the speakers do not have any disclosures relevant to today’s...

1

Copyright © 2010 Institute for Healthcare Improvement

L30: How to Improve Quality and Flow While Reducing Cost

Eugene Litvak

Peter Lachman

Jason Leitch

Copyright © 2011 Institute for Healthcare ImprovementSlide 2

Disclosures

The speakers do not have any disclosures relevant to today’s session and discussion

2


Aims

� As funding constraints have increased under health care reform, the way hospital services are designed and managed has become paramount. This session will demonstrate that there is a cost-effective way to reorganize services to deliver safe and effective care.

� Participants will learn that the management of variability and the application of operations management theory are keys to a financially viable and safer health service. These principles will be illustrated by case studies and participatory work.


Learning Objectives

� After this session, participants will be able to:

� Develop an understanding of the theory of managing operations

� Demonstrate how to improve flow in hospital settings

� Identify ways to decrease variation in delivery of care

3


SESSION 1 INTRODUCTION TO THE PROBLEM

The growing finance gap

Peter Lachman


Setting the Context

� Is the central problem in healthcare cost or quality?

� Does your organization need additional resources?

� Is your biggest problem deciding where to close services or how to improve quality ….or both?

� Does managed care/capitation accompanied by reduced budget leads to poorer quality of care.

4


International Comparison of Spending on Health, 1980–2008

* PPP=Purchasing Power Parity.

Data: OECD Health Data 2011 (database), version 6/2011.

Average spending on health

per capita ($US PPP*)

Total expenditures on health

as percent of GDP

7Source: Commonwealth Fund National Scorecard on U.S. Health System Performance, 2011. 7


Outcome rankings

Source: Calculated by The Commonwealth Fund based on 2007 International Health Policy Survey; 2008 International Health Policy Survey

of Sicker Adults; 2009 International Health Policy Survey of Primary Care Physicians; Commonwealth Fund Commission on a High

Performance Health System National Scorecard; and Organization for Economic Cooperation and Development, OECD Health Data, 2009

(Paris: OECD, Nov. 2009).

5


Deaths per 100,000 population*

* Countries’ age-standardized death rates before age 75; including ischemic heart disease, diabetes, stroke, and bacterial infections.

See Appendix B for list of all conditions considered amenable to health care in the analysis.

Data: E. Nolte, RAND Europe, and M. McKee, London School of Hygiene and Tropical Medicine analysis of World Health

Organization mortality files and CDC mortality data for U.S. (Nolte and McKee, 2011).

Mortality Amenable to Health Care

Source: Commonwealth Fund National Scorecard on U.S. Health System Performance, 2011. 9


Spending more does not improve quality

CMS data:

Higher spending

states have

poorer quality

Source: Baicker K, Chandra A. Medicare spending, the physician workforce, and beneficiaries' quality of

care. Health Aff (Millwood). 2004 Jan-Jun;Suppl Web Exclusives:W4-184-97.

6


By income and insurance status, 2010U.S. average

Percent of adults ages 19–64 with any medical bill problem or outstanding debt*

* Problems paying or unable to pay medical bills, contacted by a collection agency for medical bills, had to change

way of life to pay bills, or has medical debt being paid off over time.

Data: 2005 and 2010 Commonwealth Fund Biennial Health Insurance Surveys.

Source: Commonwealth Fund National Scorecard on U.S. Health System Performance, 2011. 11

Medical Bill Problems or Medical Debt


The NHS

£15-20bn

productivity

challenge

7


Traditionally, healthcare

providers have prospered

without needing to focus on

costs and waste…

─Emphasis on revenue growth

─Little price transparency

─Less competitive than most industries

─Typically funded with insured dollars

…but several forces suggest that

this will not continue

─Cost inflation>incomes

─ Increased out-of-pocket funding

─CMS initiatives focusing on efficiency and cost

─Threat of disruptive, lower cost entrants

─Better credit rating for high safety/high quality providers

Providers Will Find Increasing Pressures to Focus on Costs and Waste


Question

Can we look at ways of delivering health care and increase quality and safety?

8


Session 2Introduction to Management

Operations

Eugene Litvak

Institute for Healthcare Optimization


Center for Medicare & Medicaid Innovation

Innovation Advisors Program

The Innovation Center seeks to deepen the capacity for

transformation by creating a network of experts in

improving the delivery system for Medicare, Medicaid

and CHIP beneficiaries.

The Innovation Advisors Program will inspire

dedicated, skilled individuals in the health care system

to deepen several key skill sets, including:

� Health care economics and finance;

� Population health;

� Systems analysis; and

� Operations research

9


Changing Patient Flow through the Parts of the System May Not Optimize Overall Patient Flow

� Can one reduce ED overcrowding just by improving patient flow through the ED?

� Can one improve ICU patient flow by increasing its size?


Reducing the Time of Patient Transfer…

… between two hospital units or the length of stay in the unit may not be right goal

� Do we really want to transfer patients from the ED to Med/Surg bed in one hour?

� Should we “fight” for reducing ICU length of stay?

10


High Census May Not be a Good Reason for Adding More Beds

� Midnight or midday census? The answer might be …“neither”

� Does the patient mix matter? Which patients mix?

� Does the size of your hospital/unit matter?

� What would you do when your waiting room is full?


Is Your Goal to Have a High Utilization of Your Resources?

� Is this a right goal?

� What happens when your resources are highly utilized?

� When utilization is high enough?

