Choose by Topic
Health Data Stewardship and Its Importance in Healthcare Analytics
Would you believe me if I told you that some of the most important healthcare staff never see patients but instead provide health data stewardship? These data stewards may not even sit within hospital or clinics? What if I told you that some of the most vital roles in supporting improved care delivery never went to medical school? Truth is, every healthcare organization has them — content experts that know every nook and cranny of data collection for mission critical software systems.
These healthcare data stewards are the unsung heroes of healthcare. The majority of their work is often taken for granted. That is, when they do their job well, things go smoothly, so clinicians and hospital staff can document patient care and ensure proper billing/collection. These knowledge experts, in a real sense, are stewards over critical systems of patient records.
Finding healthcare data stewards within a health system
How do you find these stewards? The answer is easy. When something goes awry with a given data system, these are the folks called upon to fix the problem. What kind of problems, you may ask, are they solving? They vary. It could be something as simple as granting a new employee access to use a system of record. A more challenging request could be to merge duplicate records for the same patient into one consolidated and complete record. A more serious but frequent scenario may be that in the wake of a failed clinical outcome, an audit trail of what was done and/or not done needs to be shared with an oversight committee. In each of these scenarios, these data stewards are filling critical roles in making visible an electronic representation of workflow to non-technical people.
Health data stewards are keepers of tribal knowledge. Their intimate and expansive knowledge of how data is collected to represent workflows across different systems has become a goldmine for analytics. Analysts working with data stewards can shave days’ worth of time (and cost) off the report development process. This is possible because data stewards know how processes are documented, and where to find the data that might otherwise get overlooked. Data stewards may also help to fast track report credibility by ensuring analysts know all the possible locations to find and mine data from source systems.
Expanding role of healthcare data stewards with enterprise data warehouses and analytics
Up to this point, we have talked about the role of data stewards within the data systems themselves. Over the last 15 years, however, there has been an emerging and growing need for data stewards to expand their influence in some other arenas. Data stewards are now playing key roles in the field of enterprise data warehousing (EDW) and analytics.
A data warehouse may loosely be defined as a massive repository of data from multiple systems of record. Think of it this way. An analyst charged to write a report on diabetics who are not being well-managed may struggle to build a trusted report. If the analyst were to use the electronic medical record (EMR) as the source for data extraction, they may find an appropriate cohort of diabetics but may miss whether these patients are being well-managed because they are only looking in the EMR. The EMR may have lab orders and results —all indications of good care delivery.
Suppose however, that the hospital system also has an ancillary lab system that also collects lab orders and results. Those may or may not be found within the EMR if there is not a dedicated feed to tie these systems together. Or perhaps there are claims data systems that also contain limited clinical data on the hospital’s self-insured population. The claims data is not likely going to interface with the EMR, so claims on diabetic care received would not be found in the EMR. Moreover, the analyst may only have access to query the EMR and may not have the ability to query claims or the ancillary lab system. If this is the case, the analyst’s diabetic report would paint an incomplete picture of diabetics not receiving appropriate care.
But what if you could put direct copies of data from the EMR, claims, and the ancillary lab systems into one massive data repository: an EDW? Having data available in one location would provide the analyst with the ability to build a more complete picture of diabetics and their care. Data stewards over the EMR, ancillary labs, and claims data systems would be valuable resources for the analyst to write their report because the data stewards can show the analyst all the places to look (within their copied sources) for care documentation.
There’s just one problem: getting anything more than a few spare moments of a data steward’s time can be difficult because health data stewardship isn’t part of their job description. Few healthcare organizations have proactively formalized the data steward role. Rather, it is usually thrown onto someone with an already full plate. It may be that a data steward’s day job is that of billing supervisor, or department manager, or even a nurse manager.
Formalizing the role of the healthcare data steward
While it may seem difficult to justify at first, organizations committed to using healthcare analytics to improve clinical quality, outcomes, and patient satisfaction while lowering costs (The Institute for Healthcare Improvement’s Triple Aim) need to formalize the role of the health data steward. Primarily, they need to carve out time in those experts’ weeks to devote their knowledge to filling gaps, answering questions, and ensuring the analytics are producing high-quality, accurate, and actionable information. Successful healthcare organizations typically dedicate 25 percent or more of their experts’ time to support these sorts of quality improvement initiatives.
Yes, we said “experts” plural because there will likely be several health data stewards. That can be a culture shock to hospitals and health systems that have been focused on running lean. In order to free up that time, the health system will need to reassign some of the data stewards’ work. But the investment will ultimately return many times its value as the organization realizes the advantage of trusted and timely analytics.
The tipping point to recognizing the need for health data stewardship
When an organization launches an EDW, data stewardship and data governance become important issues that merit careful thought and strategy. With the EDW populated, the health system is now able to get a holistic view of the patient experience across all of the systems of record. At this point the question changes from ‘Can we?’ to ‘How can we?’ An EDW must still respect the rules of Health Insurance Portability and Accountability Act (HIPAA) and patient privacy, however.
