Hi Paige, thanks very much for your comments and suggestions! I will pay attention later on! ... View more

It's okay, @PaigeMiller , you are welcome! ... View more

Hi @PaigeMiller I agree with you that double ampersand is not needed here, please see details in my other threads, one ampersand can resolve the macro. However, I am curious about how @nsns use double ampersand to resolve the macro (which obviously resolved twice) and what does the macro look like 😀 ... View more

You do not always need a period (a dot) to resolve a macro, except the macro value only constitutes part of your string then you need a dot to separate the two parts. For examples (I typed code here and run the code in windowing environment 9.4 base, btw😀 and this does not make any difference just somehow want to mention it btw😀) %let var=employee; %put Employees resolves to &var.s as it is.; %let lib=weight; %let table=itt; %put The dataset weight.itt resolves to &lib..&table as it is.;       ... View more

And also I forgot to write the proc print step in the second macro in my previous thread, here is the correction: %macro popset(table,var); %local table var; data &table._&var; set &table; if &var='y'; run; proc print data=&table._&var; run; %mend popset; %popset(weight,itt); %popset(weight,pp); ... View more

Please discard my first %let macro in my last thread (the way I create that %let macro was odd and funny, I don't know what's wrong with my mind then 😀). Here is the right one (and you do not need to use one more ampersand to resolve the macro): data weight; input subjid weight itt $ pp $; datalines; 101 50 y y 102 55 y y 103 58 y y 104 60 n y 105 66 n y 106 67 y n 107 68 y n 108 70 y n 109 71 y y 110 72 y y 111 80 n y 112 49 y n 112 48 y n 114 49 y y 115 45 y y ; run; proc print data=weight; run; /**create %let macro to print dataset for different population**/ %let pop=itt; data weight_&pop; set weight; if &pop="y"; run; proc print data=weight_&pop; run; %let pop=pp; data weight_&pop; set weight; if &pop="y"; run; proc print data=weight_&pop; run; ... View more

Thanks for the suggestion Paige! I did not use %local statement in a macro before and do not know whether I use it correctly, in this post I think you can help with my question so I referred to you in this thread. Sorry for it if it looks irrelevant! ... View more

Hi, so you solved the problem yourself through adding one more ampersand & before the macro &pop.? And then it is resolved to a value like 'ITT' and 'PP', isn't it 😀 , so this macro need to be resolved twice (i.e., double ampersand && should be applied) into the correct value, and the macro you are using is a big complex nested macro which I cannot imagine what does it look like 😀 ... View more

 @nsns I have another question for you: why there is a period (or dot) behind your macro variable &pop (,please see the picture, I marked it out with a red circle)? @PaigeMiller , I am not sure whether the %local statement in my second macro write correctly, could you help confirm it? Thanks! ... View more

Hi nsns, thank you for telling me what 'PP' stands for (I don't know what is per protocol analysis population though😀 ), I create a dataset like this and a macro according to what you described, hope this explain part of your problem. Please let me know if it does not solve your question😀. /*create dataset*/ data weight; input subjid weight itt $ pp $; datalines; 101 50 y y 102 55 y y 103 58 y y 104 60 n y 105 66 n y 106 67 y n 107 68 y n 108 70 y n 109 71 y y 110 72 y y 111 80 n y 112 49 y n 112 48 y n 114 49 y y 115 45 y y ; run; proc print data=weight; run; /*create %let macro to print dataset for different population*/ %let flag=y; data weight_itt; set weight; if itt="&flag"; run; proc print data=weight_itt; run; data weight_pp; set weight; if pp="&flag"; run; proc print data=weight_pp; run; /*another way to create this macro*/ %macro popset(table,var); %local table var; data &table._&var; set &table; if &var='y'; run; %mend popset; %popset(weight,itt); %popset(weight,pp);   ... View more

'ITT' stands for 'Intend To Treat' population. I forget what 'PP' stands for. I read the terms in some materials about clinical trial dataset before. 'ITT' should be the column name if I remember it right, and has value of "Y" (which flags this subject belongs to ITT population) and "N" (which flags this subject does not belong to ITT population). Mean statistics is calculated base on different populations, sometimes you need to report statistics for all samples(or subjects, participants), other times you need statistics for the ITT population only. ... View more

You are welcome, Jeff! In fact I personally think Tom's solution is brilliant, particularly because of the statement: days=min(date2, intnx('month', month,0,'e'))-max(month, date_1)+1, and the proc sort and proc report steps which helps me review what I have learnt. I like quickbluefish's solution as well, which is a bit too advanced for me. I appreciate this forum, I can learn a lot here and see how others think and write SAS code.       ... View more

Sorry for my English, by "arithmetic formula" I mean math expressions. ... View more

Hi abhi309,   For your requirement of "if the first prescription is within 180 days from 1st Jan 2020 then I don't want to include any subsequent prescriptions until the day difference between the consecutive prescriptions are more than 180 days and including all the prescriptions thereafter. But if the first prescription is after 180 days from 1st Jan 2020 then I can include all the prescriptions without bothering about the day difference", it can be simplified as: "do not include any prescription within 180 days from 01JAN2020". And thus my answer is created like this.   I also created a data set which calculate the days between two consecutive prescriptions for one subject(or participant, or patient whatever term you use) for possible further usage. The reference code for using LAG function to calculate numeric difference of any consecutive values for one subject is from SAS Programming 3: Advanced Techniques, course notes PDF, on page 32, no.4.   data preclaim; input PT_ID PRES_DATE date9.; format pres_date date9.; datalines; 1 5Jan2020 1 8Feb2020 1 10Jan2021 1 8Feb2021 2 15Aug2020 2 19Sep2020 2 18Oct2020 3 9Jan2020 3 10Oct2020 3 15Oct2020 4 10Jul2020 4 15Jul2020 4 20Mar2021 ; run; proc print data=preclaim;run; /*add include flag as 0 or 1*/ data preclaim1; set preclaim; if pres_date<'01jan2020'd+179 then include=0; else include=1; run; proc print data=preclaim1;run; /*calculate the number of days between any two consecutive prescription claims for one subject, for futher usage*/ data preclaim2; set preclaim; by pt_id; presdlag=lag1(pres_date); if first.pt_id then presdlag=.; diffdays=pres_date-presdlag; format presdlag date9.; run; proc print data=preclaim2;run; ... View more

Hi, Jeff, I did not get what you said. I have several questions on the rules you calculate the days for those months: (1) could you please tell how did you come up with the number of 22 days for that January? You can use arithmetic formula to show it, such as 31-9=22, or tell the start day and end day in your calculation, such as start day is 01Jan2024 and the end day is 22Jan2024. (2)why there is only 28 days in that February? Thanks. ... View more