Jump to content

  • Quick Navigation
Photo
- - - - -

Huge disparity in lab results-third party labs


  • You cannot start a new topic
  • Please log in to reply
14 replies to this topic

#1 TimG

TimG

    Grade - SIFSQN

  • IFSQN Senior
  • 335 posts
  • 114 thanks
145
Excellent

  • United States
    United States

Posted 04 January 2021 - 09:25 PM

Ok, I want to get some feedback on this one since I'm not as up on the lab side of things. I have a supplier of a product who provides a CoA for that products lot prior to our accepting it. My main concern in this instance is iron. Our supplier shows CoA's in the 3-7ppm typically. We had some iron issues in our final product so one of the things I did was send out this product to a third party for analysis. Of the 17 samples sent, I have an average of 64.88ppm iron (some spiking 110,133,145) which is way different from what the CoA is showing. Contacted the supplier, explained this is a big deal, they pretty much said "no way" and pulled 3 samples, one to be sent to their independent third party lab, to our independent third party lab, and to our in house global R&D team.

 



#2 Charles.C

Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 18,276 posts
  • 5120 thanks
1,124
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 04 January 2021 - 09:49 PM

Ok, I want to get some feedback on this one since I'm not as up on the lab side of things. I have a supplier of a product who provides a CoA for that products lot prior to our accepting it. My main concern in this instance is iron. Our supplier shows CoA's in the 3-7ppm typically. We had some iron issues in our final product so one of the things I did was send out this product to a third party for analysis. Of the 17 samples sent, I have an average of 64.88ppm iron (some spiking 110,133,145) which is way different from what the CoA is showing. Contacted the supplier, explained this is a big deal, they pretty much said "no way" and pulled 3 samples, one to be sent to their independent third party lab, to our independent third party lab, and to our in house global R&D team. So far, from MY third party lab I have an iron result of 62.7ppm , from THEIR third party lab they show an iron of 6ppm.

The methods are different but they are comparable ITSM methods.

 

Has anyone else run into completely off the wall analysis like this? I almost wonder if they are screwing up in their dilution factor at my third party lab and the decimal should be 1 to the left.

 

Hi Tim,

 

A few questions.

 

What is the specific item ?.

 

What is ITSM ? (Google failed).

 

Are all results for iron expressed as ppm of  "Fe" ?

 

What methods are being used ?

 

How do you know if samples being distributed are the "same" ?

 

A typical way to compare labs is to send a well defined "dummy" sample within the samples of specific interest.

 

PS - reason for queries -

 

(1) Some Products demand specific methods for accuracy ? (from bitter experiences).

(2) Different methods can give different results (eg due [1]). The method should ideally be validated. Laboratories sometimes develop short-cut procedures which may fail for atypical samples.

(3) Different samples can obviously give different results.

 

JFI, one common method for measuring ppm levels of Fe in food is atomic absorption (AA). This can readily distinguish between 5 and 50ppm for appropriate samples. However foods typically require "ashing" prior to AA analysis. The former can cause some variations.

 

Nonetheless, for similar samples, validated methods should not give variations of magnitude such as you describe.


Edited by Charles.C, 05 January 2021 - 06:26 AM.
extended

Kind Regards,

 

Charles.C


#3 pHruit

pHruit

    Grade - FIFSQN

  • IFSQN Fellow
  • 1,629 posts
  • 669 thanks
369
Excellent

  • United Kingdom
    United Kingdom
  • Gender:Male
  • Interests:Composing/listening to classical music, electronics, mountain biking, science, sarcasm

Posted 05 January 2021 - 08:37 AM

Charles has asked most of the pertinent questions, but just a few more points that may be worth considering:

 

  • Are both of the independent labs accredited, and if so, are the methods they've used actually included within the scope of accreditation? I appreciate that this doesn't directly offer an explanation for the disparity between the results, but I have seen labs leave some bits outside their scope and had slightly surprising results on occasion, as compared to those who put it in scope and are perhaps more familiar with the method. (Obviously noting that accreditation doesn't always equate to competence... ;) ).

 

  • Have you and your supplier asked the labs to check what they've reported? Yes, their verification and signing process for certificates should catch typos, but again I've seen it happen, so it is very definitely worth asking...

 

  • At risk of sounding very patronising (not intended!), there is definitely no confusion in interpreting the units of measure in which the results are being reported? Again it's one I've seen - just such a mistake is arguably responsible for unintentionally starting my career in the food industry!


