Taking Control of Your Health: A Q&A With the Author of The Decision Tree


For much of human history, if you came down with a fever (or pretty much anything else) your doctor would “bleed” you – and most likely go on his way without washing his hands. While technology has drastically increased our understanding of human anatomy and disease since those days, many useful technologies are never applied to health care — or, worse, drive health care costs ever higher while offering dubious benefits. We have more data than ever about our own health and the causes of illness, yet assimilating those data to make intelligent, cost-effective healthcare decisions can seem impossible.

Thomas Goetz, the executive editor of Wired, has written a new book that he hopes will give people some tools in this effort. The Decision Tree: Taking Control of Your Health in the New Era of Personalized Medicine tackles everything from genetic testing and pharmacogenomics to the challenges of long-term behavioral change. Goetz has agreed to answer some of our questions about his book:


What exactly is a decision tree, and how can it change the way people make health care decisions?


One of the biggest challenges in healthcare is how to give individuals a way to make better decisions – a way for them to engage with the huge amount of information that our doctors or care providers or the Internet offers us.

“Engagement improves our odds.”

Thomas Goetz

Helping people make good decisions is essential, because research shows that when people engage in their health, they tend to have better outcomes. Engagement improves our odds. This is true when people are trying to engage in preventive medicine, when they’re facing a diagnostic test, or when they’re weighing various drugs or treatments. At all of these moments, we are trying to navigate our way to a good choice, a choice that corresponds with our values and gives us our best chance for a desired outcome. But this is an incredibly difficult process for nearly all of us, because the information comes in so many unfamiliar forms – genetic results, statistical probabilities, side effects – and there is so much uncertainty in the equation. So we often leave decision-making up to our doctors, or we don’t really engage in the information in a way that might otherwise improve our outcomes.

As I was writing the book, researching all these new tools for engaging with information in more meaningful ways, I realized that there was a way to visualize this whole approach of engaged, patient-centric healthcare. We can give people a structure for thinking about their options, for thinking about how one choice leads to or precludes others, and then perhaps people will have an easier time taking advantage of all the superb progress that modern medicine can provide. That’s what led me to the notion of a decision tree. A decision tree, essentially, is a flow chart – it’s a kind of algorithm, a way to factor in various strands of information and probabilities to move towards an optimal outcome. So it’s a way of thinking about our healthcare as a series of deliberate choices, where we are the primary decision-makers. When we take on that role, we tend to have better health.


What role should genetic testing play in medicine?


The experts that I talk to in the book are incredibly hopeful about how our growing knowledge of genetics might improve medical diagnosis and treatment. But it’s important to realize that, when it comes to disease, our DNA is not entirely our fate. Yes, there are certain diseases that are entirely genetic, but for the vast majority of diseases, genetics only play an influential factor alongside our environmental exposures and behaviors (basically, all the stuff that we ourselves do). To me, this means two things: First, DNA isn’t something we should be fearful of; it’s not destiny, it’s simply an influence that we should consider and try to understand. And secondly, this means that DNA is a welcome “baseline” – a starting place for us to understand what we may want to be mindful of, down the line. Knowing our genomes can help us flag risks so that we can then deploy other behaviors – diet, drugs, what have you – to navigate away from those risks. As genetic sequencing gets cheaper, this is where I think DNA testing will end up: a test at birth that is used alongside our blood type as a dataset that may be revealing but not decisive. It’ll be one more arrow in the medical quiver that can help our doctors (and us, too) understand what might be going on in our bodies.


Behavior change is hard, yet it’s often a requirement for improved health. What are the secrets to successful behavior change?


I wish it was as easy as spilling the secrets. There’s been no shortage of seven or seventeen things that will change your life, and I won’t pretend I have my own set. But I do know that behavior change is probably the single greatest challenge in healthcare today – so many of the things that people suffer from and die of are based in our habits or antics, from smoking to diet to sloth. But that also makes behavior change a great opportunity, because if we can get people to embrace change and succeed, then you’re really going to improve and save lives.

