Evaluating your provider network behavior to improve patient retention

By Jeff Herman, Senior Vice President, Product Management, WebMD Ignite

This article is the second in a series that shines light on the key drivers of physician referrals. The series draws on Chartis’ work with health systems and WebMD Ignite’s data analytics and insights on consumer access and referrals. You can also reference the first article in this series, where we unpack the impact of patient experience on provider referrals.

It’s no surprise that health systems want to drive patients to their facilities — it’s the first step to creating lifetime value for the organization. But even with the most sophisticated marketing capabilities, it’s only possible to realize additional patient volume if your provider network is set up to capture and route patients into your facilities. As Chartis has noted, identifying and intervening with provider behavior is complex. But ultimately, the endeavor will be a worthwhile undertaking for your organization.

In this post, we’ll unpack the key steps health systems must take to understand their markets. We’ll discuss the benefits and challenges of using the data you have available, and share insights (and surprises) from a real-life example from a large integrated delivery network in the Northeast.

Better understand your market using these key metrics

First, you want to quantify which service lines have the material encounter volume and contribution margin to make an impact. Next, you’ll need to identify both your network’s referral loyalty (how often PCPs and other referral sources choose to refer into your provider network) and your network’s rendering loyalty (how often specialists choose to treat patients within your facilities). Once established, you can link these two behaviors at the patient level, to determine where they are dropping out of the care continuum.

This analysis allows you to understand that:

• By identifying key referral patterns — and their points of failure — you can improve your network’s effectiveness, enhance patient continuity of care, and boost revenue generation.
• Analytics allow you to shortcut directly to the most needed areas of improvement.
• You need to establish relationships with new providers, strengthen bonds with existing ones, and make new referral connections between them.
• Your patient marketing can only go so far if you don’t have an effective provider network ready to capture it.

The data can be used at various levels to support you at different points in your strategy. Use higher level-insights, such as network integrity by specialty, to point you in the right direction. Don't be afraid to dig deeper: differentiate the analysis by region or sub-specialty. Be sure to support your provider conversations with specific, hard data points, such as individual provider-to-provider relationships and measured lost volume. And don’t forget to consider other factors beyond claims analysis. There may be operational variables, such as capacity and ease of access, that also come into play. Having all of this information provides a more comprehensive overview, and helps inform actionable next steps.

Be aware to the challenges that come with using claims data

While the concept of the analysis is straightforward, there are operational challenges. For one, claims data can be expensive to acquire, store, and manage. And by its nature, claims data is inherently incomplete. It requires enhancements to become useful, including:

• Aggregating claims from as many sources as are available, while avoiding duplication.
• Allowing tokenization of patients across multiple data sources. This provides the longitudinal patient tracking that needs to be superimposed on the dataset.

While it’s true that identifying providers can be fairly straightforward due to NPIs, identifying facilities is considerably less so. Hospital sites can be easy to ascertain due to clear names and the presence of inpatient volume. But clinics and provider groups are notoriously difficult to assign to an integrated delivery network (IDN). In the claims data, many do not have names that clearly assign themselves to the parent system (e.g., “Cardiovascular Associates”) and are listed with a billing address instead of the actual site of care.  

If you have a data science team in house, you can apply logic to triangulate multiple data points, such as the roster of providers, a shared billing address with other known facilities, references, and others. Remember, too, that raw claims data will represent multiple claims per point of care encounter. Surgeries, for example, would include separate claims for the facility, the surgeon, and the anesthesiologist. To avoid duplication, these will need to be rolled into a single encounter-level metric.

A real-life case study yields proof points to facilitate discussions

We ran an in-depth analysis for a large health system in the Northeast. They were looking to identify the areas of their network where the greatest patient leakage was occurring.  

The basics of the analysis provided the top line information they needed, such as the network integrity score by specialty, so they could prioritize by intervention. But as we dug into the drivers of leakage within each of the services, there were some important discoveries. Specifically:

A low network integrity score may have a lopsided driver.
When looking at the rank order of their network integrity scores by specialty, we examined what was lowest on the list. Surprisingly, 80% of the referrals for this specialty were remaining in-network. How? Because the specialists themselves were leaking patients to competing facilities — 95% of the time. The inverse was also true. When referrals were received, the specialists were highly loyal and kept the patients in-network.

Declining volume may not be the result of your provider network.
For each of the health system’s key services, we trended the volume of patient cases that were captured at their facilities over a five-year period. Two service lines saw consistent decline in captured cases quarter over quarter (Q/Q). To examine the driver of these losses, we evaluated three metrics overlaid by quarter: the total market volume of the service, the health system’s network integrity score, and their resulting market share.

Q/Q for cardiac surgery, the total market volume was decreasing, while network integrity and market share were steady. When examining these in tandem, we could see that the decline of their captured volume was the result of a shrinking market, not their provider network integrity.

On the contrary, Q/Q for neurosurgery, total market volume was steady, while network integrity and market share were declining. Neurosurgery’s volume loss could be attributed to their provider network. As market volume remained steady, the rate of their provider group’s capture dropped over time, which was reflected in their market share.

While your employed providers are more loyal, you can’t extricate affiliated providers from the equation. 
A patient’s pathway includes three milestones: a referring provider, a rendering provider, and site-of-care facility. If we require that at least one of the two providers (either referring or rendering) was employed, the network integrity score shot up. In this case, from 44% to 70%. That jump was even more drastic when restricting to surgical cases: from 38% to 82%.

A health system can facilitate discussions with providers to unpack their decision points that drive them to practice out-of-network, and offer suitable alternatives that remain in-network. One such discussion revealed that electrophysiology services weren’t available in-network in a specific area of their market, highlighting an actionable next step for the health system: hire an electrophysiologist in their north region.

Final thought

Claims data analysis can provide a powerful tool for health systems to evaluate their provider network. While there are operational challenges to gathering and analyzing the data, the results are well worth the effort, either by your internal teams, or by partnering with a provider solutions company. Using analytics can make your provider network more efficient, improve your patient’s continuity of care, and generate additional revenue.  As one health system leader concluded, “These insights are invaluable because now we have the data behind our suspicions.”

Chartis’ companion article to this blog post offers an executive overview of building your organization's overarching referral management strategy.