Part 1 of 2: Optimizing provider data and streamlining analytics processes

Sept. 3, 2019
Smart people should be focused on intelligent activity, not frustrating data wrangling

After working for some time as a young software developer for a large hospital system, I oversaw my first enterprise data warehousing (EDW) project, called the “Practitioner Master File.”  The team consisted of the EDW manager, a Database Architect (DBA), and our Chief Medical Information Officer (CMIO), who lent his subject matter expertise as needed. After two years of diligent work, I developed a great appreciation for the many internal and external data sources involved and the numerous complexities related to the importance and priorities of various elements for practitioners.  Important provider data are spread across many systems, and there are numerous governance and technical challenges involved when coalescing them into a single version of truth.

Today, I realize that many healthcare organizations across the country have still not succeeded in such an endeavor. To understand best practices for optimizing provider data and streamlining associated analytics processes, let’s look at the current situation in the healthcare industry, a strategic approach to solving the challenges, and the benefits gained by doing so.

Understanding the problem

For many reasons, providers and payers often find it challenging to develop something similar to a Provider Master File. Key data elements reside within many source systems, and state or national rosters are incomplete and inaccurate, especially as they relate to group, employer, and location. Modern electronic health records (EHRs) have been beneficial in many ways, but most have provider lists that consist of little more than National Provider Identifier (NPI) and name.

While this information may be useful for security and access, it lacks the many data elements needed for effective downstream analytics. With no single version of truth for all provider data fields, analysts and report writers are forced to manually reconcile provider data across five to ten systems in order to meaningfully group information. This slows the productivity of provider-oriented analytics significantly. On the frontend of these systems, provider information such as phone/fax, practice, address, and even specialty will change with frequency, often requiring several phone calls and e-mails to get outdated or incorrect information updated. 

While a dozen or more data sources may have important provider data, there are typically five key internal systems and two external systems that are most important for harmonization: 

1. Internally, credentialing systems have the most robust information for providers with privileges, HR, or workforce systems for visibility into employee/contractor ID with service line, department, and some cost information.

2. Ambulatory emergency medical records (EMRs) provide information about where providers are practicing.

3. Acute EMRs illustrate on whose behalf providers are working or performing procedures within the hospital. 

4. Marketing has all of the consumer-friendly specialty terminology that are patient/member/consumer-facing on an organizational “find a doctor” website. 

5. The enrollment system is essential to understanding payer organization associations. 

6. Externally, data from National Plan and Provider Enumeration System (NPPES) and the Center for Medicare and Mwdicaid Services (CMS) serves as supportive sources of provider information. 

Of course, the issue with all of these separate systems is that they are individual data silos with various source technologies and diverse information models, file formats, and update frequencies.

I struggled with these same challenges at hospital systems over the last 20 years. We were innovators and visionaries who invested heavily in data management and analytics, with dozens of people dedicated to both areas. However, as years turned into decades, we each experienced the law of diminishing returns when trying to do things manually or with piecemeal technology.

At some point, there is no longer an appetite for continuing to hire full-time-employees or throwing smart people at the problem. To be sure, those smart people should be focused on intelligent activity, not frustrating data wrangling.

In my broader work with the National Quality Forum and the Healthcare Data Analytics Association, I saw dozens of the most prestigious healthcare organizations in the country struggling with the same issue.

Next week I’ll explain how to take a better approach to collecting, organizing and using data.

Photo 107269241 © Alenavlad | Dreamstime.com
Photo 46773848 © Fortyforks | Dreamstime.com
Photo 271525294 © (null) (null) | Dreamstime.com
Photo 255305755 / 2023 © Nataliia Mysik | Dreamstime.com