You are on page 1of 3

# ),.

lJLATING
A STANDARD DEVIATION
leviation [s] is a statistic which quantifies how close numerical values
dues)are in relation to each other. The term precision is often used
eably with standard deviation. Another term, imprecision, is used to express
art numerical values are from each other. Standard deviation is calculated for
ducts from the same data used to calculate the mean. It provides the
an estimate of test consistency at specific concentrations. The repeatability
ay be consistent (low standard deviation, low imprecision) or inconsistent
larddeviation, high imprecision). Inconsistent repeatability may be 'due to the
nvolved or to a malfunction. If it is a malfunction, the laboratory must correct
TI.

## )Ie to get repeated measurements of the same specimen as close as possible.

ision is especially needed for tests which are repeated regularly on the same
rack treatment or disease progress. For example, a diabetic patient
care situation may have glucose levels run every 2 to 4 hours. In this case, it
It for the glucose test to be precise because lack of precision can cause loss
lbility. If there is a lot of variability in the test performance (high imprecision,
ard deviation), the glucose result at different times may not be true.

W~-jl~J~J
.........................................
, """""""""'" '.

'/'/.:~~=":"\
:'..) ...J ,.) .

.'\~~~~.//~

.../:: :?..............

! :~:~~/})
) ,.
Basic Lessons
Although most calculators and. spreadsheet programs automatically calculate standard
In Laboratory
deviation, it is important to understand the underlying mathematics.
Quality Control

## To calculate a standard deviation for a set of QC values

(X1l X2J ...xn) use the formula:

I(xn - X)2
s =
n-1

where: s = standarddeviafion
x = mean(average)of the QC values
I(xn - X)2= the sum of the squaresof differencesbetween
individualQC valuesand the mean
n = the numberof valuesin the data set

To calculate the standard deviation for the normal level of control (Levell) in Table 1,
begin by calculating the mean [x]:

## x = 4.0 + 4.1 + 4.0 + 4.2 + 4.1 + 4.1 + 4.2 mmol/L + 7

x = 28.7 mmol/L + 7
x = 4.1 mmol/L

## s~ ~ I(xn - Xj2 n-1 .

s= (4.0- 4.1)2+ (4.1- 4.1)2+(4.0-4.1)2+ (4.2- 4.1)2+ (4.1- 4.1f+ (4.1- 4.1)2+ (4.2-4.1f
6

6

## s = ~ 0.01 + 0.0 + 0.01 + ~01 + 0.0 + 0.0 + 0.01

s = ~ 0.04 + 6
s = 0.082 or 0.10 (Rounded)

The standard deviation for one week of testing of the normal potassium control level is
0.082 mmol/L.8 Now that the amount of precision is known, some assumptions can be
made about how well this test is performing.

8This type of standard deviation is called between run standard deviation because the data used to calculate
18 the statistics came from different analytical.runs.
) .
CREATINGA LEVEJY-JENNINGS
CHART
Standard deviation is commonly used for preparing Levey-Jennings (L-J or LJ) charts.
The Levey-Jennings chart is used to graph successive (run-to-run or day-to-day) quality
control values. A chart is created for each test and level of control. The first step is to
calculate decision limits. These limits are :t1S, :t2s and :t3s from the mean. The mean
for the Levell potassium control in Table 1 is 4.1 mmol/L and the standard deviation
is 0.1 mmol/l.11 The:t 1s, :t2s and :t3s quality Gontrollimits are calculated as follows:

## '(:t1s) 4.1 - 0.10)(1)= 4.0

4.1 + 0.10)(1) = 4.2

## (:t2s) 4.1 - (0.10)(2) = 3.9

4.1 +(0.10)(2)=4.3

## (:t3s) 4.1 - (0.10)(3) = 3.8

4.1 + (0.10)(3)= 4.4 '

These ranges are used with the mean to construct the Levey-Jennings chart
as shown ,in Figure 3.

## Unassayed Chemistry Control, Lot No. 12345

LEVEL 1 (Normal Control)
Test: Potassium

' _

## ~E 4.2 -I I --t 1 1 1 ( I '1"' I I---mlm'--lmml--- --I Imm!- +1s

. ,,'
.5. 4.1 -f--':;f--- r--- 1
- I I I '
,.L---,l---r---t---t---1"---+--+---~---t I---+--+--+---t MEAN
CD
01 I i
-_u,- ---u:
~
a:
4.0' _m-- --m- --m:
--uu
uml-U_--I ~-I f--'---I I I-U---I i I !- -1s

:,,:,1 2 3
[,::':::::"':::::::.:..:
4 5
r
6 7 8
III.C:::..:::.::
9
Run Number
10 .11 12' 13 14 15 16 17 18 19

The Levey-Jennings chart we have developed can be overlaid onto a bell-shaped curve to
illustrate the overall distribution of quality control values (see Fig. 4 on the following page).

11Rounding of the mean and standard deviation to the nearest tenth is allowable in this example because
potassiUm results are generated and reported to the nearest tenth. The standard deviation of 0.08 mmol/L
is rounded to 0.10 mmol/L - _I: