The Value of Good Data in Healthcare Delivery

by admin on February 23, 2009

Written by Ellen and the rest of the MIT Sloan Baobab Team

Our G Lab project revolved around looking at the benefits of implementing an electronic data system in healthcare clinics in Malawi.  Our partner organization, Baobab Health Partnership, developed a touchscreen appliance that can be used both for patient registration and as a clinical support tool for healthcare workers.  Baobab is based in Lilongwe, Malawi, and at the moment, the software is specifically tailored for anti-retroviral therapy (ART) clinics in the country. One key benefit of the system is improving the quality of the data collected at clinics (there are many others, but they will not be discussed here).  Right now, most ART clinics use a paper-based system to collect and store patient information.  This raises a host of issues relating to the quality of data kept by the clinics.  We heard stories of incomplete or lost records and witnessed first-hand how human error can unintentionally produce inaccurate data when manual systems are used.  As clinics expand their patient base, these issues will become increasingly problematic.  By having an electronic system at the point-of-care, so the argument goes, you can reduce data errors or incompleteness, and thus improve the quality of the data gathered.  Why does this matter?

Each quarter, the Malawi Ministry of Health (MOH) visits every ART clinic to collect the data in order to aggregate it across the entire country.  This data is used to identify needs, observe trends, direct resources and set national policy regarding HIV/AIDS.  Clinics are asked to compile data themselves in advance of the visit.  In the quarterly visits we attended, we joined a team of MOH staff who pour over the registration books and tally up the numbers using a paper and pencil.  The staff average roughly 4 hours at a clinic with approximately 1500 patients.  If you can imagine a clinic with over 10,000 patients, the task seems quite daunting and much more time intensive.  It is easy to see how data irregularities can occur.  We were told by an MOH official that one 2006 audit (we were unable to locate a copy of it) showed a 10% difference between the numbers found by the MOH and those reported by the clinics.  In other words, there was a 10% difference in the numbers compiled by the clinic staff in preparation for supervision visits, and the numbers compiled by the supervision team.  In a country like Malawi that purchases $15M of ARVs per year, 10% amounts to a discrepancy of $1.5M.

After working together as a team in Malawi for three weeks, I left the team and returned to Kenya where I had the privilege to meet staff members from the Partnership for Supply Chain Management, Inc, a team of sixteen private, non-profit and faith-based organizations who implement the PEPFAR-funded USAID Supply Chain Management System (SCMS) program.  The staff members I met with were primarily concerned with supplying laboratory equipment and supplies.  They faced huge challenges in implementing their program in a cost-effective manner because they lacked quality data from laboratories.  None of the laboratory supplies they purchased were available in-country.  With long and highly variable lead-times, it is difficult and extremely costly to make last-minute purchases.  It is also a challenge to ensure quality supplies are purchased particularly where a cold chain is required.  Timely and accurate data is critical to ensuring Kenya’s labs are fully stocked with a high quality product.  SCMS staffers really liked the idea of having an electronic system that made it easy to track usage of laboratory supplies without over-burdening already overworked lab techs.  They were particularly in one that can be used at the point-of-care rather than for retroactive data entry.  Dare I say they could even have more accurate forecasts?

Incidentally, in the logistics realm, the USAID DELIVER project has created ton of training materials, case studies, guidelines and tool kits to set-up and manage a logistics information system.  The DELIVER website is packed with things to download.  I actually came across quite a few of them posted in healthcare clinics and pharmacies in Malawi and Kenya.  Although Baobab is not an inventory management tool, here is another paper-based solution that faces the same problem of data quality.

Returning to the original question, we should ask ourselves is a paper-based system sufficient?  For managing inventory?  What about managing patients?

While we at the MIT love technology solutions, implementing an electronic data system comes at a price.  In the case of Baobab Health, the Malawi MOH is looking to roll out an electronic data system to the largest sites throughout the country over the next five years.  The cost of implementation, including maintenance and ongoing training is in the millions.  Despite the costs, there are lots of beneficial ways the data collected by the system can be used on an aggregate level as well as on the individual patient level which we wrote about for our project.  The challenge our team faced is that while we know clinics may not maintain inventories or deliver healthcare services in the most efficient manner, we have no way of quantifying exactly how much money could be saved or how service levels could improve with better quality data.  We do know that as more patients move from first line to second line ARV regimens, procurement and inventory management of ARVs will become increasingly complex and reliant on good data as the number of different drugs needed in substantial quantities proliferates.  It is already incredibly complex in laboratory management, as I saw in Kenya.  The Baobab system has a lot of potential to help with these issues by improving data quality, and is an exciting piece of technology to look at.