11


Can Your Hospital or Physician Office Reduce Overcrowding?

� Should this be your goal?

� Do you want to attract more patients?

� Would more patients reduce overcrowding and wait times?


What is the Right Goal for Patient Flow Improvement?

� Can Operations Management help you to achieve this goal?

� There is a right goal for patient flow improvement – patient throughput and access to care while improving or controlling quality of care!

12


The Central Health Care Problem: Cost or Quality?

To Treat or To Save Money?


Operations Management: An Art or a Science

� Real life problems are not labeled

� Intuitive solutions are frequently wrong

13


Operations Management

� Complex scientific decisions in uncertain environment


Boundaries: Scientific Limitations, Intuition, and Politics

� Good theory or good results?� To be right or to be helpful? (Mad bear

story)� What does a model? Simplification of

reality or replacement of reality?� A model for people or a model for robots?� Unobtainable (?) data� To be “good” or to be “lucky”?� Problems unsolved and problems

unsolvable

14


Operations Management

� Main Components

� Optimization

� Queuing Theory

� Network Theory

� Inventory Management


Optimization

� Linear Programming

� Nonlinear Programming

� Dynamic Programming

� Stochastic Programming

15


Queuing Theory

� Service Policies

� Queues without priorities

� Queues with priorities

� Queues with losses


Network Theory

� Classical transportation problem

� The shortest path problem

� The longest (critical) path problem

� Flows in the network

� The assignment problem

16


Classical Transportation Problem

S1

S2

S3

Sm

-D1

-D2

-D3

-Dn


The Shortest Path Problem

1

7

2

3

4

5

6

8

17


Critical Path Problem (Example)

5

42

3

23

4

21

4

323


Critical Path Problem (Solution)

0

5

42

3

23

4

21

4

323

18


Optimal Assignment Problem

0

Employee Job

1 1

22

33

44


Inventory Management

� Inventory location

� Optimal inventory management when demand is uncertain

19


Inventory Location Problem (Example)

0

1

1

1

1

1

1


0

0.5

1

1

1

1

1

Inventory Location Problem (Solution)

0.5X

20


Session 3 – Table workWhy is Operations Management Theory Relevant to Health Care?

Jason Leitch


A case history - discuss in your groups

You are the duty manager on for the dayat your hospital. What are the key issuesthat face you during the day to ensurethere is a good flow and no delays?

10 minutes

21


Feedback

� How did you assess the problem?� What data did you have?� What solutions did you have?

10 minutes


Session 4Introduction to Queuing

Eugene Litvak

22


Two Medical ICUs with the Same

Patients Acuity

5 beds

Average LOS = 2.5 days

Admission rate = 1pt/day

10 beds


Admission rate = 2pts/day

Do they have the same waiting times to be admitted to these units?


Ťw = 0.13 days

1st ICU 2nd ICU

Ťw = 0.012 days

23


How Many Beds/Nurses/Exam Rooms Do You Need?

Suppose you are coming to a bank…


How Many Beds/Nurses/Exam Rooms Do You Need?

Suppose you are coming to a bank…

24


Service Mechanism

� Single Server00000

� Multiple Servers� Serial Servers

Parallel Servers

S

S S

SS

S


Types of Service

� Queues with waiting

� Queues with rejections

25


Service Discipline

� First come, first served

� Last come, first served

� Random

� Priorities


Service Time

� Constant service time� Exponential service time� Non-stationary service time� Correlated service time (peak, off-peak

hours)

26


Arrival Patterns

� Regular arrivals with unpunctuality� Request n arrivals at the time (tn+ εn)

� Aggregated (group) arrivals� Discrete-time arrivals (t1, t2,…, tk)� Non-stationary arrivals (peak, off-peak

arrivals)� Correlated arrivals (length of the queue)� Arrivals in a continuous flow (gas)


Main Characteristics

1. Queues with rejections� Absolute throughput – the average

number of requests served during one time unit

� Relative throughput – the ratio between the average numbers of requests served and arrived during one time unit

� Average number of occupied service units

� Average number of free service units� Probability of rejection

27



2. Queues with waiting� Average length of queue – the average

number of requests waiting for service� Average number of requests in the

system – the average total number of request being served and waiting for service

� Average request’s waiting time� Average request’s time in the system –

the average total time service and waiting time for a request


Notations and Parameters� S - Number of servers� λ - Average number of arrivals per unit of time, λ =

1/ta (ta – average interarrival time interval)� µ - Average number of request served per unit of time,

µ = 1/ts (ts – average service time interval)� ρ = λ/µ� Ŭ = p/ S – average system’s utilization rate � pn – The probability of the system to be in the state n

when n requests “arrive” to the system:

28


Queues with Waiting (Parallel Servers)

S

S

S


Notations and Parameters

� No requests (n = 0)p0 = [1+ ρ/1!+ ρ2/2!+…+ ρ(s-1) /(S -1)! + ρs/S!(1+

ρ/S)]

� No queue (0 ≤ n ≤ S)pn= p0ρ

n/n!

� Queue (n > S)pn = p0ρ

n/S!S(n-s)

Queue with Random Arrivals and Exponential

Service Time (M/M/S)

29



� Probability P(Ss) that all the servers are serving requests can be computed by the following formula: P(Ss) = ρs ⋅ S ⋅ p0/S! ⋅ (S - ρ)

� The average number (Řl) of requests in line can be computed by the following formula:Řl = P(Ss) ⋅ ρ/(S - ρ) = [ρ(s+1) ⋅ S ⋅ p0]/[S!(S - ρ)²]

� The average number (Řs) of requests in service is equal to: Řs = ρ

� The average number (Ř) of requests in system can be computed by the following formula:Ř = Řl + Řs


� The average waiting time (Ťw) for a request can be computed by the following formula:Ťw = Řl/λ

� The average system time (Ť) for a request can be computed by the following formula:Ť = (Řl + Řs)/λ

30


Ťw

λλλλ

λλλλ = 0.8 ⋅⋅⋅⋅ µµµµ ⋅⋅⋅⋅Sρρρρ = 0.8 ⋅⋅⋅⋅ S


Applying Queuing Theory: Examples

1. Triage

2. Two ICUs

31


Example : Triage on Monday

On Monday morning at 8am, there are 2 triage nurses on duty. They can each serve one patient every 3 minutes. The Monday morning arrival rate is 20.8 patients/hour.

What is the waiting time?


� The average patient waiting time (Ťw) can be computed by the following formula:� Ťw = Řl/λ

� The average number (Řl) of patients in line can be computed by the following formula:� Řl = P(Ss) . ρ /(S-ρ) � Probability P(Ss) that all S “servers” are

busy:� P(Ss) = ρs . S . p0/S! . (S- ρ) (Table)

32


λλλλ, µµµµ , S

λ= 20.8 , µ= 1/3min = 20, S = 2

ρ = λ/ µ = 1.04

P(Ss) (table) ≈ 0.33

Řl = P(Ss) . ρ /(S-ρ) = 0.33 . 1.04/(2 - 1.04) ≈ 0.36

Ťw = Řl/λ = 0.36/20.8 = 0.017 h = 1.04 min


Example: Triage on Friday

� On Fridays at 4 PM, the arrival rate is about 10.3 patients/hour. There is one triage nurse at this time, who can serve one patient every 3 minutes.

� What is the waiting time?

33



λ= 10.3 , µ= 1/3min = 20, S = 1

ρ = λ/ µ = 0.52


Řl = P(Ss) . ρ /(S-ρ) = 0.52 . 0.52/(1 – 0.52) ≈ 0.56

Ťw = Řl/λ = 0.56/10.3 = 0.054 h = 3.28 min


Two Medical ICUs with the Same

Patients Acuity

5 beds


Admission rate = 1pt/day

10 beds


Admission rate = 2pts/day

Do they have the same waiting times to be admitted to these units?

34



λ = 1 , µ= 1/2.5 = 0.4, S = 5

ρ = λ/ µ = 2.5


Řl = P(Ss) . ρ /(S-ρ) = 0.13 . 2.5/(5 – 2.5) ≈ 0.13

Ťw = Řl/λ = 0.13/1 = 0.13 days

1st ICU



λ = 2 , µ= 1/2.5 = 0.4, S = 10

ρ = λ/ µ = 5

P(Ss) (table) ≈ 0.036

Řl = P(Ss) . ρ /(S-ρ) = 0.036 . 5/(10 – 5) ≈ 0.036

Ťw = Řl/λ = 0.036/2 = 0.018 days

2nd ICU

35


Ťw = 0.13 days

1st ICU 2nd ICU

Ťw = 0.018days


Session 5Introduction of Variability Concept

and Methodology

Eugene Litvak

Institute for Healthcare Optimization

36


Quotes from the 2006 IOM report The Future of Emergency Care in the U.S. Health System (Hospital-Based Emergency Care: At the Breaking Point)

� “Hospitals have direct control over operational efficiency, and have a number of variables within their control. They include such factors as impatient bed capacity, ancillary service delays, the scheduling of services and support staff…”

� “4.1 Hospital chief executive officers should adopt enterprise-wide operations management and related strategies to improve the quality and efficiency of emergency care.”



� “4.3 Training in operations management and

related approaches should be promoted by

professional associations; accrediting

organizations, such as the Joint Commission on

Accreditation of Healthcare Organizations

(JCAHO) and the National Committee for

Quality Assurance (NCQA)…”

37


Gaps cited by the Institute of Medicine in Crossing the Quality Chasm (2001)

� Ineffectiveness of care� Lack of efficiency in delivery system� Inadequate safety� Insufficient patient-centerness� Inadequate timeliness of care


•Patient Safety

• Nurse understaffing/overloading

• ED diversions/access to care

• High cost

• Hospital resources needed for bioterrorism

preparedness

Addressing variability is necessary, although not

sufficient, to satisfactorily resolve these problems.

Major health care delivery problems:

38


How unsmooth census looks like?(no holidays, no weekends, weekdays only)


How did we staff, and how do we staff?

39


Systemic Effects of Peak Loads

� Internal Divert –Patients sent to alternative

floors\Intensive Care locations

� Internal Delays – PACU backs up

� External Divert - ED divert

� Staff overload – medical errors and inability to

retain staff

� System Gridlock – Increase in LOS

� Decreased throughput and revenue


Controlling the total cost, without knowing cost of delivery, decreases

quality.

