With the proliferation of Electronic Medical Record (EMR) and Enterprise Resource Planning (ERP) systems across the healthcare industry over the last decade, healthcare institutions now have more data (and data mining tools) at their fingertips than at any time in history.
On the surface, a highly distributed, self-service information access model appears to be the ideal goal of an EMR or ERP because it can eliminate bottle necks associated with a traditional report writing group handling the requests of an entire organization. Instead, this model can promote efficiency and more timely decision-making and management because it allows users to access information at their convenience at any time.
"On the surface, a highly distributed, self-service information access model appears to be the ideal goal of an EMR or ERP because it can eliminate bottlenecks associated with a traditional report writing group handling the requests of an entire organization"
Children’s Health in Dallas recently embarked on an Analytics Process Discovery to validate our self-service model, breaking down the project down into two phases: Discovery and Process Improvement Recommendations.
The Analytics Discovery Process
Children’s Health allows users to access information directly from data in our EMR and ERP systems, which have been in place for approximately 10 years (with over 12,000 reports being generated during this time). In order to support advanced data reporting needs (or to help groups whose staff are unable to use the reporting tools), we maintain a centralized data intelligence group, which receives approximately 200 requests a month.
To validate the process, we interviewed user groups across the organizations that were accessing information in an effort to understand what reports they were running, which tools they were using, and we inventoried the reports and tools that groups were using to access data.
We quickly founda general lack of organization and standardization of reporting. Multiple groups were creating the same information, but not always in the same manner, resulting in conflicting numbers. This phenomenon appeared to be partially related to employee turnover, which resultedin a general lack of knowledge of the report’s background and intent. Additionally, we found multiple iterations of essentially the same report because some users felt it was easier to create a new report than to modify an existing one. In other words, we had a bloated inventory without effective management.
Data reconciliation was another area of concern. Through we require certification in order to access EMR and ERP data models; numerous users expressed the thought that, if the report ran, it must be correct.Given the discovery of so many duplicates and inconsistent reports, we saw this perception as an opportunity for additional training and education, and also an opportunity to more frequently review reports to confirm the intent and purpose of the reports is being met accurately.
Recommendations from the Discovery Process
Governance: The single most important recommendation gleaned from this project is that reports and information requests should be part of a governance process, where a multi-disciplinary group ensures that information is being produced consistently and with proper definition—and that the information being created is consistent with metrics and measurements that are in line with those of leadership. Implementing a strong governance process could restore standardization and definition of data, lessen some of the individual interpretation that is occurring today, and contribute to a more transparent environment.
Clean-up: Since the distributed environment is part of the culture, a ‘clean up’ effort must occur—reports that are outdated or have not been used in a specified time-frame need to be eliminated. Reports with multiple iterations should be consolidated to reduce clutter, and a plan for general maintenance going forward needs development.
Information accuracy: We must establish consistent sources of truth for data and communicate these sources so users will know where to go for the information they need. Since EMR data can be replicated in downstream systems, it’s important to socialize the sources of truth because not all systems are updated at the same time. For example, if one user creates a report from the EMR and another creates the same from the Cost Accounting system, both systems will eventually have the same data, but the timing in which the systems are updated may be different—so a source of truth must be communicated for consistency and accuracy.
Just as we must clean up our reporting inventory, we must also clean up our toolset inventory. Over the course of time, we have accumulated multiple (sometimes costly) tools, which contribute to the problem of multiple report variations, impacting the consistency and standardization of our data.
Finally, training and education must be developed beyond the certification. We want users to understand data beyond the simple drop and drag of an object in a tool—they need to understand the functional aspect of the data as opposed to just the technical machination of it.
In our view, implementing these recommendations will result in improved data management and consistent measurements to help our organization meet the challenges that we face in today’s healthcare environment.