Optimize internal and external data sourcing to meet business requirements, apply quality considerations, and manage data provisioning agreements consistently.
Most organizations acquire data from either external or internal sources. It is frequently found that the performance of sources (e.g., the timeliness and completeness of data provided) does not meet the organization’s expectations. Similarly, data quality can be sub-optimal for the organization’s needs. The questions for the organization in this process area address best practices for: defining data sourcing requirements, acquiring and providing data that meets quality expectations, managing agreements, and interacting with providers.
Effective management of providers helps assure quality data delivery, assists with selecting the service providers for the organization’s needs, and improving the cross-organization match rates for patient identity when data is shared. It is recommended that service level agreements (SLAs) be established by the receiving organization, with both external and internal data providers, specifying what data will be provided, the quality rules it should conform to, and performance thresholds (i.e., minimum expectations for timeliness, completeness, providing updates, etc.). It is also advised to specify in the SLA or contract that the vendor must provide an organization its own patient data at no cost for purposes of analysis.
Data sourcing decisions need to reflect the needs of data consumers to ensure that requirements can be satisfied with appropriate data sources, whether internal or external. This process ensures that delivered data will meet business needs. An organization should determine parameters against which alternative sources will be compared and selected. This will help the organization discover if data elements provided by a source need to be combined (e.g., Patient First Name and Patient Last Name combined into Patient Name in the destination), split (e.g. Patient Name split into Patient First Name and Patient Last Name), or have business rules applied (e.g., new data element Mother Maiden Name) to satisfy business requirements. Mapping of a data source (i.e., data elements provided by the source, mapped to the destination data store(s)) is enabled by precise business terms and definitions, data profiling analysis results, as appropriate, and clear data requirements.
For patient demographic data, the receiving organization should include in the SLA that a data provider should engage in bi-directional communication when patient identity matching cannot be accomplished. The provider should communicate a report that contains the number of matching failures per upload, accompanied by delivery of the records that were involved in the failures. In addition, a data provider should communicate the algorithm(s) that they employ for matching, with an accompanying list of data elements employed. Mandatory versus optional data elements should be included, as appropriate.
Data quality criteria (e.g., thresholds, targets, and quality rules) are integral to data sourcing decisions and should be used to compare sourcing alternatives and develop SLAs with the selected data providers.
Assessing the quality of sourced data over time should be based upon metrics (e.g., number of records provided with no birth date, etc.) that are established through data quality processes (See the Data Quality category) and performance statistics, such as data error handling, provider responses to requests, and issues escalation. Data integrity can be enhanced by implementing appropriate controls for authorization and access (See Data Standards).
Finally, the organization should regularly monitor delivered data, at a frequency commensurate with its criticality, to ensure that criteria continue to be satisfied. To resolve issues and work with providers to improve data delivery, it is useful to know how providers produce and deliver their data. This knowledge is best acquired and applied prior to connection; it will assist the organization in working with the provider to correct errors, apply additional quality rules as warranted, and thus increase consumer confidence in the data.
An organization that implements sound provider management processes will realize the following benefits:
Data requirements define the data attributes needed to implement a new process, including formats, ranges, and other critical metadata. Sourcing specifications define how data providers will be selected and provide existing authoritative data sources. Likewise, it should provide guidance on selecting an authoritative source when none exists.
Example Work Products
Data quality rules and thresholds are defined in terms of data quality dimensions (See Data Quality Assessment). These requirements should be included in SLAs by key consumers with guidance and oversight from data governance.
An SLA specifies delivered services, service measures, levels of acceptable and unacceptable service, and expected responsibilities, liabilities, and actions of both the provider and customer in anticipated situations. Data governance stakeholders should monitor assessments to ensure the results are in line with the agreements.
Example Work Products
Defects and deviations should be defined within SLAs in terms of data quality dimensions. For example, if a 2-digit month, 2-digit day, and 4-digit year with seperating dashes (##-##-####) is required for Date of Birth, then 09/14/1969 while in a legitimate format, would be considered a defect. Were the date entered as 09/14/69 it could be counted as 2 defects (i.e., / vs -, 69 vs 1969).
Deviations occur when defect rates exceed defined thresholds. For example, a 20% defect rate in formating would violate a 90% threshold by 10%. The process for holding a provider accountable for defects and deviations should provide guidance on the prioritization of defects and their resolution.
Accordingly, critical defects might be defined as any defects beyond the threshold defect hurdle rate. The process may further specify critical defects to be resolved within one hour up to the point where the defect equals the defined threshold. Then additional defects should be adressed after other critical defects in other dimensions are resolved.
Example Work Products
1.1 Has the organization applied requirements for patient demographic data to data sourcing specifications and agreements?
2.1 Are data quality rules and thresholds embedded into service level agreements with data providers?
3.1 Is there a process to hold providers accountable for defects and deviations from data quality thresholds defined in the service level agreement?