Sunday, January 9, 2011

Hospitalists and Code Status Discussions Upon Hospital Admission: The Importance of Framing

"Frame" sculpture near Strongstry, England
(photo by Liz Dawson) /  CC BY-SA 2.0
Consider a case:  A 60 year-old man is admitted to the hospital with failure to thrive secondary to metastatic cancer.  The physician who admits him asks the question out of the blue, "If your heart stops or you stop breathing, would you want us to attempt to resuscitate you?"  How many times is this exact conversation taking place right now and what do you think the answer usually is?  (The phony stat: "One out of every x minutes, a doctor conducts a code status discussion in a manner that may lead to a decision which is unlikely to help the patient achieve their goals of care.")

This type of scenario is similar to one of the scenarios examined in a recent cross-sectional, observational study conducted at the University of California, San Fransisco published in the Journal of General Internal Medicine.  While the study didn't determine what the above x is, it adds details to our understanding of code status conversations that occur upon hospital admission.  In the study, initial hospital evaluations were recorded and transcribed and then a quantitative content analysis was conducted to examine the nature of code status discussions conducted.  The main goals of the study included describing the code status conversations that took place, understanding the nature of those conversations, and determining whether the conversations adhered to standard recommendations (as outlined in Table 3 of the paper).  Of the 80 encounters recorded, code status was discussed during the hospitalist's initial evaluation in 19 encounters.

Physicians were more likely to discuss code status if:
  • The patient was older.
  • Patients whose death or ICU admission within a year would not surprise the physician (Odds Ratio of 4 with Confidence Interval approaching 1). But physicians did not have a code status discussion with 66% of patients who met this criterion.  
Code status discussions lasted for a median of 1 minute.

Back to the case above.  In only one encounter (which was similar to the case above, although the physician's approach to the code status conversation was not described) was prognosis addressed.  In this case, it was the patient who was savvy enough to understand that prognosis is an important part of the "frame" of the code status conversation, so he brought it up to the physician:
PATIENT: [My oncology physicians] really haven’t provided much information on prognosis…I think that’s a discussion that I need a little more information. It’s still a little abstract. I mean, my
general thing [is] to say, “No, I don’t want heroic measures.”…But that seems like a fairly drastic response, unless it’s just…I mean, you know, I just don’t feel like I know what the words mean enough to know what I’m saying.
The heart of the results comes from understanding that in spite of the growth of the palliative care movement and related educational initiatives, many physicians (the average physician in this study finished med school in the 21st century) still may not understand the basics of code status discussions such as those highlighted in Table 3 of the study, such as the need to:
  • Seek mutual understanding between the physician and patient regarding prognosis and goals of care before initiating a discussion about code status.
  • Approach the code status discussion like an informed-consent conversation (replete with a comprehensive details about the risks, benefits/outcomes, and alternatives).
  • Make a recommendation to the patient that accounts for both the patient's prognosis and goals of care.
One uncertainty addressed in the discussion section is the fact that it's unclear when it is appropriate to address code status during an initial encounter (at the time of hospital admission).  The authors make the point that the Patient Self Determination Act may have led some physicians to conclude that code status should be addressed during every admission.  This may have led to the unintended consequence of some having a brief, poorly conducted conversation with every patient rather than a high quality conversation with those who most need it.

A code status discussion at the time of hospital admission differs from an advance care planning conversation in the clinic or during a palliative care consultation in some important ways, most of which are related to the fact that the provider has just met the patient:
  • Less likely for the provider to possess an understanding of the patient's narrative longitudinally.  How does the present admission fit into the course of the patient's illness and entire life?  How would resuscitation or an ICU course fit into that narrative?
  • Less patient trust of the provider.  (Assuming the patient has a good relationship with their chronic provider)
  • On the other hand, the hospitalist admitting the patient might be able to predict with greater certainty the circumstances under which a patient might code (eg Patient now has acute renal failure with a K of 5.8) so the conversation could be more concrete.
  • There is little control over the setting in which the conversation occurs. 
Understanding the effect of psychological framing is a key to understanding the appropriate conduct of these conversations.  Framing is the basis for many of the recommendations outlined in Table 3 and also helps explain the importance of having some understanding of the patient's narrative when approaching this conversation.  It's also important to understand that the "frame" can be communicated over the course of an interview (or multiple interviews).

