Wednesday, May 18, 2016

Just ask-tell-ask: Physician-Surrogate discordance about prognosis in the ICU


by: Alex Smith, @AlexSmithMD


I'm sure that all of us have encountered this scenario: the doctor thinks patient's prognosis is not good, but the family thinks it's not so bad. 


Why does this happen?  A nice study published yesterday in JAMA sheds considerable light on the issue.  Doug White and colleagues interviewed 229 surrogates of critically ill ICU patients and their ICU physicians at UCSF.  All patients were on a ventilator (breathing machine) and had been in the ICU 5 days. In 53% of instances, the physicians and surrogates held discordant estimates of the patient's prognosis, defined in this study as at least a 20% difference in likelihood of survival to hospital discharge. 


The differences were 4:1 slanted in favor of more optimistic prognostic estimates on the part of surrogates.  One particularly unique feature of this study was the use of qualitative open-ended questions to understand the reasons surrogates were more optimistic.


It turns out that there were two main sources for the discordance between ICU and surrogate perspectives:
  1. The surrogates misunderstanding the ICU physicians prognostic estimate.  The surrogates thought that the ICU physician's estimate was more optimistic than the ICU physician's actual estimate.  In other words, there was a communication breakdown between what the ICU physicians reported and what the surrogates best guess of the ICU physicians prognostic estimate.
  2. Differences in beliefs.  The quotes from the qualitative analysis were illustrative.  (1) Surrogates felt it was important to hold an optimistic estimate because maintaining hope was beneficial to the patient, "I'm trying to think positive;" (2) The surrogate knew the patient had strengths that the physician was unaware of, "[The doctors] don't know his will to live;" and (3) Religious beliefs, "[It's] up to God."
Who was right?  Both were able to predict who survived and who died better than chance alone.  The physicians were slightly (and statistically significantly) better than the surrogates.  Remembering that a coin flip or 50% represents chance alone, the surrogates were 24% better than chance alone.  The physicians were 33% better than chance alone.


Major takeaways:
  • The authors were careful to use the word "discordant" rather than "disagreement."  Disagreement would have implied a conversation about prognosis took place where the doctors and surrogates disagreed.  While 80-90% of surrogates and physicians reported that a conversation about prognosis took place, it's not clear what happened in that conversation.  My guess is the doctors "disclosed" their estimate of prognosis, perhaps being more optimistic with surrogates than they reported for the study.  It's very possible, perhaps likely, that the physicians did not ask the surrogates their own perspective of prognosis, so there was no opportunity for disagreement. 
  • Just ask-tell-ask.  The ask-tell-ask framework is palliative care bread and butter, yet the importance of using it simply cannot be overemphasized.  Some modifications are in order based on these findings: (1) Ask what the surrogate thinks the prognosis is, and what informs that perspective, including desire for optimism, sources of strength and resilience, and religious beliefs; (2) tell what you think the prognosis is and discuss your reasons; (3) ask again to see if this information has changed their views of prognosis.  See this model conversation from ePrognosis about how to use Ask-Tell-Ask.
  • JAMA publishes qualitative research?!?  Off the top of my head, I think it's been some 20+ years since JAMA published any qualitative research.  Hurrah! 













2 comments:

Anonymous said...

I think it's important to recognize that prognosis can be a very fluid concept especially in the ICU setting, where the clinical situation can change very rapidly, from hour to hour, even minute to minute. The intra-individual estimate of the prognosis can vary widely, let alone the inter-individual estimate.

Lynn said...

Alex makes a good point that this study was about "discordance" and not "disagreement." However, I think the findings are most useful for when there is an actual disagreement. This study helps us understand where families are coming from. And it reminds us that we need to ask families where they're coming from - i.e. their values and the patient's values. Some discordance in prognostic estimate itself is not a problem as long as the providers understand the what's most important for the patient in the event of "the worst" happening.