Managing Variation Lessons From Uk Walleyache

Managing Variation in Demand:
Lessons from the UK National

Health Service
Paul Walley, M.B.A., lecturer in operations management, University of Warwick,
Coventry, England; Kate Silvester, M.B.A., FRCOphth, Osprey Programme Coach,
West Midlands Strategic Health Authority, Birmingham, England; and Richard Steyn,
MS FRCSEd(C-Th) DRGOC, cardiothoracic surgeon, Heart of England Hospital,
Birmingham, England [First Page]
[309], (1)
................................................................................................
Lines: 0 to 36
E X E C U T I V E S U M M A R Y
———
Managers within the U.S. healthcare system are becoming more aware of the im- * 41.60103pt PgVar
pact of variation in demand on healthcare processes. The UK National Health ———
Service provides a prime example of a system that has experienced the conse- Normal Page
quences when the issue is not dealt with satisfactorily, having suffered from ex- * PgEnds: Eject
cessive queues for a prolonged period. These delays are mostly caused by a lack
of attention to variation and inappropriate responses to the queues, rather than a [309], (1)
capacity shortage.
A number of collaborative programs recently have come to grips with many of
the causes of the queues in both elective care and emergency care. Although there
are still areas that need large-scale improvement, good progress has been made,
especially within emergency care. The authors of this article have acted as technical
advisors to a number of these improvement programs and have been able to doc-
ument many of the practices that have helped to reduce or eliminate unnecessary
queues and delays across the 200 sites in England that have 24-hour emergency
care facilities. Local program managers at these sites continuously reported progress
for a period of 18 months. A number of important lessons for both the design and
control of healthcare processes have emerged from the collaborative work. These
lessons focus on understanding and measurement of demand, capacity planning,
reduction of introduced variation, segmentation and streaming of work, process
design, capacity yield management, and measurement of variation.
For more information on the concepts in this article, please contact Mr.
Walley at paul.walley@wbs.ac.uk. Part of this work was funded by a research
grant to Paul Walley at the University of Warwick.
309
Photocopying and distributing this PDF of the Journal of Healthcare
Management
BookComp/ is prohibited
Health Administration Press/ Journalwithout the permission
of Healthcare Management of 309
/ Vol. 51, No. 5/ Page Health
1st proof
Administration Press, Chicago, Illinois. For permission, please fax
your request to (312) 424-0014 or e-mail hap1@ache.org.
J o urna l o f H e a lt hca re Ma nagement 51:5 September /October 2006
I n an article, Noon, Hankins, and

Cote (2003) ably demonstrated the
impact of variation in demand and
to combine the learning from both
elective care and emergency care. The
management of variation has since
treatment times on healthcare systems. become one of the improvement
Where variation occurs in congested priorities within the NHS (NHS
systems, queues are likely to form. If Modernisation Agency 2004). This
the system is operating very close to article highlights the lessons learned
capacity or if the degree of variation about managing variation from this
is high, these queues are likely to get collaborative work.
out of control, resulting in reduced
service efficiency and service quality. Q U E U E T H E O R Y Q U I C K LY
The UK National Health Service (NHS) REVISITED [310], (2)
is a classic example of a system that Queues form most commonly when
has seen how queues can get out of arrival patterns or service times have a
Lines: 36 to 50
control, partly because staff historically random element. Noon, Hankins, and
———
had very little knowledge about the Cote (2003) provide typical examples
0.0pt PgVar
causes of the queues. At one point in of such occurrence, such as at clinics, ———
the 1990s, more than a million people same-day surgery centers, and intensive Normal Page
were on record as waiting for treatment care units. The level of queuing is PgEnds: TEX
within the NHS, and probably more influenced by a number of factors,
patients were ill but not registered including the distribution of arrival
[310], (2)
on the waiting lists. Emergency care patterns, the spread of service times,
systems suffered as badly, with patients and the size of the service system.
who needed emergency admission Generally, the greater the variation in
trapped in queues in emergency arrival or service time, the more likely
departments. The phenomenon of it is that there will be a queue and the
“trolley waits”1 became widespread. expected queue will be long. As systems
Patients often had to wait for days start to work closer to their effective
before they could be admitted onto an capacity limit (i.e., as they become
appropriate ward. congested), the queue length generally
In 2001 the NHS Modernisation becomes longer.
Agency started a series of collaborative At higher levels of utilization,
projects, based very much on the Insti- queues grow rather rapidly, and the
tute for Healthcare Improvement (IHI) situation can get out of control. In
model, to improve parts of the system. hospital wards, for example, the
The Emergency Services Collaborative scale of the system also influences
(ESC) managed the improvements the likelihood of queues (Kolesar and
in emergency care within the 200 Green 1998). For a large emergency
hospitals in England that have 24- hospital with no elective patient
hour emergency departments. This was streams, average utilization of ap-
followed by a broader initiative, the proximately 80 percent should not
Improvement Partnership for Hospitals, cause unacceptably low bed availability
which has taken similar approaches (Bagust, Place, and Posnett 1999).
310
Management
of Healthcare Management of 310Health
/ Vol. 51, No. 5/ Page 1st proof
Managing Var iation in Demand
However, a pediatric unit with 40 acute flow through in a factory-like manner

