For years, the government has collected a rich store of data about the performance of individual dialysis facilities. But it has kept nearly all the information secret from those it might benefit most: Patients.

Now ProPublica has obtained this data under the Freedom of Information Act. We are making a comprehensive set of clinic records publicly available for the first time on our website.

Patients and others can search for a clinic and see how it compares on 15 key measures, ranging from mortality and hospitalization to transplant rates and infection control. Also on the site are historical reports dating to 2002.

Release of the data is long overdue, patient advocates say.

"It gives you a snapshot of what a clinic is about," said Roberta Wager, a past president of the American Association of Kidney Patients who works as a nurse and patient educator at several dialysis clinics in Texas. "This is your life. Wouldn't you want to have everything in your favor?"

There are almost 400,000 Americans who depend on chronic dialysis to do what their failed kidneys cannot, a number that has grown swiftly over the past two decades, spurred by epidemics of obesity and diabetes.

More than 5,000 facilities have sprung up to provide them with care, stretching into the nation's most rural areas and competing for patients in urban and suburban areas.

Patients today have more choice than ever. Yet most pick centers based on convenience, or on what their doctors suggest, with little notion that even clinics within the same communities can have substantial disparities.

In more than 200 counties nationwide, the data show, the gap between facilities with the best and worst patient survival, adjusted for case-mix differences, is greater than 50 percent. In areas such as Allegheny County, Pa., or Franklin County, Ohio, each with upwards of two dozen clinics, the differences are even more substantial, exceeding 200 percent.

There is also wide variability in how often patients at different clinics are hospitalized for septicemia. Although septicemia cases can be unrelated to dialysis, it is a significant risk for patients, who typically have their blood cleaned of toxins three times a week. Nationally, the rate was about 12 percent a year for 2006 to 2008. But in dozens of counties, the spread between facilities with the highest and lowest rates was more than 25 percentage points.

The federal government, which pays for most dialysis under a special Medicare entitlement created in 1972, has collected data on clinic outcomes and practices patterns for decades.

The Centers for Medicare and Medicaid Services contracts with the University of Michigan's Kidney Epidemiology and Cost Center to carve this information into customized reports for each facility in the country, comparing its results to expectations based on case mix and national averages.

For years, the government has shared the reports with state health agencies to guide inspections and with dialysis facilities.

But very little information has reached patients.

CMS' Dialysis Facility Compare website characterizes patient survival rates broadly as "As Expected," "Better than Expected" or "Worse than Expected." It shows how clinics do in managing patients' anemia and in delivering treatments that remove adequate amounts of waste.

But the site offers no hard numbers on mortality and no information on other measures featured in the complete dialysis facility reports.

Why have these measures been withheld? Dr. Barry Straube, CMS' chief medical officer, gave several reasons. There was concern that some data was incomplete, or that it had been submitted inconsistently by providers, he said in an October interview.

Some measures are adjusted for differences in patient characteristics, such as age or complicating medical conditions, to make comparisons more reliable. But these adjustments have come under fire from some providers for lacking sufficient refinement and for not capturing factors -- such as patients' compliance with treatment plans or socioeconomic or educational differences -- that can influence outcomes.

That can lead to "a very unfair picture of quality in a particular unit," Straube said.

The two chains that dominate the U.S. market for dialysis services, DaVita Inc. and Fresenius Medical Care North America, said they support more transparency in general but cautioned that there are caveats to the material previously undisclosed by CMS.

"The currently available data could be expanded and improved, and the release of Dialysis Facility Reports is only helpful if providers, both large and small, hold themselves to the same consistent, rigorous standards in reporting," a spokesman for DaVita said in a written statement.

Fresenius said it was critical that yardsticks used to measure performance were proven fair: "The testing of measures is a science that is important to ensure that dialysis facilities with valid distinctions in their demographics, makeup and operations are fairly accounted for in the quality measures."

One provider has published its data: Northwest Kidney Centers, a nonprofit that operates 14 dialysis centers in the Seattle area. Joyce F. Jackson, the chief executive, said the organization's quality committee voted to put the reports online last year.

"Having this information validates for us how we're doing—we have a baseline," she said. The board members concluded patients could benefit from it, too.

Consumer advocates say that whatever the data's shortcomings may be, its release is a critical step toward patient empowerment. With more information, patients can compare services independently and learn what questions to ask, said Lisa McGiffert, campaign director for Consumers Union's Safe Patient Project.

Providers also respond differently when they know their results will be out there for all to see.

"Transparency is the way to make improvements happen," McGiffert said.

No one would suggest using outcome data in isolation to assess a dialysis center.

Wager said she urges prospective patients to interview current ones and to tour facilities. She also suggests gathering information about staff members' experience level and turnover rates.

Still, she said she wishes she'd had the data available today when she went on dialysis herself in the early 1980s, before receiving the first of two transplants.

"I had nothing, no frame of reference," she recalled. "It can be overwhelming for the typical patient to understand the numbers, but it should be the patient's right and the patient's decision to have them."