Six years ago, federal health officials were confident they could save taxpayers hundreds of millions of dollars annually by auditing private Medicare Advantage insurance plans that allegedly overcharged the government for medical services.

An initial round of audits found that Medicare had potentially overpaid five of the health plans $128 million in 2007 alone, according to confidential government documents released recently in response to a public records request and lawsuit.

But officials never recovered most of that money. Under intense pressure from the health insurance industry, the Centers for Medicare and Medicaid Services quietly backed off their repayment demands and settled the audits in 2012 for just under $3.4 million — shortchanging taxpayers by up to $125 million in possible overcharges just for 2007.

Medicare Advantage is a popular alternative to traditional Medicare. The privately run health plans have enrolled more than 17 million elderly and disabled people — about a third of those eligible for Medicare — at a cost to taxpayers of more than $150 billion a year. And while the plans generally enjoy strong support in Congress, there are critics.

"It's unclear why the Obama Administration allowed CMS to overpromise and under-deliver so badly on collecting these overpayments," Sen. Chuck Grassley, R-Iowa, told Kaiser Health News in an email response to the findings.

He said CMS "should account for why this process seems to be so broken and why it can't seem to fix it, despite recommendations to do so. The taxpayers depend on getting this process right."

The failure to collect also alarmed Steve Ellis, vice president of the budget watchdog group Taxpayers for Common Sense in Washington.

"They need to put up a bigger and stronger fight to make sure these programs are operated on the straight and narrow," Ellis said.

Yet outside of public view, federal officials have been losing a high-stakes battle to curb widespread billing errors by Medicare Advantage plans, according to the records obtained through a Freedom of Information Act lawsuit filed by the Center for Public Integrity.

The Center for Public Integrity first disclosed in 2014 that billions of tax dollars are wasted annually partly because some health plans appear to exaggerate how sick their patients are, a practice known in health care circles as "upcoding."

Last August, the investigative journalism group reported that 35 of 37 health plans CMS has audited overcharged Medicare, often by overstating the severity of medical conditions such as diabetes and depression.

The newly released CMS records identify the companies chosen for the initial 2007 audits as a Florida Humana plan, a Washington state subsidiary of United Healthcare called PacifiCare, an Aetna plan in New Jersey and an Independence Blue Cross plan in the Philadelphia area.

The fifth one focused on a Lovelace Medicare plan in New Mexico, which has since been acquired by Blue Cross.

Each of the five audits, which took more than two years to complete, unearthed significant — and costly — billing mistakes, though the plans disputed them.

For example, auditors couldn't confirm that one-third of the diseases the health plans had been paid to treat actually existed, mostly because patient records lacked "sufficient documentation of a diagnosis."

Overall, Medicare paid the wrong amount for nearly two-thirds of patients whose records were examined; all five plans were far more likely to charge too much than too little. For 1 in 5 patients, the overcharges were $5,000 or more for the year, according to the audits. None of the plans would discuss the findings.

As preliminary results of the audits started to roll in, CMS officials outlined steps to recover more than $128 million from the five plans at a confidential agency briefing in August 2010, according to a policy memo prepared for the meeting. The records don't indicate who attended.

That day, CMS set Humana's payment error at $33.5 million, PacifiCare at $20.2 million, Aetna at $27.6 million, Independence Blue Cross at nearly $34 million and Lovelace at just under $13 million. Those estimates were based on extrapolation from a sample of cases examined at each plan.

CMS "has developed a process for moving forward with payment recovery," according to a briefing paper from the 2010 meeting.

But that process fizzled after two years of haggling with the plans and insurance industry representatives, who argued the audits were flawed and the results unreliable. In August 2012, CMS gave in and notified the plans it would settle for a few cents on the dollar.

"Given this was a new process, the decision was made at the time to tie repayments to the actual claims reviewed as part of the 2007 pilot audit," said CMS spokesman Aaron Albright. "For subsequent audits, we said we intended to determine repayments by extrapolating the error rate of the sample of claims reviewed to all claims under the contract." Albright said more of the audits are underway. Allowing the insurers to dodge liability dealt a serious blow to the government's efforts to crack down on billing abuses — a setback one taxpayer advocate called alarming.

"That's a very bad way to operate the system." said Patrick Burns, acting executive director and president of Taxpayers Against Fraud in Washington, on hearing of the outcome. "Nobody is held accountable."

Indeed, CMS kept the settlement terms under wraps until 2015, after an inquiry by Grassley. The senator had requested details about Medicare Advantage fraud controls in response to articles published by the Center for Public Integrity.

In a July 31, 2015 letter to Grassley, CMS Acting Administrator Andy Slavitt attached a table that showed the five plans repaid just under $3.4 million. The letter didn't mention the earlier estimate that the government was due $128 million. Grassley said it should not have taken the FOIA lawsuit to make that information available to the public.

