Ma. Teresa M.Mercado ● 2005-09-13 11:30 (6971 d 02:40 ago) Posting: # 31 Views: 19,773 |
|
May I know the conduct of BA/BE in your labs. Are all subjects done in 1 instrument (HPLC) by 1 researcher or in 3 different HPLCs by 3 diff researchers. |
Helmut ★★★ Vienna, Austria, 2005-09-19 18:18 (6964 d 19:53 ago) @ Ma. Teresa M.Mercado Posting: # 32 Views: 14,578 |
|
Hi Teresa! Samples from subjects should be analyzed in a cross-over manner, i.e., sample1/period1–sample1/period2, sample2/period1–sample2/period2, etc., thus avoiding sample carry-over – which could be a major issue if analyzing samples at random. If possible (no time pressure), samples from the entire should be run by the same analyst on one machine. We once got problems during an inspection (analyses were performed on three identical machines), and the inspectors wanted us to show that the results from the three instruments were ‘interchangeable’, which is clearly nonsense (since a complete set of calibrators and QCs was run in all batches on each instrument), but we had to show it anyhow. If you have to run the analytical part faster, I would suggest to show ‘no differences between machines’, i.e., run one complete set of calibrators and QCs on each machine and demonstrate statistically no difference (e.g., paired t-test: equality of intercept and slope of regression line, or orthogonal regression [also known as unweighted Deming regression] of responses from both machines for intercept=0 and slope=1). If you plan to include more than one machine in actual runs, it would be wise to include such a demonstration as a part of your pre-study validation. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |