ITAC notes - August 16, 2012


Announcements (5 min)

DHTS Epic Update, ---Art Glasgow (20 min)

Project Updates: 0365, MOOC, Box, --- Lynne O'Brien, Mark McCahill (20 min)

SABA Learning management System, --- Kim Andrews, Brian Aucoin (30 min)

GigU Update, --- Kevin Davis (15 min)


Announcements (5 min)

Robert Wolpert, guest-chaired the last meeting of the ‘11/'12 academic year.

Announcements will be rolled into "Project Updates" in a later part of the meeting so that Art Glasgow can get started on the DHTS Epic Update.

DHTS Epic Update, ---Art Glasgow (20 min)

Robert introduced Art Glasgow, who is the CIO for DHTS.

Art gave an update for the project. They have been working on for the last year preparing for this project. This project has been branded by Duke as "Maestro Care". This is an enterprise wide electronic health records system.

On July 18th they went with the first go-live. There are four waves of go-lives for ambulatory care.  There will then be a revenue cycle and then hospital go-lives.

The scope of going live is large. To get ready to go live, the team was working to bring across information from 138 clinical systems so that the system would be usable for clinics on day one. Much data had to be preloaded;

  • 36.2 million encounters
  • 4.5 million vitals
  • 3.8 million records
  • 3 million radiology reports
  • 333,000 future appointments

The size of the data conversion effort was massive, aside from all the configuration and workflow.

Wave one included 33 clinics on July 18th. This wave was the primary care clinics and included about 14,000 users. About 300 of the users were super-users who were part of building the content set and the workflow. About 600,000 patient contacts per year will be covered by this go-live. This was the smallest wave. This wave itself is bigger than a typical community hospital. Overall wave one was a resounding success from a physician satisfaction standpoint.

We also changed the way we support go-lives as well. This is the largest go-live that Duke Medicine  has ever done at one time. We brought in additional support desk staff and additional product specialist. Epic sent 12 specialists to be on-site with us.  We also put clinic liaisons and training specialists on site.  For the first week we did 3 conference calls a day; morning , noon & evening. This seemed to work pretty well. We also did hire actual go-live support to be in every clinic. Duke Primary Care also planned to go to a 50% schedule.  They planned to do this for about 3 weeks.  They are back to full schedule now & returned to full schedule faster than they anticipated.  We worked with them to cut to 50% so that they wouldn't be overwhelmed at the beginning. We also did go-live with another software package.  It wasn't just Epic that went live.

We built a foundation for a variety of things

  • Standardization across the system more than 240 clinics.
  • Documentation and billing discipline
  • Evidence based treatment
  • Rapid deployment of treatment changes
  • Compliance with legal requirements and better practices

We are now really focused on is wave two. It's a bigger wave, and it's an execution against specialties. This wave will be critical. The good news is that wave one was a success.  Wave two is happening and specialists will have to get on board as well.  The nurses have been the biggest help.  They see the value for patient care and they see how it brings together a lot of disconnected workflows.  They are anxious and excited to get involved.  We're actually re-opening the discussion of taking nurses out the hospital and training them to do  wave two go-live support and back-filling them with temporary nurses in the hospital.

We had an after action from Epic on wave one. Not everything went perfectly, but we got there and we got there as one of the top four or five smoothest that they've ever seen. This was largely due to some Herculean efforts of individual people.

Day 1 statistics;

  • 1400 patient encounters
  • 600 closed on day one
  • 1140 charges dropped by 9am
  • 1000 sessions by 11am
  • Average patient wait time was 22 minutes (this was partially because schedule was reduced by 50% - urgent care clinics got pounded)

Day 1 tickets;

  • 500 total
  • 225 open
  • Security issues dominated early
  • Fixed two ePrescribing issues
  • Progress notes weren't going to eBrowser
  • Results routing for anticoag drug

We saw typical issues that weren't huge patient safety issues.  All were closed within reasonable time.

To Date stats;

  • Average 1577 encounters (Almost back to full schedule)
  • 77 percent are closing the same day (Huge number)
  • Average time to close is 2 hours 30 min
  • 72,128 charges processed through 8/10/12
  • 24 min average wait time

We also went live with another software package called M*Model. M*Model is a voice recognition package that allows the doctor to put on a headset or microphone and instead of typing into the hub record they can just speak. There are two reasons we did this. The first is for provider adoption. We needed to give them something that was as easy as possible, like speaking into a microphone.  The second is that we don't want people free texting into the system. We want them putting structured data in, so that the data has the same semantic interoperability when we move it over to the research side of the house. M*Model was a somewhat controversial choice. Nuance is what you would typically think of as the leader in this space.  Our deployment model was heavily virtualized and the M*Model Citrix deployment model gave us some potential benefits.

We did have some issues. There was an incompatibility bug within a text edit module in Epic that caused crashes day one. We then found some performance issues day two. The good news is that it's working and we got it working within about ten days.

Significant system changes were installed July 30. System use Aug 7-11;

  • 78 unique users
  • 423 session (5.4. sessions per user)
  • Averaging 18 min of audio per session.

Billy Willisadded that this was difficult because it was very visible. There was much anticipation. It's pretty new.  There's a new architecture and some load balancing that is being tweaked.

Robert Wolpert asked if the audio files are saved. Billy said no, they are deleted when the information becomes discreet data within the electronic health record.

Art Glasgow continued, - we've deployed it in a somewhat novel cloud based architecture. We have doctors who travel from clinic to clinic. Their voice file and information has to travel with them via the cloud, otherwise they would have to carry a machine around.

We also did conversions over to MyChart.  MyChart is the Epic provided portal.  We also have HealthView, which is the self-developed portal. Each has things that the other doesn't. The long-term plan is to merge them together, but to get ready for that we have to convert all the accounts as well.  We converted under 400,000 accounts, both primary and proxy, over to the front end so that we can convert over. We're going to do that within wave two or three.

The last piece is after visit summaries. An after visit summary is when you see the doctor, he tells you what's wrong, gives you your medicine and hands you a piece of paper that says what we did and what we agreed on for you to do. This is important for two reasons. First, it's good medicine and it's good customer service. Equally important, it's a "meaningful use requirement".  The meaningful use program allows doctors to get additional money for meeting certain criteria under a pay for performance plan.

The meaningful use target is 50% printed and provided within 3 days. Already on day-one we are far ahead of the metric. This is due to the usability of the system.

To-date we've had;

  • 2541 tickets logged
  • 338 open (24 percent enhancement requests, 59 percent application, 18 percent technical, 3 percent device, 2 percent security)

Tickets open per hour is trending down. We're also in the middle of a consolidation. There are less calls and call times are down.

That's wave one. We're already focused on wave two, which goes live in 46 days. September first opens content build for wave three. We're going to get better at this every time and hopefully have more process discipline every time.


Robert Wolpert asked if the first wave the easy cases.  Art answered- "yes and no".  It's easier because Duke Primary Care is more standard already. However the actual discipline is more unpredictable than the specialties. Billy added that they expect the first wave to be harder because it was the first wave. Before the first wave we didn't have a system, now we do. So the challenges will be different.

John Board asked if we have a good system for failover.  Art answered that it's better than some, not as good as others. It is deployed under Citrix, so it's a virtual mode. Most of the failures we've had haven't had anything to do with the application itself. They've had to do with a dropped session, a system incompatibility or a device problem. Art feels comfortable with the survivability of the system . Billy added that we did actually fail over from the primary to the secondary and back within the data center.  It's not a hot swap by any means. We don't have a test-clinic environment.

Do you see an issue with the data being used by the clinical research side? We don't think so. Leading researchers are helping us with the build to help us identify what they need up front.

Art added a couple of anectdotes;

  • Day one, the first two appointments for go-live were two of Art's kids.
  • The third day the doc clinic was already personalizing the app and running circles around attendings.

Tracy added congratulations, for the highly successful first wave.


Project Updates: 0365, MOOC, Box, --- Lynne O'Brien, Mark McCahill (20 min)

Mark McCahill reported on Office 365. Office 365 is moving forward at a reasonable clip. We've got a test environment. Last Friday we got access to the magical code that allows us to do directory synchronization from two active directories which is key because we want to do both Duke Health and the University into one place. We also have started testing the migration of mailboxes. That testing is going reasonably well. It's looking like it will be feasible to move a bunch a people when we actually get around to doing that.

The holdups are that we're waiting for the production instance. We have a test instance, but not the production instance in hand. We're actually waiting for two production instances. One will be for Duke Health with the University and the other for Alumni. Microsoft has given us assurances that this will happen real soon.  

For migration testing we are doing trial migrations of real mailboxes to see how long it takes. The model will be to get the mailboxes staged up and then finalize the migration in a short period of time.  That way we can get the vast majority of the users mail moved and then do the last little bit when we do the switchover. We're not ready to commit to a time-line, but it looks like we could probably be doing some of the migration in the fall.

Charlie, - right now were optimistic that we can do a big bang transition of the Sunmail students. Mark added that the goal would be to do this while they are here, because then they can support each other. We want to get them done before they bail out for the semester.

