The part I'm getting stuck on is that there's not a single maximum value to continue the numbering from since my data is grouped. I can't restart the numbering because I want to preserve the old data, which isn't always sequential and doesn't always start at 1.   Here is the data I'm starting with:   data old; INPUT patient $ repeat_instance episode_number visit_date :MMDDYY10.; FORMAT visit_date MMDDYY10.; datalines; A 1 1 01/12/2026 A 2 2 01/16/2026 A 4 3 02/04/2026 B 1 1 01/05/2026 B 2 2 01/09/2026 C 1 1 01/10/2026 D 2 1 01/18/2026 D 3 2 01/31/2026 D 5 3 02/05/2026 ; run; Here's the new data I want to add:   data new; INPUT patient $ repeat_instance episode_number visit_date :MMDDYY10.; FORMAT visit_date MMDDYY10.; datalines; A . 4 02/12/2026 B . 3 01/16/2026 B . 4 02/04/2026 C . 2 03/05/2026 D . 4 03/09/2026 D . 5 04/10/2026 ; run; I'm combining the data in a simple data step and sorting:     data combined; set old new; run; proc sort data=combined; by patient episode_number; run; This is what the dataset combined looks like: patient repeat _instance episode_number visit_date A 1 1 01/12/2026 A 2 2 01/16/2026 A 4 3 02/04/2026 A . 4 02/12/2026 B 1 1 01/05/2026 B 2 2 01/09/2026 B . 3 01/16/2026 B . 4 02/04/2026 C 1 1 01/10/2026 C . 2 03/05/2026 D 2 1 01/18/2026 D 3 2 01/31/2026 D 5 3 02/05/2026 D . 4 03/09/2026 D . 5 04/10/2026   This is what I want: patient repeat_instance episode_number visit_date A 1 1 01/12/2026 A 2 2 01/16/2026 A 4 3 02/04/2026 A 5 4 02/12/2026 B 1 1 01/05/2026 B 2 2 01/09/2026 B 3 3 01/16/2026 B 4 4 02/04/2026 C 1 1 01/10/2026 C 2 2 03/05/2026 D 2 1 01/18/2026 D 3 2 01/31/2026 D 5 3 02/05/2026 D 6 4 03/09/2026 D 7 5 04/10/2026   Basically, I want repeat_instance to be the max value of the old data, grouped by patient, plus 1. I've used the retain function when numbering from scratch, or for continuing numbering that's not grouped.  I'm thinking of using the lag function but can't quite figure out how that would work beyond the first new value of the group. Here's what I've tried with the lag function and the resulting data: data lag; set combined; by patient; lag_repeat_instance=lag(repeat_instance); run; data lag_ct; set lag; by patient; retain repeat_instance; if repeat_instance=. then repeat_instance=lag_repeat_instance+1; run; patient repeat_instance episode_number visit_date lag_repeat_instance A 1 1 01/12/2026 . A 2 2 01/16/2026 1 A 4 3 02/04/2026 2 A 5 4 02/12/2026 4 B 1 1 01/05/2026 . B 2 2 01/09/2026 1 B 3 3 01/16/2026 2 B . 4 02/04/2026 . C 1 1 01/10/2026 . C 2 2 03/05/2026 1 D 2 1 01/18/2026 . D 3 2 01/31/2026 2 D 5 3 02/05/2026 3 D 6 4 03/09/2026 5 D . 5 04/10/2026 .    I was hoping there is a better solution that I'm not thinking of. Thanks in advance!   ... View more