No matter how the information is presented, poor skills in numerical concepts and basic probability may prevent women from accurately assessing their benefit from mammography screening, according to a study in the current Annals of Internal Medicine. "This is troubling information, since women are increasingly being asked to make their own decisions about screening based on their personal evaluation of the risks and benefits," says Dr. Lisa Schwartz, co-author of the study with Dr. Steven Woloshin. Both are general internists at the Veteran's Affaris Medical Center in White River Junction, Vt. and assistant professors of medicine at Dartmouth Medical School.
A recent consensus panel convened by the National Institutes of Health declined to make a recommendation on screening mammography for women aged 40-49, advocating instead that women assess the need for screening with their doctors. "This assumes that patients understand quantitative information," says Schwartz. "In fact, there is evidence to suggest that many people do not work well with numbers."
In the study of 287 women, more than half were unable to guess corrrectly the number of times heads would come up in 1,000 coin tosses. "This is consistent with findings of the National Adult Literacy Survey," Schwartz says. "Having difficulty with mathematical concepts is very common."
Schwartz and colleagues mailed 500 women one of four questionnaires that differed only in the way they presented statistical information about how the average woman's risk of dying from breast cancer is reduced by screening mammography. The four formats presented information about the potential benefits of mammography in ways that health providers commonly use to communicate such information. The women were chosen at random from a registry of female veterans maintained at the Veterans Affairs Medical Center in White River Junction, Vt. Questionnaires were filled out and returned by 287 women.
Some questionnaires presented the risk reduction as a percentage, such as "a 33 percent reduction in risk;" others as absolute numbers, as "4 in 1,000 risk reduction." Based on the statistical information supplied by the questionnaire they had received, the women were asked to estimate their own risk of dying from breast cancer with and without annual mammography. The women were also asked other questions designed to test numeracy ? that is, their fluency with numerical concepts and probability.
Investigators found that as the number of correct responses to the three numeracy questions increased, the percentage of women who accurately gauged the risk reduction of mammography increased. Adjustment for age, income, education and frame of the information had little effect on the relationship between accuracy and numeracy.
Almost a third of the women were unable to answer any of the numeracy questions correctly. The greatest difficulty encountered was converting between percentages and probability ? 46 percent of the respondants were unable to convert 1 percent to 10 in 1,000; 80 percent were unable to convert 1 in 1,000 to .1 percent.
"Much of the work on decision-making in medicine has been done on framing ? the way information is presented. For example, researchers have shown that people find the same treatment more attractive if its risk is expressed in terms of a gain ? 95 percent survive ? than in terms of a loss ? 5 percent die," says Schwartz. "Though our findings confirmed that framing matters, they also show that low levels of numeracy hinder women from applying information accurately no matter how it is framed."
Other authors on the study were Dartmouth Medical School faculty Dr. William C. Black and Dr. H. Gilbert Welch , who is based at the White River Veterans Affairs Medical Center.
###
Journal
Annals of Internal Medicine