A Holistic Approach to ICD-10 Testing for Healthcare Providers

A Holistic Approach to ICD-10 Testing for Healthcare Providers

A lot has been said about the many challenges of upgrading to ICD-10; in fact, it’s often considered the Y2K of healthcare. This white paper presents a unique approach to leverages a Golden Data Set of dual coded medical and claims records and Test Automation Accelerators, to deliver high quality end-to-end ICD-10 testing

By: Yaron Kottler, Subba Jevisetty and Tom Parnon

A lot has been said about the many challenges of upgrading to ICD-10; in fact, it’s often considered the Y2K of healthcare. From a testing perspective, the challenges can be summarized as follows:

  • Multidisciplinary skills required for successful testing including skills in IT, revenue management, clinical documentation, medical coding, and more.
  • An unprecedented number of parallel system upgrades leading to a substantial need for both testing specific to ICD-10 as well as extensive regression testing.
  • Scheduling and resource availability at multiple partners: payers, clearinghouses and HIEs.
  • Availability of accurately-coded test data.
  • The need for test data synchronization across provider and payer test environments (e.g. members, plans).
  • Depending on test approach and test data approach, the need for member information de-identification to avoid HIPAA issues related to PHI data.
  • Revenue impact testing resulting from DRG shifting.
  • Testing to identify Clinical Documentation Improvements (CDI).
  • Defining the correct approach for High Volume end-to-end, internal, and external testing.

There are two distinct approaches to addressing these ICD-10 testing challenges, both relating to the source of the test data to be used. One approach is to use crosswalks to map ICD-9 codes to ICD-10 codes. The alternative is to use a test data repository that includes dual-coded medical records along with test automation accelerators to load the test data.

Though the crosswalk approach is more widely known, it suffers from the following limitations:

  • Only about 5% of ICD-9 codes map directly to ICD-10 codes
  • All crosswalks are software programs that make assumptions about how to map the remaining 95% of codes that don’t directly translate and correlate
  • Crosswalks generate problems whenever their mapping algorithms do not match a particular provider’s environment
  • Crosswalk software is costly

ICD-10 testing that employs dual-coded medical records along with accelerators to efficiently load the test data into the provider’s systems is a superior solution. When executed correctly, providers can expect the following benefits:

  • A complete end-to-end testing solution which covers internal and external functional and business process testing, revenue impact testing, CDI testing, coder readiness testing, etc.
  • The only approach that enables accurate revenue impact testing
  • An approach which supports the organizations CDI and coder training initiatives
  • The most efficient approach to testing resulting in dramatically fewer false negatives and non-bugs
  • The least expensive approach to end-to-end ICD-10 testing often costing less than half vs. other approaches.

The rest of this White Paper describes a real-life step-by-step example tracing data flow through a typical ICD-10 end-to-end test process.

Testing Approach

A complete End-to-End ICD-10 testing solution should include the following components:

  1. Internal Testing: how the information flows within the provider system, as well as ensuring that ICD-10 codes are processed correctly from beginning to end within
  2. External Testing: how transactions are sent and received between external trading partners, such as payers, clearinghouses, billing services, etc. through channels already in use
  3. Revenue Impact Testing: ensures the processing of data, either ICD-9 or 10, does not negatively impact revenue generation, and that there are no vast differences in the revenue generated between ICD-9 and ICD-10 data
  4. End-to-End Business Process Testing: making sure the full spectrum works together and meets the expectations of everyone involved (from doctors and patients to billing services and clearinghouses)
  1. Risk Management: mitigate the highest risk to businesses, such as in reference to revenue or ease of operations
  2. Clinical Documentation Improvement testing and validation: because ICD-10 can sometimes require more coding, this validates the appropriateness of, identifies deficiencies in, and recommends improvements to the new coding


The most efficient way to accomplish these tests is by using Golden Data, a set of dual-coded (in both ICD-9 and ICD-10) medical records, and automated accelerators. The Golden Data is peer-reviewed and validated by clinical/coding experts. The purpose of the automated accelerators is to improve the speed and accuracy of data entry without the need of human input, resulting in faster and more efficient testing which is reflected in higher quality and greater monetary savings.

The overall steps of the testing process should be as follows:

  1. High risk scenarios from the impact analysis will be used as the basis for test strategy, planning, and test design
  2. A test plan is created based on the high-risk scenarios
  3. Test data is created from the dual-coded clinical data repository based on the high risk scenarios
  4. Test design and test scenarios will be created using the high risk scenarios and the corresponding clinical data
  5. Data entry design will be done by pairing patient and partner (plan) information with clinical data
  6. Data entry automated accelerators will be customized for provider’s specific business rules and processes
  7. Clinical data will then be entered using automated accelerators into the provider’s front end systems/applications
  8. Remaining test execution will be done by executing the workflows through back end systems via billing, claim creation/submission, and clinical data warehousing.
  9. The output ICD-10-based claims will be verified for accuracy.
  10. For external testing, both ICD-9 and 10 claims will be submitted and the payments/remittances will be verified and compared for revenue impact due to DRG shifts.


A successful risk-mitigation strategy is one that selects scenarios for testing based on a combination of the following risk categories:

  • High volume
  • High cost/revenue
  • High complexity, or likely points of failure
  • Anticipated opportunities for improvement of existing processes
  • Known to be problematic today (under/over reimbursed)


A high-quality test database and adequate test environments need to be planned and managed throughout the test execution.

This risk-based test strategy, combined with a dual-coded, medical data-based testing approach, is very cost effective, efficient, and holistic. A high level end-to-end testing approach is depicted below.

This approach is detailed in steps with illustrations below.


    • Identify high risk scenarios from production ICD-9 claims using impact analytics data





  • Select clinical data records from the clinical data repository for code which corresponds with both ICD-9 and 10


  • Enter selected clinical data through medical records systems like EPIC, Cerner, etc. by associating with real production patient demographics, if test patient demographics data is not set up across the end-to-end process between providers and payers.

    1. The bulk of the test data will be automatically injected using automated accelerators
    2. A small percentage of test data will be entered manually by looking at the clinical data record information
  • Execute the workflow through billing and claim creation
  • Verify actual output 5010 claims (837) records against expected 5010 claims(837) records
  • Resolve the differences and report the test results 
  • External and Revenue Impact Testing: Claims will be created using both ICD-9 & 10 data and the claim reimbursements/payments will be compared for neutrality and variance. Test patient demographics will be replaced with real production patient data (optional).
    Asynchronous testing – same claims can be sent to multiple payers by automatically replacing with real patients enrolled in each specific payer.

Stand Alone Revenue Impact Testing: Since the Golden dataset contains 5010 claims (837) in both ICD-9 & ICD10, revenue impact testing can be started concurrently with other testing cycles as early as vendor products and payer system are ready. ICD-9 and ICD-10 5010 claim reimbursements/remittance advices (835) would be compared and resolved for unexpected variances.


Clinical Documentation Improvement (CDI):

Dual coded medical records can be helpful in CDI in the following ways:

  1. Current production medical records will be dual coded along with documentation and coding improvement suggestions. This will help the staff understand the problems with their documentation/coding practices and also to learn how to document and code correctly.
  2. Provider’s ICD-10 readiness can be assessed based on their documentation/coding practices.
  3. Provider’s staff can be trained in documentation and coding for medical records.


The changes which are coming will undoubtedly prove to be unique and challenging for everyone involved: hospitals, individuals like doctors and patients, insurance providers, clearinghouses and billing services, and even testers like us. The best way to manage these changes is to embrace an approach based on Golden Data and Test Automation Accelerators. This is the best way to ensure higher quality and can provide savings estimated at over 50%.