MDL rules and regs.

[JLCookael]

What is the best source for the rules and regulation on calculated MDL's. What I am looking for is limits for RSD across the 7 replicates, %recovery relative to spike concentration for each replicate and the average for all replicates, limits for MDL relative to spiked concentration,

I have read 40 cfr 136 app. B Maybe its just me but it is not really clear.

[Balderquell]

It depends on who you ask and which regulatory body you have to answer to. Some methods are give specific accuracy and precision requirements.

If you're environmental, check with your state. If your state is Nelac, check the nelac documents. Refer to the last few pages of the published method.

[JLCookael]

We are environmental. We are Nelac. Nelac says nothing specifically. All methods I am familiar with refer to 40 CFR 136 app B

[Balderquell]

You're in Tyler? I worked in Dallas for several years and I don't remember ever seeing a TCEQ required limit on RSD for MDLs. If you need a limit, your lab may have to control chart it somehow. You might also check the TRRP document.

The RSD will factor into your calculated MDL, however. If the RSD is too high, then you might not get the calculated MDL you need. And that's how we "scored" our MDL studies. We set a target MDL, usually less than half our MQL or lower if a particular client needed lower, and if the results of our MDL study met or exceeded the target MDL, we considered it acceptable.

If your RSD is coming out on the high side, you might see if you can change your MDL spike amount.

What method are you running?

[JLCookael]

I am running 8260/624, 8270/624, 8081/8082/608, tx1005. Thanks for all of the comments. I have come to the conclusion that MDL limits that I mentioned do not exist ( at least formal limits by regulatory body) and limits that I was exposed to in the past were some sort of internal limits.

[gpronger]

The problem with the historic MDL procedure (40CFR Part 136 Appendix B) was that in many ways the calculated MDL did not coincide with reality often significantly over-estimating the method's real sensitivity (the MDL being well below what the lab could actually see).

With your analyses for instance, when you run an MDL, we are quantitating from typically the primary ion, but to have a confirmed detect we need to be able to see the secondary ions. These may be only 10% of the base peak and will very often become no better than background noise at the calculated level of the MDL. Also, the MDL procedure is based upon the assumption that the slope of our calibration passes through the intercept. Now for volatiles most analytes probably are pretty close to this, but with semivolatiles, if you consider the nitrophenols, benzidine, etc., we very often do not have a zero intercept. In this case, we can run the spike, and so long as we have decent reproducibility, we get a low MDL - which is actually between the actual (positive) intercept of the line, and the zero. In other words, we end up with a detection limit that we fundamentally can't see.

Due to these issues, a lot of labs implemented secondary criteria to guard against having a calculated MDL which the lab knows it can't see. Whether it was by adopting recovery criteria, performing some type of spiked verification of the MDL, etc., it is not uncommon to see labs adopting secondary criteria.

Depending upon the State and your customer, they may have required secondary criteria that they would want to see you performing.