When talking about clinical trial, what is the first impression in your mind? The following characters maybe the most popular ones: large amount of cost, long duration of investigation. As a clinical data manager, I would like to share the insights of the development of clinical data management.
Traditionally, data manager is asked to be focus on keeping data completeness, integrity and data accuracy. Actually, all these are factors are related with data integrity. Moving forward, people began to talk about data quality and some concepts were introduced as ALCOA and ALCOA+. “A” stands for attributable. “L” means legible. “C” is for contemporaneous. “O” means original and the last “A” stands for accurate. Still, these requirements can contribute to data integrity but not quality. For example, site really enter the data of smoking into database. According to source data, the subject consume one package of cigarette/week. PI correctly entered the data into database. This is fully compliant with ALCOA requirement. However, when we look into the data, every subject in this site took one package of cigarette per week. This looks strange as all subjects enrolled into this site with such a consistent habit of taking cigarette. After investigating the process of collection data, we found the PI always asked the subjects how many cigarettes you have for a week. Most of the subjects from the site mentioned he/she can not describe an exact number but they all confirmed they do not take a lot. Then, the investigator consider ‘not a lot’ equals one package of cigarette per week. From this example, you can see the data recorded on CRF can not exactly reflect the real situation. Even though they did not take a lot, they may not take the same amount and the data should not be entered as the same.
Then, how we can identify the high-risk situation? What we expected is the exact number instead of a general estimation. I believe that identifying this quality risk data is the direction for data managers. We should be not only focus on traditional data clean process to find the logical error, but also focus on the risk management. We need to change our mindset from following checklist or data validation/cleaning plan to be focus on data quality.
In the past, we put most of our efforts on data integrity checks. We setup hundreds of edit checks for each trial and thousands of queries automatically generated and answered. However, a research shows that only 3.7% of the data entered in eCRF was changed after entry in EDC with only 2.6% of changes not attributed to Source Data Verification (SDV). It seems a lot of efforts were paid but not result in finding out quality issue. This shows the importance of risk management to be applied to data handling. We need to put our efforts on key data and identify the high risk data.
Some methods have been introduced to explore risk. Reviewing data trend is a very good way to find the discrepancy. For example, a subject always got the same data for the systolic blood pressure and diastolic blood pressure all through different visits. You may need to check with the investigator why the data never changed. Does the machine show enough precision data? Is there any mis-practice to get the examination done? Another example is a site may get a significant high amount of PD cases reported. We may need to analyze with monitors to see whether all the PDs are related with the same process. Should we provide extra training to the site staff? For data managers, we should identify the data trends of the potential risk and also guide team to look into the high risk data, investigate for the root cause and then, get the solution.
For the new era of clinical trial, everyone will utilize the advanced technology to get the routine tasks done. People will have more time to work on complicated tasks and identify the risks that may impact on the quality. Getting quality data and managing the risky data will lead the new direction of clinical trials.
• A reflection Paper on the impact of the clinical Data Management—SCDM white page
• FDA Guidance and EMA Reflection paper on Risk‐based Monitoring (2013)
• RBM Interactive Guide.