Talking, not training, increased the accuracy of physicians' diagnosis of their patients' preferences for colon cancer screening.

This was a secondary analysis and was not powered for physician accuracy of preference assessment. The authors did not adjust for multiple tests because the analysis was exploratory. The sample came from academic medical centers in the Northeast United States, and patients were predominantly white, educated, and had high health literacy, so the results may not generalise broadly. No data on patient insurance status or type was collected. Time spent on the discussion was estimated by patients or physicians, not objectively recorded, and the quality of the discussion was not measured. Physicians completed surveys after visits, and memory, caseload, and time between visit and survey could have affected their reports. The preference analyses used 319 matched responses, not all 382 matched surveys. The study tested primary care physicians, not nurses, physical therapists, occupational therapists, or speech-language pathologists.

The study was funded by the Patient-Centered Outcomes Research Institute. All authors report grant support from PCORI. Dr. Sepucha reports grant funding from the Agency for Healthcare Research and Quality and a contract with Blue Cross Blue Shield of Massachusetts. Dr. Atlas reports grant funding from the National Cancer Institute and the Risk Management Foundation of Harvard. Dr. Fairfield reports grant funding from the National Institutes of Health. Dr. Simmons reports paid expert testimony from the US Department of Justice for testimony on cancer screening in primary care malpractice cases.

Do not conclude that spending longer on the discussion causes more accurate preference assessment. Time was self-reported, not objectively measured, and this was an exploratory secondary analysis not powered for accuracy.