Surviving Sepsis

“In April 2008, a 70 year old, independent lady with no previous comorbidities. became a grandmother for the first time and was looking forward to watching her family grow up. She developed a cough & became breathless and presented to her local hospital. She was admitted, and developed severe sepsis & septic shock secondary to her community acquired pneumonia and died within 7 hours. Her sepsis was not recognised, and antibiotics & fluids were not given in a timely manner. The patient’s family and the well-meaning and competent medical and nursing team were devastated.”

So begins the account of a real patient story that compelled Dr Matt Inada-Kim and colleagues to tackle the problem of managing sepsis within their practice.


  1. Dr Ken Catchpole says

    This is a great project, which Signal Detection Theory might further inform.

    A fair proportion of “failures to rescue” and the ‘deteriorating patient’ problem are not because we don’t know that a patient is really sick or that we don’t deliver treatment when we raise the alarm – but in going from one to the other; and specifically, in the penalty for ‘false alarms’. Signal detection theory tells us that yes/no (‘should we raise the alarm?’) decisions on the presence/absence of a ‘signal’ (sepsis/not sepsis), are based not only on how we make the diagnosis but on the rewards or penalties for doing so (which sets our underlying propensity to raise the alarm). If we expect certain patients to be septic we are more likely to call it (regardless of what the symptoms say); but if we are heavily penalized for calling ‘sepsis’ when the patient turns out be to not, we are less likely to call it (again, regardless of symptoms). The higher the penalty for such ‘false alarms’, the more evidence we will need and the sicker the patient will become before we raise the alarm. That’s partly why we wait so long to call it. So achieving better response is as much about rewarding raising the “sepsis” alarm, regardless of whether the patient is really septic or not, as it is about diagnosis or response.

    In part this project is about improving the response once identified, which is great – and that alone may have discouraged the normal “why on earth did you call me?” penalty for false alarms; but also increased the expectancy around having a sepsis Dx per se.

    It would be interesting to see how this intervention affected the number of false alarms. I suspect it will have increased them – which means that unless those ‘false alarm’ rewards and high sepsis expectations continue, it will slip backwards again. This happens when clinicians become frustrated when continuously called to a patient for “no reason”, or when time or cost pressures discourage the response. Encouraging those false alarms is essential if you want to detect more genuinely septic patients in a timely fashion.

    • Tod Guest says

      I think that is such an important point which is relevant time and time again in incidents regarding delayed treatment of the deteriorating patient. Providing simulation training for junior teams on these kinds of cases is in my view effective at teaching the process of care and the prioritisation and urgency of treatment interventions, but when it comes to ‘raising the alarm’, this is hampered by the challenge of engaging senior clinicians in addressing the problem raised here of false alarm penalties.

      I will be trying to raise the profile of this and integrate it into my training efforts.

Leave a Reply

Your email address will not be published. Required fields are marked *