Did you know 77% of the cost to develop a 510(k) device is spent in clinical research and regulatory submissions? It’s 80% for PMA devices.
“A lot of that cost is due to how poorly we handle data,” explained Jim Rogers, President of Nextrials, at the 10x Medical Device Conference.
Jim’s solution is “why didn’t we think of this before” simple: Re-purpose electronic health record (EHR) data as the source data for clinical research.
Jim: “FDA wrote a Guidance. It came out September 2013 to give people action items around using electronic source.” If you re-use the data you already have, you’ll:
- Eliminating duplication of data entry.
- Eliminating or reducing transcription errors.
- Remotely accessing your source data.
And Jim anticipates your question: “There’s so many EHR systems, how could we ever begin to integrate with them?” Watch the video for his answer.
Jim Rogers: Hi. My name’s Jim Rogers. I’m the President of Nextrials. We’re a software and services company focusing on clinical research. And today I want to talk about the cost of doing clinical research.
As you can see here, some pretty frightening numbers. 77% of 510(k) devices, 80% of PMA devices, that cost is spent on clinical research and regulatory submissions. And a lot of that cost is due to how poorly we handle data. Everywhere from how the site records the data, how the sponsor collects the data, how we clean up the data and how we verify the data.
So what we propose is to use electronic health records to increase the efficiencies. Use electronic health records as the source data. A lot of the data that you are collecting for clinical research is already collected in electronic health records. There’s a number of benefits to using electronic health records as your source data.
Eliminating duplication of data entry. Eliminating or reducing transcription errors. And then another key ability that this gives you is the ability to remotely access your source data. So now your CRA’s are doing it remotely as opposed to flying out to the different sites.
Now these bullet points did not come from the Nextrials Marketing Department. They came from the FDA. The FDA wrote a Guidance. It came out September 2013 to give people action items around using electronic source.
Now I’m talking primarily about electronic health records. But if your devices are feeding information directly into the EHR, this applies to you and this Guidance applies to you.
A couple of key points in the Guidance is that the FDA states that they promote the use of electronic source data. Now the FDA does not use the word promote very often. So that’s very key. Another key point, they don’t expect, they’re not going to assess whether or not the EHR systems are 21 CFR Part 11 compliant. So they don’t have to meet that criteria to be used for clinical research.
So let’s break this down. Just like the sites, Electronic Data Capture has done a lot of wonderful things for the industry. Primarily, sponsors are the beneficiaries of that. What we’ve done though is we’ve made the sites do a lot of data entry which is tedious and time-consuming. They also have to learn to use multiple EDC systems. And this creates a burden on the sites and studies have shown 20%-50% of investigators would do one study and then they’re done. It’s just not worth it to them. But if we’re getting the data out of the EHR systems, we’re reducing the amount of data entry we’re asking them to do, the amount of training on different EDC systems is reduced and the data collection matches the site workflow.
And this is something I didn’t appreciate until we got in this. So here’s an example. You may design a form that has 20 data points on it. Those 20 data points exist on six different EHR screens. So now they’re going back and forth between screens trying to fill out one form. But if we’re getting that data directly from the EHR system, it’s pulling across all the EHR data to fill out that one form.
Data quality, something we always struggle with and again, whenever you’re doing duplicate data entry and I’m preaching to the choir here, you’re going to introduce error because the way it works at the sites is they’re over here typing into the EHR system, they’re swiveling around on the chair and typing that information again into an EDC system.
Studies have shown that that creates an error rate greater than 4%. But if you’re getting the data directly out of the EHR system in a valid data process, now there’s validation involved with this. You can’t just hook up any software. I’m not going to gloss over that fact. But this has interesting survey results. 92% of the sites said, more than 80% of the data you are asking them to collect is already collected in their EHR system. And 70% said, every data point you’re asking us to enter for you, we’ve already entered into our own EHR system.
Source Data Verification, a very expensive part of clinical trials. If you can access the source data remotely, you’re much more efficient and you can do it at a much lower cost. And think about this, if you’re getting it directly from the EHR system in a valid data process, you can trust that data better. If your primary efficacy data points are coming directly from the EHR, if your device is then putting data into the EHR and now that’s extracted directly out for clinical trials, the data’s much more reliable.
Now there’s a lot of misconceptions about EHRs because people think about how they used to be and this crowd is a little smarter than most but I hear people say, “Oh sites, most of them use paper charts.” How many here have their doctor use electronic health records that you know of? Mine rolls in with a laptop so I know he’s using it.
People think they’re just building systems or they’re just big Word documents we place in a paper chart. That’s not true. A lot of the data is structured. Some isn’t. Some is unstructured. I know Libbe is going to talk about unstructured data. And then the other thing, people throw up their hands and goes, “There’s so many EHR systems, how could we ever begin to integrate with them?”
So let’s look at where we are today. 70%-80% already have EHRs. Greater than 90% by 2019. And this was driven mostly by the American Recovery and Reinvestment Act where the federal government spent $11 billion in incentives to drive EHR adoption.
So how does it work? From within the EHR system, the site coordinator pulls up a CRF. It pops up on our screen. The EHR system in the background will auto-populate all the data fields that we’ve mapped. Now the data that’s not in the EHR, obviously those fields are going to remain blank. Then the coordinator will fill-in the missing fields, hit submit and it’s now in the EDC system.
So the implementation of this, the way we went about this is we know to gain acceptance, we’ve got to do most of the work as a software vendor. So for the site, all they have to do is point their EHR system towards our servers and set a configuration for the study. We do the rest.
So every year for the last eight years, we’ve been going to a software validation conference called Connectathon in the dead of winter January in Chicago, now Cleveland and these are all the EHR systems that we’ve validated with. You see some big names here. Epic, Cerner, Allscripts. Together they represent about 60% of the hospital market.
Now, that’s great for data collection but our goal is to have a clinical study done where the site coordinator never leaves their EHR system. They never login to an EDC or clinical trial database. So we’ve done extensions to the standards and the standards were designed by the FDA, IHE which is the standards group on the EHR side and CDISC, the standards group on the life sciences side.
So we’ve built extensions. You could now enroll a patient from within the EHR system. Even if it’s a randomized trial, you can enroll a patient. All the edit checks you typically see in the EDC system were run in the EHR. You could pull up all your queries and go through them and resolve them. Again, all within the EHR system. And what we found is the problem is not, not enough data. The problem is too much data.
We did a study and we went in to collect all the medications that were taken 30 days prior to enrollment on the study. You pull up someone who’s been around for a while, there’s a lot of medications in there. So we built a way to filter and select from multiple data points. We also built visual cues so you can look at a form and see which data came from the EHR and what was manually entered. And then the last thing, we built a flag to tell you when
the data is mismatched between the EHR and the EDC.
How does this happen? You collect the data. Week later, the coordinator changes that in the EHR. This system will break up both data points and you decide which one you want to use.
So we did a quick pilot, 4 sites, 40 patients, all in Greenway EHR. 75% of the data was auto-populated. The sites love the experience.
So just to wrap it up, we’re always striving for faster, cleaner data. It’s supported not only by the FDA but the EMA which wrote a reflection paper for E Source. It’s implemented by multiple EHR systems. It reduces the amount of work that the sites have to do and it reduces overall study cost.
Thank you.
Joe Hage: Thank you Jim.
Recent Comments