Will mailbox quotas go up? - Yes 25 gig, currently Exchange quotas are 2 Gig.

Lynne O'Brien reported on MOOCs. (Massive Open Online Course)  Coursera Moocs were announced on July 17th. In less than a month we've had 130,000 people sign up for 8 courses. One is over 40,000, the smallest is over 5,000. That's electrophysiology... there are 5000 people signed up for electrophysiology. Lynne encouraged people to look at the landing page. Requirements for the astronomy include the ability to do sample math problems. Fourteen thousand people have signed up anyway.

Now we have to actually produce the courses. We have a mixture of things going on. For some we are having to really rethink the course. One of the especially challenging things is figuring out how to add interactive components. We currently have several staff members taking courses offered by other schools to get the student experience. Lynne is taking a class on world music. Some people complain about difficulty and some saying that this is supposed to be hard, if it's an Ivy League course. We have a variety of production methods planned. Some classes will be done in studio. Others are doing it with a kit on their own. The first class will start on Sept 24th

Robert asked if somebody is keeping track of resources and cost model.   Lynne said "yes", but I think we'll need the whole year to figure that out. We have a very lightweight weekly survey that we're having instructors and support staff fill out to get an idea of time spent on this project.

There was discussion about how discussion boards are used for quantitative courses that involve equations and the difficulty of blending this activity into the board capabilities.

One of the things that makes this interesting is that we're trying figure out if the number of participants makes a difference.  Is it better for students to work in small groups or large groups? The company founders think that you need 1,000 active participants for the discussion boards to work effectively. This means that you may need as many as 10,000 students enrolled to have 1,000 actively participating. It's going to be very interesting to think about what kind of data we might get out of such a system that will help us understand some of these kinds of questions. We have a faculty group that will be convening to think about assessment and data analysis. We want to think about what is it that we want to know, so that we can get the appropriate data.

Billy Willis asked if we are going to be doing surveys.  Lynne said "yes", surveys and data. One interesting things is that it makes you rethink about all the roles and what it means to be a student. Some don't care about certificates. They aren't taking quizzes, but they are participating.

How are essays graded?  They are peer graded. Each student is assigned to grade 5 others, then your own. You don't see your grade until you do your grading.

There was discussion about grading and about on-line cheating.

A question was raised about competitors to Coursera.  - right now it's the only one we furthest along with.  We've had a number of phone calls with Sebastion Thrun at Udacity .  We have another call lined up with him. We'd like experiment with that one as well to experiment with different platforms and approaches.   Another systems that has emerged is Ed X, which started out as MIT X.  Harvard joined them and then more recently Berkley. 

Tracy added that we want to be sure that when we produce materials that we can reuse them in any number of places.  

It was noted that some companies are trying to contact individual faculty directly.

Mark gave an update on personal storage. We are still waiting on BOX to do a BAA.  The BAA would allow Duke Health to join in. We would like to move from AFS for personal home directories to SIFs We're at the point where we can allow an opt-in option. An application called Portage will let you move your home directory over to SIFS. The reason you might do that is that you don't have to have an AFS client on your computer, you can use the built-in support. The performance is better than the AFS clients on Mac or Windows.  We're just at the point of being able to do a soft rollout. The documentation just got turned on at OIT website. We think we will have it up and running by about the 10th of September.  People who don't care about home directories can start now.  There are also about 250 people doing CGI scripts. We will talk with those people directly.

For Box, we're waiting on the migration to SIFs. The public labs will mount either.

Another topic is collaboration tools. There's been a lot of activity around Webex tools . We will be doing a Webex "meeting place" that people can opt into. The opt-in will happen through the site-license website that Evan has. Hopefully that can rollout sometime in October in a big way.  We'll probably pilot that with a small number of people first.  There is also Quad which is Cisco's collaboration product which has rebranded as Webex Social. We're looking at a rollout of Webex Social in October. There's a collaboration tool called Webex Social.

Robert asked if we going to sign a deal with Box? Tracy answered that we are working to get them to sign a BAA. We are told that Box has agreed. Once an agreement has been signed we will go ahead and rollout for non-hospital users while the BAA details are hammered out.

Charlie noted that Box will be the place to do file-level collaboration and SIFs will be the place for secure home directory. There may be migration to Box, because it may be easier to get to from home, etc.

Mark noted that we haven't put the link onto the OIT site, but users will soon be able to go to the link that lets you opt in to SIFS. The link is

Will the SIFS system act as a web-page server? - Not at this point. It's coming. It works in tests, but needs a little more testing before we move into production.

For CGI users we're looking at moving them to a separate CGI environment that will be just straight CGI.


SABA Learning management System, --- Kim Andrews, Brian Aucoin (30 min)

Kim Andrews works in Learning and Organizational Development. Brian Aucoin works at DHTS in the health system side. They are both working together to implement SABA.

We anticipate questions about the difference between Sakai and SABA. Our focus is to manage employee performance and improvement as opposed to delivering courses. Sakia can be open to the public . SABA is employees only.  SABA is basically a corporate LMS.

We're dealing with competencies and performance against competencies. One of the biggies for employees is tracking. SABA will track certification and will provide automatic notification to managers.  It can also track CEU credits because we have some employees who are required to maintain credits for certification or licensure. We have some federal requirements for training such as Title 9 or harassment training. This will provide auditable tracking for such training.  The focus is really from an employee perspective.

For SABA content isn't so much the issue. There's not much of a difference regarding the size or delivery of the content.   You can create content in any of a number of formats; Captivate, movies, pdfs Word Documents. All can be done in both.  The access to create course in Sakai is fairly open. In Saba you have to be an LMS administrator to create a course.  It's a more tightly controlled environment. Both Sakai and SABA can handle rich media content.

Sakia has collaboration tools. Saba does not. It does have the ability to hook into something like Webex but it doesn't have built-in collaboration tools. We can buy the social platform tools, but we didn't purchase that module.

We can do rich assessment in Sakai and in SABA. You can also use other authoring tools like Captivate to create content. Of course there's reporting based on assessment information.

Handicap accessibility is similar for both.

The health system has been using SABA for a while. LMS is coming to the University. The one thing that has been holding it back is authentication. They are working on Shibboleth integration so that it can be used on the University side. This is expected in September.

Brian said that the Health System came on line June 1, 2011. Right now we have 550 on-line offerings in the system.  We have 1600 classroom offerings.  We've been using SABA to manage the Maestro Care training. We've been training the wave one and wave two people via the application. In the past year there have been 38,000 registrations and completions in the system. It will be used by more than 25,000 staff, just for Maestro Care.

The benefits for Duke as an employer are that we can manage some required training. We can automatically assign training to certain groups of people based on their job or their HR data.    Managers can go in and view and assign and manage learning plans for their employees. They can create training plans for certain employees. They can get automatic email notifications for required training when it comes due.

There is reporting for managers and employees. They can view records to see what has been completed and what hasn't.

As a community we can take advantage of our shared resources.

Billy asked if we are still pursuing the SAP portal integration. Brian said "yes" we are testing the Shibboleth integration. The idea is that we would go to "Work at Duke".

Robert asked who will see this.  Kim said that staff and faculty, but that to begin with only for required training.

There was additional discussion about how it might be used going forward.

Susan Gerbeth-Jones asked if it can plug into  Kim said that it's not necessarily plugged in, but info from can be pulled in to SABA.

Julian explained that is a comprehensive set of on-line training materials in the IT training space. Any application has a set of training pieces.  Duke has a site license for does internal tracking for completion of modules. This tracking information can be imported to the SABA space.


GigU Update, --- Kevin Davis (15 min)

Kevin gave an update on the Gig.U project.

Gig.U is a consortium of about 3 dozen American public and private universities that came together last year to try to advance the cause of increasing broadband in our communities around the campuses. There has been national leadership from the past FCC Chief of Staff who was the architect of the national broadband plan at the beginning of the of Obama administration. This consortium went out last fall and issued an RFI to vendors in the industry and vendors who might be interested in the industry. Some are existing vendors, some wanting to be at the table and some are dis-intermediated vendors who see a business interest behind broadband.

There are two premises behind the reason for GigU. One is that we just don't have a market that is competitive to bring the kinds of speeds in Europe and countries like South Korea and Japan to the United States.  Given the current nature of the regulatory and market environment it just doesn't make sense for vendor to come in and run their own backbone into a city and then run the middle mile fiber throughout the city and then run the last mile fiber to all the houses. It's hard to make a case for that in a broad sense.  We think that university communities like Durham, Chapel Hill and Raleigh have areas that would demand these speeds because of the presence of the universities and research and teaching. If you can find the right areas and change the math to incent the delivery of these types of networks, you can start delivering these sorts of speeds.

In the case of the Triangle, UNC, Duke and NC State are all engaged and the goal is to bring these kinds of speeds to the Triangle region.  The universities have assets and investments that we've made on campus.  There are state activities like MCNC. These are the types of pieces that can help to lower the math.

