In this webinar, SCDM invites CRF Health to discuss the principles of managing eCOA data in clinical trials. We will use experience from real life studies to illustrate how to handle issues clinical data managers typically encounter. Participants will increase their knowledge of how to collect high quality eCOA data and how to conduct an eCOA trial successfully from a clinical data management point of view.
[webinar begins at 00:27]
Hello everyone. Welcome to this SCDM webinar on Managing eCOA Data - Principles and Best Practices. My name is Dianne. I am the Educational Project Manager at the SCDM headquarters in Brussels, Belgium.
So just a few housekeeping notes before we start. Kindly make sure that you’re on mute throughout the presentation. There will be a 60-minute presentation today followed by enough time for your questions. There is a question-and-answer box on the right side of your screen, so please enter your questions there for us to respond to later. If you encounter any technical issues or if you have any questions during the session, you can use the same chat box, and I will try to help you as fast as possible. And finally, please do not worry too much about taking notes. The recording of this session and the slide deck will be shared with you afterwards.
Now I’m very happy to be joined today by two presenters: Amy Bradley and Rauha Tulkki-Wilke from CRF Health. I will now pass the floor on to them, to start this presentation by telling us a bit more about themselves and then develop today’s topic.
Thank you Diane. So this is Rauha Tulkki-Wilke. I’m the Vice President of Product and Service Management at CRF Health. I first came to the industry around 16 years ago, when I joined CRF Health, and we started developing what were then called electronic patient diaries, for collecting primarily patient-reported outcomes in clinical trials. And during this time, it’s really evolved into eCOA, that we’ll be talking about today, so electronic clinical outcome assessments, which is a much broader concept. So electronic tools for collecting outcomes data has been my day job now for about 15 years. And I’m based in Finland, in a technology center at CRF.
And I’d like to hand over to Amy for a brief introduction and then we’ll jump into the session.
Thanks Rauha. My name is Amy Bradley. I am the Manager of Global Data Management at CRF Health. I’ve been with CRF Health for two years, and prior to CRF Health I was with Icon Medical Imaging. So in some way or another I have been involved in data collection for about 12 years. But with eCOA it’s only been the last two years. So I’m very happy to be joining Rauha today for this presentation.
Next slide please. Thank you. So in today’s session, we put together learning objectives for the audience of what the session will give you today. So one of the most fundamental ones is to understand what eCOA is, and I’m sure there are plenty of you who know that. But let’s go through that, so we all have the same basic understanding. And how eCOA is captured. We’ll also understand the principles and the regulations guiding eCOA data management and perhaps eCOA capture in general. We’ll understand the typical scenarios encountered by clinical data managers and learn methods to identify and handle issues in eCOA data.
And the structure of today’s webinar will be the following. So we’ll first take a look at the eCOA basics to make sure we have the same baseline to start with. And then we’ll get into the regulatory framework to see what are the basic principles, the basic regulations and concepts that we are working within. Managing eCOA data from perhaps more of a content perspective, what’s included, what are the specific considerations and activities. And then after that, jumping into more like the operational side of things which is to do with planning data management during study set-up phase as well as working on a trial and study close-out. And last but not least, we’ll walk you through some of the typical scenarios to prepare for.
So let’s start with what electronic clinical outcome data, that is eCOA, are. And it’s really an umbrella term that covers four different types of outcome assessments. There is the most typical one, which is the patient reported outcomes. So this is information that comes directly from the patient. In a typical scenario, this would be collected on a diary at home or using questionnaires during site visits. Then another form—and this is an area that is increasingly being captured electronically today—are clinician-reported outcomes. So this is information about the patient that comes from someone with specialist training who is observing the patient and discussing with them. So, for example, an oncologist or a psychiatrist or nurse or other specialist role. Then, the third category are observer reported outcomes. So this is information about the patient again, but comes from someone with no specialist training but who knows the patient really well to report how they are feeling and functioning. Typically these—both clinician and observer reported outcomes—are relatively objective, to ensure that the patient’s subjective view is not obscured by someone else. So observers are typically parents, caregivers, or other people close to the patient. And then the fourth category are performance outcome measures. So this is information about the patient’s performance. So a good example of this would be the six-minute walk test where the patient is performing something and it’s being measured.
