In the case of working with Healthcare Claims Databases, patients (Member_ID) with a certain disease (ICD10_Level1_code) need to be excluded.
How can I use JMP to exclude any “patient” that has had a certain disease at least once within the data?
Thank you for your attention.
Someone will probably be able to give you a way to do this via scripting, but until then, this is a somewhat roundabout way to accomplish this. You could use Analyze-Tabulate putting ID in the rows and the disease code in the columns. Make that into a data table. Select all the rows (IDs) that have a 1 in the disease code column you want and then subset those rows (no columns needed). Then you can join the subset list of rows to the original data set using the IDs (including the non-matches, but don't exclude multiples). Then create a column with an IF statement that if the IDs match (between the original ID column and the one from the subset) then =1, otherwise 0. Then you can select the 1s and exclude them.
Continuing with the interactive solutions (not scripted), find the disease code in one of the rows, right-click it, and choose Select Matching Cells. Then select Rows > Exclude. Alternatively, you could select Rows > Row Selection > Select Where and define the search target here. Then exclude the select rows as before. Finally, you could select Rows > Data Filter. Add the column with the disease code. Be sure to check the Include mode and the Inverse option. Find the code and select it in the list.
These alternatives all affect the row state. All the platforms pay attention to the row state and respond to changes in the row state. This allows you to perform an analysis or plot data and quickly change from the entire data set to any subset and immediately see the result.
I don't think this will accomplish what is being asked for - it is easy to exclude the rows with that disease code, but the question was how to exclude all the patient IDs that have that disease code and there are multiple observations for each patient ID. So, a patient may have one observation with that particular disease code and a bunch of other observations with other disease codes. Then, it is a bit more complicated to exclude the patients with that disease code rather than just the observations with that disease code.
Run the code below to make an example table, and select a specific condition. Then click on the heading of the 'Patients' column, and use 'Row > Row Selection > Select Matching Cells'.
NamesDefaultToHere(1); // Make some data dt = NewTable("Data", NewColumn("Patients", Numeric, Nominal, Formula(RandomInteger(1, 25))), NewColumn("Condition", Numeric, Nominal, Formula(Random Integer(1, 10))), AddRows(100) ); dt << runFormulas; // Use a Distribution to see what happening dt << Distribution( Nominal Distribution( Column( :Patients ) ), Nominal Distribution( Column( :Condition ) ), Histograms Only ); // Select the required condition(s) Wait(2); dt << selectWhere(:Condition == 1);
There are no labels assigned to this post.