Code status conversations are not part of the Review of Systems.  It's not another box to check or part of "Disposition" in the assessment and plan.  If the conversations are considered as such, there is a very high risk that the frame will be neglected leading both the provider and patient away from factors important to consider. 

There are little things that any physician can do even at 2 am to help build the frame regardless of whether you plan on discussing code status during that first visit, such as:
  • In every patient with a potentially life-limiting illness (or their family member if patient is delirious), ask at least one question that addresses their understanding of their disease/prognosis, how the disease has impacted them, and possibly also what they expect/hope for in the future.  This is most appropriately done while discussing the history of present illness or past medical history.  During a palliative care consult, this will normally lead to a longer conversation.  During a hospital admission, that luxury of time isn't usually there, and the patient may need to be told, "There are some important issues here that we'd like to try to address while you're in the hospital."  These types of questions will not only help you determine how to approach code status discussions, but also the overall care of the patient. 
  • If a patient clearly understands that their prognosis is very poor and indicates that comfort is a major goal, that's an invitation to recommend a DNR (even during the first conversation).  In other words, why even burden the patient with a question about whether they want resuscitation measures thus running the risk of making them feel like they would be rejecting a therapy that you are offering to them?  Just recommend DNR with a clear explanation for how DNR fits with the person's goals.  This may not be the only case where a DNR recommendation is appropriate, but if there is prognostic or goal of care uncertainty during the first visit, recommending a DNR on that visit may be premature.  However, it will likely still be appropriate to educate on the benefits of resuscitation measures (which are usually less than the average patient assumes.)
  • Regardless of your recommendation, if you discuss code status during the initial evaluation (or really whenever you discuss it), frame the recommendation in the context of the goals of the hospitalization/overall goals of care and how you're going to go about achieving those goals. 
  • (EDIT prompted by comments: Of course, always ask about the presence of advance directives and clarify who the patient's surrogate decision-maker is.  When the patient has an AD, it usually helps to ask questions about how the AD came about IE what prompted the patient to make their particular directive, what were they concerned about when they completed the paperwork, etc.  Patients will often bring up the experience of a loved one who has died, which raises another point: As the provider elicits the family history, this is another part of the frame.  eg After asking "How long ago did your mother die?" consider the follow-up question "what was her death like?")
The default full code status of hospitalized patients is a unique scenario for patients (outside the emergency medicine realm) where consent for a major procedure is implied.  Not only is consent implied, but there's no way for a patient to withhold assent once the conditions come about where resuscitation measures are indicated.   I remind trainees that the default to full code exists because if you don't initiate ACLS immediately, there will be little chance of the measures working for any patient.  The default does not exist because every patient wants to be coded or because it's necessarily appropriate to code every patient.  Sensitive communication (double entendre) at the time of admission should be aimed to identify patients for whom a resuscitation attempt might "break the heart" of the patient (if they were to find out about it) or family.  Such patients include a) those who would not want to receive these measures under any circumstance and b) those who might be predisposed to say "do everything" to the question in the first paragraph but who might readily change their mind if provided information on outcomes. (ie People who have the misconception that after resuscitation, they'll stand a good chance of sitting up eating ice cream the next day and be on their way home on the following day.)

There's of course another group of patients for whom resuscitation measures are unlikely to be beneficial but who still say "put me through it."   It would be foolish for us to expect the code status issue to be fully resolved upon admission although an urgent palliative care referral would be appropriate if death appears imminent.  A palliative care referral will hopefully guarantee excellent support for the patient, family, and providers regardless of the course of therapy.

Indeed the results of the study reflect only two hospitals.  While there is likely variation throughout the country and certainly amongst individual providers, I'm guessing that the dedicated readers of this blog will agree that improvement is necessary across the board.  How do you think we should get there?

Pallimed | Blogger Template adapted from Mash2 by Bloggermint