Try lag3. ... View more

I think this tackles it: data have; id=1; model='a'; grade=1; amount=10; output; id=1; model='b'; grade=2; amount=10; output; id=2; model='a'; grade=3; amount=20; output; id=2; model='b'; grade=3; amount=10; output; run; proc sort data=have out=have_sort; by id descending grade descending amount model; run; data want (drop=max_amount); set have_sort; by id; retain max_grade max_amount max_model; if first.id then do; max_grade = grade; max_model = model; max_amount=amount; end; if grade > max_grade then do; max_grade = grade; max_model = model; end; else if grade=max_grade and amount > max_amount then do; max_grade = grade; max_model = model; end; run; You could also do this in proc sql qith nested queries, but that second criteria would be a bit tricky.  ... View more

The value to compare to is still the ICD10 code. If you want that second grouping, the way I would do it is to add another format to your proc format step. proc format; value $ICD 'A000'-'A099'='Intestinal_infection' 'A150'-'A199'='TB' 'A200'-'A280'='Zoonotic' 'A300'-'A499'='Other_bacterial' 'A500'-'A640'='STI' 'A650'-'A699'='Spirochaetal' 'A800'-'A890'='Viral_CNS' 'A900'-'A990'='Viral-fever' 'B000'-'B099'='Viral_skin' 'B150'-'B199'='Viral_hep' 'B200'-'B240'='HIV' 'B250'-'B349'='Other_viral' 'B350'-'B490'='Mycoses' 'B500'-'B640'='Protozoa' 'B650'-'B839'='Helminthiases' 'B850'-'B890'='Other_infestation' 'B900'-'B949'='Sequelae' 'B950'-'B990'= 'Other' 'C000'-'C970'= 'Neoplasms' 'D000'-'D489'='Other_neoplasms' 'D500'-'D899'='Blood_disorder' 'E000'-'E899'='Endocrine' 'F000'-'F990'='Mental_disorder' 'G000'-'G998'='Nervous' 'H000'-'H599'='Eye_adnexa' 'H600'-'H959'='Ear_mastoid' '1000'-'1990'='Circulatory' 'J000'-'J998'='Respiratory' 'K000'-'K938'='Digestive' 'L000'-'L998'='Subcutanous' 'M000'-'M999'='MSK' 'N000'-'N999'='Genitourinary' 'O000'-'O998'='Pregnancy' 'P000'-'P969'='Perinatal' 'Q000'-'Q999'='Congenital' 'R000'-'R999'='Clinical_laboratory' 'S0000'-'S9999'='Injuries' 'T0000'-'T983'='Multiple_injuries' 'V010'-'V990'='Transport_accident' 'W000'-'W990'='Other_accidents' 'X000'-'X840'='Other_injuries' 'Y100'-'Y989'='Sequelae' 'Z000'-'Z999'='Potential_hazard' other='Ungrouped' ; value $regroup 'A000'-'B990'='Infections' 'C000'-'D489'='Neoplasm' 'D500'-'D899'='Blood_dis' 'E000'-'E899'='Endocrine' 'F000'-'F990'='Mental_dis' 'G000'-'G998'='Nervous' 'H000'-'H599'='Eye_adnexa' 'H600'-'H959'='Ear_mast' '1000'-'1990'='Circulatory' 'J000'-'J998'='Respiration' 'K000'-'K938'='Digestive' 'L000'-'L998'='Subcutanous' 'M000'-'M999'='MSK' 'N000'-'N999'='Genitourinary' 'O000'-'O998'='Pregnancy' 'P000'-'P969'='Perinatal' 'Q000'-'Q999'='Congenital' 'R000'-'R999'='Clinical_laboratory' 'S0000'-'Y989'='Injuries' 'Z000'-'Z999'='Other' other='Other' ; data ICD10; set have; New_code = icd; Regroup = icd; format new_code ICD. Regroup regroup.; run;   You can also get your code to work by using  New_code = put(icd,ICD.); to change the value of New_code before hitting it with your else if code.  ... View more

You need to change your syntax to: if (conditions) then do; stmt; smtmt; stmt; end; else if (conditions) then do; stmt; smtmt; stmt; end; else if (conditions) then do; stmt; smtmt; stmt; end; else if (conditions) then do; stmt; smtmt; stmt; end; else do; stmt; stmt; stmt; end; The data step isn't interpreting the "and"s the way you think it is.  ... View more

You could loop through the arrays and return the min greater than 0 with something like: data test2; set have; array min_SL_W{*} _numeric_ PRE_SL_W_count Post_SL_W_2wk Post_SL_W_6wk; min1 = 0; do i =1 to dim(min_SL_W); if min_SL_W{i} < min1 and min_SL_W{i} ne 0 then min1 = min_SL_W{i}; end; if min1=0 then min1=.; ... repeat for each array ... run;   ... View more

True, but glancing at the rest of the macro, I figured it was pretty safe to assume that was not the intention. ... View more

Looking at this quickly, it looks like you are calling the macro with a bunch of parameters: %tottable (var1, cross1, cross2, cross3, cross4, cross5, col1, 1, col2, 2) but the only one that I can see is defined by the time you call it is var1.  ... View more

You would include it every time you want to check if there was an error and abort if there is. You don't have to include it if you want to continue anyway. ... View more

Sorry, it took a bit for SAS to boot up for me to test that. I edited my above post to reflect the actual behavior.   TRANWRD will replace using a space for the empty string in the third argument.   If you want to remove the space, you can use TRANSTRN and TRIMN like so: newString = transtrn(dataelement,'0',trimn('')); ... View more

If all you want to do is remove (all) 0s, you can use TRANWRD.   newString = tranwrd(dataelement,'0',''); that will replace 'A015478' with 'A 15478'. ... View more

There is!   data test; set sashelp.cars; run; proc sort data= test; by typ; run; *this code yields an error message because typ doesn't exist; *this is where I want my SAS to stop; %macro OnlyRunWithoutError; %if &syserr = 0 %then %do; data test2; set test; run; *so, I want SAS not to execute this line of code if SAS faces an error earlier; %end; %mend OnlyRunWithoutError; %OnlyRunWithoutError; ... View more