Statistical quality control in hematology analyzers

The statistical quality control carried out in hematology analyzers has many important
differences from the corresponding techniques in the clinical chemistry analyzers. These
differences are due to reasons such as the high stability of cytometry technology, the small
biological variation of some hematology parameters, the big reagent vials and the small time
lasting of the hematology controls.
Because of the above reasons Levey-Jennings charts in hematology analyzers are different
from corresponding charts in clinical chemistry. For instance the hematology Levey-Jennings
charts have only three lines (upper and lower limits and central line). The reason is that these
Levey-Jennings charts are not created statistically from a normal distribution of former quality
control data, which is not possible because of the very small variation of hematology quality
control values. In hematology analyzers the upper and lower control limits act as the
“specification’s limits” in industry quality control.
The small biology variation of many hematology parameters made many researchers to
established quality control methods based only on patients results. Such suitable parameters
are the erythrocyte indexes (MCV, MCHC, MCV) with the smaller biological variation (due not
only to biology but mostly to the hematology analyzer’s technology). These attributes of them
inspired Brian Bull (an American Hematologist) to establish a new quality control method widely
known as “Bull’s algorithm”.
Bull’s algorithm (also known as method) detects systematic errors in MCV, MCHC and MCV
and consequently in HgB, Hct and RBC. His method is a kind of moving average. Its main idea
is to estimate the mean value of the last twenty patients’ values, including in them the mean
value of the batch of the previous twenty values. The algorithm itself is a quite complicated
equation which eliminates the outliers and estimates the moving average of the last twenty
values. Bull’s algorithm has been proved quite effective in detecting small systematic errors
(almost 1%) not only in erythrocyte indexes but also in almost all the hematology parameters. It
uses all patients’ data without exception. The last fact made Bull’s algorithm the cheapest
quality control method in laboratory medicine.
Hematology quality control samples last only 20 – 30 days and are very expensive, when, on
the other hand, whole blood samples are stable in the refrigerator for 24 hours. These facts led
some researchers to find methods which are based on the repetitive analysis of patient
samples. These methods are known as “retained patient specimens”.
In 1988 Cembrowski (Canadian clinical chemist) established the most effective “retained patient
specimens” method. It was based on the repetitive analysis of the same patient samples
between two successive days. His method is known as “m/nlim”.
-       “Lim” stands for the quality control limit. It is equal to the double of the standard deviation of
the repetitive analysis (2 x SD).
-       “n” stands for the number of patients’ samples which will be analyzed twice.
-       “m” stands for the portion of “n” number of samples which is permitted to be out of limits
(“lim”).
 
Statistical simulations created by Cembrowski proved the effectiveness of his method.
According to him the best combination of “m”, “n” and “lim” is 2, 3, 2 or 2/32s.
 
Concluding, three different methods are in the disposal of the laboratory in order to detect the
analytical errors in hematology laboratory. Levey-Jennings detects systematic and random
errors. On the contrary, Bull’s algorithm and “retained patient specimens” detect only systematic
errors, but they have the advantage of the low cost. Laboratory can choose the best
combination of the three.

 Tελευταία ενημέρωση:  Κυριακή Ιανουαρίου 20, 2013