The Surprising Benefits of Bad Healthcare Data
Healthcare organizations that strategically gather and analyze healthcare data can dramatically improve outcomes and reduce costs. But health systems and their data consumers (frontline staff, leadership, finance, IT, patients, etc.) are starting to feel overwhelmed and burdened by the volume of information they encounter each day; especially when bad healthcare data is involved.
Healthcare organizations can ease the bad data burden by creating a culture that embraces data transparency; one that empowers data consumers to operate as problem solvers, working together to find and fix bad data. Health systems that accept bad data as an inherent part of being an analytics organization respond to data issues more quickly; they encourage open communication and feedback, and dedicate the necessary time and resources to act on that feedback.
This article explains exactly how health systems should approach bad data, from understanding where it comes from to avoiding the temptation to scrub (modify) it. The article also provides a pragmatic definition of data transparency, reveals the surprising benefits of bad healthcare data, and provides three steps for creating a culture that considers transparency an absolute necessity. Bad data is an unavoidable, necessary step on the always winding road to outcomes improvement.
What Is Bad Healthcare Data and How Does it Happen?
Bad healthcare data is unwanted or incorrect healthcare data stored in health system applications. Applications store digital records that may contain bad data for several reasons:
- No application is perfect. Input mistakes are inevitable. The information stored is only as good as what was entered into the application.
- Some applications have advanced rules to increase user convenience and store data more efficiently. Sometimes, rules are wrong and incorrect information is captured.
- Calculations applied to data may be incorrect.
- Health system or healthcare rules change, invalidating existing data.
Health systems use a variety of applications, each with their own complexities and challenges. These applications house a variety of data, including EMRs, lab data, patient satisfaction data, inpatient and outpatient billing, general ledger values, revenue cycle figures, human resources information, risk management information, and provider certification tracking. The sheer volume of information and disparate data systems creates opportunities for bad data.
Even the best healthcare organizations in the world encounter bad data scenarios. Bad data will exist as long as human beings are involved and healthcare evolves. Fortunately, there are endless opportunities for improvement; opportunities only bad data can reveal. Health systems should strive to identify bad data, share feedback regarding bad data, and analyze sources of bad data. When these data due-diligence tasks are complete those same health systems should follow through fixing bad data with the ultimate goal of getting a little closer to the truth each time. With time, decisions become data driven and systems start making more accurate predictions.
How to Find Bad Healthcare Data
There are several ways data consumers find bad healthcare data:
- Using applications regularly: consistently using applications tends to uncover bigger data issues.
- Operational reporting: monitoring daily, up-to-date trends from the application.
- Analytic reporting/reporting done from a reporting system of disparate data sources, such as an enterprise data warehouse (EDW): seeing data in multiple contexts from a variety of sources makes it easier to identify source data problems.
Of all the ways consumers find bad data, it’s easiest to spot data problems from analytic systems (analytic reporting) because they look at trends over time; uncommon trends typically point to data issues.
Raw Data Is Best: Resist the Temptation to Scrub
When data consumers find bad data in the extraction process, they’re often tempted to modify the data as it’s loaded into a reporting or analytical system, such as an EDW. Modifying or scrubbing bad data may create a more pleasant-looking report. Sometimes data is scrubbed for fear of having to explain what is being reported. This can happen with any type of data, but is most common with numbers and complex calculations, such as finance values, that are time-consuming and difficult to explain.
Scrubbing data is also tempting when health systems don’t have control over the source system. For example, they may receive claims data in a flat file. In this situation, health systems should continue to fix the problem at the source and be pragmatic and transparent about scrubbing rare exceptions.
There’s a strong temptation to scrub bad data to avoid making a change in the source system—a difficult process that requires coordination and hard work. But the essential element of true data transparency is making healthcare data visible to consumers in its raw or unchanged state. Brave health systems will avoid the temptation to modify bad data by realizing its potential for good. They will continue to share data, dedicate resources to receiving feedback, and demonstrate a willingness to make necessary changes.
The Surprising Benefits of Bad Healthcare Data
Given the numerous temptations health systems face to scrub bad data, it’s important to remember why bad data is good.
Benefit #1: Provides Valuable Feedback to Application Users
Data consumers are the application users; they want to know if they’re making mistakes. Bad data reveals correctable human error-related input mistakes.
Benefit #2: Inspires an Improvement Culture
Being transparent with bad data tends to elicit negative reactions, which can speed up the process of fixing the source problem. This is especially valuable if the organization insists on resolving problems at their source rather than modifying reporting systems. By making the problem visible until it’s fully resolved, organizations inspire an improvement culture that embraces changes and prioritizes quality.
Benefit #3: Creates a Snowball Effect of Success
Fixing bad data and achieving success inspires data consumers to follow the problem-solving process. One small success will most likely snowball into more.
Benefit #4: Leads to More Accurate Data
As data becomes cleaner and more reliable through feedback, a culture of change develops; one that is dedicated to resolving problems instead of ignoring them. Data becomes more accurate and valuable for identifying outcomes improvement areas, measuring baselines for clinical, financial, and waste management improvements, and reporting the progress of interventions in calculating improvement measures and ROI.
Overcoming the Trough of Despair
The trough of despair (see Figure 1) exists wherever healthcare data exists. What begins as hope and euphoria from data consumers turns into despair and finger pointing as the EDW lifecycle starts to mature and bad data emerges.
Healthcare organizations can overcome the trough by creating a data transparency culture based on feedback and teamwork, which can turn despair into perseverance and commitment (the final stage of the trough). Don’t work around bad data; commit to the continuous cycle of fixing it.
