Tuesday, March 22, 2011

Prognosis is Unimportant


I wanted to highlight a recent article in the NYTimes by Peter Bach entitled, "After a Diagnosis, Looking for a Magic Number" He recounts how after his wife was diagnosed with breast cancer, as a cancer epidemiologist himself, he arranged to see a breast cancer oncologist and researcher and asked him about her prognosis. This researcher refused, saying that as soon as the patient and husband could point to a "probability of recurrence would cause us to make different choices", he would not provide them this information.

The comments seemed quite divided. Some respondents thought the doctor was wise, that providing a "number" would only distract from the goal of doing everything you can to maximize your life in the limited time you have. Others thought that the doctor was condescending, that the doctor wasn't treating these patients as adults.

As a researcher who thinks about prognosis, I am clearly not unbiased in my views, so would love folks thoughts on this.

Not surprisingly, I don't agree w/this oncologist's view. I think it is very difficult to be the 1st one to plant the stake in the ground: It's hard for a patient or loved-one to say if the risks are >50%, then we would (or would not) do X. Similarly, it is hard for the doctor to broach this subject because predicting the future opens us up to being proven wrong. However, because it is hard and uncomfortable on both sides, this conversation is too easily put off, until it is too late. Thus, I believe that the default posture of clinicians should be to discuss prognosis in cases of possibly life-limiting illness.

That said, this specific situation of a husband who is a physician-epidemiologist may be a reasonable situation to not discuss prognosis. The husband is going to be able to get ballpark estimates himself, and I think the marginal increased precision of prognostic estimates from the breast cancer specialist is probably of dubious value.

by: Sei Lee

9 comments:

ken covinsky said...

On the surface, it seems inappropriate for the oncologist to refuse to answer a question about prognosis. It's the decision of the patient and family how much information they want to have.

As you allude though, there is probably another dynamic at play here--as Dr. Bach is hardly the typical spouse in terms of his background knowledge--and the Oncology Doc probably knew him pretty well. So, the oncologist knew that Dr Bach already had a very good ballpark estimate of the answer--but was asking for a very expert precise estimate.

And in reality, precise estimates are only useful at the epidemiologic level---they are not as useful at the patient level, because there is so much noise around the estimates, and the population estimates can never account for all the characteristics of the patient in front of you.

So, this was probably less a refusal to provide an answer to a question about prognosis, but a way for the oncologist to offer supportive advice to his colleague--basically telling that it was fruitless to overly obcess about the exact prognosis---because the exact number did not matter--and his way of saying, "no matter how much information you have, you will not be able to control the outcome, so what you two you need to do is go on living your lives."

Christian Sinclair, MD said...

I am confused as to the title of the post. "Prognosis is unimportant" seems to go against the position you defend later in the post. Could you clarify why you choose this title Sei?

I thought the doctor in the article did a poor job of shirking his professional duty.
I think the majority of people are not looking to the doctor as the sole provider of prognostic data as covered in the 2011 AAHPM State of the Science lecture: http://www.ncbi.nlm.nih.gov/pubmed/20228686

The main result from that study was: Less than 2% (3 of 179) of surrogates reported that their beliefs about the patients' prognoses hinged exclusively on prognostic information provided to them by physicians.

Ken, you state in your comment: "And in reality, precise estimates are only useful at the epidemiologic level---they are not as useful at the patient level, because there is so much noise around the estimates, and the population estimates can never account for all the characteristics of the patient in front of you."

I would say that prognostic estimates communicated with acknowledgment of uncertainty and with ranges (not precise definite times) serves a very important purpose to good planning on the individual level. As long as we continue to say we are poor at predicting we will continue to be poor at predicting. We need a lot more research into prediction science.

I also commented on this article over at Pallimed: http://www.pallimed.org/2011/03/what-are-my-chances-doctor-narrative.html

catherine.dubeau said...

I actually thought this was a great example of what we do in geriatrics, which is focus on implications for function and goals to drive treatment decisions (and advance directives). Also, that these implications can and often do change over time. This approach moves us off the focus on numbers (what my father, in his wisdom, calls "Jimmy the Greek medicine"), and onto the person (and family, etc) in the context of their life. If the number/odds isnt going to change what you do at the moment, then it does become less relevant. The writer and his wife may go back and decide that yes, knowing the percentages may make a difference for action at this moment (eg, if her income is crucial to a college fund for their child) - and the oncologist at that point should then discuss them.

