2008/07/17 09:50 Bill Rouse, "Modeling & Managing Complex Systems: A Case Study of Healthcare Delivery", ISSS Madison 2008

ISSS Madison 2008, 52nd Annual Meeting of the International Society for the Systems Sciences

This digest was created in real-time duringthe meeting, based on the speaker's presentation(s) and comments from the audience. The content should not be viewed as an official transcript of the meeting, but only as an interpretation by a single individual. Lapses, grammatical errors, and typing mistakes may not have been corrected. Questions about content should be directed to the originator. The digest has been made available for purposes of scholarship, posted on the ISSS web site by David Ing.

Intro by Gary Metcalf

Bill Rouse, Georgia Institute of Technology, Industrial and Systems Engineering and College of Computing:


Tennenbaum Institute:  cross campus

Work on one problem:  how do complex organizational systems change or not change

Looked at 200 companies

  • Of 100 companies in the Dow over a hundred years, only GE has lasted
  • Creative destruction is good for the economy, but not good for the company

Looking at American health system

50% of costs increases are due to technological innovation, how can we afford this

Complex adaptive systems

  • Clinton administration discovered that you can't deal with this through command and control
  • Large collection of entrepreneurs, no one in charge
  • How can we influence them?

Stakeholders and interests

  • Who takes risks?
  • Look at the impact of the health care system on health
  • Conclusion:  the health care system has 10% of influence, 40% of lifestyle, 40% is parents, 10% is environment

Economist article on healthcare

  • Last page (25) every dollar saved is somebody's income
  • This is why it's hard to change the system

Value chain on disease detection:  how many on the people?

  • How much spent not on research, but on improving research?  Cut out of budget
  • NIH putting all of money into medical research may not be the right place

Assuming physicians

  • Networks of networks, certifying positions
  • Complex social system

Try to map the system, map the market in healthcare

  • How many companies to get a cell phone?  6000 companies for this phone, and there's 6000 different ones for another phone

Retail marketplace, study limited to Fortune 1000

  • Retail is a system that manages the complexity for you


  • e.g. if computer doesn't work, a lot of questions about operating, etc., that people don't want to know


  • Probablistic model to understand state of a network
  • Given any random transaction, the state is the nodes of the network
  • How much information would have to be processed to calculate the complexity of the network?
  • Can create a complexity surface
  • Complexity of healthcare is 27 bits:  for any random consumer transaction, if wanted to know involved nodes, would have to ask a billion questions

Retail:  only half of companies in Fortune 1000

  • Retailers know how to manage complexity, to make it easy for consumers


  • Airlines make taking a flight easy (although annoying)

Telecom is going through convergence with media

  • e.g. if watching news on tv, could grab tv onto cell phone, then throw onto car
  • Companies that reduce complexity for consumers will succeed


  • Consumer-directed healthcare, where people have choice, have an 8% adoption rate:  people want choices, don't want to have to be smart about the network

Two principles

  • We may not want to be innovative, but can be innovative in creating value for business
  • If not creating B2C value, B2B doesn't matter
  • The magnitude of B2C complexity reflects the market maturity
  • In order to decrease complexity to the end consumer, should increase complexity in the business where you can best manage it

Want to design complexity into a system in ways that we can manage it

  • Can manage design, development, manufacturing, sustainment
  • Can't manage economies, markets, competitors, end users

Enterprise agility

  • Used to focus on optimizing systems
  • It works well when the design envelope remains the case
  • Fragile when outside the design envelope
  • e.g. making Cadillac Escalade, except the market has change
  • In June, the sales at Toyota went down 20%, but up at Honda
  • Corollas not on the lot, Honda has flexible manufacturing, putting Accords and Civics on the same line

Another talk:  don't want agile systems or agile information, want agile decision-making, because otherwise don't have resources to take advantage

Management implications:

