Near the top of a long, interesting story in Sunday’s Washington Post, in which he argued that improved medical treatments are a key factor in the rising cost of health care, reporter David Brown wrote the following sentence: “Over the same period, the charges for treating a heart attack marched steadily upward, from about $5,700 in 1977 to $54,400 in 2007 (without adjusting for inflation).”
That same day, the influential progressive blogger Kevin Drum, who has beat this drum before, took Brown to task:
I continue not to understand why anyone would write this. Why not this instead?
Over the same period, adjusted for inflation, the charges for treating a heart attack marched steadily upward, from about $20,000 in 1977 to $54,400 in 2007.
Technically, Brown’s wording is correct. But it’s not helpful, since most people don’t have even a vague notion of how much cumulative inflation there’s been since 1977. The revised wording, however, is helpful: it gives people a correct impression of how much more we spend treating heart attacks these days. Namely, two to three times as much as 30 years ago.
This wasn’t just a slip of the keyboard. Brown and his editor obviously made a deliberate decision to use nominal figures even though this doesn’t give the average reader a very good idea of how much costs have actually risen. I’d sure like to hear their explanation for why they made this decision.
I thought I’d like to hear the explanation, too, as Drum’s point about the relative merits of adjusted and unadjusted dollar figures is both persuasive and widely held—there’s a reason, after all, that inflation-adjusted dollars are called “real dollars.” So I e-mailed Brown and Frances Sellers, deputy national editor at the Post, each of whom quickly acknowledged that it was a “mistake,” in Brown’s words, not to supplement the nominal data with adjusted figures.
Brown also said that, while reporting the article, he had converted the 1977 figure to 2009 dollars using a tool on the Post’s in-house Web site. His decision not to use the adjusted figures, he said, may have been influenced by a conversation with Dr. Anne Elixhauser, a senior research scientist at the federal Healthcare Cost and Utilization Project, which provided the data for costs in 2007 (and for an accompanying chart showing cost trends from 1993 to 2007, which also used unadjusted numbers). Brown asked Elixhauser if her agency adjusted its figures for inflation. “She said they didn’t,” he told me, “because doing so is in itself controversial.”
It’s true that there is some disagreement about adjusting medical costs—but the issue is mostly over how, not whether, to adjust the figures. Some experts prefer to adjust using a standard tool that reflects changes in the general economy, Elixhauser told me. Others prefer a different tool that treats medical costs differently. (She now prefers the former approach, but has in the past used the latter.) By providing raw data in nominal dollars, HCUP allows analysts to choose the approach they prefer; however, she recommends that dollar figures be adjusted.
Elixhauser did add that some professional colleagues prefer to use raw, unadjusted figures, and though she disagrees, she thought doing so was “defensible.” “It’s not black and white,” she said.
That’s a more generous position than the one taken by the journalist and author David Cay Johnston, who won a Pulitzer for his coverage of tax policy while at The New York Times. “It is literally worthless to tell you the numbers over the thirty-year period without telling you the rate of inflation,” he said, when asked about the Post piece.
There is one circumstance in which it’s appropriate not to adjust for inflation, Johnston said. That’s when tracking the relationship between two figures over time—for example, how the change in medical costs compares to the change in the median wage, or how home prices compare to median family income. But otherwise, “anytime you use data that go more than a couple of years, you should adjust.” And in periods of especially high inflation, that window narrows.