DALLAS - Just minutes after Thomas Eric Duncan arrived for a second time at the emergency room, the word is on his chart: "Ebola." But despite all the warnings that the deadly virus could arrive unannounced at an American hospital, for days after the admission, his caregivers are vulnerable.

The Centers for Disease Control and Prevention has pointed to lapses by the hospital in those initial days. And Duncan's medical records show heightened protective measures as his illness advanced. But either because of a lag in implementing those steps or because they were still insufficient, scores of hospital staffers were put at risk, according to the records.

The hospital's protective protocol was "insufficient," said Dr. Joseph McCormick of the University of Texas School of Public Health, who was part of the CDC team that investigated the first recorded Ebola outbreak in 1976. "The gear was inadequate. The procedures in the room were inadequate."

Duncan's medical records, provided by his family to The Associated Press, show Nina Pham, the first Texas nurse to be diagnosed with Ebola, first encountered the patient after he was moved to intensive care at 4:40 p.m. on Sept. 29, more than 30 hours after he came to the ER. Nearly 27 hours later, Amber Joy Vinson, a second nurse who contracted the disease, first appears in Duncan's charts.

Because doctors and nurses are focused on logging the patient's care, they may not always note their own safeguards in the medical records. In Pham's first entry, she makes no mention of protective gear. When she logs again the following morning, she specifically mentions wearing a double gown, face shield and protective footwear, equipment she mentions again in later entries.

In the first apparent mention of Vinson - identified by just her first name in another nurse's notes - she is said to have worn personal protection, including a hazardous-materials suit and face shield.

It's unclear whether those initial interactions with Duncan represent the time at which a breakdown in protection led to the infections, or whether such lapses persisted during the remainder of the patient's 11-day stay. At least 70 workers are named in the records as being involved with Duncan's care in that period.

Dr. Tom Frieden, the CDC director, told lawmakers during a congressional hearing Thursday that he did not know how the nurses got infected, only that "possible causes" had been identified.

In comments a day earlier, he gave a clue: "For the first several days of the patient's stay, before he was diagnosed, we see a lot of variability in the use of personal protective equipment."

Because Ebola has an incubation period of up to 21 days, those who cared for Duncan at the start of his second hospital stay will not be considered safe from infection until Monday. Those with him at the time of his death will not emerge from monitoring until Oct. 30.

Duncan first arrived at Presbyterian on Sept. 25 but was sent home. When he returned by ambulance at 10:07 a.m. on Sept. 28, he was sicker and probably more contagious. Staff noted immediately upon his intake that he had recently arrived from Liberia.

Five minutes later, a nurse notes that he is in a room and had "put on airborne precautions." Nine minutes after that, Duncan is seen by a doctor who writes that Ebola is a possibility and notes that he "followed strict CDC protocol" by being "masked, fully gowned and gloved" when treating Duncan.

The doctor makes no mention of eye protection such as goggles or a face shield, which are considered basic equipment in Ebola guidelines issued by the CDC. There is also no initial mention of foot coverings, which are suggested when such patients have diarrhea or vomiting, as Duncan did, according to the records.

It is also not clear when Duncan was safely isolated, though the records refer to him being in an "isolation room" in the ER on Sept. 29.

Numerous entries in the records of Duncan's stay at the hospital - both in the ER and later in intensive care - make note of precautionary measures. Many other entries are silent on the issue, and the mention of hazardous-material suits does not appear in Duncan's records until after his diagnosis is confirmed on Sept. 30.

Hospital officials say Duncan was immediately put in isolation in a private room and that staff adhered to CDC guidelines on protective gear, even though those guidelines changed during the course of Duncan's stay.

"The CDC guidelines changed frequently, and those changes were frustrating," hospital spokesman Wendell Watson said.

Dr. Victoria Sutton, a member of Texas Gov. Rick Perry's newly appointed infectious-disease task force, said the issue was not protocols, but preparation.

"I think the problem is there wasn't enough time to do training," she said.

The hospital has denied the allegations of several of its nurses who anonymously aired concerns through a statement issued by the National Nurses United union. Among their complaints: that Duncan was kept for hours in an area of the emergency department where seven other patients could have been exposed; that a nurse supervisor faced resistance from higher-ups when she said he should be moved to an isolation unit; and that even after the patient was isolated, hospital workers came and went from his bedside without proper protection, then walked through halls that were not properly cleaned.

"If any of those allegations - let alone more than one - are correct, if they are valid, then obviously his whole hospitalization put health care workers at risk," said Dr. William Schaffner, an infectious-disease specialist at Vanderbilt University Medical Center.

A half-dozen doctors and nurses make notes in Duncan's chart during the first 24 hours of his stay in the ER. As his first night at the hospital fades into the following morning, his condition worsens. A doctor notes he is suffering and deteriorating. At one point, he asks for a diaper because he's too exhausted to get up.