Basic & Intermediate Systems

Clinical Website, www.bio-rad.com/diagnostics Bio-Rad Laboratories Main Office,

4000 Alfred Nobel Drive, Hercules, California 94547, Ph. (510) 724-7000, Fx. (510) 741-5824
of Quality Control
Diagnostics Also in: Regents Park, Australia, Ph. 61-2-9914-2800, Fx. 61-2-9914-2888

Group
Vienna, Austria, Ph. 43-1-877-8901, Fx. 43-1-876-5629 Nazareth, Belgium,
Ph. 32-9-385-5511, Fx. 32-9-385-6554 Rio de Janeiro, Brazil, Ph. 5521-3461-5202,
Fx. 5521-2224-6524 Montreal, Canada, Ph.1-514-334-4372, Fx. 1-514-334-4415
for the Clinical Laboratory
Prague, Czech Republic, Ph .420-2-41430532, Fx. 420-2-41431642 Shanghai, China,
Ph. 86-21-64260808, Fx. 86-21-64264988 Copenhagen, Denmark, Ph. 45-4452-1000,
Fx. 45-4452-1001 Espoo, Finland, Ph. 358-9-804-22-00, Fx. 358-9-804-11-10
Marnes-la-Coquette, France, Ph. 33-1-4795-6000 Fx. 33-1-4741-9133 Munich, Germany,
Ph. 49-89-318840, Fx. 49-89-318-84100 Quarry Bay, Hong Kong, Ph. 852-2789-3300,
Fx. 852-2789-1257 Haryana, India, Ph. 91-124-6398112, Fx. 91-124-6398115 Rishon
Le Zion, Israel, Ph. 972-3-9514127, Fx. 972-3-9514129 Milan, Italy, Ph. 39-02-216091,
Fx. 39-02-21609-398 Tokyo, Japan, Ph.81-3-5811-6290, Fx. 81-3-5811-6282 Seoul,
Korea, Ph. 82-2-3473-4460, Fx. 82-2-3472-7003 Miami, Florida, Ph. 305-894-5950,
Fx. 305-894-5960 Mexico D.F., Mexico, Ph. 5255-5534-2552, Fx. 5255-5524-5555
Veenendaal, The Netherlands, Ph. 31-318-540666, Fx. 31-318-542216 Auckland,
New Zealand, Ph. 64-9-415-2280, Fx. 64-9-415-2284 Oslo, Norway Ph. 47-23-38-41-30,
Fx. 47-23-38-41-39 Warsaw, Poland, Ph. 48-22-3319999, Fx. 48-22-3319988
Amadora, Portugal, Ph. 351-21-472-7700, Fx. 351-21-472-7777 Moscow, Russia,
Ph. 7-095-721-14-00, Fx. 7-095-721-14-12 Singapore, Ph. 65-6415-3188, Fx. 65-6415-3189
Johannesburg, South Africa, Ph.27-11-442-85-08, Fx. 27-11-442-85-25 Madrid, Spain,
Ph. 34-91-590-5200, Fx. 34-91-590-5211 Sundbyberg, Sweden, Ph. 46-8-555-127-00,
Fx. 46-8-555-127-80 Glattbrugg, Switzerland, Ph. 41-61-717-95-55, Fx. 41-61-717-95-50
Bangkok, Thailand, Ph. 662-651-8311, Fx. 662-651-8312 Hemel Hempstead, United
Kingdom, Ph. 44-208-328-2000, Fx. 44-208-328-2550

©1991 Bio-Rad Laboratories, Inc. • Lit. No. Q-1102 4/05 • All Rights Reserved • Printed in the USA
QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 1

What is Quality Control?

Quality Control in the clinical lab o rat o ry is a system designed to
increase the pro b ability that each result rep o rted by the lab o rat o ry
is valid and can be used with confidence by the physician making
a diagnostic or therapeutic decision. Quality control (QC)
procedures function by detecting analytical erro rs; ideally any
error large enough to invalidate the medical usefulness of
l ab o rat o ry results should be detected. In pra c t i c e, m a ny QC
procedures operate by submitting controls (sample mat e rials well
ch a ra c t e ri zed by previous testing) to the laboratory testing
process, and comparing the results of current testing to an
expected ra n ge of values derived from previous testing.

What are the basic quality control statistics?

The expected ra n ge of values for a control is calculated by use of
re l at ively simple statistics. These statistics incl u d e :
mean (x̄)
standard deviation (s)
coefficient of variation (CV); and
the standard deviation index (SDI).

Mean
The mean is defined as the
arithmetic average of a set of
data points. It is expressed as:

where:
xi = each data point
n = the number of
data points in the set
The mean describes the "central tendency"
of the data set. In the clinical lab, the mean identifies the "target
value" of a set of data points, usually QC or patient data.
2 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 3

Standard deviation is quantified using the following formula:

The mean is the fundamental statistic used for comparison or for

c a l c u l ation of other statistics. The National Committee for Clinical
L ab o rat o ry Standards (NCCLS) recommends that at least 20 data
where:
points collected from 20 or more "separate" runs be used to
2
establish laboratory target values for control (x ) = the sum of the squares of each value of x
2
mat e rials. Lab o rat o ries should establish their (x) = the sum of all data points squared
own target values using manu fa c t u rer assay n = the total number of data points in the set
values only as guides. Provisional targe t
Using the previous LDH example, the standard deviation is
values may be established by running 20
calculated as follows:
rep l i c ates in less than 20 ru n s , but the
provisional values must be replaced x = 587
after data from 20 sep a rate runs is (x)
2 2
= (587) = 344,569
accumulated.3 H oweve r, for 2
p u rposes of this discussion, o n ly
(x) /n = 344,569/5 = 68,913.8
2
five data points will be used in the (x ) = [(120)2 +(115)2 ...n] = 69,015
fo l l owing illustrat ive examples. s= (69,015 – 68,913.8)/4
s = 5.03 IU/L
EXAMPLE: LDH QC values of {120, 115, 110, 119,123 IU/L}
represent a data set of 5 points. The sum of the Limits for data accep t ability are defined using the standard deviation
set is 587 and the mean is 587 ÷ 5 = 117.4 IU/L. statistic. The ra n ge for the 1s limit would be calculated as:

Standard Deviation (Mean) +/– (1)(s)

