To help assure healthcare privacy and security in automated systems,
usage data needs to be collected. This data will be reviewed by
administrative staff to verify that healthcare data is being used in
accordance with the healthcare provider's data security requirements
and to establish accountability for data use. This data collection
and review process is called security auditing.
This document defines the format of the data to be collected and
minimum set of attributes that need to be captured by healthcare
application systems for subsequent use by an automation-assisted
review application. The data includes records of who accessed
healthcare data, when, for what action, from where, and which
Marshall Informational [Page 2]
RFC 3881 Security Audit & Access Accountability September 2004
patients' records were involved. The data definition is an XML
schema to be used as a reference by healthcare standards developers
and application designers.
This document consolidates previously disjointed viewpoints of
security auditing from Health Level 7 (HL7) [HL7SASIG], Digital
Imaging and Communications in Medicine (DICOM) Working Group 14,
Integrating the Healthcare Enterprise (IHE) [IHETF-3], the ASTM
International Healthcare Informatics Technical Committee (ASTM E31)
[E2147], and the Joint NEMA/COCIR/JIRA Security and Privacy Committee
[NEMASPC]. It is intended as a reference for these groups and other
healthcare standards developers.
The purposes the document fulfills are to:
1) Define data to be communicated for evidence of compliance with, or
violations of, a healthcare enterprise's security and privacy
policies and objectives.
This document defines the audit message format and content for
healthcare application systems. The focus of auditing is to
retrospectively detect and report security/privacy breaches. This
includes capturing data that supports individual accountability
for patient record creation, access, updates, and deletions.
This document does not define healthcare security and privacy
policies or objectives. It also does not include real-time access
alarm actions since there is a perception in the healthcare
community that security measures that inhibit access may also
inhibit effective patient care, under some circumstances.
2) Depict the data that would potentially reside in a common audit
engine or database.
Privacy and security audit data is to be collected on each
hardware system, and there are likely to be separate local data
stores for system-level and application-level audits. Collating
these records and providing a common view - transcending hardware
system boundaries - is seen as necessary for cost-effective
security and privacy policy administration.
The data definitions in this document support such a collation,
but the technical implementation alternatives are not covered in
this document.
Marshall Informational [Page 3]
RFC 3881 Security Audit & Access Accountability September 2004
3) Depict data that allows useful queries against audited events.
Audit data, in its raw form, reflects a sequential view of system
activity. Useful inquiries for security and privacy
administration need workflow, business process, organizational,
role, and person-oriented views. Data definitions in this
document anticipate and support creating those views and queries,
but do not define them.
4) Provide a common reference standard for healthcare IT standards
development organizations.
By specifying an XML schema, this document anticipates extensions
to the base schema to meet requirements of healthcare standards
bodies and application developers.
This document specifies audit data to be collected and communicated
from automated systems. It does not include non-automated processes.
Data for events in the above categories may be selectively collected,
based on healthcare organization policy. This document does not
specify any baseline or minimal policies.
For each audited event, this document specifies the minimal data
requirements plus optional data for the following event categories:
1) Security administrative events - establishing and maintaining
security policy definitions, secured object definitions, role
definitions, user definitions, and the relationships among them.
In general, these events are specific to the administrative
applications.
2) Audit access events - reflecting special protections implemented
for the audit trail itself.
3) Security-mediated events - recording entity identification and
authentication, data access, function access, nonrepudiation,
cryptographic operations, and data import/export for messages and
reports. In general, these events are generic to all protected
resources, without regard to the application data content.
Marshall Informational [Page 4]
RFC 3881 Security Audit & Access Accountability September 2004
4) Patient care data events - documenting what was done, by whom,
using which resources, from what access points, and to whose
medical data. In general, these audits are application-specific
since they require knowledge of the application data content.
Security subsystems found in most system infrastructures include a
capability to capture system-level security relevant events like
log-on and security object accesses. This document does not preclude
such functions being enabled to record and supply the data defined in
this document, but transformation of the collected data to the common
XML schema definition may be necessary to support requirements
consolidated auditing views.
Application-level events, such as patient record access, are not
captured by system-level security audits. The defined data support
applications' record access auditing for healthcare institutional
security and privacy assurance plus related policy administration
functions.
System-local data definitions for collection and storage of audit
data, prior to transformation to a common schema and transmission to
a common repository, are not included in this document.
This document anticipates, but does not define, end-uses for the data
collected.
The typical healthcare IT environment contains many systems from
various vendors and developers who have not implemented common or
interoperable security administrative functions. This document
anticipates a requirement to transmit data from several unrelated
systems to a common repository. It also anticipates the aggregated
data which may then be queried and viewed in a variety of ways.
There are distinctions of detail granularity, specificity, and
frequency between audit data required for surveillance versus
forensic purposes. While some surveillance data may be useful for
forensics, the scope of this document is limited to surveillance.
This document does not address access real-time policy violation
alarm actions. There is a perception in the healthcare community
that security measures which inhibit access may also inhibit
effective patient care, under some circumstances.