beds could not operate effectively with with few in-process waits. However,
such a high general level of utilization we can readily see an immediate
due to its smaller scale (60 percent contradiction. Focused facilities
utilization would be more common). deliberately create smaller pools of
Small intensive therapy units must demand, thus increasing the input
operate at even lower levels of effective demand variation. We must therefore
utilization, unless they are prepared to be very careful when we divide demand
divert patients elsewhere on a regular into streams to achieve operational
basis. efficiency, ensuring that the operational
benefits are not counteracted by a
FOCUS OR CARVE OUT lack of demand or resource pooling. [311], (3)
Queue design can inadvertently make Similarly, we must not put two
queues unnecessarily worse. It is incompatible demand sources together,
Lines: 50 to 60
usually better to have one queue being ostensibly to reduce variation, only
———
dealt with by many servers than to split to find that the in-process variation is
0.0pt PgVar
the demand into multiple queues and too high. In the United Kingdom, we ———
dedicate individual servers to those now recognize that demand-splitting Normal Page
queues (as long as this has no effect on does not necessarily coincide with an PgEnds: TEX
service times). The multi-queue system increase in efficiency. Thus, it is referred
will normally experience longer average to as “carve out” (i.e., unnecessary
[311], (3)
waits because capacity is more likely ring-fencing of resources that increases
to be wasted. For instance, people variation). We must also be patient-
sometimes join the wrong queue and focused enough to recognize that some
get stuck behind someone who takes variation is necessary, and so some
a long time, while other resources are in-process variation might be tolerated
idle because no one has gone to a free (McLaughlin 1996).
server. This is why resource pooling is
regarded as an effective way of dealing METHODOLOGY
with some types of variation. An action research program was set up
One other way to lessen queue in July 2001 to study the ideal design
problems is to reduce service time of emergency care. Two healthcare
variation. This can be achieved, in communities (with a catchment of
theory at least, by streaming patients approximately 500,000 people in
into groups that exhibit similar service each) were studied in depth. The study
characteristics. Taken to an extreme, we examined demand for unplanned
can get “focused” healthcare processes, healthcare—specifically, how to stream
such as surgi-centers or the classic demand so that effective processes
example of the Shouldice hernia repair could be designed to meet the demand.
center (Yang et al. 1992; Heskett 1993). The early lessons were applied to eight
In these situations, in-process variation additional communities and then
is drastically reduced, as processes adapted for the national scale program,
are highly standardized and patients the Emergency Services Collaborative,
311
Management
of Healthcare Management of 311Health
/ Vol. 51, No. 5/ Page 1st proof
which lasted 18 months at each site. helped reduce the impact of variation
The final program closed at the end of on the system’s performance.
2004. The national program grouped
30 communities across England, and Really Understand Demand
they were given financial support and Before the collaborative program
advice on how to improve emergency started, demand was believed to be
care flows. All 200 sites with 24-hour random and therefore could not
emergency care facilities participated in be forecast. We have found that
the ESC program. not only is it possible to forecast
Local program managers submitted demand, it is also essential to make
monthly reports to the national appropriate plans for emergency
program team. These reports detailed demand. We found that time of day [312], (4)
