Difference between revisions of "Trustworthy Healthcare Provider"

From IDESG Wiki
Jump to: navigation, search
(Digital Service Endpoints)
(Solutions)
Line 73: Line 73:
 
* Note that the patient may be considered to be bound to their medical records, but that is an organizational relationship that is core to the patient matching problem. While it is not feasible for a medical record to have an identifier (index key) that is itself cryptographically bound to the patient given current US law, it is possible to include cryptographically bound identifiers in the medical record.
 
* Note that the patient may be considered to be bound to their medical records, but that is an organizational relationship that is core to the patient matching problem. While it is not feasible for a medical record to have an identifier (index key) that is itself cryptographically bound to the patient given current US law, it is possible to include cryptographically bound identifiers in the medical record.
 
* The user agent will typically also create a binding between the subject and any service provider indicated the consent granted by the subject to the provider. In most cases the subject will have the ability to dissolve the binding and withdraw the consent. Contract law may override this ability to withdraw consent in some cases.
 
* The user agent will typically also create a binding between the subject and any service provider indicated the consent granted by the subject to the provider. In most cases the subject will have the ability to dissolve the binding and withdraw the consent. Contract law may override this ability to withdraw consent in some cases.
 
===Protection of the Patient Credentials===
 
These are the requirements from NIST SP 800-63-3.
 
 
* Identity Assurance Level 2 requires the real-world existence of the claimed identity and verifies that the applicant is appropriately associated with this real-world identity. IAL2 introduces the need for identity proofing. Since many covered entities already perform identity proofing, the remaining problem is to move that proof from the covered entity into the patient credential on the smart phone.
 
* Authenticator Assurance Level 2 provides high confidence that the claimant controls authenticator(s) bound to the subscriber’s account. Proof of possession and control of two different authentication factors is required through secure authentication protocol(s) e.g. FIDO U2F (Web Authentication)or Client to Authentication Protocol (CTAP FIDO?) with the smart phone (with a trusted execution environment) itself.
 
 
The wiki page [[Phone as Health Care Credential]] describes how these requirements can be met in a patient-friendly manner.
 
 
===Use of [https://tcwiki.azurewebsites.net/index.php?title=Native_App Native Apps] to hold the patient's PHI===
 
While there may be multiple ways to protect patient's PHI while in the user's possession, this paper explores two use cases.
 
#1 The app is considered to be a HIPAA covered entity and will only accept and disburse patient data to other covered entities. Exfiltration of data from this [Trusted Federation] is severely restricted, but still possible.
 
#2 The app is not a HIPAA covered entity and is not bound by the federal regulations that apply to covered entities. Apps with little regard for the patient's information may be able to convince users to share data which is not covered.
 
The above addresses information that starts out covered by HIPAA. Other data that is not initially covered includes in-home telemetry data and user entered data include patient approved medical directives.
 
====App as HIPAA covered entity====
 
There are severe federal penalties for exposing covered data.
 
* The app must be approved by a trusted authority to even open a communications channel to an EHR.
 
* This applies to downloading data from the EHR, as well as uploading it to other sites.
 
* Potentially exfiltration of data from the domain of HIPAA covered entities could be down WITH EXPLICIT APPROVAL by the user of the app.
 
* Rules of consent to release data could be rigorously enforced and violator could be thrown out of the federation. All federation access would be verfied for every connection to any other federated entity.
 
 
====App is not protected by HIPAA====
 
Only regular FTC protections for user private information apply.
 
* There is an effort to create a "Code of Conduct" for apps that can acquire PHI from an EHR that is in development by the CARIN alliance.
 
* It is not clear that if a large company, like Apple, created a Health app, would then decide not to agree to the code of conduct. Could they be excluded from access PHI given the access rules of the ONC?
 
 
===Recovery and Redress===
 
# lost phone
 
# phone migration
 
# Phone hijacking
 
 
===Access Codes===
 
The access codes are created with the permission (consent) of the Patient or the Physician. Each grants access to a limited set of data.
 
* Patient Credential access code provides access to the identify proofing evidence created by a Covered Entity.
 
* Medical Records ([https://tcwiki.azurewebsites.net/index.php?title=EHR EHR]) access code provides access to a particular patient EHR or a subset of the data in an EHR.
 
 
===Biometric Factors===
 
* Finger print
 
* Palm print
 
* Iris scan
 
* Face scan - particularly dynamic scans where the patient must rotate the head in response to commands from the authenticator.
 
* Behavior
 
 
===Knowledge Authentication Factors (from KBA)===
 
* Medical records have information that can be used to assure that the patient is correctly identified and that the patient is matched to the correct record. Clearly there are privacy concerns that must be met if this method is to be acceptable to the patient.
 