So here’s what I did find: there are two essential things that work to get people to change. First is feedback – giving people some sense of where they stand, and a personal understanding of where they have to go. Research shows that when people can see their personal progress, they tend to have higher rates of success at changing their behaviors. The second thing that works is communities or groups. When we are one among many, all working on the same goal, people tend to have much better success. In part this is because we can compare ourselves against our peers; but it’s also about communication and confidence.

These two principles have been well known in public health circles for decades; they’re the magic behind everything from Weight Watchers (which really does work) to smoking cessation. They work, but they take significant resources and discipline. But here’s the thing: the Internet and information technologies are now making these strategies less resource intensive. Feedback can be as easy as clicking a few buttons on your iPhone, and “communities” is just another word for “social networks.” This is the moment we’re at today, and I think we’ll start to see some remarkable innovation in making these strategies more available to more people.


Now that we have screening tests for so many diseases, why don’t we just screen everybody for every disease?


Short answer: False positives would overwhelm true positives. Long answer: Screening tests are a terrific strategy for spotting disease early, and saving lives. They’re the reason cervical cancer is no longer a leading cause of death among women – thank you, Dr. Papanicolaou for the Pap smear – and they’re the reason the death rate from heart disease is only a third of what it was in 1950 (taking somebody’s blood pressure is a very simple sort of screening test).

But a screening test requires a calibration between the cost of the test, the accuracy of the diagnosis, and the population of people at risk. The calibration results in a kind of arithmetic: If a test is very cheap and very accurate – like a blood pressure test – then by all means it should be widely deployed, pretty much universally. So it’s suggested that everyone over 18 get a cholesterol test. But as a test gets more expensive and more intensive, we need to limit the population. So a colonoscopy is a superb way to detect colon cancer early, but experts only recommend that men over 50 get the test. Even if you put cost aside, it’s unrealistic to give every test to every person because we’d lose the positive signals in the noise of false positives – those people who the test says have a disease, but actually don’t. False positives, of course, are a normal result of screening, and ideally any positive test will be followed up with a second, more precise (and probably more expensive) test. But unless we limit the number of people tested to begin with – unless we choose the right pool to begin with – even the best test will generate more false positives than the system can handle.


How effective are drugs? Is there any way we can make drugs more effective?


One of the amazing things I learned in writing the book is that many popular drugs don’t really work all that well. Anti-depressants work at best around half the time; chemotherapies for cancer work at best 20% of the time. This is the hit and miss of modern pharmacology – there are no sure things. We can improve the efficacy of drugs in two ways. From one direction, we can hunt for more effective drugs. That’s been the path the pharmaceutical industry has pursued for the past 50 years, but it’s pretty much run out of steam. The other way to improve efficacy is to only give a drug to the people you know it’ll work for. So if Gleevec works for less than 10% of leukemia cases, identify what those 10% have in common – turns out it works almost perfectly for Chronic Myeloid Leukemia — and only give the drug to those people.

This idea of drug targeting (or pharmacogenomics, as it’s called) has been around for decades, but the industry has been slow to move, loathe as it’s been to leave the blockbuster model behind. But that’s starting to change, and the hope is that in the next decade there’ll be more development of drugs that work for smaller populations.


In the (increasingly unlikely) event that healthcare reform makes a comeback, what are the three reforms you’d most like to see?


I still hold out hope for healthcare reform, but three things that we should insist upon whether in a big bill or smaller legislation are:

  1. more evidence-based medicine – which involves things like comparative effectiveness research, so newer (read: expensive) drugs are compared to older (read: cheaper) drugs
  2. more pay-for-performance targets for physicians, so our doctors have greater incentive to pursue prevention, and have ways to get compensated for keeping us healthy (rather than waiting for treatments)
  3. a revised HIPAA privacy rule, so that it’s easier for doctors to share patient data with the patient, and its easier for patients to share their data with each other. These three things would go a long way towards creating a smarter, more participatory kind of healthcare.

