Cardinal Alaris Smart Pump Seminar
Yesterday I presented at a workshop in Seattle held by Cardinal Health. The topic was Wireless Infusion Information Technology and was educational – being remarkably free of sales pitches for a vendor event. There were about 17 attendees, which I thought was pretty good for a mid market city like Seattle. I promised the attendees that they could download my presentation, so here it is. As you might imagine, I talked about medical device connectivity and differences between proprietary end-to-end solutions and enterprise solutions like the one from Emergin (a Cardinal vendor-partner).
Alaris was the first medical device vendor to develop a network based solution where the network was intended to be part of the hospital IT infrastructure and not an embedded controlled private network. The decision to embrace the general purpose computing environment, versus a tightly constrained private network was radical at the time. In hindsight, given the enterprise-wide deployment of infusion pumps, Alaris had little choice – no hospital would deploy a second wireless LAN just for their pumps. Since Alaris' move, Draeger is the only other vendor I'm aware of to fully embrace the hospital IT infrastructure rather than force a private network on customers. As patient acuity continues to raise in all patient care areas and medical devices are deployed enterprise-wide, vendors will have little choice but to take the route Alaris and Draeger have taken.
Cardinal currently has 111 hospitals running their smart pumps. That equates to 32,000 pumps with 55,000 channels. M.D.Anderson in Houston has the largest single installation with 1,300 smart pumps. Alaris has installations with both InnerWireless' passive DAS (distributed antenna system) and Mobile Access' active DAS.
Do you remember all the buzz about bar coding 5 years ago? It seemed everyone from the government to the Leapfrog Group were breathless about the patient safety potential for bar codes in patient identification and meds administration. In the past 5 years, adoption of bar coding at the point of care has barely moved, while in the same period smart pumps have virtually become a standard of care. I think the fact that many infusion pumps are leased had something to do with that rapid adoption. And the poor adoption of bar coding can be attributed to one thing – workflow. The 5-rights are a workflow hit for caregivers, especially during peak busy times. If the meds administration workflow does not result in a net reduction in work, successful adoption may never be achieved. The win for caregivers is the EMAR or electronic meds administration record. The EMAR can significantly reduce meds admin paperwork, providing the nurse with a significant net benefit. Even with an EMAR, how effectively workflow is automated is the make or break adoption factor.
Also presenting was Bob Schaefer from the IT department at Kingston General Hospital. Bob described their experience justifying and installing an enterprise-wide wireless LAN. They have installed a Cisco 802.11a/b/g WLAN with 185 access points (APs) covering 800,000 square feet. They spent $200,000 on Cisco APs (plus some management tools and switch upgrades), $110,000 on site surveys, and $90,000 cable drops. The big hidden cost for them was the infection control costs of getting into the ceilings. Their annual costs are around $20,000.
In addition to designing their network for coverage they had to do some rework based on what Bob called “traffic patterns.” Users and devices seemed to congregate in higher concentrations in certain areas (especially the ED), sometimes overwhelming the AP covering that area. They used a high density deployment model in the ED where they turned down the power on their APs and doubled the AP density. On nursing units they are looking at a different deployment strategy with paired APs to double capacity. Many WLAN distributors unfamiliar with hospital environments throw up APs every 100 meters that result in poor to non existent coverage in some areas. Even experienced folks usually miss the capacity variable – you must anticipate the number of wireless devices (VoIP phones, PDAs, COWs, RFID tags, medical devices, etc.) that could congregate in a particular area and ensure your network design provides sufficient AP loading capacity.
Kingston's WLAN is segmented into 4 VLANs to provide optimized security, performance and network management for specific applications. They have a VLAN each for infusion pumps, wireless VoIP phones, point of care computing devices, and public access to the Internet. I asked Bob if they would add other wireless medical devices to their infusion pump VLAN in the future. He noted that if future devices included UDP they would, but if they did not use UDP (a hard to secure internet protocol) they would put them on a separate VLAN for security reasons. There are a number of continuous patient monitoring vendors who currently use UDP to stream physiological waveform data across the network. Bob suggested there were better ways to stream data that was easier to secure.
