Outsourced vs. In-house Healthcare Analytics: Pros and Cons
Healthcare is undergoing a major disruption that’s causing health systems to face an inevitable question: how to improve quality while reducing costs? Most organizations recognize that part of the answer lies in analytics. Using analytics, health systems can evaluate efficiency, effectiveness, and find improvement opportunities that are impossible to find any other way. But what is the best way to approach analytics?
At a high level, organizations often choose between two principal approaches to performing analytics:
- Outsourcing the analytics function to benchmarking companies/software-as-a-service. Benchmarking companies aggregate data from many different hospitals and then run it through a proprietary algorithm to come up with comparisons. For example, these companies might analyze the data to compare risk-adjusted mortality rates for sepsis among academic medical centers or small community hospitals.
- Tackling analytics in-house with a system’s own healthcare enterprise data warehouse (EDW). In this approach, a health system aggregates data from its own sources — clinical, financial, and other — to create a data foundation for improvement projects. This process of analytics deals with improving performance internally rather than comparing performance with external peers. A key component of this work is focusing on the fidelity of the data, where details matter as it rolls into the bigger picture.
Both of these analytics approaches have their place in the industry—but they serve two different functions. To help clarify what can be accomplished using each approach, we’ve outlined the pros and cons of each.
The Pros and Cons of Outsourced Analytics
Benchmarking is an excellent way for health systems to look at performance at a high level and see how it stacks up against the organization’s peers. Outsourced analytics can also provide high-level insights into needed improvements: if a system’s sepsis mortality rate is 20 percentage points higher than the industry average, there is a clear need to take steps to improve that care process.
This approach delivers summary-level outcomes metrics that serve as a valuable snapshot of performance to executives. These metrics also provide value to areas of the health system, such as the marketing department, which can use this data to demonstrate that the given healthcare organization delivers better or less expensive care than competitors.
The major drawback of outsourcing the analytics function is that benchmarking focuses simply on highlighting high-level outcomes. It does not provide insight on how to change those outcomes. For example, benchmarking data can show that an organization is in the 50th percentile of risk-adjusted mortality for sepsis, but it cannot show how to improve performance enough to be in the 80th percentile.
Another drawback to relying on benchmarking data is that this data can be unreliable. People assume when looking at benchmarked data that it compares apples to apples (i.e., it is comparing data from one academic medical center with the same data from other academic medical centers). However, because of data quality issues—and there are always data quality issues—we can’t always trust that this assumption is accurate. The reality is that not all hospitals are equally skilled or have applied resources toward delivering high-quality data to the benchmarking company.
Finally, clinicians/front-line staff might not trust benchmarked data, even if it is accurate.Physicians and nurses do not always trust the accuracy of outcomes metrics if they do not know exactly what steps the data has passed through to arrive at the metrics. This is especially true in cases where the data indicates a need for significant performance improvement. Physician trust in that data only comes through a significant fingerprinting process in which the physicians are working within a cross-functional team of data architects and analysts, where the data can be both summarized and detailed. Then, and only then, will they trust the data and lead out on improvement efforts.
Pros and Cons of In-house Analytics
The major benefit of in-house analytics using an EDW is that it allows health systems to improve—rather than just report on—performance. The EDW delivers access to fine-grained detail of the data so that improvement teams can identify the root cause of any performance problems. Based on that insight, the team can design a process or intervention to improve performance. The team can then use the EDW to measure how the intervention impacts the targeted outcome metric.
Another significant benefit of the in-house approach is the ability to determine data lineage. Rather than shipping data out to a third party and then receiving high-level metrics in return, an organization can control and track everything that happens to the data, from the source to the final output. Knowing every step the data passed through to arrive at a metric enables an organization to establish whether or not the metric is indeed meaningful.
This concept of data lineage is not only important for the accuracy of the metrics on which improvement metrics are based, it is also essential for clinician engagement. For example, imagine an organization is launching a data-driven initiative to improve sepsis care. Part of that effort involves using the data to identify a cohort of patients. This process, when done well, involves quite a bit of fingerprinting by clinicians on the frontlines. These clinicians contribute to identifying a meaningful cohort by asking questions such as, “How did we define this cohort? Did we include a lactate test? Did we take a look at broad spectrum antibiotics?”
By participating in this process, by looking at the data, giving input and seeing it incorporated, the clinicians who will be affected by the improvement initiative are validating and accepting the integrity of the cohort. It is their data, their process to change, and their outcomes to improve.
The great value of this approach is that it allows clinicians to take full ownership of the data and the improvement process. When a health system outsources, that ability is lost. The data goes out, the organization gets a metric back—and has no insight into the steps in between. It is hard to build faith among clinicians in benchmarked metrics that are simply presented to them in that manner. In contrast, through the internal process of fingerprinting and data validation, clinicians develop a sense of ownership and confidence that what the data reveals is accurate. Getting this ownership is a very critical step for getting people to trust the data, to acknowledge that there is room for improvement, and to take appropriate measures to improve processes.
The most frequently mentioned disadvantage of the in-house analytics approach is that it requires significant resources—an investment in the right personnel and the right technology infrastructure. It requires an EDW and analytics software. It also requires database experts, data architects, and analysts on staff who can manage the data, infer meaning from it, and then work to establish interventions to drive improvement.
Another limitation of the in-house approach is that it drives internal improvement but doesn’t provide insight on how an organization’s performance compares to that of its peers. It’s worth noting, however, that there are drawbacks to putting too much emphasis on comparative performance measures. Doing better than the hospital next door doesn’t necessarily indicate excellence. By focusing on internal improvement, organizations can ensure that they are truly performing to the top of their capability, not simply doing better than their peers.
Looking Backward versus Moving Forward
W. Edwards Deming, a pioneer of total quality management, famously stated, “Managing by results is like looking in the rear-view mirror.” That’s what relying on outsourced analytics that consists only of outcomes data does. It allows your organization to look back at the past but doesn’t provide insight as to how to succeed in the future. It can signal the need for a change but doesn’t help create that change.
The performance indicators that benchmarked data provides are valuable. But in today’s industry where improving performance is paramount, organizations need more tools and insight to take action based on these signals. This is where an in-house EDW and analytics come into play. These tools enable organizations to pinpoint what needs to be improved—and then move forward to drive and measure that improvement.
Would you like to use or share these concepts? Download this presentation highlighting the key main points.