A version of this story was co-produced with NPR and aired on All Things Considered. (Check here for local listings.)

In 2007, with roadside bombs exploding across Iraq, Congress moved to improve care for soldiers who had suffered one of the war's signature wounds, traumatic brain injury.

Lawmakers passed a measure requiring the military to test soldiers' brain function before they deployed and again when they returned. The test was supposed to ensure that soldiers received proper treatment.

Instead, an investigation by ProPublica and NPR has found, the testing program has failed to deliver on its promise, offering soldiers the appearance of help, but not the reality.

Racing to satisfy Congress' mandate, the military chose a test that wasn't actually proven to detect TBI: the Automated Neuropsychological Assessment Metric, or ANAM.

Four years later, more than a million troops have taken the test at a cost of more than $42 million to taxpayers, yet the military still has no reliable way to catch brain injuries. When such injuries are left undetected, it can delay healing and put soldiers at risk for further mental damage.

Based on corporate and government records, confidential documents, scores of interviews and emails obtained under the Freedom of Information Act, our investigation found:

• The people who invented ANAM and stood to make money from it were involved in the military's decision to use it, prompting questions about the impartiality of the selection process. No other tests received serious consideration. A report by the Army's top neuropsychologist circulated last year to key members of Congress labeled the selection process "nepotistic."
• The Pentagon's civilian leadership has ignored years of warnings, public and private, that there was insufficient scientific evidence the ANAM can screen for or diagnose traumatic brain injury. The military's highest-ranking medical official said the test was "fraught with problems." Another high-ranking officer said it could yield misleading results.
• Compounding flaws in the ANAM's design, the military has not administered the test as recommended and has rarely used its results. The Army has so little confidence in the test that its top medical officer issued an explicit order that soldiers whose scores indicated cognitive problems should not be sent for further medical evaluation.
• Top Pentagon officials have misrepresented the cost of the test, indicating that because the Army invented the ANAM, the military could use it for free. In fact, because the military licensed its invention to outside contractors, it has paid millions of dollars to use its own technology.
• The military has not conducted a long-promised head-to-head study to make sure the ANAM is the best available test, delaying it for years. Instead, a series of committees have given lukewarm approval to continue using the ANAM, largely to avoid losing the data gathered so far.

Several current and former military medical officials criticized the Defense Department's embrace of a scientifically unproven tool to use on hundreds of thousands of soldiers with TBIs.

"The test was not developed for the purposes of identifying the kinds of problems that we see in concussions," said Dr. Stephen Xenakis, a retired brigadier general and former adviser on mental health issues to the chairman of the joint chiefs of staff. The test was picked "without asking ourselves the questions: what are we trying to achieve here and what are we going to use the screenings for?"

Army Surgeon General Eric Schoomaker acknowledged there have been problems with the testing program and called it a "first step."

"The Army recognized all along that it was not an optimal test," Schoomaker said in a written statement to ProPublica and NPR. He added that the Army has tried to improve the ANAM test and is comparing it to alternatives.

Many experts in the field say Congress' mandate for testing went beyond what science can provide. There is no computerized test that, on its own, allows doctors to diagnose TBI.

Yet studies have indicated the ANAM shows promise when used immediately after a blast, helping doctors determine if soldiers are sharp enough to return to duty. Other studies, as yet unpublished, show the test may also be able to detect certain cognitive problems months after a brain injury.

Those most familiar with the ANAM program insist that testing can be of significant value if used properly -- and that this is where the military has gone wrong.

"We have failed soldiers," says retired Col. Mary Lopez, who used to manage the Army's testing program. "It is incredibly frustrating because I can see first-hand the soldiers that we've missed, the soldiers that have not been treated, not been identified, [or] misdiagnosed. And then they struggle."

A Test to Identify Invisible Wounds

On a crisp morning in early September, about 40 uniformed soldiers mill around outside a squat rectangular building on the Fort Lewis-McChord military base near Tacoma, Wash.