Tom Steinhauer from Cardinal wrapped things up with a great description of how and why Cardinal's bar coding solutions have evolved over time. Tom laid out each piece of technology and showed how workflows mapped back to the system, and the changes Cardinal has made over time to better match how drugs are actually administered. Of particular interest was his description of integrating their IV meds administration system with third party systems that support “pills, shots and drops” – currently not a simple endeavor.
Pictured right, folks start to trickle in during the pre seminar meet-and-greet.
UPDATE: Dan Pettus, the MC for the above seminar, emailed me with some clairification regarding their installed base. Cardinal has about 400 hospitals that use their “smart” pump – i.e., pumps with Guardrails. The 111 hospitals mentioned also include the server.
Barriers to Connectivity
There are a lot of forces driving change in health care. In connectivity, device vendors are looking to workflow automation surrounding the device to grow their business and provide competitive advantage, while health care IT (HIT) vendors look for growth in diagnostic and therapeutic processes closer to the patient. To a great extent, they’re going after some of the same markets.
Charles Fox is a clinical workflow expert who has spent years automating care delivery. Earlier today he sent me the following industry observations that I thought were particularly insightful.
There is an obvious tug-of-war in progress between HIT suppliers and medical device suppliers. For example, the infusion device suppliers are wandering into the IT space with server-based systems, gateways that are a proxy for the infusion device at the bedside. HIT suppliers are getting more serious about the infusion device business as well because it represents a piece of the care process they don’t own.From my vantage point this is driven by several factors. Infusion pumps themselves are becoming a commodity and the size of the market is tending toward shrinkage. As infusion pumps get smarter they increasingly become part of the medication administration process and less like passive purveyors of medical data. Programmable infusion pumps create a special case in the connectivity space because they represent a combination of servers and systems with interdependencies that are not passive and can do real, direct harmto a patient.
The above cause a degree of panic among infusion system suppliers who see the handwriting on the wall, “…how do we grow when we don’t own the medication process, and how do we manage the liability of introducing a programmable device into the medication process driven by orders in another system?”
There is a tendency for issues like this to put up walls between HIT and infusion system suppliers. Device suppliers can creep just so far into care delivery before they run into two major issues: how do you manage care process and workflow across dissimilar systems, and how do you decide who gets to be in control?
But some say this will all be solved by interoperability. It may be solved eventually but today device suppliers and to some extent HIT suppliers are disincented to implement interoperability because it immediately draws a line in the sand around their market.
So what is the solution? I guess I’m altruistic enough to think that a group of device suppliers could get together with a group of HIT suppliers and decide to “do the right thing” based on what is in the interest of safe and effective care. There are attempts from the device side in the form of IEEE 11073 and attempts from the care delivery side in the form of CIMIT…..with the FDA trying to shove them together. But this is another case of government having to force an issue because industry can’t get their act together.
Government mandates are a handy explanation for stock analysts worried by a charge against earnings ostensibly due to government regulations….charges that are really due to shrinking market and failure to act.
Connectivity (or medical device integration if you’re a HIT vendor) is not something you throw a product manager at and forget about. This is some serious strategic stuff. Another fact that Charles doesn’t mention is that it’s expensive too, way more than a few million dollars for a new product; especially if you screw up.
If you think about it, the big guys like GE, Siemens and Phillips have their fingers in both the HIT and the device industries. And as a rule, even they can’t integrate within their own product lines. All that opportunity, and they can’t get it together. There has always been, in my mind, an oil and water relationship between software and hardware folks even within the same company. Fundamentally they see the world differently. Companies that try to do both tend to be good only at one….and the other struggles. The hardware guys tend to be king of the mountain.Point is it may take the government to disrupt things…..but as is usual with government mandates, no one will like the outcome…..and the patient is lost in the dust.
Charles comes from the software side of things, where the hardware guys seem like the king. Ironically, the device guys think the HIT guys are king. Maybe these contrasting perspectives simply highlight the fundamental differences in the business models and interdependencies between software that runs on general purpose computers and software that runs on embedded systems.
