Best Practices for Organizing and Transforming Data into Action

The ever-increasing drive toward comprehensive, coordinated, affordable health care has made data analytics an indispensable part of the health care landscape. Even small organizations are involved in collecting large amounts of data about the work they do and the populations they serve. Similar to someone who looks in a closet full of clothes and deems they have nothing to wear, organizations may simultaneously feel like they have too much data and not enough. This feeling stems from the two key challenges most companies face working with data: organization and transforming data into action.

Collecting, storing, and extracting information from data can feel like a burden, especially for companies that find it challenging to train or retain employees with requisite data skills, sucking time and resources from their core mission. Many organizations are unsure exactly what data they have, how it is stored, or how to access it. Just like your closet at home, the instinct to save everything can make it more difficult down the line to find that item you need. All of this adds up to missed opportunities for gaining insights from already available information.

On the other hand, when it comes time to put data into action, the available data can feel incomplete or too outdated to tell the story they want. Organizations have the tools to capture data, but they fail to understand what is valuable or how to make it impactful. The reaction may be to collect more information or purchase new reporting tools, adding complexity and expense to a process they do not understand well. Often the result is that the final report or analysis is answering a different question than the author intended, driven by perceived roadblocks to data along the way.  

Third Horizon Strategies (THS) is working closely with clients to solve these data issues. Working with impact-driven health organizations, the firm has established four best practices to ensure that it answers the client’s most pressing questions and that the client understands how data can help them thrive in the evolving health care environment.

1. State your mission: It is good practice for any organization to recenter from time to time. It is especially true when incorporating new technologies such as those around data. Even if you can’t explain every detail, it should be clear from the outset why what you are doing helps support your organization’s overall mission. If it is not, don’t even start.
2. Take an inventory: Write down all the data your organization has collected, where it came from, on what schedule, where it is held, and how it is organized. Include any publicly available data you regularly reference and services you currently subscribe to. Make notes about your typical routine around how you use data.
3. Plan to scale: Most cloud services let an organization purchase as much or as little storage and computing power as needed. But there is more. Small data sets make it easy to catch errors or missing data by sight. As your number of observations grows, so does the need for suitable ETL mechanisms and quality assurance protocols. You should also understand how your organization gets its data. Is it all going through one person? What if that person leaves? Are business association agreements executed in a way that doesn’t unnecessarily restrict access across the organization?
4. Be explicit: To this point we have said very little about actual analytics. Asking good questions is the micro-level compliment to the first practice of stating your mission. When mining data for insights, it is good practice to state questions explicitly and, where possible, in yes or no terms. Does our mobile crisis center need an additional ambulance? What is the average cost of care for individuals diagnosed with a substance use disorder ( (SUD)? To answer the first question, we might examine dispatch logs around location and time of day, numbering only a couple of thousand rows. The second case requires evaluating claims data numbering in the tens of millions. By asking a specific question, we get better insight into the scope and cost of the project than we would otherwise, with a clear path forward.

Case study

Organization A specializes in residential treatment. They have little data literacy internally, so they rely primarily on outside contractors for their data needs, specifically for reporting to grant agencies. While the arrangement is satisfactory, it is becoming too expensive. Furthermore, there is no thinking beyond next month’s report regarding what to do with the information they collect. Recently the organization purchased a new business intelligence tool that is supposed to make it simple to create compelling visualizations. However, the public data that fit the tool seamlessly in demonstrations isn’t so relevant to their work, and their own data is proving difficult to unpack or outright incompatible. Data do not empower this organization. Yet when speaking with its members, they have excellent ideas about streamlining the continuum of care driven not by theory but by practice. They have scar tissue. They’re passionate about their work and know that what they do is making a difference, even if they can’t quantify it.

THS works with Organization A to identify critical domains for data collection and analysis and tie those domains directly to their mission. We take an inventory of all the data they currently collect from disparate sources, where it is held, how it is organized, and whom it passes through, withholding any judgment as to the quality of information. We move all data into a secure relational database where data are organized in tables with variables in columns and observations in rows. Information is drastically more readable and decipherable. We identify key data collection points such as intake and discharge and sharpen our focus around what they want answered. Soon they have identified outcome variables to inform impact, drawing the favorable attention of grants. Confident in their ability to manage current data flows, we form a plan for the future to leverage Fast Healthcare Interoperability Resources (FHIR) standards to exchange information with referring providers and get at the continuum of care issues they spoke about so passionately at the outset.

This case study is far from fantasy, although typing the last paragraph was fun. We want to engage users and hope to get employees comfortable using the right data tools to tell their stories. It starts small, first by getting organized and then by building confidence. Like a good wardrobe, your data solution should be functional and tailored to fit your style.