Instead of running reports against a single system, organizations now have a wealth of data from multiple systems available. This broad data set brings up new concerns. Using our diabetes report for example, if an analyst had access to the EMR system but not ancillary lab or claims data in the original systems, who is to determine if and how much access he or she could have to data from those systems now within the EDW? Data stewards can help address this problem. They can help data architects (those building the EDW) derive filtered data sets that simultaneously support HIPAA and still give the analyst access to data he or she needs to build her diabetes report.
How health data stewards contribute value to the health system
Because of the tremendous amount of tribal knowledge health data stewards possess, particularly in terms of workflow, they can quickly spot problems within analytics. Let’s explore how data stewards could help in a real-world, process improvement initiative.
Hospitals are mandated to report quality measures. A case in point is that the Centers for Medicare and Medicaid Services (CMS) requires that hospitals report all instances of catheter-associated urinary tract infections (CAUTI) acquired in an intensive care unit (ICU).
Suppose a hospital wants to reduce CAUTI infection rates system wide. A hospital submitting CAUTI quality measures will already have the data needed for those CAUTIs coming out of the ICU because it is required. While ICU data is important to meet the CMS mandated measures, it does not give a complete picture of CAUTIs across the entire hospital because catheters are used in many other departments besides ICUs. Using only the CMS data set to report CAUTIs will underreport the actual system CAUTI rate. The IT department may not know all the places catheters are used, but a nurse manager or front-line caregiver who is a data steward will.
Data stewards know where to look for information about CAUTIs and catheter insertions in general within the electronic health records (EHR). For instance, although there may be a discrete field for catheter insertion in the EHR, the data doesn’t always get entered in that particular field. There may be (and usually are) multiple methods to document a CAUTI. Only one method of documentation may be the ideal but the reality is that there may be fragmented and incomplete aspects of documentation that are well understood by data stewards.
As the analyst and data steward work together, it is not unreasonable to find a record that shows a clear indication of CAUTI (fever, blood draw, or lab results, for example) even though no catheter insertion was documented in the EMR. Finding exceptions like this provides an opportunity for nursing education and process improvement. The process improvement initiative to reduce CAUTI rates system wide will be much more effective via the partnership the data steward and the analyst to build a complete and trusted report of baseline data.
Healthcare data stewards in action
A recent project of ours demonstrates the value of carving out time for experts to act as data stewards. The executive team from a health system had complaints from the physicians that they were being under-paid for their services, and they blamed the coding department for the discrepancy. The coding department, on the other hand, said they continually had to chase down poor or incomplete documentation from the physicians.
As part of a six-week project, we enlisted the help of a data steward for physician billing. Six to eight hours a week were carved out of the data stewards’ normal duties to help us. The data steward helped us understand the workflow process used to capture information from the professionals. It was manual and archaic. Not only did the data steward help us define the current state, though, they also helped us define the ideal state for improved data capture.
Based on the health data stewards’ expertise, we automated the process to the ideal state, creating alerts across six checkpoints in the patient experience used for billing. This work allowed us to follow the processing path of billing. It quickly highlighted which physicians were following the process and which were not. We brought all of this data to the hospital executives, and showed how better documentation would not only end the complaints but also help them capture additional revenue. By conclusively identifying where the problems were, the executives were able to act on the knowledge and receive significant payment for previously unbilled physician services.
Making the data steward model work
That’s the “why” of designating data stewards. It still leaves the “how.”
It is an absolute must to secure an executive sponsor for your data steward effort. As we mentioned earlier, the people who fill these roles are typically overworked and under-appreciated. Becoming a true data steward is typically beyond the scope of what they can take on unless an executive sponsor gets senior management to view the role of data steward seriously. This is best evidenced by time dedicated for the subject matter experts to contribute regularly on an ad hoc basis. Even better, make it part of their formal job description.
A common mistake around the data steward role is that healthcare organizations may take data stewardship too far. For large organizations, the trend is to build committees, extensive charters, and endless meetings. The bureaucracy of data stewardship risks suffocation of analytic value. Analysts are dismayed by the laborious process of working through data stewards and the time it takes to get anything done. Don’t try to force growth.
Yes, legitimize the role of the data steward by carving time out for the role to be done adequately by knowledge experts. But don’t overbuild a bloated infrastructure. Let the process of data stewardship grow organically. As the value of analytics becomes apparent across the organization, match the growing demand on data stewards with incremental governance and additional executive sponsorship.
It is worth pointing out that IT should not lead the data steward effort. While IT is an important contributor in terms of building the databases and creating appropriate database security roles, they lack the knowledge about the specific systems and how data is captured during patient flow. They also don’t know the data well enough to determine who should have access to which data. IT should support, not lead the data steward effort.
Start small — then let the unsung heroes shine
Rather than building a formal process that tries to address every need up-front, it’s better to start small and let the process fit naturally into the organization. Identify a need, determine who can help, get the executive sponsor to carve out the time, and capture a win. As the organization’s analytics capabilities grow, then increase the data steward capacity accordingly. With this approach, it’s possible to build a record of wins instead of failures, which will help generate greater enthusiasm for the trusted analytics. Once you give these unsung heroes a chance to shine — and give them a little recognition when they do —the health system will reap bountiful rewards.
Have you experienced the benefit of health data stewardship? If so, how did your health system approach the reallocation of responsibilities?
Would you like to use or share these concepts? Download this presentation highlighting the key main points.