Rwanda Upgraded to OpenMRS 1.4

I want to congratulate Chase and Maros for getting Rwinkwavu and Kirehe’s OpenMRS servers upgraded to version 1.4 with sync. Jeremy and Evan, you two owe them beers for having upgraded first and finding and fixing a whole slew of bugs in 1.4 that you would otherwise have had the pleasure of experiencing firsthand.

I’d also like to share this conversation for those who couldn’t be here:

(Scene 1: Wednesday night, 10pm)

Maros: So, Darius, unless you tell us not to, we’re going to start the upgrade now.
Darius: Hmm…okay, go ahead.
Simon (BCG): WTF? You’re going to do a major upgrade on a Wednesday night?
Chase: Yes.

(Maros and Chase leave.)

Simon: You’re really okay with this?
Darius: Yes, I have complete confidence.

(Scene 2: Thursday morning, 8am. Chase and Maros are sitting around their laptops. They haven’t slept. Darius and Simon arrive.)

Simon: So?
Chase: Rwinkwavu is up and running.
Darius: Whew.

Not that I doubted for a moment.

-Darius

PS- Anyone reading this from the broader OpenMRS community, we’re going to release 1.4 as alpha now that this first round of bugfixing has happened.

OpenMRS Implementers Meeting

It was a great honor to attend the OpenMRS Implementers Meeting – 17-20 June 2008 at the Elangeni Hotel in Durban, South Africa.  It was my first OpenMRS meeting and also my first trip to Africa.

Monday was spent at an OpenRosa meeting which was moderated by our old friend and colleague, Neal Lesh.  This group has made great progress.  Here’s a description of OpenRosa:

OpenROSA is a consortium formed to create open source,
standards-based tools for mobile data collection, aggregation,
analysis, and reporting. By developing open source solutions and
conforming to standards based on the XForms specification, our
different projects can easily share code, data, ideas and
infrastructure.

The rest of the week was filled with OpenMRS.

The most interesting presentations were ones from the implementers — the ones from our own PIH team Jeremy Keeton for Lesotho, Christian Allen about synchronization with great graphics about doctors traveling by camel and many funny subliminal messages), Baobab (Mike McKay and Jeff Rafter), Millennium Village Project (Andy Kanter) and Albert Schweitzer Hospital in Haiti (James Arbaugh).  The many different uses of OpenMRS is fascinating and creative.  Albert Schweitzer has over 700K patients in OpenMRS!  MVP uses OpenMRS as a primary care system.

Two other presentations were valuable.  Ben Wolfe from Regenstrief Institute gave a demo of building odules. This was well timed since I was building my first module.  Too bad this wasn’t more of a hands-on lab class, but it was still very useful.  A panel discussion about the current concept dictionary, OpenMRS Concept Co-op (OCC), Terminology Service Bureau (TSB) and OpenEHR archetypes was a useful finale to a wonderful week.  This area needs attention, but hopefully additional resources will focus on this.  I’d like to do more work in both these areas.

It was great to meet so many active participants in the OpenMRS community and a once-in-a-blue opportunity to get the Partners In Health OpenMRS team together.

Reporting Framework at Durban

The Durban conference presented the community with a great opportunity to discuss many newly developed features soon to be released in OpenMRS version 1.3.  One of the most significant of these is the new reporting framework.

Although general in nature, the primary goal of the reporting framework is to ease development of aggregate indicator-based reports.  Given that many organizations share the same reporting requirements for agencies like Pepfar and the Global Fund, our goal is to foster collaboration and knowledge, thus reducing the work involved for all.

As developers involved with producing this framework, we entered the conference aiming to demonstrate the capabilities, to gather feedback, particularly on usability, and to rework the functionality as needed.  This proved to be a very valuable experience.  Initial feedback, early in the week, was that the interface was far too unwieldy for a non-technical user of the system.  This resulted in the first of our redevelopment efforts, a completely rebuilt report design interface, in which the user follows a configuration wizard to complete the design process.  As this was iteratively developed on site, by the end of the week we were able to demonstrate these revisions to the wider audience.  We received a very positive response and a number of useful suggestions for continued improvement.

One of the more interesting suggestions came from Christian, never shy with his opinions on user experience.  His contention is that the report design process can be driven wholly from an “example report”, expanding on the ideas utilized in the ReportTemplate module.  The basis of this design would involve starting with a template file in a format that most users are very familiar with – Excel, Word, Text, or similar, and initiating the design process from there, with wizards to help the rest of the way.

It is clear to me following the meetings in Durban that user-experience will continue to play a larger and larger role in the success of OpenMRS and, as with the reporting framework, iterative design and continued community involvement will be key to this success.