It was great to have an opportunity to take what I learned with the MIT Sloan team in Malawi and look at how healthcare is delivered in Kenya.  Maybe some readers out there have experience with data quality in other parts of the world?  What would be the impact of good quality data on your program?  How much would you be willing to pay for it? Take a look at Baobab’s website, think about your own data needs and tells us what you think!

{ 2 comments… read them below or add one }

Joaquin Mendoza March 2, 2009 at 9:23 pm

After reading this blog entry regarding the team´s GLAB experience and their emphasis on the importance of data, I could not help but think about my team´s experience in Nairobi Kenya working with a local HMO/healthcare provider. Our host was a holdings company with three companies underneath 1)Insurance Sales 2) Management 3)HealthCare. Our company had operations in 5 countries in West Africa, delivered health care insurance to middle – top income population and also operated several clinics.
Our project involved documenting key processes within the company. The need for it came from unexpected growth in business which resulted in new processes, tasks and activities where only few people or areas new about. As a result, information flow was slow or sometimes inexistent. Furthermore, pointing out to the previous blog, sharing of critical data – important for any HMO and healthcare provider – was not efficient.
As a result of this inefficiency that company had invested heavily in IT as a way to “link” the three companies under the holding company. However, the current database and IT group had not been successful in created one, all encompassing software but many individual ones that were created as patches to solve issues that kept coming up with the original software. This is similar to Baboab Health´s MIT group who noticed that implementation of state-of-the-art and technological solutions come not only at a fixed price (implementation, maintenance, labor, etc…) but unexpected costs (impatience, frustration, confusion, etc…) Furthermore, as was the case with our host, people tended to create their own “patches” or “do-it-yourself” solutions and the result of it was an increase in manual labor and more silos since people not involved with the process had no idea what this person was doing or how. The lack of information sharing also caused another issue which was that people starting becoming critical and key since they were gatekeepers of information or one of the few who understood the linkage between processes and how information affected each one.
Our team deviated from the famous MIT “high tech” solution but rather noticed that what the company really needed was a holistic view of its key processes as well a map of how information flowed – all of this done in powerpoint. Once this layout is in place then can a company sit down and determine where information and data should flow. Our team not only mapped processes within each of the three companies under the holding company but also across each one. The team also worked with our host to determine what key metrics (data) would be helpful in order to measure productivity, success, workload, pay, etc.
Overall, the lesson learned was that sometimes high tech solutions cause more problems.

Heather Pichette March 10, 2009 at 5:37 pm

After reading these entries, it made me think how much more efficient the healthcare system could be if there was more IT infrastructure. Our team was working with CIDRZ, the Center for Infectious Disease Research in Zambia, and more specifically we were working at the lab. I was extremely impressed with CIDRZ and how well-managed they were. They processed lab samples from local health clinics and for research studies efficiently and for the most part, quickly. Henry, the business manager, helped to implement some lean sigma ideas to improve the processes when he arrived at the lab. However, sending the lab results back to the health clinics seemed like a laborious task. If only the results could be sent electronically, the physicians would be able to treat the patients much more quickly with the right dose of ARVs and the patients would not have to wait many days for their lab results. Of course, one of the inherent flaws is the lack of internet in much of the country (or really slow internet). But some of the private health clinics definitely talked about getting internet connections soon, which would make this all possible.

Imagine if they could build much of healthcare system from the ground up, unlike the US which is having to “retrofit” IT onto an existing system, and anyone who has worked in the US understands the arduous task of converting from paper to electronic copies. In Zambia, many of the public health clinics that we visited had stacks of patient charts. Oftentimes the charts were overflowing and the records room was completely overflowing. One of the challenges, they said, is the duplication of patient records. A patient may visit one clinic and then choose to start going to another one (patients oftentimes visit one that is further from their home to preserve their anonymity, there is such a stigma associated with being HIV positive). Therefore, it is hard to maintain accurate and complete health records. If only the clinics could be connected to each other, imagine how much more efficient the processes would be! I was reading about a clinic in Malawi that actually texts lab results to patients. The patients can then show the text to the clinician, who can treat them on the spot. This of course eliminates the problem of having the patient return to a clinic which may be 15 km from their house, only to find out that the lab results are not yet ready and they need to return once again. For many of these patients, the cost of traveling to the site is equivalent to eating that day, and it is a difficult choice to make.

Although the IT question was not one that we addressed while working at CIDRZ, I think that it’s one where they could have a large impact on the way that healthcare is delivered and maintained. As CIDRZ scales very quickly, it is important that they are able to continue to provide the best services to their clients, and maybe this would make an interesting project for a future G-lab team?

Leave a Comment

Previous post:

Next post: