Article #9
Newsletter list 


Previous Article 

Next Article 


Medical decision making and judgment

Robert M. Hamm
Houston, Texas

This year I have worked on projects that address judgment and decision making in the medical context.

Patients' probabilistic inference. In this study (Hamm, R.M., and Smith, S.L. (1998). The accuracy of patients' judgments of disease probability and test sensitivity and specificity. Journal of Family Practice, 47, 44-52), patients read a vignette of a person seeing a doctor with a given complaint. The major disease that could produce that complaint, and the test typically used to see if the patient has that disease, were described. Patient was asked to estimate prior probability [p(disease)], test sensitivity [p(T+|D)], test specificity [p(T-|D-)], and post-test probability [p(D|T+)]. Knowledge (prior probability and test characteristics) was inaccurate; probabilistic inference (using patient's own judgments) was inaccurate; and past experience with the disease improved the accuracy only slightly. We argued that this demonstrates a need to educate patients explicitly about the possibility of inaccurate test results.

Improving physician judgment. We reviewed past attempts to improve physician decision making through the applications of judgment research or decision analysis (Hamm, R.M., Scheid, D.C., Smith, W.R., and Tape, T.G. (in press, 1998). Opportunities for applying psychological theory to improve medical decision making: Two case histories. In G.B. Chapman and F. Sonnenberg (Eds.), Decision Making in Health Care: Theory, Psychology, and Applications (pp ?-?). New York: Cambridge University Press), and focussed in particular on two projects. One of them gave physicians cognitive feedback to train them to make more accurate probability judgments, but did not change the rate of a criticizable action (prescribing antibiotics for sore throats that are probably due to viral infections). (This result was similar to one reported to this group last year by Tom Taylor.) The other study we discussed gave physicians an accurate (state of the art!) estimate of the probability a patient would die of their illness, and some info about patient values, but it did not reduce the amount of "futile" end of life interventions. Our conclusion might be characterized as "it didn't prove that our science is useless, we just have to try harder,' or 'they did not analyze the situation well enough to discover what was really going on so the information/training provided was irrelevant." What, then, would be relevant? We hope that our paper, which will appear in a volume sponsored by the JDM society, will encourage researchers to look anew at the hard problems encountered when trying to use our best scientific understanding to improve applied decision making.

Analyzing a medical decision. We are analyzing a decision point that occurs when physicians screen women for precursors of cervical cancer (Hamm, R.M., Loemker, V., Reilly, K., Johnson, G., Dubois, P., Staveley-O'Carroll, K., Brand, J., Owens, T., and Smith, K. (in press, September, 1998). A clinical decision analysis of cryotherapy versus expectant management for cervical dysplasia (CIN 1). Journal of Family Practice). If a particular condition is found, one could treat now or wait and see if it goes away and treat only if it doesn't.. If we had confidence the patient would return for the required follow up, then waiting would be better. We are doing a study of 300 patients, to see if it is possible to predict who will faithfully return for follow up in the coming year. In addition to data from the medical record and from a questionnaire the patient fills out, we have asked the physician and nurse to judge the likelihood the patient will return.. We will analyze the accuracy of these experts' judgments, as well as produce an environmental model. Is there enough info available to the physician to afford a prediction that could make a difference in who gets treated?

Contact Robert M. Hamm

Previous Article                Next Article

Newsletter list