The health care industry has become progressively more concerned with and focused on the potential of “big data,” with some going so far as to call it the miracle cure needed to fix a broken system.
While big data certainly holds promise for informing process improvement and care outcomes, it is not a silver bullet for delivering better care less expensively. Big data can provide a fascinating overview of a large population. However, harnessing that data to identify processes that can improve outcomes requires targeting a specific set of patients and developing a comprehensive picture of their health care needs. This can be difficult, given the number of organizations and technology systems in play – each of which generates a different data set and often refuses to interact with the others.
Claims data: the good and bad
Claims data, sourced from bills submitted by physicians to health insurance plans, should be viewed as horizontal; that is, all care for a patient or patient population across a specific period of time. Claims data can track patient care from system to system and across any setting: a pharmacy, lab, hospital, out-of-state provider office, etc.
While claims data shows all of the locations where a patient has received care – and the services a patient received – this data often misses crucial details. Claims information is only as good as the staff member doing the coding and is prone to errors and omissions, such as codes not billed or under-reported by each provider. In addition, claims data are routinely more than ninety days out of date by the time they reach the user. For example, claims data might show that a certain test was done, but will not show the results of that test, or the assigned diagnosis may not match the test or procedure performed.
Clinical data: the good and bad
The rise in use of electronic health records has provided physicians with access to a tsunami of readily available clinical data. Clinical data should be viewed as vertical; that is, including all services and outcomes for a single patient or patient population within a system. When taken directly from the EHR, this data provides a detailed picture of a patient’s history within an organization: who they’ve seen, tests or procedures they’ve had, and prescription lists. Clinical data is also timelier than the claims data, which can take weeks or months to be prepared and disseminated.
But again, this data usually only includes input from a single system. Relying strictly on clinical data leaves gaps where patients may have sought treatment outside of the facility or health system, such as a visit to the emergency room, an urgent care center, or an unexpected trip to a local physician while on vacation. Even seeing physicians from a competing network across town will result in “missing” EHR data.
Combining the two
Both clinical and claims data provide a history of a patient’s care, but when used alone neither provides the entire story. Providers are rightly suspicious of single-source data because it is based on inaccurate, incomplete, and out-of-date information.
With combined clinical and claims data, providers can get a holistic view of a single patient or patient population over a period of time, and track and manage that population for improved outcomes. This approach is critical to success in care management, where it is imperative to identify high-risk patients, develop care plans to manage their chronic conditions, and then track the impact of that care over time.
Using only claims data to identify high-risk patients means that warning signs in the EHR might be missed – for example, rising creatinine levels might indicate a patient with worsening chronic kidney disease, but those lab results would probably not show up on a claim. Conversely, using only clinical data from an ambulatory system might mean missing repeated emergency visits for a patient with COPD. Only with a combined data set can risk accurately be assessed.
Once an actionable population of high-risk patients have been identified, care managers can use combined clinical and claims data to follow the patient across the health care system, opening up opportunities to steer them toward the right providers and services and manage their compliance with care plans.
Take, for example, one of our large and geographically diverse ACO system clients in the Midwest. By combining their own system’s clinical record with the outcome data from up to 50 other clinical records and the claims provided from both CMS and local health plans, the system has been able to target several hundred of their most difficult patients. By using a series of care management outreach and facility contracting tools, they have lowered their ACO spend by hundreds of dollars per member while increasing their quality and pharmaceutical outcomes. Many of these interventions could not have been implemented with either just claims or just clinical records.
Conclusion
Combining clinical and claims data will ultimately help providers better manage population health, but bringing the data together is often easier said than done. Aggregating data from different sources – all of which store, structure, and share data in different formats – has proven to be a roadblock for health care organizations that want a comprehensive picture of a population’s health.
Health care organizations can get around this roadblock, however, by using software to aggregate and harmonize data from different sources. This technology allows organizations to be truly interoperable and more easily combine claims and clinical data from two or more sources.
With a complete data set, providers can see a complete picture of their patients across the spectrum of care – not only with their organization, but also outside of their network. This visibility could prevent a provider from ordering unnecessary or duplicative tests, inform them if a patient is actually following their care plan or filling their prescription, or help identify a high-risk patient in need of additional care – all insights that are necessary to manage a risk population.
