Such an identifier may be created by a publisher or archive, and there is no requirement that identifiers be unique.
Design Note: An element, rather than an attribute, was chosen as the means to capture DOIs and other identifiers so multiple identifiers could be retained.
<!ELEMENT object-id (#PCDATA) >
Text, numbers, or special characters
<abstract> Abstract; <boxed-text> Boxed Text; <chem-struct> Chemical Structure (Display); <chem-struct-wrapper> Chemical Structure Wrapper; <citation> Citation; <fig> Figure; <graphic> Graphic; <media> Media Object; <product> Product Information; <related-article> Related Article Information; <supplementary-material> Supplementary Material; <table-wrap> Table Wrapper
... <sec sec-type="results"> <title>Results</title> <p>We identified 703 abstracts, of which 687 were not relevant ... Details of randomisation procedures, treatment schedules, and numbers of patients followed up are given on the <italic>BMJ</italic> website.</p> <fig id="F1"> <caption> <p>Deaths among patients receiving day hospital care or alternative services. Odds ratios of death by end of follow up were calculated by fixed effects model. Heterogeneity between trials is presented as χ<sup>2</sup></p> </caption> <alt-text>Deaths in hospitals or with alternative services</alt-text> <graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="fors2662.f1" alt-version="yes"> <object-id>456472</object-id> </graphic> </fig> ... </sec> ...