the activities of each improvement and day of week drive emergency
team, and the data were rapidly room attendance to a great extent,
Lines: 60 to 78
synthesized to establish which types with weather and a few special causes
———
of improvements had beneficial (e.g., public holidays) also affecting
0.0pt PgVar
effects. At the end of the first wave arrival patterns. When clinicians and ———
of implementation, the program was service managers were given the data in Normal Page
formally reviewed using site visits and usable form, they were better able to * PgEnds: Eject
structured interviews with clinical and understand the concept of variation.
program personnel. In total, 80 people Once repeating seasonal demand
[312], (4)
were interviewed across 33 sites. patterns were recognized and the
Pareto principle was applied (i.e.,
80 percent of the consequences comes
LESSONS FROM THE from 20 percent of the causes), systems
EMERGENCY SERVICES could be designed to meet demand.
C O L L A B O R AT I V E There was also a small cultural shift,
The UK government has established a as staff moved away from feeling panic
series of performance targets for the during busier-than-average days and
healthcare system, rewarding hospitals began recognizing and accepting that
for performance above certain levels. some days are busy due to variation.
One of these is a four-hour wait target
in emergency departments. Emergency Maintain the Right Capacity Plans
departments are expected to assess, Problems occur when there is insuf-
stabilize, and admit to a bed or to ficient capacity to meet demand. It is
discharge 98 percent of patients within remarkable how few systems formally
four hours. Before the ESC started, tried to match capacity to predicted
on average, only about 65 percent of demand.
waiting patients got through the system
in four hours. The mean is now near Example: The daily emergency demand
96 percent across the 200 sites. The profile. At the first site we investigated,
following recommendations show how weekday demand followed a consistent
a number of simple principles have profile through a 24-hour period. There
312
Management
1st proof
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
FIGURE 1
Hourly Patient Arrival Pattern in a Typical UK Emergency Room
ER Daily Arrival Pattern
[313], (5)
% of Total per Hour
3
Lines: 78 to 103
———
2
0.0pt PgVar
———
1 Normal Page
* PgEnds: Eject
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
[313], (5)
Time of Day
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
was a morning rush of patients with the emergency room, where the peak
minor injuries and a slight blip in in capacity occurs four hours after the
demand after schools closed and on peak in demand. We should not be
into the evenings. Patients with major surprised that queues occur. However,
illnesses, usually requiring admission, staff did not always realize that the
tended to arrive more regularly during capacity plan was inappropriate. In
daylight hours, partly determined by some cases, additional staff members
ambulance availability. Figure 1 shows were deliberately introduced when the
the demand by hour. queues were at their worst—to manage
The capacity strategy for the the queues, rather than to prevent the
queues from happening.
emergency department misread
demand considerably because there is a Understand How Your System
tradition within the NHS of assigning Introduces Variation
staff on early and late shifts to cope It is normally assumed that elective
with peaks and troughs in workload demand has less variation because it
on hospital wards. Using a ward-based can be reduced by careful scheduling.
shift pattern is quite inappropriate in However, the ESC program discovered
313
Management
1st proof
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
FIGURE 2
Example of Demand Variation Between Emergency Care and Elective Care
35
30
25
Admissions (Surgery)
20
15
[314], (6)
10
Lines: 103 to 134