-Mike Seaton

Data Management in Durban

PIH’s team of OpenMRS implementers and data mangers based in Rwanda, Lesotho and Malawi conducted a data management workshop at the HISA 2008 Conference in Durban.  The purpose of the workshop was to identify common data management problems and brainstorm as a group to come up with good solutions.  In addition to the PIH team, current and prospective OpenMRS implementers based in Kenya, Uganda, Tanzania and South Africa were present at the workshop.

The problems discussed ranged from challenges with the initial implementation through to common issues faced by an established project with a large patient population.  Specific examples that attendees gave included:

1. Getting started with OpenMRS at an existing HIV clinic with approximately 6,000 historical patients, but very poor paper and electronic record keeping practices
2. Internet and/or power unavailability presenting challenges to a typical implementation
3. Proliferation of transcription errors leading to poor data quality
4. Lack of motivation from clinicians to use EMR forms leading to incomplete record sets
5. Incorrect patient identifiers resulting in lab results being assigned to incorrect patients

Most of the data management problems that were discussed were common experiences for the implementers in attendance, and a number of interesting solutions were shared during the second half of the workshop.  These included:

1. Focusing on getting a particular record set right at the beginning of an implementation over trying to get all of the data into the system at once.  In the case of the implmenter with 6,000 historical patients, this could involve building a recordset for patients currently enrolled at the clinic, and leaving out patients that were already deceased or transferred out.
2. Being flexible with power and internet requirements.  Prior to getting internet at its clinics in Lesotho, PIH developed a system for offline data entry using Excel, which were uploaded to a central server in Maseru.
3. Empowering the data entry team to notify clinicians of transcription errors on forms, and motivating the team to hold itself accountable.  The OpenMRS implementers in Eldorat have developed data accuracy targets and data clerks are rewarded for meeting them.
4. Making it worthwhile for clinicians to use forms by (a) making the forms themselves improve workflow and (b) focusing on developing useful reports.  Allowing the data entry team to be part of the process by delivering the reports to the clinicians can help build good relationships between the two groups.
5. Building an electronic system for issuing unique identifiers to patients.  In Malawi, the team from Baobab uses label printers and barcode scanners to issue ids and access patient records.

One of the overriding themes that came out of the workshop is how integral data management is with clinical practice as a whole.  From the moment a patient arrives at a hospital through to the point at which the patient’s record is entered into an EMR, there are countless opportunities for encountering data errors.  Each of these problem points poses an opportunity however, as good data management practices can greatly improve the clinical workflow as a whole.

Another theme from the discussion is that data management should not be an end point, but rather a part of a closed loop between the developers of an EMR and the clinical setting.  Some of the best solutions that were discussed in the workshop involved both a proactive data manager and an attentive development team that was able to help come up with a solution to the problem.

A Check Digit for Polyphemus

“‘Cyclops, you asked my noble name, and I will tell it; but do you give the stranger’s gift, just as you promised. My name is Nobody. Nobody I am called by mother, father, and by all my comrades.’

“So I spoke, and from a ruthless heart he straightway answered: ‘Nobody I eat up last, after his comrades; all the rest first; and that shall be the stranger’s gift for you.’

–Odyssey, Book IX

Ever since the age of myth, the human race has been engaged in an epic struggle against mistaken identity. Oedipus unknowingly kills his father and marries his mother, because he does not know their true identity. The unlucky neighbors of Baucis and Philemon in Ovid’s tragic tale could have been spared if only they had a better way to accurately identify guests. Belle was lucky enough that the Beast’s true identity was a handsome prince, but most of us would prefer to know in advance. And of course, “Nobody” knows better than the now-blind cyclops the importance of accurately identifying visitors.

In the medical world, mistaking a patient’s identity can also have severe consequences. Imagine the following scenario. Annie Nonymous goes to the clinic for a scheduled visit and has blood work done. Annie Nonymous has a system ID of 12543912. The lab technician, who enters hundreds of lab tests into his computer every day, mistakenly types in Annie Nonymous’s system ID as 12543412 (which is actually the ID for Anne Nonymous). The next day, the doctor sees Annie’s CD4 count in Anne’s file and mistakenly starts Anne on ART instead of Annie! Of course, this is a simplified scenario, and in reality there are more opportunities to catch mistakes before they result in a change in drug regimen, but medical mistakes based on mistaken identity are indeed a serious problem. In fact, according to the Institute of Medicine, in the United States alone medical errors lead to around 100,000 patient deaths per year [1].

At Inshuti mu Buzima, we are taking a big step forward to prevent mistaken patient identity. From the beginning, we have been issuing patient identifiers that are verified with a check digit, generated according to Verhoeff’s Dihedral Check Digit Algorithm [2]. A check digit is a simply a letter appended to the end of an identifier that is calculated from the rest of the identifier string. You can play with generating check digits here: http://rwanda.pih-emr.org/verhoeff.html.