They are scheduled to ship out to Afghanistan in a matter of weeks and must take the ANAM before they go. Filing into the testing facility, they take seats in front of computers.

ANAM proctor Felix Rios, a former first sergeant who served in Iraq and Afghanistan, does a quick Power Point presentation on the test. There will be 20 minutes of questions covering basic math, memory and reaction time.

Taking the ANAM is the first step in protecting soldiers from the effects of brain injuries, Rios tells the group.

"One of the best ways to tell if something's affecting you is to know how you were before it happened. That's what you do here with ANAM," Rios says.

The soldiers leave feeling comforted.

"I felt reassured," Lt. Benjamin Lewis Westman said after completing the test. It was good, "knowing that the Army is taking steps to ensure people are taken care of."

The military's foray into cognitive testing reflects the types of wounds troops have sustained fighting in Iraq and Afghanistan.

Soldiers have suffered an epidemic of concussions, also known as mild traumatic brain injuries, in bombings by insurgents. While most troops recovered quickly, some developed long-term cognitive problems. They couldn't think, read, write or remember the way they had.

Too often, their injuries were missed on the battlefield and, even after they returned home, eluded more sophisticated scanning technology: Studies showed that up to 40 percent of troops who sustained concussions went undiagnosed.

After a 2007 Washington Post series exposed the grim condition of soldiers with brain injuries at Walter Reed Army Medical Center, Congress pushed to create a program to screen soldiers for such wounds.

It's not surprising that the military, under pressure to act quickly, looked to the ANAM. There's no scientific consensus supporting one computerized neurocognitive test among the half-dozen or so available.

ANAM had been developed in-house. Starting in the early 1980s, scientists at Fort Detrick, the military's primary research base, conceived the test as a way to measure pilots' reflexes or to see if a paratrooper could think clearly after a rough landing.

The fighting in Iraq and Afghanistan spurred researchers to re-envision the test: Could ANAM help doctors assess cognitive damage from concussions? In the summer of 2007, as Congress pressed the military for a testing solution, the ANAM was administered to thousands of soldiers from the 101st Airborne at Fort Campbell in Kentucky as part of a pilot program.

"We had developed it ...and we owned it," said Lopez, who helped launch the pilot project, of the ANAM. "We pushed it and we ran as fast as possible because we knew we had a huge problem with TBI."

The Inside Track

Other factors also tilted the selection process in ANAM's favor, sparking a lingering debate over whether the interests of the test's developers trumped those of soldiers.

In February 2007, researchers at Fort Detrick paid the publisher of the Archives of Clinical Neuropsychology to release a supplemental issue on ANAM, including several articles that endorsed its use in detecting traumatic brain injury. Some of the authors had a financial stake in the test, owning patents or trademarks on it, and others received military contracts and funding to help develop the ANAM, but the details of their interests weren't disclosed.

Dr. Robert Kane, the ANAM researcher hired by the military lab to edit the journal issue, said that the journal didn't require such financial disclosures, but "in retrospect, patent information could have been provided."

In October 2007, when the military convened a panel of experts to weigh the ANAM's pros and cons, the journal issue was among its primary sources of information, according to internal emails between panel members and senior military medical officials. In addition, several ANAM developers testified at the panel's sole meeting and one even sat on the committee.

"It just screams conflict of interest," said Professor Marcia Angell, a former editor of the New England Journal of Medicine who is now a senior lecturer on social medicine at Harvard Medical School. "I mean here you have a situation where the same people are sellers, buyers and evaluators."

The ANAM patent and trademark owners interviewed for this article, including Col. Karl Friedl who helped develop the ANAM and was listed as a member of the scientific advisory panel, described their involvement in the deliberations as peripheral and said their financial interest in the test was small and had no bearing on their remarks.

"Any inference that the government inventors would make money from DoD's use of the product and would recommend it for financial reasons are completely erroneous," said Kathryn Winter, one of the ANAM researchers who attended the meeting. Winter said she had made less than $5,000 in royalties from ANAM sales so far, though she could receive more in the future.

