Work has been undertaken to produce atlases of hospital admission rates across
England. This was originally a joint project between the Unit of Health-Care Epidemiology,
Oxford University, and the South East Public Health Observatory. Now part of the
work of the Applied Health Research Unit, we hope to extend the atlases in due course.
Examples of the atlases are now available on this website. A particular feature
of the atlases is that they include analyses of linked Hospital Episode
Statistics. Thus the tables and maps include not only episode-based
admission rates but also person-based admission rates (counting people
who have several admissions as 'one person' in the person-based
rates).
Geographical units of analysis. The production of rates at the level
of local authority area is currently our main approach.
However, we have built into our analytical methods the option
also to produce tables and maps for strategic health authority
areas (which, with the merger of strategic health authorities,
we may modify to county level), and for Government Office
regions. The choice of geographical units is likely to vary
according to how common or rare the disease or operation
is.
Episode-based
and person-based rates.
Counting episodes and people using national linked HES creates
choices which include:
- whether the episode-based rates should be based on finished
consultant episodes (FCEs) or continuous inpatient spells
(CIPS). Depending on the circumstances, each has its
own strengths and weaknesses.
- time-frames
for counting 'people'. For example, whether
to count each person once, and once only, if they are
readmitted within a year; or whether to count people
once, and once only over a longer period of time, e.g.
five years. Again, there are some advantages and disadvantages
of different ways of counting 'people'.
- how to count people and reference them to a geographical
area of residence if the person moves home. Shall we
count them as one person, and one only, and assign them
to the original area of residence? Or, if they move,
shall we count them as a different 'person'
from each of two different places of residence? Again,
there are advantages and disadvantages of each method.
Principal diagnosis, or diagnoses in any position on the HES record. First operation or operations at any
point on the record. We have built in the option to analyse
any of these.
Source
of admission. For some conditions, it can be useful
to distinguish between admission rates for elective admissions,
and those for emergency admissions. When appropriate, we
will produce atlases for all admissions, elective admissions,
and emergency admissions.
For simplicity, in our first presentations,
we have chosen one set of selection criteria. We then
propose that, for some conditions, we produce sets of atlases
covering the different selection criteria for the condition.
These production runs of the atlases should be regarded as experimental
and exploratory. For example, our selection of specialties (and
ways of grouping them) and of diseases and operations (and ways
of grouping them) can easily be changed. We have included some
uncommon diseases, as well as common ones, to see what we find.
We do not expect that every user will find every set of results
equally interesting or important. We will learn as we go, and
as we receive feedback, but the aim is to be reasonably comprehensive
in a first round of analyses.
We are currently consulting with colleagues in surgery to construct
a full list of surgical operations to study.
The data are from national HES. Annual HES datasets have been kindly
provided by the Health and Social Care Information Centre each year since
1998. At the time of writing in March 2014, the Information Centre has not
yet been able to provide HES for 2012 and 2013. Neither we, nor we suspect any other
people, are in a position to understand all the local quirks that
may exist in the data for each and every location. We therefore
invite users to be cautious in their interpretation of local values
and to consider seeking advice from local experts on the likely
reliability of, or known local quirks in, local HES data.
The Atlases are in PDF format which will require Adobe Reader to view.