Although we have been issuing identifiers with check digits from the beginning, until now we had not actually been checking that check digit anywhere in the EMR. OpenMRS used to have the Luhn Check Digit Algorithm hard-coded into place, but we have recently rewritten the OpenMRS core to allow the addition of any check digit algorithm of the one’s choosing. We prefer the Verhoeff algorithm over Luhn because the Verhoeff algorithm guarantees to catch all single digit replacements and transpositions of adjacent digits. For example XXXaXXX will never have the same check digit as XXXbXXX, and XXXabXX will never have the same check digit as XXXbaXX. Luhn can also detect any single digit replacements, but it does not catch all transpositions of adjacent digits.

Now that we have the tools to detect mistaken identifiers, Cheryl, our data manager, is feverishly working to correct identifiers in the system we know are invalid, with the peace of mind to know that more will not be created. Although check digits are not able to prevent all cases of mistaken identity, we’re confident that together with the other systems of checks and balances we have in place, mankind’s ancient struggle against mistaken identity may be drawing to a close.

[1] To Err is Human:Building a Safer Health System. Institute of Medicine. Washington, DC: 1999.

[2] Wagner, Neal. “Verhoeff’s Decimal Error Detection”. The Laws of Cryptography with Java Code. p 54. San Antonio, TX: 2003. http://www.cs.utsa.edu/~wagner/lawsbookcolor/laws.pdf

OpenMRS in the news!

Coming off the heels of a successful conference in Durban, Paul Biondich from the Regenstrief Institute was interviewed by the BBC about OpenMRS. A podcast of the interview is available online, as is the article resulting from the interview.

Keep an eye out for upcoming blog entries about the OpenMRS Conference and the latest developments from PIH project sites!

Collaboration with OpenELIS, an open source lab information system

In the last few weeks the OpenMRS team along with collaborators from the University of Washington have been meeting with members of the OpenELIS team, an open source laboratory information management system (LIMS) that’s been implemented in Minnesota, Kansas, Missouri, and Vietnam. The members of this team are from the Association of Public Health Laboratories (APHL) and from labs in the US. Conversations have focused on how the OpenELIS group can learn from the community building experience of OpenMRS and how the two systems can connect to provide countries with the possibility of implementing both systems.

Our first meeting last month was extremely productive. Dr. Gail Cassell was also present. We’re now coordinating meetings to meet the two goals mentioned above: helping in creating an OpenELIS developer community and connecting both systems.

Getting Connected

The last few weeks we have been working to keep the momentum from Hamish’s visit to Malawi in early February going by laying the groundwork for a rollout of the touchscreen-based Patient Registration / Patient Diagnosis Module.

Work had already begun in late 2007 as I travelled up to Lilongwe to work with Mike McKay and the team at Baobab to build the Patient Registration component of the system. We took the basic structure of Baobab’s existing registration module, which was integrated into their ART system and based on Version 1.0 of the OpenMRS datamodel, and rebuilt it as a standalone module writing directly to OpenMRS Version 1.1. We were able to make this application work side-by-side with our current implementation of the PIH EMR in Malawi. Since then we have been using it for creating the patients in our backlog of ART data, taking advantage of the unique National ID number that the application generates. Baobab has already issued id numbers for more than 500,000 patients in Malawi, or roughly 4% of the country’s 13 million people.

Going forward our intention is to fix a few remaining bugs in Patient Registration so that it is ready for real-time use at Neno Rural Hospital, and add the Patient Diagnosis component to capture the primary and secondary diagnosis, treatment prescribed and clinician comments. In the existing workflow at the hospital, clinicians record the diagnosis in the patient’s medical passport, a data clerk transcribes the information into the ‘Outpatient Register’, one of the many medical registries issued by the Ministry of Health, and the patient proceeds to the Pharmacy. To create a smooth transition, we intend to add the Patient Registration / Patient Diagnosis module as an intermediate step between the clinical visit and the paper registry. Once the hospital staff grows accustomed to the new electronic systems, we hope to modify the workflow slightly so that clinicians can record the diagnosis information directly during the consultation.

Our next steps include:

1.) Demonstration of the existing Patient Registration module to Ministry of Health staff working at Neno Rural Hospital

2.) Generating a list of diagnoses and prescriptions common to Neno

2.) Working with Baobab Developers to write the remaining code for the Patient Diagnosis module

3.) Interior and exterior renovations at the hospital to create the Patient Registration area

4.) Creating a wireless network link between the hospital, our offices at the Neno District Assembly, other offices, and our warehouse in Neno

The final details of our plans for Patient Registration / Patient Diagnosis are falling into place, and we are excited about the prospect of having the system up and running at Neno Rural Hospital sometime in the next few months. In the meantime we are also busy entering the existing data for ART patients at Neno Rural Hospital, and planning for a more robust integration of Baobab’s Ruby on Rails interface within OpenMRS.

