President Barack Obama pledged on Wednesday that “reducing the waste and inefficiency in Medicare and Medicaid would pay for most” of his health care plan. This echoes remarks from Peter Orszag, his director of the Office of Management and Budget, who has claimed that one-third of health care spending, more than $700 billion, is wasted annually. Those Orszag comments come straight from the Dartmouth Atlas, which announced that the United States could save 30 percent of its health care expenditures if high-spending regions were more like low-spending ones. But this can’t be how we’ll pay for reform. The numbers are too good to be true.

The Dartmouth Atlas Project uses Medicare data to assess how health care resources are distributed throughout the United States. Dartmouth has long maintained that “Medicare spending is closely correlated with overall spending,” as it argued in the journal Health Affairs last year. Later, in rebutting an article I published in the same journal, Dartmouth officials asserted that “Medicare accurately measures what doctors and hospitals do.” But officials at the Center for Medicare and Medicaid Services, who have the responsibility for these data, have written in the same journal that “Medicare spending does not explain much of the variation in total per capita health care spending.”

Last month, a group of physicians, including the director of policy-reform efforts at the Dartmouth Institute for Health Policy and Clinical Practice, acknowledged in an op-ed outlining “10 steps to better health care” that because “we relied on Medicare data for our selections, it is possible that some of these regions are not so low-cost from the viewpoint of non-Medicare patients.” In other words, Dartmouth admitted that Medicare is not a proxy for the whole while also downgrading (to about 16 percent) its forecast of likely savings based on regional reforms. Orszag and the president should take note: They are unlikely to find $700 billion in Medicare and Medicaid alone.

What’s more, the Atlas findings may not even be a map to Medicare spending and savings options. Only about half of the spending for Medicare patients is actually paid by Medicare. The rest is paid by the patients’ supplementary insurance, known as “Medigap,” or by public Medicaid or by patients themselves, and a lot is simply written off (though some bad debt can be billed back to Medicare). Even Medicare’s base rates vary. While in theory the rates are national, they are adjusted for a host of regional, local and political factors — for example, a hospital in Green Bay is reimbursed at Chicago rates, thanks to the efforts of a friendly congressman. It’s all quite variable, and much of the variation depends on the economic status of patients and of their communities.

If a hospital is lucky enough to be in a community such as Rochester, Minn. (home to the Mayo Clinic), where few patients are poor and even fewer are uninsured, it can expect to make out handsomely from its Medicare patients and everyone else. This is why total health care spending in Minnesota is not rock bottom, as the Atlas suggests, but in the upper 20 percent nationally. And Mayo’s total charges for common illnesses are among the highest in its region. This, together with revenue from paying patients who come from elsewhere, allows Mayo to provide high-quality care, which all of its patients get, regardless of what they pay. Sadly, the reverse is true at hospitals where total revenue is low — without enough total funding, all patients suffer.

Orszag has argued that if Medicare spending could be as low in Newark as it is at Mayo, the nation could save billions. But this theory doesn’t hold up in practice. Consider: One-fourth of the folks in Newark live in poverty, compared with less than 10 percent of those in Rochester. And national surveys show that poor people consume more health care resources — 50 to 75 percent more than average. They are sicker and they stay sicker, despite the best efforts of physicians and hospitals. Mayo is a fine institution, but it isn’t more cost-effective than other hospitals in its home region, nor are its operations in Jacksonville, Fla., and Phoenix more cost-efficient than other hospitals in those cities. So why would it be more cost-effective in Newark?

To really achieve health care reform, and find a way to pay for it, the president will have to give up on the Dartmouth suggestion and grapple with some painful truths. First, medical care is inherently variable in different regions of the country — socio-demographic differences matter. Second, more is more and less yields less — the best care is the most comprehensive care, and it costs more. Finally, poverty is expensive — the greatest “waste” is the necessary use of added resources when coping with patients who are poor. If we want a technologically advanced, socially equitable health care system, we will have to organize our finances accordingly. There is no quick fix. That’s what we should be talking about.