Decision Science Roles in Medical Decision Making

Medical decisions are difficult for several reasons:

·         They involve multiple objectives, such as alleviating pain and suffering, quality of life, long term health impacts, cost, and so on.  

·         They involve numerous participants:  the patient, doctors, insurance companies, family, employers, and yes, even the government.  

·         More and more data (both good and not so good), is becoming available.  What information is more reliable and what treatments do the more reliable information support?

The question of who makes a medical decision is paramount.  Ideally, for most of us, it would be the patient, and ideally it would be a decision to select that alternative which best satisfies the patients objectives.  (A decreasing minority of people would like nothing to do making with the decision but just leave it up to the doctor).  Determining which alternative is ‘best’ is usually difficult because it requires a synthesis of tradeoffs among conflicting objectives, interpretation of opinions of more than one physician, and an understanding of the complex and often competing data and research results. 

Common practice today consists of a series of medical examinations, tests, and consultations, followed by a haphazard search for additional information.  The patient is often left weighing pros and cons that result from physician recommendations that are: conflicting, biased (naturally so since each understands their specialty best) , and constrained by cost/insurance considerations.  The patient must make a gut feeling choice after sleeping on an overload of information.  Moreover, the decision alternatives may be limited to a set of alternatives that have been pared down because of what some bureaucrat (does it matter if the bureaucrat works at an insurance company or the government?) has decided.

Newer techniques for helping patients decide are evolving.  These typically involve a more systematic way of gather information and presenting it to the patient (see “Weighty Choices, in Patients’ Hands”, Wall Street Journal, Aug 4, 2009 <http://online.wsj.com/article/SB10001424052970203674704574328570637446770.html>.  

However, even if the patient has the latest and best information, the decision is not likely to be one that best meets the patients objectives for at least two reasons.

First, and perhaps at the center of today’s debate on health care, is that patient  alternative choices need to be broadened by reducing restrictions imposed by insurance companies, government, and the ability to pay.  Vested interests are making it difficult to change from what is obviously a poorly functioning system where a patient is rarely even aware of the cost of alternative treatments, and  is sometimes even prevented from choosing  specific alternatives  which are deemed inappropriate by a third party.   Decision science has much too offer in this debate, but has contributed little thus far.

Second, in order for patients to make decisions that best meet their objectives, they must be able to synthesize data and information (much of which is very technical), physician opinions (sometimes from physicians in different fields, each only superciliously knowledgeable about fields outside their own), and most importantly, their own objectives (which are often conflicting).  

Medical decisions, are, by their very nature, subjective.  This is because the relative importance of the patients objectives differ from patient to patient.  But subjective doesn’t mean casual, sloppy, or inconsequential . Subjective simply means that the decision will differ from patient to patient.  There is, in fact, a single best choice for each patient.  Theoretically sound and practical techniques in the decision sciences, such as the Analytic Hierarchy Process,  have been shown to be able to help patients determine the alternative treatment best suited to their individual circumstances and objectives. (A Novel Computer Based Expert Decision-Making Model for Prostate Cancer Disease Management”, Journal of Urology 2005 Dec; 174(6) 2310-2318) http://linkinghub.elsevier.com/retrieve/pii/S0022534705009626.  Not only can such techniques help patients make decisions that best meet their objectives,  reassure patients that they are making the ‘best’ decision for them, are convincing to attending professionals and third parties who may be footing part or all of the bill, but the decisions can actually be less costly to the patient, insurance companies and the government alike.  The use of such techniques is likely to become widespread in the not to distant future.

Management is a form of herding.  Sometimes shepherding, sometimes cat-herding, but most of the time it’s much more complicated — like guiding a distrusting group of friends, who’ve grown apart, to regain the vaguely shared glory of some imagined past.  Envision herding a group of lions and lambs, foxes and hens, insects and birds to a meadow where the herder is paid based on how many people mistake this motley crew for a circus.