Now, how eCOA is being captured, we typically talk about modalities. So modalities basically mean the different ways, different modes of data collection that can be used. And now, we put there first, to be paper. And obviously it’s not an electronic means. But it’s to highlight that, still today, most of the outcomes data collected at clinical trial is actually collected on paper. And we’ll talk a little bit in a bit of time about what that portion of data is. There are a multitude of electronic modes that can be used to collect this data electronically, and typically the choice of the mode would be done based on the actual needs of that specific study. And from an eCOA vendor perspective, and so from the sponsor perspective, every study is always unique and it has unique characteristics. So it really makes sense to make this choice, with consideration. So the typical modes are using a hand-held or a mobile device—a smartphone basically—that the patient takes home with them from the study site or uses their own phone for it. So there is essentially an application on the smartphone that reminds the patient to go in, enter the data in the study-specific schedule. Then tablets are used as well. Tablets are mostly used at sites to collect site-based assessments. So when the patient comes in for a visit, they’re filling questionnaires on a tablet device, and from the table device they are sent onwards to the database. Web is one modality. So online assessments, whether the patient fills them in at home or at the study site, web is a very uniform modality in that sense, and it’s mostly used in post-marketing studies, as it’s very flexible and easy to operate because it doesn’t include any hardware. IVRS is also a modality especially used for infrequent data collection nowadays. The patient calls to your phone number, and there’s an audio of the questions and then a recording of the answers through the pad of the phone. Digital pen is also a modality, so using a digital pen on a special paper to record answers. Apps are a form of hand-held mobile device and smartphones where it basically means that the diary is an app that the patient can install on their own phone. And wearables are something new, especially for capturing the patient’s performance, for example their activity levels during the study.
So the benefits of collecting the data electronically are pretty numerous. So data management, for the audience, these are obviously very logical and natural. So with the power of technology we can put in edit checks, automatic skip logic to ensure that questions are only displayed depending on response to a previous question are shown. There can be automatic score calculations for questionnaires that include a score, or score calculations that can feed into inclusion and exclusion criteria and randomization. We can integrate with medical devices to pull in meaningful other data. So for example, spirometers for, for example, PEF values in asthma studies or glucometers in diabetes studies to get the patients’ blood glucose levels into the diary and ensure the patient can augment those into specific events. We can automatically flag unusual scores, outliers, and overall have a real-time view to the data that’s being collected. We can also do computer-adapted testing with eCOA, as well as integrate additional sensors and other potential data points. For example, we can integrate data, let’s say for example endoscopic scores, to ensure that we get a full picture of the patient’s status. And overall we get better quality data and higher compliance, or at least the compliance is more accurately known, because we can see when the patient actually filled in the data.
So as I said, the use of eCOA is increasing really really rapidly. So about two years ago, we first researched this, and the penetration within clinical trials, within phase 2, phase 3, and phase 4 trials was close to ten percent. Today, eCOA penetration is approximately 20% in phases II-IV, and increasing. It’s increasing at a rate of approximately 20-25% a year. So our estimate is that by 2018, it will reach something like 40-50%.
If we go to the next slide, we’ll see the therapeutic areas where eCOA is being used. And it ties together with the penetration to a certain extent, because there are therapeutic areas where, for example, primary end point data is collected with outcomes end points. And there, using the electronic methods is more widely recommended as well as accepted, whereas some therapeutic areas are just coming along with the most significant driver there being eCOA being the more efficient way of actually collecting the data. So if we look at these therapeutic areas where we are seeing a lot of data being collected electronically already, respiratory, pain, vaccines, neurology, oncology, and GI.
So as you can see, these are slightly different from one another. Respiratory pain as well as GI are typical home-based diaries, where the patient is entering data on a daily basis into a daily diary. And in these therapeutic areas, they were the ones where eCOA adoption started approximately 15 years ago, and they are still leading in terms of adoption. Then vaccines are slightly different. There is a daily diary but it’s primarily safety data that’s collected through electronic means. Not at all less important, but slightly different. Neurology and oncology are the ones that have come on more recently, but even there the use is pretty broad today. So as you can see, overall, there are a lot of different therapeutic areas that utilize eCOA methods today. And this is likely obscured view, because it’s from our database, but I think it still gives an indication of use of eCOA across therapeutic areas.
