We´ve got a rather strange recovery problem; during validation we always find in the 70% range about 103%, in the 130 % range about 95% of the theoretical content. We have tested two different matrices (one is lactose based hard gelatine capsule the other is an propandiol-water-mixture) and the phenomenon occured with both.
Has anybody an idea what the reason could be ??

Hello Friend Kirtin;

1- Your problem is relative to matrice effect or a degradation of your sample.

2- Do you have good resolution between yours peaks?

Good Resolution should be better that 1.5.

3- Wath is your sample concentration?

Good look my Friend

Kristin,
How are you making a calibration curve? Is it a single point forced through zero? It sounds like your slope on the cal. line is off. Your true y-intercept may not be close enough to zero to use this method. If this is the case I would re-make a range of std's covering the entire recovery range and not force through zero. Hope this helps

Daren

Sorry about spelling your name wrong, just saw it.
-Daren

Kirstin, deST is hinting that you may be operating beyond the linear range of your detector--whatever it is. UV, Antek N, LC/MS, what?

Build a 5 point calibration curve 50-150% of your expected range (50, 75, 100, 125, 150%) and report back.

Working for Merck, you may have a very interesting problem to work on, not something simple.

If values are produced based on a one-point calibration it really looks like nonlinear behaviour of the System. A linearity test could show that a) the calibration curve is not linear or b) the calibration line has an offset.
If you are not working in the linear region of the detector decrease your concentration.
If you are working in the linear region of the detector (assuming that it is UV) then look at the UV spectra at different concentrations: Are they all the same or do they change?
Keep us posted!