Precision

[kristind858]

Hello,
I just recently visited a lab that does not perform duplicate sample preps in order to determine their lab precision. They only use one sample injection and run an external standard. The lab is a contract lab and of course wants the highest amount of throughput as possible. I thought that GLP requires duplicates and even triplicates unless sample limitation is a factor, which it is not in our case. The lab in question does participate in proficiency testing which is excellent but they can't use this as an excuse to only perform one sample prep, correct? What if they are only extra cautious with their proficiency samples but rush though all other customer samples? When can they determine if there was a systematic or random error? We would never know if any error occurred and the results they send to us would be based on 1 sample prep. Do most contract labs only do 1 sample prep?

[Peter Apps]

It depends.

A method with an internal standard added at the start would usually be robust enough for single samples, especially if additions and aliquoting are tracked along the way.

With QC analyses of product from a reliable process the rate of out of specs is very low, so it may be more cost efficient to rework out of specs which are due to analytical error, than to duplicate every sample, which in effect is a 100% rework rate. The problem with this approach is that an analytical error might make on out of spec sample look as if it was in spec.

I am absolutely sure that the contract lab would be happy to run as many replicates as you wanted - as long as you are prepared to pay the higher cost.

Peter