Latest SNAP-2 manuscript published in PLOS Medicine
The latest paper from the SNAP2:EPICCS study has been published today in PLOS Medicine and is available open access from the Journal website.
You can view the open access paper here.
This study included 22,631 adult patients undergoing surgery in 271 hospitals in the UK, Australia and New Zealand. It focused on comparing the accuracy of objective risk prediction models against the clinical judgement of the perioperative care team on the day of surgery.
The study found that the Surgical Outcome Risk Tool demonstrated excellent discrimination (so it was good at "lining patients up" in order of risk) but poor calibration (it over-predicted risk, particularly in higher risk patients). Clinicians were similarly good at discrimination but even more pessimistic in terms of calibration. A new prediction model was developed and validated, combining clinical judgement with the SORT, and this demonstrated excellent discrimination and calibration.
Chief Investigator Ramani Moonesinghe said:
"There has long been controversy about the relative value of clinical judgement versus objective assessment of short-term postoperative mortality risk. We now feel confident recommending the Surgical Outcome Risk Tool over the more complex P-POSSUM model for objective risk prediction, and are delighted to publish a new, highly accurate model which blends clinical judgement with the SORT. The SORT-clinical judgement tool can be used to inform shared decision making conversations with patients, communicate patient risks with colleagues, and support patient selection for postoperative enhanced or critical care. Hopefully this new risk assessment tool will be widely adopted and support clinician and patient decision-making when contemplating and planning major surgery.
As always, we are indebted to over 3000 collaborators, and in particular the national trainee leads, Danny Wong (UK), Scott Popham (Australia) and Helen Lindsay, Lisa Barneto and Andrew Marshall Wilson (New Zealand)."
Share this :