Leadership is different: leaders are to groups what cowbells are to cows — they provide a signal that others can follow.  A successful leader polishes the bell, tunes it to the sound of success, and convinces all it is the purest note ever struck.  But a false leader cannot sustain the true tone for long.  The inherent challenge of leadership is then to stay true in times of trouble, to appear — and to be — an honest broker among competing interests.  There are not many leadership tools available, other than experience, intellect, and integrity.  Not the type of things they stock at your local Home Depot. 

One constant I have observed in great leaders is that they allow — demand really — that each person they interact with become better than that person really thought they could be.   There is complexity in this.  A great leader in times of war — Churchill comes to mind — may not be so great in times of tranquility, and of course the other way around as well. 

So, are great leaders born?  Yes, of course.  They are just not born great leaders. 

When the going get’s tough, the tough have some choices to make.  There are some schools of thought here.  The first is that this is a golden opportunity for most companies to lay off the unproductive bottom 10% of their staff.  A second is to reduce future revenue generating efforts (such as new product development projects).  A third is to ask for an across the board reduction in exempt staff salaries.  A fourth is to reduce bonuses to an absolute minimum (especially since most staff will not quit their jobs with the economy as it is).  A fifth is to shift support staff to sales efforts (hence reducing the non-productive time of sales people).  A sixth is to ask for graduated reductions in pay (those with highest salaries receiving a greater share of the cuts).  A seventh is to slash prices on all items that don’t move, while reducing prices on hot sellers less.  An eight idea I heard was to offer all customers an iPhone with their purchases, since that seems to be the only thing that’s still selling like hot-cakes.

But whatever you choose to do, how will you make the decision?  Alone with a bottle of Jack?  With Jack, and a bottle of Perrier?  Or with Jack and Pierre and Otto and Lakysha and Anil and Sergey and Wang?   Today’s companies operate in 24 time zones (or 24 home offices in one city), have wastly different local customs, regulations, and success levels.  What’s perfect for New York may not work in Tokyo, and what’s right in Berlin can seem foolish in Cape Town.

Here’s an outline of process we are using to help our customers:

Part I – And the problem is…
You have to define the problem before you plan a solution.  Too many companies attempt to make changes to their plans without defining what has changed and how the new plan should address this.

Part II — …we need a process…
You can go without a process… you just won’t be very successful.  Circles are nice for doodling.  But don’t get caught going in circles when times are tough.  Without a process, you are walking with one leg a foot shorter — and will tend to circle the room.

Part III — …that we can buy into…
So you have a process.  Great.  Does anyone buy into it?  If your process is opague and non-inclusive, no one will buy into the results.  Open it up to the relevant people and set expectations for how their input will be used.  Just because people participate, the have to know that companies are not democracies — but why not let them bring good ideas into the mix?

Part IV – … and communicate to the organization
Once the leaders decide on what to do, they have to find a way to have the followers follow.  You have your carrots and your sticks, of course.  But usually there are not a lot of carrots in a recession, so that leaves sticks…  Not a good way to have people follow leadership.  So find new carrots that don’t cost money.  Communicate, provide non-financial benefits, give stock in lieu of cash bonuses.  But whatever you do — just make sure you communicate your sincere intent (unless of course your intent it’s insincere, in which case you are unlikely to be reading this.)

So, that’s the process.  Now get going.

By the way, I make software that helps with this.  www.expertchoice.com.

Washington DC’s favorite quadrennial parlor game is underway. Who will Sens. Barack Obama and John McCain choose as their running mates?

While the needs of Obama and McCain are vastly different, the structure of their decisions are essentially the same. Each needs to weigh a range of criteria, including experience, the ability to attract constituencies, and things like age and various intangibles.

I’ve constructed a model to help each determine what really matters when choosing the right running mate, and I need your input. This isn’t an exercise to identify the best candidate, but to determine the criteria they should use when evaluating the candidates (I can add in candidate evaluations, so let me know if you’re interested in blowing this out).

This process helps eliminate the biases and personal preferences that can corrupt the decision-making process. I also feel that with these results, the right candidates will become self evident.

I’ve utilized Expert Choice’s Comparion platform for this exercise. It’s typically employed for collaborative decision-making in organizations, and I’ve had to do a little re-engineering for our purposes here.

