A number of recent studies have examined the association of health literacy with poor health status, health outcomes, and health care utilization. Most of these studies have small samples, which is understandable because in-person health literacy assessment is time-consuming and costly. But the representativeness of these samples to the general population is unknown.

As has been discussed previously, health literacy is a social construct. It is intimately connected with the socioeconomic environment and with demographics. Health literacy is also complex. A few sociodemographic measures will not account for all individual differences in health literacy. For example, some people with substantial schooling may still have inadequate health literacy, and such cases will be captured only by direct measurement.

However, when one is examining population-level interrelationships such as the extent to which limited literacy is correlated with poor health status, sociodemographic factors may drive a majority of differences in health literacy. A derived measure would allow easy quantification of this relative contribution.

The potential gains of a demographic assessment could be substantial. The derived health literacy measures would be applicable to nationally representative survey data such as those obtained in the National Health Interview Survey (NHIS), the Medical Expenditure Panel Survey (MEPS), and the Medicare Current Beneficiary Survey (MCBS). Using the derived measure, one could then exploit the richness of such datasets, examining the relationship of health literacy with health outcomes (especially rare events that are harder to investigate in small datasets) and with health care utilization.

Two main steps are involved in deriving the DAHL. First, one obtains the imputed measure of health literacy using a dataset that has a direct measure of health literacy, in this case the Prudential Survey data. The Prudential Survey includes an individual measure of health literacy based on the short-form TOFHLA instrument. It is one of the largest such health surveys, with a sample of about 3,000, and is representative of a number of regions around the country. The sample frame for the survey was all new enrollees to a Medicare health maintenance organization plan in four locations (Cleveland, OH; Houston, TX; South Florida; and Tampa, FL) during the 9 months from December 1996 to August 1997. The survey excluded those not living in the community, those with severe cognitive impairment, and those who were not comfortable speaking in either