Pat Crofoot

"more pay-for-performance targets for physicians, so our doctors have greater incentive to pursue prevention, and have ways to get compensated for keeping us healthy (rather than waiting for treatments)"

There is a tiny movement where patients pay an annual fee for a Family Doctor / practitioner. That covers all checkups, physicals, phone calls etc..

Typically, the patient carries a drug and catastrophic plan through an insurance co.. to minimize risk and keep payments low.

This lowers the costs for the patient, the doctor, and eliminates the HMO model (decreasing costs for employers), and still maximizes the lowering of risk to the patient should something happen.

Also increases competition between doctors for patient loyalty and referral.


1. Pay4Performance has little to do with prevention. It is about playing the game. Data from the UK demonstrated that most were placed in outlier categories instead. I've seen certain offices/networks massage the data to fit the desired goal (e.g. contact the patient to fulfill the goal - in writing.)

Prevention will be practiced when it is adequately reimbursed - or reimbursed as well (or better) than the procedures to treat the (hyper)acute event you seek to prevent. To be really crass - what is the cost to a practice for vaccinations (procurement, equipment for proper storage - and monitoring, the evaluation and counseling to the patient/family, the administration (MD or RN?) and overhead - rent/utilities, insurance (business/malpractice) and the system to record it all versus the reimbursement for same.

2. We think of medicine as science. Clear and Pure. It is not. It is about probabilities for a particular time or event.

Diagnoses is about probabilities - given these symptoms in this time and space configuration, in a person of this age, in this environment, etc., the likely diagnoses are X, Y and Z with X the likeliest. (The doctor who knows everything and does everything is the pathologist - but he's too late as the adage goes.) But life (and medicine) is a process.

We envision a day when medicine can be as "precise" and "accurate" as Bones McCoy seems to be on Star Trek -with instant read outs for exactly what is awry. Given knowledge of one's DNA (and epigenome) we then know EXACTLY how you will respond to treatment A, B or C. This is why you physician asks about your family. We make certain assumptions based upon initial evaluation (young, old -er than stated age, use of tobacco, alcohol, certain other pharmaceuticals.)

3. Quality - or really Value - as noted by M Porter and E Teisburg - is really about evaluation of practices based upon better delineation of the subsets, i.e. demographics and the disease treated much like the old ECOG, SWOG and CF Foundation trials and registries. Those subsets of patients may reimburse differently - and based upon those quality or value metrics. So reward those who do well with the uneducated, or the poor, or the rich or the whatever metric.

As someone from WIRED I would expect a call for some sort of national database, or at least a standard for all EHRs to easily communicate (rather than the myriad patches); the ability for the patient to access and ?comment/correct their record but not expunge any parts. (This 2ould not be accessible from the Internet because of the security level needed. (I'm not certain I trust VPNs. sorry.) It would really require a separate network, accessible only after certification/registration from some national entity.)

4. Research/Trials need to be revamped in funding and publication/dissemination. Have PhRMA and the device manufacturers a) pay fees to FDA for monitoring post approval for safety, etc that may be prorated for time from approval and separate for once generic is introduced b) pay or contribute into funding for national clinical trials clearinghouse that provides the funding for trials. This helps to assure that trials are appropriately set up to provide meaningful data (and powered, and safety, etc.). Companies could not avoid head-to-head studies (as they do now). The initial (phase I-III) trials that are used for approval might be fully funded by the company but must be approved by the clearinghouse.

5. Change Malpractice to arbitration based. This removes stigma and encourages full evaluation of each and every situation for how the process/knowledge base could have been better. Are the issues individual (related to the individual case - patient or practitioner's knowledge base) or potentially more generalizable - a class of patient conditions, a process, etc. Then improve the appropriate entity/process.

