Aims Acute upper gastrointestinal bleeding (AUGIB) is a medical emergency with a 10% mortality
risk. Dividing endoscopy services into two separate sites: elective (‘cold’ site)
and emergency (‘hot’ site) is a novel approach that avoids competition for endoscopy
slots, thus optimising patient outcomes. We aimed to describe time to endoscopy and
outcomes at an exclusively ‘hot site’ and identify predictors of delayed endoscopy
(> 24 hours).
Methods Retrospective study of all emergency gastroscopies performed at a ‘hot’ London-based
endoscopy site between 6/9/2018 – 8/5/2019. The ‘hot’ site provisions a session of
endoscopy dedicated to inpatients during Monday to Friday, and on-call emergency theatre
access during weekends if required. No elective endoscopies are performed at this
unit. A decision tree classifier was used to select features contributing to delayed
endoscopy and a multiple logistic regression model utilized these shortlisted factors
to assign values of importance with regard to delayed endoscopy.
Results 151 gastroscopies were performed during the study period (132 new admissions (87.4%),
19 in-hospital bleeds (12.6%)). 55% of patients had an endoscopy within 24 hours.
Tab. 1
Endoscopy times: Admission vs Inpatient AUGIB
|
Admission with suspected AUGIB
|
Inpatient suspected AUGIB
|
p-value
|
|
Mean Time to endoscopy (hours)
|
34.1 (Range: 1.8–175.2)
|
49.5 (Range: 1.1–336.1)
|
0.03
|
Multiple logistic regression found that patients admitted with anaemia (p < 0.001,
OR 1.54) and weekend admissions (p < 0.005, OR 1.34) were independent predictors for
delayed endoscopy.
Conclusions Almost 90% of emergency gastroscopies at our ‘hot’ site are performed for direct
AUGIB admissions. Those admitted with suspected AUGIB undergo endoscopy sooner than
those with an inpatient bleed. Primary presentation with anaemia and weekend admissions
are associated with delayed endoscopy. The majority of patients at our exclusively
‘hot’ site did not have a delayed endoscopy.