Panel members were hardly blind to the ANAM's flaws. In their report, they noted there was no scientific proof that the test could work in the field and acknowledged that it hadn't been subjected to rigorous peer review.

The committee recommended ANAM anyway.

"Nobody was enamored of it, but there was a decision made to use it," said former Navy Capt. Morgan Sammons, a psychologist who served as a co-chairman of the panel.

In May 2008, S. Ward Casscells, then the senior civilian Pentagon official in charge of health affairs, issued an order to use the ANAM across the military.

Competitors who pitched other testing software maintain the playing field was slanted and that soldiers have been ill-served.

"It remains unfathomable why the procurement process at the top was maneuvered so that soldiers were denied useful and usable screening tools for the signature injuries of these wars," said Don Comrie, chief executive of PanMedix, Inc., which markets a competitive test. "As a result we still don't know who was exposed to a blast and what, if anything, is wrong with them."

Casscells and Schoomaker say the selection process was comprehensive and fair.

But some researchers, including retired Lt. Col. Michael Russell, the head of the Army's testing program, say the National Hockey League managed to pick a better test than the Pentagon did.

The NHL evaluated five tests, looking for the option that was most accurate, best supported by research, and easiest to use for trainers and players, many of whom were not English speakers.

It chose the ImPact, a 20-minute computerized test that requires the athlete to remember and then reconstruct a series of designs and patterns. By the beginning of 2007, every player had taken a baseline test.

Several NFL teams also use the ImPact, as does Army Special Forces, which said it based its decision on a review of scientific literature.

Russell, the military's leading civilian neuropsychologist, called the ANAM a poor choice.

"If they had said, would you like to use something else, I probably would have said, yes, I'd like to use something else," he said.

Internal Skeptics Hobble Testing Effort

During and after the selection process, the testing program was met with deep skepticism from top military medical officers, internal correspondence obtained by ProPublica and NPR shows.

In a series of emails, Col. Charles Hoge, a leading Army psychiatrist, warned senior members of the medical command that the evidence supporting the ANAM, or any tool like it, was flimsy and that using it to screen soldiers for brain injuries could lead to misdiagnoses.

"Rolling out ANY diagnostic or clinical test on a population level ... without objective and reliable criteria for how the data will be interpreted and used is malpractice," Hoge wrote in November 2007 to several senior medical officials.

Lt. Col. Mike Jaffee -- a neurologist then in charge of the military's premier brain injury center and a member of the panel that recommended the ANAM -- responded that the military didn't have a better option. "There are currently no instruments that have been validated for blast or combat TBI," he wrote.

The decision to use the ANAM was based on politics, not science, Jaffee acknowledged to Hoge.

"The bottom line is because of the political situation we have been told that studying the situation before acting is not an acceptable option" to the Defense Department, he wrote.

"Congress can mandate till the cows come home," Hoge shot back. "The right thing to do is to go back and help Congress to understand what is feasible and achievable."

When Casscells issued the order to administer the ANAM to all troops before deployment, the Army's top medical officer pushed back.

In a confidential email obtained by ProPublica and NPR, then Army Surgeon General Eric Schoomaker told Casscells that more than 30 military and civilian neuropsychologists, researchers and other experts had reviewed the ANAM at an Army symposium in June 2008.

"All were very cautious about the application of this technology -- it is not FDA-approved for screening for a mild TBI/concussion and has not been evaluated for sensitivity/specificity in this setting," Schoomaker wrote. "Its use in this regard is fraught with problems."

Three minutes later, Casscells dashed off a two-sentence reply -- "Thanks for these wise caveats. Your scholarly standards!" -- but his order stood. In a recent interview, Casscells said he and Schoomaker came to an understanding on the ANAM "that half a loaf was better than none."

