Many if not all healthcare leader wants to maximize their data analysts’ value and effectiveness. One of the main reasons data analysts aren’t as effective as they could be is not having access to the right tools.
I have had really horrible experiences when it comes to managing data. Many times i told myself this is not what I had signed up for. I didn’t want to spend most of my time gathering data, validating data acquisition methods, reformatting, ensuring appropriate data types, trimming, cleaning, scrubbing, and conforming data in preparation for analysis and reporting. I wanted to put my analytical skills to use. I wanted to help these healthcare organizations solve problems and make improvements.
The root of the problem was not having the right tools to analyze data and discover insights that would drive care and process improvement initiatives. I was spending too much time collecting data and not enough time transforming it into meaningful analytics. In this blog, I describe four strategies that will empower your data analysts to transform data into meaningful improvements.
3 Stages of Transforming Raw Data into Meaningful Analytics
Understanding the tools analysts need to transform data requires some background knowledge. Any type of data, including healthcare data, goes through three stages before an analyst can use it to achieve sustainable, meaningful analytics:
Data capture
Data provisioning
Data analysis
Healthcare Data Analytics
The healthcare data analysis lifecycle.
Stage 1: Data Capture
An analyst’s job is impacted by the way people, processes, and devices produce and capture data. These three entities are responsible for the data’s appropriateness (did they capture the right stuff?), discreteness (did they capture it in the right format?), and ease of data extraction (was the data captured in an accessible way?).
Stage 2: Data Provisioning
Analysts need data from multiple source systems throughout the organization to produce meaningful insights. For example, an analyst assisting a team of clinicians on a quality improvement issue needs a variety of data from multiple source systems:
EMR data (for clinical observations and lab results).
Billing data (for identifying cohorts using diagnosis and procedure codes, charges, and revenue).
Cost data (for determining the improvement’s impact on margins).
Patient satisfaction data.
Aggregating data manually —pulling all of the data into one location in a common format and ensuring data sets are talking to each other (through common linkable identifiers such as patient and provider identifiers)— is extremely time consuming. It also makes data more susceptible to errors. There are more effective ways to gather data (more on this later).
Stage 3: Data Analysis
The data analysis starts after the appropriate data has been captured, pulled into a single place, and tied together. The analysis process consists of several parts:
Data quality evaluation: Analysts need to understand the data by taking time to evaluate it. They also need to note their method of evaluation, which they’ll reference when they share their findings with the audience.
Data discovery: Before attempting to answer a specific question, analysts should take time to explore the data and look for meaningful oddities and trends (this is a critical component of good data analysis.) In my experience, more than 50 percent of acted-upon analyses resulted from stumbling upon something in the discovery process.
Interpretation: Most people think of the interpretation step when they think about analyzing data, but it’s actually the smallest sub-step in a long process (in terms of the time analysts spend on it).
Presentation: Presentation is critical. After all the work spent getting data to this point, the analyst needs to tell a story with the data in a consumable, simple way that caters to the audience. Presentation principles need to be considered, such as those described by Edward R. Tufte in The Visual Display of Quantitative Information.
These three stages of insightful data analysis will drive improvements. Each step alone, however, is not enough to create sustainable and meaningful healthcare analytics. It’s equally important to empower data analysts to focus on analyzing data; not just capturing and provisioning data.
Four Ways to Optimize Your Healthcare Data Analyst’s Value
There are four ways to empower data analysts to provide the insights necessary to drive improvements:
EDW for data aggregation#1: Provide Analysts with a Data Warehouse
The most effective way to empower analysts to identify value-added improvements is by implementing an enterprise data warehouse (EDW). The EDW becomes a one-stop shop for data aggregation. Using just one login, analysts can access any data across the entire health system.
Some people don’t think an EDW is necessary; that it’s possible to bring the data together manually on an as-needed basis. This sounds fine in theory, but an EDW provides many other critical attributes: metadata, security, auditing, and common linkable identifiers. My colleague, Mike Doyle, further describes the importance of EDWs in his article “Do I really need a healthcare enterprise data warehouse?”
#2: Provide Analysts with Full Access to a Testing Environment
Give analysts ample opportunity to build, break, and rebuild data sets within the EDW. Analysts should be able to use the data warehouse like a sandbox in which they can store anything they consider useful.
#3: Provide Analysts with Data Discovery Tools
Data discovery tools, such as business intelligence (BI) tools, make it easy for analysts to explore the data and look for useful oddities or trends. But not all BI tools are sufficient for in-depth data analysis. BI tools might feature nifty charts and graphs help the masses understand what the data is saying. But they also need to make it possible for analysts to drill into the data to find trends and meaningful correlations. The right data discovery tool should enable analysts to build intertwined, insightful reports that lead to system improvements.
#4: Provide Analysts with Direction
Healthcare data analysts need direction, not step-by-step instructions about what their reports should contain. Step-by-step instructions result in one-off reports linked to very precise requests. Providing direction, on the other hand, leads to deeper, more meaningful insights that help solve problems and make improvements. The best report requests provide enough direction to put the analysts on the right track and enough leeway to encourage analysts to ask more questions as they analyze the data.
Provide analysts with direction, time to investigate the problem, and a forum for asking detailed questions. The end product will be substantially better because it includes not only what the requester initially wanted, but also additional insights that go deeper into the data—which might be exactly what the organization needs.