← SASnRCalculate the Difference Between Two Dates

Introduction

Clinical programming constantly needs the gap between two dates — study day derivation (ASTDY), treatment duration, time between adverse events.
The simplest case is a day count, but month and year intervals matter too — for example, cycles of chemotherapy or annual lab windows.
SAS uses the INTCK function with an interval keyword; R works with difftime or lubridate arithmetic.
Study day (ASTDY) is one of the most common derivations in ADaM: it counts the days from the reference start date to an event date, skipping day zero.
This lesson derives study day and treatment duration from a small adverse-event dataset with reference start dates.

Calculate the Difference Between Two Dates

SAS (Base SAS)+

*=============================================================================;
* Calculate the difference between two dates using INTCK
*=============================================================================;

*-----------------------------------------------------------------------------;
* Create sample adverse event data with reference start dates
*-----------------------------------------------------------------------------;
data ae;
    input usubjid $ rfstdtc :date9. aestdtc :date9. aeendtc :date9.;
    format rfstdtc aestdtc aeendtc date9.;
    datalines;
S001 01JAN2025 05JAN2025 12JAN2025
S001 01JAN2025 15FEB2025 20FEB2025
S002 10FEB2025 10FEB2025 10MAR2025
S002 10FEB2025 25MAR2025 30APR2025
S003 01MAR2025 15MAR2025 15MAR2025
;
run;

*-----------------------------------------------------------------------------;
* Derive study day, AE duration, and month interval
*-----------------------------------------------------------------------------;
data ae_derived;
    set ae;

    * Day count between two dates;
    day_diff = intck('DAY', rfstdtc, aestdtc);

    * Study day: add 1 when the event is on or after reference date;
    if aestdtc >= rfstdtc then astdy = day_diff + 1;
    else astdy = day_diff;

    * Duration of adverse event in days;
    ae_dur = intck('DAY', aestdtc, aeendtc) + 1;

    * Month count between reference and AE start;
    month_diff = intck('MONTH', rfstdtc, aestdtc);

    * Year count between reference and AE start;
    year_diff = intck('YEAR', rfstdtc, aestdtc);
run;

proc print data=ae_derived noobs;
run;

We create a small AE dataset with three subjects, each having a reference start date (rfstdtc), AE start date (aestdtc), and AE end date (aeendtc).
intck('DAY', rfstdtc, aestdtc) counts the number of day boundaries crossed between the two dates — this is a simple subtraction for days.
To derive study day (ASTDY), we add 1 when the event is on or after the reference date, because CDISC study day has no day zero.
For events before the reference date, the day difference is already negative, so we leave it as is — day −1 is the day before reference start.
intck('DAY', aestdtc, aeendtc) + 1 gives the duration of the AE in days; we add 1 because both the start and end day are counted.
intck('MONTH', rfstdtc, aestdtc) counts how many first-of-month boundaries the interval crosses — it does not count partial months.
intck('YEAR', rfstdtc, aestdtc) works the same way for year boundaries.
We run PROC PRINT and confirm that subject S002's first AE on the same day as reference start gives ASTDY = 1, not 0.

R (tidyverse)+

# =============================================================================
# Calculate the difference between two dates using lubridate
# =============================================================================
library(tidyverse)
library(lubridate)

# -----------------------------------------------------------------------------
# Create sample adverse event data with reference start dates
# -----------------------------------------------------------------------------
ae <- tribble(
  ~usubjid, ~rfstdtc,     ~aestdtc,     ~aeendtc,
  "S001",   "2025-01-01", "2025-01-05", "2025-01-12",
  "S001",   "2025-01-01", "2025-02-15", "2025-02-20",
  "S002",   "2025-02-10", "2025-02-10", "2025-03-10",
  "S002",   "2025-02-10", "2025-03-25", "2025-04-30",
  "S003",   "2025-03-01", "2025-03-15", "2025-03-15"
) %>%
  mutate(across(rfstdtc:aeendtc, ymd))

# -----------------------------------------------------------------------------
# Derive study day, AE duration, and month interval
# -----------------------------------------------------------------------------
ae_derived <- ae %>%
  mutate(
    # Day count between two dates
    day_diff = as.numeric(difftime(aestdtc, rfstdtc, units = "days")),

    # Study day: add 1 when the event is on or after reference date
    astdy = if_else(aestdtc >= rfstdtc, day_diff + 1, day_diff),

    # Duration of adverse event in days
    ae_dur = as.numeric(difftime(aeendtc, aestdtc, units = "days")) + 1,

    # Month count between reference and AE start
    month_diff = interval(rfstdtc, aestdtc) %/% months(1),

    # Year count between reference and AE start
    year_diff = interval(rfstdtc, aestdtc) %/% years(1)
  )

print(ae_derived)

We parse the character date columns into Date objects with ymd() from lubridate, applied across all three date columns at once using across().
difftime(aestdtc, rfstdtc, units = "days") returns a difftime object; we wrap it with as.numeric() to get a plain number.
Study day follows the same rule as SAS: add 1 when the event date is on or after the reference start date.
AE duration adds 1 to the day difference to count both the start and end day, matching the SAS derivation.
interval(rfstdtc, aestdtc) %/% months(1) performs integer division of a lubridate interval by one month — this counts whole calendar months, just like SAS INTCK('MONTH').
The same pattern works for years: interval() %/% years(1) counts whole calendar years elapsed.
We print the result and confirm that subject S002's first AE on the same day as reference start gives astdy = 1.

Happy learning!

Search lessons

Introduction

Calculate the Difference Between Two Dates