Understanding the 4 Steps of Data Testing
Successful data conversion projects are often a result of great planning, due diligence and clear expectations. However, when it comes to data validation and testing, things can get tricky. Everyone knows that projects dealing with mission-critical data involves attention to detail by the project team, but sometimes the need for is lost on some of the other parties involved in testing, including end users.
The questions and tips below provide background on the data testing process and will help you determine both the level of testing and amount of outside staff involvement needed for an efficient and accurate data conversion.
Step 1: Subject Matter Expert Testing
During this process, experts in both the source data system and the new data system test data and identify any issues they see, especially from a support standpoint. The subject matter expert testing will validate general criteria to be tested by end users and make sure that all necessary elements are included in the formal testing plan.
- What is determined: How the data move and formatting are perceived from a workflow, utilization, and process standpoint.
- Common issues uncovered: Faults in transference, scope, omission and logic.
Step 2: End User Acceptance Testing
End user acceptance testing occurs in three phases: small-scale, large-scale, and full-scale. This testing method uncovers data conversion issues experienced by end users. Upon successful completion, this iterative testing will determine when data conversion can start production data loads.
- What is determined: Utility of data from a user and provider standpoint in variable scenarios.
- Common issues uncovered: Data not present in a readily useful or efficient manner, data missing details or including incorrect details and iterative corrections needed to optimize the system.
Step 3: Testing Preparation
Efficient and accurate testing requires a solid testing plan that includes enough detail that users understand what is expected of them. Leaning on super-users for testing is one way to make testing more efficient, but sometimes that isn’t possible. Some other considerations when preparing for testing are:
- Do testers understand what is being testing? What criteria are testers evaluating (ex. MRN’s, encounters, documents, results)?
- Have you identified availability of testers and confirmed with their respective department manager? Will testing require more than one session, and has that been factored into the project timeline?
- If super-users don’t exist or are unavailable, does your plan account for time required to train testers?
- What is an acceptable error rate per testing element? How are issues and irregularities being recorded and shared with the project manager and team? Is focused retesting planned once issues are mitigated?
- Are your testing worksheets simplified and clear in expectations of the elements tested? Do they have a clear pass/fail field and a space for details for testers to record more specific information? Will you have enough detail documented during testing to create remediation plans for data that fails validation?
Step 4: Testing
Testing can be a long and grueling process that requires a lot of time and attention from a lot of individuals in your organization. Luckily, there are ways that the burden of testing can be diminished, resulting in more efficient testing and a decreased resource drain. Some easy ways to reduce the demands of testing are:
- Extract specific cases and documents that are abnormal. These could include summaries and details that are not commonly used and could lead to unnecessary issues being reported.
- Identify and exclude test patient or case data. Test patient data is commonly found in production environments. Identifying and excluding this data early limits the amount of data for testing and conversion.
- Identify and remove corrupt source data from extraction records. Mistyped and incomplete data are common, and past system failures in the source system could leave data sets missing or render them unextractable.
- Provide unique identifiers for each field to ensure elements are not repeated in your test plan. Without unique identifiers, it becomes more difficult to accurately catch and address issues.
- Include each format of inpatient and outpatient data, document types, results, and case types in every sample testing phase to aid in issue identification tied to a specific type of data.
- Plan for data cleanup ahead of conversion that includes the Master Person Index (MPI) and reconciliation of source data.
Data conversions are never a walk in the park, but by having a solid plan and clear expectations, you can help mitigate issues before they arise and create a more efficient process. Testing is also simplified when data validation goes smoothly.
Have questions about an upcoming data extraction or conversion, or need support? Let us know and we can help.