Take-out Pizza Example

40


Litvak E. & Long MC. Cost and Quality Under Managed Care:

Irreconcilable Differences? American Journal of Managed Care,

2000; 6 (3): 305-312.

http://www.ajmc.com/files/articlefiles/AJMC2000MarLitvak305_312.pdf

Litvak E. "Optimizing patient flow by managing its variability". In

Berman S. (ed.): Front Office to Front Line: Essential Issues for

Health Care Leaders. Oakbrook Terrace, IL: Joint Commission

Resources, 2005, pp. 91-111.

http://www.ihoptimize.org/Collateral/Documents/English-

US/front%20lines%20chapter.pdf



� “By applying variability methodology, queuing theory and the I/T/O model, hospitals can identify and eliminate many of the patient flow impediments caused by operational inefficiencies”

� “By smoothing the inherent peaks-and valleys of patient flow, and eliminating the artificial variabilities, that unnecessarily impair patient flow, hospitals can improve patient safety and quality while simultaneously reducing hospital waste and cost”

41


The Ideal Healthcare System

(100% efficiency)

1. All patients have the same disease with

the same severity.

2. All patients arrive at the same rate.

3. All providers (physicians, nurses) are

equal in their ability to provide quality

care.


Can your health care delivery system become a Toyota product line?

42


Natural Variability

I) Clinical Variability

II) Flow Variability

III) Professional Variability

−Random

−Can not be eliminated (or even reduced)

−Must be optimally managed

}


Designing and Testing Complex

Mechanical Systems: Family Car

• Hitting a pothole vs. high speed impact against

the wall

• Health care “financial bumper”

• Are the stresses an intrinsic part of health care

delivery?

43


What makes hospital census variable?


What makes hospital census variable?

� If ED cases are 50% of admissions

and…

� Elective-scheduled OR cases are 35% of admissions

then…

� Which would you expect to be the largest source of census

variability?

44


The answer is…

The ED and Elective-Scheduled OR have approximately equal effects on census variability.

Why?Because of another (hidden) type of variability...


Artificial Variability

� Non-random

� Non-predictable (driven by unknown individual priorities)

� Should not be managed, must be identified and eliminated

45


Why managing variability today is more important than before?


Does the healthcare system need more capacity?

90

46


At what cost?

� Typical cost of new capacity

� Inpatient beds - $1M in capital and $250K-800K annual operating expense

� Operating rooms - $2 – 7 Million, $250K+ annual operating expense

� Major imaging (CT, MRI, PET/CT, etc.) –approx. $1M+

� Cardiac Catheterization Lab – approx. $2M

� Nursing and other provider shortages?

Slide provided by Sandeep Green Vaswani, Institute for Healthcare Optimization

91


Is this about access to care?

47


Floors

ICU

ED

Variability and access to care

Scheduled

demand


48


Can you solve your ED, ICU or Med/Surgunits overcrowding without smoothing elective admissions?

Can you match your capacity and demand without throwing excessive resources at the system by other means without smoothing elective admissions?


After elective patient flow is smoothed (and only then!) the right size of the hospital units could be determined for both scheduled and unscheduled admissions

49


Do you always place every patient into the appropriate bed? What happens if you do not?

How important is to know the right size of the unit?


Rapid Response Team

Does the Rapid Response Team helps at your hospital ?

Why?

Litvak E, Pronovost PJ. Rethinking rapid response

teams. JAMA. 2010;304(12):1375–6.

98

50


Elective Surgical Requests vs Total Refusals

Michael L. McManus, M.D., M.P.H.; Michael C. Long, M.D.; Abbot Cooper; James

Mandell, M.D.; Donald M. Berwick, MD; Marcello Pagano, Ph.D.; Eugene Litvak,

Ph.D. Impact of Variability in Surgical Caseload on Access to Intensive Care

Services, Anesthesiology 2003; 98: 1491-1496.


Variability and Readmissions

Does variability affect readmission rate?

“The main outcome variable is unplanned patient readmission to the neurosciences critical care unit within 72 hrs of discharge to a lower level of care. The odds of one or more discharges becoming an unplanned readmission within 72 hrs were nearly two and a half times higher on days when ≥9 patients were admitted to the neurosciences critical care unit …” *)

“The odds of readmission were nearly five times higher on days when ≥10 patients were admitted …” *)

*) Baker, David R. DrPH, MBA; Pronovost, Peter J. MD, PhD; Morlock, Laura L. PhD, et al. Patient flow variability and unplanned readmissions to an intensive care unit.Critical Care Medicine: November 2009 - Volume 37 - Issue 11 - pp 2882-2887

51


Is this about nurse staffing as well?

Can you provide an adequate nurse staffing without smoothing elective admissions?


Impact of Census Variability on Patients per Nurse Ratio

http://www.ihoptimize.org/what-we-do-methodology-artificial-variability-healthcare-

provider-nurses-other-providers.htm

52


Variability and Quality of Care*

� Inadequate numbers of nursing staff contribute to 24% of all

sentinel events in hospitals. Inadequate orientation and in-

service education of nursing staff are additional contributing

factors in over 70% of sentinel events*

� “…higher numbers of nurse hours per patient, larger

proportions of RNs and high levels of competition with other

hospitals were all correlated with higher levels of NQF Safe

Practices adoption.”**

*Dennis S. O’Leary, JCAHO (personal communication)

** http://nursing.advanceweb.com/News/National-News/Magnet-

Designated-Hospitals-More-Likely-to-Adopt-NQF-Safety-Practices.aspx


Variability and mortality

“Each additional patient per nurse was associated with a 7% increase in the

likelihood of dying within 30 days of admission and a 7% increase in the odds of

failure-to-rescue”*

* Linda H. Aiken, Sean P. Clarke, Douglas M. Sloane, Julie Sochalski, and Jeffrey H. Silber.

