Why Population Health Management Strategies Require Both Clinical and Claims Data (Executive Report)
patient care. This type of data answers the questions: what is effective, why is it effective, and how can providers leverage the lessons learned to improve care for other patients. Some typical categories of clinical data include patient demographics (age, sex), medical history (vital signs), allergies, immunizations, problem list, diagnoses, procedures, lab tests, medications, and providers (with specialty).
When the care delivery team combines clinical data with patient satisfaction data, they gain additional knowledge about how patients perceive the care processes, which is critical to providing a quality experience. In addition, clinical data shows how any variations in these processes will impact the cost, quality, or outcome of the patient’s encounter. Clinical data is also timely, since it’s collected in real time at every patient visit and allows modifications in the care delivery processes as they occur.
However, despite the strengths of clinical data, its use for population health management is limited because of the reasons below:
|“Clinical data contains essential information for population health management and needs to be obtained despite the barriers.”|
Clinical data doesn’t span the continuum of care. Clinical data provides great detail about a patient’s medical profile and treatment patterns during a specific encounter. However, the details are generally limited to the acute care, ambulatory setting or physician’s officevand are difficult to aggregate into a cohesive view.
Clinical data is siloed. The data in one facility is generally siloed from the others, another major drawback of clinical data. If a patient goes to several places for care, such as a specialist’s office, a pharmacy, or a rehab center, the disparate systems usually don’t share information across the care continuum.
Clinical data is unstructured. Rather than being input into pre-defined fields in a standard form the way a claims system is, 80 percent of clinical data is entered as unstructured text notes, making extraction and normalization of the data difficult.
CHALLENGES THAT MUST BE OVERCOME BEFORE ACHIEVING THE BEST SOLUTION: THE COMBINATION OF CLINICAL AND CLAIMS DATA
While neither clinical data nor claims data provide enough information on their own for population health management, when combined, the two types of data provide the best of both worlds: up-to-date clinical detail along with a complete view of a patient’s care throughout their entire patient care experience. With this comprehensive picture, healthcare providers and payers can manage their populations more effectively and realize the Triple Aim of higher quality, lower costs, and a better patient experience. Yet, to achieve this ideal solution, there are many challenges providers and payers must first overcome.
The greatest challenge to data sharing has primarily been cultural because data has been used as a pawn in contract negotiations between payers and providers. And while they may talk about sharing data, there’s still significant resistance to actually sharing the data. For information sharing to work, payers and providers will have to learn to trust each other, and that’s a huge hurdle to overcome.
Patient Cohort Definition Challenges
Most organizations use groupings of administrative (claims) codes, such as CPT, ICD-9, and ICD-10 to define cohorts. This approach is a valid starting point, but such macro level groupings often miss patients who should have been included. Adding clinical data, such as lab results and imaging studies, is necessary to refine the cohort and ensure the full eligible population is included.
Even though it will be difficult to group specific patient populations into disease cohorts, it’s a foundational step and will give providers and payers the ability to answer questions such as the following:
- What is this population’s utilization of healthcare services across the continuum?
- How much do the population and individual members cost the organization?
- What is the risk associated with this population and specific members?
- What are the rules for accurately attributing members to physicians?
Registry Challenges for Patients with Multiple Conditions
Grouping patients is complicated when they suffer from multiple chronic diseases, such as diabetes, heart failure, or asthma — all major focuses of population health management. One reason for this difficulty is the complexity of identifying patients who may not present with a primary complaint related to these chronic conditions.
For example, suppose the primary reason a patient is admitted to the hospital is for a total knee replacement procedure. Yet the patient also happens to suffer from heart failure and diabetes. The patient is assigned a primary diagnosis related to the knee surgery for billing purposes, but the physicians and coding staff may omit other diagnoses that aren’t relevant from a billing perspective.
Now, based on this patient’s primary diagnosis code, he appears in hospital data as an orthopedic patient. This presents a problem when the organization is defining its heart failure or diabetes population because the patient may be excluded from the other cohorts. Only clinical data offers the insight necessary to identify this patient as one who belongs in the heart failure or diabetes registry. This task is best accomplished through the establishment of an enterprise data warehouse (EDW) and the use of advanced applications for analyzing data.
Identifying patients who belong in multiple registries becomes even more difficult when attempting to create those disease registries across the full continuum of care. Yet, despite the many challenges, finding a solution is necessary in order to manage population health given the tectonic shift occurring in today’s healthcare industry and payment models.
Billing Code Variation Challenges
Currently, care settings have many variations with their billing codes and billing methodologies. These variations and the lack of clinical specificity attached to the codes makes it difficult to define specific