← SASnRTranspose data - long to wide

Introduction

Clinical data often arrives in long shape — one row per subject per test — but reports and downstream tools usually want wide shape with one row per subject and one column per test.
SAS uses proc transpose with a by, var, and id statement to pivot the data.
Tidyverse uses pivot_wider() from tidyr, specifying the id, names, and values columns.
Base R uses reshape() with direction = "wide".
This lesson pivots a LONG dataset of lab results (usubjid, lbtestcd, lbstresn) into a WIDE dataset with one column per lab test.
After running, the WIDE dataset should have one row per usubjid and one column per lab test code — HGB, ALT, and AST in this sample.

Transpose data - long to wide

SAS (Base SAS)+

R (tidyverse)+

library(tidyverse)

long <- tribble(
  ~usubjid, ~lbtestcd, ~lbstresn,
  "1001",   "HGB",     13,
  "1001",   "ALT",     30,
  "1001",   "AST",     23,
  "1002",   "HGB",     12,
  "1002",   "ALT",     28,
  "1002",   "AST",     15
)

wide <- long %>%
  pivot_wider(
    id_cols     = usubjid,
    names_from  = lbtestcd,
    values_from = lbstresn
  )

In tidyverse, pivot_wider() from tidyr is the direct counterpart of PROC TRANSPOSE.
id_cols = usubjid names the column that defines rows in the wide output — one row per usubjid.
names_from = lbtestcd names the column whose values become new column names — HGB, ALT, and AST here.
values_from = lbstresn names the column whose values fill the cells in the wide layout.
If several rows share the same usubjid and lbtestcd, pivot_wider() collapses them to a list — add values_fn = mean or similar to aggregate explicitly.
We run and open the result: one row per subject, with lab-test codes as columns; the shape should match the SAS WIDE dataset exactly.

Happy learning!

Search lessons

Introduction

Transpose data - long to wide