*Copyright @ www.mycsg.in;

How SAS stores dates

SAS stores dates as ordinary numeric values — specifically the number of days elapsed since 01 January 1960
The date 01 January 1960 itself is stored as 0, dates before it are negative, and dates after it are positive
Because a SAS date is just a number you can add, subtract, and compare dates using ordinary arithmetic
A date has no inherent display format — the same numeric value looks like a plain integer until you attach a date format
Understanding this storage model is the foundation for all date work in SAS

Demonstrate the underlying numeric value

The DATA step below assigns a known date and prints it twice: once as a raw number and once with a DATE9. format applied
Both variables hold exactly the same value — only the display changes when a format is attached

data dates_demo;
    date_raw   = '01JAN2020'D;
    date_fmt   = '01JAN2020'D;
    format date_fmt date9.;
run;

Inspect the output dataset — `date_raw` will display as a large integer (21915) and `date_fmt` will show as 01JAN2020
Count manually: 2020 minus 1960 is 60 years, which is approximately 21915 days, confirming the storage model

Date literals

A date literal is the most direct way to write a fixed date in SAS code
The syntax is a date string in single quotes followed by the letter D: `'ddMONyyyy'D`
The month must be the three-letter English abbreviation in any mix of upper and lower case
SAS converts the literal to the corresponding numeric day-count at compile time

/*
date1 = '01JAN2020'D;   * 01 January 2020;
date2 = '15mar2023'D;   * 15 March 2023  - lowercase month is also valid;
date3 = '31DEC1999'D;   * 31 December 1999;
*/

Common date informats - reading dates into SAS

An informat tells SAS how to interpret a character string and convert it to the internal numeric date value
The most common date informats are listed below — the width in the informat name matches the expected character width of the source string

/*
DATE9.     reads  '01JAN2020'  (ddMONyyyy, 9 characters)
DATE7.     reads  '01JAN20'    (ddMONyy,   7 characters)
YYMMDD10.  reads  '2020-01-01' (yyyy-mm-dd, 10 characters with separator)
YYMMDD8.   reads  '20200101'   (yyyymmdd,   8 characters no separator)
DDMMYY10.  reads  '01/01/2020' (dd/mm/yyyy, 10 characters with separator)
MMDDYY10.  reads  '01/15/2020' (mm/dd/yyyy, 10 characters with separator)
*/

Use the informat that matches the exact format of the incoming data — width and separator must align
If the source data is already numeric (a SAS date stored as a number) no informat is needed

Reading dates from character variables using INPUT

The `input()` function converts a character value to numeric using a named informat
The resulting variable is a SAS date and can be formatted and used in date arithmetic

data subjects;
    infile cards truncover;
    input usubjid $ rfstdtc $10.;
cards;
1001 2023-01-15
1002 2023-02-20
1003 2023-03-05
1004 2023-04-10
;
run;
 
data subjects;
    set subjects;
    rfstdt = input(rfstdtc, yymmdd10.);
    format rfstdt date9.;
run;

The variable `rfstdtc` is a character string in YYYY-MM-DD format, typical of ISO 8601 dates used in clinical data standards
The `input()` call with `yymmdd10.` converts the string to a SAS date number and stores it in `rfstdt`
The `format` statement attaches DATE9. so the value displays in a readable form — the underlying number is unchanged
Verify the output: `rfstdt` should show dates like 15JAN2023 beside the original character string

Common date formats - displaying SAS dates

A format controls how a SAS date number is displayed — it does not change the stored value
Multiple formats can be applied to the same variable for different outputs

/*
DATE9.     displays  01JAN2020
DATE7.     displays  01JAN20
YYMMDD10.  displays  2020-01-01
YYMMDD8.   displays  20200101
DDMMYY10.  displays  01/01/2020
MMDDYY10.  displays  01/15/2020
WORDDATE.  displays  January 1, 2020
WEEKDATE.  displays  Wednesday, January 1, 2020
MONNAME.   displays  January
YEAR4.     displays  2020
*/

The format to use depends on the output destination — DATE9. is common in SAS reports, YYMMDD10. is standard for ISO-compliant outputs
Always attach a format before printing or exporting dates to make them human-readable

Applying different formats to the same date value

data formats_demo;
    dt = '15MAR2023'D;
 
    dt_date9   = dt;  format dt_date9   date9.;
    dt_yymmdd  = dt;  format dt_yymmdd  yymmdd10.;
    dt_ddmmyy  = dt;  format dt_ddmmyy  ddmmyy10.;
    dt_word    = dt;  format dt_word    worddate.;
    dt_year    = dt;  format dt_year    year4.;
run;

