Should the variability (e.g., RSD < 40%) from recovery experiments be used for the acceptability of a method? (please explain)

Answer Explanations

• Yes

  Expert 9
  Yes, an assessment of both the accuracy and the reproducibility of a method is necessary to evaluate the reliability of results produced.
• Yes

  Expert 1
  Yes, many regulatory methods allow for a wider range of variability.  However, this should be based on study objectives and the variability should also be considered while interpretating the data.  So, it is not a matter of allowing 40%RSD, but it is about the purpose/goal of the study and how the data will be used and interpreted, which dictates allowable levels of variability.  While image based methods can have high variability, mass spec-based methods can have tighter RSD of <20%.    I suggest that this is explained in the text, especially after line 137.  
• Yes

  Expert 3
  Not only will it confirm consistent performance and support the validation of the method, it also reflects the precision of the method. 
• No

  Expert 2
  Even though some recovery experiments are acceptable when the average recovery is reasonable and consistent across concentration levels a large variability in recovery for example RSD > 20–30% might suggest matrix effects or unstable extraction and in my opinion should be not considered by itself, an official acceptance criterion. 
  
  
• No

  Expert 7
  First, recovery variability (RSD) is just one factor to be considered for acceptability of a method. There are also other factors to be considered, accuracy and precision across multiple concentrations, linearity of the method, robustness of the method, stability under different conditions of the method, variability of the method between different analysts and instruments, and so on. 
  
  Secondly, RSD should be explained in context. For regulated or critical applications, a lower RSD is expected, while for certain low-risk or exploratory settings, a higher RSD should be acceptable.

  
  
  
• Yes

  Expert 4
  I think that RSd can indeed be of the criteria for acceptability of a method. If the RSD is high or even very high, then this sheds doubts on the usefulness and reliability of the outcomes of an analysis. However, I think that in such a case the method may still be suited for obtaining semi-quantitative information that could for instance be useful for further analytical efforts.
• No

  Expert 5
  Too general of a question.  Is this assuming recovery of analyte standards from a 'standard' matrix? 
  Is the method specific to only one analyte/matrix pair? Or is it intended to be generalized to 'biological tissue'? If so, how many of those analytes/matrix combinations have been tested?
  How many analysis were completed to generate the RSD?
  Is this the RSD for one set of experiments done once by one analyst in one laboratory? Or is it from an interlaboratory study?
  RSD is a good 'rule of thumb' measure, but it is no substitute for full method validation.
  Personally, I would never deem a RSD<40% result as an acceptable methodology, more of a guideline as to whether this method may be a useful starting point.
• No

  Expert 6
  When we talk about recovery in analytical method, typically acceptability is around 80%-120%. For biological tissue, I personally think 60%-140% is minimum acceptability. Outside of this range, it suggests the matrices introduce either positive or negative background effect to the analyte. I think this recovery range is more critical than RSD. We could have a consistent matrices effect that the recovery is about 20%, but the repeatability (RSD<40%) is good with 18%,19%,20%,21% and 22%. In this case (low recovery, small RSD), the method is still defective.
• No

  Expert 8
  Not in all cases, but yes in some cases.  Everything is context dependent.  The quality control strictness may be different between a student's research project and an analysis of samples for regulatory reasons by an accredited lab.