About sks521

sks521 · ‎07-03-2019

Just to give you an idea of what I mean if it's not clear; PKid -3 -2 -1 0 1 2 3 total visits 2 2 6 0 6 0 0 0 14 3 3 13 4 15 4 0 0 39 5 2 5 13 1 4 0 0 25 So -3, -2, -1, 0, 1,2,3 are time periods and each id had different frequency of visit during each period, this is only the intervention group (by status variable) How do I take average visits per pkid during each time period in a data step, when the data is in the following long form; pkid visit month2 status 1 1 -3 eligible 1 1 -3 eligible 1 1 -2 eligible 1 1 0 eligible 2 1 -1 intervention 2 1 2 intervention 2 1 3 intervention 2 1 0 intervention

sks521 · ‎07-03-2019

Sorry about confusion, I should've been more explicit. I have created a variable to indicate my time periods in a separate step. I also did create a variable called 'visit' to indicate a visit. Have also tried the code you have suggested but it still doesn't solve the problem. It doesn't give me average number of visits per discrete id for each time period by status

sks521 · ‎07-03-2019

Mathematically speaking I am looking for an output like; Average visits per ID for pre-intervention period= total visits/total number of ids (discrete ids attending the hospital during that period) For example; average number of visits during pre-intervention; Eligible= 4.44 Intervention= 5.66 Average number of visits during intervention; eligible=5.33 Intervention= 6.66 Average number of visits post-intervention; eligible=3.99 Intervention=6.22 Something like that. Logic; for e.g. id '1' would have had 20 visits over the course of study- 3 during pre-intervention, 7 during intervention and 10 during post-intervention. Similarly, ID '2 ' would have had 9 visits; 1 during pre-intervention, 0 during intervention and 8 during post-intervention. So I want to create average visits per discrete ID for each period. I hope it's clear now? Please ask if it's not clear. Thanks for your time on this

sks521 · ‎07-03-2019

Well, this sql gives me following output; id average 1 4.00 2 4.66 3 7.88 Doesn't really solve the problem. Still not the average per time period, I'm afraid!

sks521 · ‎07-03-2019

Mathematically speaking I am looking for an output like; Average visits per ID for pre-intervention period= total visits/total number of ids (discrete ids)

sks521 · ‎07-03-2019

PROC MEANS DATA=saltaire SUM MEAN MAXDEC=0 ; VAR visit; by pkid; run; the output looks like; PKid=437 Analysis Variable: visit Sum Mean 20 1 so this gives me sum of visits for each ID separately. How do I convert this into average fro each time period?

sks521 · ‎07-03-2019

I have posted the code in my previous reply. Thanks S

sks521 · ‎07-03-2019

Thanks Kurt, that's the problem, I don't know how to calculate mean by ID. Might be a simple step but it's because I've been struggling with it too much and my brain has given up probably.

sks521 · ‎07-03-2019

forgot to mention, the code that I wrote only gives me a sum of visits during each time period but not a mean by number of ids in each time period.

sks521 · ‎07-03-2019

So I have created a variable called 'month2' to get 7 time periods from the 'eventdate' variable and another variable called 'visit' to indicate one visit for each 'eventdate' then wrote this code; proc summary data=saltaire; by status month2; var visit; output out=sum n=n; run; What I am not sure about is, how to take into account visits per number of IDs by status and month2.

