HELP with longitundinal base programming - I have done this before but it's not working now.

[mgrzyb]

I need help with some longitudinal programing which I have done before!

But this time, after I assign first obs for counting observations, I get a weird result for id=55.

I included the program here (I would not accept my sas files -sorry). 

- The log says it all worked but I get the wrong results.

- I attached the database - I changed dates so there are no HIPAA violations

- At end, I attached the output.

I am sorry for a basic ? but i'm tired and have spend 2 days. 

 IT WILL NOT ALLOW ME TO ATTACH THE DATA. SO I WILL COPY IT HERE. IT'S KIND OF LONG.

  OH, IS THERE ANY WAY TO LEFT JUSTIFY THE OUTPUT?

　

Program:

ODS HTML close; ODS listing; /* this is SAS printout */

options linesize=132 pagesize=60 center date;

options pageno=1;

LIBNAME help 'C:\sas\a_study\post1help';

libname library2 'c:\sas\a_study\post1help';

options fmtsearch=(library2);

dm 'out; clear; log; clear;';

data one;

set help.onepost;

proc contents; proc print;

run;

 　

data counts; /* gets me the total counts of visits for each person */

set help.onepost ; *found no duplicates;

  by id;

   if first.id=1 then n_visits=0;

   n_visits + 1;

   if last.id=1 then output;

proc print data=counts; run;

OUTPUT 

The SAS System

Obs id initial_hr_date test_hr_date prehr posthr n_visits1234

1	03/01/2018	03/02/2018	160	94	10
3	11/27/2017	11/28/2017	249	133	20
55	08/20/2010	08/23/2010	118	145	57
56	06/17/2010	06/18/2010	172	88	6

* Read the count (proc prints)   Obs55 should be 4/30/10 for both dates!

Notice how 55 does not have its original data.

The dates are mixed up and

the first obs should be 4/30/10;

/*

Why does it change like that?

The others don't. */

                  * Read the count (proc prints)

                  Notice how 55 does not have its original data.

                        The dates are mixed up and

                        the first obs should be 4/30/10;

/*

Why does it change like that?

The others don't. */

THANKS!

