Hello, When a data set contains Missing data at some time points, How do we handle such values in PHX WNL, earlier in WNL 5.2 we had the option of “To set or change default workbook options” where in “To specify a character or string to represent missing data, enter it in the box labeled Missing Value. Any cell containing this string will be grayed and treated as missing in (excluded from) modeling and analyses” Now in PHX do we have such kind of similar feature, if so Please specify how to do so Regards Kaushik
Kaushik, Phoenix has no explicit missing data value. If a numeric cell is empty then it is treated as missing. If you perform numerical operations on a column containing both numbers and text then those cells which are not numbers are all treated as missing. The advantage of this that I see is you can simply use your data as it came form the lab, NS, BLQ, <10, all of these will be treated as missing in graphs, NCA, Desc. Stats etc. but if listed in a table the original value is still reported. Simon.
Hi Kaushik, agree with Simon; always format concentrations as text. Now for the tricky part: Phoenix interpolates between numeric values in NCA – which is terrible if you use the linear trapezoidal. Example: IV, C0 100, alpha ln(2), sampling at 0,1,2,4,8,12,16. Intentionally I forced PHX to use the last three data points for the estimation of lambda-z: Theoretical AUCinf = C0/alpha = 144.2695. lin/lin AUClast 157.91321 AUCinf 157.91541 (bias +9.46%) lin/log AUClast 144.26730 AUCinf 144.26950 (bias 0.00%) Now I replaced the 4 h value with “Missing” lin/lin AUClast 189.553831 AUCinf 189.55603 (bias +31.39%) lin/log AUClast 144.26730 AUCinf 144.26950 (bias 0.00%) As you can see only lin/log gives an unbiased AUC in case of missing values. Never, never use the linear trapezoidal! If you have extravascular data use lin up/ log down.
Hi Simon & Helmut Thanks for the response, I proceeded with your suggestion and i was able to complete the analysis If we have any profile with all ZERO data points and for Lambda_z calculation should we use Best fit method, what if i use Time range option it gives a msg that error in time Can you also help me understand what would be the best method in that case Regards Kaushik
Hi Kaushik, the logarithm of zero is undefined (valid only for â„[sub]+[/sub]). Try it with your pocket calculator. What do you want to achieve? A profile without any suitable data is noninfomative.
Thanks Helmut - you beat me to the response on these two ;0) Kaushik Remember also that, WNL will not compute kel for increasing concentrations (nor should it - it’s and elimination constant) so if you find your profile is ‘flat’ with non-zero values then you may also get this warning “*** Warning 14530: Lambda_z could not be estimated. No parameters could be extrapolated to infinity. See Core Output for more details” This may indicate you have not sampled for sufficiently long time to assess Lz and Thalf Simon.