What Predicts Agreement about Prognosis Between Patients and Physicians?

Studies suggest that many patients with cancer are more optimistic about their life expectancies or chances of cure compared to concurrent estimates from their physicians. In this study, we tested the hypothesis that consultations in which physicians used unambiguous quantitative prognostic communication would be associated with more accurate (concordant) prognostic understanding between patients and their physicians. We also explored the extent to which key patient, physician, and consultation characteristics predicted prognostic agreement. Methods and Subjects: We studied 232 patients having first consultations with 40 physicians from two academic institutions in the HEMA-COMM study. The HEMA-COMM study is an observational study that evaluates doctor-patient communication from patient and physician surveys, patient interviews, and audiotaping and coding of the consultations. The primary outcome for this analysis was prognostic agreement, calculated by subtracting patients' post consultation prognostic estimates from their physicians' post consultation estimates for life expectancy and likelihood of cure. Potential explanatory variables include patient, physician and consultation characteristics, and the occurrence and clarity of prognostic discussions as coded from the consultation audiotapes. Multilevel models tested whether specific physician communication behaviors (provision of quantitative prognostic estimates, inviting questions, or suggesting that prognostic estimates may not apply to the individual, so called “hedging”) predicted patient-physician agreement on prognostic estimates. Results: Median duration of the consultations was 80 minutes. Most patients (over 90%) desired prognostic information, 92% wanted this information in quantitative terms, and 77% of patients said that chance of cure would influence their treatment choice. After the consultation, 85% of patients said their physicians discussed prognosis. Coding of the audiotapes showed that in 77% of consultations, oncologists discussed life expectancy (or survival) in quantitative terms, and in 51% of consultations they discussed chance of cure. One third of patients, however, overestimated their life expectancy by two or more categories, and 34% overestimated chance of cure by 20 percentage points or more. Patients who were younger, less optimistic about their prognoses prior to the consultation, did not have a treatment plan prior to their visits, or were judged by their physicians as “actively seeking prognostic information” were more likely to agree with their physicians about prognosis. Physician communication behaviors such as provision of quantitative prognostic information without hedging and inviting questions were not predictive of patient-physician agreement about prognosis. Conclusion: Providing patients with a clear numerical prognosis does not correlate with improved concordance between physicians and patients about predicted life expectancy and chance of cure, and appears to be insufficient to achieve understanding on its own. Additional approaches to assess and achieve understanding are needed.

No relevant conflicts of interest to declare.