The practical way to distinguish a UDI device identifier (DI) from production identifiers (PI) is to ask whether the segment describes the model or the instance. Under 21 CFR 801.3 and 830.3, the DI is the mandatory, fixed portion identifying the labeler and the specific version or model; the PI is the conditional, variable portion identifying lot or batch, serial number, expiration date, manufacturing date, or, for certain HCT/Ps, a distinct identification code [2][3][4]. The same distinction reappears in FDA guidance: a UDI is composed of a DI and a PI, and the DI is the lookup key in the Global Unique Device Identification Database [2][3][6]. The two segments can appear adjacent on a label, so separating them requires attention to issuing agency syntax rather than positional assumptions alone [1][3].

Regulatory Definitions Set the First Boundary

The regulatory language gives a stable separation rule. FDA defines the device identifier as the mandatory, fixed portion of a UDI that identifies the specific version or model of a device and the labeler of that device [3][4]. The production identifier is defined as the conditional, variable portion that identifies one or more items only when they appear on the label: lot or batch, serial number of a specific device, expiration date, manufacturing date, and, for an HCT/P regulated as a device, the distinct identification code required by 21 CFR 1271.290(c) [3][4][8].

This fixed-versus-conditional contrast explains many observed label differences. A DI changes when the version or model attributes change within limits set by the labeler; the PI changes across production runs or individual units [4][5]. FDA adds that a UDI may comprise a DI only in some situations, such as a class II device that carries none of the five PI elements on its label, while a class I device is not required to include a PI at all [3]. The labeler may still voluntarily include a PI on a class I device [3].

GUDID Stores the DI, Not the PI

Database behavior offers a second way to separate the concepts. GUDID contains only the device identifier, which is the primary key for looking up device information; production identifiers are not submitted to or stored in GUDID [2][6]. The GUDID guidance repeats the equation UDI equals DI plus PI and states that the DI, together with associated data attributes, constitutes a DI record [6].

Because the PI is absent from GUDID, someone attempting to reconcile a scanned label against the database should expect the model-level DI attributes to match while lot, serial, and date values remain outside the record [6][7]. This also means the DI is the durable data key for reporting and master data exchange, whereas the PI is transaction-level information captured at the point of use. The distinct identification code for HCT/Ps may take the form of a donation identification number, serial number, lot number, or a combination of these production identifiers [6].

H3: A parsing checklist for label review

A suggested field review sequence for regulatory affairs or data operations staff is: first, identify the issuing agency from the label or package; second, locate the agency-specific delimiter for the device identifier, such as GS1 (01), HIBCC +, or ICCBBA =/; third, treat that DI segment as the fixed, model-level key and verify it against the GUDID DI record; fourth, classify each remaining UDI segment by its delimiter as lot, serial, expiration date, manufacturing date, or distinct identification code; and fifth, confirm that no non-UDI element has been inserted into either segment [1][2][3][6].

This sequence is illustrative, not a substitute for the issuing agency's published guidelines. Labelers must submit data based on the issuing agency's guidelines, and FDA provides links to check digit calculators and device identifier checker tools for GS1, HIBCC, and ICCBBA [1][6].

Issuing Agency Delimiters Mark the Segment Boundary

The three FDA-accredited issuing agencies use different data delimiters, and those delimiters are the most concrete label-level signal. GS1 uses application identifiers such as (01) for the device identifier, (17) for expiration date, (10) for batch or lot, and (21) for serial number [1]. The FDA document includes the example (01)51022222233336(11)141231(17)150707(10)A213B1(21)1234, where the fixed (01) segment is the DI and the later segments are production identifiers [1].

HIBCC and ICCBBA use different syntax. HIBCC employs a plus sign based delimiter set for the DI and additional delimiters such as $, $$+4, $$+5, or $$+6 for lot, expiration date plus serial number combinations, and /S for a supplemental serial number [1]. ICCBBA examples use =/ for the device identifier, =, for serial number, = for the distinct identification code, => for expiration date, =} for manufacturing date, and &,1 for MPHO lot number [1]. A suggested workflow is to identify the delimiter prefix first, then classify each segment, because visual adjacency of a DI and PI can otherwise hide the boundary [1][3].

Plain-Text and AIDC Forms Must Carry Both Segments

The label itself must present the UDI in easily readable plain-text and in AIDC form [2][3]. Under 21 CFR 801.40(b), both forms must include a device identifier segment and a production identifier segment that conveys one or more of the identifiers enumerated in the PI definition when included on the device label [3]. FDA advises that the DI segment and the PI segment cannot include non-UDI elements or data delimiters for non-UDI elements [3].

Data delimiters perform the parsing function across both forms. They allow users to parse the DI and PI in plain text and to verify that the information encoded in the AIDC form matches the easily readable plain-text form, and they enable parsed entry into electronic systems after scanning [3]. This is not merely a formatting preference: the delimiter structure is what lets a scanner send the model-level DI to the record while retaining the instance-level PI elements for clinical documentation [3].

Worked Example and Situations Where PI Is Optional

Consider a suggested hypothetical evaluation of a GS1 plain-text string on a fictitious label: separate every parenthetical application identifier, then assign (01) to the DI bucket and (10), (17), and (21) to the PI bucket [1][3]. If a class II label carries only a lot number, the PI is that lot segment; if the same model label has no PI elements, the UDI may be DI-only [3]. The fixed portion remains stable across those variations, which is the central test for distinguishing the two concepts [2][3].

The optional nature of the PI does not blur the definitions. FDA specifies that a class I device is not required to include a PI, but the labeler may voluntarily include one, while the UDI still contains the mandatory DI [3]. A device may bear both a Universal Product Code and a UDI under the voluntary labeling provisions, but the UPC is a separate retail product identifier and does not change the internal DI-plus-PI composition [4][7]. The boundary therefore rests on regulatory definitions and delimiter syntax, not on the number of symbols on the label [1][3][4].