The Best Definition of Data Transparency
PC Magazine defines data transparency as having two parts:
- The ability to easily access and work with data no matter where they are located or what application created them.
- The assurance that data being reported are accurate and are coming from the official source.
But the best definition of data transparency adds two additional important elements:
- Sharing data even if it ends up being inaccurate: health systems should strive for accuracy; they should believe what they’re sharing is as accurate as possible. But they should understand that bad data is inevitable. And that sharing data—good or bad—with the right consumers is an important part of increasing data accuracy.
- Sharing data in its raw or unchanged state (data that isn’t modified at the time it moves from the source system to the reporting system): transparent data is accurate to the extent it hasn’t been modified, touched, or meddled with.
Data transparency means giving the right data to the right people at the right time; nothing more and nothing less. It means trusting data consumers with raw, unchanged data.
What Data Transparency Is Not
Data transparency doesn’t mean knowingly sharing bad data with consumers, hoping they’ll discover the problems for you. Data transparency means sharing data that, to the best of your knowledge and ability, is accurate and properly sourced (knowing that bad data is inevitable).
Data Transparency does not mean making all data visible to everyone. Healthcare organizations must continue to abide by government and internal regulations and data governance mandates in terms of how data is managed and with whom data is shared. Transparent data isn’t synonymous with ungoverned data; all healthcare data must be governed.
How Data Transparency Improves Healthcare
Data transparency makes it possible for healthcare organizations to put themselves in a continuous, sustainable cycle of identifying and solving problems. The organizational culture that embraces data transparency also embraces change and improvement. After all, the main motivation behind data transparency is improving outcomes. Real-world examples of data transparency leading to major healthcare improvements abound:
Test Patients in Production EMR
In one example, test patients in the production EMR were skewing reports. The rule (filter) designed to keep test patients out of the EMR wasn’t working. As a result of showing bad data in reports from the EDW, the health system found thousands of test patients. The bad rule was found and fixed in the source system, as were the records. Readmissions and length of stay (LOS) accuracy significantly improved.
Erroneous Discharge Dates
In another example, one system discovered that discharge dates were recorded as future dates (2020, for example). The system discovered this bad data while creating analytic reports with date filters. The source of the problem was, again, an incorrect rule during an application data migration (old EMR to new EMR). Fixing these dates resulted in more accurate reports and lead to more effective interventions.
Blank Diagnosis Data
Blank diagnosis data should be impossible, but in this next example, older records were created before a rule was in place to prevent it. The blank data resulted in year-to-year analysis issues. Fixing the issue in the source system resulted in more accurate trend information.
Missing Critical Demographic Data
In the final example, one system discovered missing demographic data used for population health management while looking at exceptions or “other” categories of demographics information, which turned out to be blank. Fixing this issue improved the system’s ability to better manage the health of patients.
Empower, Share, and Act to Create a Data Transparency Culture
Data transparency is essential for improving data accuracy, but changing an entire organization’s perspectives on bad healthcare data isn’t easy. There are several steps health systems can take to start building an improvement culture that embraces data transparency as a core value.
Step #1: Empower (Encourage Staff to Provide Feedback)
Empowering data consumers to share feedback and report bad data is the critical first step in creating a culture of data transparency. Healthcare leaders should have an open door policy for receiving feedback—positive or negative.
Step #2: Share (Create a Feedback Mechanism)
The next step is creating a feedback mechanism that reassures staff that their feedback is relayed to the appropriate person, team, or department.
Step #3: Act (Dedicate Time and Resources to Respond to Feedback)
The third step is realizing the importance of dedicating time and resources to receiving, analyzing, and responding to feedback; make changes based on feedback to encourage more feedback. Acknowledge and thank providers of feedback and highlight feedback that leads to improvement.
The biggest deterrent to creating a culture of change is hoarding data based on the mindset that data consumers won’t understand it. Share data, encourage and empower data consumers to react to it, provide the appropriate feedback mechanisms, and do something constructive with the feedback. Systems can’t fix problems unless they know they exist. Embrace scrutiny and criticism as a valuable part of the improvement process.
Health systems that implement this empower, share, act feedback cycle will begin to see success as feedback turns into tangible improvement; as bad data turns into good data. All it takes is one positive change for staff to feel inspired and motivated to participate in the feedback cycle that’s resulted in tangible improvements. These steps will drive system wide demand for good data, even if it means taking the brave first step of exposing bad data.
Get Started Today
Start working toward a data transparency culture now; don’t wait until your data is clean before sharing it with the appropriate consumers—involve data consumers in the cleaning process early.
Getting started with data transparency is similar to getting started with an outcomes improvement project. Don’t tackle everything at once; develop a prioritization strategy that slowly but surely chips away at data problems. Start by targeting areas (such as a system with known problems) that have the biggest potential positive impact to the organization (e.g., immediate use case or biggest ROI). Hone in on the first focus area based on the enthusiasm, support, and willingness of its data consumers, teams, and departments. Systems will have the most success when they strike a balance between solving the biggest problems with the most impact and engaging the right data consumers; consumers that understand the value of data transparency and are willing to make necessary changes.
There isn’t one decent excuse for not being transparent with healthcare data. Health systems should acknowledge the benefits of bad healthcare data outlined in this article (provides valuable feedback to data consumers, inspires an improvement culture, creates a snowball effect of success, and leads to more accurate data) as they relentlessly pursue higher quality data. Health systems brave enough to share bad healthcare data and allocate the necessary resources to fixing it will slowly but surely improve data accuracy. And improving data accuracy improves outcomes.