← SASnRExtract Date and Time Parts from a Datetime Value

Introduction

Lab results and vitals often carry full datetime stamps — we need the date portion for merging with other datasets and the time portion for ordering records within a day.
SAS stores datetimes as seconds since midnight on 01JAN1960; extracting the date and time parts requires dedicated functions.
R stores datetimes as POSIXct objects (seconds since 1970-01-01 UTC); extracting parts uses lubridate helpers or base formatting functions.
This lesson extracts date and time components from a small set of lab datetime values and shows how to get hours, minutes, and seconds individually.
Understanding the underlying storage model is important — it explains why we cannot simply assign a datetime to a date variable without an explicit extraction step.

Extract Date and Time Parts from a Datetime Value

SAS (Base SAS)+

*=============================================================================;
* Extract date and time parts from a datetime value
*=============================================================================;

*-----------------------------------------------------------------------------;
* Create sample lab data with datetime stamps
*-----------------------------------------------------------------------------;
data labs;
    input usubjid $ lbtestcd $ lbdtm :datetime20.;
    format lbdtm datetime20.;
    datalines;
S001 ALT   15JAN2025:08:30:00
S001 AST   15JAN2025:14:15:30
S002 ALT   20FEB2025:09:00:00
S002 CREAT 20FEB2025:23:45:15
S003 ALT   01MAR2025:00:00:00
;
run;

*-----------------------------------------------------------------------------;
* Extract date portion, time portion, and individual components
*-----------------------------------------------------------------------------;
data lab_parts;
    set labs;

    * Extract the date part (SAS date value);
    lbdt = datepart(lbdtm);

    * Extract the time part (seconds since midnight);
    lbtm = timepart(lbdtm);

    * Extract individual time components;
    lb_hour   = hour(lbdtm);
    lb_minute = minute(lbdtm);
    lb_second = second(lbdtm);

    format lbdt date9. lbtm time8.;
run;

proc print data=lab_parts noobs;
run;

We create a small lab dataset with datetime stamps in the SAS datetime literal format: ddMONyyyy:hh:mm:ss.
datepart(lbdtm) extracts the date portion and returns a SAS date value (days since 01JAN1960) — ready for merging or date arithmetic.
timepart(lbdtm) extracts the time portion and returns the number of seconds since midnight — we format it with time8. for display.
hour(lbdtm), minute(lbdtm), and second(lbdtm) pull out individual time components as integers.
Subject S003 has a midnight timestamp (00:00:00); DATEPART still returns the correct date, and TIMEPART returns 0.
Subject S002's second record at 23:45:15 tests the late-night edge: HOUR returns 23, not rolling over to the next day.
The DATETIME20. informat reads the full datetime literal; if we only need the date, it is more efficient to store dates as SAS date values from the start.
We run PROC PRINT and confirm that the extracted date and time parts reconstruct the original datetime when combined.

R (tidyverse)+

# =============================================================================
# Extract date and time parts from a datetime value
# =============================================================================
library(tidyverse)
library(lubridate)
library(hms)

# -----------------------------------------------------------------------------
# Create sample lab data with datetime stamps
# -----------------------------------------------------------------------------
labs <- tribble(
  ~usubjid, ~lbtestcd, ~lbdtm_char,
  "S001",   "ALT",     "2025-01-15 08:30:00",
  "S001",   "AST",     "2025-01-15 14:15:30",
  "S002",   "ALT",     "2025-02-20 09:00:00",
  "S002",   "CREAT",   "2025-02-20 23:45:15",
  "S003",   "ALT",     "2025-03-01 00:00:00"
) %>%
  mutate(lbdtm = ymd_hms(lbdtm_char))

# -----------------------------------------------------------------------------
# Extract date portion, time portion, and individual components
# -----------------------------------------------------------------------------
lab_parts <- labs %>%
  mutate(
    # Extract the date part (Date object)
    lbdt = as_date(lbdtm),

    # Extract the time part as hms object
    lbtm = hms::as_hms(lbdtm),

    # Extract individual time components
    lb_hour   = hour(lbdtm),
    lb_minute = minute(lbdtm),
    lb_second = second(lbdtm)
  ) %>%
  select(-lbdtm_char)

print(lab_parts)

We parse datetime strings with ymd_hms() from lubridate, which creates POSIXct objects — the R equivalent of SAS datetime values.
as_date(lbdtm) extracts the date portion and returns a Date object, matching the behavior of SAS DATEPART.
hms::as_hms(lbdtm) extracts the time portion and returns an hms object that displays as "HH:MM:SS" — analogous to SAS TIMEPART with TIME8. format.
hour(lbdtm), minute(lbdtm), and second(lbdtm) return individual integer components, matching the SAS HOUR, MINUTE, and SECOND functions.
An alternative for the time string is format(lbdtm, "%H:%M:%S"), which returns a character string instead of an hms object.
The hms package is loaded separately because it is not part of core tidyverse, even though lubridate is often loaded alongside it.
We select out the original character column to keep the output clean, and print to confirm the extracted parts match the SAS output.

Happy learning!

Search lessons

Introduction

Extract Date and Time Parts from a Datetime Value