We've been much more engaged with communities. We've been involved with the Research Triangle Foundation and various Chambers of Commerce. There are a variety of questions that are on the table.

  • Who are the users?
  • What are the use cases?
  • What are the initial demands?
  • Where are assets in place already that can be leveraged?

 Chapel Hill did already plan to directly sell this service and were stymied by legislative action.

We've had folks like city and county managers engaged in Durham.   We've done a variety of outreach activities.  Over the last 3-4 months as we've been developing an RFP document. The RFP is intended as a regional document similar to other efforts like transit, road planning, water and waste water.  The intent of the RFP is intended to allow a range of carriers.  The RFP is structured with community oriented parts.  You would want to have community peering between various groups, even if different vendors come in.

Use cases vary by community. Healthcare is one of the questions.

We are looking at what local businesses are saying they need. Work from home is another component. This could cut down office space costs and commuting costs by using remote employees. We see these as some of the initial drivers, but no use cases actually justify this project completely. Because of this we need to look at the most opportunistic areas. Examples might be the Erwin Road corridor and areas where there is a high concentration of students.  Another might be downtown or RTP where key major employers might be interested.  We're trying to pull together an RFP document from a political and economic development perspective that ties to city and county planning development. We're trying to pull together all this information to create a well-supported RFP document for city council review that can be issued to hopefully get respondents in the door.

Other communities are seeing some progress that is either directly GigU linked, or related. Oldtown, Maine and Gainesville, Florida have announced small scale fiber rollouts that are linked to GigU and the areas around their campus.  UF is doing an "innovation square" project in an area adjacent to their campus. In Seattle, throughout the GigU initiative there has been a lot of talk about the dark fiber that they have access to that was built out for the municipality. They are looking to lease and make available to private builders who would run a network over that kind of asset.

Tracy brought up the Kansas City project with Google Fiber. There is a lot interest in what is happening in KC. We are interested in getting feedback on what they doing. National excitement has been generated in many communities by Google Fiber. It's one of the factors in why the communities are coming to the table and saying,  "How do we stay involved?"  We've all been watching Kansas City very closely. The model announced is 70$mo for 1G synchronous connectivity. Here, $70/month is around 30 down 5 up.

In Kansas City, Time Warner has been trying to sign people up to long term deals. Google has countered by offering that for a one-time fee of $300 to pay for the connection you get 5M free for seven years versus seven years at $30/40 month.  This creates the incentive to build out lots of houses.

Google strategy has also included the idea of the "fiberhood".  They will bring service to a neighborhood if enough people in the neighborhood sign up. Rates vary by density. They promise to connect schools, hospitals etc. within a fiberhood for free.

 It is also interesting that Google is building out their own infrastructure. They are clearly pushing a lot of their platforms into the systems.  

Kevin showed a slide with a map that illustrates hot areas and corridors in Chapel Hill. We are also working through the Durham Chamber of Commerce to meet with members and using surveys with to ask what kind of speeds would be needed and where.  We are also gauging interest in signing up. We're also working with folks in Student Affairs to understand where Duke in bringing on-line facilities and where are the hot corridors to hit. Erwin Road and 9th Street are areas that make sense for students. They are also opportunistically places where we have fiber assets that may be able to be contributed or shared in some way.

Chapel Hill's case has been that businesses don't necessarily see value. Another strategy is to bridge out so that in the future you are passing through dense areas that might join in.

We don't know how many communities will end up with these systems. But, it's the best opportunity we have to get a project like this in front of local and national stakeholders and start making progress, even if it's only in a small way.

Billy Willis asked if it is legal in NC? Tracy said "yes" according to the opinion of several legal counsels. Kevin added that the main issue is municipal assets that might be leveraged.  Their opinion is that they don't see a problem with it.

Robert asked if anyone on the team looking at the dynamics in other areas like Australia, Korea, Japan. Kevin said that the model in many cases has been used very sparsely in the US. The idea is an open access network where the municipality builds out the fiber to everyone's home and other companies sell the logical services over it. In Sweden it's a very popular model. It means you have a choice of 30 broadband providers all over one infrastructure connection coming into your house as well as possibly smart-grid and other applications.

Robert asked if there was any Chamber interest as a tool to attract high wage, low pollution industry. Kevin said that the Durham Chamber has been more engaged than any of the Chambers.

Billy asked if the primary target the home or the business.  Kevin cited examples of both.

Robert asked what date is action likely for this effort. Kevin said that there will most likely be public hearing for RFP approval to go out to the various municipalities and then we'll see when we hear back from vendors.

The meeting adjourned at 5:30