  • Value philosophy, focus on outputs, not inputs
  • e.g. focus on health outcomes
  • Why bother to have healthcare?  So that we have a productive workforce, not just so that we can feel bad for the sick?
  • Then, how to create most value?
  • Milliken Institute:  Cost of healthcare in the U.S., make diseases cost X, loss in productivity when people are sick is 5X
  • Maybe should increase healthcare spending
  • Total cost 19% cheques to healthcare provider, 48% due to absenteeism

Organizational behaviours, traditional system vs. complex adaptive system

  • Want to arrange incentives and inhibitions so that people do the right thing
  • Measure outcomes, don't have people tell you activities, which is inefficiency
  • Problem with managers is that they don't know what they want
  • Need to be agile to move to needs
  • Want personal commitments of individuals
  • Heterarchy over hierarchy, else lose time to market and innovation
  • Network is self-organized and morphing
  • Worked with IBM to look at social network, VP said doesn't relate to organizational chart, will start managing by the social network

If you're in right column, and pretend you're not, delusion

  • Wrote book on 13 organizational delusions

Information systems can help

  • What is the state of the system?
  • Not balance sheet and income statement, which is the core of scorecarding
  • Productive discussion in itself on what is the state
  • What could go wrong, and still have the decision make sense?  What could undermine?

Boeing Dreamliner:  materials in South Carolina, documentation in Italian

  • People didn't understand implications of what they're doing

Case study:  network models, from top-down

  • How to gain control of health care costs?
  • We need a learning health care system that gets smarter
  • How much does it have to learn?  What's the benchmark, how much smarter does it have to get?
  • 50% increase of cost in healthcare is due to technology:  MRIs, angioplasty, hip replacements
  • How can we afford the innovation?

More than 50% of the people have a mobile phone

  • Each person buys it themself

In the U.S., only one payer for healthcare

  • Would have a problem is there was only one payer for cell phones

Have technological innovation

  • Increases efficiency
  • Increases effectiveness
  • Decreases risk
  • This results in increased use
  • With improved care, people come back and do it again:  the average number of hip replacements is 3, get it early, then replace again later in life
  • Increased use --> increased expenditures

Simple models

Model 1:  If we increase cost at the same rate as increasing use

  • Would have to reduce cost per use every year, very aggressive targets
  • Model doesn't work
  • No way to achieve cost reductions in model
  • Doesn't reduce overall costs
  • Instructive, but doesn't take you much farther

Model 2: Production learning, e.g. cell phones, aviation

  • Every time double production, half the cost
  • If assume 5% growth in demand (modest), or 10% growth ...
  • The more demand, the cheaper it gets
  • A race between demand and sales
  • Demand is swamping efficiencies
  • If have 10% growth rate, just getting costs down 25%, but demand has quadrupled
  • Can't really get there, although can improve things
  • Doesn't suggest where learning is happening

Model 3:  Process model

  • Technology and labour, where technology costs are going down, but not fast enough
  • Supposed technology improvement at 70% is the target
  • Assume healthcare at same rate as GDP
  • If 5% growth rate
  • If have 15% growth rate and GDP is flat, need a 40% improvement in labour, which is extremely ambitious and haven't been seen before

Presented to medical community:  How good would you have to get?

  • Would have to be impressive

Practicalities:  how to limit growth of healthcare to GDP?

1.  Could limit the amount of use, e.g. only one hip replace

2.  Reduce cost of labour:  less labour per use

  • e.g. number of person hours per car
  • Nurse practitioners and assistant physicians instead
  • Self-service, e.g. reservation like airlines

Commonwealth foundation: 15 ways to bend the growth curve

  • Plays into learning models

Putting together a letter for the new president

  • 5th recommendation:  can't improve health care, unless improve education, since 40% of costs are related to lifestyle
  • Reduce tobacco use, obesity, then wouldn't have 10% to 15% growth curve
  • Seven recommendations reduce use rates
  • Nine recommendations reduce costs
  • Mayo:  80% fewer activities per patient, since all physicians are on salary, compensated for health outcomes
  • Mayo:  if come and work with a doctor, first physician you meet is responsible, even if it has nothing to do with their speciality, changed only if there's an explicit negotiated handoff


  • Efficiencies required to gain are immense

We spend all money on clinical practices (people), but we should be spending some time on delivery operations (processes)

  • They think about their own units, e.g. MRI people think about MRI
  • Thinking about whole process give different perspective
  • Delivery operations happen in the content of system structure (organizations), who have budgets
  • Healthcare ecosystem (society), like to think it's private businesses, but there's lots of companies who are really public-private and attached to governments

What can we do?

