Understanding the 4 Steps of Data Validation
Data services is a complex and constantly changing field. Because of the constant change, data conversion firms and vendors are in high demand. As a healthcare system leader, it is necessary to understand the data conversion process, define your data details and expectations and ask the right questions up front to ensure the highest degree of accuracy during your conversion.
Of all processes within a data conversion project, testing and validation are always the tallest hills to climb. Below are questions you should ask and answer to ensure data validation go smoothly and efficiently.
Step 1: Detail a Plan
Creating a roadmap for data validation is the best way to keep the project on track. In order to create an effective roadmap and project plan, there are lots of questions that need to be answered.
- What is the overall expectation regarding the validation process?
- What benchmarks are in place to monitor project progress? Is there enough flex-time built into the schedule to allow an on-time completion despite issue mitigation?
- What issues exist in the source data that could bleed over to the new system? How will these issues be addressed?
- Will validation be used as a tool to correct data in the source before final conversion, or is data to be moved “as-is”?
- How many iterations of data validation are planned? If major issues are found in validation, is there a way to continue the conversion while remediating?
Step 2: Validate the Database
This step of testing and validation ensures that all applicable data is present from source to target. Consider it a basic reconciliation of raw numbers.
- What is determined: Number of records, unique IDs, size of data if no transform is used, comparison of source and target based on data fields, specific criteria (number of pages, documents, visits, encounters, patients, etc.)
- Common issues uncovered: Incongruent or incomplete counts, duplicate data, improper formats, mapping of fields between source and destination, “null” value fields
Step 3: Validate Data Formatting
When a transformation takes place within clinical systems, legal compliance is key. This can include maintaining page-breaks in text reports, inclusion of overlays for formatting of headers and footers, and verifying additional elements such as signature blocks and macro driven sections of text.
- What is determined: The overall health of the data in ability to be clearly understood by end users in the target system, and necessary changes are needed to ensure identical output in viewing and printing.
- Common issues uncovered: Page breaks, poorly-scanned original documents, placement of headers/footers when extracting text, missing text often caused by source system macros not populating directly to a text record.
Step 4: Sampling
Before converting data, you need to test it. But how much should you use for the validations mentioned above? Answer these questions to find out.
- Was sampling accounted for in your original plan? If so, what percentages of data should be used?
- If you have 1 million documents, what total percentage should be tested to ensure success from your institution’s standards? For the bulk of conversion, load-phase raw counts are best used to determine validity from a database perspective. Do you know of other expectations?
- What is an acceptable error rate for data? Whether it is due to error on source system input or source related issues, there will be errors to remediate.
- Can solutions be design to mitigate issues with custom data sets and increase success?
Have questions about an upcoming data extraction or conversion, or need support? Let us know and we can help.
Working your way through validation and need to start thinking about data testing? Check out our blog on planning for successful data testing.