Marshall Informational [Page 5]
RFC 3881 Security Audit & Access Accountability September 2004
This document does not define any data for patient care consents or
patients' permissions for data disclosure. It is conceivable that
the proposed audit data could be input to such applications, however,
assuming strict access controls for audit data have been established.
This document does not define system-specific or application-specific
data that may be collected and reported in addition to the defined
elements. For example, it is conceivable that audit mechanisms may
be useful for tracking financial or payroll transactions. At the
same time, this document does not preclude extending the XML schema
to incorporate additional data.
There is a potential requirement for a set of administrative messages
to be sent from a central source to each participating system to
uniformly specify, control, enable, or disable audit data collection.
Such messages are not included in this document.
This document does not include any definitions of conformance
practices. Instead, it anticipates that standards development
organizations that reference this document may specify their own
conformance requirements.
The process of assuring that security policies are implemented
correctly is essential to information security administration. It is
a set of interrelated tasks all aimed at maintaining an acceptable
level of confidence that security protections are, in fact, working
as intended. These tasks are assisted by data from automated
instrumentation of system and application functions.
Data gathered from a secured environment is used to accumulate
evidence that security systems are working as intended and to detect
incidents and patterns of misuse for further actions. Once messages
have been collected, various reports may be created in support of
security assurance and administration information requirements.
When a site runs multiple heterogeneous applications, each
application system may have its own security mechanisms - user log-
on, roles, access right permissions and restrictions, etc. Each
application system also has its own security log file that records
security relevant events, e.g., log-in, data access, and updates to
the security policy databases. A system administrator or security
auditor must examine each of these log files to find security
Marshall Informational [Page 6]
RFC 3881 Security Audit & Access Accountability September 2004
relevant incidents. Not only is it difficult to examine each of
these files separately, the format and contents of each file may be
confusingly different.
Resolving these issues requires a framework to:
- Maximize interoperability and the meaningfulness of data across
applications and sites
- Minimize ambiguity among heterogeneous systems
- Simplify and limit the costs of administrative audit tasks.
One of the leading concerns about auditing is the potential volume of
data gathering and its impact on application system performance.
Although this document does not prescribe specific implementations or
strategies, the following are meant as informative guidance for
development.
1) Audits should be created for transactions or record-level data
access, not for individual attribute-level changes to data.
2) This document does not discourage locally optimized gathering of
audit data on each application system. Instead, it anticipates
implementation-defined periodic gathering and transmission of data
to a common repository. This common repository would be optimized
for after-the-fact audit queries and reporting, thus unburdening
each application system of those responsibilities. It is also
important to keep the message size compact so that audit data will
not penalize normal network operation.
3) On each application system, a variety of policy-based methods
could be employed to optimize data gathering and storage, e.g.,
selective auditing of only events defined as important plus
workload buffering and balancing. Data gathering itself should be
stateless to avoid the overhead of transactional semantics. In
addition, prior to transmission, some filtering, aggregation, and
summarization of repeated events would reduce the number of
messages. Audit data storage and integrity on each application
system need only be scaled for relatively low-volume and short-
duration requirements, yet be consistent with implementation-
defined minimums for holding the data for subsequent collection.
4) Leveraging existing data collection should be considered. For
example, most commercial security subsystems record events in a
local common log file, so the log file data can be extracted for
communication to a common repository. Also, it is common in some
systems' designs to have a transaction log for data reconstruction
Marshall Informational [Page 7]
RFC 3881 Security Audit & Access Accountability September 2004
in event of database loss, so collecting data-update audit data
within this subsystem could reduce impact on application system
performance.
5) A security audit repository would gather all audit message data
from the different applications in one database with one standard
structure. This would allow easier evaluation and querying. Once
a suspicious pattern has been found in the audit log repository,
investigation might proceed with more detail in the application
specific audit log. The presence of a common repository also
simplifies and streamlines the implementation of policies for
audit data storage, integrity, retention, and destruction.
The following identifies representative trigger events for generating
audit messages. This is not a complete list of trigger events.
For those events arising in the security infrastructure the "minimal"
and "basic" level of auditing as outlined in the Common Criteria
[ISO15408-2] should be used as a reference standard.
This group includes all actions that create, maintain, query, and
display definitions for securing data, functions, and the associated
access policies. For each trigger type, the creation, update or
amendment, deletion, and activation or deactivation are auditable.
This includes creation, modification, deletion, query, and display of
security attributes for data sets, data groups, or classes plus their
atomic data elements or attributes.
This includes, for example, creation, modification, deletion, query,
or display of security attributes and auditable events for the
application functions used for patient management, clinical
processes, registry of business objects and methods, program creation
and maintenance, etc.
This includes all activities to create, modify, delete, query, or
display security domains according to various organizational
categories such as entity-wide, institutional, departmental, etc.
Marshall Informational [Page 8]
RFC 3881 Security Audit & Access Accountability September 2004
This includes all activities that create, modify, delete, query or
display security categories or groupings for functions and data such
as patient management, nursing, clinical, etc.
