulate about the prevalence of limited health literacy and its association with the use of health care services [5‐7, 10‐13] and health outcomes[8, 14‐16].

Patients with chronic conditions such as diabetes face even greater communication challenges. The complexity of diabetes care requires an informed individual who can seek, obtain, and comprehend information to engage in the management of his/her health. Health outcomes for adults with diabetes are better for those who can optimally incorporate self-management of their diabetes into their daily lives[17]. Decreasing elevated BP, LDL, and A1C in patients with type 2 diabetes reduces the risk of cardiovascular and microvascular events by about 50 percent[18]. The American Diabetes Association recommends that best management of diabetes can be obtained with a glycated hemoglobin (A1C) < 7%, LDL < 100 mg/dl, and BP < 130/80 mmHg[19].

Recent cross-sectional studies have assessed the association between literacy and diabetes care with conflicting outcomes. Williams et al.[9] did not find a significant correlation between literacy and health outcomes while Schillinger et al.[16] reported worse glycemic control and higher rates of retinopathy and cerebrovascular complications in subjects with lower literacy. Ross et al.[20] did not find a significant correlation between the literacy of children and their glycemic control but they did observe a significant relationship between the literacy of the mother and her child's glycemic control. Two recent studies by Rothman and colleagues[21, 22] examined the relationship between literacy and achievement of goal A1C specifically in disease management programs and reported greater benefit to patients with I literacy.

Given the complexity of diabetes management, we hypothesized that adults with diabetes and limited literacy would be less likely to achieve the recommended goals for A1C, systolic blood pressure (SBP), diastolic blood pressure (DBP), and low density lipoprotein (LDL). We also hypothesized that adults with diabetes and limited literacy would report more complications related to their diabetes.

Of the 1,576 patients we attempted to contact to participate in the interview, 36% (570) were not reached or declined. The 7,801 non-interviewed subjects of VDIS, (which includes those who declined and those never contacted) were younger (mean age of 63 versus 65, P < 0.001); more likely to be men (49% versus 46%; P = 0.06); had a lower baseline A1C (mean of 6.9 versus 7.1, P < 0.001); and a higher baseline LDL (mean of 106 versus 102, P = 0.002) than those who completed the interview. Due to missing data on 4 subjects, we had complete questionnaire results and S-TOFHLA scores on 1002 subjects with confirmed diabetes. The demographic characteristics of the study population reflect the population of northern New England (Table 1)[30]. The subjects were predominately white, educated, and older with a mean age of almost 65 years. Most had health insurance, were diagnosed with diabetes on average 7 years and almost half had an A1C < 7%. Overall, the mean A1C was 7.12%, the mean LDL was 102 mg/dl, and the mean BP was 140/78 mmHg. The mean S-TOFHLA score was 29.7 with a range of 0–36.

One-hundred and five (10%) subjects had inadequate literacy, 66 (7%) had marginal literacy, and 831 (83%) had adequate literacy (Table 1). Compared to those with adequate literacy, subjects with inadequate or marginal literacy were significantly older; less educated; less likely to be married; poorer; less likely to have private insurance and more likely to have Medicare or Medicaid coverage. Those with lower literacy had been diagnosed with diabetes for a longer time; reported less alcohol use; were more likely to have depression; had a lower diastolic blood pressure; more self-reported retinopathy, stroke or coronary artery disease; and were more likely to take medication for diabetes and hypertension.

In unadjusted linear regression analyses there were no significant relationships for any of the four outcome variables except diastolic blood pressure. However, after adjusting for age, sex, race, marital status, insurance, income, duration of diabetes, diabetes education, depression, alcohol use, and medication use specific to each outcome, literacy was no longer independently associated with any of the outcomes of interest (Table 2).

Repeating the analyses after excluding subjects who reported a history of stroke (N = 118) or dementia (N = 5) did not identify a significant relationship between S-TOFHLA score and A1C, LDL, or blood pressure. Likewise, excluding subjects with poor vision or other physical impairments (N = 23) revealed no significant associations between literacy and physiologic control of diabetes.

More complications related to diabetes were reported by the subjects with inadequate or marginal literacy. However, the differences between the groups disappeared after adjusting for confounders. (Table 3).

After controlling for potential confounders, we did not find a significant association between literacy and health outcomes for glycemic control, blood pressure, or dyslipidemia in adults who receive diabetes care in community primary care settings. We also did not find an association between literacy and complications of diabetes. This is counter to our expectations, particularly for a study population with a chronic disease requiring self-care to achieve optimal health. Our results raise questions regarding how health literacy is assessed, how it may vary across populations, and how patients may compensate for low literacy.

