Clinical Study Closeout Procedures and How They Vary
I recently presented an interactive discussion on study closeout procedures at the DIA Clinical Data Quality Summit. The format of the conference was to split the group into three smaller groups and have each group cycle though three sessions (study startup, study conduct, study closeout). This format offered an opportunity to see how different groups with similar backgrounds in clinical research would react to the same topic of study closeout.
Of all the 20+ topics discussed during the study closeout session, there are three that brought up the most discussion:
- Protocol Deviations – Who should capture the deviations, what is considered a violation/deviation and where does the information reside? A search of FDA warning letters brings up many occurrences of ‘protocol deviations’ and ‘protocol violation.’ Partial text from a 483 FDA warning letter reads:
Pursuant to 21 CFR 812.100 and 812.110(b), you are required to conduct your clinical investigation in accordance with the signed agreement and the investigational plan.
At least two subjects who should have been excluded from the study because they exceeded the Body Mass Index (BMI) criteria of greater than 35 BMI, calculated as specified in the protocol, were enrolled in the study.
If you wish to utilize exclusion criteria other than those specified in the protocol, you should contact the study sponsor before enrolling subjects under those criteria. Except in certain emergency situations, prior approval by the sponsor is required for changes or deviations from the investigational plan. FDA/IRB approval are also required if the changes or deviations may affect the scientific soundness of the plan or the rights, safety, or welfare of human subjects. 21 CFR 812.150(a)(4).
It is alarming that many companies are still struggling with determining how to spot potential deviations and violations. It was widely agreed that many deviations are found programmatically (out of window checks, out of range criteria, etc.), some are listed on the CRF in other places (inclusion/exclusion criteria, meds that may be disallowed, etc.) and some are found during monitoring visits.
It was agreed that all team members (clinical study managers, study monitors and data managers) should be responsible for reviewing data and determining if there are any violations/deviations, but there is no clear path for making these determinations.
Where should protocol deviations be captured? CDISC recommends not capturing protocol deviations or violations on the case report form (CRF). “The general recommendation is to avoid the creation of a protocol deviations CRF (individual sponsors can determine whether it is needed for their particular company), as this information can usually be determined from other sources or derived from other data.” (From the CDASH Standard v1.1).
The top three places that protocol deviations were documented are:
- Quality Assurance – In the days of paper CRF data collection, the QA group primarily focused on making sure the data in the source matched the data used in the analysis. Now that many clinical studies collect data in a variety of ways (EDC, EPRO, EHR, Central Labs/EKGs, etc.) the role of the QA audit is not as clear. During the discussions, attendees suggested the following items should be reviewed in a quality audit:
- Correct subjects were enrolled
- Randomization was handled correctly
- GCP guides are followed
- Informed consent processes was followed
- Accountability of study supply/test articles
- Data Review Meetings– The majority of the attendees conduct a meeting to review the data periodically. This meeting had several names:
- Medical quality review
- Blinded data review
- Cross functional review
Data review meetings are essential. They allow clinical study managers the insight to know how a study is progressing, ensure the correct data is being captured and make sure study closeout procedures are being followed. Some companies hold meetings involving all data periodically, some review only clean patients when 33% and 66% of patients are complete, and others report on only clean, completed visits. Other companies prefer to hold meetings with all data regularly (clean or not), and some only on patients with serious adverse events (SAE) and discontinuations. A few companies do not host data review meetings at all.
Typically, data review meetings consist of the following performed tasks:
- Aggregate view of the data
- Review of all tables, listings and summaries (on blinded data)
- Review of safety listings
- Review of coding
- Protocol deviation review
- Blinded data should be reviewed by all parties: DM, clinical, stats, regulatory, medical monitor
It was unanimously agreed that many of the issues we face during study closeout are because of lack of communication between departments. An example would be the statistics group may not see the data until the study is almost complete, or the data management group may not have input to the protocol.
How do you handle study closeout procedures, including protocol deviations, quality assurance and data review meetings? Are there any tasks that you perform that are not listed here? Please submit your comments below.