Despite IOM's report, there is still a desire to find "someone" to blame rather than merely improve so "it" doesnt happen again.



Who says that behavior change is "hard"?

I've found that I can change my health-related behaviors when I choose to. For example, I once decided to eat less food and get more exercise -- and so I ate less food, and I got more exercise, for over a year, and I pretty easily achieved (and largely maintained, to the extent that I chose) a normal weight. The fact is that it's no harder to put a small serving of chicken and two vegetables on the dinner table than it was to put a large serving of deep-fried chicken and one vegetable and two servings of starch and half an ounce of butter on my plate.

At some level, you really just have to decide to do it -- yes, re-deciding that every single minute of every day, in some cases, but it turns out that the potatoes didn't actually leap off the shelf at the store, carry themselves into my home, cook themselves, and force me to eat them.

Do you mean that changing someone else's behavior is hard?

That seems highly likely to me, but I don't think that it's my business to change someone else's behavior.

I know someone who has weighed about 400 pounds for her entire adulthood, with heart disease and diabetes and other predictable problems. She's almost certainly going to die young.

She does not choose to eat a healthy amount of food because (I'm not making this up) she believes that life isn't worth living if you can't eat as much as you want of whatever you want, whenever you want it.

Is it my job to make her value living longer over indulging her childish delight in a pig-sized dinner? No.

Is she choosing to commit suicide with her fork? Yes.

Is changing her behavior hard? No: It's impossible for me to change it -- and it's easy for her. She simply has to decide that changing her behavior is absolutely THE most important thing in the world -- more important than the comfort of a stuffed belly, more important than showing the world that she can do whatever she wants, more important than anything.

This is a decision that I seriously doubt she'll ever make, but if she made the decision, the change would follow naturally and easily. Many thousands of people with cancer, and survivors of heart attacks, have made these kinds of decisions and successfully changed their behaviors. It's not actually hard, once you're motivated enough.

In the meantime, she's paying money into Social Security that she's highly unlikely to live long enough to collect: She's "driving up health care costs", but the net effect is that she's subsidizing retirement for healthy people.



I (dis)like the way the conversation just brushed right passed the largest set of factors: environmental exposures and social ecology.

Paul Storey

check out this article from the weekend's
NY Times.

and the PBAC decision from 2007

The PBAC is the Australian expert advisory committee that advises on the subsidy of drugs on the national health system the Pharmaceutical Benefits Scheme. See here

It seems that the PBAC creates a totally different (unique?)
perspective on drugs - "is it worth more than the comparator?" in addtion to the value vs risk notion.

There probably isn't another unbiased, transparent and
accountable body in the world publishing research on the
economic value of drugs. At least, not one that has
manufacturers putting their best foot forward in a genuine
attempt to make the case.
Also, with the PBAC, there is a genuine incentive for pharma to get approval and therefore seek to deliver improved health outcomes. The PBAC isn't just trying to make the manufacturers jump through hoops, which is what also makes this information unique.



@htb - I'm not one for whom behavioral change is especially easy or hard, but I wouldn't presume to say that my personal experiences can be extrapolated to others. The book Good Calories, Bad Calories convinced me that when it comes to losing weight in particular, some people have metabolisms that will literally cause them to starve to death before they will get down to a normal weight. Their bodies are somehow programmed to make sufficient glucose available to the bloodstream only when their body fat percentage has risen above a threshold. Granted, in most cases that threshold can be lowered by consuming fewer carbohydrates (and less sugar, in particular), but there do appear to be those for whom obesity is nearly unavoidable.


PaulD: how did those genes survive the last few millennia of human history?


PaulD (@6), I don't believe that there are any women for whom a body mass index of 70 (yes, *seventy*) is genetically or metabolically unavoidable -- and that's the BMI that you achieve when you are five-foot-three and weigh 400 pounds.

The issue for me is whether you and I, in the name of 'public health' and 'her best interests', should be able to tell an otherwise responsible adult how much food to eat.