This includes all activities that create, modify, delete, query or
display the allowable access permissions associated with functions
and data, such as create, read, update, delete, and execution of
specific functional units or object access or manipulation methods.
This includes all activities that create, modify, delete, query or
display security roles according to various task-grouping categories
such as security administration, admissions desk, nurses, physicians,
clinical specialists, etc. It also includes the association of
permissions with roles for role-based access control.
This includes all activities that create, modify, delete, query, or
display user accounts. It includes password or other authentication
data. It also includes the association of roles with users for
role-based access control, or permissions with users for user-based
access control.
This reflects a basic policy decision that an event should or should
not be audited. Some, but not necessarily all, triggers or use cases
must create an audit record. The selection of what to audit depends
on administrative policy decisions. Note that, for integrity, this
event should always be audited.
Marshall Informational [Page 9]
RFC 3881 Security Audit & Access Accountability September 2004
This includes instances where audit data is viewed or reported for
any purpose. Since the audit data itself may include data protected
by institutional privacy policies and expose the implementation of
those policies, access to the data is highly sensitive. This event
should therefore always be audited.
This includes instances where audit data is modified or deleted.
While such operations are sometimes permitted by systems policies,
modification or destruction of audit data may well be the result of
unauthorized hostile systems access. Therefore, this type of event
should always be audited.
This includes successful and unsuccessful attempts from human users
and automated system. It also includes re-authentication actions and
re-issuing time-sensitive credentials such as Kerberos tickets.
This includes user invocation of application or system functions that
have permission definitions associated with them. Note that in a
Discretionary Access Control environment not all functions require
permissions, especially if their impact is benign in relation to
security policies.
The following are examples of trigger events relevant to healthcare
privacy. The actual triggers for institutional data access, policies
for non-care functions, and support regulatory requirements need to
be identified by application-domain standards developers and system
implementers.
Marshall Informational [Page 10]
RFC 3881 Security Audit & Access Accountability September 2004
This includes all functions which manipulate basic patient data:
- Create, e.g., demographics or patient profile
- Assign identifier, e.g., medical record number
- Update, amend
- Merge/unmerge, e.g., combine multiple medical records for one
patient
- Import/export of data from/to an external source, including
printing and creation of portable media copies.
- Delete, e.g., invalid creation of care record
This includes all functions which associate a subject of care with an
instance of care:
- Create, e.g., demographics or patient profile
- Assign encounter identifier
- Per-admit
- Admit
- Update, amend
- Delete, e.g., invalid creation of encounter record, breakdown of
equipment, patient did not arrive as expected
This includes all functions which associate care plans or similar
protocols with an instance or subject of care:
- Schedule, initiate
- Update, amend
- Complete
- Cancel
This includes specific clinical episodes within an instance of care.
Initiate:
- Update, amend
- Resolve, complete
- Cancel
Marshall Informational [Page 11]
RFC 3881 Security Audit & Access Accountability September 2004
This includes clinical or supplies orders within an instance or
episode of care:
- Initiate
- Update, amend
- Check for contraindications
- Verify
- Deliver/complete - including instructions
- Cancel
This includes various health services scheduled and performed within
an instance or episode of care:
- Schedule, initiate
- Update, amend
- Check for contraindications
- Verify
- Perform/complete - including instructions
- Cancel
This includes all medication orders and administration within an
instance or episode of care:
- Order
- Check
- Check for interactions
- Verify
- Dispense/deliver - including administration instructions
- Administer
- Cancel
This includes staffing or participant assignment actions relevant to
an instance or episode of care:
- Assignment of healthcare professionals, caregivers attending
physician, residents, medical students, consultants, etc.
- Change in assigned role or authorization, e.g., relative to
healthcare status change.
- De-assignment
Marshall Informational [Page 12]
RFC 3881 Security Audit & Access Accountability September 2004
This section defines and describes the data in the XML schema. The
actual XML schema definition is in section 6.
The proposed data elements are grouped into these categories:
1) Event Identification - what was done
2) Active Participant Identification - by whom
3) Network Access Point Identification - initiated from where
4) Audit Source Identification - using which server
5) Participant Object Identification - to what record
The following data identifies the name, action type, time, and
disposition of the audited event. There is only one set of event
identification data per audited event.
Description
Identifier for a specific audited event, e.g., a menu item,
program, rule, policy, function code, application name, or URL.
It identifies the performed function.
Optionality: Required
Format / Values
Coded value, either defined by the system implementers or as a
reference to a standard vocabulary. The "code" attribute must be
unambiguous and unique, at least within Audit Source ID (see
section 5.4). Examples of Event IDs are program name, method
name, or function name.
For implementation defined coded values or references to
standards, the XML schema defines these optional attributes:
Attribute Value
-------------- --------------------------------------------
CodeSystem OID reference
CodeSystemName Name of the coding system; strongly recommended
to be valued for locally-defined code-sets.
DisplayName The value to be used in displays and reports
OriginalText Input value that was translated to the code
Marshall Informational [Page 13]
RFC 3881 Security Audit & Access Accountability September 2004
To support the requirement for unambiguous event identification,
multiple values may not be specified.
