N. Tucson Blvd. Suite 9
Tucson, AZ 85716-3450
Phone: (800) 635-1196
Hotline: (800) 419-4777
of American Physicians and Surgeons, Inc.
A Voice for Private Physicians Since 1943
Omnia pro aegroto
Total Quality Management:
Can It Be Applied to Medicine?
by Gerry Smedinghoff, Consulting Actuary
The phrase “managed care” is to American medicine in the 1990s what the term “quality” was to the auto industry during the 1980s. The quality principles
that William Edwards Deming and Joseph Juran taught the Japanese in the 1950s have been transformed into a quasi-religion known as Total Quality Management
(TQM) that is now being applied to virtually every product and service.
In medicine, the general concept of quality along with the specific applications of TQM is often peddled as a panacea that will miraculously bring about
dramatic cost reductions, while improving the general public health so much that the demand for medical care will drop to a minimal and manageable level.
And why not? If it worked in Detroit, why can't it work in a hospital or doctor's office?
First, human beings aren't inanimate machines uniformly built according to engineering specifications. Thus, many basic TQM principles simply
do not apply. Moreover, many TQM principles are being violated when applied to medicine.
The Artisan Concept: Worker Autonomy, with Self-Control and Self-Inspection
Until the quality revolution in the 1980s, American factories were plagued by excessive worker supervision, excessive product inspection, and an absence
of worker responsibility. One of the most basic quality principles, the “artisan concept” popularized by Joseph Juran, defines three fundamental assumptions about
any task performed by a worker: (1) The worker knows the job; (2) the worker knows when the result is unacceptable; and (3) the worker has the knowledge and
authority to bring unacceptable performance within the tolerance of acceptable performance.
Not only does managed care grossly violate the artisan concept, but it turns it upside down. Utilization Review (UR) is a structure of checks, barriers, and
supervision of physicians and hospitals. Physicians, unlike workers on the assembly line, apparently cannot be trusted to practice their profession without being
continually badgered and molested.
To compound the problem, the people in the UR roles are usually less medically qualified than the attending physician and, being on the other end of a phone
line, always less
knowledgeable about the situation. This is the polar opposite of a TQM principle, and no other business operates this way.
What if the chief financial officer of General Motors had to have a union assembly line worker approve his annual targets for return on capital? What if an airline
pilot had to radio the flight attendant before initiating each step in landing a jet? How, with managed care, can the wrong principle applied in the wrong way
somehow produce desirable results?
This brings up the evolution of the term “managed care.” Physicians have always “managed” their practices. With “managed care,” medical care is now being
managed by administrators who have no knowledge of medicine and no contact with the patients.
There are two corollaries to the artisan concept. The first says that if a factory worker doesn't know more than anyone else about the confines of his workspace,
than he's unqualified and shouldn't be there. While this rule is usually invoked to discipline managers from excessive supervision of their blue-collar work force,
it applies doubly to work delegated to expert professionals. It would make no sense to hire Leonardo da Vinci to paint your portrait, only to have someone hover
over his shoulder continually.
The second states that the path to higher levels of quality is to cease reliance on inspection — or, “you can't inspect quality into a product. ”
To improve quality, one must either retrain the workers or improve the process. Hiring more inspectors just adds costs at a point in the process where it's too
late to make improvements.
So how do managed care entities determine whether their UR functions are effective? Does a UR function define its success by the number of physician requests
for diagnostic tests and surgical procedures it denies? Does this mean that if it denied every physician request it would be 100% successful? There is also the
countervailing factor: once physicians learn that roughly 20% of their requests will be denied, they simply increase the number of requests by a compensating
margin. This results in a lose-lose outcome: more inspectors at greater cost with no reduction in “defects.”
You Get What The System Delivers
One of Deming's quality maxims states that you'll eventually get whatever your system is designed to deliver. If a system provides a worker with defective parts,
you'll end up with defective products. If you design a system to admit fewer people into the hospital, that's what you'll get. Whether this translates into better
medicine is anyone's guess.