What’s needed is a new kind of company, a chimera that blends the best part of both HIT and medical device companies. One company that seems to be transforming themselves into something new and unique is Cardinal Health. Current business opportunities at and around the point of care are tremendous for a start up or an established company with the right kind of solutions. The kings that look so strong and capable today could be shadow of their former selves in a few short years.
Pictured right is the mythic chimera.
Children's Hospital in Denver Buys New Network
Prompted by the construction of a new hospital, the Children's Hospital in Denver is installing a new enterprise-wide network. The project will link 11 Childrens Care centers, 2 other campus facilities, 400 outreach
clinics, and 1.44 million square feet in a new 10 story 236 bed hospital
building. The deployment will include 800 access points, 5,000 VoIP phones, centralized
voicemail, conferencing, and other collaborative applications. The
hospital has already installed a distributed antenna wireless system
from Mobile Access.
Whew! And all for a mere $25 million.
pick systems that will not require a forklift upgrade” by the time the
ROI is realized. In choosing technology, he said, the hospital looked
for “end-to-end” or “big-picture” offerings rather than “individual
components “you had to tie together with wires.” He said the hospital
preferred to work with one vendor rather than multiple vendors. It
wanted to streamline – and make things the same wherever possible. For
example, it ordered the same cables for all sites.
Sadly there's no mention in the story about, you know, the really interesting stuff – like existing medical device networks, alarm notification and systems integration at the point of care. What applications is the network going to support? How far into the future did they look, and what applications or requirements did they anticipate? They've got coverage covered with Mobile Access, but what about capacity – how many wireless devices in any one location do they plan to support in 5 years? While not mentioned in the story, Emergin is reportedly part of the solution.
Frymire claims their payback on the new network ranges from less than a year to seven years.
The average return on the investment, he said, is estimated at five-six years. Calculating the return on something like this is sort of silly – you can't run a hospital these days without a network.
Geez, I just realized I wrote this whole post without mentioning the network vendor – like there's really any choice other than Cisco. I'm surprised the Cisco PR person didn't work “medical grade network” into the story somewhere.
Pictured right is a satellite view of the Children's Hospital Denver – it's a big place – from Keyhole.com (now Google Earth), via the Andy Blog.
InnerWireless About to Release Spot RFID System
Chris Click and Tom Eagle from InnerWireless were kind enough to set up a conference call to update me on their new Spot indoor positioning system.
As they approach GA (that's “general availability” in vendor-speak),
InnerWireless has a total of 3 beta sites. Here's a quick run down:
St Lukes Hospital in Kansas City – this is an ED tracking application (both staff and patients) integrated via Pango to a McKesson application. Intel provided the project management and InnerWireless provides the infrastructure. The beta is live and working.- Comers Children's Hospital at the University of Chicago – this is a straight on asset tracking installation. The goals are to of course validate system works, and then convert asset over-buys into savings. This site is also up live and working.
- Vasser Brothers Medical Center in
Poughkeepsie, NY- This is a combo asset and patient tracking installation. They installed the infrastructure last September, and plan to go live in November.
IBM is going to do workflow research to benchmark some of the dynamic
processes in acute care delivery.
On the product front InnerWireless has been working to reduce tag size while maintaining a 3-5 year battery life (at a 1 minute or greater reporting frequency). A 5 second
reporting interval gives a 1 month battery life. At 1 minute and above, battery
life becomes more of a battery shelf life issue. The final production tags are sealed and sterilizable (gas autoclave or disinfectant wipe). Testing for temperature sterilization is underway, but not expected to be a product claim.
InnerWireless considers the beacon to be their big differentiator. The beacon is what communicates with the tags to gather data to determine location. The beacons use an 802.15.4 mesh network to communicate with the RTLS engine. They are battery powered and don't require either a wired network connection or a power outlet. They are what is euphemistically called “lick and stick.” You can get more technical details from my interview with Alastair Westgarth, Senior VP Product Service Line Management here.
