RESUMO
OBJECTIVE: To assess whether medical orders within Physician Orders for Life-Sustaining Treatment (POLST) forms reflect patients' preferences for care at the end of life. DESIGN: This cross-sectional study assessed the agreement between medical orders in POLST forms and the free-form text documentation of an advance care planning conversation performed by an independent researcher during a single episode of hospitalization. SETTING AND PARTICIPANTS: Inpatients at a single public university hospital, aged 21 years or older, and for whom one of their attending physicians provided a negative answer to the following question: "Would I be surprised if this patient died in the next year?" Data collection occurred between October 2016 and September 2017. MEASURES: Agreement between medical orders in POLST forms and the free-form text documentation of an advance care planning conversation was measured by kappa statistics. RESULTS: Sixty-two patients were interviewed. Patients' median (interquartile range) age was 62 (56-70) years, and 21 patients (34%) were women. Overall, in 7 (11%) cases, disagreement in at least 1 medical order for life-sustaining treatment was found between POLST forms and the content of the independent advance care planning conversation. The kappa statistic for cardiopulmonary resuscitation was 0.92 [95% confidence interval (CI): 0.82-1.00]; for level of medical intervention, 0.90 (95% CI: 0.81-0.99); and for artificially administered nutrition, 0.87 (95% CI: 0.75-0.98). CONCLUSIONS AND IMPLICATIONS: The high level of agreement between medical orders in POLST forms and the documentation in an independent advance care planning conversation offers further support for the POLST paradigm. In addition, the finding that the agreement was not 100% underscores the need to confirm frequently that POLST medical orders accurately reflect patients' current values and preferences of care.