All five variables hold the same numeric value — the output dataset will show different display strings side by side
Confirm that each column shows the expected date representation for 15 March 2023

Core date functions

TODAY() and DATE() - get the current date

`today()` returns today's date as a SAS date number — it takes no arguments
`date()` is an alias for `today()` and behaves identically
Both functions are commonly used to calculate elapsed time or flag records relative to the current date

data current_date;
    today_val = today();
    date_val  = date();
    format today_val date_val date9.;
run;

Both `today_val` and `date_val` should show the same date when you run this step

MDY() - construct a date from year, month, and day components

`mdy(month, day, year)` returns the SAS date for the given calendar components
The argument order is month first, then day, then year — note this is not the same as the display order of most date formats
`mdy` is useful when the input data has separate numeric year, month, and day columns that need to be combined into a single date variable

data mdy_demo;
    infile cards;
    input usubjid $ birth_year birth_month birth_day;
cards;
1001 1985 4 22
1002 1990 11 3
1003 1978 7 15
;
run;
 
data mdy_demo;
    set mdy_demo;
    brthdtc = mdy(birth_month, birth_day, birth_year);
    format brthdtc date9.;
run;

The three numeric components are combined into the single SAS date `brthdtc`
Verify that the displayed dates match the year, month, and day columns in the input

YEAR(), MONTH(), DAY() - extract components from a date

`year(date)` returns the four-digit year as a number
`month(date)` returns the month number 1 through 12
`day(date)` returns the day of the month 1 through 31
These functions are useful when you need to compare or group records by a specific date component

data subjects;
    set subjects;
    rfst_year  = year(rfstdt);
    rfst_month = month(rfstdt);
    rfst_day   = day(rfstdt);
run;

The three new columns extract the year, month number, and day from `rfstdt` which was derived earlier
Confirm that the extracted values match the visual date displayed in `rfstdt`

Date arithmetic

Difference in days between two dates

Because SAS dates are integers you can subtract one from another to get the number of days between them
The result is simply the numeric difference — no special function is required for day-level differences
A common clinical use is calculating the number of days a subject has been on study

data subjects;
    set subjects;
    cutoff_dt  = '30JUN2023'D;
    days_on_study = cutoff_dt - rfstdt;
    format cutoff_dt date9.;
run;

`days_on_study` is the integer number of days from `rfstdt` to the cutoff date
Verify the values by mentally counting the approximate months between the start dates and 30 June 2023

Adding and subtracting a fixed number of days

To shift a date forward or backward by a known number of days, simply add or subtract that integer
For example, adding 7 gives the date one week later; subtracting 30 gives the date approximately one month earlier
This works directly because SAS dates are stored as day counts

data subjects;
    set subjects;
    visit1_due  = rfstdt + 7;    * one week after start date;
    screening_dt = rfstdt - 14;  * two weeks before start date;
    format visit1_due screening_dt date9.;
run;

Inspect `visit1_due` — it should be exactly 7 days after `rfstdt` for every subject
Inspect `screening_dt` — it should be exactly 14 days before `rfstdt`

Key points to remember

SAS dates are stored as integers counting days since 01 January 1960 — they are plain numbers not text
Attach an informat when reading character date strings into SAS and a format when displaying or exporting dates
The most common informats are DATE9. YYMMDD10. DDMMYY10. and MMDDYY10. — choose the one that matches your source data layout
`today()` returns the current date, `mdy()` builds a date from components, and `year()` `month()` `day()` extract components
Day-level differences and offsets use plain subtraction and addition because dates are integers
Part 02 of this topic covers interval functions `intck()` and `intnx()` for month-level and year-level date calculations

*Copyright @ www.mycsg.in;

How SAS stores dates

Demonstrate the underlying numeric value

SAS Log

Dataset View

Date literals

SAS Log

Common date informats - reading dates into SAS

SAS Log

Reading dates from character variables using INPUT

SAS Log

Dataset View

Common date formats - displaying SAS dates

SAS Log

Applying different formats to the same date value

SAS Log

Dataset View

Core date functions

TODAY() and DATE() - get the current date

SAS Log

Dataset View

MDY() - construct a date from year, month, and day components

SAS Log

Dataset View

YEAR(), MONTH(), DAY() - extract components from a date

SAS Log

Dataset View

Date arithmetic

Difference in days between two dates

SAS Log

Dataset View

Adding and subtracting a fixed number of days

SAS Log

Dataset View

Key points to remember