The Japanese have long held - and to some degree still hold - an advantage in this area of worker autonomy because trust comes naturally to Japanese managers,
while Americans continue to struggle with the concept. The best way to assure that a job gets done right is not to increase worker supervision, but to educate,
empower, and trust the person assigned to the job.
This notion of trust has filtered down into nearly every level of the U. S. economy — except medicine. An assembly worker for an American auto company with
a sixth-grade education can now be trusted to pull a cord and shut down the production of an entire plant, putting millions of dollars at risk, if he perceives a quality
problem. Yet as a result of the infamous Stark amendment, a physician can't refer a patient for a routine test to a laboratory facility in which he has a financial
Why are physicians the one and only occupation in the economy that can't be trusted to practice in the interest of their employers, customers, and society? If you
purchase a Hewlett-Packard computer, the salesman will also recommend that you purchase a service agreement from — surprise! — Hewlett-Packard. When
you bring your Buick into a GM dealership for service, the mechanic will install replacement parts from AC Delco, a wholly owned subsidiary of — surprise again!
— General Motors.
It makes no sense to trust a physician with your life, while simultaneously deeming him too crooked to order a diagnostic test worth $75.
Eliminate Numerical Quotas
One of the common traps that derails TQM efforts is to equate numerical measurements with quality outcomes. Quality does not reside in precise abstract
numbers, but in vague and temperamental perceptions that reside within the customers' minds. Deming called this trap “a denigration to counting”: accumulating
and adding up numbers because they're available, and relying on those numbers because it's the path of least resistance. To counter the natural tendency to fall
prey to this error, Deming continually reminded managers that “the most important numbers in any business are unknown and
The medical profession is full of examples of denigration to counting: rates of cesarean deliveries, hysterectomies, and hospital admissions. Regardless of what
these numbers are, managed care entities invariably seek to reduce them. What level of cesarean deliveries or hysterectomies denotes quality health care? If the
answer is always “more” or “less” (as with welfare spending or the size of the defense budget), then you're operating in the realm of politics, not quality
How many phone calls should a customer service representative handle in one hour? If you're concerned about quality, that's the wrong question. Phone
representatives shouldn't be measured against numerical quotas, but in terms of satisfied customers. How many cesarean deliveries should an obstetrician perform?
As many as circumstances call for. Counting merely measures usage; it doesn't indicate value.
Reduction and Elimination of Variance of Outcomes
When dealing with inanimate substances and mechanical parts, performance can be measured and controlled to a high degree of precision. This is why TQM is
a science: the circumstances that produce a single quality part, if replicated, will continue to produce quality parts until the system or process is thrown out of
If a process is producing acceptable parts one minute and defective ones the next — without any rhyme or reason — then there is too much slack or variance in
the system. The objective of TQM is to reduce the variance of the system parameters so that all the parts produced are to uniform standards. This can be done
on an assembly line working with inanimate substances to produce standardized products. Basic chemical elements can be made to conform to exact standards
under controlled conditions; human beings cannot. A TQM fundamental in manufacturing and service industries is to prevent and eliminate exceptions
and outliers. Yet medicine is the diagnosis and treatment of sick people — i.e. exceptions and outliers to the natural state of health. In other words, much
of the practice of medicine, by definition, lies outside the boundaries of the practice of TQM.
Sick patients often get well without any medical intervention. Some even get well in spite of medical intervention. This never happens with machines. Your
car's ability to stop won't improve if you continue to ignore warnings to replace the brake pads.
This is why we have medical quackery and not engineering quackery. As soon as someone claims to have discovered a mechanical process, the scientific
community demands that the results be replicated exactly as prescribed. And should the advertised results fail to materialize on a uniform and universal basis
(as with the “cold fusion” phenomenon a few years ago), the originating scientists are exposed as a fraud and ostracized from the scientific establishment. Yet
in medicine, research foundations will claim success in treating a disease by stating that, under certain loosely controlled conditions, mild levels of success can
be obtained a certain percentage of the time. This is considered to be valid medical research.