#4 TimG

TimG

    Grade - SIFSQN

  • IFSQN Senior
  • 335 posts
  • 114 thanks
145
Excellent

  • United States
    United States

Posted 06 January 2021 - 03:00 PM

Hey guys, sorry for the day late reply but I was going back and forth with the third party lab I use to try to get to the bottom of this. Some updates:

  • My third party lab did review their data to make sure they didn't make a calculation error (although I still feel this is the best explanation)
  • They are going to pull the retain and ship it to my global I&D department so I can actually test the same sample they tested
  • Sulfuric acid (90%) is what's being analyzed
  • Test method by my third party lab is SW-846 6010C and they list iron as mg/L
    • Test method by suppliers third party lab is AFPC XI-14B and they list iron in PPM
  • ​Both labs are accredited according to their websites, to ISO standards

I'm in a catch-22 with this one because we've spent quite a bit of time/resources on this based off what our third party lab results have been showing. So either my third party lab isn't competent enough to run the testing and report results accurately (bad because now it will call a bunch of other testing into question) or our supplier has, hopefully inadvertently, been misrepresenting data for the past few years.

 

The good news is this has very little impact on the food safety of the final product that actually gets shipped. We have a positive release program and have our own global lab test all lots of our final product prior to me signing off on it being ok to ship. It has great impact on price points of materials, carbon packs, unworkable product, etc. The $$ stuff..



#5 Scampi

Scampi

    Fellow

  • IFSQN Fellow
  • 3,338 posts
  • 907 thanks
504
Excellent

  • Canada
    Canada
  • Gender:Not Telling

Posted 06 January 2021 - 03:08 PM

Is the calculation to covert ppm to mg/L or mg/L to ppm correct?   I've seen that calculation incorrect a number of times on CoA's


Please stop referring to me as Sir/sirs


#6 TimG

TimG

    Grade - SIFSQN

  • IFSQN Senior
  • 335 posts
  • 114 thanks
145
Excellent

  • United States
    United States

Posted 06 January 2021 - 03:18 PM

Isn't it a 1/1 conversion? mg/L is ppm, right? Once again, I'm not a lab guy, but I checked a few different sources and they all say similar to the following:

1 mg/L = 1 parts per million (ppm) for dilute aqueous solutions. For example, a chlorine concentration of 1.8 mg/L chlorine is equivalent to 1.8 ppm chlorine.



#7 Scampi

Scampi

    Fellow

  • IFSQN Fellow
  • 3,338 posts
  • 907 thanks
504
Excellent

  • Canada
    Canada
  • Gender:Not Telling

Posted 06 January 2021 - 03:24 PM

When I was in cannabis we had some trouble with the conversions---it's not always that straight forward

 

the 1:1 ratio generally only speaks to aqueous solutions

 

https://www.epa.gov/...me relationship.

 

https://www.research...f_Major_element

 

Ask both labs to report BOTH sets of results and then you can compare apples to apples


Please stop referring to me as Sir/sirs


#8 pHruit

pHruit

    Grade - FIFSQN

  • IFSQN Fellow
  • 1,629 posts
  • 669 thanks
369
Excellent

  • United Kingdom
    United Kingdom
  • Gender:Male
  • Interests:Composing/listening to classical music, electronics, mountain biking, science, sarcasm

Posted 06 January 2021 - 03:24 PM

Isn't it a 1/1 conversion? mg/L is ppm, right? Once again, I'm not a lab guy, but I checked a few different sources and they all say similar to the following:

1 mg/L = 1 parts per million (ppm) for dilute aqueous solutions. For example, a chlorine concentration of 1.8 mg/L chlorine is equivalent to 1.8 ppm chlorine.

Sometimes, maybe ;)

1mg/kg could also be legitimately recorded as 1ppm. The lab will have measured in mg per unit of sample (rather than directly in ppm), unless some software/equipment make the conversion for them.
This is all fine if they've measured mg/L and reported as ppm, but if they've measured as mg/kg then this would only directly equate to mg/L if the density of the liquid sample is precisely 1kg/L.

Like Scampi, I've seen this one cause confusion and errors - taking a "mg/kg" ppm value and assuming it is a "mg/L" ppm value makes a large difference if your product has a density that is either significantly greater than, or significantly less than, 1kg/L.

 

Nonetheless I wouldn't expect this type of potential error to account for the scale of the disparity seen in the two results you have.



#9 TimG

TimG

    Grade - SIFSQN

  • IFSQN Senior
  • 335 posts
  • 114 thanks
145
Excellent

  • United States
    United States

Posted 06 January 2021 - 03:51 PM

I think it is going to end up being a calculation error. I took their 19 results for iron and averaged them, came back with 64.88 . My average expected iron result is 4-8 ppm. Which would mean if they were exactly 1 order of magnitude off on their calculation, they'd be right where I would expect the results to be, 6.49



#10 Scampi

Scampi

    Fellow

  • IFSQN Fellow
  • 3,338 posts
  • 907 thanks
504
Excellent

  • Canada
    Canada
  • Gender:Not Telling

Posted 06 January 2021 - 03:59 PM

That's what this forum is for!!!

