Population Health Preparation Begins with Data Accuracy

During the Jan. 11 Talk Ten Tuesdays broadcast, I talked about the five most expensive conditions and the clinical documentation integrity opportunities associated with them. My topic for this article is the connection between data and population health based on an article by Dr. Anil Jain in Hospitals and Health Networks titled At the Heart of Population Health: Data.

Hospitals, payers, physician practices, and other healthcare organizations collect a lot of data. That data is used by the Centers for Medicare & Medicaid Services (CMS) in the development of any new payment methodology. Organizations collecting data should assess its use and management within their organization. Why is the data collected? Is it being reported at some point?  

There are five key types of data:

1. Administrative data, which includes demographics and medical codes
2. Adjudicated claims data, which includes data that has been subjected to a series of edits
3. Clinical data, which is supplied by the electronic health record
4. Patient-generated data, which includes health risk assessments, information gathered via mobile apps, and wearable sensors
5. Unstructured data, which is also supplied by the electronic health record. Dr. Jain reports that approximately 80 percent of the data in the electronic health record is unstructured.

While there is not a gold standard for population health data, analysis of your administrative data is encouraged as a beginning step. Here are some other first steps to take in order to understand your data:

1. Admission source accuracy; this data field has become an issue with regulatory agencies. The admission source provides information regarding where the patient originates. Can you identify which document is used as the authoritative source for the information within your organization?
2. Discharge disposition/status accuracy; this data field can impact reimbursement as well as value-based purchasing. Some Medicare-Severity Diagnosis Related Groups (MS-DRGs) are impacted by the discharge status, such as in MS-DRGs 280-285, which vary based on whether a patient has an acute myocardial infarction and a discharge status of alive or expired.  Value-based purchasing captures cases with planned readmissions, which are indicated by discharge status 81-95, depending on the discharging institution. Do you have a coding policy that provides guidance for selecting the discharge status?
3. Coding accuracy; this factor is typically part of any organization’s compliance plan. Improving coding accuracy can increase any organization’s confidence in its data quality. Typically, a compliance plan includes a coding review on an annual basis. It is important to know where there may be knowledge gaps, as well as planned expansions for new medical services, to create an education plan.
4. Charge integrity; this is the accuracy of charging for services, supplies, and procedures for patient episodes/encounters. Charge integrity ensures that the chargemaster/charging mechanism is used appropriately and that no gaps exist. Since the Current Procedure Terminology (CPT) codes are updated annually and the Healthcare Common Procedure Coding System (HCPCS) are updated quarterly, a minimum of an annual review is recommended. Inappropriate charging, such as the charging of too many units of a medication, can be a regulatory audit focus. Missing items in the chargemaster/superbill can cause an organization lost revenue.

As confidence increases in the administrative data accuracy, further inquiries into other data types can move forward. Also remember that implementing good data practices now will assist in the transition to population health and ICD-11. The final beta version of ICD-11 is expected to be released in 2018, according to the World Health Organization (WHO). In the years to come, the focus will be on the data.