*Copyright @ www.mycsg.in;

Data Step - Dates Part 02: DATETIME Values, DATEPART, TIMEPART, and Datetime Arithmetic

Overview

• Part 01 (SAS_DATASTEP_L201) covered SAS date values: integers representing the number of days since 01-JAN-1960, date literals, informats, date component functions, and day arithmetic.
• This lesson introduces the related but distinct concept of datetime values in SAS, which store both a date and a time of day in a single numeric value.
• Datetime values are stored as the number of seconds elapsed since midnight on 01-JAN-1960. They are used when you need to record or calculate the exact time of an event, not just the date.
• Typical use cases in clinical data: adverse event start and stop datetime, sample collection datetime, ECG or vital sign recording time.

SAS Datetime Values and Literals

• A SAS datetime value is a numeric variable that counts seconds from midnight 01-JAN-1960 (not days).
• The relationship between a date value and a datetime value is: datetime = date_value * 86400 + seconds_since_midnight.
• Datetime literals in SAS code use single quotes with the suffix DT: '01JAN2024:08:30:00'dt
• The format of the datetime literal is: 'DDMONYYYY:HH:MM:SS'dt — note the colon separating date and time portions.
• To display a datetime value in a readable form, apply a DATETIME format (e.g., DATETIME20.) or a DATE9./TIME8. format after extracting the date or time portion.

data work.datetime_demo; * Datetime literal ; dt1 = '15MAR2024:09:30:00'dt; dt2 = '15MAR2024:14:45:30'dt; dt3 = '16MAR2024:07:15:00'dt; * Display formats ; format dt1 dt2 dt3 datetime20.; run; proc print data=work.datetime_demo; run; Copy Code View Log

• DATETIME20. formats the numeric datetime value as a readable date-time string (e.g., 15MAR2024:09:30:00).
• Inspect WORK.DATETIME_DEMO to confirm the three datetime values display correctly.
• The raw numeric value of dt1 will be a very large integer (seconds since 01-JAN-1960) — the format makes it human-readable.
• Always apply a DATETIME format when displaying datetime variables; without a format the raw second count is meaningless to a reader.

View Data

Reading Datetime Values with Informats

• When reading datetime strings from raw data or character variables, use a datetime informat to convert them to SAS datetime values.
• Common datetime informats: DATETIME20. for 'DDMONyyyy:HH:MM:SS' strings, YMDDTTM. for ISO 8601 format strings like '2024-03-15T09:30:00'.
• The INPUT function with a datetime informat converts a character datetime string to a SAS numeric datetime value.

data work.events; input subject $ event_dtc $20.; datalines; 001 15MAR2024:09:30:00 002 22APR2024:14:00:00 003 01MAY2024:08:15:30 ; run; data work.events_parsed; set work.events; event_dt = input(event_dtc, datetime20.); format event_dt datetime20.; run; Copy Code View Log

• event_dtc is read as character ($20.) because the raw data is a text string.
• INPUT with DATETIME20. converts the character string to a SAS datetime value stored in event_dt.
• Confirm WORK.EVENTS_PARSED shows event_dt as a formatted datetime, not a raw number — if it shows a large integer the FORMAT statement may be missing.

View Data

DATEPART() and TIMEPART() Functions

• A SAS datetime value encodes both date and time. Two functions extract each component separately.
• DATEPART(datetime) returns the SAS date value (integer days) corresponding to the date portion of the datetime. Apply a DATE format to display it.
• TIMEPART(datetime) returns the SAS time value (seconds since midnight) corresponding to the time portion of the datetime. Apply a TIME format to display it.
• Once extracted, the date portion is a normal SAS date value and can be used with all date functions (YEAR, MONTH, DAY, INTCK, INTNX, etc.).

data work.extracted; set work.events_parsed; event_date = datepart(event_dt); event_time = timepart(event_dt); event_year = year(event_date); event_hour = hour(event_time); format event_date date9. event_time time8.; run; Copy Code View Log

• event_date is a SAS date value — apply DATE9. to display it as 15MAR2024.
• event_time is a SAS time value (seconds since midnight) — apply TIME8. to display it as HH:MM:SS.
• YEAR(event_date) extracts the 4-digit year; HOUR(event_time) extracts the hour of day as an integer.
• Check WORK.EXTRACTED to confirm all four derived variables have the expected values for each event row.

