Monday, July 29, 2019

"I'm Not Comfortable With"....Analyzing Decisions Involving Risk

A woman with upper gastrointestinal bleeding is admitted to the ICU with significant hemoglobin drop from baseline several months before.  The INR was >10 on admission, presumably reflecting sphlanchnic hypoperfusion during the bleeding episode (decreasing clearance or increasing effect of warfarin) in addition to consumption of clotting factors during bleeding.  The INR is reversed with vitamin K, and an EGD is performed showing a nonbleeding vessel which was clipped.  The following day, the patient is up and walking around and eating a full liquid diet with no further evidence of bleeding.  MAPs are in the mid to low 60s, and she is "called out" for a transfer to the regular medical ward.  The resident is reportedly "not comfortable with" having the patient on the floor with the low mean arterial pressures.  How best to analyze the situation?

(I will only briefly note that the expression of discomfort describes an emotional reaction that may or may not be concordant with a logical and factual analysis of the situation at hand.  In this case I suspect it has something to do with the availability heuristic, where dramatic events [gushing blood] are estimated to have higher probabilities than they warrant.)

The discomfort seems to stem from a concern that the patient will deteriorate on the regular medical ward and require interventions that are not available there, prompting readmission to the ICU and incurring the risk of a missed opportunity to provide necessary care during any delays in executing the transfer.  Factors that may raise that concern are marginal hemoglobin values (7-8) especially considering a baseline of almost twice that several months before, the size of the GIB and the resulting hemodynamic instability, the marginal MAPs at the time of transfer.  Mitigating factors are the absence of ongoing bleeding, the inference that any marginal MAPs must be from a residual volume deficit (she does not have coincident sepsis), that the vessel was successfully clipped, and the patient is up and walking about in spite of the measured MAPs.  But how are we to integrate this into an estimate of the risk that may be incurred from transfer?

One approach is to estimate the base rate of a "bounce back" to the MICU following a transfer to the medical ward after GIB or any other diagnosis for that matter.  Is it 90%, 50%, or 10%?  It is certainly not 90% or 50%.  It may be 5% (one in 20 patients) or even 20% (one in 5 patients) but it is unlikely to be higher than that - else there would have been meetings to address a too high rate of readmissions.  Let's say it's 10%.

A clinical deterioration rate requiring MICU readmission of 10% may or may not be too high.  What is important to note is that the rate provides the upper bound for the amount of benefit of allowing the patient to remain in the MICU for an additional day.  That is, even if the rate goes to 0% by the next day, the maximum amount of benefit in avoiding clinical deterioration on the medical ward is 10%.  Although most rebleeds happen early, not all do, and it is more likely that if there is rebleeding upwards of 50% of it will happen over the next several days, rather than all of it in the subsequent 24 hours.  If this is a reasonable estimate, then the maximum benefit of another day in the ICU is saving/avoiding about a 5% risk of clinical deterioration on the floor.  The marginal risk reduction decreases the closer you get to zero.  Note also that some residual risk remains until and even after discharge from the hospital for several weeks or months - risk rarely goes to zero, so we will always have to accept some risk.

Other factors to consider are the consequences of a clinical deterioration.  That the patient tolerated such a large bleeding episode before treatment provides some reassurance that she will tolerate it again, after treatment which can be inferred to decrease both the probability and the severity of a recurrence (since the INR is no longer 10, the vessel is clipped, etc).

The point is, we can take our feelings of being uncomfortable, rather than as a guide to what to do, as a prompt to consider the decision more fundamentally to determine if there are ways of rationally appraising the situation.  In this and many cases there are, if we take a moment to do some back of the envelope calculations about probability, risk, uncertainty, and consequences.

The patient did not, in fact, deteriorate on the regular medical ward and was discharged to home in excellent condition.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.