In late June, Atul Gawande was named CEO of the headline-grabbing new healthcare venture from Amazon, JPMorgan and Berkshire Hathaway. Gawande, a surgeon, professor and writer, shot to fame after publishing a 2009 New Yorker article titled “The Cost Conundrum.”
In it, he explored the anomaly that was healthcare spending in McAllen, Texas. Despite having the lowest household income in the nation, McAllen ranked among the highest in healthcare spending. It was dubbed “required reading” by President Obama and served as a guidepost for policymakers as they carved out the Affordable Care Act.
But there was a big problem with the article that few pundits picked up on—it was partially based on Medicare data that was FOUR YEARS OLD.
On the heels of “The Cost Conundrum,” Trilliant Health CEO Hal Andrews did his own analysis of McAllen’s healthcare spending—using all-payer data from 2007—and came to a different conclusion.
The unfortunate truth is health system executives and policymakers have been making strategic decisions using less-than-ideal data for decades. Inpatient Medicare data, while comprehensive, is anywhere from nine to 24 months old. State data is sometimes older. In the last decade, access to inpatient commercial claims data has cracked the door to better insights, but that, too, has limitations. Not only is commercial data dated, it’s often incomplete.
So, when an analytics vendor promises to inform your multi-million-dollar decisions, how much do you know about the data they are working with? It pays to know before you sign on because incorrect data can be misleading and potentially detrimental for strategic planning.
Here are the questions you must ask when selecting an analytics partner for your hospital or health system.
- What are your data sources?
While commercially insured patients are widely considered the lifeblood of a hospital, many health systems still use legacy Medicare data to formulate strategic plans. And many analytics vendors justify this practice by telling themselves—and prospective clients—that Medicare is directionally correct. But that couldn’t be further from the truth, as evidenced by the rapidly changing healthcare landscape where value-based care is replacing fee-for-service, payers are becoming providers, and consumers are holding the reins on their healthcare decisions.
The bottom line: A single data source cannot do the job that was meant for a multitude of data sources, which should include commercial payers, Medicare, Medicaid, Medicare Advantage and managed Medicaid.
- Do you have outpatient data?
Historically, hospitals and health systems have developed strategic growth plans based upon information about inpatient volumes and market share. But with the escalating migration of care from acute care hospitals to ambulatory settings, particularly in states without Certificate of Need, it is critical to understand outpatient market share. Any reputable vendor should have access to outpatient data.
- How recent is the data used in your analytics?
One of the challenges of developing strategic plans is data latency. When it comes to Medicare, every analytics vendor is limited by the way in which CMS releases information. What differentiates vendors is the latency in the commercial payer data that they provide. Ideally, your data vendor will receive commercial claims data on a monthly basis.
- What percentage of total claims do you have for my market?
No vendor has or ever will have 100% of all claims in any market, so be wary of any vendor who touts such comprehensive access. But just as concerning are analytics vendors who have access to less than 50% of the claims in the nation. Some are even powering their analytics platforms on less than 25% of available data. Knowing the claims coverage that a data vendor has for your market(s) is essential to understanding whether you are making strategic decisions with confidence.
- Do you use data modeling?
To make up for a lack of data, many analytics vendors resort to a practice known as modeling to attempt to get an accurate view of the market. These models often combine whatever data they have with government projections of market growth, health trends and other assumptions. The problem is, modeling tends to be overly optimistic, meaning hospitals and health systems end up making multi-million-dollar decisions using a flawed, unreliable data model.
The better practice is to use actual data. But in order to do that, there has to be enough data for you to make decisions with confidence.
- Are there any affiliations with your company we should know about?
Like hospitals and health system, many industries are being consolidated, giving new meaning to the term “conflict of interest.” If it’s not apparent from the analytics vendor’s website, ask about any affiliations, such as parent company and a comprehensive list of subsidiary companies.
- How do you cleanse the data?
Claims data is notoriously riddled with errors—and payers don’t mind, because they need only a small bit of information to either approve or deny the claim. Information like referring provider adds no value to the payer but can be immensely valuable for a health system. Unfortunately, values like these are missing on more than 40% of claims. How does your vendor account for this missingness in their data sets?
- How seamless is implementation?
Many analytics vendors need access to your data—whether state data or from your EMR. These processes can be both time and labor intensive. Make sure you understand the implementation process, including timelines and resources required. Look for a platform with a streamlined implementation plan to include training and ongoing support, as well as the ability to run your own queries and export report data and visualizations. After all, an analytics platform is only useful if your team can actually use it.