5
———
0 0.0pt PgVar
———
01-Apr 08- 15- 22- 29- 06-May 13- 20- 27- Normal Page
Date * PgEnds: Eject
Elective Emergency
[314], (6)
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
that much in-hospital variation in of delays, or not ill but have not left
demand is actually introduced by the the hospital for some reason.
planning processes. We found some surprises when we
investigated the source of variation in
Example: Variation in emergency and elective demand. Figure 2 shows a comparison
admissions. In most UK hospitals between emergency demand and
some wards take both emergency elective demand variation. In most
patients and elective patients. Lack of
circumstances, elective demand
availability of beds is often a major
variation is greater than emergency
cause of delay for patients who need
demand variation because of the
emergency admission. Most hospitals
way in which operating theaters are
claim to work at average bed-occupancy
scheduled. Hospital planning systems
levels of more than 90 percent,
meaning that congestion and queuing create their own problems by making
are significant problems. Audits suggest elective schedules busiest on the days
that typically 60 percent to 70 percent when emergency demand is also
of the patients occupying those beds expected to be high.
are receiving active treatment and the When we compare admissions and
rest are either waiting for initial visits departures, we see that not only are
by doctors, stuck in the system because they partly out of sync, but that there is
314
Management
1st proof
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
FIGURE 3
Example of Demand Amplification in Healthcare Processes
Admissions vs. Discharges
200
180
160
140
120
Patients
100
[315], (7)
80
60
Lines: 134 to 155
40
———
20 0.0pt PgVar
0 ———
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Normal Page
Time (Days) PgEnds: TEX
Admissions Discharges
[315], (7)
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
a difference in the amplitude in the be discharged. There is no measured

