← SASnRConvert Character Strings to Dates

Introduction

SDTM dates arrive as ISO 8601 character strings (e.g., "2025-01-15") and must be converted to proper date types before we can do arithmetic.
Other source systems may use different formats like "15JAN2025" or "01/15/2025", so we need to handle multiple patterns.
SAS uses the INPUT function paired with an informat that matches the character layout.
R tidyverse relies on lubridate helper functions named after the component order: ymd(), dmy(), mdy().
Base R uses as.Date() with a format string built from strftime tokens like %Y-%m-%d.
This lesson parses several common clinical date formats and shows how each language handles bad or unexpected input.

Convert Character Strings to Dates

SAS (Base SAS)+

*=============================================================================;
* Convert character strings to SAS date values
*=============================================================================;

*-----------------------------------------------------------------------------;
* Create sample data with character date strings
*-----------------------------------------------------------------------------;
data raw_dates;
    length iso_dt $10 sas_dt $9 us_dt $10;
    input iso_dt $ sas_dt $ us_dt $;
    datalines;
2025-01-15 15JAN2025 01/15/2025
2025-06-30 30JUN2025 06/30/2025
2025-12-01 01DEC2025 12/01/2025
;
run;

*-----------------------------------------------------------------------------;
* Parse each format using INPUT with the matching informat
*-----------------------------------------------------------------------------;
data parsed_dates;
    set raw_dates;

    * ISO format YYYY-MM-DD: use yymmdd10.;
    dt_from_iso = input(iso_dt, yymmdd10.);

    * SAS date9 format ddMONyyyy: use date9.;
    dt_from_sas = input(sas_dt, date9.);

    * US format MM/DD/YYYY: use mmddyy10.;
    dt_from_us = input(us_dt, mmddyy10.);

    format dt_from_iso dt_from_sas dt_from_us date9.;
run;

*-----------------------------------------------------------------------------;
* Handle bad data: INPUT with ?? suppresses error notes
*-----------------------------------------------------------------------------;
data bad_dates;
    length raw $10;
    input raw $;
    datalines;
2025-01-15
NOT_A_DATE
2025-13-01
;
run;

data parsed_bad;
    set bad_dates;
    dt = input(raw, ?? yymmdd10.);
    format dt date9.;
run;

proc print data=parsed_dates noobs;
run;
proc print data=parsed_bad noobs;
run;

We create three character date columns in different formats: ISO ("2025-01-15"), SAS date9 ("15JAN2025"), and US slash format ("01/15/2025").
input(iso_dt, yymmdd10.) reads the ISO format; the informat name describes the order: year-month-day, 10 characters wide.
input(sas_dt, date9.) reads the SAS-native ddMONyyyy format, which is common in CDISC-related displays.
input(us_dt, mmddyy10.) reads the US slash format; this informat handles both slash and dash separators.
Each INPUT call returns a SAS date value (days since 01JAN1960); we apply format date9. so it displays in human-readable form.
When a character string cannot be parsed, INPUT returns missing and writes a note to the log; adding ?? before the informat suppresses those notes.
We demonstrate bad data ("NOT_A_DATE" and an invalid month "2025-13-01") to show that both produce missing with ?? in place.
We run PROC PRINT on both datasets and confirm that the three valid dates all resolve to the same underlying value regardless of input format.

R (tidyverse)+

# =============================================================================
# Convert character strings to dates using lubridate
# =============================================================================
library(tidyverse)
library(lubridate)

# -----------------------------------------------------------------------------
# Create sample data with character date strings
# -----------------------------------------------------------------------------
raw_dates <- tribble(
  ~iso_dt,       ~sas_dt,      ~us_dt,
  "2025-01-15",  "15JAN2025",  "01/15/2025",
  "2025-06-30",  "30JUN2025",  "06/30/2025",
  "2025-12-01",  "01DEC2025",  "12/01/2025"
)

# -----------------------------------------------------------------------------
# Parse each format using lubridate helper functions
# -----------------------------------------------------------------------------
parsed_dates <- raw_dates %>%
  mutate(
    # ISO format YYYY-MM-DD
    dt_from_iso = ymd(iso_dt),

    # SAS-style ddMONyyyy format
    dt_from_sas = dmy(sas_dt),

    # US format MM/DD/YYYY
    dt_from_us = mdy(us_dt)
  )

# -----------------------------------------------------------------------------
# Handle bad data: lubridate returns NA with a warning
# -----------------------------------------------------------------------------
bad_strings <- c("2025-01-15", "NOT_A_DATE", "2025-13-01")
parsed_bad <- ymd(bad_strings)

print(parsed_dates)
print(parsed_bad)

Lubridate provides helper functions named after the date component order: ymd() for year-month-day, dmy() for day-month-year, mdy() for month-day-year.
ymd("2025-01-15") parses ISO format; lubridate is flexible about separators, so dashes, slashes, or even no separators all work.
dmy("15JAN2025") handles the SAS-style format; lubridate recognizes abbreviated month names in English automatically.
mdy("01/15/2025") reads US slash format; the function knows to look for month first, then day, then year.
All three functions return Date objects, so the results are directly comparable and ready for date arithmetic.
When a string cannot be parsed ("NOT_A_DATE" or an impossible date like month 13), lubridate returns NA and prints a warning — it does not stop execution.
For bulk parsing where the format might vary row to row, parse_date_time() accepts a vector of candidate formats and tries each in turn.
We print both results and confirm that all valid dates match across the three parsed columns.

Happy learning!

Search lessons

Introduction

Convert Character Strings to Dates