Five basic network architectures should be consider when deploying an interoperable HIT platform, which are shown in the figures below with each circle representing a node (a node is a computer or computerized device with a communication link to the network):
  • STAR is a centralized structure, which is sometimes referred to as hub and spoke, in which all node communications are controlled by a node hub in the center
  • MESH is a federated/distributed node structure in which any nodes can communicate with any other nodes
  • TREE is a store and forward or distributed STAR, in which certain nodes can communicate with certain other nodes in a hierarchical manner (as reflected in an organizational chart).
  • BUSS is broadcast structure in which every communication is sent to all the nodes, but only the ones designated to receive it can access it.
  • RING is a structure in which each node is connected to two other nodes, which means when distant nodes communicate, the message must travel through all intermediate nodes.
Each has a use that can be related to a business need.

The STAR has a background steeped in mainframe System Network Architecture (SNA) technology. Dumb terminals acted as user nodes or logical units and the application environment was maintained by the mainframe. The growth of Automatic Teller Machines (ATMs) grew out of this construct. It wasn’t necessary to have two ATM’s talking to each other. All communications in the STAR must pass through the hub (or intelligent node point) at the center, which means it has a single point of failure. All policy decisions in this architecture are centralized, which is similar to organizations with a central decision-making body.

The MESH grew out of the need for a distribution network in telephone, power, and water or oil pipeline businesses. Characteristics of the MESH include the following:
  • It may have many interconnections since any nodes can connect to any other nodes.
  • Each node has a purpose and the intelligence to make decisions in near real time. For example, determining what processes to monitor and control, databases to access, data to send to other nodes, data to accept from other nodes, data transformation and analytics to perform, reports to generate, etc., as well as when to do these things.
  • Mainframe applications were the original nodes; you could construct a mesh of interconnected mainframe applications. Some of the applications (e.g., spreadsheets), however, just got too expensive to be maintained on the mainframe. So, when the Personal Computer (PC) and telecommunications were integrated together, the intelligent node was born and applications were migrated off the mainframe and onto the nodes. The computing industry enabled this by giving birth to operating systems on PCs that could leverage Transmission Control Protocol/Internet Protocol (TCP/IP). This network management technique is analogous to how you carry on a conversation between two people on a telephone call. You are the intelligence that manages the protocol (TCP) and the telephone switched network provides the under lying transport and routing (IP). You don’t care where the other people are located or how your call is routed; all you need are their telephone numbers or computers’ IP addresses. The underlying network takes care of all of that before you’re connected. If the telephone network can’t find the remote party, it returns busy. In the Internet world, email was invented and you busy signals became a thing of the past. The serving nodes just placed your mail in queue. That is the same thing being done for voice mail.
  • When the nodes in MESH become “hardened” — meaning that they become self-healing and have alternate routing paths available for traffic to flow in other directions given a failure of any one node or edge — the system is made more reliable, available, sustainable and supportable (the “RASS” criteria).
  • The MESH architecture of distributed nodes satisfies many of the needs of interoperable RHIOs because (a) it allows for immediate node to node communications without the need to traverse a network to get a permission (or policy) from the hub and (b) it meets RASS criteria. These advantages are critical in healthcare since it is a 24X7 operation managing a large and complex assembly of professionals and data sources at many different locations, and handling a full range of patient services (i.e., catastrophic, at-risk, chronic, and end-of-life care).
  • This means that when we discuss HIT interoperability, the MESH network suits it best because it enables any node to communicate with any other node at any time, from anywhere, and by anyone, thereby covering all people, places, and conditions.

The TREE architecture reflects the way many organizations work. i.e., by having hierarchies determine who receives and can respond to a communication. However, in a distributed world of “shared governance” — a system that enables independent parts to operate collectively — the TREE is ineffective. For example, a church depends on the local autonomous parishes and parishioners to work together for the common good of the community. The local parish, in this sense, is an intelligent node in a network of parish nodes. The DOD is another example since it expects units on the ground to operate autonomously, while coordinated efforts by many related groups focus on a specific objective. In other words, when parishes or units are connected to each other by common concerns, they link together into the MESH architecture. Since Regional Health Information Organizations (RHIOs) are about shared governance and Health Information Exchanges (HIEs) are about data sharing to improve care, the MESH is the best architecture for them.

The BUSS architecture is one in which each communication is sent to every node, but only the nodes for which the communication is specifically destined can receive and respond to it. An analogy is the broadcast media — a single transmitter, multiple radio or TV broadcasts, and many receivers. This is a very cumbersome architecture to use when connecting all the physicians’ offices, hospitals, clinics, ancillary service providers, patients and payers involved in a RHIO.

The RING architecture can be viewed as a train moving passengers or coal. The RING routes a token around to each node. If a node must communicate with another node, it must signal to one another and then pass the data or information through each node in-between them. As with the BUSS, this is a very cumbersome architecture to use when connecting everyone in a HIE or RHIO.

Incorporating Other Architectures in a Hybrid Mesh Network

A MESH can incorporate other architectures within it by connecting associated nodes; see the example below:

Using a Node Mesh Network in Health Information Exchanges and RHIOs

A Node Mesh Network satisfies HIEs' RHIOs' business needs.

HIE and RHIO Business Needs

  • Each of the nodes must have the ability to communicate with each other to share information in a way that enables everyone to gain and use the knowledge they need to make good decisions
  • The information shared during these communications support decisions enabling individuals to save a life, move a large volume of patients or victims in an emergency (e.g., evacuating a city), and to participate in ad hoc or established multi-disciplinary teams collaborating to diagnose patients and carrying out plans of care.
  • The communications may occur at any time, from any place, and be between any authorized persons and organizations.
  • The communicated information must be protected from piracy and available for extended periods, e.g., HIPAA requires healthcare data be kept for seven years and for a patient it should be available for a lifetime.
  • The information must enable prompt detection of and reaction to errors and fraud.
  • The information must provide the depth of understanding needed to support efficient and effective care.
  • The nodes must support shared governance, enabling multidisciplinary teams to collaborate in the decision-making process.
  • The HIT systems that support the information must be hardened and RASS compliant without bottlenecks and gridlock. It must solve the “last mile” problem, i.e., provide a stable, secure, and economical way to deliver connectivity at the final leg where a node sending a communication is received by other nodes.
  • The HIT systems should be low cost, with the potential to pay for itself over time through reduced transaction fees or reduction in the cost of care and better use of resources.
  • The HIT systems must be secure, providing encryption, authentication and authorization before any information is accessed or transmitted.
  • The HIT systems must accommodate paper-based (non-IT) facilities and legacy data systems.

Recommendations for HIEs and RHIOs

The node-to-node MESH network architecture meets HIEs' and RHIOs’ business needs best because it:
  • Has maximum flexibility:
    • Allows anyone to communicate with anyone else in any way, while at the same time allowing individuals in NON-MESH architecture to participate in the MESH.
    • Can use multiple connectivity options, i.e., radio transmission, satellite transmission, wire transmission, wire less transmission.
  • Has maximum reliability since it leverages the most reliable network in the world, i.e., the switched network.
  • Is the least expensive to deploy.
  • Is the most robust since there is no single point of failure.
  • Has unlimited scalability.
  • Promotes shared governance, while it supports rapid decision making at the level of the individual.
  • Provides early detection and correction of healthcare errors when and where they occur.
  • Enables patient profile data to be linked with RFID, thereby supporting information exchange in both IT-competent organizations and paper-based ones.

Next: The HIT Gap

external image meter.asp?site=s24wellness