variation. Hospitals succumb to a increase in readmission of patients
dynamic similar to the bullwhip effect, discharged just before public holidays,
which amplifies input variation (see indicating that the patients discharged
Figure 3) (Warburton 2004). The have not been forced out.
demand amplification is caused by The lack of discipline in discharge
practices such as batching or grouping procedures is significant, especially
patients in pursuit of efficiency. For in relation to the cost of care. Some
example, consultants in the ESC patients who are well enough to
hospitals often complete just two go home on a Thursday afternoon
discharge rounds per week because of might not be discharged until the
the time it takes to tour the hospital following week, adding many days
and discharge patients. The peak days to their treatment episode. The
for discharge are all before holiday recognition of this problem has also
weekends, which suggests that many highlighted one of the common
people have been stuck in beds, waiting mistakes within the system. Both
for the discharge rounds to occur managers and clinicians who survey
even though they are well enough to the scene in an emergency department
315
Management
1st proof
filled with 20 or so patients waiting on maximizes the opportunity to generate

trolleys for admission understandably efficiency gains through process
assume that the bed shortage is the redesign while limiting the effects of
root cause of the problem. However, demand pool reduction. Four patient
when more beds are added to the streams were introduced nationally for
system capacity, the problem can the ESC:
become worse. This is because a bigger
hospital requires longer discharge 1. Minor patient treatment process
rounds. Doctors compensate for the 2. Rapid assessment, monitoring, and
extra time needed by doing fewer discharge for patients who do not
rounds, thus increasing bed capacity need admission
variation. Consequently, the queues 3. Surgical admission process [316], (8)
in the emergency department become 4. Medical patient admission
worse, not better.
Lines: 155 to 183
Although these streams required some
———
Split the Demand Around Processes compromises, such as no specialist
0.0pt PgVar
Guidelines are needed to manage the streaming for mentally ill patients or ———
trade-off between demand pooling and patients with respiratory illness, the Normal Page
variation reduction through focus. We approach worked well. * PgEnds: Eject
discovered that demand splitting occurs Some hospitals have successfully
unnecessarily due to conventions in increased bed availability by organizing
[316], (8)
clinical practice. wards based on patient length of
stay (LOS). Although this may seem
Example: Queues for scanning processes. a slightly arbitrary way of splitting
One hospital within the collaborative demand at first glance, the approach
based its schedules around the skill sets works well. Wards develop their
of the radiographers, allocating each own protocols based on the natural
person specific slots to use the hospi- speed and intensity of the treatment.
tal’s computerized tomography (CT) Short-stay wards (one to three days
scanners. Patients were not allowed to LOS) have fast access to diagnostics
move between radiographers; rather, and rapid admitting and discharge
sessions were dedicated to specific tasks procedures. Their entire approach
(e.g., “simple heads”) and degrees of focuses on high volume and fast
urgency (urgent, 2-week queue, 8-week throughput of patients who will get
queue, and so on). We found that the well quickly after a short intervention.
hospital had split demand for CT into Medium-stay wards (typically 3–14
72 separate queues, making the system days LOS) deal with patients who
almost unworkable. By pooling these receive substantive treatment and who
queues, queue levels often naturally fall need time to recover; these wards
to below the waiting times reserved for can be designed to treat patients in a
urgent patients. steady but predictable manner. Long-
Splitting demand into a small stay wards (typically more than two
number of process-based streams weeks LOS) have the capacity to deal
316
Management
1st proof
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
FIGURE 4
The Redesign of Minor Patient Treatment
a) Conventional model (four in-process queues)
Wait Wait Wait Wait

Reception Triage Assess Treat Discharge
b) See and Treat model (one in-process queue)

[317], (9)
Reception Wait Assess, treat, and discharge

Lines: 183 to 208
••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ———
0.0pt PgVar
———
