This situation could arise in a variety of health care delivery contexts. In this particular case, a large national company manages a Risk Evaluation and Mitigation Strategy (REMS) for a manufacturer, as mandated by the FDA. The backbone of the program is a large SQL database, through which new data flows in and data output is generated. The HSM Group performs regular and critical data quality checks on both the incoming and outgoing data files.
The database includes contact information, identifiers, relationships, application information, and enrollment information for physicians, hospitals, and clinics. To ensure data quality, HSM runs pre-defined quality checks to spot data errors and call out data fields that should be cleansed. Because the database is continually evolving, HSM performs data quality checks every other week as well as additional quality checks on an ad hoc basis.
To perform the quality checks, HSM uses powerful SAS data management and statistical software. More than 150 data checks are performed each time a new dataset is created. The data quality checks are important since a missing value or a wrong format may cause an entire hospital to receive drugs for which it is not approved. Some of the common data checks performed include:
- Required fields: Some data fields should be populated for all records while others are allowed to have missing or null values.
- Data formats: Many data fields adhere to a specific format such as numeric, text, and date.
- Out-of-range values: Some data fields should only contain a limited range of numeric values or allow only specific strings of text.
- Relational database checks: Some of the data tables have a one-to-one relationship with other tables meaning that, for every unique identifier that exists in one data table, there should be a matching record in another table.
- Rule checks: Checking data rules is one of the most important aspects of the entire process. Data fields may be dependent on the content of other data fields, requiring complex scenarios to be evaluated quickly. SAS data management and statistical software enables this rule checking.
You probably know HSM as an exceptional partner for providing high quality, custom quantitative and qualitative research. For more information about our strategies for managing “failed” quality checks, data quality plans for Excel files, tracking errors and call-outs, and tips for optimal communications and data/process organization, please contact HSM’s project leader Brett Plummer, PhD, at email@example.com or 480-947-8078, extension 313.