Hospital Nurse Staffing and Patient Mortality, Nurse Burnout, and Job Dissatisfaction.

JAMA, 2002; 288: 1987:1993

Litvak E, Buerhaus PI, Davidoff F, Long MC, McManus ML,

Berwick DM. “Managing Unnecessary Variability in Patient Demand

to Reduce Nursing Stress and Improve Patient Safety,” Joint

Commission Journal on Quality and Patient Safety, 2005; 31(6): 330-

338.http://www.ihi.org/NR/rdonlyres/E18D05FD-F4E5-448D-8CBE-

217CB5C03B7C/0/ManagingUnnecessaryVariabilityinPatientDemand.pdf

53


Example:

Assumptions:

� 200 surgical beds

� average census for surgical beds 160

� staffing level 40 nurses (1 nurse per 4 patients)

� average residual from 160 patients census is 20% or 32 patients

� patients are distributed evenly between the nurses

How the mortality rate will change with 20% increase in surgical demand?


Results:

� 32 additional patients will be distributed evenly between 32 nurses: 1 additional patient per nurse or 4 + 1 = 5 patient per nurse

� these 32 nurses now will take care of 160 patients, whose mortality rate increases by 7%

� if these additional 32 patients will be distributed evenly between 16 nurses, then each such nurse will take care of 4 + 2 = 6 patients

� these 16 nurses now will take care of 96 patients, whose mortality rate increases by 14%

54


Nurse Staffing and Inpatient Hospital Mortality, Needleman J., Buerhaus P., et al.

N Engl J Med 2011; 364:1037-1045, March 17, 2011

� “There was a significant association between increased mortality and increased exposure to unit shifts during which staffing by RNs was 8 hours or more below the target level “

� “The association between increased mortality and high patient turnover was also significant “

107

Patient Mortality and Patient Flow


What is easier: to talk to your colleagues or to the lawyers?!

http://www.nhmedmallawyer.com/blog/post/show/hospital-staffing-and-its-effect-on-quality-care

http://www.healthleadersmedia.com/content/LED-269595/PDH-Understaffing-a-Possible-Factor-in-Deaths-at-CRMC##

55


Five ways of staffing

Nursing August 2011http://journals.lww.com/nursing/Fulltext/2011/080

00/Nurse_staffing,_hospital_operations,_care_quality,.1.aspx


Session 6Applications to the health care problems

Applying theory to the front line

Eugene Litvak

Jason Leitch

Peter Lachman

56


Managing variability in patient flow:Success stories

Managing Patient Flow: A Focus on Critical Processes

http://store.trihost.com/jcaho/product.asp?dept%5Fid=34&catalog%5Fitem=712


Boston Medical Center (OR)

• Reduced ambulance diversions by 20%

• Increased surgical volume by ≈ 20%

• Reduced last-minute postponement of elective surgeries by 99.5%

• Reduced by 0.5h nurse hours per patient day in the step-down unit

57


Cincinnati Children’s Hospital Medical Center

�Weekend waiting times:

Despite a 37% increase in the case volume in July-August as

compared to January-March the waiting times for unscheduled

procedures has been reduced by 34%

� Weekday waiting times:

Despite a 24% increase in the case volume during September-early

October, as compared to January-March, the waiting times for

unscheduled procedures has been reduced by 28%.


Cincinnati Children’s Hospital Medical Center

� Throughput increase

Originally yearly equivalent to the addition of one more operating room to the OR (which would increase capacity by 1/21 or 4.8%),without any of the associated capital or operating costs. For the last two years 7% annual increase in surgical volume without additional OR(s).

� Reduction in Overtime

Overtime hours decreased by an estimated 57% between September 18, 2006 and the first week of January 2007. If OR operating costs are estimated at $250/room hour, then these savings are equivalent to $10,750/week, or $559,000 annually.

58


Their opinion…

“this is the best thing for ortho since I have been here…we get our cases done earlier which makes our families very happy...patients don’t have to wait NPO until the evening. The weekends are unbelievably good, the OR hostility that was previously present has been virtually eliminated”. (Ortho surgeon, Division director)


“I feel there is an improvement in our time and efficiency when assigning staff.We are assigning add-on staff the daybefore instead of pulling them fromother rooms...” (OR nurse)

59


Elliot Hospital (Maternity ward)

• Eliminated morning chaos

• Reduced the demand for additional staffed beds by 6 positions (50%)

• Improved quality of care


BMCCincinnati

Children’s

Elliot

Is it feasible for other hospitals to

manage variability in patient flow?

60


� Provide the resources (e.g., staffing) sufficient to meet

current patient peaks in demand - historical scenario

� Staff below the peaks and tolerate ED diversions,

nursing overloading and medical errors - current scenario

� Smooth artificial variability and provide the

resources to meet patient (vs. schedule) driven peaks in

demand. Variability methodology can quantify and

justify such additional resources

Three alternatives:

119


What would be national return on investment

from applying these concepts?