with complex discharge problems, the delays. First, minor injuries were Normal Page
multiple comorbidities, or and usually graded as nonurgent in the PgEnds: TEX
rehabilitation treatment. triage system, which divides the
demand into five categories based on [317], (9)
Design Processes to Absorb Variation urgency. Patients in the lowest category
The collaborative programs take a often gave up before being treated
“whole systems” view of healthcare. because the queues were too long.
Hospitals in the ESC try to optimize Second, patients were put through a
the whole system rather than optimize
multistage process involving separate
every subcomponent. Many existing
queues for reception, triage, assessment,
processes were divided up, with clinical
investigation, treatment, and discharge.
and technological boundaries that
Each stage of this process had service
patients needed to cross as they move
time variation that caused interstage
through the system. If these boundaries
queues. Figure 4 shows how this
can be removed by merging processes,
process was transformed.
total variation is reduced and the
One hospital consultant and
patients experience fewer queues.
her team developed their own
Example: “See and treat” for minor injuries. method of treating minor problems,
Patients with minor injuries were often now called the “see and treat”
experiencing very long delays in their approach. In this approach, patients
treatment. It was common for someone with minor conditions are booked
to wait for six hours to receive ten and streamed into an integrated
minutes of clinical attention (e.g., process; they are not left waiting
fitting a flexible bandage on a sprained during assessment, treatment, and
ankle). There were two reasons for discharge. The multiskilled staff absorb
317
Management
1st proof
service time variation because of the system failures. Delays can be caused
design configuration. In operations by any of the following:
management terminology, this redesign
has moved the process from a “long 1. Hospital-acquired infections
and thin” structure to a “short and fat” 2. Patients getting bedsores
structure, which is much more flexible.
3. Waits for inpatient diagnostics
Now that triage has been removed,
there is additional treatment capacity 4. Lack of physiotherapist/occupa-
because the highly skilled triage nurse tional therapist assessment
can now be used to treat patients 5. Planning errors
instead of managing patient flow. This 6. Not informing relatives of the
[318], (10)
process innovation has been one of the discharge plan
fastest ever to be adopted in the NHS.
7. Pharmacy closures after 5.00 p.m.,
Minor patient waiting times have fallen preventing patients from picking Lines: 208 to 255
by 70 percent. up their medication before they go ———
home 0.0pt PgVar
———
Look at Capacity Yield Losses Long Page
All of these quality-related PgEnds: TEX
A number of hospitals have now
problems reduce bed capacity, forcing
virtually eliminated bed shortages
the system to operate in a more
and have reduced the number of [318], (10)
congested manner. Our estimates
patients “outlying” on inappropriate
suggested that typically 25 percent of
wards. Ward staff set an expected date
bed capacity was used wastefully.
of discharge as soon as a patient is
admitted, with a project plan for what
Use Process Control Measures to
needs to happen before discharge can
Understand Variation
take place. The tasks to be scheduled
One of the approaches that the NHS
might include, for example, a visit
has imported from the United States
from an occupational therapist to
is the use of statistical process control
determine the patient’s ability to cope
(SPC) to understand variation at a
with his or her disabilities after going
local level. The IHI and authors such as
home. This makes the processes more Carey (2002) and Wheeler (2000) have
deterministic, and controlling variation provided a methodology that is prac-
is a consequence. tical enough to be implemented on a
The discharge planning document large scale. Although use has not spread
unexpectedly became a goldmine as far as we would like, SPC is used to
of information for the ESC. When
discharge dates are extended beyond • monitor demand changes,
the original plan, the document can
• measure performance over time,
be studied to understand the reasons
patients have been delayed. A high • understand the behavior of processes,
proportion of the discharge delays have • develop a body of knowledge about
been caused by quality errors and other systems, and
318
Management
1st proof
• measure the impact of small-scale that take variation into account. A