The standard deviation (s) quantifies the degree of dispersion of
Consequently, the 1s ra n ge (limit) for our LDH example is
data points about the mean and is used to set limits upon which
c a l c u l ated as:
control result acceptability is determined. Quality control data
often exhibit a "normal" or Gaussian distribution around the mean. 117.4 IU/L – 5.03 IU/L = 112.4 IU/L
117.4 IU/L – 5.03 IU/L = 122.4 IU/L
In a Gaussian distribution:
The 1s range is 112.4 to 122.4 IU/L.
• 68.3% of values are within ± 1.0 standard
deviation of the mean Approximately 68% of future data should be between 112.4 and
• 95.5% of values are within ± 2.0 standard 122.4 IU/L. Approximately 32% should be less than 112.4 IU/L or
deviations of the mean greater than 124.4 IU/L.
• 99.7% of values are within ± 3.0 standard
deviations of the mean
4 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY
The 2s range (limit) is calculated as:
(Mean) +/– (2)(s)
117.4 IU/L – (2 x 5.03 IU/L) = 107.3 IU/L
117.4 IU/L + (2 x 5.03 IU/L) = 127.5 IU/L
The 2s range is 107.3 to 127.5 IU/L.
Only about 4.5% of future data should be less than 107.3 IU/L or
greater than 127.5 IU/L; i.e., only one result in 20 should be beyond
these limits.
The 3s range (limit) is calculated as:
(Mean) +/– (3)(s)
117.4 IU/L – (3 x 5.03 IU/L) = 102.3 IU/L
117.4 IU/L + (3 x 5.03 IU/L) = 132.5 IU/L
The 3s range is 102.3 IU/L to 132.5 IU/L.
O n ly about 0.3% of future data should be less than 102.3 IU/L or
gre ater than 132.5 IU/L. It would be ve ry unusual to obtain a re s u l t
beyond these limits.
In the clinical lab o ratory, these ra n ges (limits) are used to
d e t e rmine the acceptability of a test run not only on the basis of a
single data point but on groups of data points as well. This topic is
p resented in the next section.
S t a n d a rd dev i ation is also va l u able when comparing methods or
eva l u ating new instruments. A method or instrument with a low
s t a n d a rd dev i ation produces consistent results. The lab using an
i n s t rument or method wh i ch has high standard deviations will have
less certainty about the accura cy of diagnosis or the effe c t ive n e s s
of tre atment because of test result va ri ab i l i t y. In other wo rd s , high
s t a n d a rd dev i ations (poor pre c i s i o n , gre ater variability) can affe c t
the integrity of all results.
The method or instrument selected should provide a standard
deviation wh i ch is medically accep t abl e.4
5
Coefficient of Variation
The coefficient of va ri ation (CV) is a measure of variability. A method
or instrument's CV is ex p ressed as a perc e n t , and is calculated as:
CV(%) = (Standard deviation (s) ÷ Mean)(100)
The CV for our LDH example would be:
(5.03 IU/L / 117.4 IU/L)(100) = 4.3%
The CV is useful for comparisons of precision at diffe re n t
concentrations as long as the materials used are similar and CVs are
determined under similar conditions. This statistic is commonly
used to compare manufacturer claims, CAP survey results, and peer
group QC reports. It can also be used as a par t of the internal
quality control system when performing patient precision testing,
which is presented later.
Standard Deviation Index
Another statistic which is helpful to evaluate performance is the
standard deviation index (SDI). This statistic, which is usually
obtained by participation in an external QC or proficiency testing
program, is used to compare a laboratory's results to its peer group.
It allows the lab to gauge its performance when compared to the
peer group.
The SDI for the mean is calculated as follows:
SDI = (lab mean – peer group mean)
peer group standard deviation
The target SDI is 0.0. This would indicate that the lab's
p e r fo rmance is identical to the peer group ave rage. Accep t able
SDI values are between +/–1.0. Any test/method/instrument wh i ch
has an SDI between +/–1.0 and 1.5 may have a pro blem and the
l ab should inve s t i gate. The lab must tro u bleshoot and correct any
test/method/instrument wh i ch has an SDI of +/–2.0 or gre at e r. The
re l at ive importance of the SDI statistic does dep e n d, however, on
the size of the peer gro u p .
6 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 7

The SDI statistic can be used also as part of the lab o rat o ry ' s
i n t e rnal QC system wh i ch is presented later in this document. It is
also useful in interp reting pro ficiency testing. The lab o rat o ry ' s
rep o rted result replaces the lab mean in the equation for SDI. In
this case, SDI values exceeding 2 or 3 suggest a pro bl e m .
What are the performance challenges?
NOTE: If any of the following rules is violat e d, the tech n o l ogist
must rev i ew the perfo rmance of the test, consult
tro u bleshooting guides, perhaps perform maintenance, c o rrect
a ny identified pro blems or dep a rt u re from protocol, and notify
the supervisor who will make decisions about rep o rting results
and re running the test.

