How
much do you know about your testing lab?
[Published as 'Precisely accurate' in Occupational Health
and Safety Canada magazine- October/November 1999]
by Heather McLeod, Senior Technologist - Microscopy
Lab techs are a special breed. Share a twinkie with one,
and you'll get exactly half. Fifty percent, by weight and
by volume, not 50.1 or 49.9 per cent. Including the filling.
These are very precise people. They don't pour anything
out of a container without reading the label three times.
Lab techs check and recheck calculations. They make diamond
cutters look sloppy.
One drop of water too many can dilute a sample to the
wrong consistency and throw off all the calculations. Let
a stirring rod touch the counter top, and the next time
it's picked up it could contaminate the whole sample. Unthinkable!
Who else do you know who would soak glassware overnight
in nitric acid, rinse it four times in distilled water,
then rinse it again in water treated by reverse osmosis?
Only a lab tech testing for metals.
So why do you need to know this?
Let's say there's a dust in the air giving your staff
sore throats. Or your product is coming off the line with
orange spots. Your effluent has to meet new permit regulations.
Or your health and safety committee makes a recommendation
to test the air in the workplace. All of these mean collecting
samples and sending them to the lab for testing, but you've
never dealt with an analytical laboratory before. Who are
those people in the white coats? What are they doing behind
those laboratory doors? Before you pick up that phone or
start up that air pump, you may want to know the things
that will help you understand the peculiar world of the
lab tech and the testing laboratory.
1. The whole and some of its parts
You ask the lab how much sample they need. They tell you.
But be careful, they say it needs to be representative.
Of what? Of the whole.
There are entire books written on how to take a representative
sample. The simplest way to get one is to make sure what
you are sampling is well mixed. If it's something like
a dirt sample, it's easy. Spread it out and mix it. Draw
in sections from the outside of the pile and scoop them
toward the middle. Go into the middle and scoop toward
the outside. Repeat many times. Once well mixed, take a
sample not just off the top, not just from the bottom,
not from just the inside or the outside, but from the top
all the way down to the bottom, inside to outside. Otherwise,
you may get just the lighter material on the top, or the
denser material on the bottom. The sample you have taken
represents the whole.
When you are sampling from a very large source, such
as an effluent, sample intermittently. Take the same amount
every hour, half hour, or whatever makes sense. The testing
procedure itself may have guidelines on sampling. Ask the
lab doing the testing, or check with the body setting the
regulations. The test results are only as good as the sample
you've taken. Or, to put it another way, if the sample
you submit is not truly representative of the whole, the
test results though accurate for the sample will be
worthless.
2. Size matters and other lab truths
If there's one ongoing request from a lab, it's to send
more sample. There are times when you only have a little
bit of sample, or you only had one opportunity to sample
and took all you could, and they understand that. But
if you have more, send two or three times what they need.
It's not uncommon to have companies that produce 500
tons per day of product not want to part with 28 grams.
That may sound like a lot, but it's only one ounce. Surely
you can spare one ounce? Please?
If it's a water sample, send a lot. What is a lot of
metal by drinking water standards is not a lot by testing
standards. They'll need a lot of water to get a sample
they can measure accurately.
There are reasons why the lab may need more. Some methods
require results reported on an oven-dry weight basis. This
means the lab must use a portion of the sample to determine
the weight after oven drying to remove moisture. That 10-gram
sample you sent may only be six grams after oven drying.
Lab techs are only human, and occasionally a sample gets
spilled. (Of course it's never happened to anyone I know.)
You may need to run additional tests. All indicators
may point to the sample containing metals. Through analysis,
the lab finds no significant metal contamination, so they
dig a little deeper and come up with another potential
contaminant. If you've shipped extra, they can go ahead
with the analysis. And it didn't cost any more to send
25 grams instead of 10 grams.
The lab may wish to run a duplicate. This will give you
and the lab peace of mind should the results be higher
or lower than you expected. Often times lab personnel will
do a duplicate without your knowledge, as part of their
ongoing quality assurance program. Occasionally, they have
been known to do a repeat test on your sample when the
results are out of line with previous work they have done
for you. They anticipate your questions.
And if you think that the result is unusual, please ask.
You're not insulting them. Lab managers are professionals,
and they don't mind questions. They'll do their best to
explain their confidence in the result.
3. Precisely accurate
Precision and accuracy. Like the Baldwin brothers, you
know they're different but you can't remember which is
which.
Accuracy is how close you are to the true result. If
you take a sample and spike it with 5.25 milligrams per
gram (mg/g) of lead, how close the result comes to 5.25
mg/g is the accuracy of the test.