Catherine DuBeau

Anonymous said...

I live in Florida. Every year, as summer aproaches, in preparation for Hurricane season, I buy water, canned foods and batteries. I am a realist. I live in a hurricane prone place. On the othe hand, I also hope and pray that these hurricanes won't hit my part of the world, because I am an optimist and because, from time to time, we are spared.
My point is that we as clinicians can be both, bearers of bad news and cheerleaders. Patients need to know what they are up against. Dr. susan LeGrand, from the Cleveland Clinic prepares her patients from day one for the day when the conversation will change from prolonging life expectancy to concentrating on comfort and quality of life.
We have all had patients who beat the statistics, but they remain the exception and, until we have better tools with which to battle severe illnesses, we owe it to our patients to be there with truth and hope.

Marian Grant said...

While I can see that in this case, the oncologist assumed the patient's physician husband could access prognostic information on his own, not discussing it at the outset sets the expectation for what can and will be discussed in the future. That means that other things related to future planning might also not get covered, which is a bad strategy. Prognosis should be honestly provided, especially if patients/families seek that information. A good friend of mine has serious lymphoma and the 50/50 chance she was given helped her make important decisions and plans for her young family.

Sara Perry said...

One possible scenario is that the oncologist in the article had previously been more precise in his prognoses but after one or more of his patient/families did not use the information well, he decided he wouldn't give that info again.

But that would be withholding important info from the many pt/families who would use it well. As a hospice social worker, I have seen many more pt/families use an estimated prognosis wisely than not.

Similarly, I have heard patients and family members say to me something like the following: "If the doctor had given us more information about the prognosis and the pros and cons of receiving another round of chemo, we would never have chosen to have the treatment. It absolutely wiped him out and he went downhill very quickly after that. It wasn't till afterwards that we learned from other sources that the odds of it doing more good than harm had been very low. We thought that if the doctor was offering the treatment, s/he believed it was very likely to do more good than harm. Otherwise, why in the world offer it?"

gsnipes said...

Why not REALLY treat the patient/family as adults and, if the physician HAS a reliable statistical estimate of prognosis, ask them what form of information would be most helpful to them? That would, of course, require one to spend a little time explaining the limitations of statistics so as not to frighten them. But if they want to know, and we can tell them, we have the obligation to do so, and then help them put it in perspective in their particular situation.

Gail Austin Cooney said...

Because I'm both a hospice & palliative medicine professional and in treatment for a recurrence of ovarian cancer, this is a subject of GREAT interest to me! I went back and read both the original NYT article and "The Median Isn't the Message" by Stephen Jay Gould. Then, I reflected back on my own approach to prognosis - and recognized that it has evolved over time.

When originally diagnosed in 2008, my first reaction was despair. I have cared for many women dying of ovarian cancer and it can be a difficult and challenging disease. Luckily, the initial treatment for Stage III ovarian is fairly straightforward: surgery, then intraperitoneal chemotherapy. As I progressed through treatment, I began to move closer to Stephen Gould's approach: over one third of women DON'T die in the first 3 years - why shouldn't I be one of THOSE?

It's really a position of optimism, not denial. When I make choices today, I am very aware that I may not be alive 5 years from now. But, perhaps I will. And I make life choices that accommodate both outcomes. As Gould states, if the median survival is 3 years, then 50% of people live longer - that's a lot of people! (OK - I'm mixing up my data but I think that's a good thing - I don't really care if it's a third or a half - either one is OK for now!)

When discussing prognosis with patients and families, I think that the same approach can be used. As treatments become increasingly burdensome and toxic, however, it's important to include how MUCH time a treatment is likely to add to survival - when it comes down to months rather than years, I may be making different choices.

Or, maybe not. We'll just have to see!!

sei said...

Wow! What a wonderful diversity of thoughtful opinions!

My take-away from this discussion is how much my views are formed by my experiences. Of being the palliative care consultant who is asked to consult on a patient with a life expectancy of hours and days. How it often seems that the family is ready for a symptom oriented approach, but that in the name of not taking away hope, previous doctors had chosen not to discuss prognosis.

An ER attending once told me that he hates whatever drug is recommended for outpt treatment of bronchitis because he sees all the cases where that drug fails. Maybe my views about prognosis suffer from the same bias--I don't necessarily see the cases of prognosis discussions that go well.