Now, a typical eCOA system would consist of three different parts. It all starts with design, so almost every eCOA system comes with a design tool. And the design tool is then used either by the study team or together with the study team by the vendor to design an eCOA solution for a specific study, pulling the instruments from a library, making sure they fit into the study flow, being able to simulate that and make sure the solution fits the study needs. Then there are collection methods, whether those are devices, online apps, or IVRS, there are different collection modalities that this design can be installed to. And then there is a management tool, so after the data has been collected to the study database, it can be managed through a web portal, as well as exported to other systems, as well as integrated in real time if necessary to other systems. Also data management is usually done through the web portal if it’s not done in an external system.
So just to summarize the eCOA basics. There are the different types of outcome assessments: patient reported, clinician reported, observer reported, and performance outcomes. And the benefits of eCOA really centre around leveraging the power of technology in terms of guiding data collection, controlling data collection, producing more powerful calculations, getting higher quality data with better compliance. And there are a multitude of electronic media types: hand-held and mobile, tablets, online, IVRS, digital, and apps on patients’ own phones, and wearables and other medical devices. We are expecting eCOA penetration to reach around 40-50% by 2018. And it’s already used in a wide array of therapeutic areas, we are only expecting that to increase going forward. And when you are taking and use an eCOA system you will most likely see it consisting of three different parts: a design tool, collection methods, and a management portal.
So that concludes our eCOA basics section. And we are ready to move to the next one, which is the regulatory framework.
So I’m sure many if not all of you are familiar with the ALCOA principle. But it’s pretty significant within eCOA. So let’s go through it relatively quickly. So essentially, regulatory agencies, and FDA specifically, require that data is being collected in accordance with the ALCOA principles. So ALCOA stands for attributable, legible, contemporaneous, original, and accurate. And let’s go through these in more detail.
So attributable means that whatever data we collect, whatever eCOA data we collect, we can always attribute it to the actual person who entered the data, whether they are a caregiver, clinician, or a patient, they’ve logged into the system with their own credentials. We always know who entered the data. It’s not a question which trial the data belongs to, we will always know that through system access. We will know which site the patient belongs to, for whom that data is being entered, and we can always track it down to the data point level. So all of this is visible through audit trails and through the data records in eCOA systems.
The second principle is legibility of the data. Is it clear enough to read? And only actually on paper this is a significant question, especially on patient diaries, you can’t really limit the patients from entering anything on paper. So sometimes that has resulted in entries that are not particularly legible or are difficult to put into a database. So eCOA data in general is more legible, as it’s collected via a system that guides and controls the data entry. We have drop-down menus to ensure that we don’t have incorrect entries. Even if free text fields are being used, they are legible in the sense that the text is entered through a keypad. Multiple choice questions, radio buttons, all of these different data entry methods, all guide the data entry to a direction where it’s easy to read and accurate.
The third principle is that of data being contemporaneous. So basically, answering the question of was it recorded as it happened. And this is extremely important in electronic clinical outcome assessments, and especially for patient-reported outcomes, which are typically collect at home between visits. So eCOA systems are built from two perspectives. Firstly, they prompt subjects to enter the data at the right time. And they also prevent patients from entering the data at the wrong time. So most of the systems deploy very specifically defined data entry windows for specific time points. So for example, if there is a morning diary and an evening diary, there is a window of time to enter the morning entries and there is a window of time to enter the evening entries. And in order to reach high compliance, the system generates alerts and reminders to subjects, and they also generate alerts of missing data for site personnel and sponsor personnel to make sure they can reach out to the patient if it seems to be a pattern, to educate them and simply remind them of the importance of entering data.
The fourth principle is that of data being original. So is it the first place that the data is recorded? With eCOA, data is collected right at source, so an eCOA entry is typically the earliest record that’s been documented. And the great thing with this is that it eliminates any transcription errors, as well as the burden of transcribing the data, and increases the data quality.