 

As I mentioned, have BOTH labs report BOTH values from here on in----that way you don't have to rely on your calculations for final release---you don't need/want that weighing on you---leave it to the lab folks


Please stop referring to me as Sir/sirs


#11 pHruit

pHruit

    Grade - FIFSQN

  • IFSQN Fellow
  • 1,629 posts
  • 669 thanks
369
Excellent

  • United Kingdom
    United Kingdom
  • Gender:Male
  • Interests:Composing/listening to classical music, electronics, mountain biking, science, sarcasm

Posted 06 January 2021 - 04:00 PM

I think it is going to end up being a calculation error. I took their 19 results for iron and averaged them, came back with 64.88 . My average expected iron result is 4-8 ppm. Which would mean if they were exactly 1 order of magnitude off on their calculation, they'd be right where I would expect the results to be, 6.49

It certainly seems plausible, and is easily done. But then I've also seen genuine results come back an order of magnitude out of spec...

It'll be very interesting to see what result your I&D people get!



#12 beautiophile

beautiophile

    Grade - MIFSQN

  • IFSQN Member
  • 171 posts
  • 56 thanks
17
Good

  • Vietnam
    Vietnam
  • Gender:Male

Posted 07 January 2021 - 03:27 AM

The water density is roughly 1kg/L, so that ppm can represent both mg/kg and mg/L unless other solvent is clarified or the concentration is greater than 1000ppm.

There are several statistic methods to assess inter-lab variations such as ANOVA, proficiency test, Youden. Governmental authorities often have to deal with these things for final judgments in case of incidents needing analyses by different labs, e.g. pollution of water réservoir. You can inform your case to both labs. I believe they know the concepts and will proceed against the problems.


Edited by beautiophile, 07 January 2021 - 03:36 AM.


#13 Charles.C

Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 18,276 posts
  • 5120 thanks
1,124
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 07 January 2021 - 05:45 AM

Offhand, this analysis sounds pretty easy.

 

As previous post, most analysts simply switch mg/L, mg/kg,  ppm freely around for aqueous solutions.

 

Do you actually care whether 6 or 60 ppm ? Is this high grade sulphuric acid with a spec. for impurities?

 

1st analysis method looks high tech., 2nd method is Google unknown.

 

Rather hard to believe an accredited lab would make a factor 10 error.

 

Like I said, you need a reference dummy sample.

 

Or a 3rd (4th?) lab :smile:

 

Still wondering what ITSM meant ? (In The Same Manner ?)

 

Are you sure the samples were split from the same original concentrate ?


Kind Regards,

 

Charles.C


#14 Spidey

Spidey

    Grade - MIFSQN

  • IFSQN Member
  • 106 posts
  • 31 thanks
40
Excellent

  • United States
    United States
  • Gender:Female

Posted 07 January 2021 - 04:11 PM

I wonder if it would be worth it for you to purchase a proficiency testing kit, which would have known values, and send that to each lab for testing?

 

Right now, all samples are coming from the suspect supplier and the conflict is between the supplier and your lab.  You could do as Charles suggested and pull another lab into the mix, I had that same idea, but if you continue to get conflicting results, another lab it seems like a waste of time and money.

 

By nature, proficiency testing samples are a predetermined concentration.  Additionally, the companies that make these samples, trend the results of all of the labs who send in results and are required to account for differences between actual concentration and recovered concentration.

 

Alternatively, you could purchase standards of a known concentration and send them to each lab for testing.

 

I believe the biggest issue with the current method being used is that the actual concentration of the samples being tested is unknown.  If you can get a third party sample of a known concentration, I believe that has the potential to clear up a lot of questions.



#15 Charles.C

Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 18,276 posts
  • 5120 thanks
1,124
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 08 January 2021 - 03:14 AM

I wonder if it would be worth it for you to purchase a proficiency testing kit, which would have known values, and send that to each lab for testing?

 

Right now, all samples are coming from the suspect supplier and the conflict is between the supplier and your lab.  You could do as Charles suggested and pull another lab into the mix, I had that same idea, but if you continue to get conflicting results, another lab it seems like a waste of time and money.

 

By nature, proficiency testing samples are a predetermined concentration.  Additionally, the companies that make these samples, trend the results of all of the labs who send in results and are required to account for differences between actual concentration and recovered concentration.

 

Alternatively, you could purchase standards of a known concentration and send them to each lab for testing.

 

I believe the biggest issue with the current method being used is that the actual concentration of the samples being tested is unknown.  If you can get a third party sample of a known concentration, I believe that has the potential to clear up a lot of questions.

 

Hi Spidey,

 

Thks input.

 

Indeed, IMEX of AA, standard solutions of elements like Fe are readily available, eg -

 

https://www.merckmil...www.google.com/

 

@TimG, JFI I wondered as to whether the samples of sulphuric acid sent out were pre-diluted (conc.H2SO4 is quite nasty stuff) ? Could be another potential source of error.


Kind Regards,

 

Charles.C





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users