Methods: In a randomized controlled survey experiment, a demographically diverse sample of female Internet users imagined making decisions regarding prenatal genetic screening tests and subsequent possible amniocentesis. Participants were randomized to one of two pre-test risk status groups ("high risk" = 12.5/1000 risk of fetal chromosomal problems versus "low risk" = 2/1000). Post-test risk was equalized for all participants at 5/1000. In addition, participants received their test results in one of three formats: One group only received a numerical post-test risk estimate, while two other groups received numerical information plus a qualitative label that described the test as either "normal/abnormal" or "positive/negative." Participants rated their pre- and post-test risk perceptions as well as their interest in pursuing amniocentesis to identify genetic problems with certainty (but with possible miscarriage risk).
Results: When participants received test results in numerical form only, pre-test risk status had no effect on interest in amniocentesis (51% vs. 47%, p>0.4). When qualitative labels were included, however, reactions were amplified. Even though the post-test risk level was the same for all study participants, previously "low risk" groups receiving "positive" or "abnormal" test results were more interested in pursuing amniocentesis ("positive": 58%; "abnormal": 56%) than previously "high risk" groups receiving "negative" or "normal" test results (both 37%, p's<0.0001). We also observed similar, significant variations in respondents' post-test perceptions of the risk of fetal chromosomal problems.
Conclusions: The seemingly innocuous practice of labeling a screening test result as "normal/abnormal" or "negative/positive" appears to result in greater changes to risk perceptions and behavioral intentions than presentation of the numerical test results alone. In effect, such labels create "diagnostic momentum," inducing people to make subsequent decisions based more on relative changes in risk than on the absolute test result. This finding has broad clinical implications for how test results of all types should be discussed with patients. Further research is needed to assess whether qualitative labels can similarly influence physician behavior.