And finally, the accuracy of the data. Are all the details of the data correct? And this is a two-fold question. Part of the accuracy comes from the fact that it’s been collected with very well defined instruments that cannot allow entries outside of the range. But part of the accuracy of eCOA data is about well-designed instruments that patients understand consistently and correctly. So that’s an important part of getting accurate data to ensure that patients consistently understand the instruments and especially the questions correctly. And the system makes sure that the entries are complete and compliant.
Moving on from ALCOA principles to the question of who controls and owns the data. So an important concept is that sites control eCOA data, and per regulations they must be in control of the source data at all times during the trial. This means in practice that they will need to have access to the data through the web portal at all times. And the eCOA web portals always give the ability to the site to access all the data, including audit trails, even though most of the time a site would access aggregate data through reports to see if there’s something they need to do. But it’s important that the eCOA web portal allows that access to the finest detail to all the data that the patients have entered. It also means that we need to ensure that—and eCOA systems ensure this—that we need to get sites approval for data changes and sometimes this includes pushing back on sponsors on what data changes are acceptable. And an important part of being in control of the data is that there is a robust audit trail for data changes.
Now, given all of this, it’s good to recognize that sometimes technology is challenging for sites. They may not be that technologically savvy. They might be at times uncomfortable with eCOA technology and may need training. But with proper training and experience, sites have demonstrated that they can use eCOA really well.
So one of the regulatory concepts tied to the previous slide is that the eCOA vendor is a Trusted Third Party. So in the FDA guidance for industry on patient reported outcome measures, that FDA released in December of 2009, they state that sponsors should not have exclusive control over ePRO source, and it must be controlled directly by the investigator or under their control via Trusted Third Parties. And within the set-up where there is a sponsor, a CRO, an eCOA vendor, and a site, the eCOA vendor acts as the Trusted Third Party that the investigator has delegated the hosting of the data for the duration of the trial.
The flow of the source data is also an interesting concept. And this was obviously when we started collecting eCOA a while back, the question of how you define the flow of source data and the path of source data in the system throughout the course of the trial. So this is what the industry concluded on. So the data is entered usually by hand to a diary. We call the diary a transient data collector. And it’s migrated onwards from the diary to the eCOA vendor database. And this migration is encrypted, it goes through the public internet usually, so it needs to be encrypted. And at the time when the database confirms that the data has been moved over to the database, it’s considered that completing the eSource migration to the database. Now, data is usually also left on the device, to a certain extent, to ensure that the rules and the calculations run properly on the diary. So for example, if there is a, let’s say, compliance calculation for the last seven days, or there’s some other logical functions that depend on the data previously entered, that data is left on the device to ensure those calculations and rules work. But otherwise, the actual source data is considered to be migrated to the database. And that’s happening until the end of the study. Now, in the end of the study, there are two different migrations that happen. The first one is that to the sponsor database, so we would migrate the eSource to the sponsor database via a CD or a validated data transfer. And a site-specific eSource copy is migrated to each investigator to the site, to ensure they can also hold the source at the site for inspections and other purposes.
So summarizing the regulatory framework. The regulators will look to answer that the eCOA data meets the ALCOA principle. And that should typically be of no real concern, as we have good best practices for eCOA and systems, generally speaking, ensure that this always happens. They will look to confirm that the investigators or Trusted Third Party have control of the data. And it’s important to understand the source data movement during the trial. So from the patient, through the transient data collector, to the eCOA database, and finally being migrated to the sponsor database and to each participating study site to the actual investigator.
So that concludes the regulatory framework section. And at this point, I’d like to hand over to Amy Bradley, who will talk about managing eCOA data in practice.