Eugene Litvak (IHO), Maureen Bisognano (IHI). More

Patients, Less Payment: Increasing Hospital Efficiency

In The Aftermath Of Health Reform. Health Affairs,

January 2011 (30):176-180

61


OECD Acute Care Bed Occupancy

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%90%

88%86% 86%

84%

79% 79% 79%76% 76%

73%

67%

Acute Care Bed Occupancy 2009

OECD Health Data

Slide provided by Sandeep Green Vaswani, Institute for Healthcare Optimization


• Based on AHA 2010 data, overall nationwide hospital inpatient

occupancy was about 66%

• Even if one were to assume that all admissions are urgent in nature

(statistically random arrivals), 80% occupancy should be achievable

(based on queuing methodology) without compromising access or

quality of care1

• Potential savings of tens of billions of dollars annually2

1 From Front Office to Front Line- Joint Commission Resources, chapter 4:

Optimizing Patient Flow by Managing its Variability

http://www.ihoptimize.org/Collateral/Documents/English-

US/front%20lines%20chapter.pdf2 Recent IOM book:

http://www.nap.edu/openbook.php?record_id=12750&page=294

National Opportunity – An Example

62


National Opportunity – An Example

“Early hospital adopters such as Cincinnati Children’s Hospital (CCH) have been able to increase

hospital throughput capability by more than 15%1. If the experience of early hospital adopters proves

generalizable, it will reduce U.S. hospital cost per admission by ~15 percent. Since hospitalizations,

including outpatient procedures, consume over 30 percent of US health care spending, this single

improvement would reduce the cost of U.S. health insurance by roughly 4-5 percent if hospitals pass

savings through to insurers and insurers, in turn, to insurance buyers. It is also likely to reduce hospital

mortality by sparing hospital nursing units preventable bulges in the number of new patients they must

admit in a single day.” 2

These 4%-5% mean over $1 trillion in 10 years.

1 http://www.ihi.org/ihi/files/WIHI/WIHI_20091202_Patient_Flow.mp3

2 E. Litvak, A. Milstein, M. Smith. “Engineering Science And Episode-

Based Hospital Payment”

http://healthaffairs.org/blog/2010/09/15/engineering-science-and-

episode-based-hospital-payment/


Readings

http://www.ihoptimize.org

http://www.ihoptimize.org/knowledge-center-publications.htm

hManaging Patient Flow in Hospitals:

Strategies and Solutions, Second Edition

2009. 166 pages. ISBN: 978-1-59940-372-4

http://www.jointcommissioninternational.org/

Books-and-E-books/Managing-Patient-Flow-

in-Hospitals-Strategies-and-Solutions-Second-

Edition/1497

63


Some helpful links:

• http://www.ihoptimize.org/what-we-do-methodology-artificial-

variability-healthcare-provider-administrators.htm


variability-healthcare-provider-surgeons.htm


variability-healthcare-provider-nurses-other-providers.htm

• http://www.rwjf.org/pr/product.jsp?id=50488

• http://www.ihoptimize.org/Collateral/Documents/English-US/MPF09-

Pages24-25.pdf

• http://www.ihoptimize.org/Collateral/Documents/English-

US/Cincinnati_Childrens_Questionnaire.pdf

• http://www.ihi.org/ihi/files/WIHI/WIHI_20091202_Patient_Flow.mp3

• http://www.ihi.org/NR/rdonlyres/E632D3DD-CFA8-425A-A676-

11AA11B354CA/0/ORInitiativeClip20090917.mp3


Summary

� Scientific managing variability in patient flow is absolutely necessary to increase overall hospital patient throughput while improving quality of care, patient safety and reducing nursing workload.

� It requires rigorous data analysis, scientific management of operations, clinical and organizational behavior expertise.

64


Great Ormond Street

� Analysed our data to see variability

� Then developed programmes to implement change

� The following data is from one if the workstreams


The Experience at Great Ormond Street

� Tertiary and quaternary hospital� No Emergency Room� However variability is still a problem� How do we deal with this

� Analysis of data� Understand the problem � Develop the model for system wide

change

65


Where will this work?

� Consider areas where you do not know how to solve the blocks in the system� Emergency room� Inpatient beds� Diagnostic procedures� Operating room� Outpatients

� i.e everywhere!!