changes. key learning outcome is a better
understanding of the implications
In a few cases, management of demand pool splitting. Anyone
reporting has been transformed by involved in the design of healthcare
presenting performance data in SPC systems can usefully apply the lessons
charts to show sustainable trends in about demand streaming. There is
improvement, rather than presenting little point in constantly pooling
simplistic two-point comparisons of demand, because eventually managerial
last quarter with this quarter. diseconomies of scale will appear.
In addition, clinical specialization
CONCLUSIONS creates smaller pools of demand that [319], (11)
The discussion in this article identifies are difficult to manage because they
seven recommendations to help introduce greater variation. There is a Lines: 255 to 283
managers understand, measure, control, balancing point where processes have
———
and reduce demand and capacity a big enough demand pool to control 0.0pt PgVar
variation. The UK healthcare system variation, but the scope of the tasks ———
has a long way to go, but it is now within the process is narrow enough to Long Page
at a stage where a small number of be manageable. PgEnds: TEX
its hospitals have achieved progress
across their entire span of activity.
Note [319], (11)
Understanding the effects of variation 1. Trolley waits occur when patients who
and the appropriate responses to it is require admission to a specialist wait
not intuitive for either managers or on trolleys not beds in emergency room
clinicians. It is difficult to convey the cubicles or the hospital corridor.
message that queues appear even in
places where there is not a long-run References
capacity shortage and that throwing Bagust, A, M. Place, and J. W. Posnett. 1999.
extra resources at problems does not “Dynamics of Bed Use in Accommo-
dating Admissions: Stochastic Simula-
routinely solve them and can actually
tion Model.” British Medical Journal 319
make the situations worse. Actions (7203): 155–58.
taken to manage queues, such as Carey, R. G. 2002. Improving Healthcare with
clinical prioritization, generally make Control Charts. Milwaukee, WI: ASQ
Quality Press.
the queues longer. Like many other
Heskett, J. L. 1993. Shouldice Hospital Limited,
professional groups, clinicians can Case 9–683-068. Boston: Harvard Busi-
be slow to adjust their practices to ness School, Publishing Division.
respond to obscure concepts such as Kolesar, P. J., and L. V. Green. 1998. “Insights
on Service System Design from a Nor-
these.
mal Approximation to Erlang’s Delay
The NHS experience should serve Formula.” Production and Operations Man-
as a warning about what can go wrong agement 7 (3): 282–93.
if variation is not understood and McLaughlin, C. P. 1996. “Why Variation Re-
duction Isn’t Everything: A New Paradigm
managed. As part of this improvement
for Service Operations.” International
journey, the NHS has learned a lot Journal of Service Industry Management 7
about how to better design processes (3): 17–30.
319
Management
1st proof
NHS Modernisation Agency. 2004. Ten Warburton, R. 2004. “An Analytical Investi-
High Impact Changes for Service Improve- gation of the Bullwhip Effect.” Production
ment and Delivery: A Guide for NHS and Operations Management 13: 150–60.
Leaders. London: Department of Wheeler, D. J. 2000. Understanding Varia-
Health. tion: The Key to Managing Chaos, 2nd ed.
Noon, C. E., C. T. Hankins, and M. J. Cote. Knoxville, TN: SPC Press.
2003. “Understanding the Impact of Yang, S., C. P. McLaughlin, R. W. Vaughan,
Variation in the Delivery of Healthcare and J. J. Aluise. 1992. “Factory Focus in
Services.” Journal of Healthcare Management Hospital-Owned Ambulatory Surgery.”
48 (2): 82–98. International Journal of Service Industry
Management 3 (4): 63–75.
[320], (12)
P R A C T I T I O N E R A P P L I C A T I O N
Lines: 283 to 312
James B. Conway, CHE, senior fellow, Institute for Healthcare Improvement,
———
Cambridge, Massachusetts 0.795pt PgVar
———
Normal Page
A cross the United States, the healthcare system struggles with issues of demand
and capacity and associated waits, delays, and diversions. Hospitals closed to
emergency ambulance patients for parts of days are commonplace, and diversion
PgEnds: TEX
[320], (12)
has been referred to as an epidemic. The Institute of Medicine (2006) recently
reported that 45 percent of emergency departments reported going on diversion
at some point in 2003; overall, it is estimated that 501 ambulances were diverted
that year. Waits in emergency departments of five or more hours are not usual, and
patients routinely leave the emergency service without being seen. The most recent
national data available, collected in 2002 by the Lewin Group and reported by the
American Hospital Association (Lewin Group 2002), show the average time waiting
for transfer from an emergency department to an acute or critical care bed was
about 3.2 hours. In hospitals on diversion more than 20 percent of the time, the
wait was 5.8 hours. More than half of urban hospitals reported being on diversion
some times, and one in eight urban hospitals reported being on diversion 20
percent or more of the time. Many believe that additional resources must be added
to appropriately resolve the issue of capacity, and the need to address these issues
is receiving considerable attention.
John Kelsch, an executive at Xerox and a consultant for the Quality of Health
Care in America project, is reported to have said, “To do things differently, we must
see things differently. When we see things we haven’t noticed before, we can ask
questions we didn’t know to ask before.” This article is an important contribution
and a jarring reminder to us as leaders to look differently at things “we already
know.” The results this study presents from 200 hospitals with emergency depart-
ments are striking and are no doubt humbling for many readers. After 18 months
of intervention, data show that these emergency departments assess, stabilize, and
320
Management
1st proof
admit to a bed or discharge 96 percent of patients within four hours, an achieve-