If an RTLS (real time location system) costs between 75cents to $1.00 per square foot, the InnerWireless Spot system would
be in the lower half of that cost range.
General Availability for Spot is planned for December 2006. Right now InnerWireless is focusing a
lot on their contract manufacturing because prospective customers are
looking for thousands of tags per site – that will require high
manufacturing volume from the get-go. Needless to say, InnerWireless is taking pre-orders and scheduling initial installations now. (I guess that's why the called me…)
At first thought, InnerWireless may seem late to the RFID party. I'm wondering if they've hit optimal timing. Pioneers like Radianse have built market awareness of RTLS solutions
along with their company – now the notion is pretty well established, early innovators have installed and it
appears the early majority of the market may be ready to adopt. As we enter the tornado phase (of market development (a Geoffrey Moore term), first-to-market will count for less than good execution in sales, installation and support.
While on the call, I asked Chris and Tom the biggest sales objections that they get from hospitals. The biggest objection is no surprise; it revolves around existing WiFi infrastructure. “We already have an infrastructure deployed, why not add software and tags and go?” is what they hear frequently. Hospitals assume there is an inherent cost savings by leveraging existing WiFi infrastructure – WiFi based RTLS vendors have effectively created that meme.
The second biggest sales objection is a resistance to tracking staff – this revolves around union's concerns (both real and anticipated), and staff's feelings of Big Brother and a resulting lack of privacy. Like the first objection, this one can be overcome with some explanation and education.
The most critical issue in selecting an RTLS is resolution or positional accuracy (besides cost). Quick and dirty zone level accuracy is fine for most asset tracking applications. But clinical applications like nurse call automation or alarm notification (you know, like- where is that ambulatory patient in VTAC?) require room level accuracy. Logistics applications for emergency and perioperative departments also require greater than zone level accuracy. Patient flow applications like Awarix and StatCom use position data to automate transactions and document workflow. I'm not saying you can't get better than zone level accuracy from a WiFi based RFID system, you can. But you certainly won't get it from a network with only passable coverage and access points thrown up every 100 meters.
InnerWireless makes a “room level” claim (95% assured) with Spot, and has done testing down to the 1 to 3 foot range. One of InnerWireless' greatest strengths is their in-building RF engineering expertise, taking in to account things like RF propagation through walls and objects (food carts, carts full of saline solution, etc.) found in a hospital. They've done their best to leverage that strength to create an indoor positioning system that delivers optimal accuracy – and maybe overcomes some of the inherent limitations in the 2.4 GHz band.
Pictured right is someone installing a “lick and stick” beacon.
DISCLOSURE: InnerWireless arranged the conference call that resulted in this post (and they provided the snazzy photo too – a real plus!). Since they're about to release their Spot system I thought the topic was newsworthy enough for the blog. I was not paid for this post, nor would I accept remuneration for any writing here. (I have thought about selling ads or sponsorships.)
I welcome calls and the occasional press release from any vendor or provider who'd like to chat about products, their experience and/or the industry. As long as the result is informative and newsworthy (and I have the time), it will most likely make it into a post. I assume that everything discussed is confidential unless it's
understood up front to be public; I'll ask permission if I want to
write a post about it.
This site gets between 300 and 400 visitors per day now, has over 200 people who subscribe to emails or RSS, and has an average visit length of between 2:30 and 3 minutes. And of course my readers are the smartest people in the industry.
UPDATE: Here's a technical article on ZigBee that might interest the rocket scientists out there (it's over my head). The story does raise the perennial IT system question about scalability – a topic I wish I'd asked about during my call with Tom and Chris. Perhaps someone can comment on that topic in respect to InnerWireless' Spot system specifically, and ZigBee in general.
Housekeeping
A number of readers have been thoughtful enough to tell me of some text getting covered or cut off by the right hand column. I've deleted the “Yesterday's Top 5 Links” box (it was sort of boring anyway) and republished the home page. That should fix the problem. The roll over link score from My Blog Log will still work, and I'll still get the statistics on which links you find the most interesting.
If (or when) anyone has any other problems, please let me know.
