Source
VA Outcomes Group, White River Junction, VT 05009, USA. Lisa.Schwartz@Dartmouth.edu
Abstract
OBJECTIVE:
To establish the reliability/validity of an 18-item test of patients' medical data interpretation skills.
DESIGN:
Survey with retest after 2 weeks. Subjects. 178 people recruited from advertisements in local newspapers, an outpatient clinic, and a hospital open house.
RESULTS:
The percentage of correct answers to individual items ranged from 20% to 87%, and medical data interpretation test scores (on a 0- 100 scale) were normally distributed (median 61.1, mean 61.0, range 6-94). Reliability was good (test-retest correlation=0.67, Cronbach's alpha=0.71). Construct validity was supported in several ways. Higher scores were found among people with highest versus lowest numeracy (71 v. 36, P<0.001), highest quantitative literacy (65 v. 28, P<0.001), and highest education (69 v. 42, P=0.004). Scores for 15 physician experts also completing the survey were significantly higher than participants with other postgraduate degrees (mean score 89 v. 69, P<0.001).
CONCLUSION:
The medical data interpretation test is a reliable and valid measure of the ability to interpret medical statistics.