Precision is how small an increment you can read. Is
your result 5 mg/g or 5.2, or 5.25? Precision is a big
thing in lab testing. Two things limit precision. Sample
size, as mentioned above, and the sensitivity of the testing
equipment.
Think of the precision of the testing equipment as a
timepiece. With a sundial, you can estimate time to the
nearest five or 10 minutes. The clock on your VCR is probably
accurate within a minute or two. A quartz wristwatch can
tell the time to the nearest second. Stopwatches can get
you down to fractions of a second. As technology advances,
instruments get more and more precise.
If the lowest an instrument measuring oil and grease
can detect is two parts per million (ppm), and your sample
is one ppm, there is no way of knowing that exact amount.
The instrument will read it as < 2 ppm oil and grease.
It could be 1.5 ppm, 0.5 ppm, or none at all. But, with
the precision of that instrument, this is all they can
tell you. That is why lab reports never say there is "none",
only less than the detection limit.
4. What the blank do you need to run that for?
Blanks and standards confuse a lot of people. You thought
all you had to do was take a sample, but now the lab
wants a "blank" or a "standard" sample. Why?
When you take in four dirty shirts to the cleaners, do
they make you take in a clean one for reference? No. But
the lab may ask you for a "clean" filter paper when you
bring in a filter you have used to take an air sample.
There are a few reasons for this.
The blank filter can be used as a measure of the whole
process. Has your supplier given you "clean" filters? Have
you stored them properly to avoid contamination? Have they
been contaminated on the trip to and from the sampling
site? Your blank will tell you this. Imagine you're testing
the air for fibres, asbestos fibres perhaps, in an area
where asbestos is being removed. Now, just imagine that
there is a source of contamination where you are unpacking
your equipment. An unexpected source of fibres, unrelated
to asbestos. Before you even sample, your filter is filled
with fibres. If you sent this sample to the lab, your results
would be sky high, not reflecting the true hazard. Now,
send a blank filter to the lab, and they'll see the same
high results there. Using this knowledge, they can work
on removing the source of contamination, and getting accurate
results. The lab needs a blank to calibrate their instruments.
The blank sets the baseline, telling the instrument where
zero is.
Standards serve a different purpose. A standard is a
sample of the material you are trying to quantify. If you
are testing for something common like zinc, you won't need
to supply a standard. The lab will have one. But, if you
are testing for contamination from oil X lubricating your
machinery, the lab does not have oil X on hand. Oil X is
probably a blend of different materials, oil A, oil B,
binder Y and whatever else those brilliant oil chemists
have mixed together. The testing instrument will have been
given measured does of oil X in a range of dilutions. The
lab may also need a sample of your original uncontaminated
product to further isolate your contaminant.
5. They're good, but they're not that good
You've got dust. You've got complaining workers. Get this
tested, your boss says, see if it's bad. So you call
the lab. Can you tell me if this is harmful? And before
you get started, can you tell me how much this will cost?
Unfortunately, there is no machine that tests for toxic
materials. The lab needs to have an idea of what they're
looking for. That's why you may be asked a whole string
of questions. The lab manager is not being nosy, he's narrowing
down the list of suspects. If he asks about other work
being done in the area, he's looking for clues, not company
secrets. Part of being a testing lab is a strict code of
confidentiality.
And when you ask when the results will be ready, please
be aware that sometimes five days means five days. That
is how long it will physically take for the tests to be
completed. For example, a five-day BOD (biochemical oxygen
demand) measures the amount of dissolved oxygen in your
effluent consumed by micro-organisms in five days.
Chances are your lab doesn't work weekends unless you've
arranged it in advance. If timing is critical, ask. But
be warned, your lab techs may get cranky if you submit
a sample dated February 10 on July 28 and ask for rush
service!
6. It's hard to tell you what isn't there
Logic may tell you that if there's no contaminant X in
your sample, analysis should be easy. After all, what
do they have to do? Just tell you it's not there. They
don't have to isolate, identify, or quantify. Easy, right?
Not always.
Imagine it this way. You're looking for your friend Mr.
X at a crowded party. If you spot him right away, it's
easy. He's there. If you don't, you have to go room to
room and confirm that everyone you see isn't Mr. X. Only
then can you say for certain, "This party contains no Mr.
X". Some analyses are like that, and a good technician
keeps looking until he or she is sure.
7. SOS, save our shipper
Shipper/receiver is one of those under-appreciated jobs.
Every specimen is identified, counted, logged, preserved
and sent to the correct department for analysis. Opening
packages all day, it's a little like Christmas and a
lot like Russian roulette.
