← SASnRRetain first value of a subject onto other records

Introduction

A common derivation in clinical data: capture the first score (baseline, first visit, first entry) for each subject and copy that value onto every other record of the same subject.
The output keeps the same number of rows as the input; each row just gains an extra column holding the subject's first-visit value.
SAS does this in a DATA step using RETAIN plus the first. automatic variable on the BY group.
Tidyverse uses group_by() + fill() (or the shorter first()) to push the first value down through the group.
Base R uses ave() with a custom function that always returns the first element of each group.
This lesson copies each subject's visit-1 score onto visits 2 and 3 so the result is easy to inspect.

Retain first value of a subject onto other records

SAS (Base SAS)+

*=============================================================================;
* Create the input dataset;
*=============================================================================;

data fs;
    input usubjid visitnum score;
    datalines;
101 1 20
101 2 10
101 3 30
102 1 90
102 2 10
102 3 100
;
run;

*=============================================================================;
* Sort the dataset by usubjid and visitnum;
*=============================================================================;

proc sort data=fs;
    by usubjid visitnum;
run;

*=============================================================================;
*Retain the first score onto all records;
*=============================================================================;

data fs01;
    set fs;
    by usubjid visitnum;

    retain first_score;
    if first.usubjid then first_score = score;
run;

To carry the first score of each subject forward onto every record, we sort by subject and visit, then use RETAIN with first.usubjid.
SAS uses BY-group processing: after PROC SORT by usubjid visitnum and a by statement in the DATA step, SAS exposes automatic flags first.usubjid and last.usubjid.
retain first_score; tells SAS not to reset first_score to missing at the start of each row — it holds whatever value we assigned on the previous row.
if first.usubjid then first_score = score; captures the score from the very first row of each subject. Because of the retain, that value stays in place for every subsequent row of the same subject.
After running, we expect fs01 to have all six input rows plus a new first_score column equal to the first visit’s score within each subject.

R (tidyverse)+

#==============================================================================;
#Retain first score;
#==============================================================================;

fs <- tribble(
  ~usubjid,~visitnum,~score,
  101,1,20,
  101,2,10,
  101,3,30,
  102,1,90,
  102,2,10,
  102,3,100
)

fs01 <- fs %>% 
  arrange(usubjid,visitnum) %>% 
  group_by(usubjid) %>% 
  mutate(
    seq = row_number(),
    first_score = if_else(seq==1,score,NA)
  ) %>% 
  fill(first_score,.direction="down")

To carry the first score of each subject forward in tidyverse, we arrange, group and then fill the missing values downward.
arrange(usubjid, visitnum) is the tidyverse analogue of PROC SORT BY usubjid visitnum.
group_by(usubjid) scopes the next operations to each subject — equivalent to the SAS by usubjid.
Inside mutate(), we use row_number() to tag the first row of each subject. We assign the score to first_score only for the first row; other rows get NA.
fill(first_score, .direction = "down") then propagates that first value downward, filling every subsequent NA within the same group — the tidyverse equivalent of the SAS RETAIN.
After running, we check fs01 and confirm first_score equals the first visit’s score for every row of the subject.
A shorter alternative is mutate(first_score = first(score)), which gives the same result without the seq/fill dance — worth keeping in mind for real code.

Happy learning!

Search lessons

Introduction

Retain first value of a subject onto other records