But in two animal studies in San Antonio before the war in Iraq began in 2003, the results of which were shown to The Sun by a retired Army researcher, HemCon was found to be no more effective than gauze, and in one case perhaps less effective.

According to Army documents, after the second study was completed and eight pigs treated with HemCon fared roughly the same as eight pigs treated with gauze, Army scientists who had helped develop the dressing removed results from a pig that was treated with HemCon but quickly died. They determined it was an "outlier," or a fluke. Without it, HemCon came out statistically superior.

Statisticians interviewed by The Sun said that removing subjects from such a small study is questionable and should be disclosed prominently in the research's conclusions. Schoomaker and people involved in the advisory committee's deliberations told The Sun they never knew of the failed tests.

Once HemCon was in the combat zone, the Army published a series of anecdotal reports suggesting that it was effective, but the dressing was not universally embraced. Medics in Iraq told The Sun that the dressings were too small and rigid for battlefield injuries, and a supply officer said troops were discarding them unused. The Navy's top trauma surgeon publicly challenged HemCon's effectiveness. Then, early last year, in a new San Antonio study, just one of 10 bleeding pigs treated with the dressing survived. The Army ordered the dressing removed from the field last summer in favor of a newer dressing.

Schoomaker said HemCon was never ideal but was the best product available at the time. An Army researcher in San Antonio told The Sun, however, that the service's enthusiasm seemed driven more by hope than science. "It just never lived up to the hype, even in our own laboratories," he said.

Finding: A blood-clotting dressing sent to Iraq late last year as a possible replacement for HemCon was distributed even as Army scientists cautioned against using it to treat soldiers. It was quickly recalled by Army leaders when tests on pigs revealed potentially deadly complications.

That dressing, called WoundStat, was touted as "significantly more effective" than other dressings in the same San Antonio study that questioned HemCon's effectiveness and led to its recall.

The authors of the study, however, cautioned that WoundStat posed "a potential safety risk that must be investigated in survival studies before this product can be recommended for use by the military." They feared that because it is a loose clay material rather than a bandage, WoundStat could drift into veins and arteries and cause potentially deadly clots.

But those findings, which were shared with The Sun, did not appear in the scientists' final report and were never published. Schoomaker decided in June to send WoundStat to Iraq - unaware, he says, of the scientists' complete assessment.

After WoundStat was already in Iraq, the same San Antonio scientists tested the product last December on eight pigs with bleeding neck wounds. Within several hours, according to Army officials and other accounts relayed to The Sun, seven of the animals developed a complete blockage of the blood vessels at the site of their wounds. On Dec. 18, the Army ordered WoundStat removed from the field.

The Army is still trying to determine whether the product was used on wounded troops in Iraq or Afghanistan, and whether any complications arose, said Col. Paul R. Cordts, a vascular surgeon and the Army's director of health policy and services.

Finding: Liberal use of a blood-clotting drug, injected copiously into wounded soldiers in 2005 and 2006, was the Army's "standard operating procedure" more than a year before a study into the use of the drug in trauma patients had been completed. The drug has since proven largely ineffective in three unpublished Army studies and potentially dangerous in at least one, and it is no longer routinely used.

Army officials said in interviews two years ago that Recombinant Activated Factor VII, a drug approved by the FDA only for treating hemophilia, arrived in the trauma hospitals in Iraq in early 2004, soon after they got an early look at data from what was then the only clinical trial conducted with trauma patients without hemophilia.

According to Army documents provided by doctors deployed to Baghdad, however, the first guideline from San Antonio on using Factor VII was issued a year earlier, before the war began. It cited only laboratory studies, a handful of anecdotal treatments in Israel and clinical tests on pigs as references.

The guideline recommended giving the drug to almost any casualty with a traumatic injury, even when the extent of bleeding was unclear. The Army didn't begin formal research into Factor VII until more than 2,000 doses had been shipped to hospitals in Iraq, according to documents from the Army's Medical Research and Materiel Command.

In 2006 doctors at the main combat hospital in Baghdad proposed greatly curtailing Factor VII's use, noting safety concerns and their own review of unpublished data that found "no mortality benefit" when the drug was used. They proposed giving it only to patients with "blunt" injuries, and withholding it from the other 90 percent of casualties with "penetrating" injuries.

The doctors say they transmitted their proposal up the medical chain of command at about the same time The Sun reported on Factor VII's possible link to deadly blood clots, but never heard back. Schoomaker and other Army officials said they don't recall seeing it.

Sometime around January 2007, according to documents and electronic messages shown to The Sun, one of the service's top transfusion researchers, using data collected from several hundred patients in Iraq, concluded that Factor VII was "associated with increased mortality" and at best had no measurable effect when variations in injuries and treatments were considered. Two other Army researchers separately reached similar conclusions last year, according to summaries and internal messages shown to The Sun.

None of the research was peer-reviewed or vetted by statisticians, a process that can often lead to different conclusions. And Army officials said the research critical of Factor VII did not look at enough patients to be statistically relevant. Yet the Army's most recent published research related to Factor VII - one study in mid-2007 based on 61 patients, and one in early 2008 based on 124 patients - analyzed smaller groups of patients. Both concluded that the drug is beneficial.

A large clinical trial testing Factor VII in trauma patients was canceled last June by the drug's manufacturer when it became clear halfway through the study that it wouldn't show that the drug was working.