About jenim514

jenim514 · ‎06-22-2016

Hi! I'm wondering if there is an option to include in the Proc Logistic preocedure that would output the number of observations used for each variable? for example, if i am using gender as an independent variable, can I add something to the procedure that would output # males and # female observations in addition to all the statistical output? Thanks!

jenim514 · ‎06-16-2016

@AstoundingI think I found my error! I have duplicate clinics in table B and I think that was dublicating observations in table A

jenim514 · ‎06-16-2016

So, using my real numbers (rather than hypothetical) and my actual log...this is what I am seeing...The final table has far more observations than i want. 119 data dmisid; 120 set mtf; 121 parcost=coalescec (mtf, parcost); 122 run; NOTE: There were 436487 observations read from the data set WORK.MTF. NOTE: The data set WORK.DMISID has 436487 observations and 102 variables. 113 Proc sql; 114 create table mtf.mtf_size as 115 SELECT a.*, b.parent_size, b.dmis_id, b.facility_type_code, b.facility_name 116 from dmisid a inner join mtf.facility_size_all b 117 on a.parcost=b.dmis_id; NOTE: Table MTF.MTF_SIZE created, with 3692241 rows and 106 columns. ***Also note, Table B mtf.facility_size_all has 5267 observations(the unique clinics).

jenim514 · ‎06-16-2016

@AstoundingTable B just contains a list of Clinics and their specific features. Each observation is a unique clinic (e.g.Parent_id). Now table A could have multiple ID's that match a single clinic (parent_id) in Table B...so the variables for that ID match would populate table A. So the end table would always need to be 500 obs.

jenim514 · ‎06-16-2016

Hi! Looking for some help with a sql join. Table A has all the observations I need (500 obs) so the final table should have 500 obs regardless of Table B. (which has 300K obs) Table B has some additional variables I need only if the Parent_ID (Table B) match the DMID_ID in Table A. I tried an inner join and left join but did not get correct output. Help is appreciated!! Proc sql; create table mtf.mtf_size as SELECT a.*, b.parent_size, b.dmis_id, b.facility_type_code, b.facility_FIPS_country_code, b.dmis_parent_id, b.facility_service_code from dmisid a inner join mtf.facility_size_all b on a.dmid_id=b.parent_id; quit;

jenim514 · ‎06-14-2016

it always helps to have that second pair of eyes! Yes, spelling correction fixed the problem. Thank you!

jenim514 · ‎06-14-2016

Hi! I need some help troubleshooting this syntax. Basically, I want to count all 'Oral Steroid' prescription fills by patientid. if there is an steroid use (i.e. gt 0) then variable Controlled='No', else 'Yes' . The output I am getting is not suming steroids (i.e. tot_steroids) by patient id. I is giving me a running total down all observations...and summing something untrelated to steroid use...but not sure what is it summing, and it is outputting TWO varialbes for 'tot_steroid'?? Help is appreciated! data mtf.control_status; set mtf.ipop_mtf_pdts (rename=(G=Drug_type)); by patientid; if first.patientid then do; steroid=0; tot_steriod=0; *initiate back to zero with each new patient id; end; if drug_type='Oral Steroids' then steroid=1; else steroid=0; tot_steroid + steroid; *summing oral steroid pharmacy data; if tot_steroid gt 0 then Controlled='No'; *at some point this patient used oral steroids; else Controlled='Yes'; if last.patientid then output; *the last patientid with summed pharmacy claims; run; Output looks like this (by patientid) SAS Output Obs Drug_type steroid tot_steroid Controlled 1 Rescue 0 3 No 2 Control 0 3 No 3 Control 0 3 No 4 Rescue 0 3 No 5 Control 0 3 No 6 0 3 No 7 Control 0 3 No 8 Control 0 21 No 9 Rescue 0 21 No 10 Oral Steroids 1 28 No 11 Rescue 0 28 No 12 Rescue 0 28 No 13 Rescue 0 28 No 14 Rescue 0 28 No 15 Control 0 28 No

jenim514 · ‎06-02-2016

Hi! I need help calculating the number of actual days a person was in my 5 year study. So, looking accross observations, I need to calculate number of days between the first encounter date 'encdate' and last encounter date by 'patientid'. Assistance is apprecited!

jenim514 · ‎05-31-2016

Hi @MikeZdeb. A while ago you gave the the solution to a question I have about identifying observations in a rolling 12 month period. Can you help me modify this syntax to identify encounters that have 5 ER visits in any rolling 12 month period (regardless of diagnosis 'dx'). I feel like if should be similar to this syntax, but not quite sure how I should work this. All the observations in the table are ER visits, but I just need those that have 5 or more in any rolling 12 month period (table contains 5 years of ER data). Thank you!!

jenim514 · ‎05-23-2016

Hi! I need to create a count visits by patient and age. I have attempted with the following syntax, but age is not counting correctly data util_sum; set asthma.household_female; by patientid patage; if first.patientid=1 then N_util=0; if utilization=1 then N_util +1; if last.patientid=1 then output; keep patientid patage n_util; run; I also tried using proc means, but age is still not counting correctly. proc means sum data=utilization by patientid; class patage; var utilization; output out=var_sum sum=; run; Help is greatly appreciated!!

