The two most important numbers in American health care are 5 and 50. Five percent of people account for about 50 percent of the health system's spending.

Given that health-care spending is almost one-fifth of our economy, that means the health problems of 15 million Americans are consuming almost one-tenth of our GDP -- around $1.5 trillion.

The conclusion is clear: If you want to cut costs -- and relieve suffering -- you somehow need to help that 5 percent.

I spent Thursday at a conference dedicated to doing just that. "Promoting Good Care" brings together some of the organizations dealing with some of the toughest, costliest most complex cases in medicine.

There's Health Quality Partners -- you may remember them from "if this was a pill, you'd do anything to get it" -- which deals with elderly patients who are chronically ill. There's Jeffrey Brenner's Camden Coalition -- you may remember them from Atul Gawande's dive into "hotspotting" -- which deals with the sickest, costliest patients in one of the poorest communities in the country. There's the Nurse Family Partnership, which works with young, at-risk mothers. The list goes on.

What was amazing about the conference is that though these groups serve the sickest people in the country, the discussion had almost nothing to do with treating actual health problems. It was about treating everything else.

All of these organizations have learned the same lesson: The problem with the 5 percent is not simply that they're extremely sick. It's that they're extremely hard to help.

These people aren't just ill. They're poor. They have mental health problems. They're dependent on wheelchairs. They have dementia or brain damage. They live in unsafe homes. They don't have cars. They're agoraphobic. They're worn down by chronic pain. They're stubborn. They're flaky. They're angry and they're fearful. They're bad at talking to authority and working through bureaucracies. They're wary of the medical system and cowed by doctors who don't seem to have time for them. They're locked into bad habits and used to bad environments.

Our health-care system can deal with very sick. Our health-care system is arguably the best in the world at dealing with the very sick. What we're bad at is dealing with everything that happens outside the hospital -- all the things that keep making these people very sick. And so long as all those other things go unfixed, these people keep getting sick, and they keep racking up huge bills -- not to mention facing enormous suffering.

Yale's Elizabeth Bradley conducted a striking study making this point. She begins with a graph you've seen before. The United States spends much more on health care than does any other developed nation:

But our health outcomes don't reflect our investment in health care. We're 25th in maternal mortality; 26th in life expectancy; 28th in low birth weight; 31st in infant mortality.

The reasons behind those low rankings are complex. But one is that even as we spend lavishly on health care we're stingy on the other kinds of spending that support good health:

Other developed countries spend twice as much on social services as on raw health care. The United States spends half as much on social services as on raw health care. Further research Bradley conducted confirms the result: More social spending is associated with better health outcomes, even when the United States is removed from the sample.

Bradley's research included interviewing medical workers. A common theme, she reports, is that "the health care sector is bearing the brunt of an inadequate social service sector." That's a burden it simply can't bear.

You could hear that clearly from the groups at the conference. They were achieving remarkable health outcomes, but it wasn't because they'd invented new pills, or new surgeries. It was because they were doing the difficult, unglamorous work of providing social services and ongoing support to people who are otherwise very sick and very alone.

They visit them every week. They remind them to take their pills. They help them coordinate mental health care. They watch for signs of domestic abuse. They explain what doctors are too busy to clarify. They answer questions people think are too stupid to ask. They work on all the things that make the 5 percent so difficult for the medical system to help so that the only thing the medical system needs to do is work on what happens when the 5 percent become really, truly sick.

It's easy to hear about the 5 percent and assume all they need is more or better medical care. But that assumption is part of why we're doing such a terrible job managing their health -- and their costs.