Hospital safety and complexityBMJ 2011; 342 doi: https://doi.org/10.1136/bmj.d1320 (Published 01 March 2011) Cite this as: BMJ 2011;342:d1320
All rapid responses
While a gap of two years may not be the most speedy of rapid responses it is perhaps relevant to note that in a recent article on optimum occupancy in psychiatric hospitals I have attempted to bridge the gap between the queuing theory approaches to occupancy and the use of occupancy as a measure of busyness (1).
It would seem that the supposed industry standard of 85% occupancy actually represents close to a maximum sustainable occupancy to avoid adverse patient outcomes rather than the 'optimum'.
Hopefully this somewhat belated answer to your question may be of wider usefulness.
1. Jones R. Optimum occupancy in psychiatric hospitals. Psychiatry On-Line 2013. Available at: http://www.priory.com/psychiatry/psychiatric_beds.htm
Competing interests: The author provides consultancy to health care organisations
It is all too easy to describe problems in hospitals as the result of
the fact that hospitals are "complex systems". But sometimes problems
arise because of a lack of willingness to adopt simple solutions, not
because the problem is too complex to solve.
Too many investigators look at the problems of bed management in
hospitals as being the result of a collision of outside pressures and
factors the hospital can't control (random arrivals of patients,
constrained budgets for additional beds). Wouldn't it be better, they say,
if there were more beds so we could avoid the problems of cross infection
of patients and cancelled operations that arise when the beds are too
Few seem to recognise that most of the problems are self-inflicted.
Most of the factors that create high bed occupancy are entirely within the
control of the hospital (not that they are actually under control).
Occupancy is a function of the patterns of arrival and discharge. The
pattern of discharges is mostly under the control of hospitals, and many
of the arrivals are also controllable. Emergency arrivals are--obviously--
not controllable, but are mostly very predictable. What I have seen in the
UK is a broad failure of hospitals to attempt to control anything leading
to a train crash of poor coordination over demand for beds.
The need for emergency beds typically peaks between midday and mid
afternoon. But elective patents are often told to arrive first thing in
the morning. More importantly few are discharged until late in the
afternoon. I estimated that in some UK hospitals, moving just 30% of
discharges to the morning would leave perhaps 10 percentage points (ie bed
utilisation would go from 85% to 75%) more free beds during the day when
they are required.
And we also know that we keep too many people in beds when they don't
need to be there. Audits based on clinical notes (by the Oak Group or
McKesson: see www.oakgroup.co.uk or www.mckesson.co.uk) suggest that, in
typical english hospitals, about half the patients don't need to be in a
hospital bed. Few health systems have used this knowledge to drive change
or lower the bed utilisation for those who do need it.
So high bed utilisation isn't the problem at all: it is a symptom of
a failure to coordinate or manage the arrival and discharge of patients.
The supposed side effects of high utilisation (cancelled operations,
hospital acquired infections...) are actually caused by poor coordination
of care. And if we coordinated the flow of patients well, we wouldn't have
to worry about the "optimal" level of bed utilisation.
Competing interests: I have offered advice on bed occupancy to national health bodies in the UK.
I thank Dr Callanan for his comments about my letter.1 In a 250 word
letter with a maximum of 5 references it is necessary to be brief.
Unfortunately, no reference to bed-occupancy was able to be provided.
For some time we have been interested in the effect of bed occupancy
on transmission of multiple antibiotic-resistant organisms (MROs). It is
difficult to do this in hospitals that regularly have average bed
occupancy exceeding 90%. We are therefore forced to use "overcrowding" as
a proxy e.g. access block, cancellation of elective admissions or surgery
due to insufficient theater time or ICU beds, numbers of patients in
outlying wards. There is a great deal of evidence linking "overcrowding"
and adverse events (see references quoted in the references below).
We have described some episodes where access block to an infectious
diseases ward appeared to promote MRO transmission.2 In a recent study
involving a Bayesian Network3 we found that "overcrowding" did not appear
to promote MRSA transmission above endemic levels in one hospital.
