Data Processing

Ravi Kainth

Data Processing
For a clinical trial to produce data for analysis, the procedures for the entry and the processing of that data must be well established. The purpose of data processing is to ensure that data are reliable, complete and accurate. Data processing includes: Receiving, Entering, Cleaning, Coding, SAE Reconciliation

Data Processing
The goal of data processing is to efficiently produce quality data for analysis. The Institute of Medicine defines quality data as data that supports conclusion and interpretation equivalent to those derived from theoretical error-free data

Data Processing
Throughout a clinical trial, each component and step in data processing should ensure that an appropriate level of quality is maintained. Per ICH E6, Good Clinical Practices: Quality Control should be applied to each stage in the data handling process to assure that data are reliable and processed correctly.

Minimum Standards
Utilize written procedures that describe the data processing steps and required quality level. Ensure that all procedures, guidelines, working practices, or references are current and available to employees for the tasks they perform. Provide descriptive guidelines for staff who write queries for investigator sites.

Minimum Standards
Apply Quality Controlto each stage in the data handling process to ensure that the data are reliable and processed correctly.1
An example of a quality control measure is to periodically inspect samples of data and take corrective action if errors are found. Data of poor quality (errors) are classified as being the result of poor processes.

Best Practices
Address the purpose, characteristics and complexity of each study in the data cleaning procedures. Define the list of critical variables (usually those of primary and secondary safety and efficacy and subject identifiers) before the data cleaning activities are defined. Monitor data processing production functions adequately to assure stable and desirable quality levels that are consistent with the needs of the trial. Provide easy access for all employees to the documentation for all current procedures, guidelines, working practices, or references for the tasks they will perform.

Best Practices
Employees at the same level as those who will perform the processing steps should test new CDM production processes by using the proposed working practices, guidelines or references in a test environment prior to production work being done. Provide CRF completion/data entry instructions and helpful tips for both paper and electronic systems. For EDC, such a document might contain information clarifying how to enter particular forms or information about data cleaning.
Example: For the purpose of data cleaning, if the patient withdrew from the study while on study medication, there should be instructions to look for an adverse event where the action taken on study medication is drug stopped. Please note; In a paper-based system, there also should be datareview guidelines that include issues such as that described in the previous example.

Best Practices
Inform sites of timelines for data entry, running entry checks, and replication of data. Establish database quality criteria. Establish a quality control plan with the study team that appropriately addresses the primary efficacy data.

Overview
The Data Processing covers the following: Data Receiving Data Entry User Manual for EDC Data Cleaning Documenting Data Changes Medical Coding Protocol Amendments

Data Receiving
Throughout the industry there exist a multitude of ways to receive data for clinical trials. (fax, mail, delivery companies, couriers, hand-delivered by monitors, entry over the Web, etc.) Regardless of the data acquisition mechanism, the process by which data are received, confirmed as received, and made available for data processing should be documented in sufficient detail so that the origin of the data and the data received is known.

Data Receiving
Data receiving procedures should facilitate timely, high-quality data processing. Procedures should be in place to ensure blinding of subject identifying information (i.e., name, address, social security number) submitted to the data center, unless collection of these data is provided for in the informed consent, protocol, and local regulations.

Data Receiving
Most reporting and tracking systems can be programmed to follow a patients progression through the study as well as predict the last patients final visit dates, which helps to monitor the timeliness of the sites handling of study data.

Data Entry
Data Entry is the process of trancribing the Clinical Data from the source document of CRF into a electronic Database. Data Entry Techniques: Independent double data entry with a third person (two people enter same data independently, while the third person resolves discrepancies between first and second pass entry)

Data Entry
Double data entry with blind verification (two people enter the data independently and any discrepancies are resolved during second pass entry) Double entry with interactive verification (second entry operator resolves discrepancies between first and second pass entry and is aware of the first entered values) Single entry with a manual review; and Single entry with no manual review.

Data Entry
Differences exist between the type of training required for heads-down data entry and single-entry used primarily in EDC. In heads-down entry, the skill emphasis is on the number of keystrokes made and specific training on the database, versus singe-entry, which requires skill emphasis on the understanding of the trial.
FDA and ICH regulations do not require double entry or any other specific data-entry process. The skill level of the resource and the time available should be contributing facts in choosing a data entry process.

Data Entry
Evolution of Data Entry and Data Processing More companies are conducting trials by EDC. personnel are trained on the specific EDC system being utilized in the study as well as the protocol and key data issues. In many EDC systems, there is a source document verification tick box or field that is checked by the monitor once source verification has occurred. The box or field becomes unchecked when data in the panel are changed.

Data Entry
The sites should be provided clear guidelines regarding the timing from a patients visit until data are entered into the e-CRF or onto the paper CRF. In EDC trials, since the sites are entering and cleaning the data, resources are necessary to manage the data. This may include -training the sites on the system, measuring site progress on data entry and cleaning, working through late forms and data discrepancies with the sites, looking at aggregate data for outlying sites, identification of data trends, coding and data transfers.