GOSH Admissions By Date

Program for Management of Variability in Health Care Delivery Boston

University Health Policy Institute

0

20

40

60

80

100

120

140

160

Sun/

01/0

4/07

Sun/

15/0

4/07

Sun/

29/0

4/07

Sun/

13/0

5/07

Sun/

27/0

5/07

Sun/

10/0

6/07

Sun/

24/0

6/07

Sun/

08/0

7/07

Sun/

22/0

7/07

Sun/

05/0

8/07

Sun/

19/0

8/07

Sun/

02/0

9/07

Sun/

16/0

9/07

Sun/

30/0

9/07

Sun/

14/1

0/07

Sun/

28/1

0/07

Sun/

11/1

1/07

Sun/

25/1

1/07

Sun/

09/1

2/07

Sun/

23/1

2/07

Sun/

06/0

1/08

Sun/

20/0

1/08

Sun/

03/0

2/08

Sun/

17/0

2/08

Sun/

02/0

3/08

Sun/

16/0

3/08

Sun/

30/0

3/08

Count

Admission Date

Grand Total Total Mean

Copyright Great Ormond Street

66


Admissions By Urgency and Date



0

20

40

60

80

100

120

140

160Su

n/01

/04/

07

Sun/

15/0

4/07

Sun/

29/0

4/07

Sun/

13/0

5/07

Sun/

27/0

5/07

Sun/

10/0

6/07

Sun/

24/0

6/07

Sun/

08/0

7/07

Sun/

22/0

7/07

Sun/

05/0

8/07

Sun/

19/0

8/07

Sun/

02/0

9/07

Sun/

16/0

9/07

Sun/

30/0

9/07

Sun/

14/1

0/07

Sun/

28/1

0/07

Sun/

11/1

1/07

Sun/

25/1

1/07

Sun/

09/1

2/07

Sun/

23/1

2/07

Sun/

06/0

1/08

Sun/

20/0

1/08

Sun/

03/0

2/08

Sun/

17/0

2/08

Sun/

02/0

3/08

Sun/

16/0

3/08

Sun/

30/0

3/08

Count

Admission Date

Grand Total Total Mean Elective Elective Mean Emergency Emergency Mean



Elective Inpatient Admissions by Use of Theatres: Non-holiday Weekdays

Program for Management of Variability in Health Care Delivery

Boston University Health Policy Institute

0

5

10

15

20

25

30

35

40

45

50

4/2/

2007

4/9/

2007

4/16

/200

74/

23/2

007

4/30

/200

75/

7/20

075/

14/2

007

5/21

/200

75/

28/2

007

6/4/

2007

6/11

/200

76/

18/2

007

6/25

/200

77/

2/20

077/

9/20

077/

16/2

007

7/23

/200

77/

30/2

007

8/6/

2007

8/13

/200

78/

20/2

007

8/27

/200

79/

3/20

079/

10/2

007

9/17

/200

79/

24/2

007

10/1

/200

710

/8/2

007

10/1

5/20

0710

/22/

2007

10/2

9/20

0711

/5/2

007

11/1

2/20

0711

/19/

2007

11/2

6/20

0712

/3/2

007

12/1

0/20

0712

/17/

2007

12/2

4/20

0712

/31/

2007

1/7/

2008

1/14

/200

81/

21/2

008

1/28

/200

82/

4/20

082/

11/2

008

2/18

/200

82/

25/2

008

3/3/

2008

3/10

/200

83/

17/2

008

3/24

/200

83/

31/2

008

Non-theatre Non-theatre Mean Theatre Theatre Mean


67


Inpatient vs. Day Case Admissions: Non-holiday Weekdays Only



0

10

20

30

40

50

60

70

80

90

100M

on/0

2/04

/07

Mon

/16/

04/0

7

Mon

/30/

04/0

7

Mon

/14/

05/0

7

Mon

/28/

05/0

7

Mon

/11/

06/0

7

Mon

/25/

06/0

7

Mon

/09/

07/0

7

Mon

/23/

07/0

7

Mon

/06/

08/0

7

Mon

/20/

08/0

7

Mon

/03/

09/0

7

Mon

/17/

09/0

7

Mon

/01/

10/0

7

Mon

/15/

10/0

7

Mon

/29/

10/0

7

Mon

/12/

11/0

7

Mon

/26/

11/0

7

Mon

/10/

12/0

7

Mon

/24/

12/0

7

Mon

/07/

01/0

8

Mon

/21/

01/0

8

Mon

/04/

02/0

8

Mon

/18/

02/0

8

Mon

/03/

03/0

8

Mon

/17/

03/0

8

Mon

/31/

03/0

8

Count

Admission Date

DC DC Mean IP IP Mean



Theatre Cases by Date: Non-holiday Weekdays Only



0

10

20

30

40

50

60

Mon

/02/

04/0

7

Mon

/16/

04/0

7

Mon

/30/

04/0

7

Mon

/14/

05/0

7

Mon

/28/

05/0

7

Mon

/11/

06/0

7

Mon

/25/

06/0

7

Mon

/09/

07/0

7

Mon

/23/

07/0

7

Mon

/06/

08/0

7

Mon

/20/

08/0

7

Mon

/03/

09/0

7

Mon

/17/

09/0

7

Mon

/01/

10/0

7

Mon

/15/

10/0

7

Mon

/29/

10/0

7

Mon

/12/

11/0

7

Mon

/26/

11/0

7

Mon

/10/

12/0

7

Mon

/24/

12/0

7

Mon

/07/

01/0

8

Mon

/21/

01/0

8

Mon

/04/

02/0

8

Mon

/18/

02/0

8

Mon

/03/

03/0

8

Mon

/17/

03/0

8

Mon

/31/

03/0

8

Count

Admission Date

DC DC Mean IP IP Mean Outpatient Outpatient Mean


68


Discharges by Date



0

20

40

60

80

100

120

140

160

Sun/

01/0

4/07

Sun/

15/0

4/07

Sun/

29/0

4/07

Sun/

13/0

5/07

Sun/

27/0

5/07

Sun/

10/0

6/07

Sun/

24/0

6/07

Sun/

08/0

7/07

Sun/

22/0

7/07

Sun/

05/0

8/07

Sun/

19/0

8/07

Sun/

02/0

9/07

Sun/

16/0

9/07

Sun/

30/0

9/07

Sun/

14/1

0/07

Sun/

28/1

0/07

Sun/

11/1

1/07

Sun/

25/1

1/07

Sun/

09/1

2/07

Sun/

23/1

2/07

Sun/

06/0

1/08

Sun/

20/0

1/08

Sun/

03/0

2/08

Sun/

17/0

2/08

Sun/

02/0

3/08

Sun/

16/0

3/08

Sun/

30/0

3/08

Count

Date

Discharges Overall Mean



Discharges by Date: Non-holiday Weekdays



0

20

40

60

80

100

120

140

160

Mon

/02/

04/0

7

Mon

/16/

04/0

7

Mon

/30/

04/0

7

Mon

/14/

05/0

7

Mon

/28/

05/0

7

Mon

/11/

06/0

7

Mon

/25/

06/0

7

Mon

/09/

07/0

7

Mon

/23/

07/0

7

Mon

/06/

08/0

7

Mon

/20/

08/0

7

Mon

/03/

09/0

7

Mon

/17/

09/0

7

Mon

/01/

10/0

7

Mon

/15/

10/0

7

Mon

/29/

10/0

7

Mon

/12/

11/0

7

Mon

/26/

11/0

7

Mon

/10/

12/0

7

Mon

/24/

12/0

7

Mon

/07/

01/0

8

Mon

/21/

01/0

8

Mon

/04/

02/0

8

Mon

/18/

02/0

8

Mon

/03/

03/0

8

Mon

/17/

03/0

8

Mon

/31/

03/0

8

Count

Admission Date

Discharges Overall Mean


69


Summary

� While day case patients comprise majority of admissions, true inpatients have most impact