Outbreaks appeared to be related to a complex multidimensional threshold
involving "overcrowding", hand hygiene, MRO prevalence, screening
intensity and isolation bed availability, and probably other factors. We
appeared fortunate to be on the "safe" side of the threshold, the nature
of which we are trying to understand. It seems probable that
"overcrowding" may be more important with new VRE isolates and this may
make sense because this organism is capable of prolonged survival on
environmental surfaces. VRE has become a major problem for many hospitals
and we are extending the Bayesian Network analysis to this organism.
There has been considerable recent interest in bed occupancy in
Australia4 (see also the articles that have cited this article). Obviously
one figure for "ideal" occupancy is unlikely to fit all hospitals or even
a single institution at all times. For example, a busy public hospital
that is required to take all comers and is adjacent to a large freeway
containing many overpowered motor vehicles is likely to differ from a
private hospital in an affluent suburb, although the schedules of the
latter may increasingly be disrupted by non-elective surgery, particularly
So what sort of bed occupancy should be aimed for? The answer appears
to be "it depends". However, faced with day to day reality, the central
authorities controlling large public hospitals need to make a decision and
80% to 85% is widely, if anecdotally, quoted although the 85% level
apparently arises from earlier work on queues.4 Moreover, it appears to
have the support of several experienced clinicians to whom I have spoken.
As an aside, it is probable that the cost of lowering higher bed occupancy
levels would be repaid substantially in reduced adverse event rates (the
cost of treating potentially preventable adverse events is substantial)
and better staff management. We need to know the true cost of re-work in
public hospitals that have become highly "efficient" and this should
include costs to patients (who may require extended convalescence on
welfare) and society as well as to the hospitals.
Complex System research may of course be refining this 80% to 85%
level for particular types of hospitals.4 However, it seems unrealistic,
at least in the foreseeable future, to be able to run a complex computer
program in a busy public hospital at intervals to determine optimum bed
occupancy, and then to implement it.
Since the mid-1980s we have been required to become more "efficient"
and "productive". A variant of Baumol's Cost Disease that describes the
economic relationship of technological change and complexity tells us that
this is not so easy; cutting bed numbers to promote "efficiency" may have
unintended and perhaps unforeseen consequences. Now Complexity and Network
Science tell us that sustainability and resilience are most important,
that some redundancy is essential for resilience, and that as we become
increasingly efficient we simultaneously become increasingly vulnerable to
failures. We are thus faced with a dilemma that cannot easily be resolved.
It is made even more difficult by our worship of an unsustainable economic
system based on growth, consumption and debt that causes environmental
damage and appears to be nearing bankruptcy.5 Still, lowering average bed
occupancy in busy public hospitals to an average of, say, 85% may still be
feasible and very worthwhile.
1. Morton A "Hospital safety and complexity" British Medical Journal
2. Morton A, Whitby M and Looke D "Isolation ward access block"
Healthcare Infection 2009;14:47-50.
3. Waterhouse M, Morton A, Mengersen K, Cook D and Playford G "Role
of overcrowding in methicillin-resistant Staphylococcus aureus
transmission: Bayesian network analysis for a single public hospital"
Journal of Hospital Infection (in
4. Bain C, Taylor P, McDonnell G and Georgiou A "Myths of ideal
hospital occupancy" Medical Journal of Australia 2010;192:42-43.
5. Jackson T "Prosperity without growth" Earthscan, London 2009.
Competing interests: No competing interests
I wonder if the evidence for optimal bed occupancy is there - in this
article, is it one of the few statements that is not referenced. I have
yet to see convincing proof of this view but would be delighted to see the
evidence of the statement.
Redundancy is often a required feature for any complex system to work
successfully but the exact figure for such redundancy particularly in a
"many product, customised production"model appears more "perceived wisdom"
than "gospel truth".
Competing interests: No competing interests