Thank you very much. So when considering data management with eCOA, a lot of the data management issues that you face are similar to what you would face when collecting data on paper, but also eCOA also introduces some other reasons why data management is a critical piece to the data collection process. And this slide just talks about some of the reasons why the data management function is so critical with eCOA data collection. And so challenges might result in data quality issues that require cleaning and data corrections, and that’s not unlike something we would see with data collected on paper, but for the same reasons, the challenges presented by eCOA will result in the need to clean the data and update the data. Sites and patients can make mistakes. And so that’s another reason why we would need to manage the data, in an effort to correct those mistakes. And certain types of issues require different handling and identification of the issues. So depending on the design of the project, there may be a need to monitor, identify, and handle certain issues, and that may need to be done on a study-specific basis. And also eCOA is relevant for central monitoring. So the information that’s collected electronically, the availability of that data is much sooner than what we would see on paper. And so that lends itself well to central monitoring. The data is sensitive to site performance, it’s collected continuously and with high frequency. So this provides a strong indicator for risks.
So when we’re talking about data changes with eCOA, eCOA is often the eSource. And it’s designed to record the data at the time of collection. Missing or incorrect data cannot be recovered. And so therefore data management typically focuses on preventing, stopping, and handling issues that cause missing or incorrect data. Data changes are primarily limited to administration data such as subject numbers. Structural data, such as visit numbers or study periods, objective or evidence data such as the number of pills or a.m. or p.m. Subjective patient entries are not modified. For example, pain scores are missing entries, so anything subjective coming directly from the patient, those entries will not be altered at all.
And as the data typically is end-point or safety data, unaddressed issues might have severe impact to the trial validity and results. So it’s important that the review and monitoring of the data is taking place consistently throughout the project so that the delivery of the data is as clean as possible.
So managing eCOA data overview. So there’s three phases really, when we’re talking about eCOA. The set-up portion of the project, the data capturing and processing, and then finally study closure. So when you’re talking about set-up, the important thing to understand and the important things to consider are the instruments that will be used for the project, how they will be presented to the patient or the clinician or the observer, how the design of the study and the device will handle that instrument. And then you need to consider the data structure, so the programming of the information in the device, and how the data relationships affect other parts of the database. And then of course you need rules and definition around the data that’s being collected, and how that will translate into the deliverable to the customer. And then for data capture and processing, this is where we’re talking about central eCOA monitoring. So device management, so that’s talking about the device inventories, which sites’ patients have the devices that are being used for the project, sponsor oversight and trial management. Then for active data management, this is where we’re talking about reviewing the data that’s been collected and identifying trends. One primary example is if there is possibly a non-compliant patient or a non-compliant site, this is when we would want to identify that issue and take steps to address the issue and then prevent it from happening again. And then data cleaning, issue identification, resolution, residual effect handling, and DCF processing. And we’ll talk a little bit about this further in the upcoming slides. And then finally, data finalization, QC, and documentation. And then study closure, data and documentation delivery.
So device management. In addition to managing inventories—so you have devices with the eCOA vendor, you have devices at the various depots, the sites, and with the patient—but there’s some other issues to be considered when talking about device management. Sometimes you may see that there may be multiple devices for the same patient. So when you’re looking at a database, you might see that Patient 1 at Site 1 has data in two different devices, so that’s something that happens not uncommonly. Most of the reason is because the patient lost the device so they needed to be re-issued a second device. Another device management scenario is a battery drain can cause time synchronization issues with the database. Or we might see that devices are in incorrect status, so a patient or a site forgot to move a patient from a screening period into a treatment period, for example. And use of patients’ own devices may result in different data issues, for example inaccurate time stamps is another factor to be considered when talking about device management.
So active data management. It’s important that this is considered throughout the course of the trial and not just in the beginning or not just in the maintenance phase or not just at the end, but from beginning to end it’s important that the active data management function is being performed. Proactive management of the data ensures data quality, completeness, correctness, and consistency and integrity. Active data management is tailored to each trial focusing on specific trial characteristics. Final use of the data, so one project may require certain data management functions that are different from another, and that is dictated by the final use of the data or how they are going to use the data after it’s delivered. Typically, active data management will include protocol compliance, eligibility, trial stage shifts, subject identifiers, redundant data, missing data, data transfers, common repeated issues in data cleaning, and patient reporting patterns. And as I mentioned earlier, data is continuously monitored to identify and address issues using interactive reports and analytics across the eCOA database and external data, for example IVRS data and EDC data.