In the manufacturing of products, variance of results is an anathema. In medicine, it's acceptable and inevitable. Honda expects all of its fuel injection systems
to perform to exact specifications uniformly. Honda's ultimate objective is to produce cars of superior quality that are indistinguishable from each other. They
want the car buying public to think, “if you've driven one Honda Accord, you've driven them all.”
But obstetricians don't measure the duration of a mother's labor against a single standard any more than they expect every baby they deliver to be the same size,
weight, and sex. Unlike the automotive engineers at Honda, a physician practices his craft with the knowledge that “when you've seen one patient, you've seen
Finger Pointing and Counter-pointing
The insurance industry, with its army of actuaries and arsenal of numbers, has led the battle to exercise greater control of the “health care delivery process.” Two
of their most contested points in the implementation of managed care are the incidence rates of hysterectomies and cesarean deliveries. The fact that these vary
widely is cited as evidence of poor quality.
This argument is founded on three premises: (1) People are the same all over the country, and therefore, medical practice should be just as uniform; (2) wide
variances in medical practices are evidence of sloppy, and perhaps even aggressively unethical practice; and (3) practice patterns in the areas of lower rates of
certain procedures are, by definition, the areas with higher quality care.
First, more than anyone else, actuaries are aware that the geographic distribution of the population of the United States is most definitely not uniform. Areas
vary by ethnic groups, age distribution, lifestyles, standard of living, and personal preferences. Moreover, all of these categories of geographic variances can have
significant effects on the numerical quantifications of aggregate data on medical practice.
Second, although the financing and payment system doesn't reflect it, the practice of medicine is a consumer service, subject to the capital resources of the
sellers and the personal preferences of the buyers. While a hysterectomy or cesarean delivery may be viewed as an indiscriminate commodity that can be bought
or sold on a trading floor by an HMO or insurance company, it's actually a matter of professional judgment and expertise for the physician and a personal evaluation
of cost, risk, and rewards to the patient.
The insurance industry certainly doesn't measure up any more favorably on its own standards. Insurance rates, cost structures, and profit margins vary from
company to company just as much, if not more, than measures of medical practice. Actuaries and administrators, who are so bold to question the practice patterns
of physicians, are not about to let physicians turn the tables on the them and dictate how to underwrite employer groups, price health insurance, or pay claims!
An actuary would never think of barging into the cockpit of an airplane and hovering over the pilots telling them how they should proceed through their flight
checklist. Yet medical care, which also may be a matter of life and death, is in danger of being micro-managed by these same “professionals” who casually sort
physicians into “high” or “low” quality categories based on impersonal aggregated numerical comparisons.
It makes no sense to give the local TV weatherman a raise for a nice sunny day or to hold him accountable for the damage done by a tornado. Similarly, the
medical profession is often helpless to reverse the progression of certain diseases. The best doctors can do is to provide an incomplete but informed professional
opinion and recommend alternatives, the outcomes of which cannot be assured, as so many factors are beyond their control.
TQM confines itself to inanimate objects and solidifies proven processes and products. Medical science operates in the unproven realm of the interaction
of the human mind and body and continually pushes against the envelope of the unknown. Since patient outcomes can't be predicted with the certainty of
mechanical processes, physicians need to be evaluated by different criteria.
Before physicians agree to be measured against any “quality” standards, the basic terms should be defined and goals agreed to. Is quality defined as a lowered
risk in delivering a newborn baby? Or does it mean a rate of cesarean deliveries comparable to Ontario, Canada?
Finally, physicians should be left alone to practice their craft, just like any other profession from accountant to auto mechanic. They also should be judged
and held accountable according to standards that reflect the uncertain nature of the environment in which they practice.
Reprints available from:
Association of American Physicians and Surgeons
1601 N. Tucson Blvd. Suite 9, Tucson, AZ 85716
Pamphlet No. 1069, December, 1999