Our results are consistent with those of Williams et al.[9] who reported no significant association between control of blood glucose and blood pressure and lower literacy despite a strong correlation between inadequate literacy skills and worse disease knowledge. Although small sample size (352 patients with hypertension, 55 with diabetes) may have been a factor in that study, our results are similar in a much larger population.

In a study that examined the literacy level of children with type 1 diabetes and the literacy level of their mothers, Ross et al. reported no significant correlation between the mean annual A1C of the child and the child's literacy level[20]. However, a significant relationship was noted between the mean A1C of the child and the literacy level of the mother (r = 0.28; P = 0.01) after adjustment for the child's age and sex, duration of diabetes, daily insulin dose, the child's literacy score, and social class[20].

Similarly, Schillinger et al.[16] found that among 408 patients with type 2 diabetes seen in public hospital clinics, inadequate health literacy as measured by the S-TOFHLA, was independently associated with poor glycemic control (OR = 0.57; 95% CI 0.32–1.00; P = 0.05) retinopathy (OR = 2.33; 95% CI, 1.19–4.57; P = 0.01) and cerebrovascular disease (OR = 2.71; 95% CI, 1.06–6.97; P = 0.04). They controlled for age, sex, race/ethnicity, education, language, insurance, social support, depressive symptoms, complications of diabetes, diabetes education, duration of diabetes, and diabetes treatment. One explanation for our conflicting results may be related to differences in the populations under study. Schillinger's study involved a younger, more ethnically diverse population with poor glycemic control (mean A1C of 8.5%). Eighty-five percent of their population reported a race/ethnicity other than white, 46% did not complete high school, 32% had no health insurance, and 93% reported an annual household income less than $20,000. Another explanation for the difference between our outcomes and those reported by Schillinger et al. may be that our measurements were not exactly the same in all domains. We used different measures to assess depression and social support and different criteria for determining the presence of complications associated with diabetes.

Rothman et al. report that literacy was a statistically significant effect modifier in determining how patients with diabetes did in a disease management program; the patients with lower literacy were more likely to achieve goal glycemic control[21, 22]. Consistent with our pre-intervention patient characteristics Rothman et al. reported similar baseline A1Cs between low and high literacy patients. However, in their study participation in a disease management program showed that literacy does appear to be a factor in diabetes control.

Our finding of a lack of significant association between literacy and disease outcome may be because compensatory strategies may have already been implemented to accommodate literacy barriers. It may be that achievement of some goals is more dependent on practitioner action than patient self-care skills[22]. Compared to the self-care associated with diabetes, the patient's role in management of cholesterol and blood pressure does not require as much complex reasoning and management. The older adults in the VDIS study may also have had support systems in place (help of a spouse, visiting nurse) to assist with diabetes management making health literacy less relevant. Medication tolerability may also be a factor for outcome achievement that goes beyond literacy.

Another possible explanation for our finding of a lack of a significant association between literacy and health outcomes is that optimal self-management of diabetes may not be solely dependent on reading ability. In addition to print literacy, health literacy includes numeracy, oral literacy, culture and context[2]. The published literature on "health literacy" typically reports measures of reading ability and rarely, if ever, addresses the broader domains of health communication[14, 31, 32]. While there may be great value in assessing health literacy to address communication barriers between patients and health care providers, we still do not know what aspects of health literacy may be important to communication, how to measure most aspects of health literacy, how to intervene, and if interventions will improve outcomes of care, particularly for adults with diabetes.

There are several additional limitations to our study. The cross-sectional study design does not allow us to measure incident outcomes or assign cause and effect. The generalizability of the sample is limited because it comes from a single region and is racially homogenous. There is a potential self-selection bias in the population of subjects who agreed to be interviewed, who differed significantly from the non-interviewed VDIS study population who were younger, more likely to be male, had better A1Cs and worse LDLs. The younger age group may have been more likely to be working and have other competing role demands limiting their availability to engage in an interview. It is also possible that subjects who declined participation in the interview process had lower literacy skills. However, all the subjects were outpatients with a diagnosis of diabetes and are representative of patients cared for in many community based primary care settings.

Our study suggests that literacy, as measured by the S-TOFHLA, is not associated with A1C, LDL, or BP in a cross-sectional study of older adults with diabetes under relatively good glycemic control. The common complications of diabetes are not associated with limited literacy in this population. Additional studies to examine the optimal measurement of health literacy and its relationship to health outcomes are needed.