Rationale
This identifies the audited function. For "Execute" Event Action
Code audit records, this identifies the application function
performed.
Description
Indicator for type of action performed during the event that
generated the audit.
Optionality: Optional
Format / Values
Enumeration:
Value Meaning Examples
----- --------------------- ----------------------------------
C Create Create a new database object, such
as Placing an Order.
R Read/View/Print/Query Display or print data, such as a
Doctor Census
U Update Update data, such as Revise
Patient Information
D Delete Delete items, such as a doctor
master file record
E Execute Perform a system or application
function such as log-on, program
execution, or use of an object's
method
Rationale
This broadly indicates what kind of action was done on the
Participant Object.
Marshall Informational [Page 14]
RFC 3881 Security Audit & Access Accountability September 2004
Notes
Actions that are not enumerated above are considered an Execute of
a specific function or object interface method or treated two or
more distinct events. An application action, such as an
authorization, is a function Execute, and the Event ID would
identify the function.
For some applications, such as radiological imaging, a Query
action may only determine the presence of data but not access the
data itself. Auditing need not make as fine a distinction.
Compound actions, such as "Move," would be audited by creating
audit data for each operation - read, create, delete - or as an
Execute of a function or method.
Description
Universal coordinated time (UTC), i.e., a date/time specification
that is unambiguous as to local time zones.
Optionality: Required
Format / Values
A date/time representation that is unambiguous in conveying
universal coordinated time (UTC), formatted according to the ISO
8601 standard [ISO8601]
Rationale
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.
Notes
In a distributed system, some sort of common time base, e.g., an
NTP [RFC1305] server, is a good implementation tactic.
Description
Indicates whether the event succeeded or failed.
Marshall Informational [Page 15]
RFC 3881 Security Audit & Access Accountability September 2004
Optionality: Required
Format / Values
Enumeration:
Value Meaning
---- ----------------------------------------------------
0 Success
4 Minor failure; action restarted, e.g., invalid password
with first retry
8 Serious failure; action terminated, e.g., invalid
password with excess retries
12 Major failure; action made unavailable, e.g., user
account disabled due to excessive invalid log-on attempts
Rationale
Some audit events may be qualified by success or failure
indicator. For example, a Log-on might have this flag set to a
non-zero value to indicate why a log-on attempt failed.
Notes
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.
Description
Identifier for the category of event.
Optionality: Optional
Format / Values
Coded value enumeration, either defined by the system implementers
or as a reference to a standard vocabulary. For implementation
defined codes or references to standards, the XML schema defines
these optional attributes:
Marshall Informational [Page 16]
RFC 3881 Security Audit & Access Accountability September 2004
Attribute Value
-------------- --------------------------------------------
CodeSystem OID reference
CodeSystemName Name of the coding system; strongly recommended
to be valued for locally-defined code-sets.
DisplayName The value to be used in displays and reports
OriginalText Input value that was translated to the code
Since events may be categorized in more than one way, there may be
multiple values specified.
Rationale
This field enables queries of messages by implementation-defined
event categories.
The following data identify a user for the purpose of documenting
accountability for the audited event. A user may be a person, or a
hardware device or software process for events that are not initiated
by a person.
Optionally, the user's network access location may be specified.
There may be more than one user per event, for example, in cases of
actions initiated by one user for other users, or in events that
involve more than one user, hardware device, or system process.
However, only one user may be the initiator/requestor for the event.
Description
Unique identifier for the user actively participating in the event
Optionality: Required
Format / Values
User identifier text string from the authentication system. It is
a unique value within the Audit Source ID (see section 5.4).
Rationale
This field ties an audit event to a specific user.
Marshall Informational [Page 17]
RFC 3881 Security Audit & Access Accountability September 2004
Notes
For cross-system audits, especially with long retention, this user
identifier will permanently tie an audit event to a specific user
via a perpetually unique key.
For node-based authentication -- where only the system hardware or
process, but not a human user, is identified -- User ID would be
the node name.
Description
Alternative unique identifier for the user
Optionality: Optional
Format / Values
User identifier text string from authentication system. This
identifier would be one known to a common authentication system
(e.g., single sign-on), if available.
Rationale
In some situations a user may authenticate with one identity but, to
access a specific application system, may use a synonymous identify.
For example, some "single sign on" implementations will do this. The
alternative identifier would then be the original identify used for
authentication, and the User ID is the one known to and used by the
application.
Description
The human-meaningful name for the user
Optionality: Optional
Format / Values
Text string
Marshall Informational [Page 18]
RFC 3881 Security Audit & Access Accountability September 2004
Rationale
The User ID and Alternative User ID may be internal or otherwise
obscure values. This field assists the auditor in identifying the
actual user.
Description
Indicator that the user is or is not the requestor, or initiator,
for the event being audited.
Optionality: Optional
Format / Values
Boolean, default/assumed value is "true"
Rationale
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 a another user.
Description
Specification of the role(s) the user plays when performing the
event, as assigned in role-based access control security.
