A record of an event relevant for purposes such as operations, privacy, security, maintenance, and performance analysis.

Based on IHE-ATNA.

Partitions the audit event into one or more categories that can be used to filter searching, to govern access control and/or to guide system behavior.

Categorization might be done automatically (inferred by code) or manually by user assertion. The absence of a category may limit the ability to determine when the element should be handled, so strong consideration should be given to how systems will be able to determine category values for legacy data and how data that cannot be categorized will be handled. As well, some categories might not be mutually exclusive, so systems should prepare for multiple declared categories - even within a single category 'axis'. In general, there should not be a 'strong' binding ('required' or 'extensible') on the category element overall. Instead, the element can be sliced and bindings can be asserted that apply to particular repetitions.

Describes what happened. The most specific codes for the event.

This field enables queries of messages by implementation-defined event categories.

Indicator for type of action performed during the event that generated the audit.

This broadly indicates what kind of action was done on the AuditEvent.entity by the AuditEvent.agent.

Indicates and enables segmentation of various severity including debugging from critical.

ATNA will map this to the SYSLOG PRI element.

The time or period during which the activity occurred.

The time or period can be a little arbitrary; where possible, the time should correspond to human assessment of the activity time.

The time when the event was recorded.

This ties an event to a specific date and time. Security audits typically require a consistent time base (e.g. UTC), to eliminate time-zone issues arising from geographical distribution.

In a distributed system, some sort of common time base (e.g. an NTP [RFC1305] server) is a good implementation tactic.

Indicates whether the event succeeded or failed. A free text descripiton can be given in outcome.text.

when a code is given there must be one code from the given codeSystem, and may be other equivilant codes from other codeSystems (for example http response codes such as 2xx, 4xx, or 5xx).

In some cases a "success" may be partial, for example, an incomplete or interrupted transfer of a radiological study. For the purpose of establishing accountability, these distinctions are not relevant.

Indicates whether the event succeeded or failed.

when a code is given there must be one code from the given codeSystem.

In some cases a "success" may be partial, for example, an incomplete or interrupted transfer of a radiological study. For the purpose of establishing accountability, these distinctions are not relevant.

Additional details about the error. This may be a text description of the error or a system code that identifies the error.

A human readable description of the error issue SHOULD be placed in details.text.

The authorization (e.g., PurposeOfUse) that was used during the event being recorded.

Record of any relevant security context, not restricted to purposeOfUse valueSet. May include security compartments, refrain, obligation, or other security tags.

Use AuditEvent.agent.authorization when you know that it is specific to the agent, otherwise use AuditEvent.authorization. For example, during a machine-to-machine transfer it might not be obvious to the audit system who caused the event, but it does know why.

Allows tracing of authorizatino for the events and tracking whether proposals/recommendations were acted upon.

Allows tracing of authorization for the audit event and tracking whether proposals/recommendations were acted upon.

The patient element is available to enable deterministic tracking of activities that involve the patient as the subject of the data used in an activity.

When the .patient is populated it shall be accurate to the subject of the used data. The .patient shall not be populated when the used data used/created/updated/deleted (.entity) by the activity does not involve a subject. Note that when the patient is an agent, they will be recorded as an agent. When the Patient resource is Created, Updated, or Deleted it will be recorded as an entity. May also affect access control.

This will typically be the encounter the event occurred, but some events may be initiated prior to or after the official completion of an encounter but still be tied to the context of the encounter (e.g. pre-admission lab tests).

Links the audit event to the Encounter context. May also affect access control.

This will typically be the encounter the audit event was created during, but some audit events may be initiated prior to or after the official completion of an encounter but still be tied to the context of the encounter (e.g. pre-admission lab tests).

An actor taking an active role in the event or activity that is logged.

An agent can be a person, an organization, software, device, or other actors that may be ascribed responsibility.

Several agents may be associated (i.e. have some responsibility for an activity) with an event or activity.

For example, an activity may be initiated by one user for other users or involve more than one user. However, only one user may be the initiator/requestor for the activity.

When a network are used in an event being recorded, there should be distinct agent elements for the known actors using the network. The agent with a network detail would be the responsible agent for use of that network.

The Functional Role of the user when performing the event.

Functional roles reflect functional aspects of relationships between entities. Functional roles are bound to the realization/performance of acts, where actions might be concatenated to an activity or even to a process. This element will hold the functional role that the agent played in the activity that is the focus of this Provenance. Where an agent played multiple functional roles, they will be listed as multiple .agent elements representing each functional participation. See ISO 21298:2018 - Health Informatics - Functional and structural roles, and ISO 22600-2:2014 - Health Informatics - Privilege Management and Access Control - Part 2: formal models.

For example: assembler, author, prescriber, signer, investigator, etc.

The structural roles of the agent indicating the agent's competency. The security role enabling the agent with respect to the activity.

Structural roles reflect the structural aspects of relationships between entities. Structural roles describe prerequisites, feasibilities, or competences for acts. Functional roles reflect functional aspects of relationships between entities. Functional roles are bound to the realization/performance of acts, where actions might be concatenated to an activity or even to a process. See ISO 21298:2018 - Health Informatics - Functional and structural roles, and ISO 22600-2:2014 - Health Informatics - Privilege Management and Access Control - Part 2: formal models..

For example: Chief-of-Radiology, Nurse, Physician, Medical-Student, etc.

Reference to who this agent is that was involved in the event.

This field ties an audit event to a specific resource or identifier.

Where a User ID is available it will go into who.identifier. Where a name of the user (human readable) it will go into who.display.

Indicator that the user is or is not the requestor, or initiator, for the event being audited.

This value is used to distinguish between requestor-users and recipient-users. For example, one person may initiate a report-output to be sent to another user.

There can only be one initiator. If the initiator is not clear, then do not choose any one agent as the initiator.

Where the agent location is known, the agent location when the event occurred.

Where the policy(ies) are known that authorized the agent participation in the event. Typically, a single activity may have multiple applicable policies, such as patient consent, guarantor funding, etc. The policy would also indicate the security token used.

This value is used retrospectively to determine the authorization policies.

For example: Where an OAuth token authorizes, the unique identifier from the OAuth token is placed into the policy element Where a policy engine (e.g. XACML) holds policy logic, the unique policy identifier is placed into the policy element.

When the event utilizes a network there should be an agent describing the local system, and an agent describing remote system, with the network interface details.

When a network protocol is used the endpoint is associated with the agent most directly using the endpoint. This is usually the software agent that has implemented the application level protocol. Preference is to define network in terms of a Reference(Endpoint), or URI; use string only when address or hostname is all that is known. When encoding using string it is best to encode using the formal canonical host name, but if you can't, then you can encode numeric in Literal address form using square brackets '[]' as a v4 string (in dotted notation), or v6 string (in colon notation).

When remote network endpoint is known, another agent representing the remote agent would indicate the remote network endpoint used. Convention is to indicate data flowing from Source to Destination. The convention for Search, given data flows both ways (query parameters vs results), is to have the Source as the initiator of the Search Transaction, and the Destination the responder to the Search transaction.

The authorization (e.g., PurposeOfUse) that was used during the event being recorded.

Record of any relevant security context, not restricted to purposeOfUse valueSet. May include security compartments, refrain, obligation, or other security tags.

Use AuditEvent.agent.authorization when you know that is specific to the agent, otherwise use AuditEvent.authorization. For example, during a machine-to-machine transfer it might not be obvious to the audit system who caused the event, but it does know why.

The actor that is reporting the event.

The event is reported by one source.

Events are reported by the actor that detected them. This may be one of the participating actors, but may also be different. The actor may be a human such as a medical-records clerk disclosing data manually, that clerk would be the source for the record of disclosure.

Logical source location within the healthcare enterprise network. For example, a hospital or other provider location within a multi-entity provider group.

This value differentiates among the sites in a multi-site enterprise health information system.

Identifier of the source where the event was detected.

This field ties the event to a specific source system. It may be used to group events for analysis according to where the event was detected.

Code specifying the type of source where event originated.

This field indicates which type of source is identified by the Audit Source ID. It is an optional value that may be used to group events for analysis according to the type of source where the event occurred.

Specific instances of data or objects that have been accessed.

The event may have other entities involved.

Required unless the values for event identification, agent identification, and audit source identification are sufficient to document the entire auditable event. Because events may have more than one entity, this group can be a repeating set of values.

Identifies a specific instance of the entity. The reference should be version specific. This is allowed to be a Parameters resource.

Use .what.display when all you have is a string (e.g. ParticipantObjectName).

Code representing the role the entity played in the event being audited.

For some detailed audit analysis it may be necessary to indicate a more granular type of entity, based on the application role it serves.

Security labels for the identified entity.

This field identifies the security labels for a specific instance of an object, such as a patient, to detect/track privacy and security issues.

Copied from entity meta security tags.

The query parameters for a query-type entities.

For query events, it may be necessary to capture the actual query input to the query process in order to identify the specific event. Because of differences among query implementations and data encoding for them, this is a base 64 encoded data blob. It may be subsequently decoded or interpreted by downstream audit analysis processing.

The meaning and secondary-encoding of the content of base64 encoded blob is specific to the AuditEvent.type, AuditEvent.subtype, and AuditEvent.entity.role. The base64 is a general-use and safe container for event specific data blobs regardless of the encoding used by the transaction being recorded. An AuditEvent consuming application must understand the event it is consuming and the formats used by the event. For example, if auditing an Oracle network database access, the Oracle formats must be understood as they will be simply encoded in the base64binary blob.

The DICOM AuditMessage schema does not support both .name and .query being populated.

Tagged value pairs for conveying additional information about the entity.

Implementation-defined data about specific details of the object accessed or used.

The name of extra detail provided in the value. This element is the tag for the value. Where a simple string is used for the tag name, use the CodeableConcept.display element.

The value of the extra detail.

Should not duplicate the entity value unless absolutely necessary.

The entity is attributed to an agent to express the agent's responsibility for that entity in the activity. This is most used to indicate when persistence media (the entity) are used by an agent. For example when importing data from a device, the device would be described in an entity, and the user importing data from that media would be indicated as the entity.agent.

A usecase where one AuditEvent.entity.agent is used where the Entity that was used in the creation/updating of a target resource, is not in the context of the same custodianship as the target resource, and thus the meaning of AuditEvent.entity.agent is to say that the entity referenced is managed elsewhere and that this Agent provided access to it. This would be similar to where the Entity being referenced is managed outside FHIR, such as through HL7 V2, v3, or XDS. This might be where the Entity being referenced is managed in another FHIR resource server. Thus it explains the provenance of that Entity's use in the context of this AuditEvent activity.