� Substantial variability in elective admissions� Theatre cases comprise large majority

� Wasted bed & theatre capacity� Improved scheduling of elective

admissions, especially theatre cases, needed


Boston University Health Policy Institute Copyright Great Ormond Street


Recommendations

1. Central management of admissions2. Establishment of a central ‘patient flow team’3. Central management of operationally-relevant

information systems4. Improve collection and reporting of flow data5. Separate emergency and elective beds6. Separate resources for day case and inpatients7. Determine best management strategies for ‘high utiliser’

patients8. Reconfigure wards into larger units


70


And how did we respond?


0

5

10

15

20

25

30

35

13

-De

c-0

8

27

-De

c-0

8

10

-Ja

n-0

9

24

-Ja

n-0

9

07

-Fe

b-0

9

21

-Fe

b-0

9

07

-Ma

r-0

9

21

-Ma

r-0

9

04

-Apr-

09

18

-Apr-

09

02

-Ma

y-0

9

16

-Ma

y-0

9

30

-Ma

y-0

9

13

-Ju

n-0

9

27

-Ju

n-0

9

11

-Ju

l-0

9

25

-Ju

l-0

9

08

-Aug

-09

22

-Aug

-09

05

-Sep

-09

19

-Sep

-09

03

-Oct-

09

17

-Oct-

09

Ab

so

lute

Co

un

t

Date

Cardiac VFM Pilot: Weekly Theatre Throughput and Critical Flow Failures

Weekly theatre throughput

Weekly theatre critical flow failures

Increasing NHS + IPP throughput

Decreasing critical flow failures (cancellations,

emergency refusals and nursing shifts lost to

sickness)

New solution go live

PDSA: flow management taken on by operational manager

Reduced variability -process more predictable

High variability -process unpredictable


71


0

100

200

300

400

500

600

700

800

900

1000

13

-De

c-0

8

27

-De

c-0

8

10

-Ja

n-0

9

24

-Ja

n-0

9

07

-Fe

b-0

9

21

-Fe

b-0

9

07

-Ma

r-0

9

21

-Ma

r-0

9

04

-Apr-

09

18

-Apr-

09

02

-Ma

y-0

9

16

-Ma

y-0

9

30

-Ma

y-0

9

13

-Ju

n-0

9

27

-Ju

n-0

9

11

-Ju

l-0

9

25

-Ju

l-0

9

08

-Aug

-09

22

-Aug

-09

05

-Sep

-09

19

-Sep

-09

03

-Oct-

09

17

-Oct-

09

Cumulative throughput

Cumulative critical flow failures



0

5

10

15

20

25

30

35

13

-De

c-0

8

27

-De

c-0

8

10

-Ja

n-0

9

24

-Ja

n-0

9

07

-Fe

b-0

9

21

-Fe

b-0

9

07

-Ma

r-0

9

21

-Ma

r-0

9

04

-Apr-

09

18

-Apr-

09

02

-Ma

y-0

9

16

-Ma

y-0

9

30

-Ma

y-0

9

13

-Ju

n-0

9

27

-Ju

n-0

9

11

-Ju

l-0

9

25

-Ju

l-0

9

08

-Aug

-09

22

-Aug

-09

05

-Sep

-09

19

-Sep

-09

03

-Oct-

09

17

-Oct-

09

31

-Oct-

09

Ab

so

lute

Co

un

t

Date

Cardiac VFM Pilot: Weekly Theatre Throughput and Critical Flow Failures

Weekly theatre throughput

Weekly theatre critical flow failures

Linear (Weekly theatre throughput)

Increasing NHS + IPP throughput

Decreasing critical flow failures (cancellatio

ns& emergency refusals)

New solution go live

PDSA: flow management taken on by operational manager

Reduced variability - process failures better managed


72




Patients:

• Reduced waiting time and improved access to care

• Reduced mortality and medical errors

Nurses:

• Reduced overtime

• Reduced workload

73


What is here for me?

Physicians:� Reduced waste of time� Increased patient throughput� Reduced overtime

Hospital:� Better utilization of resources � Reduced hours of ED overcrowding� Staff and patient satisfaction� More staffing resources: better tolerating peak loads� Reduced mortality and medical errors� Reduced length of stay� Increased hospital throughput and revenue


What is next?

74


1. Provide the resources (e.g., staffing) sufficient to

meet current patient peaks in demand -

historicalscenario

2. Staff below the peaks and tolerate ED diversions,

nursing overloading and medical errors - current

scenario

3. Smooth artificial variability and provide the

resources to meet patient (vs. schedule) driven

peaks in demand. Variability methodology can

quantify and justify such additional resources

Three alternatives:


END!!