Jack Mason, the master blogger at HealthNex, has organized a group of bloggers to embark on a Blogposium. The goal is to turn our individual blogs into a "collaboration
confederation" that will work together to add dozens of new entries to
the Clinical Informatics Wiki (Clinifowiki), an extension of the Clinical Informatics
Review. This experiment in "collaborative, paticipatory blogging" is our contribution to the industry and will hopefully add important content to the Clinfowiki. Here's a list of participants and their topics:

News about the Blogposium has been picked up by eWeek and Business Innovation Insider.

So, here's the first draft of my contribution, Integrating Medical Devices into EMRs. Everyone's encouraged to provide comments, edits, suggestions, and criticism - either via email, comments to this post, or at the post on Clinfowiki.

Introduction and Background

Quantitative data from medical devices makes up a significant portion of a patient's electronic medical record.

The typical patient's electronic record will include trended vital signs data, notations for certain medical device alarms, and documentation concerning therapy provided. The types of data acquired from medical devices include discreet numeric data, waveform data, and events like alarms or when therapy was started and stopped. Some device data needs to be manually qualified, such as breath sounds or the patient's position when reading blood pressure.

Problems with the manual recording of medical device data are well understood. These problems include illegibility and incorrect data entry through transposition of data or entering correct data into the wrong field. Manual recording of data frequently results in delays in data availability; caregivers are frequently interrupted while capturing vital signs, resulting in delays in getting data entered into the chart.

Users adopting an EMR without medical device data integration often complain about "double entry" of data. This attitude is not the result of a workflow change (the only difference is typing data into a computer vs. writing it into a paper chart), but rather a change in the user's expectations. Users of an information system expect things to be automated and when basics like data acquisition remain manual tasks they become frustrated, impacting adoption.

Medical Device Categories

Devices feeding the EMR can be divided by location, one group of devices is found at the point of care, the other lies beyond the point of care in diagnostic and therapy departments. Devices found beyond the point of care are usually integrated through departmental information systems (PACS, cath reporting systems), or traditional physician or technologist reporting systems. Point of care devices can be divided into 4 categories; continuous-data devices, spot data devices, portable therapeutic devices, and portable diagnostic devices.

Continuous-data devices include patient monitors (including telemetry transmitters), "smart" infusion pumps and ventilators. These devices generate continuous data like waveforms, and generate life-critical alarms that must be rapidly communicated and displayed. These types of devices monitor patient parameters and provide surveillance through the display of real-time waveforms and/or alarms.

Spot data devices are used to take readings of physiological parameters on an as needed basis. The most common example is the spot vital signs monitor. These devices are moved from patient to patient to record vital signs at a frequency ordered by the attending physician. In the past some vital signs monitors were connected for extended periods to patients to take readings on a regular basis, e.g., every 5, 10 or 15 minutes.

An example of a portable therapeutic device would be a dialysis machine. These types of devices are rarely integrated; the technician must manually document their episode of care.

The final medical device category is the portable diagnostic device. This category represents quite a range of devices. Traditional portable diagnostic devices include ECG carts and portable diagnostic imaging modalities. Lab test are moving to the point of care with a variety of devices, some wirelessly enabled, and some deployed as "mini-labs" that are located on nursing units. Connectivity for this category of device is challenging, and rapidly evolving. Fortunately for those deploying EMRs, portable diagnostic devices will be integrated into separate diagnostic information systems that will feed diagnostic reports into the EMR.

Patient Context

Patient safety requires that data from medical devices be reliably associated with the correct patient. How patient context is established and maintained varies with the type of connectivity used, and impacts workflow and the reliability of data that flows into the EMR.

Medical devices that rely on RS-232 serial ports for connectivity are "dumb" - they know they are hooked up to a patient because they're generating data, but they have no knowledge of the patient's identity. Serial port connectivity relies on hardwired connections to a specific serial port and software on a back-end server to know which patient goes with which port. Patient context is typically established by a user at a central station who knows that patient X is in room Y. This type of connectivity is hard to deploy in a wireless environment because there is no fixed association between a port and a patient room or other physical location.

Medical devices with network connectivity are "smart" because they know the identity of the patient they are connected to. Establishing patient context at the point of care can be done by entering the patient's name or ID into the medical device, or selecting the patient from a list that is displayed on the medical device. This method is ideal for wireless connectivity because it is independent of physical location. Devices with network connectivity have the ability to display the patient's name, providing a visible check for proper identification and patient context.

There are a few devices that provide serial connectivity patient context through a hardwired network connection. These devices mimic a serial port by using a fixed IP address or MAC address to identify the device at a central station where patient context is managed. Current practice in hospital network management precludes the use of this scheme except on private networks.