And so some of the typical scenarios or issues that you will see are non-compliant patients, missed visits, devices in incorrect status, duplicate patients, PIN security issues, study-specific issues, and monitoring trends. So these are the high-level main categories that you will see when you are managing data throughout the course of the trial. And it’s important to note that the data manager can analyze the clinical data to find trends and issues and be able to highlight them for the study team and vendor for resolution.
So cross checks between data sources. So the IVRS is a master data source for most of the administration and structural data. However, additional data may be needed from an EDC or less common sources, such as lab scores. And usually batch transfers that are run just before data is monitored. Validated system set-ups done during study set-up. This would require sponsor or CRO to manage the vendors involved. So if we needed to, for example, cross-check our eCOA database with an IVRS database, we would need for the sponsor and the CRO to facilitate that exchange. And real-time integration is used more frequently. So for web service-based, that real-time integration is used more frequently.
So when we’re talking about data changes, or data clarification forms, the issues or discrepancies can really be identified by the eCOA vendor, can be identified by the sites, or the sponsor. And the eCOA database will allow the creation of a data change form. So these data change forms can be created by any of the parties that identified the need for the change. But the important piece here is that the site has the ability to approve those changes. And where appropriate, the sponsor will have the ability to approve the changes as well. And then once those approvals are obtained, the eCOA vendor can update the database with the requested change.
So this information is taken from our own database, but I expect that it’s probably consistent across the board. So obviously the most common changes that we see are changing form data, and by that I mean administrative data, such as visit numbers for example. As mentioned previously, we would not be changing any subjective data entered by the patient. But that is largely the majority of the data change forms that we see. But we also see changing subject information, adding forms, removing forms, merging devices, changing patient status, removing a subject, and changing a site. When I talk about removing a subject, we’re not actually removing a subject from the database, but the patient or the subject will appear as “marked as removed” in the database. So the information always stays within the database, but it will have an indication of “marked as removed.”
So for monitoring the data, there will likely be some normal typical reports that you will see, such as compliance reports, visit schedule, period status, subject administration information. But a project may also require some unique reports that address the needs for that particular project. And it’s important that these are identified in the data management plan, so the data manager needs to have a really good understanding of the design of the project, what is being measured, what some of the important areas to monitor are, based on the protocol. And all of that should be defined in the data management plan. It’s important that the reports that are used are built to highlight issues that need attention across the database. So again, it’s important that the person authoring the data management plan has a very good understanding of the project and the information that’s being collected and what it’s going to be used for, so that they can define what types of reports will be needed to monitor that data.
So when we’re talking about managing eCOA data, the key things to keep in mind is that there needs to be a method to resolve challenges with the eCOA data that is collected. eCOA allows for effective risk-based monitoring. Data management typically focuses on preventing, stopping, and handling issues that cause missing or incorrect data. Device management and active data management comprise the majority of the data changes. And reports are utilized to monitor particular areas of interest, such as compliance rates.
So that concludes the managing eCOA data section. And so now we can move on to planning data management and working on a study.
So when you’re planning data management, the discussions around data management for a project should start at the outset. The initial discussion should talk about the overall data and activity architecture, protocol design, and operational considerations. And so again, probably at the kickoff of a meeting, this is when these discussions should start. In the set-up phase, the end-to-end data flow should be reviewed to ensure an optimal set-up and design to proactively prevent issues. So the design should really take into consideration what the end use of the data is going to be. And the design should also accommodate an effective way to handle and prevent issues. And this serves as a basis for defining and planning the active data management to be performed in the trial to manage issues. It’s important to note, though—and we’ll talk about this in subsequent slides—that the planning of the data management sort of is an ongoing process throughout the trial. But the important part here is to note that it should begin at the outset.
So we talked about the data management plan a little bit ago. We just wanted to point out some important things that need to be taken into consideration when writing the data management plan. And some of them are the checks. What checks does the customer need to be flagged throughout the course of the trial. How often. A very well articulated definition of the checks needs to be included in the data management plan to prevent any misunderstanding. The query method. Earlier there was a slide about who can create DCFs, or data change forms, who can approve them. That will need to be defined in the data management plan as well. How soon should those queries be approved and implemented? What is the escalation procedure for unresolved queries? It would be important to include all of that information within the data management plan. As far as definitions, what does “non-compliance” mean in this project? So that definition will change from project to project. Is time synchronization relevant for this project? So many many times, time synchronization is not an issue in a project. But if it is a factor, that needs to be handled up front and a method for handling that needs to be defined in a data management plan. If there is anything that is excluded that should be important to note, that should be also noted in the data management plan. And then the frequency. How often will checks be performed and the queries rendered? That should be defined in the data management plan. And also, this is above and beyond the reports that we talked about earlier, those will also be in the data management plan.