Obs   id   initial_test_hr      hr_date date   prehr    posthr

1 1 01/30/2018 02/19/2018 135 105

2 1 02/19/2018 02/20/2018 162 83

3 1 02/20/2018 02/21/2018 198 94

4 1 02/21/2018 02/22/2018 184 99

5 1 02/22/2018 02/23/2018 153 105

6 1 02/23/2018 02/26/2018 161 127

7 1 02/26/2018 02/27/2018 134 148

8 1 02/27/2018 02/28/2018 174 129

9 1 02/28/2018 03/01/2018 206 152

10 1 03/01/2018 03/02/2018 160 94

11 3 10/05/2017 10/30/2017 137 160

12 3 10/30/2017 10/31/2017 162 109

13 3 10/31/2017 11/01/2017 305 176

14 3 11/01/2017 11/02/2017 207 211

15 3 11/02/2017 11/03/2017 175 96

16 3 11/03/2017 11/06/2017 145 78

17 3 11/06/2017 11/07/2017 172 175

18 3 11/07/2017 11/08/2017 186 95

19 3 11/08/2017 11/09/2017 261 173

20 3 11/09/2017 11/10/2017 148 118

21 3 11/10/2017 11/13/2017 166 134

22 3 11/13/2017 11/14/2017 197 157

23 3 11/14/2017 11/15/2017 351 162

24 3 11/15/2017 11/16/2017 231 100

25 3 11/16/2017 11/17/2017 235 100

26 3 11/17/2017 11/20/2017 126 88

27 3 11/20/2017 11/21/2017 144 113

28 3 11/21/2017 11/22/2017 241 268

29 3 11/22/2017 11/27/2017 129 153

30 3 11/27/2017 11/28/2017 249 133

31 55 04/20/2012 04/20/2010 154 128

32 55 04/20/2010 05/10/2010 164 113

33 55 05/10/2010 05/11/2010 135 107

34 55 05/11/2010 05/12/2010 138 112

35 55 05/12/2010 05/13/2010 109 119

36 55 05/13/2010 05/14/2010 155 107

37 55 05/14/2010 05/19/2010 189 149

38 55 05/19/2010 05/21/2010 235 138

39 55 05/21/2010 05/24/2010 190 201

40 55 05/24/2010 05/25/2010 181 116

41 55 05/25/2010 05/26/2010 190 165

42 55 05/26/2010 05/27/2010 177 170

43 55 05/27/2010 05/28/2010 158 111

44 55 05/28/2010 06/01/2010 188 116

45 55 06/01/2010 06/02/2010 230 100

46 55 06/02/2010 06/08/2010 127 120

47 55 06/08/2010 06/09/2010 187 132

48 55 06/09/2010 06/10/2010 196 167

49 55 06/10/2010 06/11/2010 193 149

50 55 06/11/2010 06/15/2010 150 81

51 55 06/15/2010 06/16/2010 165 119

52 55 06/16/2010 06/17/2010 172 128

53 55 06/17/2010 06/18/2010 105 123

54 55 06/18/2010 06/21/2010 208 107

55 55 06/21/2010 06/22/2010 163 85

56 55 06/22/2010 06/23/2010 180 98

57 55 06/23/2010 06/24/2010 196 81

58 55 06/24/2010 06/25/2010 202 118

59 55 06/25/2010 06/29/2010 183 85

60 55 06/29/2010 06/30/2010 181 147

61 55 06/30/2010 07/01/2010 151 119

62 55 07/01/2010 07/02/2010 184 154

63 55 07/02/2010 07/06/2010 199 93

64 55 07/06/2010 07/08/2010 191 101

65 55 07/08/2010 07/09/2010 187 94

66 55 07/09/2010 07/12/2010 36 156

67 55 07/12/2010 07/13/2010 146 96

68 55 07/13/2010 07/19/2010 267 130

69 55 07/19/2010 07/21/2010 152 99

70 55 07/21/2010 07/23/2010 265 104

71 55 07/23/2010 07/26/2010 164 158

72 55 07/26/2010 07/27/2010 181 93

73 55 07/27/2010 07/28/2010 250 111

74 55 07/28/2010 07/30/2010 251 104

75 55 07/30/2010 08/03/2010 178 125

76 55 08/03/2010 08/04/2010 224 93

77 55 08/04/2010 08/05/2010 195 145

78 55 08/05/2010 08/06/2010 246 123

79 55 08/06/2010 08/10/2010 155 121

80 55 08/10/2010 08/11/2010 307 91

81 55 08/11/2010 08/12/2010 179 106

82 55 08/12/2010 08/13/2010 258 207

83 55 08/13/2010 08/16/2010 198 77

84 55 08/16/2010 08/17/2010 208 123

85 55 08/17/2010 08/19/2010 225 155

86 55 08/19/2010 08/20/2010 206 115

87 55 08/20/2010 08/23/2010 118 145

88 56 06/01/2010 06/09/2010 151 91

89 56 06/09/2010 06/10/2010 189 108

90 56 06/10/2010 06/15/2010 156 94

91 56 06/15/2010 06/16/2010 162 107

92 56 06/16/2010 06/17/2010 183 116

93 56 06/17/2010 06/18/2010 172 88

[Kurt_Bremser]

I see no problem. You get all 4 variable values from obs 87 of onepost, and the correct count of 57 (obs 31 to 87).

[mgrzyb]

I think I send the wrong part of the dataset for id=55. let me look again. 

[mgrzyb]

I posted the wrong data but the right out put. 

If you notice id=55, the 1st visitnumber should be that where there is 4/20/10 and 4/20/10.  And it is not.  I copied it on a word document this time.  

Feel free to test my program and see what you get.  

I can't get the 1st visit to be the 1st observation for id 55, which should be 4/20/10 and 4/20/10.  That person had the 1st initial consult on 4/20/10 and went to the treatment to treat the heart rate the same day.

I APOLOGIZE greatly if I got the data mixed up. 

But this is the correct data here. 

It is in excel. I hope it attaches. I don't know how to make these things attach. I tried everything!

[mgrzyb]

I have to take a nap.  I am so tired.  Please help me. 

[Tom]

You mentioned DATES but your posted code is doing NOTHING with the date variables.  What do you want to do with the date variables?

Let's assume that in addition to counting observations you want to find the first and the last date per ID.

data want ;
  set help.onepost ; 
  by id date ;
  retain n_visits first_date ;
  format first_date last_date date9. ;
  if first.id then n_visits=0;
  if first.id then first_date = date ;
  n_visits + 1;
  if last.id ;
  last_date = date ;
run;

Make sure that your retained variable (N_VISITS and FIRST_DATE) are NOT already on the input dataset.  If they are then either use different variable names or add the DROP= dataset option to the SET statement so the old variables are not read into the current step.

PS Learn to use the tools on the Forum editor menu to paste text, programs and images.  This will make your postings easier to read.

[mgrzyb]

Thxs. I’ll check it out later tonight.

[katemcguckin]

Hi Tom,

I tested your code. I do not understand the date comment.  I have:

initial_hr_date and test_hr_date.

It gets stuck when sorting by id date   - I hv no date variable.

Do I need to change initial_hr_date to firstdate and test_hr_date to last_date?

Could you write it out? I am so confused. 

Thanks.  

[katemcguckin]

Tom, I get this in my log:

BY variable date is not on input data set HELP.ONEPOST.

[Kurt_Bremser]

---

@katemcguckin wrote:

Tom, I get this in my log:

BY variable date is not on input data set HELP.ONEPOST.

---

If you want code that has been tested against real SAS data, you have to post such. Excel spreadsheets are useless for this. They only leave us guessing how your SAS datasets might look.

Post data in a data step with datalines, and we have something to work with.

[Tom]

Adapt the concepts to your data.  You seem to have two date variables.  What do you want to do with them?  If you do nothing, like in your original post, then you will output the values that are on the last observation for the group.

[katemcguckin]

Hi,

You know. It dawned on me when I use the long data and program it, it's all ok.   I think I confused myself so much.  

I want to know tge days between treatmeants time from 1st visit to last visit,

I did those using code from Cody's book.

I am sorry I bothered you.  I sometimes over think and confuse myself.

Thanks for your time, though!  I am sorry.