In the article, Louise Walter, MD, UCSF assistant professor of medicine in the geriatrics division at the SFVAMC, acknowledges the importance of measuring the quality of medical care, but suggests that the current methodology for such measurement needs to be reexamined.
Walter based her comments on results of a study she and three UCSF/VAMC colleagues conducted of the medical records of the 229 patients selected in 2002 by national VA auditors to measure colorectal cancer screening at the San Francisco VA Medical Center. The study was conducted after clinicians at the San Francisco facility had been told that the center's screening rate of 58 percent in 2002 failed to meet the national VA target rate of 65 percent. Walter and her colleagues decided to conduct their own audit of the national VA' s audit of colorectal cancer screening at their facility to determine if there was a quality of care problem.
In their study of the records, the researchers found that many of the patients selected for audit by the national VA were elderly with severe illnesses, making the physical risks and other negative aspects of colorectal cancer screening outweigh the benefits. For example, she said, the dire health of patients with metastatic prostate cancer and end-stage renal failure increases the risks of colorectal cancer screening, which include perforation of the colon and cardiac complications. In her view, the decision by the SFVAMC physicians not to screen such patients represented good quality care, even though the decision was counted as poor quality care by the screening measure. In addition, she wrote, the many patients who refused colorectal cancer screening were counted as receiving poor quality care since they were not screened, when, in her opinion, she says, informed screening decisions that incorporate patient preferences should count as good quality care.
Instead of equating quality of care simply with numbers of patients screened -- a principle applied at the VA nationwide -- Walter advocates that quality be defined by how well screening is targeted to patients likely to benefit, considering prognosis and patient preferences. If measures of quality are not thoughtfully specified, they may become, she writes, "scientifically insupportable algorithms that classify high rates of a procedure as good care regardless of who received it, why it was performed, or whether the patient wanted it."
"We should aim for good clinical decisions 100 percent of the time, not 100 percent screening rates," Walter said. "We need to give doctors the opportunity to examine if the patient needs the test or not and encourage them to talk with patients about these recommendations."
The VA system -- the largest health care system in the United States -- has been considered a U.S. leader in patient care following a system-wide reorganization in the mid-1990s. In 2001, The Institute of Medicine recommended that many practices adopted by the VA, including performance measurement and reporting, be applied to the American medical system as a whole. (The Department of Veterans Affairs manages 158 hospitals, 163 nursing homes, 854 outpatient clinics, rehabilitation programs and scores of other facilities.)
In her essay, Walter praises the VA for its improved patient outcomes and an intellectually free environment, noting that her research was funded by a VA Career Development Award in Health Services Research and Development.
"Quality measurement sounds like it's easy, but it is actually very hard to accomplish," Walter said. "Our current measures of quality need to be critically evaluated to determine if they are measuring important aspects of quality care, such as good clinical decisions and informed decisions with patients."
Co-authors are Natalie P. Davidowitz, BA, data processing coordinator; Paul H. Heineken, MD, UCSF clinical professor of medicine and SFVAMC associate chief of staff of ambulatory care, and Kenneth E. Covinsky, MD, MPH, UCSF associate professor of medicine and SFVAMC staff physician.