Optionality: Optional; multi-valued
Format / Values
Coded value, with attribute "code" valued with the role code or
text from authorization system. More than one value may be
specified.
The codes may be implementation-defined or reference a standard
vocabulary enumeration. For implementation defined codes or
references to standards, the XML schema defines these optional
attributes:
Marshall Informational [Page 19]
RFC 3881 Security Audit & Access Accountability September 2004
Attribute Value description
-------------- --------------------------------------------
CodeSystem OID reference
CodeSystemName Name of the coding system; strongly recommended
to be valued for locally-defined code-sets.
Display Name The value to be used in displays and reports
OriginalText Input value that was translated to the code
Rationale
This value ties an audited event to a user's role(s). It is an
optional value that might be used to group events for analysis by
user functional role categories.
Notes
Many security systems are unable to produce this data, hence it is
optional.
For the common message, this identifier would be the one known to
a common authorization system, if available. Otherwise, it is a
unique value within the Audit Source ID (see section 5.4).
Consider using a globally unique identifier associated with the
role to avoid ambiguity in auditing data collected from multiple
systems.
Role ID is not a substitute for personal accountability.
Ambiguities arise from composite roles and users with multiple
roles, i.e., which role within a composite is being used or what
privilege was a user employing?
The network access point identifies the logical network location for
application activity. These data are paired 1:1 with the Active
Participant Identification data.
Description
An identifier for the type of network access point that originated
the audit event.
Optionality: Optional
Format / Values
Marshall Informational [Page 20]
RFC 3881 Security Audit & Access Accountability September 2004
Enumeration:
Value Meaning
----- --------------------------------
1 Machine Name, including DNS name
2 IP Address
3 Telephone Number
Rationale
This datum identifies the type of network access point identifier
of the user device for the audit event. It is an optional value
that may be used to group events recorded on separate servers for
analysis of access according to a network access point's type.
Description
An identifier for the network access point of the user device for
the audit event. This could be a device id, IP address, or some
other identifier associated with a device.
Optionality: Optional
Format / Values
Text may be constrained to only valid values for the given Network
Access Point Type, if specified. Recommendation is to be as
specific as possible where multiple options are available.
Rationale
This datum identifies the user's network access point, which may
be distinct from the server that performed the action. It is an
optional value that may be used to group events recorded on
separate servers for analysis of a specific network access point's
data access across all servers.
Note
Network Access Point ID is not a substitute for personal
accountability. Internet IP addresses, in particular, are highly
volatile and may be assigned to more than one person in a short
time period.
Marshall Informational [Page 21]
RFC 3881 Security Audit & Access Accountability September 2004
Examples
Network Access Point ID: SMH4WC02
Network Access Point Type: 1 = Machine Name
Network Access Point ID: 192.0.2.2
Network Access Point Type: 2 = IP address
Network Access Point ID: 610-555-1212
Network Access Point Type: 3 = Phone Number
The following data are required primarily for application systems and
processes. Since multi-tier, distributed, or composite applications
make source identification ambiguous, this collection of fields may
repeat for each application or process actively involved in the
event. For example, multiple value-sets can identify participating
web servers, application processes, and database server threads in an
n-tier distributed application. Passive event participants, e.g.,
low-level network transports, need not be identified.
Depending on implementation strategies, it is possible that the
components in a multi-tier, distributed, or composite applications
may generate more than one audit message for a single application
event. Various data in the audit message may be used to identify
such cases, supporting subsequent data reduction. This document
anticipates that the repository and reporting mechanisms will perform
data reduction when required, but does not specify those mechanism.
Description
Logical source location within the healthcare enterprise network,
e.g., a hospital or other provider location within a multi-entity
provider group.
Optionality: Optional
Format / Values
Unique identifier text string within the healthcare enterprise.
May be unvalued when the audit-generating application is uniquely
identified by Audit Source ID.
Marshall Informational [Page 22]
RFC 3881 Security Audit & Access Accountability September 2004
Rationale
This value differentiates among the sites in a multi-site
enterprise health information system.
Notes
This is defined by the application that generates the audit
record. It contains a unique code that identifies a business
organization (owner of data) that is known to the enterprise. The
value further qualifies and disambiguates the Audit Source ID.
Values may vary depending on type of business. There may be
levels of differentiation within the organization.
Description
Identifier of the source where the event originated.
Optionality: Required
Format / Values
Unique identifier text string, at least within the Audit
Enterprise Site ID
Rationale
This field ties the event to a specific source system. It may be
used to group events for analysis according to where the event
occurred.
Notes
In some configurations, a load-balancing function distributes work
among two or more duplicate servers. The values defined for this
field thus may be considered as an source identifier for a group
of servers rather than a specific source system.
Description
Code specifying the type of source where event originated.
