CHALLENGES TO NUMERACY ASSESSMENT AND RISK COMMUNICATION IN AN UNDERSERVED POPULATION

Purpose: We performed a series of face-to-face interviews followed by a written survey among socioeconomically diverse primary care patients to assess literacy and numeracy barriers to shared decision making.

Method: In prior interviews among underserved patients in similar populations, we found that >25% of patients are functionally illiterate, reading at < 3rd grade level. Because many numeracy measures are self-administered, requiring the respondent to read, we have developed four simple picture-based questions that do not require reading to assess the ability to calculate a natural frequency (2 out of 10), a fraction (two-thirds of nine), and two percentages (30% of 10 and 30% of 100). We compared math skills measured by these questions to results of the Subjective Numeracy Scale (SNS), an 8-item written survey (Fagerlin et al. 2007). Demographic data were collected. Reading level was measured according to the REALM instrument.

Result: We interviewed 150 patients, 59% were African American, 39% white, and 1% Asian. Mean age was 54 and 42% were women. Median reading level was 7^th-8^th grade and 24% read at < 6^th  grade level. Using the picture-based questions, all but one respondent could calculate a natural frequency, 42% were able to calculate a fraction and 62% could calculate both percentages. Median total SNS score was 3.5 (1-6 likert scale) and median total score for picture-based questions was 3 of 4 correct. Correlation between the overall math skills measured by the picture-based questions and the total SNS score was (0.25, p<0.0001). However, individual SNS items did not correlate well with picture-based skills assessment in lower reading levels. e.g. there was a negative, nonsignificant correlation between the ability to correctly calculate a fraction and the SNS item “How good are you working with fractions?” among patients with < 6^th grade reading skills (-0.004, p=0.96). The correlations were (0.20, p=0.016) among patients with 7^th- 8^th grade reading skills and (0.15, p=0.08) for those with high school reading skills. Similar results were observed in comparisons between the ability to calculate a percent using picture-based questions and SNS items assessing percents.

Conclusion: Our results suggest that numeracy measures with written self-administered instruments may not perform well among patients with lower literacy skills. Picture-based assessments provide an alternative means of estimating numeracy in these populations.