View Data

Datetime Arithmetic

• Because datetime values are stored in seconds, datetime arithmetic is performed in seconds — unlike date arithmetic which is in days.
• To find elapsed time between two datetime values: elapsed_seconds = end_dt - start_dt.
• Divide by 3600 to convert to hours, by 60 to convert to minutes, or by 86400 to convert to days.
• This is essential for calculating duration of adverse events, time to first response, or time from dose to sample collection.

data work.durations; input subject $ start_dtc $20. end_dtc $20.; datalines; 001 15MAR2024:09:30:00 15MAR2024:14:45:00 002 22APR2024:08:00:00 23APR2024:08:00:00 003 01MAY2024:07:15:00 03MAY2024:19:30:00 ; run; data work.elapsed; set work.durations; start_dt = input(start_dtc, datetime20.); end_dt = input(end_dtc, datetime20.); elapsed_sec = end_dt - start_dt; elapsed_min = elapsed_sec / 60; elapsed_hr = elapsed_sec / 3600; elapsed_days = elapsed_sec / 86400; format start_dt end_dt datetime20. elapsed_hr 8.2 elapsed_days 8.3; drop start_dtc end_dtc; run; Copy Code View Log

• Both start_dtc and end_dtc are character strings — INPUT with DATETIME20. converts each to a SAS datetime value.
• The arithmetic is straightforward because both values are in the same unit (seconds).
• Subject 001: start to end is 5 hours 15 minutes = 18900 seconds = 5.25 hours.
• Subject 002: exactly 24 hours = 86400 seconds = 1 day.
• Subject 003: 2 days 12 hours 15 minutes — verify elapsed_days is approximately 2.51.
• Inspect WORK.ELAPSED and confirm the elapsed values match expected calculations.

View Data

Converting Between Date and Datetime

• Sometimes you need to convert a SAS date value to a datetime value — for example to compare a date against a datetime in a merge or WHERE condition.
• Multiply a date value by 86400 to produce the midnight datetime for that date: datetime_at_midnight = date_value * 86400.
• Conversely, DATEPART() extracts the date from a datetime as shown above.
• Never compare a date value directly to a datetime value without conversion — they use different scales (days vs seconds) and the comparison will produce incorrect results.

data work.conversion_demo; date_val = '15MAR2024'd; dt_midnight = date_val * 86400; dt_noon = dt_midnight + (12 * 3600); back_to_date = datepart(dt_noon); format date_val date9. dt_midnight dt_noon datetime20. back_to_date date9.; run; proc print data=work.conversion_demo; run; Copy Code View Log

• DT_MIDNIGHT is the datetime value for 15MAR2024 at 00:00:00.
• Adding 12 * 3600 = 43200 seconds gives the datetime for 15MAR2024 at 12:00:00 (noon).
• DATEPART(dt_noon) recovers the original date value — confirm it equals date_val.

Key Points

• SAS datetime values are seconds since midnight 01-JAN-1960 — distinct from date values which are days since the same epoch.
• Datetime literals use single quotes with the DT suffix: '15MAR2024:09:30:00'dt.
• DATEPART() and TIMEPART() split a datetime into its date and time components respectively — apply DATE and TIME formats to display them correctly.
• Datetime arithmetic is in seconds: divide by 3600 for hours, 86400 for days to convert to meaningful units.
• To convert a SAS date to a datetime, multiply by 86400 to get the midnight datetime for that date.
• Never compare a SAS date directly to a SAS datetime — convert one to the other's scale first.