I don’t know if anybody has come across this issue: When defining a column as ‘Carry’ in an NCA setup (Phoenix 6.1 or 6.2) and the column has missing values, it is actually not carried over in the NCA results worksheets, i.e. simply dropped. This was not the case in WNL 5.x. Other types of columns which are not carried over are probably those with very large numbers. As a source dataset, I had a SAS transport file with a numeric variable which had a DATETIME format. The SAS viewer shows this as DATE:TIME (e.g. 01JUN2011:16:24:34) but in Phoenix it shows as a very large number (e.g. 1.673957E+09). When defining this variable as Carry, it is dropped in the NCA results even if I define it as a Text variable.
Hello, This behaviour existed in WNL 5.3 and it is due to carry alongs that are time dependent (i.e. including missings). The program checks to see if the column has the same value for each time point and if it does (time independent) then it is put in the final parameters worksheet as only one entry is allowed there. If the column is time dependent then it is still there but only in the Summary worksheet as it changes over time. If the values change during the profile there was no good way to determine which one was correct and thus this was reported and fixed. If you need to know specifics on when this was implemented, I can dig around a little further and provide you more details. Hope this helps, Ana
It seems to me that WNL 5.2 did not show this behaviour, so you may have changed that in 5.3. Thanks, now I understand the logic. So a workaround would be to either use these variables as Sort variables or to complement (fill the missing cells) the data before doing an NCA if the value is always the same.
Your workaround is correct.
Trying our this further (i.e. taking these variables as sorting criteria), I realized that the option ‘Exclude profiles with Insufficient data’ in the NCA setup does not work properly in Version 6.1, i.e. it doesn’t really exclude these profiles in the NCA results. This seems to be fixed in V6.2.
There was an issue in 6.1 fixed in 6.2 as indicated by ‘Issues Corrected’ in the Release Notes. “Excluding subjects with insufficient data kept one record in output (QC 8745): Excluding subjects with insufficient data was still keeping one record in output. This has been fixed in Phoenix 1.2.”
Hi Ana,
I am using Phoenix WinNonlin 6.3. Can the ‘Exclude Profiles with insufficient data’ option in NCA be customized? for example, exclude profiles that have less than 4 measurable concentrations time points.
Lisa
Dear Liza,
I’m not Ana, but let me help you
Even in latest Phoenix 7 there is no way for customizing this option, sorry.
All rules are explicitly described in WNL User’s Guide.
I suggest to use some workaround with Data Wizard and Enumerate tool.
Thank you.
Lisa
Lisa,
the way I would do it is;
-
run a descriptive stats object on your data first, to get the number of data points per profile.
-
Then you need to merge the N from result sheet from descr. stats with your dataset.
-
Filter using the Data Wizard object where you filter to remove profiles with less than 4 data points.
Let me know if this is not clear.
Simon.