The Tipping Point

When I wanted to figure out whether it was time to replace our 2004 Toyota Camry, which runs beautifully and is the most comfortable and reliable car we’ve ever owned, I pulled out its repair records and graphed money spent on fixing the car as a function of time. What I was looking for was an inflection point or, as Malcolm Gladwell might call it, a tipping point. I wanted to know whether our annual maintenance expenses for the car had begun to creep up or, more importantly, if the curve had abruptly gotten a lot steeper. I was pleasantly surprised to discover that after the first couple of years, when our only costs were for oil changes, the annual costs had remained rock stable. The car has 125,000 miles on it; at this rate, it might make it to 200,000.

Now I can’t be certain that this calculation has any prognostic significance. It’s possible that there could be one very expensive year, a year when some particularly costly part needs to be replaced, but that large outlay is a blip on the curve; after that one year, annual expenditures might go back to normal. Drawing conclusions after looking at the high expenditure year and not waiting to see what happened the next year could lead to the erroneous conclusion that the vehicle had begun a period of relentless decline. And I have no evidence that even if my system makes sense when applied to cars, it would also work for people, but it set me to wondering.

The problem with people, and the reason that I’m hoping the car analogy might prove helpful, is that sometimes it’s hard to know when to stop “doing everything,” as patients often put it. Occasionally, it’s easy: there are patients, for example, with widely metastatic cancer who are clearly and unequivocally dying. They may not want to accept that the end is near, but it’s obvious to any good oncologist or palliative care physician. It is often obvious to families and even to the patient as well. There are patients with very advanced heart failure whose hearts can barely pump any blood and who are far too sick to withstand highly invasive treatment such as a heart transplant; they also may not want to acknowledge that they are dying. Their situation may be compounded if their doctors also don’t want to admit that they're dying, but there is compelling objective evidence that they are at death’s door.

Much of the discussion about death and dying these days centers on patients like these. And yes, it can be difficult emotionally to let go, but the truth is that these aren’t the hard cases. The hard cases are people who aren’t, as Katy Butler put it in her book about a better path to death, knocking on heaven’s door. The challenge is to figure out when to pull back before you reach that point and not necessarily to pull out all the stops, to go from maximal medical care to an exclusive focus on comfort, but rather to something in between. I have talked a great deal about this kind of intermediate care, particularly for frail older people. But how do you know when it’s time to switch gears—not necessarily from fourth gear to first gear, but perhaps to second or third?

Palliative care physicians have to deal with this question all the time. We have to figure out “eligibility criteria” for palliative care consultations. A recent state initiative in Massachusetts mandates that “suitable patients” be told about the availability of palliative care for people in their condition. Some useful guidelines have been developed: patients with heart failure who have had two or more hospitalizations or emergency room visits in the past year are good candidates; patients with dementia who have lost a great deal of weight or have had recurrent infections would benefit. Several interesting studies have identified “markers” for the beginning of frailty: a serious fall, for example, or admission to a nursing home. But the reality is that we don’t have a great way of identifying patients who are likely to be entering the final phase of their lives. It would be very useful to figure out who such patients are by using “administrative databases,” that is, insurance company records of doctor visits or hospitalization or lab tests. Hence my interest in inflection points.

I noticed recently that someone in my family had an abrupt increase in the monthly number of doctor visits, lab tests, and procedures. Instead of seeing her primary care doctor once or twice a year, she was going every month or two. Instead of very rare visits to sub-specialists—a trip to the ear doctor once a year to have wax removed, a visit to the dermatologist once a year for a skin exam, and an annual appointment with the eye doctor—she was seeing these specialists more frequently, along with other doctors: a surgeon, a rheumatologist, a gastroenterologist. I couldn’t help wondering whether she had passed a tipping point.

We have to tread carefully here. After a single, isolated acute illness, a patient might have a transient increase in “health care utilization,” as the economists put it. When mapping trajectories in the last year or two of life, we know that some patients don’t follow a steady path but instead become transiently frail—say after a small stroke or a hip fracture—only to return to their previous level of functioning after rehab.

So we have to plot out doctor visits and lab tests over a long enough period of time to be able to distinguish between a blip and an inflection point. If we insist on too long a time frame, then the approach ceases to be useful—by the time we know for sure that there was an inflection point, the person could be dead. 

I never used to like the people-as-machine metaphor—patients in many ways aren’t like machines whose parts wear down. But maybe, just maybe, old people are a bit more like old cars than I thought.