II. Standards and Suggested Practices
E. Data entry
In general, there are two basic steps in the CTR data entry process (Figure 4). The first step begins with the transferof data by the CTR service provider to the health department. The specific steps in the data transfer process used by service providers depend on the format of the data to be transferred. If a hard-copy form is used for documenting client data, HIV test forms may be mailed or hand-delivered to the health department. Electronic data may be transferred via a secure, Web-based system (e.g., PEMS), or discs can be mailed or hand-delivered to the health department. Appropriate security measures must be taken during any data transfer.
In the second step, data from each service provider are integrated into a central database at the health department, where they are stored for management, cleaning, analysis, and utilization. If service providers send hard-copy CTR forms to the health department, these data may be integrated into the health department’s central database via key entry or scanning. If electronic data are transferred via a Web-based system or hand-delivered disc, they can be integrated into the health department’s central database. Although we use the general term data transfer, many Web-based systems allow service providers and health departments to electronically share data as the data are entered in the system. In these circumstances, a data transfer step is not required; rather, these new CTR data are immediately made available to health departments for integration into their central database.
CTR data quality can be influenced in the data entry phase—during the data transfer from the service provider to the health department or during the integration of data into the central database. In addition to the steps of data transfer and integration, data verification between health departments and service providers must be conducted to verify data accuracy and completeness, enter missing data, or correct data errors. This verification process may take place before or after data are integrated into the health department’s central database. Although standards specific to data screening and error correction are addressed in the data management and cleaning section of the 2009 QA Standards (Section F), these corrective steps should be conducted by health departments and service providers, as appropriate, during both the data entry and data management phases.
Figure 4. CTR data entry methods
Standard E1: Health departments should establish and communicate to service providers the requirements for data transfer and should develop a system for monitoring the transfer and the receipt of CTR data.
Health departments must clearly communicateto service providers the requirements for data transfer, including which variables should be transferred and the procedures and schedule for data transfer. Health department staff may consider the following:
- Clear expectations and agreements between health departments and service providers increase the likelihood of timely data transfer. If different schedules or procedures are required for different types of data (e.g., HIV-positive test results), the health department should clearly communicate these variations to service providers. Health departments should regularly remind service providers of the deadline for data transfer.
- If service providers send hard-copy CTR forms to the health department, staff should check each form to make sure that forms are not torn, folded, stapled, paper-clipped, three-hole–punched, taped, or wrinkled. The health department should have a system by which service providers can easily order additional CTR data collection forms.
- Service providers should have a log for tracking the transfer of CTR data to the health department. This tracking system could include the date of data transfer to the health department, the number of forms or records transferred, and the ID numbers of the forms that were transferred.
- Health departments should have a tracking system to document the receipt of CTR datafrom service providers and a clearly communicated policy for the timely transfer of data. Recognition should be given to service providers who consistently send their data to the health department by the data transfer deadline.
- When service providers are late in transferring data, health department staff (e.g., program managers, contract managers, coordinators) should follow up promptly to identify issues and if necessary, provide technical assistance.
- Web site with information on agency forms and materials (San Francisco) [cited previously as Resource 12]
Complete, updated list of CTR templates, forms, policies and procedures, and other materials that service providers need to fulfill their data requirements: a model of the kind of easy-to-access and easy-to-use information that health departments can make available to service providers.
Standard E2: Health departments should establish clear procedures for integrating data and timelines that are based on the health department’s method of data entry (e.g., scanning, key entry, electronic transfer).
Develop data entry procedures for integrating multiple data sources and formats into a single database without compromising the integrity of the data. Health department staff may consider the following:
- If CTR data forms are scanned, provide staff with procedures that clearly explain the steps of the process, from scanning through data integration into the health department’s central database. Errors can be detected and resolved as data are scanned and through the data verification and validation process. For example, during scanning, staff may notice a pattern of illegible handwriting, leading to problems with optical character recognitions (e.g., a “G” is read instead of a “6”). Verification and validation programs can be written to detect invalid associations between variables (e.g., ZIP codes and state). This guidance, which can be included in the protocol, should include detailed instructions about how to use the scanner and the scanning software, how to log scanned forms, and how to troubleshoot if scanner errors occur.
- If data are entered manually, keystroke errors can easily be made when translating data from the form into the computer. Data entry staff should become very familiar with data collection forms and may also benefit from checklists, variable guidance, and notes. Regular checks should be conducted to compare hard-copy forms with key-entered data in order to identify patterns of errors made during data collection or entry.
- If the database is composed of relational data sets, a linking variable should be associated with each record in the data sets (e.g., unique ID number) that allows records to be integrated into all data sets.
- Regardless of whether data are scanned, key-entered, or accessed electronically, a timeline should be established for integrating data from service providers into the health department’s central CTR database.