Scanning the pile of boxes and envelopes, the shipper
can tell from the courier labels what company the package
is from. Some days that's invaluable. Some days the box
contains a few samples and nothing more. Now, we know you
are the lab's favourite customer, and they love hearing
from you, but their system gets clogged up with all these
other customers. So when they don't know which of the 500
people at company X has sent this sample, or what test
they want, all the shipper can do is get out the Ouija
board and hope for help from beyond.
I don't want to nag, but packing is important. If you're
sending a dust, please put it in a container, then into
an envelope. The shipper's just not expecting to open an
envelope with a cloud of dust and a note saying "please
test for asbestos". It makes her nervous. So do wet sample
bottles put into a cardboard box. The box gets wet, a bottle
falls through, the bottle breaks. This may be an effluent
that you are required to test weekly. If the sample's gone,
it's gone. And you have no results.
I know you heard this already from your third grade teacher,
but write the sample description clearly. Otherwise, you
may get a report on a sample named "flaw cream" instead
of "flow stream". Oh, and permanent ink is ideal. It's
hard to read the sample description backward and smudged
from the palm of your hand.
When shipping dangerous goods you need, well, a dangerous
goods label. You are required by law to fill it out yourself.
The lab shipper would love to help you by telling you what
the pH of your sample is over the phone, but she can't.
Check it out with the Department of Transport. There's
a short course you can take. Your boss will give you the
day off, and I think they serve you lunch.
And remember, if you jam the sample into a small-necked
container using a screwdriver, your lab tech will need
a coat hanger and a stick of dynamite to get it out. (Plus,
the results you end up with may be for the screwdriver,
not the sample.)
8. Talk to us. Please.
Lab work doesn't begin and end with testing. You can call
any time for help suggesting and find procedures, and
recommendations for other facilities if the test is outside
the lab's expertise.
The lab manager will give you an idea of when your results
will be ready, so you can plan your next move. Sometimes
samples need to be shipped immediately after sampling.
He'll warn you.
If you have an in-house procedure, let them know. They
can adapt to your needs. If you have your own testing lab,
an outside lab can provide cross-checks for your own quality
control.
You may need special containers to sample into, or specially
prepared containers. A lot of the time, the lab can provide
you with these. Just ask! If you're sampling for oil and
grease, not only is a clean glass container essential,
but a little nitric acid in the bottom will ensure the
sample is at the required pH.
And don't think you can use those plastic containers
you've got sitting around to ship your chemical oxygen
demand test sample. The plastic will breathe and affect
the amount of oxygen in the samples. Your results will
be meaningless.
You may not realize how much time your lab spends ensuring
quality results. An average of 12 to 15 per cent of the
lab technician's time is spent on duplicate testing, or
other quality assurance work. Some testing procedures require
two quality assurance samples for every one sample tested!
Some labs check their results in a round-robin sample
exchange with other labs, often the only option for non-environmental
testing. Or, a lab may keep a large supply of sample on
hand to run monthly checks, to make sure results are consistent.
9. Confidentially yours
Remember, your lab is impartial. They don't make the regulations,
and they don't enforce them.
If your results indicate a problem, they are there to
let you know. They don't tell anyone else, it's all strictly
confidential. Think of your lab contact as a psychiatrist.
Sit back, tell them your problems. You'll feel much better.
10. Worse than a blind date?
The phone book's open to "analytical laboratories". Now
what? The biggest ad? The most technical sounding name?
How do you pick a quality lab?
A few things are universal to any profession. How long
have they been in business? What education or experience
do the laboratory managers have? Don't sell experience
short, it's just as important as education. Working your
way through "real world" problems is invaluable.
Ask about the lab's accreditation. How do they check
the accuracy of their results? One respected standard for
analytical laboratories in Canada is CAEAL, the Canadian
Association of Environmental Analytical Laboratories. There
are other standards, such as the ISO, or International
Organization for Standardization. These independent companies
certify quality. They come to the lab, inspect the lab's
procedures, people and performance. The lab receives suggestions,
which are then implemented and the lab is re-examined.
This may be repeated several times until the lab meets
the evaluation standards, giving them the "seal of approval" in
the form of laboratory certification. But, the lab cannot
then rest of its laurels. It must consistently achieve
acceptable results for samples supplied by the certifying
body. The lab is supplied with a chart that identifies
its ranking with other labs in the program.
Of course, there are exceptions to certification and
quality assurance. Some tests are subjective, meaning the
person doing the test is making a judgment based on his
or her experience. An example of this would be unknown
contaminant identification. Your lab person may look at
a small white particulate and know it's starch. She can
show you how she's looked at it under the microscope under
certain light conditions and seen the distinctive "church
windows" that confirm starch. She can't offer you certification
results, only her experience and certainty.
No, most lab techs can't leap tall buildings in a single
bound, but how many superheroes would have the patience
to watch a thermometer creep up one degree at a time?
back to the top
|