##
LABORATORY ERROR AND THE LEAN SIX SIGMA PROCESS

Originally lean and six-sigma were separate ideas
designed to achieve two related metrics: time and error. Lean was designed to
eliminate non-value-adding steps and six sigma aimed to reduce variation in
process. The Lean Six
Sigma projects comprise the Lean's waste elimination projects and the Six Sigma
projects based on the quality characteristics of a process. Lean
is a process adopted to eliminate waste from a process, first practiced and
then formalized into the Toyota Production System.

The objective of lean was to reduce time; the objective of
six-sigma was to reduce error. Today both are combined to form lean and
six-sigma. Lean six sigma measures the amount of non-value adding steps in a
process to reduce variation and improve performance of a process, as part of
its core metrics. A process sigma represents the capability of a process to
meet (or exceed) the process requirements. It reflects the number of defects
(errors) per million opportunities (DPMO). The sigma refers to the number of
SDs from the mean a process can be before it is outside the acceptable limits.
E.g. if sodium has six sigma performance, then the mean could shift by six SDs
and till meet the laboratory requirements. A 6 sigma process has narrow process
SD and produces only 3 errors for every million tests. A 3 sigma process has
much wider SD and produces about 26,674 errors per million tests.

There are various ways to calculate the sigma of a process.
In order to calculate the sigma, defects must be clearly defined. The most
straight forward method uses the process yield-the percentage of times that a
process is defect free. Another simple method is to calculate the DPMO. Both of
these methods then require finding the process sigma on the process sigma
chart.

Six sigma metrics can be plotted graphically. This chart incorporates many of the measures like total allowable error for given analyte or given process, systematic error and imprecision. Systemic error and imprecision are derived from the COM experiments and AE is by the specification given by CLIA regulations.

Six-sigma is an evolution in quality management which is widely implemented in business and industry in new millennium. Six sigma metrics are being adopted as universal measure to maintain quality in process. The principles of six-sigma go back to Motorola’s approach to TQM in the early 1990s. This means that variation upto 6 sigmas or 6 standard deviations should fit within the tolerance limits for the process; hence the name six sigma. For this development Motorola won the Malcolm Baldridge Quality award in 1988.

Six sigma provides a framework for evaluating process performance and
process improvement to reduce variation. The goal for process performance is
illustrated in figure below which shows the quality requirements for that
measurement or process.

Any
process can be evaluated by determining how many sigmas fit within the
tolerance limits. There are two methods for assessing process performance in
terms of sigma metric. One approach is to measure outcome by inspection. The
other approach is to measure variation and predict process performance.

Conversion to sigma metric is done
by using standard table available. In health care organization a defect rate of
0.033% (333 DPM) is considered excellent, where error rates from 1% to 5% are
often considered acceptable. A 5.0% (50000 DPM) error rate corresponds to 3.15
sigma performance and 1.0% error rate corresponds to 3.85 sigma. Six sigma
shows that the goal should be error rate of 0.1% (4.6 sigma) to 0.01% or 100
DPM (5.2 sigma) and ultimately 0.001% (5.8 sigma).

The application of sigma metrics
for assessing analytical performance uses the variable obtained during method
validation studies, like accuracy, precision, PPP, NPP, sensitivity,
specificity parameters and that available from internal and external quality
control processes.

For the particular method for given
analyte, the allowable error, method bias, method CV, can be obtained from
external quality assessment programs or regulatory requirements (like US
Clinical Laboratory Improvement Amendment [CLIA] criteria for acceptable
performance in proficiency testing). Process variation and bias can be
estimated from method validation experiments, peer comparison data, proficiency
testing results and routine QC data.

In the laboratory sigma performance
of the method can be determined from imprecision: SD or CV and inaccuracy
(bias) observed for a method and quality requirement (allowable total error, TE

_{a}) for the test. [Sigma = (TE_{a}– bias)/SD]. Sigma metric from 6.0 to 3.0 represents the range from best case to worst case respectively. Methods with sigma performance less than 3 are not considered acceptable for production.