Reader Bernard Farrell (who's got a terrific diabetes oriented blog) ask about my recent slam on USB connectivity for medical devices.

I'm intrigued by your comment "The good news is that USB connectivity is pretty
crude and poorly suited for most medical device connectivity applications and
should be eclipsed by Ethernet and wireless LANs." Given how successful USB has
been in the camera and MP3 device marketplace, why do you see it as 'crude'? It
may eventually be replaced by wireless, but I don't think this will happen until
wireless boards are as cheap as USB is right now. Anyway, I'm just curious here
and I've love to know more.

Medical device connectivity is all about workflow, not just getting data out of a device and into some software application. At one extreme you have the user look at a device, manually record the data, go to an information system, pull up the appropriate patient and key in the data, indicating the time the data were "acquired". This process is obviously labor intensive and prone to many types of errors.

The best kind of connectivity is that which becomes invisible, requiring the minimum intervention on the part of the user. The best connectivity at the present time is wireless where the user easily establishes patient context at the patient attached device. Data is then acquired automatically (the frequency and specific data elements) according to a predefined protocol or configured by the user. While attached to the patient, the device system manages the connectivity and patient context (not a trivial task in some use models) and notifies the user only when there is a problem requiring manual intervention. Finally, when the device is removed from the patient, the patient context and data connection are broken down, freeing system resources for other patients and connections.

USB connectivity is "crude" in that most of the manual steps outlined above are still manual; the USB only eliminates the need to manually record and enter data. The user must still be in the presence of the device to extract the data (via the USB drive), and must convey it over to the receiving computer, pull up the correct application, indicate where the data is to go, and properly identify the patient. The two problems here are staff productivity (and the likelihood that data collection will be late or not occur at all), and the risk of attributing the data to the wrong patient. Those are two big problems, both of which could result in an adverse event.

There are other forms of "crude" connectivity that suffer from similar productivity and safety problems. Connectivity that does not establish patient context in the patient connected device (common with serial interface connections) has patient identification risks. The device docking approach used by many point of care diagnostic devices runs the risk of delays in getting data into the chart - or possibly never getting into the chart. And wired connectivity of any kind is a frequent victim of forgetfulness, yanking sockets out of walls and/or damaging cables as devices are moved while attached to the network.

Connectivity also impacts the patient. Wearability, mobility, and patient comfort are affected by the quality of the connectivity implemented by a device. Poor implementations can result in false positive alarms, leads-off situations, and extra work and hassle toileting the patient. Even open heart patients are encouraged to start waking shortly after surgery, and a portable device that's not wireless is, well, silly.

To all of this, you can add my previous observations on the data security risks posed by USB connectivity. Pictured right is another snazzy USB drive, known as the stick.