Schoomaker subsequently issued an order mandating the use of the ANAM across the Army, prompting another round of internal protests. In a November 2008 email, one senior adviser sent Schoomaker a Power Point presentation saying that ANAM test results were "misleading" and that using them could jeopardize the credibility of the military medical establishment.

Faced with the backlash, Schoomaker modified his order, issuing a follow-up that limited the test's use in critical ways.

He decided that soldiers would not have to take the test upon their return to the U.S. from the battlefield, though Congress required post-deployment testing by law.

Schoomaker also ordered that soldiers who scored badly -- or "red" in testing lingo -- on their pre-deployment tests would not be referred for follow-up evaluations to see if they had an undiagnosed brain injury. In a written statement to ProPublica and NPR, Schoomaker said soldiers with symptoms were urged to see doctors, but not based on ANAM scores. "For some people 'red' is their normal score," he said of the test.

Several neurologists, in and out of the military, said they considered Schoomaker's order to ignore low baseline scores to be unethical. The Navy and Marines, unlike the Army, refer troops for further attention if they score poorly on their pre-deployment ANAM.

"It's our obligation as medical providers to our patients to try to figure out if there is something going on that needs to be treated," said Navy Commander Jack Tsao, a neurologist who runs the ANAM program for those branches of the service.

Lopez said that by ignoring poor baseline scores, the Army risked sending soldiers with cognitive problems into war.

"This is horrible and it goes against our medical ethics and moral responsibilities," she said. "They just lock up these results in a box and never look at them before clearing soldiers to deploy."

Flaws in Implementation

Despite the many concerns with the ANAM, many experts say the test could have helped detect brain injuries in soldiers if it had been used properly.

Instead, the military has implemented the test in ways that have undercut its value, according to interviews and internal emails obtained by ProPublica and NPR.

One example: Troops take the ANAM just once before deploying, even though some of the test's developers have found that users should take it several times to produce a more accurate baseline score.

Without a reliable baseline, Russell and other specialists said, it's impossible to measure the change in a soldier's cognitive abilities after a blow to the head. In a 2010 report, one military researcher called the testing effort "fundamentally flawed" because of the lack of accurate baselines.

More than 1 million soldiers have taken pre-deployment ANAM tests, but medical officials have requested test scores for comparative purposes just 11,000 times since 2008.

Doctors face substantial obstacles in accessing the information.

There's still no computer network that collects and stores test results or integrates them with the military's overall medical system. A doctor in Afghanistan who wants an injured soldier's pre-deployment score has to call into a hotline where an employee creates a PDF with the test results, manually deletes private information such as Social Security numbers, and emails or faxes back the file.

Fewer than 3,000 requests for ANAM test results have been made from the war zones in the last three years, according to recent estimates. Yet more than 90,000 troops suffered traumatic brain injuries during that time, according to the military's official figures.

Reports from the field indicate that the test is not being used consistently when soldiers sustain possible concussions.

This summer, Robert Parish, a neuropsychologist treating troops in Afghanistan, sent an email to colleagues saying the ANAM was more useful than some thought but remained little-used.

"I have corresponded quite a bit with the ANAM staff recently and discovered, quite to my astonishment, I am currently the only one in any combat theater who is actively using the [ANAM] database in evaluations," Parish wrote.

"Free" Test Costs a Bundle

When the military chose the ANAM, one of its selling points -- at least, in the eyes of some involved in the decision -- was that it was free.

"The ANAM, as I understand it, was owned by the Department of Defense. It was developed in conjunction with the Army. So there was no cost to it," said Dr. Paul Hammer, director of the Defense Veterans Brain Injury Center.

Lt. Col. Jaffee, who sat on the panel that recommended ANAM, has said in emails and slideshow presentations that ANAM was picked in large part because it was "the only tool available free of charge to DoD."

But even though the ANAM was developed by military researchers, the Defense Department no longer owned the test by the time it went looking for a screening tool.

This was not unusual: To spur innovation, the agency allows government scientists to license their inventions to outside businesses as long as the military shares in the profits.