 
===Federated Trust===
 
*In a federated system there will be at least one trust anchor that provide trust certificates to the members of the federation.
 
*[[Trust Framework Membership Validation ]] describes the API to establish trust among covered entities and for the benefit of the patient.
 
*[https://github.com/TransparentHealth/poet Pre Oauth Entity Trust] describes a means to represent third-party application endorsement for health care applications. POET’s goal is to help consumers distinguish between applications that have an endorsement versus applications that have no pedigree (i.e untrusted and could be malicious).
 
  
 
===Record Matching===
 
===Record Matching===

Revision as of 20:01, 18 September 2019

Full Title or Meme

There are two contexts where Trustworthy Healthcare Provider is defined and so there are two memes that it covers:

  1. Providers that share trust among themselves to know that Protected Health Information (PHI) can be shared as the correspondent is a recognized HIPAA-covered entity.
  2. Providers that may be trusted by patients (or other end users) with their PHI with their Consent.

Goals

  1. Trust Registry of IDEF (Trusted Identifiers in Cyberspace) for healthcare industry in US.

Context

  • The wiki page Health Care Profile establishes the context for this page.
  • Details of the Trustworthy Healthcare Ecosystem explores the full ramifications to every aspect of information exchange in the ecosystem.
  • For a healthcare trust ecosystem to have value for a provider those providers must agree among themselves as to the criteria for entry into the registry of that ecosystem.
  • For a healthcare trust ecosystem to have value for a patient, these criteria are important:
  1. The patient can know that their medical and other records are safe within any provider's Electronic Health Record (EHR) database.
  2. The user can determine the trustworthiness of other providers that are seeking access to their medical and other records.
  3. Trust begins when a doctor sees a patient for the first time with a current complaint. The patient provides some identification and subjective information about their history and health problem and then the doctor does an objective (clinical ) exam which may or may not validate the initial complaint. That is the start of a trust relationship.
  4. After the patient has visited a PCP (Primary Care Physician) they are entitled to acquire their medical records. Where the records are stored digitally, the patient must be give online access.

Real World Members

  • The patient and the physical places where the patient goes are all legal entities and have legal obligations.
  1. Patient - is the person receiving care and the one that "owns" the rights to the information. (In some cases the patient allows other users access to their PHI.)
  2. Patient Support - is source of the user phone or computer and the code running on the user's phone or computer. It is trusted to protect the patient's credentials and medical records from disclosure.
  3. Covered Entities - Any entities that is covered by HIPAA rules including the places that a patient goes for medical care. They will all be recognized as such in the digital world by virtue of trust certificates. Some of these will perform Identity Proofing. The services that they provide are listed below.

Names use within the Ecosystem

There are a variety of identifiers used by the real-world entities described above. These are the handles to which trust is assigned.

Problems

  • If a PCP (Private Care Practice) is the source of Identity Proofing to be used with other providers, can we provide the standards that would relieve them of liability for misuse?
  • For IAL2 some means of protecting the patient authentication credentials must be available, e.g. FIDO U2F (Web Authentication) or the smart phone (with a trusted execution environment) itself.
  • Medical records can apply to both state and transaction records.
    • When the patient asks for records they have the right to get everything that is permitted by law.
    • When a physician makes a referral they typical send the relevant information relating to that condition (with patient consent).
  • The following problem areas were identified at the CARIN Health Care Digital ID Summit 2019-06-04. In order to best proceed the definitions have been amended slightly to make the terms into distinct entries in a Trustworthy Healthcare Ecosystem taxonomy of Identification.
  1. Identity - is the matching of the real live biological human to the records necessary for billing and patient care. The level of identity proofing that occurs here will be situationally determined as would be the case (e.g.) for new versus returning patients. It will only be established by patient-present use cases, except in some low volume use cases.
  2. Authentication - is the matching of the user of a digital computer system to the patient identified in problem 1. Our scope in this ecosystem is limited to the digital exchange of credentials and trust statements over the web.
  3. Trust and Federation - is the trust between different organizations in different federations as well as trust of the patient for the web sites that they visit.
  4. Consent - is required for all information-sharing events. In some cases blanket consent has been provided. In some venues consent has a limited life time.
  5. Matching of the Patient - to their electronic health records can occur in both patient-present and remote access use cases. Prior to the time of patient matching, Identity (problem 1) and Authentication (problem 2) must have been completed. (But note that matching an incoming patient to the medical records may be included as a part of the Identity and Authentication solutions.) The most severe matching problems, when the patient is non-responsive, are out of scope here. The most common use case is presumed to be the passing of medical records from the PCP to some other provider and back again as modified.

