| To Register SMDM Homepage | |
|
|
|
||||
Purpose: To determine how presenting pediatricians with a diagnostic test’s sensitivity and specificity affects their estimate of the posttest probability.
Methods: We mailed a four-page questionnaire to 1502 pediatricians practicing in the U.S. randomly selected from the AMA master list, and sent up to 4 times to non-responders. The questionnaire presented a vignette of a 5-month old girl with perioral cyanosis and a hacking cough; subjects were instructed to assume that the likelihood that she had pertussis was 30%. They were then told that the direct fluorescent antibody (DFA) for pertussis sent on her was negative. We randomized subjects to one of three decision support groups (control, technical or non-technical decision support). Controls received no additional information. The technical decision support group was presented the sensitivity (50%) and specificity (95%) of the DFA for pertussis. The non-technical decision support group was presented the DFA’s sensitivity and specificity, and a non-technical explanation of these terms. Subjects were asked to estimate the likelihood that this patient had pertussis. We conducted t tests to evaluate how decision support affected estimates of posttest probability and chi-square tests to compare the proportions in each group who chose a probability of 50%.
Results: 653 subjects returned completed surveys: 202 in the control, 231 in the technical and 220 in the non-technical decision support groups. The estimated response rate was 52%. 81% of participants were board certified in pediatrics and their mean age was 43 years. There were no significant differences between randomization groups in any subject characteristic. The correct posttest probability was 18%, but it was substantially overestimated in all three groups (mean 41, median 50, range 0-100, inter-quartile range 15-50). Subjects who received technical decision support had a mean posttest probability not significantly different from controls (41% vs. 38%, p = 0.16). In contrast, subjects who received non-technical decision support estimated a significantly higher posttest probability than controls (45% vs. 38%, p<0.001). Subjects who received technical or non-technical support were more likely than controls to chose a probability of 50% (17% vs. 38%, p <0.0005; 17% vs. 41%, p <0.0005).
Conclusions: Presenting non-technical decision support about the sensitivity and specificity of a diagnostic test worsened pediatricians’ estimates of disease probability. Decision support may have unexpected effects on medical decision making.
See more of Oral Concurrent Session A - Clinical Strategies and Guidelines
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)