Cloud EHRServer (beta)

Working with anonymous clinical information

This guide will help you understand the design principle of having only anonymous information stored in the EHRServer.

Introduction

Most health information systems manage clinical and demographic information on the same system, and even store both types of information on the same physical database. There are some reasons why this might not be a good idea, and, technically speaking, not the best software architectural design.

The two main reason why having a physical separation of clinical and demographic data are: Security and Enabling Data Reuse.

Security

In the last couple of years, a lot of data breaches and security incidents occurred in the health care domain. Just Google "Health Data Breach" and you will find some results.

Think of what would happen if clinical information was stolen, but it didn't contain any demographic data. No names, no addresses, no phone numbers, no SSNs, no document numbers, nothing, just plain clinical data. It would be worthless for the stealers, while privacy and confidentiality is intact. This is one advantage of having only anonymous clinical information.

Enabling Data Reuse

The primary use of clinical information is health care. There is a wide range of uses for this kind of information beyond health care. This is called "secondary uses". The differences between primary and secondary uses are well defined by the ISO 18308 standard.

Between secondary data uses we can find:

  • Education
  • Research
  • Decision Support
  • Management
  • Policy-Making
  • Statistical Analysis
  • ...

None of these secondary uses need to identify individuals. So, granting access to the "Clinical Repository" for people or systems to enable the secondary uses won't put privacy and confidentiality at risk if all the clinical information is anonymous. This is a core concept of the EHRServer design but is nothing new: this requirement is part of the openEHR specifications.

Key design concepts

Meaningless identifiers

Since EHRs need to have some kind of identifier, we have choosen to use UUIDs as identifiers because:

  1. Are meaningless, are not person or patient identifiers
  2. Universally unique, allows distributed generation, maintainin uniqueness
  3. Are compliant with the openEHR specifications

No references to person or patient real-life identifiers

Real person or patient identifiers like SSN, National Identification Card numbers, Drivers License umbers or Medical Record Numbers shouldn't be used in the Clinical Repository. For example if you use the REST API to create an EHR, a patient ID should be provided. We recommend to use UUIDs for that. So no personal IDs are stored in the EHRServer or communicated ot it. This reduces the impact of any type of security incidents associated with the clinical repository, keeps the information anonymous, and maintains privacy and confidentiality.

Keep generated EHR and patient UUIDs safe

When creating an EHR, a patient ID should be provided. Above we recommend that ID to be an UUID. To have access to the EHR, client apps should store the patient UUID and/or the EHR UUID. Those identifiers are the only link between your patient's demographic data and the clinical data stored in the EHRServer.We recommend to store those identifiers in a safe place, encryption is encouraged, allowing decrition only for authorized users, and, if possible, making a little dificult to find the link between a patient demographic record and it's EHR and patient UUIDs. One idea would be to store the UUIDs in a different database, alongside with the patient ID in the demograpihc repository. These are all extra security measures. Even if a hacker can access your patient's demograhpic data, might not have access to the UUIDs. If they gain access to both, they still will need the Private API Key to get data from the EHRServer. Of course the API Keys should also be secured, but that is not related with this specific guide.

Information linkage

Users should be authenticated and authorized to access both demographic and clinical data. Users with access only to clinical data won't be able to know the identities of the patients associated to each EHR. This kind of access can be granted to allow secondary uses, like Clinical Research. Which user has access to which data is controlled by you. You also control the demographic repository and the UUIDs.

Accessing clinical information

To access data of a specific EHR, or to cimmit data to it, client apps should provide the EHR UUID to the EHRServer REST API. Only authorized users should be able to get the UUID and access the REST API with that UUID. The UUID access should be controlled by your application, while the EHRServer verifies authorization rules internally.

Conclusion

Remember to use UUIDs not real IDs for patients when communicating with the EHRServer. Also take into account that security is a multi-factor concept, and part of the security care relies on how your apps are developed. Take into account the recommendations of this guide to reduce risks while enabling clinical data reuse.