"If you ask anyone, "When was the last time you had a mammogram, or a cholesterol test?' almost all people will think it's been a shorter time than it's actually been. Over-reporting happens to all of us," said lead author Kevin Fiscella, M.D., M.P.H. "But our study found that some minorities tend to over-report even more often than other groups of patients. This is a problem that has to be recognized and fixed."
Fiscella, an associate professor of Family Medicine and Community and Preventive Medicine at the University of Rochester Medical Center, said he is not sure why the discrepancy in reporting exists. One theory supported by social psychology research is that African American or Latino patients may feel some pressure to defy stereotypes. So, when doctors or nurses ask them about an event for which they are uncertain, they err on the side of caution and reply affirmatively, he said.
The study is to be published Sept. 29, 2006, in the open access journal BMC Health Services Research.
The University of Rochester collaborated with colleagues from the University of California in Sacramento to analyze 49,645 patient records, matching Medicare claims with patient surveys of the last preventive screening test they had received. The researchers made statistical adjustments for age, gender, income, education level and health status.
With the exception of PSA testing for prostate cancer, results showed that minority patients were more likely to report taking part in a test in the absence of a submitted Medicare claim. In addition to over-reporting errors, the results also might suggest incorrect or inconsistent Medicare billing practices by providers who serve minorities, the study said.
As a frequent advisor to several national groups concerned with health care disparities, Fiscella said the research has several implications:
* First, the study shows that ethnic disparities still exist in health care, despite a widespread belief that the gap is closing.
* Second, it shows that self-reported data based on patient memory contains too much human error and the federal government should not rely exclusively on it when deciding policy issues, such as the allocation of resources to eliminate health care disparities. Insurance claims may represent a less biased means for national tracking of ethnic disparities.
* And perhaps most importantly, the study helps to shed light on vexing questions such as why black women are dying from breast cancer in higher numbers than white women, despite national data suggesting that black women are getting screening mammograms, and therefore have the same access to earlier and better detection as white women.
The authors recommend further studies to clarify the causes of discrepancies in disparity estimates.