The number of disciplinary actions taken by the nation's hospitals against physicians has declined during the past decade, raising questions about the adequacy of information in the national computerized data bank that hospitals and health plans rely on for information about doctors' records, researchers reported last week in the Journal of the American Medical Association (JAMA).

Results of the study, conducted by researchers from the University of Washington and the federal Bureau of Health Professions, underscore growing concern among advocates for patients and federal health officials about the utility of the National Practitioner Data Bank.

Launched in 1990, the data bank is designed to create a permanent paper trail of malpractice payments and disciplinary actions against physicians, thereby preventing bad doctors from moving from hospital to hospital or state to state undetected and with impunity.

The study is the latest indication that the data bank may have unintended consequences as some doctors seek to evade scrutiny. A 1995 report by the inspector general of the Department of Health and Human Services found that underreporting to the data bank appears to be a serious problem.

Since the data bank began operating in 1991, "physicians have become totally intransigent about settling a case in a reasonable way," said Roger A. Rosenblatt, a physician at the University of Washington and coauthor of the JAMA study. "Its existence has transformed physician behavior. The contortions people go through to stay out of the data bank are truly stunning."

Federal law requires that hospitals and health plans query the data bank before granting a doctor privileges and every two years after that. The data bank is closed to consumers at the insistence of the American Medical Association. The AMA, the nation's largest and most powerful physicians' group, has long argued that patients would misinterpret malpractice information contained in the data bank because even good doctors get sued. Currently the data bank contains files on more than 130,000 physicians, about 20 percent of the nation's doctors.

In addition to malpractice payments, the data bank contains reports of disciplinary actions taken by hospitals, professional societies and state medical boards. Hospitals are required to notify the data bank if a doctor's privileges are revoked, surrendered or suspended for more than 30 days. Suspensions for fewer than 30 days are not reportable.

As a result, hospital officials and patient advocates say, a growing number of hospitals are imposing 29-day suspensions and other penalties that don't have to be reported.

To gauge the impact the data bank has had on hospital disciplinary actions, which are chiefly restrictions of privileges, Rosenblatt, physician Laura-Mae Baldwin of the University of Washington and their colleagues examined all reports to the data bank of actions taken to restrict privileges between 1991 and 1995.

They found that about 75 percent of 4,743 hospitals reported taking no actions against any doctor over a three-year period. Baldwin's team also reported that the percentage of hospitals taking at least one action affecting privileges decreased from 11.6 percent in 1991 to 10 percent in 1995.

That decline has continued, according to officials at the Health Resources and Services Administration (HRSA), which oversees the data bank. HRSA officials said that in 1997, 950 suspensions of doctors' privileges were reported to the data bank, a number that dropped to 800 in 1998.

Those numbers are far below initial projections. Based on estimates by the American Hospital Association (AHA), data bank officials expected to receive about 14,000 reports of suspensions or other hospital disciplinary actions annually.

Mary Grealy, chief Washington counsel to the AHA, said that she does not regard the decline documented in the study in JAMA as "dramatic" and that the source of the initial projections was not clear.

"I think we have to remember that hospitals' taking adverse action [against a doctor who has privileges] is very, very serious," she said. "We look for a variety of ways to work with physicians to improve the quality of care."

Querying the data bank is not the only way hospitals check a doctor's record, according to Grealy. "There are a lot of other checks and balances in place."

Baldwin said her team discovered significant variation among hospitals. Rural hospitals were more likely than urban hospitals to have no discipinary actions to report. So were hospitals in the South and the 269 institutions--all of them academic medical centers--that comprise the Council of Teaching Hospitals for the Association of American Medical Colleges.

Hospitals in the three states--California, New Jersey and West Virginia--that impose fines of more than $5,000 for failing to report actions on privileges to state medical boards had significantly higher rates of reporting to the data bank than states without such laws, the researchers found.

Although this study did not address the practice, there are other ways a doctor's missteps may not be recorded in the data bank.

The most common is known as "the corporate shield." Under the 1986 law that created the data bank, payments made by doctors to settle malpractice cases must be reported. But payments made by health plans, professional corporations, group practices or hospitals do not need to be reported.

Plaintiffs' lawyers say that as a condition of settlement, a doctor often insists on being dismissed as a defendant in a lawsuit, thereby avoiding a data bank report. Hospitals, which are often named as co-defendants, typically accede to such requests in an effort to minimize the unwanted publicity that results from having a claim go to trial.

Hospitals have to balance the desires of two constituencies, Rosenblatt noted. "They have to protect patients as well as as make sure they have the allegiance and loyalty and dedication of their [medical] staff," he said.

"It's a common subterfuge," Washington malpractice specialist Jack H. Olender said of the corporate shield.

Olender cites a recent case brought by a woman who sought treatment from a physician at a Washington area hospital after she stepped on a nail. The woman was not given a tetanus shot and developed a near-fatal infection that required amputation of her foot. She subsequently sued the doctor and the hospital for malpractice.

Olender said the case was settled before trial for nearly $1 million after his client agreed to drop the doctor as a defendant. Because the payment was made on behalf of the hospital, the doctor escaped a data bank report.

HHS has proposed regulations that would close the "corporate shield" by requiring that the name of the doctor whose action was the basis of a paid claim be reported to the data bank.