Health Literacy and Heart Failure Care in Minority Communities

IN RESPONSE:

We appreciate Drs. Pignone and DeWalt's comments. In our trial, we measured patients' health literacy, however, not general literacy. These concepts differ considerably. In preliminary analyses, we found no evidence that low health literacy was correlated with the effectiveness of nurse management in terms of either of the primary outcomes of physical functioning or all-cause hospitalization. Specifically, we measured physical function by using the Short Form-12 (SF-12) physical component score, which we obtained at baseline and every 3 months through the end of the intervention at 12 months. We estimated a linear regression of the change in SF-12 physical component score on treatment assignment (nurse management vs. usual care), an indicator for inadequate health literacy, and the interaction between inadequate health literacy and treatment assignment. Although the coefficient on treatment assignment was statistically significant, indicating a positive effect of nurse management, the coefficient on the interaction term between nurse management and inadequate health literacy (P > 0.20) provided little evidence that the intervention was more effective for patients with inadequate health literacy at baseline. Alternative specifications of this model, including models with additional control variables for sociodemographic factors and longitudinal models that used intervening data points at 3, 6, and 9 months, yielded similar null findings on health literacy. For all-cause hospitalizations, we estimated Poisson regression models with total hospitalizations over the 12-month study as a function of treatment assignment, inadequate health literacy at baseline, and the interaction between these 2 variables. The coefficient on the interaction term was not significant (P > 0.20).

These null findings on health literacy are particularly vexing because they tend to discount a potential reason why our intervention succeeded where similar interventions have not. Continued research on why disease management works in some patient populations but not in others is clearly needed and is a focus of our continuing efforts.