Emergency medical admissions can be predicted
Dear Sir: re. Emergency admissions: a journey in the right direction?
The authors introduce their report with the following observation;
‘Emergency admissions to hospital are, by definition, unpredictable and
unexpected in the individual case, even where the system has been properly
set up to cater for them.’ This is self evidently correct. Having a
system properly set up to cater for individuals in a timely manner
requires an ability to predict demand upon whole services. In our
hospital we have a system that predicts the emergency medical admissions
on a daily basis.
Our Trust comprises a district general hospital (Blackpool Victoria
Hospital, BVH) of 937 beds and four other sites with a further 206 beds.
At BVH 286 beds are currently available for emergency medical admissions.
The Accident and Emergency department and Medical Assessment Unit (MAU)
are situated on BVH site. All emergency medical admissions to the Trust
pass ultimately through the MAU. We have collected data for emergency
medical admissions for the past nine years from the hospital information
system. The median (range) number of daily admissions in the past year was
38 (16 to 68).
Whilst the daily and weekly admission figures within any given year
do not appear to follow a pattern the number of admissions on
corresponding days each year is broadly reproducible: e.g. the number of
admissions on the Monday nearest to 1st April each year is very similar to
that from previous years. By using annual data we have constructed an
iterative model that allows us to calculate the mean (+/- standard
deviation) medical admissions for any given day of the year. Any
deviation from the model provokes an investigation using the principle of
‘special cause’ from statistical process control methodology. When we
plot real time data against our model we find a very good correlation. The
actual medical emergency admissions have been within the model’s
predictions for 301 of 365 days in the calendar year ending 31st October
2007. For the 64 days when the model was inaccurate the median excess
number of emergency admissions was 4.
We hope to roll out this model to the clinical management team for
daily use in the near future. By using data for elective admissions to
other specialties (which are readily available) we shall be able to
predict when there are likely to be pressure points in the system and so
manage them proactively. This will allow us to prevent any cancellation
of routine work and permit planned opening of extra beds in preparation
for times of increased medical admissions.
MJO'D was clinical lead for the Improvement Programme for Health at BVH. VJE is Service Improvement Manager at BVH.
Competing interests: No competing interests