In February 2006, the lab that employed the ANAM researchers licensed the ANAM to the University of Oklahoma, which would help refine the test at its cognitive research lab. The university then struck a deal with Vista Partners, a contractor based in Colorado, to market the test and handle sales.

So far, the university and Vista have sold the ANAM to a handful of civilian hospitals and researchers, but their biggest customer has been the military.

The Defense Department pays Vista about $2 million a year in user fees for the ANAM, according to contracting documents and interviews. Oklahoma University gets a 12 percent cut of sales, sending a portion back to the military lab that invented the test. The Defense Department also pays the university $1 million a year for making improvements to the test, Russell said.

On top of that, the military has paid Eyak Services, an Alaska-based contractor, $30 million over the last four years to give soldiers the ANAM test. An additional $6.8 million has gone to another contractor, Evolvent, to build an electronic system to collect and distribute test results, though this effort remains incomplete.

The program's total price tag tops $42 million, records show.

Rep. Bill Pascrell, D-N.J., who helped author the bill to create the program, is disappointed with the way the money has been spent.

"This is not adequate," said Pascrell in an interview with ProPublica and NPR. "You're doing harm to these veterans and these wonderful warriors and their families, and we're not going to put up with it. This is not what we paid for."

As Problems Emerge, Little Action

Military officials openly acknowledged problems with the TBI testing program at two congressional hearings last year.

In April, Dr. Charles Rice, the president of the Uniformed Services University of the Health Sciences, testified to a House Armed Services subcommittee that the ANAM might not accurately reflect soldiers' cognitive abilities. The next day, Schoomaker was even more blunt, telling lawmakers the test was no more accurate than a "coin flip."

Following those hearings, Pascrell and two other congressmen requested a comprehensive report on the program.

In September, Russell delivered a scathing 537-page report that slammed nearly every aspect of the ANAM program.

The report -- which was obtained by ProPublica and NPR, but has never been released publicly -- begins by saying the "program's history is more troubled than commonly understood."

Russell went on to lambaste the manner in which the test was chosen: "The selection of ANAM was nepotistic, and the long delay in examining alternative instruments is baffling."

In the field, Russell said, the test had failed at "a basic level." It was unreliable and soldiers' scores often reflected factors such as fatigue that were unrelated to TBI.

"This is unacceptable if the ANAM is to be considered a TBI test," Russell wrote.

There was one bright spot amid Russell's criticism: Though the ANAM could not diagnose TBI, he said, it could help doctors evaluate when soldiers who had suffered concussions were mentally fit to return to duty.

Overall, however, Russell concluded that the problems with ANAM were so severe that "it appears that we may be doing our soldiers a disservice with the present baseline program."

Pascrell said he was infuriated by Russell's findings.

"There is no question in my mind that the Department of Defense violated the very essence of the law that we passed" to require testing, he said. In particular, he said, he was frustrated by the lack of post-deployment testing and the delay in comparing ANAM to other tests.

There are few signs that the changes Pascrell wants are imminent.

When the military initiated the testing program in 2008, it also created a panel of civilian experts to monitor it and other TBI-related issues.

At a meeting in August 2011, the panel discussed whether ANAM should be replaced, expanded, or used routinely post-deployment. Then it endorsed the status quo, partly to preserve the usefulness of data from tests already administered.

The committee's recommendations could be "expressed in one sentence, which is 'Continue to do what we're doing with the ANAM, but don't do more at this point,'" said Dr. Kurt Kroenke, the Indiana University professor of medicine who led the discussion.

When S. Ward Casscells, the former head of Pentagon official health affairs, ordered the ANAM into use in May 2008, he depicted it as an interim measure and promised the military would launch a comprehensive study comparing it with other brain-injury tests.

Emails obtained by ProPublica and NPR show the comparison was supposed to be completed by November 2009, but it remains unfinished.

A recent report from the Government Accountability Office said the Defense Department doesn't expect the results until 2015 -- two years after the last troops are scheduled to leave Afghanistan.