Keeping Score on Doctors: Report Says Flaws in Counting Must be Addressed to Ensure Accuracy
DURHAM, N.C. – Agencies that rank doctors and hospitals need
to make sure they are comparing apples to apples, or rankings
can become skewed and unfairly penalize high-quality medical
professionals, according to cardiologists James Jollis of Duke
University Medical Center and Patrick Romano of the University
of California, Davis.
In a critical analysis published in the April 2 New England
Journal of Medicine, the physicians warn that the methods of
data collection are not accurate enough to make results
trustworthy. But, they add, if simple corrections are made,
such scorecards can be a good way to ensure quality care.
"We are questioning whether these types of scorecards are
really accurate enough to make available to consumers," Jollis
said in an interview. "Hospital 'scorecards' are here to stay,
so as physicians we have the responsibility to be sure the
methods used to generate rankings are as accurate as
The researchers analyzed the Pennsylvania Health Care Cost
Containment Council's 1996 report on how well doctors and
hospitals in the state fared in taking care of heart attack
patients. They found while the method used was, in general,
sound, it had several flaws that could have skewed results.
They chose Pennsylvania because it was the first state to
implement a government-sponsored statewide ranking.
"The Pennsylvania report is a good intermediate step, but we
need better systems for reporting outcomes before these type of
rankings will be realistic," Jollis said.
The researchers say before such ratings systems are adopted
and results made public, they should be subject to peer review
to be sure the most accurate information is reported to the
In their analysis, Jollis and Romano argue that using
information from hospital bills and insurance claims is not an
accurate way to gather information about patient outcomes. They
say such data, while more standardized than handwritten
physician charts, often are not a clear reflection of quality
of care or outcomes.
Jollis and Romano argue that it is particularly crucial for
agencies rating doctors to account for differences among
"Medicine, particularly with heart disease patients, is a
highly interactive enterprise, with many physicians involved in
a single case," Jollis said. "A heart attack patient may be
treated by an emergency room physician, and transferred to a
cardiologist's care if the case becomes particularly
For example, the researchers say 14 percent of patients
assigned by Medicare as having been cared for by cardiologists
were actually admitted by internists or family physicians. The
ideal monitoring system, they say, should focus on the doctors
who have the greatest influence on outcome, which for heart
attack is the first physician involved, since early
intervention is crucial in heart attack treatment. The
Pennsylvania study didn't report the responsible physician that
In addition, the Pennsylvania study had other important
shortcomings, they say. Each hospital stay was counted
separately, so if a patient was transferred from one hospital
to another for further treatment and died while in the second
hospital, that patient was counted as a survivor at the first
hospital but as a death at the second, and the methods used to
define disease severity didn't include preexisting conditions
that could have influenced how well patients did.
"Report cards are widely used by government, insurance
companies, HMOs, and employers," Jollis said. "However, the
quality of data is not yet up to the task of confidently
identifying the best practices."
Jollis and Romano argue that future scorecards should:
Broaden the definition of acute myocardial infarction, or
Use a consistent definition of responsible physician;
Redefine complications so preexisting conditions are
Combine information from all hospital stays for each
The researchers argue that doctors and hospitals need to
recognize that the era of "scorecards" for medical
professionals is here to stay, and if they want to make sure
they are rated accurately, they should develop standardized
ways of reporting patient data.
"As physicians, we already devote a large amount of time to
documenting patient care, but in a narrative form that is not
readily comparable," Jollis said. "If we redirect our efforts
toward recording key data in a universally comparable format,
we could avoid the types of inaccuracies that come from trying
to extract outcome data from hospital claims."