"Perhaps adding insult to injury, these numbers might never have seen the light of day without a lengthy lawsuit," Grassley said this week.

Paying based on risk scores

When Congress created the current Medicare Advantage program in 2003, it devised a new way to pay the health plans.

The method, phased in starting in 2004, seemed simple enough: pay higher rates for sicker patients and less for people in good health using a formula called a risk score.

But CMS officials soon realized that risk scores rose much faster at some plans than others, a possible sign of upcoding, or other billing irregularities, records show. These overcharges topped $4 billion in 2005, one CMS study found.

The special audits, called Risk Adjustment Data Validation, or RADV, were designed to identify, and hold accountable, health plans that couldn't justify their fees with supporting medical evidence.

Until these audits, CMS "pretty much went on the honor system with the plans," an unnamed agency official wrote in an undated presentation.

In the five 2007 pilot audits, two sets of auditors inspected medical records for a random sample of 201 patients at each plan. If the medical chart didn't properly document that a patient had the illnesses the plan had reported, Medicare wanted a refund. Auditors gave the plans the benefit of the doubt when auditors couldn't agree, according to the CMS briefing paper.

Finally, CMS applied a standard technique used in fraud investigations in which the payment error rate is extrapolated across the entire health plan, which greatly multiplies the amount due. CMS said it was conservative in assessing the penalties and allowed the plans to appeal.

Appeals or no, the health plans recoiled at the prospect they could be on the hook for millions of dollars they hadn't budgeted for and didn't believe they owed. The actual 2007 overage for the 201 Humana patients, for example, was $477,235. Once extrapolated, it soared to $33.5 million.

Michael S. Adelberg, a former CMS official who is now an industry consultant in Washington, said that in retrospect the audit process was "probably rushed."

Adelberg said the audits "raised strong industry concerns" on a variety of fronts, from whether CMS had the legal authority to conduct them to the soundness of their methods. CMS stands by its audit techniques and has defended RADV as the only way it can assure plans bill honestly.

Yet agency records released through the FOIA case suggest CMS lacked the will to press ahead with extrapolated audits for Medicare Advantage plans given the fierce industry backlash — even though they do so in overpayment cases targeting other types of medical providers.

One confidential CMS presentation dated March 30, 2011, notes that officials had received more than 500 comments expressing "significant resistance" to the RADV audits.

The presentation goes on to say the audit program's success depended on its "ability to address the challenges raised."

CMS didn't overcome those challenges. Instead, it agreed to settle the five initial audits for $3.4 million, just what it found in the patient files it reviewed — without the extrapolations. And the center did the same for 32 additional 2007 audits, which officials had predicted would refund up to $800 million to the federal treasury. In the end, CMS wound up with $10.3 million from the 32 plans.

The RADV program's shortcomings, though little known to the public, haven't gone totally unnoticed. The program was the target of a sharply critical May 2016 report by the Government Accountability Office, which noted that Medicare Advantage plans have overbilled the government by billions of dollars, but rarely been forced to repay the money or face other consequences.

The GAO, the watchdog arm of Congress, called for "fundamental improvements." The watchdogs also found that CMS has spent about $117 million on the audits, but recouped just under $14 million.

Government officials didn't dispute that the RADV process had taken far too long and yielded way too little. But while CMS has resumed extrapolated audits, there's little evidence it is speeding things up.

CMS expected to complete extrapolated audits for payments made in 2011 and finish the job in early 2014, agency records show.

But it has yet to do so. In late December, an agency spokesman said he had no new information about when the 2011 audits would be finished or how much the government would collect.

While the industry awaits the results, it has hardly warmed to the process.

America's Health Insurance Plans, an industry trade group, argued in a June 2016 position paper that RADV was "not yet stable and reliable," adding that the audits "could disrupt the care being provided by plans that are working hard to meet the needs of their enrollees."

John Gorman, a former government health care official and current industry consultant, said he expects RADV to forge ahead under the incoming administration. But he predicted efforts to collect overpayments will "slow down" because the Trump team will prove to be "more sympathetic" to business interests than the Obama administration. The Trump transition office did not respond to a request for comment.

Gorman said that while career civil servants at CMS decide which plans get audited, how much to assess the health plans as a result rests with "political appointees" who are susceptible to industry lobbying, which he termed "the old Potomac two- step."

But Grassley said he is determined to keep a close eye on the audit program. "I intend to press the incoming administration on holding CMS accountable for overpayments that harm taxpayers," he said.

Taxpayer advocate Ellis said with so much public money at stake, the government needs to step up its game.

"You can presume that the more people get away with overpayments the more they are going to take," he said. "As the program gets bigger the problem get bigger."

Copyright 2017 Kaiser Health News. To see more, visit Kaiser Health News.