Upwards of 12 percent of U.S. cancer patients are initially misdiagnosed, a new study suggests, leading to repeat testing, treatment delays, increased health-care costs and patient anxiety.

Improper tissue and blood sampling, sometimes coupled with inaccurate reading of hospital lab results, are the twin culprits for cancer diagnosis error, according to a team of Canadian, Chinese and American researchers.

A wide range of cancers -- including both gynecologic and non-gynecologic disease -- are vulnerable to such detection mistakes, while the absence of uniform standards to evaluate error frequency across U.S. hospitals complicates efforts to combat the problem, the study authors said.

The authors point out that health-care centers are also often reluctant to tackle the issue head-on for fear of the adverse legal and public relations consequences related to error disclosure.

Nevertheless, they stress that diagnostic confusion rarely poses a significant threat to the long-term health of a patient.

"I want to make clear that the major consequence is not that patients unnecessarily have organs removed or have a false diagnosis of cancer, but rather that they have cancer and it is not diagnosed," said study author Dr. Stephen S. Raab, a professor of pathology at the University of Pittsburgh School of Medicine. "Fortunately, patient symptoms almost always lead clinicians to do the right thing and retest, and this will lead to finding the cancer and a proper diagnosis."

Raab and his team observe that because U.S. hospitals lack national standards to help guide efforts to monitor error occurrence, it is difficult to compare the frequency, cause and effect of cancer misdiagnoses across facilities.

The researchers nonetheless sought to do exactly that, by focusing on four American medical institutions located in the mid-Atlantic or Midwestern region of the country.

Since 2002, the four centers have participated in an ongoing five-year project to reduce cancer misdiagnoses and improve error detection.

The researchers first standardized the process by which all four institutions reviewed problematic diagnoses -- meaning those instances in which lab work examining two separate specimens taken from a single patient resulted in conflicting diagnoses.

Raab and his colleagues then focused exclusively on those examples of diagnosis discrepancy that had been collected in 2002, analyzing patient tissue samples, blood work and medical records.

In the Oct. 10 online issue of the journal Cancer, the authors report that the institutions varied widely in their diagnosis error tallies -- a fact they attributed to a subjective bias in error evaluations that the newly established standardization process did not completely eliminate.

For example, doctors at one institution very often disputed the findings of another facility in terms of the assignment of error cause or the patient harm an error was determined to have provoked.

"Harm," the authors noted, signified a host of clinical consequences, ranging from the minor inconvenience of needing an additional blood test to the much more serious problem of a six-month delay in proper diagnosis and treatment. Loss of a body part, or even death, was also an extremely rare but theoretically possible harmful consequence of misdiagnosis.

Gynecological diagnosis errors (involving such tests as Pap smears and cervical biopsies) occurred somewhere between just under 2 percent to just over 9 percent of the time, depending on the facility. In 46 percent of such cases, such errors were found to have had no negative treatment consequence, while an almost equal number of cases did provoke some form of harm.

Non-gynecological diagnosis errors (involving such tests as bronchial lung brushing and biopsies) were found in the range of between 5 percent and nearly 12 percent. No harmful consequences were provoked in 55 percent of such cases, while harm of some kind was evident 39 percent of the time, the study found.

In all the institutions, the majority of errors seemed to have occurred during laboratory-setting searches for evidence of abnormal cells, rather than during either surgical tissue extraction or the post-lab pathologist review of test results.

Despite their findings, the researchers emphasized that their efforts to create a uniform assessment system across medical care centers is still in the early stages, making it difficult to draw definitive conclusions about either error frequency or the effect of errors on cancer patient care.

In that light, they suggested that even the highest error frequency percentages revealed among the 2002 patient pool should be viewed as low-range estimates.

Even at the observed rates the potential number of Americans who encounter diagnostic error is large, they noted.

About 150,000 cancer patients who undergo Pap tests annually may be subject to such mistakes, the researchers estimated, while a similar number of non-gynecological patients may also face diagnosis problems each year.

Noting that yearly an estimated 128,000 Americans have to deal with some kind of error-related "harm," Raab and his team called for the medical community to actively support the establishment of error assessment criteria and monitoring guidelines.

"The concept of standards just doesn't exist yet," said Raab. "So, there are no really good ways to evaluate and compare the quality of testing and no standard for the measuring of harm as a result of error. Clearly, there's enormous differences between institutions and we have to figure out why."

"It's a real problem," added Raab, "because people don't want to say they're making an error. But these findings are the result of standardizing the process, and we learned aspects of doing that, and so now we're heading towards bringing the error frequency to a norm, at least across these four institutions."

Dr. Len Lichtenfeld, deputy chief medical officer of the American Cancer Society, expressed enthusiasm for the attempt to get a better handle on the big diagnostic picture, but he said that cancer patients should not meanwhile become unduly concerned about such errors.

"Patients need to know that, yes, there is real interest and concern about the quality of care in medical practice in general," he said. "But for the most part, care is excellent."

"Even though there's a discrepancy which is classified as an error, most of the time no meaningful harm came to that patient," added Lichtenfeld. "It doesn't mean there weren't some errors, but it's not as bad news for the patient as it might appear to be in terms of numbers. So, I would not have people become panicked."