Worst method write-ups

[DR]

Interesting... We've gone the other way on the assumption that the method was validated with the recipe and that the name was what was likely to be less carefully considered on review.

[Kenn]

...when you work in a CRO and have to decide whether to prepare solutions from customer provided methods as written, or in a manner that will provide what the header says you're making...

In my line of work customer supplied methods are a joke 90% of the time and rarely can be followed. I don't know how many times I have had to explain to a client why his method made to quantitate his compound at mg/mL concentrations in water will not work in something like saltwater or pond scum if you want an LOQ of 50 ppt. It is truly to the point where I glance at them to see what the mass is I should be looking for when infusing the compound and the mobile phases and gradient just so I know how polar it's going to be when I go to chromatograph it. The rest is typically not applicable.

I do like the beautiful chromatography they present that typically has no labeled concentration. It's real easy to chromatograph something at 100 ppm in reagent grade water, now quantitate it in saltwater dosed at 100 ppt. It's usually "I'll look at the same mass transitions that are in your method but will probably change the sample solvent composition, the isocratic run will be changed to a gradient in which case the mobile phase solvent and column will probably change too. Other than that I'll use your method".

[Kenn]

I was looking at one the other day that went into great detail on the instrument and conditions and how they prepared their stocks and standards. Then they proceeded to say samples were extracted with a combination of methanol:water with aliquots from sample extracts cleaned up using Oasis HLB columns (6mL 200 mg) before LC/MS/MS analysis.

Excuse me?

I guess they didn't think the "combination of methanol:water" they loaded on that column or what they used to elute it was important?

This had to be a case of a scientific report writer somewhere who didn't understand the data and decided to just shorten things up a little. I guess you can't say she was wrong.

[DR]

My current favorite is the liquid pharma product method whose authors insist on including a specific gravity factor that renders wrong units for the calculation. The best part is the huge number of approving signatures of people who have approved of the method, the validation, the transfer, and the revisions that followed. None of them seem capable of checking a calculation, just to see if it actually comes out mg/mL or % of label claim.

[Consumer Products Guy]

When I was a young pup, a co-worker's procedure (PhD yet) had us extract a bar of soap, filter, then dilute with solvent. Her analyte peak eluted on the shoulder of the solvent and some issues with the HP 5830 integration, packed column days. So I simply eliminated the solvent dilution step, and injected less, then the analyte peak eluted where the baseline was now flat.

[tlahren]

I see a lack of information often in non-analytical journals. I usually have to contact the author for more information. Where I'm at, the chemists/analysts are usually asked by the primary author to write up the methods/materials section while they focus on the "story" of the results. After they parse our section to make it "journal" sized we get to review it again to make sure it still has all the important information and units etc. This process works well.

[James_Ball]

Thinking about units that don't work and unrealistic quantitation levels. We have a procedure that looks for Picrate to analyze Picric Acid or Ammonium Picrate in water around a closed military base. We developed it using LCMSMS and can get 1ppb detection limits with direct injection of the water sample. Another lab was using our method to perform the same analysis and when the customer questioned the results we checked their data. They were using HPLC/UV, their low calibration standard was 2ppm which gave a barely visible peak yet they still claimed the 1ppb detection limit we did. They even had a detection limit study they ran by injecting seven replicates of a 20ppm standard, which statistically did give an MDL of 1ppb. Even though practically 1ppm would have been pushing it to even see a peak.

[MSCHemist]

There were a series of papers on a new GC amino acid derivatization technique with chromatograms (of active frustrating compounds) that looked like a kovats series. I tried several times and could not duplicate the results and am very skeptical of the papers.

[MichaelVW]

It's kind of a stretch to call this a method, but a couple years ago we were shopping for a microwave digestion system. The sales guy brought a demo machine out and showed us how to use it. I asked about safety features, given that we would be microwaving flammable solvents.

"You can use hexane. It isn't flammable."

There was an awkward pause until one of my coworkers mumbled something about, um, yeah, maybe we will double-check on that later... lets move on to the next question.

[DR]

New favorite, just because the same omission is seen in 4 or 5 iterations of the method for related compounds:
There is one named component with a relative response factor. There is no such RRF in the calculation provided! We were left to guess whether to add it to the numerator factors or denominator factors based on the UV spectrum of it vs. the API.

Clearly, nobody is proofing these things, and this is from a big name pharma company.

Scary.

[James_Ball]

New favorite, just because the same omission is seen in 4 or 5 iterations of the method for related compounds:
There is one named component with a relative response factor. There is no such RRF in the calculation provided! We were left to guess whether to add it to the numerator factors or denominator factors based on the UV spectrum of it vs. the API.

Clearly, nobody is proofing these things, and this is from a big name pharma company.

Scary.

That reminds me of when I first began in the environmental lab, the state sent out the list of analytes for the new landfill monitoring. My favorite was that I had to analyze for 1,2-Dibromomethane.

I always listed on the report for that compound DNE instead of ND. The lab manager asked me what DNE stood for, I told him "Does not exist"

[ScientiaPotentiaEst]

Not too long ago I had a conversation with a QAQC manager that went something like this:

QAQC: "Can you tell me how {regression model} works?"

Me: "Yeah, like {this}. At least, I'm fairly sure. The software doesn't actually give details, but I ran some quick calculations and it seems to check out."

QAQC: "Okay, wonderful. EVERYone in this lab is using that model, but you are the first person who has actually been able to explain it to me. The rest just saw that it was an option in the software and used it without thinking about it."

Me: *eyes widen* "Oh, well.... okay."