Value: Either increase health outcomes or decrease the cost of outcomes

  • Processes
  • Our health care is used as a consumable
  • Growing up, value:  I'm better off if you're educated, and I'm better off if you're healthy
  • Now, the idea of public good has faded

Now, trying to take the network model, and make it more hierarchical, and look at how complexity is arranged at levels


  • Because healthcare is a complex adaptive system, can't do command and control
  • Can improve delivery by looking at processes
  • Was viewed as a technician, but now see engineers helping
  • Complexity of healthcare as 27 bits is only Fortune 1000, but it's instructive
  • Case study shows would have to do an amazing job of learning
  • National academy:  can we think of the education system as a system?
  • If can't fix K-12 education, can't fix the system
  • Kia (cars) located a plant in southwest Georgia, discovered that Georgia high school graduates don't have skills to do automobile assembly
  • Disappearing manufacturing jobs, no manufacturing:  they're high-tech jobs


Management vs. leadership.  What is leadership, and should it be farmed outside of organization?

  • This is one issue that we've been looking at
  • Retail industry has doubled in size, half of companies have disappeared, and profits haven't increased
  • Differentiation:  leadership, not just being able to articulate a vision, but leadership involvement, going out and walking around, being interested in being involved
  • Study of mobile enterprise, where people work where they area:  key variable of adoption of the technology and successful deployment is leadership
  • After it's been deployed, leadership is less important, because social network takes over
  • Leadership is focused on influencing the organization, not commanding it; setting values of the organization, getting involved, and showing the value of being involved
  • Not a leadership expert
  • Can't contract out leadership

Elderly, mothers and children lose out?

  • No, it's a time thing
  • Elderly, will see more involvement in the future
  • Last year of life could be seen as a net drain, but we all want good health care, and that's a result of being productive for the rest of life

Mayo Clinic negotiating handoff with the system.  Assume life threatening disease, e.g. prostate cancer, no health care worker.  Who can help to make the decision?

  • Experience with integrated information system, had confusing symptoms, turned out to be a B-12 deficiency, all doctors had all of the same information, including results within speciality
  • More collaborative
  • Don't think that there will be a physician where one person understands everything

40% of diseases could be prevented with improved lifestyles.  Possible to work within the health silo, improves outcomes.  Argue that could improve outcomes further, if had user-centric design, where users have input into income, employment, improving systems from below.  Evidence-based policies from below.

  • Book subtitle:  human-centred design
  • IBM asks:  how should we solve this?  Put 11-year-olds on this, have created a game, hope to open up the game soon
  • To meet the needs of a health care professional, the game is too complicated for the layman
  • Game playing can help with health care

Manchester U. project on flexibility.  Looking at network orchestration.  Definition of orchestration?

  • Study a few years ago, compared symphony, jazz, puppetry
  • Orchestration depends on the art
  • The way the leader works is different
  • In theatre, the leader orchestrates during rehearsals, but the during the show, the play is given to the cast
  • In symphony, the leader stays involves
  • Don't have a definition of orchestration, but have studied different ways of doing it
  • Much more mentoring

Individual players working for the system?

  • Happens over time, have to deal with the status quo
  • Now, looking at how the U.S. government spends money
  • When an operating cost versus an investment?
  • e.g. supposed can reduce cost by $100 billion later, but costs $5 billion now?  Government only has balance sheet, no income statement
  • How can we get people to think about investments different by operating costs?
  • Giving talks, reviewing legislations, lots of little things
  • Inhibition:  shouldn't kill people
  • Playing the game:  don't spend anything
  • Privacy regulations also impact health information systems