Interview With CDISC Device Team Co-Founder Carey Smoak

15 May, 2016, by ClinCaptureTeam

Carey Smoak is the co-founder and co-leader of the CDISC Device team. He has authored and co-authored more than 30 papers on the topic and has done numerous presentations at PharmaSUG, SAS Global Forum, the Bay Area SAS Users Group, the Bay Area CDISC Implementation Network, and other venues. Carey Smoak has a master’s degree in public health and currently works for Portola Pharmaceuticals as a Director of Statistical Programming.

ClinCapture : Could you please tell us about yourself and your history with CDISC standards?

Carey Smoak: I started my career working as an epidemiologist in various medical schools, and I made the switch to working for biotech and pharmaceutical companies as a SAS programmer in the mid-1990s. I’ve worked in the pharma, biotech, and medical device industry for more than 20 years, and I have been involved with CDISC for the past 10 years. I co-founded the medical device CDISC team, but I’m also a part of the submission data standards team for CDISC.

ClinCapture : Tell us about the regulatory pathway to make CDISC standards required for FDA submissions.
Carey Smoak: A couple of years ago, in December 2014, the FDA put out the requirement notice that clinical studies for drugs needed to be submitted in the CDISC format: The official deadline for this requirement is the summer of 2016. Sponsor companies now coordinate with the FDA ahead of time regarding their specific projects and requirements in order to be ready when they start their study, and ultimately when they get to the data submission phase. Drug developers should also be aware of the information that the FDA maintains on their website about the latest applicable standards. For medical devices, it’s a little bit different as device standards are still under development. We have now developed therapeutic domains for devices and we are close to having a controlled terminology for medical devices. In order words, we’re completing the requirement to having an established set of standards for devices. As far as FDA regulations, the CDRH main set of guidance follows a 5-year cycle, and the current guidance does not include any requirements about CDISC standards. The Center for Devices and Radiological Health (CDRH) is the branch of the United States Food and Drug Administration (FDA) responsible for the premarket approval of all medical devices, as well as overseeing the manufacturing, performance and safety of these devices. At this point, it’s debatable whether or not it will get into the next CDRH cycle, so we might still be a few years away to fully required CDISC standards for devices, but there is a lot of discussions indicating that it will be required in the very near future.

We have now developed therapeutic domains for devices and we are close to having a controlled terminology for medical devices.

ClinCapture : Who are the biggest advocates of CDISC standards: Sponsors, vendors, or the FDA?
Carey Smoak: It is definitely a push from pharmaceutical and biotech companies. Most drug developers actually had to come up with their own standards, which was a very natural thing to do. A report written in the mid 2000’s, the Gardner report, showed that by implementing standards, developers can save a significant amount and time and money. So from the perspective of sponsors, having standards really does make a lot of sense. It starts with Case Report Forms (CRFs), the setup of the database, setup of edit-checks, standardization of programs, and outputs… for collecting, managing data and reporting clinical data. Today, most companies are interested in standards but the adoption seems to be the hard part. They seem to want to follow CDISC standards, but also maintain their own standards. Some companies call it CDISC+, where they take CDISC standards, and add on to the existing framework. Then when they submit the data to the FDA, these sponsors just take away their company’s standards.

ClinCapture : Let’s talk about CROs and technology vendors, what should sponsors look for when selecting a drug/device development partner? What are the key criterias to verify that vendors are knowledgeable on the topic, and CDISC compliant?
Carey Smoak: Working with CROs is a different type of problem. I find that CROs sometimes have their own standards as well. It is better for sponsors to establish CDISC standards internally and then push them out to the multiple CROs instead they’re working with. That way, you have a much better chance of having something that’s compatible across trials and stakeholders, which is important if you’re doing an integrated analysis. In my experience, pushing the standards that you have established to CROs get more consistent results.

ClinCapture : Let’s talk about cross company collaboration: Could CDISC standards allow companies to exchange data or collaborate on different de-identified sets of data?
Carey Smoak: Common standards and domains will certainly make this a lot easier. That is actually one of the reasons the FDA is so interested in standards. They could run an analysis, let’s say, across a group of companies that are producing beta blocker drugs, which would let them look at much larger sets of data. This could empower them to identify rare adverse events for a certain type of blood pressure medication for example. My impression is that the FDA does not have enough data in CDISC quite yet to do that kind of meta-analysis with common standards.

Actually, as much as 40% of data used in clinical trials come from sources other than EDC systems.

 

ClinCapture : Are CDISC Standards also helping for interoperability between multiple systems, such as EHR and EDC systems?
Carey Smoak: Indeed, there is a lot more data out there than just the clinical data collected for clinical trials on case report forms. Actually, as much as 40% of data used in clinical trials come from sources other than EDC systems. For example, from central labs, but also from other sources. You may have a lot of data that are images, which means there are lots and lots of non-standardized data out there and in electronic health records. The FDA and sponsor companies have ways to access that data, for example you may need to go into electronic health records to gather more information when a serious adverse event occurs. Electronic health records could also be a quick way to collect a large amount of data instead of entering it into an EDC system. If you have an electronic health record with age, race, sex, height, weight, vital signs, and more, why not use it as a way to capture the data without entering it twice. Another area that is also growing is the realization that clinical trials do not represent the real world. Clinical trials are based on highly selected sets of population, and that is how we decide to get drugs approved for the entire population. There has been more and more talks about how to leverage big data and data science within pharma companies to look at other types of databases. For example, a database that collects hospital admissions could be mined and become a very interesting part of clinical trial data.

ClinCapture : Who would have the authority to mine that data, or at least oversee permissions to mine data across private and public sources?
Carey Smoak: It definitely goes against the traditional way we think of the EDC as a validated database. In the real world, we may be looking at data in a non-validated way. It is still unclear what the regulatory requirements should be here, as well as the technology, but it certainly is a area of need, because clinical trials do not represent the real world. As drug developers, that is very much in our interest to see how things are out there in the real world.

ClinCapture : Thank you. Is there anything else about the CDISC world that you would like to add?
Carey Smoak: Yes, the CDISC world really is trying to step up. In the early years, there was just a base model of domains that covered traditional things like medical history, vital signs, etc. We standardized the things that were easy to standardize across most common areas. What is very exciting is that CDISC is now at the cutting edge with therapeutic user guides. There is about 40 of them, and they’re trying to address specific therapeutic areas like cardiovascular disease, Alzheimer’s disease, etc. Two years ago, when I was at the CDISC interchange, I participated in a team that designed standards for CRFs to study Ebola. It was nice to sit and help researchers collect data for such an important problem that is Ebola research. I think CDISC is really becoming able to solved medical problems in a faster way, and to respond to crisis.