It takes about five minutes to do, and I think you’ll find it interesting and hopefully a little enlightening. I wish I could automatically update the results here on the blog…but I’ll be updating them here daily.

Click Here to Participate in the Survey

Click the two images below to view the preliminary collaborative results. Some interesting results so far…

- Much more emphasis is placed on the need for “Experience” regarding Obama’s Veep than McCain’s.

- McCain needs to focus more on a Veep that can attract the conservative base than Obama needs to worry about the liberal base.

- McCain needs someone that has business/management/economics experience, but people don’t think Obama needs the same. That said, Obama may need someone with foreign affairs experience.

- McCain may need to worry more about picking someone with negative baggage.

- Both candidates shouldn’t ignore the need for their running mate to help carry a specific state or region.

I’ve pasted the model for Obama and McCain below. Note that I’ve kept it rather simple for the sake of time and having fun with it. I’m open to suggestions or criticisms.

Sen. Obama
Experience (primary criteria)

- Brings foreign affairs and/or military experience to the ticket (sub criteria)
- Brings business, management, or economic experience to the ticket
- Brings political and/or policy achievement to the ticket

Constituencies
- Attractive to the liberal base
- Attractive to the women vote
- Attractive to the middle class
- Can help carry a swing state or region

Intangibles
- Age (brings balance to the ticket)
- Gets along with Obama / a team player
- Lack of negative baggage

Sen. McCain

Experience (primary criteria)

- Brings foreign affairs and/or military experience to the ticket (sub criteria)
- Brings business, management, or economic experience to the ticket
- Brings political and/or policy achievement to the ticket

Constituencies
- Attractive to the conservative base
- Attractive to the women vote
- Attractive to the middle class
- Can help carry a swing state or region

Intangibles
- Age (brings balance to the ticket)
- Gets along with McCain / a team player
- Lack of negative baggage

The Deciders…

Cooperation and collaboration may rival computers and communications technology as the most improved aspects of business in the past 50 years.   Optimists hope that management by fiat, grunts, and growls will sometime soon be fully replaced by a conscious attempt to arrive at the best solution, regardless of its source.  This, I’m afraid is folly.  Two major problems remain to be solved before we herald in the era of kumbayaesque business practices:

1. The transaction cost of keeping everyone happy, and
2. The elimination of poutology as a career strategy.

Let’s treat them in turn. 

Transaction costs is a technical term from economics.  In it’s simplest form this can be thought of as just what it costs to exchange things in a market place.  More generally, transaction costs can be seen as the direct and indirect costs of obtaining a certain outcome.  So, in a business decision environment we can view transaction costs as the price of paying off the foot-draggers, saboteurs, whiners, and single-focus freaks.   In a highly collaborative environment where consensus and collaboration are valued as an end in themselves (because they are believed to bring about better decision options and execution actions) this cost is not negligible.  Who hasn’t seen a decision delayed, implementation altered, or lowest-common-denominator choices prevail, because someone appeared to object.  This is especially prevalent in hiring decisions, where a single ‘uhmm’ can be seen as the prelude to a veto war. 

Poutological behavior is more pernicious.  While more common among friends and family, pouting is often employed effectively as means to an end in business.   This refers to the act of pouting to gain leverage later.  In a sense, this is a particular type of anchoring bias.  If I pout about a decision today, but then acquiesce ‘because I’m a team-player’ then next time people will, whether they realize it or not, take into account my “magnanimous” gesture.  The net effect is that the pouter has greater leverage coming into the next round of collaborative decision making.   A pouter who is perceived as being invaluable has double leverage since the implicit threat of abrupt departure hangs over the team like Damocles’s sword.

Tomorrows collaboration tools and team training in business school have to do more than just help us collaborate and achieve consensus.  Managing the process of collaborating, disarming the poutologists, and reducing transaction costs are very important to help us avoid the politics of teams, and the dreaded ‘groupthink’ which poorly configured teams often sink to.  Such tools will allow us to prioritise tasks, and maintain a goal-oriented focus, thereby making politically motivated pouting as transparent as a child’s crossed arms and foot-stomping. 