Optionality: Optional
Marshall Informational [Page 23]
RFC 3881 Security Audit & Access Accountability September 2004
Format / Values
Coded-value enumeration, optionally defined by system implementers
or a as a reference to a standard vocabulary. Unless defined or
referenced, the default values for the "code" attribute are:
Value Meaning
----- ------------------------------------------------------
1 End-user interface
2 Data acquisition device or instrument
3 Web server process tier in a multi-tier system
4 Application server process tier in a multi-tier system
5 Database server process tier in a multi-tier system
6 Security server, e.g., a domain controller
7 ISO level 1-3 network component
8 ISO level 4-6 operating software
9 External source, other or unknown type
For implementation defined codes or references to standards, the
XML schema defines these optional attributes:
Attribute Value
-------------- --------------------------------------------
CodeSystem OID reference
CodeSystemName Name of the coding system; strongly recommended
to be valued for locally-defined code-sets.
DisplayName The value to be used in displays and reports
OriginalText Input value that was translated to the code
Since audit sources may be categorized in more than one way, there
may be multiple values specified.
Rationale
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.
The following data assist the auditing process by indicating specific
instances of data or objects that have been accessed.
These data are required unless the values for Event Identification,
Active Participant Identification, and Audit Source Identification
are sufficient to document the entire auditable event. Production of
Marshall Informational [Page 24]
RFC 3881 Security Audit & Access Accountability September 2004
audit records containing these data may be enabled or suppressed, as
determined by healthcare organization policy and regulatory
requirements.
Because events may have more than one participant object, this group
can be a repeating set of values. For example, depending on
institutional policies and implementation choices:
- Two participant object value-sets can be used to identify access
to patient data by medical record number plus the specific health
care encounter or episode for the patient.
- A patient participant and his authorized representative may be
identified concurrently.
- An attending physician and consulting referrals may be identified
concurrently.
- All patients identified on a worklist may be identified.
- For radiological studies, a set of related participant objects
identified by accession number or study number, may be identified.
Note, though, that each audit message documents only a single usage
instance of such participant object relationships and does not serve
to document all relationships that may be present or possible.
Description
Code for the participant object type being audited. This value is
distinct from the user's role or any user relationship to the
participant object.
Optionality: Optional
Format / Values
Enumeration:
Value Meaning
----- -------------
1 Person
2 System Object
3 Organization
4 Other
Marshall Informational [Page 25]
RFC 3881 Security Audit & Access Accountability September 2004
Rationale
To describe the object being acted upon. In addition to queries
on the subject of the action in an auditable event, it is also
important to be able to query on the object type for the action.
Description
Code representing the functional application role of Participant
Object being audited
Optionality: Optional
Format / Values
Enumeration, specific to Participant Object Type Code:
Value Meaning Participant Object Type Codes
----- -------------------- ----------------------------------
1 Patient 1 - Person
2 Location 3 - Organization
3 Report 2 - System Object
4 Resource 1 - Person
3 - Organization
5 Master file 2 - System Object
6 User 1 - Person
2 - System Object (non-human user)
7 List 2 - System Object
8 Doctor 1 - Person
9 Subscriber 3 - Organization
10 Guarantor 1 - Person
3 - Organization
11 Security User Entity 1 - Person
2 - System Object
12 Security User Group 2 - System Object
13 Security Resource 2 - System Object
14 Security Granularity 2 - System Object
Definition
15 Provider 1 - Person
3 - Organization
16 Data Destination 2 - System Object
17 Data Repository 2 - System Object
18 Schedule 2 - System Object
19 Customer 3 - Organization
20 Job 2 - System Object
21 Job Stream 2 - System Object
Marshall Informational [Page 26]
RFC 3881 Security Audit & Access Accountability September 2004
22 Table 2 - System Object
23 Routing Criteria 2 - System Object
24 Query 2 - System Object
A "Security Resource" is an abstract securable object, e.g., a
screen, interface, document, program, etc. -- or even an audit
data set or repository.
Rationale
For some detailed audit analysis it may be necessary to indicate a
more granular type of participant, based on the application role
it serves.
Description
Identifier for the data life-cycle stage for the participant
object. This can be used to provide an audit trail for data, over
time, as it passes through the system.
Optionality: Optional
Format/Values
Enumeration:
Value Meaning
----- --------------------------------------
1 Origination / Creation
2 Import / Copy from original
3 Amendment
4 Verification
5 Translation
6 Access / Use
7 De-identification
8 Aggregation, summarization, derivation
9 Report
10 Export / Copy to target
11 Disclosure
12 Receipt of disclosure
13 Archiving
14 Logical deletion
15 Permanent erasure / Physical destruction
Marshall Informational [Page 27]
RFC 3881 Security Audit & Access Accountability September 2004
Rationale
Institutional policies for privacy and security may optionally
fall under different accountability rules based on data life
cycle. This provides a differentiating value for those cases.
Description
Describes the identifier that is contained in Participant Object
ID.
Optionality: Required
Format / Values
Coded-value enumeration, specific to Participant Object Type Code,
using attribute-name "code". The codes below are the default set.
