Characteristics of High-Quality Data

High-quality clinical research data have several interrelated characteristics. Accuracy means that data correctly represent what was observed, measured, reported, or documented. Completeness means that required data are present or that missingness is clearly documented. Consistency means that data do not contradict related information elsewhere in the dataset. Timeliness means that data are collected, entered, reviewed, and available within expected timeframes. Validity means that values conform to predefined formats, ranges, and rules. Reliability means that measurements are reproducible under comparable conditions. Integrity means that data are protected against unauthorized or undocumented change.

These characteristics should not be treated as abstract ideals. They have practical implications for every part of a study. A temperature value of 370 degrees Celsius is invalid because it falls outside a biologically plausible range. A participant recorded as male with a positive pregnancy status may indicate inconsistency, a data entry error, or a special clinical circumstance requiring clarification. Validation reduces preventable errors and improves the overall quality and reliability of the clinical database.

Completeness ensures that all required data are captured. Where data are missing, the absence should be clearly documented and justified. A missing primary outcome threatens completeness.

Timeliness refers to the prompt collection, entry, review, and availability of data within defined timeframes. Delays in data entry or reporting can affect monitoring, decision-making, and overall study progress. A laboratory value entered three months after collection may raise concerns about timeliness.

Data interpretability and usability are critical in practice. Even when data are accurate and complete, they may still be difficult to understand or analyze without proper documentation. A key tool that supports this is the data dictionary, which defines variable names, formats, coding schemes, and allowable values. While a data dictionary is not itself a data quality characteristic, it plays a crucial role in enabling consistency, validity, and accurate interpretation of data. Without it, datasets may be difficult to use, share, or reproduce, even if the underlying values are correct. A dataset exported without a data dictionary may be difficult to interpret, even if the values themselves are accurate. Quality is also contextual. Meeting technical quality standards alone does not guarantee that data are suitable for analysis. Data must also be collected and managed in a format that supports meaningful interpretation and statistical evaluation. A variable may be complete but not useful if it is collected in a form that does not support the analysis. For example, recording antimalarial treatment as a free-text field may capture information, but it may not be easy to summarize unless coding rules are later applied. Similarly, a database may have strong validation rules but still produce poor data if staff are not trained, source documents are incomplete, or workflows encourage users to bypass required checks.

The characteristics of high-quality data are closely interconnected. A weakness in one characteristic often affects others and can compromise the overall quality of the study database. High-quality clinical research data are accurate, complete, consistent, timely, valid, reliable, and maintained with integrity. Together, these characteristics ensure that clinical trial data are fit for purpose, support participant safety, comply with regulatory requirements, and provide a robust foundation for scientific and medical decision-making.

Table 1.3: Characteristics of High-Quality Clinical Research Data