Limited literacy was common in patients with hypertension cared for in both VA and the university affiliated primary care practices. Our study demonstrated that the relationship between literacy and SBP was complex and was conditional on the healthcare system in which patients received their care. More specifically, the relationship between literacy and SBP differed significantly in the two healthcare systems, with much larger differences in SBP according to literacy level for patients in the UHS than the VAHS. The difference in SBP persisted after adjusting for several variables including age, race, medication adherence, and education. In unadjusted analysis, there were similar trends in the relationship between literacy and DBP and BP control according to healthcare system; however the differences were not statistically significant after adjusting for covariates. In spite of this, the observed SBP difference of 6.1 mmHg according to literacy status can contribute to a significant increase in risk for vascular events, especially when considered over many years [30
]. These findings suggest that the relationship between literacy and SBP may vary significantly across healthcare systems.
While we did not formally measure any organizational characteristics of either healthcare system, there is reason to believe that the characteristics of the healthcare organization may significantly influence the impact of literacy on health. Changing how healthcare organizations interact with patients with limited literacy has been suggested as a way to improve health disparities related to literacy [31
]. Addressing the system of healthcare delivery is a key part of the Chronic Care Model and may be particularly important for patients with limited literacy [32
]. Others have shown that literacy may predict response to disease management interventions that change the system of care for these patients [17
]. In the growing body of literature examining the relationship between literacy and health outcomes, surprisingly few studies have been conducted in a VA setting and further research in this setting would be helpful to identify whether the relationship between literacy and health differs from other systems.
Although these results have potentially important research and healthcare implications, they should be interpreted with several caveats. The generalizability of our findings may be limited by including only patients sufficiently motivated to participate in a two year randomized controlled trial. In addition, although the combination of two separate datasets provides the advantage of greater statistical power and the ability to compare healthcare systems, these data were collected at different time points from separate randomized controlled trials. To ensure the consistency of our measured variables between studies, we examined the baseline interview from each study carefully and only include variables elicited in the same fashion. Furthermore, the two studies were conducted by the same principal investigator (H.B.B.) and study team, thereby reducing the chance of bias from differential measurement in the two healthcare systems.
An alternative explanation for our findings is that there were systematic differences between the patients in the two healthcare systems that confounded our results. Gender was heavily imbalanced between the two healthcare systems and we were unable to adjust for this variable in the combined model. Also, veterans may adopt different health behaviors as a result of their training in the military that were not captured in our available patient measurements. Although we adjusted for several patient variables that may be associated with systolic blood pressure, we did not include measures of patient knowledge, health beliefs, or health status, which have previously been associated with literacy and may differ between the two patient populations [6
]. In addition to patient characteristics, we did not explore other variables that may mediate our findings such as type of health insurance coverage or more specific clinic site level differences in how care is delivered.
Finally, although the relationship between literacy and blood pressure significantly differed between these two healthcare systems, further work including a larger number of representative healthcare systems would provide more definitive evidence that literacy's impact on disease outcomes varies across different systems of healthcare delivery. Further work should include more explicit measurement of the financial and organizational characteristics of healthcare delivery. Within a healthcare delivery system, there may be many factors that interact with patient literacy to influence health outcomes. Future studies with more detailed measurement of organizational characteristics are needed to both validate our findings and provide greater information about the factors that may mediate the interaction between literacy and healthcare delivery systems.