 

Patients queue to have their diagnoses recorded in the Outpatient Register at Neno Rural Hospital

A room adjacent to the existing registration area will be converted to allow data clerks to register patients and record diagnoses using the new Patient Registration / Patient Diagnoses module.

 

A pharmacy technician fills prescriptions at Neno Rural Hospital. The Patient Diagnoses module will allow clinicians to print labels for prescriptions during the consultation, eventually this could be linked to a wharehouse/pharmacy stock management system.

Patients queue at the Outpatient Register; construction of the new Neno District Hospital is visible in the background. Neno District Hospital is scheduled to open midyear, dramatically improving access to inpatient care for residents of Neno District.

How do you scale up (even a small system)?

Over the past 3 years, we have implemented a tuberculosis (TB) laboratory information system to connect the laboratories and health centers in the Peruvian public health system. This laboratory information system, named “e-Chasqui” after the ancient Inca messengers, permits entry of all TB results at regional and national laboratories and immediate viewing of those results by clinical staff. In addition, the system includes applications to assess quality control, generate aggregate reports, notify health centers of new results or contaminated samples, and track enrolled patients and the status of pending laboratory tests. In 2005, we started the pilot with two laboratories and 12 health centers. It took about a year of working with the labs and the health centers to both get the system exactly right and for them to realize the benefits of it. I say that because since 2006 the use of the system by the health centers has exploded, increasing over 3 fold over the past year and a half. At the request of public health officials we have now expanded the use of this system to a total of four laboratories and 42 health centers that cover over three million people in Lima. But there are still about another 70 health centers that we’ve been asked to train in the next 3 months in one of the health districts and another 60 or so in the second health district. So how am I and the e-Chasqui data administrator going to be able to do all of the activities required for the randomized controlled trial we are performing, continue to support the current health institutions and the overarching study that we are a part of, and at the same time triple the number of users of the system? Well, we’ll know the answer in a few months…

Power to the Clinics

It’s something most people take for granted when they talk of implementing an EMR system anywhere, even in a developing country. We know to ask questions about Internet bandwidth, or how frequent blackouts might be, are there staff members with enough computer skills to operate the systems, etc. But in general, we take it for granted that there will be some kind of power. The rural mountains of Lesotho, however, don’t afford us the luxury of making this assumption. The tiny Sesotho villages operate much as they did when this mountain kingdom was formed in the 1800’s. Thatch-roofed rondavels dot the mountainsides with rows of maize filling in every spot of arable land and sheep grazing on the land above that. Fire and the traditional Basotho Blanket are the only measures for keeping warm in the harsh winters and transportation is primarily on horseback, although the occasional truck does wind its way through the mountains to provide stocks of lamp kerosene, Coca-Cola and Hansa beer to the rural shops. While the power grid has made its way to the primary townships in the mountains, the “last-mile” still remains far out of reach for all of the rural villages where PIH operates. The lack of a market in the impoverished villages have made the likelihood of the grid extending into these areas any time soon low at best. Generators, with relatively low initial costs seem like a good alternative, but factor in rising fuel costs and the sheer difficulty of hauling liters of petrol up to the sites and this option too becomes infeasible. The villages do enjoy plenty of year-round sunshine however, so when the Solar Electric Light Fund (SELF) was approached by PIH to bring solar power to the clinics it seemed like a perfect solution.

After months of planning, procuring and the shipping of two containers from the other side of the world had been completed, we were ready to begin the installation of the equipment at our sites. Our team was comprised of our leader Walt Ratterman, a solar expert from SELF, three local electricians and a couple of IT guys. The first task was to unpack and sort the shipping containers full of equipment into piles for each of the four sites. From there we loaded the equipment onto 4 ton trucks for the slow and arduous journeys to the sites. The equipment included solar panels, 140lb. batteries, racks, mounting poles and a whole host of electrical equipment. Two of the trucks completed their journeys without incident, while the other two got stuck at river crossings that required their offloading and reloading onto boats or smaller 4×4 pickups for the rest of the journey. With the equipment onsite and the team in place we began assembling the various components that after a few days of work, resulted in the lights being flipped on for the first time in these villages. The idea of these rural villages skipping entire generations of electrical production and jumping right to an eco-friendly and sustainable solution is heartening indeed. Patients can now be attended to at night, doctors have the facility for crucial laboratory equipment and our laptops running OpenMRS now have a place to plug into. Soon Internet will follow and these clinics will take a leap forward into the 21st century with all the advantages that modern communication, telemedicine and electronic record keeping have to offer.

-Jeremy