Value Meaning Participant Object Type Codes
----- ---------------------- -----------------------------
1 Medical Record Number 1 - Person
2 Patient Number 1 - Person
3 Encounter Number 1 - Person
4 Enrollee Number 1 - Person
5 Social Security Number 1 - Person
6 Account Number 1 - Person
3 - Organization
7 Guarantor Number 1 - Person
3 - Organization
8 Report Name 2 - System Object
9 Report Number 2 - System Object
10 Search Criteria 2 - System Object
11 User Identifier 1 - Person
2 - System Object
12 URI 2 - System Object
User Identifier and URI [RFC2396] text strings are intended to be
used for security administration trigger events to identify the
objects being acted-upon.
The codes may be the default set stated above, implementation-
defined, or reference a standard vocabulary enumeration, such as
HL7 version 2.4 table 207 or DICOM defined media types. For
implementation defined codes or references to standards, the XML
schema defines these optional attributes:
Marshall Informational [Page 28]
RFC 3881 Security Audit & Access Accountability September 2004
Attribute Value
-------------- --------------------------------------------
CodeSystem OID reference
CodeSystemName Name of the coding system; strongly recommended
to be valued for locally-defined code-sets.
DisplayName The value to be used in displays and reports
OriginalText Input value that was translated to the code
Rationale
Required to distinguish among various identifiers that may
synonymously identify a participant object.
Description
Denotes policy-defined sensitivity for the Participant Object ID
such as VIP, HIV status, mental health status, or similar topics.
Optionality: Optional
Format / Values
Values are institution- and implementation-defined text strings.
Description
Identifies a specific instance of the participant object.
Optionality: Required
Format / Values
Text string. Value format depends on Participant Object Type Code
and the Participant Object ID Type Code.
Rationale
This field identifies a specific instance of an object, such as a
patient, to detect/track privacy and security issues.
Notes
Consider this to be the primary unique identifier key for the
object, so it may be a composite data field as implemented.
Marshall Informational [Page 29]
RFC 3881 Security Audit & Access Accountability September 2004
Description
An instance-specific descriptor of the Participant Object ID
audited, such as a person's name.
Optionality: Optional
Format / Values
Text string
Rationale
This field may be used in a query/report to identify audit events
for a specific person, e.g., where multiple synonymous Participant
Object IDs (patient number, medical record number, encounter
number, etc.) have been used.
Description
The actual query for a query-type participant object.
Optionality: Optional
Format / Values
Base 64 encoded data
Rationale
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.
Description
Implementation-defined data about specific details of the object
accessed or used.
Marshall Informational [Page 30]
RFC 3881 Security Audit & Access Accountability September 2004
Optionality: Optional
Format
Type-value pair. The "type" attribute is an implementation-
defined text string. The "value" attribute is a base 64 encoded
data.
Rationale
Specific details or values from the object accessed may be desired
in specific auditing implementations. The type-value pair enables
the use of implementation-defined and locally-extensible object
type identifiers and values. For example, a clinical diagnostic
object may contain multiple test results, and this element could
document the type and number and type of results.
Many possible data encodings are possible for this elements, so
the value is a base 64 encoded data blob. It may be subsequently
decoded or interpreted by downstream audit analysis processing.
This section contains the actual XML schema definition for the data
defined in section 5. It also provides brief guidance for specifying
schema localizations for implementation purposes.
The XML schema specified in section 6.1 conforms with the W3C
Recommendations for XML Schema structure [W3CXML-1] and data types
[W3CXML-2].
The schema specified in section 6.1 may be extended and restricted to
meet local implementation-specific requirements. W3C Recommendation
for XML Schema structure [W3CXML-1], section 4, is the governing
standard for accomplishing this.
As of the current version of this document, a public reference URI
for the base schema has not been established.
Local definitions reference the common audit message base schema.
For example, here is a schema with a local vocabulary restriction for
"Audit Enterprise Site ID" plus an extension adding a new "Audit
Source Asset Number" element.
The URI used to identify this schema (http://audit-message-uri) is a
syntactically valid example that does not represent an actual schema.
Schema validators might report an error when attempting to import a
schema using this URI.
<xs:schema xmlns:audit="http://audit-message-URI"
xmlns:xs="http://www.w3.org/2001/XMLSchema"
elementFormDefault="qualified" attributeFormDefault="unqualified">
<xs:import schemaLocation="http://audit-message-URI"/>
<xs:complexType name="LocaAuditSourceIdentificationType">
<xs:complexContent>
<xs:restriction base="AuditSourceIdentificationType">
<xs:attribute name="AuditEnterpriseSiteID" use="required">
<xs:simpleType>
<xs:restriction base="xs:string">
<xs:enumeration value="Main"/>
<xs:enumeration value="Clinic1"/>
<xs:enumeration value="Clinic2"/>
<xs:enumeration value="Radiology"/>
<xs:enumeration value="Lab"/>
</xs:restriction>
</xs:simpleType>
</xs:attribute>
</xs:restriction>
</xs:complexContent>
</xs:complexType>
<xs:element name="LocalAuditSourceIdentification">
<xs:complexType>
<xs:complexContent>
<xs:extension base="LocaAuditSourceIdentificationType">
<xs:attribute name="AuditSourceAssetNumber" type="xs:string"
Marshall Informational [Page 43]
RFC 3881 Security Audit & Access Accountability September 2004
use="required"/>
</xs:extension>
</xs:complexContent>
</xs:complexType>
</xs:element>
</xs:schema>
Audit data must be secured at least to the same extent as the
underlying data and activities being audited. This includes access
controls as well as data integrity and recovery functions. This
document acknowledges the need for, but does not specify, the
policies and technical methods to accomplish this.
