What level of evidence do we need for POLST to use it ourselves, to advocate for wider usage, and for establishing POLST completion as a quality metric?  The answers to these questions will vary.  Reasonable people will disagree.  And today, on our podcast, our guests disagree.  Firmly.  AND we are delighted that our guests modeled respectful disagreement.  With no hard feelings.  Respectful disagreement is in short supply these days. Our guests today are Kelly Vranas, pulm crit care doc who published a systematic review in JAGS of the evidence for POLST (as well as other articles here, here,and here); Abby Dotson, who is Executive Director for National POLST and Director of the Oregon POLST registry; Karl Steinberg, geriatrician and palliative care doc and President of National POLST; and Scott Halpern, pulmonary critical care physician, bioethicists and palliative care researcher who was senior author of a Viewpoint in JAMA that was critical of the concept and evidence base for POLST (and argues little has changed).  We had a full podcast, and I wasn’t able to give my take on the existing evidence for POLST, so I’ll write it here. I’m in the middle between Scott and Karl, where I suspect Kelly is, though we didn’t ask her explicitly.  On the one hand, I agree with Scott that observational studies finding those who complete a POLST stating a preference for comfort oriented care and DNR are not go to the ICU says little about the effectiveness of POLST.  Far more likely that those underlying preferences and values are what drove the findings than completion of the POLST form that codified the preferences into orders.  On the other hand, I agree with Karl that the POLST has face validity, and anecdotal evidence is overwhelming.  Certainly SOME of those avoided hospitalizations, CPR, and ICU stays were due to documentation of those orders in the POLST. The unanswered question is: HOW MUCH of the differences are due to the POLST?  It’s not zero, as many of us have had cases in which we said, “Thank god we completed that POLST, it clearly stopped X from happening.” But is it a tiny, meaningless, fraction?  Or a substantial proportion?  My guess is a small but meaningful fraction of differences in observational studies is due to completion of the POLST, though the majority of differences are due to underlying preferences and values. Is that fraction due to POLST large enough that we should design quality metrics around completion of POLST? Absolutely not. Do we need better evidence, preferably from an RCT of POLST vs no POLST? Yes. Caveat as well that RCTs should not be placed on pedestal as the only answer-  often patients enrolled in RCTs do not represent real world patients - observational studies do.  For a trial to have value, it should not exclude patients over age 80, or those with dementia, or patients residing in nursing homes.  And it must be powered to detect a small but meaningful difference, not the same level of effect seen in observational studies. OK, I’ll step off my soap box. Additional links mentioned in the podcast: Recent JGIM article on POLST in California nursing homes, hospitalization, and nursing home care Karl’s GeriPal post on appropriate use of POLST  Enjoy! -@AlexSmithMD