sks521 · ‎07-03-2019

Hi folks, I have a data set where each ID has multiple visits over time and I have 7 time points for which I need to calculate mean number of visits by ID. I have tried several codes and asked the question on this forum but couldn't get an answer. Can someone help please? Ta My data looks like; ID status eventdate age sex 1 Eligible 2016-04-14 73 male 1 Eligible 2017-02-12 73 male 1 Eligible 2018-09-03 73 male 1 Eligible 2016-01-15 73 male 1 Eligible 2016-11-17 73 male 2 Intervention 2018-01-15 73 male 2 Intervention 2016-01-12 73 male 2 Intervention 2017-04-15 73 male 2 Intervention 2018-09-22 73 male 2 Intervention 2018-07-26 73 male 2 Intervention 2018-03-15 73 male 3 Intervention 2017-02-13 73 male 3 Intervention 2017-01-12 73 male Based on eventdate, I have 7 time points; arbitrarily; if EventDate<= '28FEB2016'd then delete;*for the sake of uniformity in number of pre- and post- time periods, deleting visits before March 2016; else if '01MAR2016'd <=EventDate<= '31AUG2016'd then month2='-3';*from March 2016 to August 2016; else if '01SEP2016'd <=EventDate<= '28FEB2017'd then month2='-2'; *From September 2016 to February 2017; else if '01MAR2017'd <=EventDate<= '31AUG2017'd then month2='-1'; *From March 2017 to August 2017; else if '01SEP2017'd <=EventDate<= '28FEB2018'd then month2='0'; *From September 2017 to February 2018- intervention period; else if '01MAR2018'd <=EventDate<= '31AUG2018'd then month2='1'; *From March 2018 to August 2018- first six months after intervention; else if '01SEP2018'd <=EventDate<= '28FEB2019'd then month2='2'; *From September 2018 to February 2019- second six months after intervention; else if '01MAR2019'd <=EventDate<= '31MAY2019'd then month2='3'; *From March 2019 to May 2018- third time period after intervention but doesn't have the data till August; Help much appreciated.

sks521 · ‎07-03-2019

So I have tried to use the same syntax using more than three time pints and I got this log; 885 data SMPsum; 886 set saltaire; 887 888 retain _err_ preint4 preint3 preint2 preint1 int postint1 postint2 postint3 0; 889 array period[0:8] _err_ preint int postint; ERROR: Too few variables defined for the dimension(s) specified for the array period. 890 891 select; 892 when ( EventDate <'01MAR2016'd) pd = 1;*pre-intervention -4; 893 when ('01MAR2016'd <=EventDate<= '31AUG2016'd) pd=2; *preintrvention -3; 894 when ('01SEP2016'd <=EventDate<= '28FEB2017'd) pd=3; *preintervention -2; 895 when ('01MAR2017'd <=EventDate<= '31AUG2017'd) pd=4; *preintervention 1; 896 when ('01SEP2017'd <=EventDate<= '28FEB2018'd) pd =5;*intervention; 897 when ('01MAR2018'd <=EventDate<= '31AUG2018'd) pd=6; *postintervention 1; 898 when ('01SEP2018'd <=EventDate<= '28FEB2019'd) pd=7; *postintervention 2; 899 when ( EventDate>= '01MAR2019'd) pd =8 ;*post-intervention 3; 900 otherwise pd = 0; 901 end; 902 903 period[pd] = 1; 904 output; 905 period[pd] = 0; 906 run; NOTE: The SAS System stopped processing this step because of errors. WARNING: The data set WORK.SMPSUM may be incomplete. When this step was stopped there were 0 observations and 49 variables. NOTE: DATA statement used (Total process time): real time 0.04 seconds cpu time 0.01 seconds Ho do I deal with that please? Thanks S

sks521 · ‎07-03-2019

Hi Rick, Would you mind looking at my example data set and post example syntax please? Thanks S

sks521 · ‎07-02-2019

I did run this code but the output doesn't look right; Pkid Nobs Mean 1 5 20879.60 2 3 20758.33 3 5 20879.60 Looks like it has calculated mean out of Date value while I need mean number of visits per following duration; pre-intervention, intervention and post-intervention. Thanks

sks521 · ‎07-02-2019

Thanks but if I wanted to have 6 or 7 time periods instead, how will I write the syntax then? Best Sarwat

Online Status	Offline
Date Last Visited	‎02-21-2020 01:05 PM

Re: Creating new variables

Re: Creating new variables

Re: Creating new variables

Re: Creating new variables

Re: Creating new variables

Re: Creating new variables

Re: Creating new variables

Creating new variables

Re: ICD-10 grouping

Re: ICD-10 grouping

Re: Grouping ICD-10 codes

Re: Bringing data into SAS using Datalines option

Re: Finding duplicates

Re: Finding duplicates

Re: SAS macro

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Re: Means procedure

Means procedure

Re: Summing

Re: Difference-in difference trends plot

Re: Summing

Re: Summing