There are three kind of data bigots out there:

1. Those who think their data is precise and torture it to justify any decision they’ve subconsciously already made.
2. Those who think their data is imprecise and then ignore it to justify a decision they’ve subconsciously already made.
3. Those who think Mr. Data should not have been given status as a sentient being, and hence that he was property.

The first two kinds of data bigotry are very common.  They both arise from the same particular psychological bias called “confirmation bias.”  Those of you who have or have been children should be familiar with it:  “You always give him more” is a childish confirmation bias, where the child will take any evidence of advantage to a sibling to be confirmation that the sibling is favored.  Similarly, children will neglect evidence that disconfirms this belief.  Asked if it’s not true that she also got ice cream, the child’s immediate response will be: “Yes, but I behaved better and deserved it” or some similar conclusion.

In business, where we would hope children are not left in charge, we see this same confirmation bias.  An initial hypothesis is supposedly ‘tested’ by systematically collecting and interpreting data to fit it.  I once frequented a little coffee shop that had a great location within walking distance of a major university campus and few competitors.  It should not have been able to fail.  I noticed however that the place kept getting more run down and sloppier as the weeks went by.  Fewer products were on the shelves and not as many people were sitting in there drinking coffee.  It was astounding–I actually saw people turn heel and walk out.  It was an example of a double confirmation bias.  The customers thinking the place was run-down and shabby didn’t trust it to deliver, and the owners seeing customers turn away didn’t want to invest in inventory and upgrades.  Soon it closed, but new owners re-opened a coffee shop and had great success.

The flip-side of using data to test a hypothesis is to deny the availability of data, thus making it impossible to test the hypothesis.  This make it very simple to justify an initial decision.  The example that comes to mind here is strategic decisions where sometimes data is used to justify a decision, but more often than not one or more people will claim no data is available and unguided intuition must be used.  As previous posts here indicate, this is just surrendering to laziness — not a good way to run business.

But why was Commander Data usually referred to as “Mr. Data?”

A company of two has politics in its decision making.  These office politics grow as the number of employees grows.  To examine this, let’s separate between three types of office politics: 1. Those that arise from specialization of functional skills; 2. Those that arise from differences in people’s personalities; 3. Those that arise from internal competition in the company and each employees’ trade-off between corporate and personal success.

We can see the importance of collaboration and establishing priorities as we examine the politics of decision making from these three perspectives. 

First we have the politics that arise from a function-based assessment of how to prioritize a company’s risks and opportunities.  The marketing/financial/operations/sales/etc. perspectives all have value, but they differ in how the evaluate available information, and in the time horizon they consider.  The marketer’s ‘luxury’ of focusing on the future may not be terribly relevant in the mind of a sales manager that needs to meet this quarter’s sales quota.

Then there are personal differences that give rise to politics.  Some people thrive on risk-taking and value a big upside.  Others may think steady-going is the way to go.  Some people like everyone to get along and need consensus building before making decisions.  And then there are the sociopaths and sycophants, the ignorant and the lazy, the smart and the arrogant.  The wallflowers and the wannabes, the we-don’t-do-it-like-that-here old-timers, the know-it-alls, and of course the never-do-wells. 

Lastly there is the simple truth that people change jobs, and that they want to work on things that makes them look good in the eyes of prospective future employers.  Hence, people would rather lead their own cool sexy project with the big budget, than work on someone else’s boring old infrastructure effort.  People’s behavior in order to get their project funded is often the source of office politicking.

Experience tells us that all companies have these dynamics, and that they are the root-cause of negative office politics.  Now it is clearly true that office-politics — especially as a mechanism for fostering healthy competitions of ideas and people — is not all bad.  The true question companies face today is how to harness their diversity of opinion, experience, and attitudes:  that is how to collaborate and prioritize.  Ironically, dispersed workplaces where people have more meetings that are not face-to-face may help here.  When run properly, telephone and online meetings can be very effective as they allow us to focus on the objectives and not the people.  Tools that enable this way of working are emerging–and will provide a way to reduce the negative aspects of office politics, while at the same time enhance the positive aspects.