Identifying Persons With Poor Health Literacy In 7-Seconds: An Important Adjunct To Improving Cancer Screening And Treatment.

6007 Background: One in five adults in the United States has low literacy skills. Low literacy has been linked to poor health status, advanced stage of presentation for cancer, and limited awareness and utilization of cancer screening tests. While assessment tools are available to measure literacy, few clinicians use them routinely. We sought to develop a brief and easy-to-use version of an existing health literacy screening test. Methods: Patients were interviewed in oncology, infectious disease, and general internal medicine clinics (N=1334). All patients completed the Rapid Estimate of Adult Literacy in Medicine (REALM); an individually administered 3–5 minute reading screening test whereby patients pronounce 66 medical words. REALM-66 scores correlate significantly with longer reading comprehension tests (r = 0.97). While word recognition tests do not measure comprehension, they identify low literacy. Results: Stepwise multiple regression was used to predict REALM total score, treating individual words as potential dummy-coded predictor variables. Variables entering the model were constrained to have a partial R-square of ≥1%, and to be statistically significant under a Bonferroni criterion. Seven words entered the model. The seven words (REALM-7) had a correlation of r=0.95 with the REALM-66 total score for the development sample, and an r>0.93 for 2 validity samples. There was an almost perfect mapping from REALM-7 scores to reading grade level based on the REALM, and no evidence of Simpson's paradox was noted for sample stratification by site, gender or race. Conclusions: It is possible to accurately assess patient health literacy and determine reading level equivalent by asking patients to pronounce only 7 words. Physicians may utilize the REALM-7 into clinical practice to identify individuals who may not understand cancer screening or treatment materials. No significant financial relationships to disclose.