How to do data integration: The process from City of Kansas City, Health Department

Pictured from left to right, Frank Thompson, Jamie Matney, Kathryn Resch. This is the KCHD team responsible for integrating data from housing case management and medical case management for the Ryan White Transitional Grant Area.

Providers need all of the available data to best serve their clients. However, some client information is not available when it could be useful to a provider, even though the information has been collected and stored in an electronic system. In addition, similar data may be defined differently by different agencies and in different data systems. Understanding and connecting these different data systems in order to have the most accurate and comprehensive picture of each patient is important, but it can be daunting.

In a previous blog post, we outlined the importance of data integration to better serve people living with HIV. The following is a brief accumulation of knowledge and experience the City of Kansas City, Missouri Health Department (KCDH) has gained during the process of integrating data from our housing case management system to our medical case management system. This is a high-level, step by-step guide to data integration by non-computer engineers. Some of the following steps we did. Some steps we learned that we should have done earlier.

Step 1: Build Consensus.
There are several arenas which first have to be discussed and resolved before data are integrated. Issues like privacy and confidentiality, data sharing, and buy-in from leadership are typical barriers that are not easily solved. Providers are reluctant to give access to personal health information and clients, especially those who have stigmatized conditions, are reluctant to have their data shared across providers. These concerns are warranted and should be addressed. These are not easy conversations to have but they must be had before moving forward.

Step 2: Decide on what data needs to be integrated.
Data is a big term. Does “data” in your context mean all of the available data on a patient or does it only mean a few specific elements? Deciding on what specific data elements are shared is another conversation between providers and clients. It may not be necessary to integrate all of the available data on a client if there are concerns with privacy and confidentiality. Deciding in advance how you want to use the integrated data will help you prioritize what data elements to select for integration. Integrating as little as two-to-five data elements can translate into big wins. Don’t feel you have to choose several data elements for integration for your project to be a success.

Step 3: Decide on what frequency data will be integrated.
Data will need to be integrated continuously, but what does this mean practically? Will the two systems always be in sync, or will files be sent daily, quarterly, or monthly? The frequency with which you integrate your data should be addressed in preliminary conversations with leadership and database administrators. Again, consider how you intend to use the integrated data elements as it may help decide what frequency you select. At times, the infrastructure and capacity of your database(s) may also influence frequency.

Step 4: Figure out the process.
Eventually, you will have to understand how the data actually gets shared from one system to another. Is this a bi-directional integration (where data from both systems is shared back and forth) or is this a uni-directional integration (where data from system A gets shared to system B, but no data from system B is shared with system A)? Figuring out the process will take conversations with administrators from both systems. Bring them onboard early and try to learn their terminology.

Step 5: Sync historical data.
Try testing data from the past first to see if and how the integration is going to work. Ideally this is done in a test environment (a dummy system that allows you to figure it out before going live). There will inevitably be issues. Don’t worry. You will be able to solve them with enough time and energy.

Step 6: Training, training, training.
People are the heart of data systems and the need for training cannot be overstated. The topic of housing may not be something that providers think about routinely. Conversely, non-medical staff may not be versed in medical terminology. Train all the staff using the systems on the new data elements, changes in procedures, and any other factors that they need to know. Be there to answer questions and plan to retrain as necessary. No one likes to learn a new system so be patient and understand their frustration.

Step 7: Meaningful use.
Getting to the point of where data is continuously being integrating is a tremendous accomplishment. Celebrate! Then get to the work of putting the data to meaningful use. If providers now have access to the information the data depicts, how does it influence their conversation with clients? Perhaps your agency will implement procedural or process improvement. Ensuring data integration ultimately positively impacts services and the client experience is the end goal.

Integrating data from multiple systems has the potential to improve decision making for providers,
increase provider acumen, improve quality for clients, and in the long run save more lives. While KCHD’s process of integrating housing and medical case management data was challenging, the rewards have been shown. In the next post in this series, we will show the outcomes of integrating data for people living with HIV.

*This project was funded by the Health Resources and Services Administration’s (HRSA) Special Projects of National Significance. Technical assistance was provided by the RAND Cooperation. We are grateful to both for their support during this process.

Health Care

Health Forward Foundation
2300 Main Street, Suite 304
Kansas City, MO 64108
(816) 241-7006