The perfo rmance ch a l l e n ges consist of seve ral rules wh i ch define 13S This rule detects random erro r.
specific perfo rmance limits. These rules are commonly re fe rred to Vi o l ation of this rule may also
as We s t ga rd rules. If any of the rules is violat e d, then the analy t i c a l point to systematic erro r. Th e
run is considered out of contro l
run may be invalid and the test results unaccep t able. The We s t ga rd
when one control value exceeds
rules are va ri e d. Some are designed to detect random error; others the mean +/–3s. This rule is
detect systematic error wh i ch may indicate a bias in the system. applied within the run only.
Six rules are commonly used by laboratories in va rious
c o m b i n ations. Rule combinations are selected by the lab o rat o ry 22S This rule detects systemat i c
and are based on the number of control levels run with each erro r. It should be applied within
and across runs. This rule is
a n a lytical run. The ove rall objective is to obtain a high pro b ab i l i t y
v i o l ated within the run wh e n
of error detection and a low fre q u e n cy of false rejection of runs. two consecutive control values
(or 2 of 3 control values when 3
The six commonly used rules are: levels are being run) exceed the
"same" (mean +2s) or (mean
12S This is the "wa rning rule." –2s) limit. The rule is violated
If one control measurement a c ross runs when the previous
exceeds the mean +/– 2s, then value for a particular contro l
the tech n o l ogist must consider level exceeds the "same" (mean
other controls in the run +2s) or (mean –2s) limit.
("within-run") and in previous
runs ("across-run") befo re R4S This is a "ra n ge" rule and it
a c c epting the run and rep o rting detects random erro r. This rule is
the results. applied within the run only. Th i s
rule is violated when the
Laborat o ries wh i ch use this rule alone in perfo rming their quality diffe rence in standard deviation
c o n t rol will fre q u e n t ly reject runs wh i ch are va l i d. According to b e t ween two consecutive contro l
values (or 2 of 3 control values
Westgard1, fa i l u re to allow for valid points between 2s and 3s will
when 3 levels are being run)
result in falsely rejecting: exceeds 4s.
• 5% of all analytical runs when using one level of contro l ;
• 10% of all analytical runs when using two levels of control; and
• 14% of all analytical runs when using three levels of control.
8 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY
41S This rule detects systemat i c
error and is applied both within
and across control mat e ri a l s .
This rule is violated within the
c o n t rol mat e rial when the last
four control values of the same
c o n t rol level exceed the "same"
(mean +1s) or (mean –1s) limit.
The rule is violated across
c o n t rol mat e rials when the last
four consecutive control values
for diffe rent control leve l s
exceed the "same" (mean +1s)
limit or (mean –1s) limit. Th i s
rule may not re q u i re rejection
of the run. Rather, it can be
an indicator to perfo rm
i n s t rument maintenance or
instrument/kit calibration.
10X This rule detects systemat i c
error and is applied both within
and across control mat e ri a l s .
The rule is violated across
control mat e rials when the
last 10 consecutive va l u e s ,
regardless of control level, are
on the same side of the mean.
The rule is violated within the
control material when the last
10 values for the same control
level are all on the same side of
the mean. This r ule may be
modified to 9 replicates when
running three control levels, or
8 rep l i c ates when running 4
control levels. This rule may
not require rejection of the
run. Rat h e r, it can be an
indicator to perform instrument
maintenance or instrument/kit
calibration.
9
What are the optional protocols and statistical challenges?
Optional protocols and statistical challenges which can be used at
the discretion of the laboratory include patient precision testing,
CUSUM, calculation of anion gap and SDI.
Standard Deviation Index (SDI)
Although SDI is a statistic usually generated by participation in an
external QC or proficiency testing program, it can be used as a tool
in monitoring internal quality control performance, as well. For
example, if the laboratory suspects that a trend is occurring,
this suspicion can be validated by
use of the SDI statistic. In this
case, rather than using peer
group data in calculating the
statistic, the laborat o ry ' s
cumulative mean and cumulative
standard deviation are used. The
SDI fo rmula is modified as
fo l l ow s : (Mean of suspect data
points – lab cumulative mean) ÷
lab cumulative standard deviation.
An SDI value greater than +/– 1.0
indicates a possible pro blem with
the test.
Cumulative Sums (CUSUM)
Another technique which can be employed
selectively on pro bl e m atic tests to detect shifts or
t rends is CUSUM (Cumu l ative Sums).
Ap p l i c ation of this technique is ve ry simple. It is set up for each
level of control using the established mean and
10 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 11

standard deviation. The laboratory establishes an upper and lower Table 1. Possible CUSUM Scenario
threshold for each level of control. Any result beyond the thresholds Control: Assay. Chem. Lot: 10001 Month: Jan 1998
Test: LDH Mean: 117 IU/L s = 5 IU/L
triggers calculation of the CUSUM. The CUSUM calculation
Lower Threshold: 112 Upper Threshold: 122 Control Limit: +/–13.5
continues for successive results until the cumulative sum either
Date Run Value Diff CUSUM
exceeds the control limit and identifies an "out of control" situation 1/1 1 120
for the test under observation, or changes sign, and the calculation 1/1 2 117
1/2 1 108 – 4.0 – 4.0 Initiate CUSUM
is stopped. 1/3 1 123 + 11.0 + 7.0 End CUSUM
1/4 1 119
Westgard recommends a (threshold/control) limit combination of 1/5 1 126 + 4.0 + 4.0 Initiate CUSUM
1/6 1 127 + 5.0 + 9.0
+/– (1s/2.7s) or +/– (0.5s/5.1s) for more particular practioners of
1/6 2 127 + 5.0 + 14.0 Out of Control
this technique.5 These combinations yield low frequency of false
rejection of valid runs and high systematic error detection. Using In the example above, the first two control results do not violate
the LDH example with a mean of 117 IU/L and a standard either the lower or upper threshold limit. Perfo rmance is
d ev i ation of 5 IU/L, the CUSUM threshold/control limit a c c ep t able. The third control result (108) violates the lowe r
combination +/– (1s/2.7s) would result in a lower threshold of 112 threshold limit by – 4 units. The fourth control result (123) is high
or (mean –1s) and an upper threshold of 122 or (mean +1s). The but for calculation of CUSUM is compared to the lower threshold
control limit would be +/– 13.5 or (±/2.7s). limit which was first violated. The difference is +11 units. The
CUSUM becomes +7. Calculation of CUSUM ends and returns to
Each control result for the test being monitored is reviewed. If the
a value of zero because the CUSUM sign changed from negative to
result exceeds either threshold limit, CUSUM is initiated. The
TABLE 1 positive. The sixth control result violates the upper threshold limit.
on page 11 difference between the result and the threshold limit is logged
Subsequent control results cause a violation of the upper control
provides along with the sign of the difference. Each subsequent control
a simple limit (+13.5) and the test is considered to be out of control.
example of
result is compared to the threshold first violated and a cumulative
CUSUM calculation ends when the method is declared out of
this protocol sum of the differences is maintained. This continues until the
control and receives corrective action.
cumulative sum changes signs or the
control limit is exceeded.
Although no Westgard rules have been broken, this scenario
demonstrates a possible bias in the system. Six of eight values are
above the mean and three values almost exceed the mean +2s limit,
which would have triggered the 12s warning, indicating the need to
review compliance with other Westgard rules. Any corrective
actions taken in response to CUSUM violation or potential
violation should be noted on the table or on a separate log sheet.
12
Anion Gap
A fre q u e n t ly neglected tool used to spot ch e ck perfo rmance of
e l e c t ro lyte analy ze rs is calculation of the anion gap wh i ch is
defined as:2
Na – (CO2 + Cl) = 5 to 14
This simple fo rmula can be applied to electro lyte data at specific
intervals (time, runs, or tests) to monitor possible analytic error. If
several samples fail this simple screen during any one day or run,
then all of the patient results should be reviewed for possible
a n a lytical erro r.
Patient Precision Testing
Although the purpose of controls is to va l i d ate analytical runs, they
also identify potential pro blems with the analytical system wh i ch
i n cludes the instrument or kit, technologi s t , a n c i l l a ry equipment and
re agents. Sometimes lab o rat o rians have difficulty deciding wh e t h e r
the control is at fault when an out of control situation occurs,
especially if it is a rep e ated occurre n c e. Patient precision testing is a
useful mechanism to distinguish between analytical system
perfo rmance and control performance. It also increases sensitivity to
random erro rs .
Patient precision testing is re l at ive ly easy to implement. The
lab o rat o ry chooses an ab n o rmal or normal patient sample for repeat
testing on the next analytical run or next day. It is a fo rm of
duplicate testing.
Limits for the maximum allowable diffe rence between duplicat e
results may be derived by either of two methods. One method
re q u i res the lab o rat o ry to perfo rm a series of rep l i c ate tests on normal
and/or abnormal patient samples, defining the absolute difference
between each rep l i c at e, and fi n a l ly establishing an accep t able ra n ge
of perfo rmance based on these rep l i c ate differences.4,6
QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 13
The other system re q u i res the use of normal and/or abnormal patient
samples and the CV for the corresponding control level. In this
system, the CV is applied to the initial patient result to define an
accep t able rep l i c ate ra n ge.6 When the sample is retested,the rep l i c at e
value is compared to the predefined ra n ge. Calculated limits for
allowable diffe rences are applicable only over a narrow ra n ge over
wh i ch the standard dev i ation (or CV) is fairly constant. Th u s , specific
concentration ra n ges must be defined for selecting patient specimens
for use as controls.
Patient precision test results can be interp reted as follows:
1. If the control and the patient precision rep l i c ate are
within accep t able limits, then there pro b ably is no
p ro blem with the control or the system.
2. If the control and the patient precision rep l i c ate are
both outside accep t able limits, then there pro b ably is
a pro blem with the system.
3. If the control is within accep t able limits, but the
patient precision rep l i c ate is outside accep t able limits,
then there is possible random error occurring or a
possible pro blem with the integrity of the
p recision sample.
4. If the control is outside acceptable limits, but the
patient precision rep l i c ate is within acceptable limits,
then there is possible random or systematic error
occurring or a possible pro blem with the integrity
of the control.
How often should controls be run?
C o n t rols should be assayed with each analytical run and placed
ra n d o m ly through the run to detect analytical imprecision. Contro l s
should also have assay values within cl i n i c a l ly significant ranges.
Use of multiple levels of control allows for better lab o rat o ry
decisions rega rding analytical error and validity of the run.3
14 QUA LI TY CON TROL FO R THE CLI NI CA L L ABOR AT O RY 15

References Suggested Reading:

1. Westgard, J. O., Barry P. L.,Hunt, M.R., Groth, T., A multi- 1. Westgard, J. O., Koch, D. D., Oryall, J. J.,Quam, E. F.,
rule Feldbruegge, D. H., Dowd, D. E., Barry, P.L., Selection of
Shewhart chart for Quality Control in Clinical Chemistry. medically useful Quality Control procedures for individual
CLIN. CHEM. 27/3, 493-501 (1981) tests done in a multitest analytical system.
CLIN. CHEM. 36, 230 (1990)
2. Weisbrot, M.D., I. M., Statistics for the Clinical Laboratory,
J.B. Lippincott Company, Philadelphia, (1985) 2. Howanitz, Peter J. and Joan H., Laboratory Quality Assurance,
McGraw-Hill Book Company (1987)
3. National Committee for Clinical Laboratory Standards.
Internal quality control: principles and definitions; Tentative
Guideline. NCCLS document C24-A. Villanova, Pa;
NCCLS; (1991)

4. Cembrowski, G.S., Carey, R.N., Laboratory Quality

Management: QC & QA, ASCP Press (1989)

5. Westgard, J. O. et. al., Combined Shewhart-CUSUM control

chart for improved quality control in Clinical Chemistry.
CLIN. CHEM.23/10, 1881–1887 (1977)

