MYTHS OF QUALITY
MYTHS OF QUALITY
A MYTH is a Mistaken Yarn, Theory, or Hypothesis!
 A word from Dr. Westgard |
MYTHS OF QUALITYA MYTH is a Mistaken Yarn, Theory, or Hypothesis! |
|
Historical myths of
cartography:
Mythical island of California
Mythical island of Friesland
Mythical islands in Lake
Superior
Modern myths of quality:
QA assures quality in healthcare
Statistical QC controls the quality
of laboratory tests.
Quality can be managed even if
the required quality isn't known.
Quality requirements need to consider
only imprecision and inaccuracy.
Current methods have better imprecision
and inaccuracy than needed.
Analytical quality is a given today.
No further improvements in analytical
quality are needed.
Need for quantitative management
of analytical quality
Did
you know that California was an island? It's well documented on
the most reputable maps of the 1600s that California was completely
surrounded by water. For example, see the accompanying map that
shows the Isle de Californie. There it is, documented in
black and white, proof that California was an island.
This map of Nouveau Mexique et Californie by Alain Mallet was published in 1683 in the Description de l'Univers (Paris). Mallet copied the flat-topped model of California that appeared in an earlier map by Sanson, who was one of the most distinguished French cartographers. It was very common for mapmakers to copy each other's work. When a new discovery appeared on one map, it was widely disseminated on many of the maps of the time. The discovery that California was an island was first documented in 1622 and persisted on maps as late as 1750, even though evidence in 1705 clearly established that this was a myth.
Actually, there is quite a history of mythical islands, suggesting that these myths are not as rare as you might expect. In the late 1500s, one of the most famous mapmakers, Abraham Ortelius, prepared a map of the Northern Atlantic that showed an island of Friesland lying a bit west and south of Iceland, complete with a detailed description of the coastline, the harbors, the people who lived there, what they looked like, and what they did for a living. The Ortelius map is beautiful, decorated with sailing ships and sea creatures, and was the most authoritative map of the area at that time. The only problem was that Friesland didn't exist. When people sailed to the new world and passed Iceland, they ascribed more and more details and reality to Friesland because they expected it was the next body of land.
|
|
Another example that is of interest to those
of us in the Midwest are the Isles of Phillipeaux and Pontchartrain
in Lake Superior. When the border between the U.S. and Canada
was settled by the Treaty of Paris in 1783, it was decided that
these islands would be part of the U.S. (The map shown here is
from 1780 by the cartographer Bonne documents the existence of
these islands, though it uses the name Minong instead of Phillipeaux
for the island in the southern or bottom part of Lake Superior.)
In the early 1800s when Wisconsin was being settled, the U.S.
government sent out surveyors to map this area more completely,
but they couldn't find these islands! They showed up on all the
maps of the time, but they didn't show up above the water. It
seems that that the explorers created these islands and named
them after the government minister who was funding their investigations.
They probably needed some preliminary findings to justify getting
more money for further explorations, somewhat akin to the process
of obtaining research funding today.
These myths are amusing in retrospect, but they were taken very seriously at the time and caused some serious problems later on. There are myths today about quality that are also taken very seriously and will cause us serious problems in the future. Some of them hit very close to home - the quality of healthcare and the quality of laboratory testing.
It's a mistaken yarn that puts a good spin on current efforts to measure the quality of healthcare. As healthcare providers, we all talk about quality assurance (QA), but our quality assurance programs (which are often required by regulation and accreditation) primarily deal with measuring performance. Quality Assessment would be a better name for these efforts. While it is important to assess quality to know how well we're doing, measuring quality doesn't assure that the necessary quality will be achieved. Achieving quality actually requires quality planning, which starts with defining the quality that is needed, then builds that quality into the process.
It's a mistaken theory that the use of statistics somehow assures that laboratory test results have the necessary quality. Virtually all laboratoryes apply statistical quality control (SQC) as part of their efforts to assure the quality of laboratory tests. While we may not understandthe theory or the statistics, we still believe that something magical happens because of those statistics. Somehow by analyzing controls and plotting results on control charts, we expect to control the quality of our testing processes, even if we don't understand how it works.
It's a mistaken hypothesis that quality can be managed when we don't know the quality that is needed. Few laboratories have defined the analytical quality that is needed for the tests they perform. How is it possible to know we are achieving the unknown? Can you manage the finances of the laboratory without knowing the budget? Don't you need to know the quality required for a laboratory test if you are to manage the quality of the testing process?
This problem with quality requirements gets to be even more complicated. Current thinking about quality goals and requirements is flawed because it considers only the stable method performance characteristics (imprecision and inaccuracy). If performance is stable, why bother doing quality control at all? If QC is necessary, then we must have to consider the performance characteristics of QC procedures (probabilities for error detection and false rejection) in our goal setting models?
The net effect of all these myths is the belief that the performance of current laboratory methods is better than required for medical needs. This belief is based on a mistaken theory for setting quality goals, a mistaken hypothesis in equating all medically tolerable variation with analytical variation, disregarding the subject's own biological variation, and the mistaken assumption that QC procedures have ideal response curves and can detect any change in performance, regardless how small.
As a consequence of these myths, there is a common feeling today that analytical quality is a given, i.e., analytical quality itself is being assumed today. In the midst of current programs in Total Quality Management (TQM) and Continuous Quality Improvement (CQI), it is often mistakenly assumed that the problems in technical quality management have already been solved. This represents the mistaken hypothesis that past efforts have resolved any technical difficulties, so now we can get on to new issues that are in vogue, such as monitoring customer satisfaction, measuring patient outcomes, etc.
The collective result of all these myths is a false sense of security regarding the quality of laboratory testing processes. Many think analytical quality is so good that there is no need for further improvements. This is the most serious myth of all because it makes us complacent about what we doing and hinders efforts to further improve the analytical quality of laboratory tests.
These myths need to be exposed if the technical management of analytical testing processes is to be improved. That's the purpose of this introduction! You need to critically assess many of the quality management practices that are accepted in laboratories today. To begin, you need to understand how quality requirements can be defined, how method performance should be measured experimentally, how the experimental data can be analyzed with statistics to estimate analytical performance characteristics, and how a decision on the acceptability can be made.
Once the performance of a method has been determined to be acceptable (Basic Method Validation), you need to select a statistical QC procedure that can detect medically important errors (Quality Planning), make routine measurements on the necessary number of controls, and interpret the control results using the appropriate decision criteria or control rules (Basic QC Practices).
Other Essays:
