HLTH 2026Vorro is at HLTH 2026, Las Vegas, Nov 15–18. Bring your toughest interoperability problem.Book a meeting
HL7 v2Data Type⏱ 6 min read

HL7 TM Data Type: Time

In short

The TM data type encodes a time of day in HH[MM[SS[.SSSS]]] format with optional timezone offset; HL7-to-FHIR mapping converts it to the FHIR time type.

Need this HL7 v2 feed built or fixed?Contact Us

The TM (Time) data type has been part of HL7 v2 since v2.1 and represents a time of day with optional sub-second precision and an optional timezone offset. It is the type used by time-only components of scheduling segments (notably the time portion of SCH appointment fields where a profile splits date and time), pharmacy administration windows, and a handful of legacy fields where time-without-date is meaningful.

Purpose

TM carries a wall-clock time independent of any calendar date. When a field needs both date and time, the standard uses DTM; when only a time is meaningful - a daily recurrence ("administer at 09:00"), a shift boundary, or a time-of-day window - TM is the correct primitive. The optional offset lets implementers anchor the time to a specific zone where ambiguity matters.

Format and constraints

The TM format is HH[MM[SS[.S[S[S[S]]]]]][+/-ZZZZ], where:

  • HH - two-digit hour, 00–23, always present.
  • MM - optional two-digit minute, 00–59.
  • SS - optional two-digit second, 00–59 (leap second 60 is not supported by most parsers).
  • .S to .SSSS - optional decimal fraction of a second, 1 to 4 digits. Five or more decimal digits is non-conformant.
  • +/-ZZZZ - optional timezone offset from UTC, expressed as a four-digit HHMM. +0000 is the explicit UTC marker; -0500 is US Eastern Standard Time; +0530 is India Standard Time. Note that the offset is HHMM, not HH:MM - no colon.

Only digits, the decimal point, and the optional sign are allowed. No colons inside the time portion. No AM/PM. Hours use the 24-hour clock.

The precision-truncation rule applies: 09 means "9 AM, minute and second unknown," not "09:00:00." Padding with zeros loses precision information permanently.

Where it's used

TM is comparatively rare because most clinical events need a date too - but it appears in:

  • SCH-11 / SCH-12 - Appointment Timing Quantity sub-components where a profile carries the recurrence time separately from the date.
  • TQ1-8 - Start time / End time bounded recurrence patterns (in implementations that split out time-of-day).
  • RXA-3 time-of-day component in some legacy administration profiles.
  • PD1-12 time-only fields in patient additional demographic profiles (rare).
  • STF-23 / STF-24 office hours / on-call time-of-day in staff profiles.
  • Custom Z-segments and profile-specific extensions where a daily recurrence time is carried independently.

In practice, most time-of-day data in modern v2.5+ messages is carried as part of a DTM; pure TM fields are far less common than DT or DTM.

Version differences

  • v2.1 through v2.4 - TM defined as HHMM[SS[.SSSS]] with optional offset.
  • v2.5+ - Standard clarified to permit fractional seconds up to 4 decimal places and confirmed the offset is +/-ZZZZ with no colon.
  • v2.5 through v2.8.2 - No structural changes. The primitive is unchanged in v2.8.1.

Common mistakes

  • Sending ISO 8601 (10:15:30) where TM is expected. Colons are not legal in TM. The value must be 101530.
  • Including AM/PM. TM is strictly 24-hour; 0915 is 9:15 AM and 2115 is 9:15 PM.
  • Sending more than 4 fractional-second digits (.123456). The standard caps fractional precision at 4 digits.
  • Stripping the timezone offset during routing. A TM of 0900+0900 (Tokyo morning) silently becoming 0900 and being interpreted in the receiver's local zone produces a 9-, 14-, or 17-hour error.
  • Confusing +0000 (UTC) with empty offset. An absent offset means "unspecified zone, infer from context"; +0000 explicitly asserts UTC.

Examples

Minimal value:

1015          ; 10:15, seconds unknown

Hour-only:

10            ; 10 AM, minute and second unknown

Fully-populated with fractional seconds and offset:

101530.5+0000     ; 10:15:30.5 UTC
094512.2500-0500  ; 09:45:12.25 US Eastern Standard

Precision-truncated examples:

09            ; hour only
0915          ; hour and minute
091500        ; hour, minute, second
091500.5      ; with one-decimal fractional second

In context - a SCH segment carrying a time-of-day recurrence:

SCH|APT-99231|||1|||||30|min^minute^UCUM|^^30^20260625090000-0500^^^^^090000-0500

Common pitfall - colon-separated and AM-marked:

0915                ; correct
09:15               ; WRONG: colons are not legal in TM
0915AM              ; WRONG: TM is 24-hour, no AM/PM marker

FHIR mapping

The v2-to-FHIR IG does not publish a per-datatype ConceptMap for TM. The mapping is implicit: TM maps to the FHIR time primitive, which uses the form HH:MM:SS[.sss]. The conversion is mechanical - insert colons between HH/MM/SS and shift the decimal accordingly:

  • 09 ↔ 09:00:00 (with a precision-loss caveat - FHIR time requires HH:MM:SS at minimum)
  • 0915 ↔ 09:15:00
  • 091530.5 ↔ 09:15:30.5

FHIR time does not carry a timezone offset. When a TM has an explicit offset (091500-0500), the converter must either drop the offset, move it to a sibling timezone field on the parent resource, or promote the value to a dateTime if a date is available.

Engine considerations

  • Precision preservation - store TM with an explicit precision flag (HOUR, MINUTE, SECOND, FRACTION) so downstream code does not silently pad to HH0000.
  • Timezone offsets - preserve the offset across the routing pipeline. Engines that strip offsets to "normalize" times destroy information that cannot be recovered.
  • 24-hour clock - TM is always 24-hour; reject AM/PM markers at parse time rather than guessing.
  • Fractional seconds cap - validate at most 4 fractional digits; truncate-and-warn rather than silently rounding when senders exceed this.
  • TM into a DTM-typed field - when a sender places a TM where a DTM is expected, most engines will reject the field. Do not synthesize a date.

How Vorro parses and produces TM

Vorro's HL7 parser stores TM as a precision-tagged time-of-day with an optional explicit offset. On inbound, the parser validates that the value contains only digits, an optional single decimal point, and an optional +/-HHMM offset; non-conformant inputs route to a curation queue. Fractional seconds beyond 4 digits are truncated and a soft warning is logged against the source. Offsets are always preserved - Vorro does not "normalize away" timezone information at the boundary because doing so corrupts cross-site scheduling.

On outbound, Vorro emits TM in the exact precision originally captured. When the source had an explicit offset, the outbound TM carries the same offset; when the source did not, no offset is invented. When converting to FHIR time, Vorro inserts colons mechanically and, if an offset is present, moves it to a paired timezone extension on the surrounding resource rather than dropping it.

Sources

← Back to HL7 v2 Guide

Ready to Integrate This Into Your Workflow?

Talk to a Vorro expert about implementing HL7 v2 in your specific environment.