- Data entry protocol for HIV counseling information form (HIV6) (San Francisco)‡
Detailed instructions and graphics for CTR data entry staff of how to process forms, what information should be on forms, what errors can be found on forms, and what information should be entered in the database.
- PEMS user manual (CDC)‡
Manual developed to help PEMS users understand and use the application (inclusive of all major releases of PEMS as of April 2008, as well as associated patches up to 3.1); includes hyperlinks that allow the document to be used as an online tool.
Standard E3: Health departments should develop guidance for communicating with service providers to verify and update data and to correct data errors.
After service providers have transferred CTR data, health departments are responsible for communicating with service providers to reconcile any data errors that are identified and update data as needed (see the data verification loop in the CTR data life cycle). QA checks should be performed on CTR data that have been transferred by service providers in order to identify potential data errors and their sources. Errors may include incomplete or inaccurate data, such as out-of-range values, transcription errors, and internally inconsistent values. It may not be possible to verify and resolve all errors. Health departments may need to prioritize verification of data to ensure that QA is performed for critical data while balancing staff workload and capacity. (For a list of common types of data errors, and possible resolutions, see Appendix C.)
Health departments can use the following strategies to ensure a successful data verification loop with service providers:
- Provide clear guidance to service providers about requirements concerning incomplete or inaccurate data. This guidance should specify which CTR data errors will be returned to the service provider for completion or correction.
- Request corrections from service providers through error reports, and then track these requests to ensure that corrections are made in a timely manner.
- Monitor errors made by specific service providers to determine whether there are error patterns that might be addressed through training or technical assistance.
- Develop guidance for handling each type of CTR data error. For example, for some types of data errors, health departments may send data back to the service provider for correction or completion. For other types of data errors, health departments may use case-by-case discretion and either leave data as missing or impute values.
- Provide clear guidance about the process and a timeline for updating data records. For example, if a client was tested by a service provider and the data were submitted to the health department before the client returned to receive test results, what are the procedures for updating the information about the provision of results to this client?
- Assign service providers to a point of contact at the health department for resolving data errors in order to develop rapport and increase health department familiarity with the data quality practices and common errors of each service provider.
- Example of a CTR data error report
Example of an error report from the health department to the service provider (including errors identified on each form to assist in correction and resubmission) and a cover letter to service providers.
Standard E4: Computer-based systems used for CTR should have programmed data entry features to improve data accuracy and efficiency.
Computer-based reporting systems can be programmed to make data entry more efficient. Skip patterns and validation checks can prevent data errors and improve the quality of data. Skip patternscan be used for conditional fields (i.e., data allowed for entry in one field are dependent on a prior response) and can make data entry more efficient and accurate. For example, if a client responds that he or she had never received an HIV test, a skip pattern in the data entry system would not allow the date of the most recent HIV test to be entered.
Figure 5. Prepopulated fields
Systems can be programmed so that some data need to be entered only once. If the same data elements, such as client demographic information, are needed later to associate the client and the services provided to that client, the demographic information can be prepopulated to reduce duplication of effort in data entry.
Figure 6. Drop-down menu or closed-ended response formats
Drop-down menus or closed-ended response formats restrict responses to a list of options, as opposed to open-ended response formats or text fields, which increase the likelihood of errors.
Figure 7. Disabled fields
Fields activated through data entry
Disabled fields can be individually tailored to prevent entry of data into fields that are not relevant (based on information obtained from a client). For example, fields for additional risk factors are disabled until information about the client’s recent sex partners is entered.
Figure 8. Skip pattern
HIV test result not provided
“Date provided” field is skipped
Skip patterns are programmed to preclude data entry in illogical or irrelevant fields. For example, information on the provision of HIV test results determines the skip pattern of the additional fields into which data can be entered. When an HIV test result is not provided to the client, the “date provided” field is disabled, and data entry continues from the “If results not provided, why?” field.
Figure 9. Out-of-range error message
Validation checks can also be used to identify missing, internally inconsistent, or out-of-range data and alert the user to correct the error.
Special considerations for data entry in clinical settings
Developing procedures for transferring and integrating data from clinical settings may present challenges for health department programs and evaluation staff. In nontraditional clinical settings for HIV testing, such as corrections facilities and emergency departments, this issue may be particularly relevant. Clinical sites that work with health departments to implement HIV testing services may already use data entry and management systems or procedures that do not easily interface with those of the health department. In addition, health departments partnering with multiple clinical sites may find that each site has its own data collection or processing system.
Ensuring that these emerging partnerships will be able to contribute high-quality HIV testing data is critical to monitoring and evaluation efforts. Health departments may need to invest extra time and staff resources in the early stages of new partnerships with clinical sites to address challenges in developing the procedures and to develop site-specific data QA protocols and training to support the consistent use of those procedures.