So we sort of went through when the discussions for and data management needs are discussed at the outset. And now we’re looking towards the end of the project and we’re approaching database lock. So some of the things that need to be considered here are that the devices that are being used for the project are returned, or if not returned, their current disposition is approved. For example, the records may show that a patient lost a device. So that disposition needs to be approved. So the important thing is the location of the devices are either returned, or their disposition is approved. The data sending is disabled. So what we don’t want is additional data to be sent to the database. So we need to lock the database in a particular time frame so that data sending disabling can be done via the eCOA vendor. We want to make sure that all the data changes are complete, so there’s no additional changes taking place to the database. And that that final data transfer after those three previous steps are complete, that data transfer is accepted by the sponsor, and the sponsor has given us approval or given the eCOA vendor approval to lock the database. So there’s a few key things that need to be ticked off before the database is locked.
So now that database is locked, we can talk about source data archiving. And some of the things to consider before we archive the data. And these should have been addressed in the previous step but still you want to make sure that the data cannot be sent from the device and data changes cannot be processed. So what we’re ensuring here is that the archiving of the data, or the data that is archived is consistent with the data that was sent in the final transfer to the customer. The study documentation is all finalized, all the signatures are obtained, all of the documentation is in their finalized status. And archival timelines—that should say method and formatting requirements—are agreed upon. So the customer should know how long it takes to archive the data, when the sites can expect to receive their archived data, and what format the archives will be in.
So we use something called a data migration plan. But something of that sort would be appropriate to define all of these things. So this data migration plan would talk about exactly what is contained within the archives, who the archives are delivered to, both at the sites and at the sponsor, how will the archive be labelled, and this document should also talk about the final decommissioning of the project, what the timelines are for decommissioning, and when the decommissioning can begin.
When we’re talking about the actual archive of the data, so each site that participated on the trial will receive an archive of their own data, they only receive their own data, they won’t receive other sites’ data. The sponsor archive will contain all of the sites’ data. And when I talk about sites’ data, I’m talking about every data point that was collected, so all of the questions that the patient answered, all of the audit trail. So you will be able to see who entered the data, when they entered the data, if the data was changed who changed it, when they changed it, what the variable or information was before the change and after, and all of the change requests. So if a change was made because of a data change form, that will also be contained within the archive. Typically the archive is saved to a medical-grade read-only CD or DVD. And a quality check is performed against the data contained on the media against the database. So a check to make sure all the patient data came over, a sampling of the forms and the data fields is performed to make sure that nothing funny happened when the data was archived onto the media.
So planning and working on a study in summary. The important points to take away from this section are that active data management starts at the beginning of the set-up with a solid data model planning and data management plan. Data is monitored via reports and cleaned via a defined process with proper authorizations for data changes. And archiving is done after database lock, producing one archive for each site, and a sponsor archive with all of the sites’ data.
So some of the typical scenarios that you’ll see when managing data for an eCOA project, these are the most common scenarios. And these were mentioned earlier. But missing data, so you will be able to see that a patient has not sent data, or a patient is experiencing technical issues preventing them from sending data. So this will be evident in the database, and this is something that can be flagged and corrected as quickly as possible. I mentioned a common occurrence is seeing a patient with data on two different devices. And this can be remedied by a data change form, where we can merge the data from both of the devices. Or a diary in an incorrect status, so a site or patient forgot to move the patient into the new status. So you will be able to see in the database that data should be in a different status, and that can also be corrected right away. So those are the most common scenarios that you will see with managing eCOA data.
So that is the end of the data management presentation for eCOA. So I will pass it over to Dianne.
[Q&A starts at 59:02]