6. Davies, O.L., Goldsmith, P.L., Statistical Methods in Research

and Production, Longman, New York (1984)
NOTES
For more information on ECS products and the Quality Control Program:
UNITED STATES
Bio-Rad Laboratories,
4000 Alfred Nobel Drive
Hercules, CA 94547
To Order Phone 1-800-224-6723
For Technical Service 800-854-6737 Phone 33-14-960-6834
Telefax 510-741-5824
AUSTRALIA
Bio-Rad Laboratories Pty. Ltd.
P.O. Box 210
Regents Park NSW
Australia 2143
Phone 61-2-9914-2889
Telefax 61 2-9914-2888
AUSTRIA
Bio-Rad Laboratories GmbH
Auhofstrasse 78D
A-1130 Vienna
Phone 43-1-877-8901
Telefax 43-1-876-5629
BELGIUM
Bio-Rad S.A.-N.V
Begoniastraat 5
B-9810 Nazareth
Phone 32-9-385-5511
Telefax 32-9-385-6554
CANADA
Bio-Rad Laboratories, Ltd.
5671 McAdam Road
Mississauga, Ontario L4Z 1N9
Phone 1-905-712-2771
Telefax 1-905-712-2990
CHINA
Beijing Bio-Rad Analytical Ltd.
14 Zhi Chun Road
Hai Dian District
Beijing 100088
Phone 86-10-205-1850
Telefax 86-10-205-1876
DENMARK
Bio-Rad Laboratories
Symbion Science Park
Fruebjergvej 3
DK-2100 Copenhagen
Phone 45-39-17-99-47
Telefax 45-39-27-16-98
FINLAND
Bio-Rad Laboratories
Pikatoma 1A
SF-02240 Espoo
Phone 358-0-804-22-00
Telefax 358-0-804-11-10
BI O- R AD LABO RATO RI ES
FRANCE
Bio-Rad S.A.
94/96 rue Victor Hugo
BP 220
94203 Ivry-Sur-Seine-Cedex
Telefax 33-14-671-2467
GERMANY
Bio-Rad Laboratories GmbH
Heidemannstrasse 164
D-80939 Munich
Phone 49-89-318840
Telefax 49-89-318-84100
HOLLAND
Bio-Rad Laboratories B.V.
Fokkerstraat 10
3905 KV Veenendaal
Phone 31-318-540666
Telefax 31-318-542216
HONG KONG
Bio-Rad Pacific Ltd.
Unit 1111
11F New Kowloon Plaza
38, Tai Kok Tsui Rd.
Tai Kok Tsui, Kowloon
Phone 852-2789-3300
Telefax 852-2789-1257
INDIA
Bio-Rad Private
C-248 Defence Colony
New Delhi 110024
Phone 91-11-461-0103
Telefax 91-11-461-0765
ITALY
Bio-Rad Laboratories, S.r.l.
Via Cellini 18A
20090 Segrate, Milan
Phone 39-2-216-091
Telefax 39-2-21609-399
JAPAN
Nippon Bio-Rad Laboratories K.K.
7-18, Higashi-Nippori 5-Chome
Arakawa-ku, Tokyo 116
Phone 81-3-5811-6290
Telefax 81-3-5811-6262
KOREA
Bio-Rad Laboratories
6F Hang Woo Building
1342-3 Secho-Dong
Secho-Ku, Seoul
Phone 82-2-508-4460-2
Telefax 82-2-538-7098
NEW ZEALAND
Bio-Rad New Zealand
Unit 15 Poland Court
21 Poland Road
Glenfield 10 Auckland
Phone 64-9-443-3099
Telefax 64-9-443-3097
SINGAPORE
Bio-Rad Laboratories Ltd.
221 Henderson Road
No. 5-19 Henderson Building
Singapore 0315
Phone 65-272-9877
Telefax 65-273-4835
SPAIN
Bio-Rad Laboratories S.A.
Avda Valdelaparra 3
Poligono Ind. de Alcobendas
E-28100 Alcobendas, Madrid
Phone 34-1-661-7085
Telefax 34-1-661-9698
SWEDEN
Bio-Rad Laboratories A.B.
Gardsvagen 7D
S 171 24 Solna
Phone 46-8-735-8300
Telefax 46-8-735-5460
SWITZERLAND
Bio-Rad Laboratories AG
Kanalstrasse 17
CH-8152 Glattbrugg
Phone 41-1-809-55-55
Telefax 41-1-809-55-00
UNITED KINGDOM
Bio-Rad Laboratories Ltd.
Bio-Rad House, Maylands Ave.
Hemel Hempstead, Herts HP2 7TD
Phone 44-1442-232552
Telefax 44-1442-259118