It is conceivable that audit data might have unintended uses, e.g.,
tracking the frequency and nature of system use for productivity
measures. ASTM standard E2147-01 [E2147] states, in paragraph
5.3.10, "Prohibit use for other reasons than to enforce security and
to detect security breaches in record health information systems, for
example, the audits are not to be used to explore activity profiles
or movement profiles of employees."
Some audit data arises from security-relevant processes other than
data access. These are the trigger events listed in section 4.1 and
4.2 of this document. Audit data, defined in this document, can
record the accountabilities for the results of these processes, as
part of a complete security implementation. A discussion of the
associated authorities, reference standards, and implementation
technology choices for the processes is outside the scope of this
document.
[E2147] "E2147-01 Standard Specification for Audit and
Disclosure Logs for Use in Health Information Systems",
ASTM International, June 2002.
[ISO15408-2] "ISO/IEC 15408:1999 Common Criteria for Information
Technology Security Evaluation, Part 2: Security
Functional Requirements", ISO, August 1999.
[ISO8601] "ISO 8601:2000 Data elements and interchange formats --
Information interchange -- Representation of dates and
times", ISO, December 2000.
Marshall Informational [Page 44]
RFC 3881 Security Audit & Access Accountability September 2004
[RFC1305] Mills, D., "Network Time Protocol (Version 3)
Specification, Implementation", RFC 1305, March 1992.
[RFC2396] Berners-Lee, T., Fielding, R. and L. Masinter, "Uniform
Resource Identifiers (URI): Generic Syntax", RFC 2396,
August 1998.
[W3CXML-1] W3C Recommendation "XML Schema Part 1: Structures",
version 1.0, May 2001.
[W3CXML-2] W3C Recommendation "XML Schema Part 2: Datatypes,"
version 1.0, May 2001.
[HL7SASIG] Marshall, G. and G. Dickinson, "Common Audit Message",
HL7 Security and Accountability Special Interest Group,
November 2001.
[IHETF-3] "IHE Technical Framework", Volume III, HIMMS/RSNA, April
2002.
[NEMASPC] "Security and Privacy Auditing in Health Care
Information Technology", Joint NEMA/COCIR/JIRA Security
and Privacy Committee, 26 June 2001.
Acknowledgments
The author gratefully acknowledges the advice and assistance of the
following people during the preparation of this document:
Carmela Couderc, Siemens Medical Solutions
Michael Davis, SAIC
Gary Dickinson
Christoph Dickmann, Siemens Medical Solutions
Daniel Hannum, Siemens Medical Solutions
Robert Horn, Agfa
James McAvoy, Siemens Medical Solutions
John Moehrke, General Electric Medical Systems
Jennifer Puyenbroek, McKesson Information Solutions
Angela Ray, McKesson Information Solutions
Lawrence Tarbox, Siemens Corporate Research
Marshall Informational [Page 45]
RFC 3881 Security Audit & Access Accountability September 2004
Author's Address
Glen Marshall
Siemens Medical Solutions Health Services
51 Valley Stream Parkway
Malvern, PA 19312
USA
Phone: (610) 219-3938
EMail: glen.f.marshall@siemens.com
Marshall Informational [Page 46]
RFC 3881 Security Audit & Access Accountability September 2004
Full Copyright Statement
Copyright (C) The Internet Society (2004).
This document is subject to the rights, licenses and restrictions
contained in BCP 78, and at www.rfc-editor.org, and except as set
forth therein, the authors retain all their rights.
This document and the information contained herein are provided on an
"AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/S HE
REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE
INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Intellectual Property
The IETF takes no position regarding the validity or scope of any
Intellectual Property Rights or other rights that might be claimed to
pertain to the implementation or use of the technology described in
this document or the extent to which any license under such rights
might or might not be available; nor does it represent that it has
made any independent effort to identify any such rights. Information
on the ISOC's procedures with respect to rights in ISOC Documents can
be found in BCP 78 and BCP 79.
Copies of IPR disclosures made to the IETF Secretariat and any
assurances of licenses to be made available, or the result of an
attempt made to obtain a general license or permission for the use of
such proprietary rights by implementers or users of this
specification can be obtained from the IETF on-line IPR repository at
http://www.ietf.org/ipr.
The IETF invites any interested party to bring to its attention any
copyrights, patents or patent applications, or other proprietary
rights that may cover technology that may be required to implement
this standard. Please address the information to the IETF at ietf-
ipr@ietf.org.
Acknowledgement
Funding for the RFC Editor function is currently provided by the
Internet Society.
Marshall Informational [Page 47]