To Register      SMDM Homepage

Tuesday, 19 October 2004

This presentation is part of: Poster Session - Clinical Strategies; Judgment and Decison Making

MORE NUMERATE PATIENTS HAVE MORE ACCURATE IMPUTED EFFICACY JUDGMENTS CONCERNING TREATMENTS OF CERVICAL DYSPLASIA

Robert M. Hamm, PhD1, Katy Duncan Smith, MS1, Sharon S. Hsieh, MS1, Crystal Turner, MS1, Kelly C. Juniper, MS1, and Donna H. McCree, PhD2. (1) University of Oklahoma Health Sciences Center, Department of Family and Preventive Medicine, Oklahoma City, OK, (2) Centers for Disease Control and Prevention, Division of STD Prevention, Atlanta, GA

Purpose and Background. To assess whether imputed efficacy (IE) improves on probability (P) judgment per se as a measure of the gist of patient knowledge of health risks and treatment benefits. By subtracting p(outcome | treatment) from p(outcome | no treatment), the effect of correlated judgment errors is removed. Higher IE accuracy in patients with higher numerical literacy (numeracy) would support the claim IE is a useful concept.

Methods. 122 female patients who had HPV DNA tests for evaluation of abnormal Pap screens were asked 10 probability questions about the progression of high risk HPV infection with and without various treatments. Measures of IE for 5 treatments were calculated. For example, from p(cervical cancer|dysplasia) if screened, and if not screened, we imputed efficacy of Pap screens for preventing dysplasia progression. Accuracy of P and IE was measured using the absolute value of the difference from the same judgments made by a panel of expert clinicians. Patient numeracy was measured using TOFHLA, REALM, arithmetic, and number comfort. Based on factor analysis, patients were assigned to high and low numeracy groups.

Results. 8 of patients’ 10 P judgments were higher than experts’, and 4 of their 5 IEs were larger than experts’. Mean absolute error of IE was smaller than MAE of both component probabilities for 3 of the 5 effects. Using the categories based on the overall numeracy factor, the more numerate patients had less variable responses than the less numerate for 7 of 10 Ps and for all 5 IEs. Accuracy of different P and IE judgments was related to different numeracy subscales. The more numerate patients were less accurate for 2 probabilities, and more accurate for 3 probabilities (all p < .05). However, their IEs were more accurate (p < .05) for 3 of 5 IEs. Conclusion. The data support that IE is a better measure of patient understanding of the gist of treatment efficacies than the component P judgments. For most concepts, IE is less variable than P judgments and less inaccurate. Compared with the less numerate patients, the more numerate patients’ IEs were less variable for all 5 concepts, and significantly less inaccurate for 3 of 5 concepts.


See more of Poster Session - Clinical Strategies; Judgment and Decison Making
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)