SAS Date Values

From its inception, the SAS System has stored date values as an offset in days from January 1, 1960. Leap years, century, and fourth-century adjustments are made automatically. Leap seconds are ignored, and the SAS System does not adjust for daylight saving time. This method of date representation means that calculations and comparisons of SAS date values will produce correct results, regardless of century.

SAS users can convert external data to or from SAS date values by the use of various informats, formats, and functions.

Reading raw data into SAS date variables

Raw data can be read into SAS date variables using the appropriate informat. For example:

data temp;
input date yymmdd6.;

If you want to be able to understand printouts of these dates, it is necessary to assign an appropriate format to the variable. For example:

format date yymmdd6.;

Converting character or numeric variables to SAS date variables

This can be done using the INPUT function. The following code extracts date of birth from PNR and writes it out as a SAS date variable (click here for a complete example):


If the values of year, month, and day are stored in separate variables, these can be written to a single SAS date variable using the MDY function:

sasdate=mdy(month, day, year);

Two-digit years (the YEARCUTOFF= option)

SAS date informats, formats, and functions all accept two-digit years as well as four-digit years. If the dates in your external data sources contain four-digit years, then the SAS System will accept and display those four-digit years without any difficulty as long as you choose the appropriate informat and format.

If dates in your external data sources or SAS program statements contain two-digit years, you can specify the century prefix assigned to them by using the YEARCUTOFF= system option. The YEARCUTOFF= option specifies the first year of the 100-year span that is used to determine the century of a two-digit year. The default value of YEARCUTOFF in version 6.12 is 1900, implying that all two-digit years are assumed to be in the 1900's. If you are working with a study where the last date of follow-up is 1992, but some individuals in your study were born in the late 1800's, you may wish to set the YEARCUTOFF option to 1893. This would lead to SAS interpreting values for year between 93 and 99 as being in the 1800's.

Let's consider an example of reading in dates with both two-digit and  four-digit years. Note that in this example the YEARCUTOFF= option has been set to 1920. (the code can be downloaded here)

options yearcutoff=1920;

data schedule;
input @1 rawdata $8. @1 date yymmdd8.;

proc print;
format date yymmdd10.;

OBS    RAWDATA           DATE

 1     651128      1965-11-28
 2     19651128    1965-11-28
 3     18230314    1823-03-14
 4     19131225    1913-12-25
 5     131225      2013-12-25

Note that the dates in observations 1 and 2 are the same (a two-digit date of 65 defaults to 1965), but the dates in observations 4 and 5 are different (a two-digit date of 13 defaults to 2013).