Working with incomplete longitudinal data - data cleaning help

[hwangnyc]

Hi everyone,

I’m working with data that has multiple visits. Baseline, 3, 6,9,12 months follows ups, five sessions in total. At each session 7 events are asked, each question will be a row in the database.  So Ideally, each case should have 37 rows.  An ideal “clean case” should look like this:

ChildID	AsthmaEventDate_HW	AsthmaEventCode	AsthmaEventCount
1944	7-Mar-14	1	0
1944	7-Mar-14	2	3
1944	7-Mar-14	3	0
1944	7-Mar-14	4	0
1944	7-Mar-14	5	18
1944	7-Mar-14	6	0
1944	7-Mar-14	7	3
1944	11-Jul-14	1	0
1944	11-Jul-14	2	0
1944	11-Jul-14	4	0
1944	11-Jul-14	3	0
1944	11-Jul-14	5	0
1944	11-Jul-14	6	0
1944	11-Jul-14	7	0
1944	6-Nov-14	1	0
1944	6-Nov-14	2	0
1944	6-Nov-14	3	0
1944	6-Nov-14	4	0
1944	6-Nov-14	5	0
1944	6-Nov-14	6	0
1944	6-Nov-14	7	0
1944	3-Apr-15	1	1
1944	3-Apr-15	2	0
1944	3-Apr-15	3	1
1944	3-Apr-15	4	1
1944	3-Apr-15	5	30
1944	3-Apr-15	6	7
1944	3-Apr-15	7	1
1944	24-Dec-15	1	0
1944	24-Dec-15	2	2
1944	24-Dec-15	3	5
1944	24-Dec-15	4	0
1944	24-Dec-15	5	8
1944	24-Dec-15	6	8
1944	24-Dec-15	7	8

 However, there are examples when the data is incomplete and missing sessions and or events (see below).

ChildID	AsthmaEventDate_HW	AsthmaEventCode	AsthmaEventCount
19	6-May-08	3	1
19	6-May-08	4	1
19	6-May-08	5	5
19	6-May-08	7	2
19	6-Jun-08	5	5
19	22-Dec-08	2	10
19	22-Dec-08	5	20

The first step that I would like to take is to see how many cases have all 7 sessions, how many have 6, etc. I’m also open to other suggestions as well.

Thanks!

[Shmuel]

1) Please note: 5 sessions time 7 events per session are 35 lines, not 37. Do I miss anyting ?

    Is session expressed by AsthmaEventCode ?

    ChildID 19 has twice AsthmaEventCode = 5. Is it possible? what does it mean? is it an error ?

2) Yo can count frequencies of AsthmaEventCode per child by:

    proc freq data= XXXX;

      table childID * AsthmaEventCode;

   run;

   Does the report produced answer to your demands?

 3)  I need more information and understanding to be mor accuracy.

Shmuel