jenim514 · ‎05-09-2016

Yes, count is correct. I think I need a data set because i will need to build graphs later. I think I can take out the marital status and just count the ER visits by year. I can add a conditional statement later.. data for ER l is binary...if there was an ER visit the ER=1. ID 3. ER $ 1. Year $ 4. 000 0 2012 000 1 2014 001 1 2012 001 1 2012 001 0 2013 001 1 2013 002 1 2012 002 1 2014 002 1 2014 Want to look like this... ID 3. ER $ 1. Year $ 4. Marital $ 1. ER_12 1. ER_13 1. ER_14 1. 000 0 2012 1 . . 000 1 2014 . . 1 001 1 2012 . . . 001 1 2012 . . . 001 1 2012 3 . . 001 0 2013 . . . 001 1 2013 .1 . 002 1 2012 1. . 002 1 2014 . . . 002 1 2014 . . 2

jenim514 · ‎05-09-2016

Hi! I am interested in ouputing a sum value to a new variable based on a condition. 1. I have a table with 100 observations- patientid with all their encounters. 2. I want to sum all married patients with ER visit in each year 2012- 2014 and output that sum by year to the last.patientid of each year. 3. The end table should still have 100 observations...just with two new sum variables. I've tried, but my table keeps reducing to one patientid with ER sum or the table stays at 100 with the ER sum on each observation line. This is the last code i tried, but not correct. Not sure where to insert output statement (not correct here!) or if this type of do statement is possible?? DATA ER1_sum; set var; by patientid; where Marital_status='Married'; if ER1=1 then do; if year='2012' then do; if first.patientid=1 then n_ER12=0; n_ER12 +1; if last.patientid=1 then output; end; if year='2013' then do; if first.patientid=1 then n_ER13=0; n_ER13 +1; if last.patientid=1 then output; end; if year='2014' then do; if first.patientid=1 then n_ER14=0; n_ER14 +1; if last.patientid=1 then output; end; end; run;

jenim514 · ‎05-06-2016

Hi! I would like some help creating a graph. I am basically trying to create three lines, bars 1. Bars show the mean of all ER visits by variable 'patientid' during during each year. 2. Line1 shows the same (mean) but just with 'married' marital status. 3. Line2 show same but with single marital status. I seemed to get the sum (default) to work ok, but when I change the stat to 'mean' it doesnt work correctly. All response variables are numeric (1) for event...but I gues I want it to sum the events and then graph the mean. It is just giving me 'mean' of 1 since all events are coded 1. Should calulate the mean in a separate data step and then do the plot or can i do that within the proc sgplot? I guess all the visuals could be lines since the means might be similar. goptions reset=goptions device=png300; ods _all_ close; ods rtf file='asthma.graph13.rtf'; proc sgplot data=asthma.graphvar; by patientid; where Visit_type='ER'; vbar year/ response=visit stat=mean fillattrs=(color=gray); vline year/ response=Married stat=mean lineattrs=(color=black thickness=10); vline year/ response=Single stat=mean lineattrs=(pattern=solid color=red thickness=10); run; ods rtf close; ods listing;

jenim514 · ‎04-22-2016

Hi! I have a variable COMB that have values in 0,1 code. Is there a way (using contains, where, index or something) that would allow me to do this: if a value contains <=4 (1's) to output to a new variable (COMB_1)? data have; format comb $ 5. datalines; 10111 01001 01000 00000 11110 11111 ; Help appreciated!!

jenim514 · ‎04-20-2016

@Astounding Worked great! Thank you!

Online Status	Offline
Date Last Visited	‎03-13-2025 05:50 PM

Re: Split header on top of Spanning Header in proc report

Re: Split header on top of Spanning Header in proc report

Split header on top of Spanning Header in proc report

Split a long text string, 200 or less using a ';' delimiter- retaining...

Control Edge from Header in proc report

Split long character string using multiple delimiters

Re: Proc Report- remove space between table and footnotes

Proc Report- remove space between table and footnotes

Re: Calling a symput date

Calling a symput date

Re: output observation at each change of occurance for 0, 1

Re: Capture text string between two delimiters

Re: macro using number as character

Re: continue indent when text wraps

Re: extract string that varies in length and number of delimiters

Trim Values that contain space

Re: wrap text in text string

Re: sql join with different values in different cases

Re: Calculate mean by ID/ Visit with multiple attempts per visit

Re: set new variable to missing if time vaue in a character string is ...

N option?

Re: Sql join to add variables to table A if there is a match in B

Re: Sql join to add variables to table A if there is a match in B

Re: Sql join to add variables to table A if there is a match in B

Sql join to add variables to table A if there is a match in B

Re: help sum calculation by patient id

help sum calculation by patient id

number of days in the study period

Re: identifying observations 12 month period

counting patient visits by patientid and age

Re: counting patient visits with two by groups

counting patient visits with two by groups

SGPlot using Mean

a condition for identifying characters in a value

Re: proc transpose long