Solutions

Relational Trust

The definition of (relational) trust is to believe in the honesty and reliability of someone or an entity you have known over time, made a good faith effort to live up to an agreement to fulfill their commitment, be it a contract or handshake agreement. ‘Trust’ is a transnational catalyst, the chemistry that initiates an interaction and reaction if not abused. (So that the trust expressed as "known to the practice" can be transmitted to other participants.)

Online Trust

  • All online service endpoints shall be equipped with capability metadata that informs communications partners as to the services offered.
  • All online service endpoints shall supply certificates that can be verified to prove compliance with minimal health industry standards.
  • All compliant services that authenticate patients with IAL2/AAL2 will supply patient credentials as requested by other compliant medical providers.
  • All services that authenticate patients will make their own determination as to the patient's identity. Patient credentials are just one component of that trust process.

Certifications today are found for:

Patient Consent

  • We need to be able to capture the patient consent in a digital message and transfer that to another provider.
    • A taxonomy for how to represent the information requirements and risks to the patient must be in use by all providers.
    • Existing taxonomies of data types in the EHR is too technical to allow patients to make informed decisions.
  • The Patient must understands what information they have consented to share and what the risks to the patient are.
    • Also why the information is required to provide that care. (Transparency)
  • When medical records come from the patient that consent would also be captured and given to the new provider.
    • We need the ability to create a consent receipt for moving medical records from one provider to another provider.

Bindings

  • In a digital world binding are formalized with a cryptographic key signing a document that associates one entity with a collection of keys and attributes. For example the X.509 certificate.
  • For the organization the trust anchor (described in federation below) works.
  • For the patient some sort of private credentials are required for a formalized binding process. If the patient's credentials are well protected they can be used with an authentication protocol to prove that the patient is present for an on-line interaction.
  • The binding is then between the patient and the protected key pair. The public part of the key pair is, in effect, an identifier of the patient.
  • Bindings can be used to assure that any part of the system remains in the state that was known to be secure. For example, software should have a way to let any communicating party know who is controlling the interaction. Devices should all be able to make a secure statement about itself. It is also desirable for an instance of an app on a device to have a unique and secure identifier. This instance represents the life-cycle of that software on that device, not just the running instance at a moment in time. Even more trivially, each communication session should be conducted over HTTPS which will also be bound to the key that was used in interchanging secure messages.
  • Note that the patient may be considered to be bound to their medical records, but that is an organizational relationship that is core to the patient matching problem. While it is not feasible for a medical record to have an identifier (index key) that is itself cryptographically bound to the patient given current US law, it is possible to include cryptographically bound identifiers in the medical record.
  • The user agent will typically also create a binding between the subject and any service provider indicated the consent granted by the subject to the provider. In most cases the subject will have the ability to dissolve the binding and withdraw the consent. Contract law may override this ability to withdraw consent in some cases.

Record Matching

No patient is fully trusted when approaching the receptionist or any health care provider beyond the personal physician. The essential problem is that mistakes happen in health care and the wrong records get attached to the wrong human being. This can cause disastrous consequences. Ensuring that the provider that is immediately attending to the patient have relevant information about the patient is essential to good outcomes.

PatientTrust.png

Next Steps

The order and extent of these items is currently arbitrary and pending review by industry experts.

  1. Approve broad plan for proceeding.
  2. Create a Patient experience set of use cases to be tested (as a part of a broader conformance verification)
  3. Establish a specific plan for the sand box.
  4. Build a trust registry that can hold configuration/manifest data and respond to API requests for:
    1. Applications that are in conformance with Evolving CARIN best practices
    2. Trusted Provider sites with EHR - deploy at least two for the sandbox to allow a sender (e.g. a PCP) and a receiver (e.g. a specialist) for testing.
  5. Build a demo native app that can show how how a patient can acquire and use an app of their own choosing with authorization code generated by PCP
  6. Enlist an existing medical software vendor to build a demo app that can show how a patient can acquire and use an app from a PCP
  7. Create a trusted third party web site that can:
    1. Demonstrate how a patient can enter and manage their (e.g.) emergency contact information that is entirely under their own control.
    2. Demonstrate Identification and Attribute Provider (IAP) with IAL2 authentication capability

References

Internal References

On Kantara wiki pages.

External References

The TEF Draft 2 supports the Cures Act’s goal of advancing nationwide interoperability and is a key component of HHS’ and the Administration’s broader strategy to facilitate nationwide interoperability. HINs must agree on a minimum set of principles that enable trust in order to facilitate interoperability and the exchange of EHI necessary to support the entire care continuum. The TEF Draft 2 establishes a uniform set of principles that all HINs should adhere to allow for the trusted and secure electronic exchange of health information. Adherence to these principles will help improve the flow of EHI, providing patients with secure access to their information when and where they need it most.