ment of a performance target set at 98 percent. In addition, minor patient waiting
times have fallen by as much as 70 percent. A number of these hospitals have now
virtually eliminated bed shortages and outliers/borders on inappropriate wards.
The authors argue that achieving these results is a product of learning that
is not intuitive for either managers or clinicians. In fact, many seemingly logical
solutions to the problem actually make the problem worse. To address issues of
variation, the authors suggest that we must
1. really understand demand,

2. maintain the right capacity plans, [321], (13)
3. understand how the system introduces variation,
4. split demand around processes, Lines: 312 to 352
5. design processes to absorb variation, ———
6. look at capacity yield losses, and * 17.21751pt PgVar
———
7. use process control measures to understand variation. Normal Page
PgEnds: TEX
Failure to design our systems for variation and to optimize flow can signifi-
cantly compromise quality of care, allow for highly ineffective use of limited re- [321], (13)
sources, and be the source of enormous frustration for patients and family mem-
bers as well as staff. The authors provide an insightful resource on addressing
variation in the United Kingdom that has considerable applicability to care in the
United States. Their approaches are also reinforced by the recent emergency service
report by the Institute of Medicine (2006); it is an important addition to resources
already published in the United States (see references below). Now the industry
must “see things differently” and move comprehensively toward any number of
collaborative efforts to apply the results for transformational improvement. Our
patients, family members, and staff deserve nothing less.
References
Institute for Healthcare Improvement. 2003. “Optimizing Patient Flow: Moving Patients
Smoothly Through Acute Care Settings.” [Online information; retrieved 04/06/06.]
http://www.ihi.org/IHI/Results/Whitepapers.
Institute of Medicine. 2006. Hospital-Based Emergency Care: At the Breaking Point. Washington, DC:
National Academies Press.
Lewin Group. 2002. “Overcrowded Emergency Departments Leading to More Diversions,
Longer Wait Times.” [Online information; retrieved 04/06/06.] http://www.hospitalconnect
.com/aha/press room-info/releasedisplay.jsp?dcrpath=AHA/Press Release/data/AHAPRESS
RELEASES-0708x9&domain=AHA.
Litvak, E., and M. C. Long. 2000. “Cost and Quality Under Managed Care: Irreconcilable Differ-
ences?” American Journal of Managed Care 6: 305–12.
Litvak, E., M. C. Long, A. B. Cooper, M. L. McManus. 2001. “Emergency Department Diversion:
Causes and Solutions.” Academic Emergency Medicine 8 (11): 1108–10.
321
Management
1st proof
Litvak, E., P. I. Buerhaus, F. Davidoff, M. C. Long, M. L. McManus, D. M. Berwick. 2005. “Manag-

ing Unnecessary Variability in Patient Demand to Reduce Nursing Stress and Improve Patient
Safety.” Joint Commission Journal on Quality and Patient Satisfaction 31 (6): 330–38.
McManus, M. L., M. C. Long, A. Cooper, J. Mandell, D. M. Berwick, M. Pagano, and E. Litvak.
2003. “Variability in Surgical Caseload and Access to Intensive Care Services.” Anesthesiology
98 (6): 1491–96.
Rozich, J. D., and R. K. Resar. 2002. “Using a Unit Assessment Tool to Optimize Patient Flow
and Staffing in a Community Hospital.” Joint Commission Journal on Quality Improvement 28
(1): 31–41.
[Last Page]
[322], (14)
Lines: 352 to 361

———
497.75pt PgVa
———
Normal Page
PgEnds: TEX
[322], (14)
322
Management
1st proof

Managing Variation Lessons From Uk Walleyache

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Managing Variation Lessons From Uk Walleyache

Hochgeladen von

Copyright:

Verfügbare Formate

Managing Variation in Demand:

Lessons from the UK National

I n an article, Noon, Hankins, and

However, a pediatric unit with 40 acute ﬂow through in a factory-like manner

ER Daily Arrival Pattern

Lines: 103 to 134

Admissions vs. Discharges

a difference in the amplitude in the be discharged. There is no measured

ﬁlled with 20 or so patients waiting on maximizes the opportunity to generate

a) Conventional model (four in-process queues)

Wait Wait Wait Wait

b) See and Treat model (one in-process queue)

Reception Wait Assess, treat, and discharge

• measure the impact of small-scale that take variation into account. A

admit to a bed or discharge 96 percent of patients within four hours, an achieve-

1. really understand demand,

Litvak, E., P. I. Buerhaus, F. Davidoff, M. C. Long, M. L. McManus, D. M. Berwick. 2005. “Manag-

Lines: 352 to 361

Das könnte Ihnen auch gefallen