Is the $300 million interactive voice response system (IVRS) industry ripe for change? Is it failing to keep pace with the timelines of industry? Is it possible to transfer randomization to the web and have the customized nuances embedded in the system?
Yes on all counts, says Ed Tourtellotte, proprietor of the 35-person Tourtellotte Solutions. We last profiled the company about a year ago. His new randomization software is called Trident IWR; you can read more about it here online and in this press release.
As befits someone who drives a Maserati, Tourtellotte prefers speed to torpor. Ease over delay. And he says flatly that existing companies in the IVR space are not able to handle all the work being directed their way. He prefers the "IWR" designation because of the inevitability of the web being used to randomize patients.
“We’ve taken the 12 week study setup time and turned it into 12 minutes,” says Tourtellotte. “This is how you should do IVR or IWR. The web is way better than the phone. We are blowing people’s hair back,” he adds, in a reference to the loudspeaker ads of another generation.
The combination of IVR and electronic data capture (EDC) has been explored by a handful of technology providers, but most sponsors seem to prefer a separate, standalone randomization system. Tourtellotte’s Wayland, Massachusetts company offers a new twist: randomization that can be driven by (and affect) drug supply tools from the same firm.
Drug Supply Simulation
There is no question that Tourtellotte is handling IVR and drug-supply consulting projects for prominent sponsors that don’t allow their names to be used. And there is also no question that the dependability of randomization is sufficiently important that the firm will have to prove itself to overtake the current established names in the IVR space, whether that's ClinPhone, Almac or Icon.
Tourtellotte says the whole present approach to IVRS is antiquated. “It’s error prone. It’s clunky,” he says. “The power of the business experts who interpret the protocol ends at the limits of the specifications they write.”
The history is interesting. Along the way to refining tcVisualize, his supply simulation software, Tourtellotte recognized he already had most of the pieces of an IVR. So, as a typical tinkerer, he set about to build the rest. And to figure out a way to get the randomization and simulation systems to talk to each other, using an XML data export. So now both Trident and tcVisualize can use the same XML file format.
That will let sponsors use their supply simulations to configure their randomization. Similarly, actual data from the randomization engine can be used to further refine their simulations of supply. The hope is to avoid any drug supply overage that may not be completely necessary, and to anticipate out-of-stock situations long before they arise.
Can IVR Scale?
Tourtellotte opines that the leading vendors of IVR solutions, while doing their best, find themselves in a tough spot. Doubly so. The current IVR systems are cumbersome for vendors to build, with programmers disappearing for weeks or months after the business requirements have been spec'd out. “All the legacy IVR vendors are trapped,” Tourtellotte says. “The IVRS industry is under phenomenal pressure. You can see major outages. They can’t scale. They’re going crazy. The scaling is the key.”
Traditional IVR systems are also unpleasant for users at clinical sites, Tourtellotte says; clinical research associates don't like punching in letters and numbers in a telephone keypad.
Under the hood of many IVR systems, he insists, there is at least one more vulnerability. A different database schema is often used for every trial, meaning that it’s hard to leverage work from one trial to another. Tourtellotte started with a blank sheet of paper and figured out that perhaps 80 percent of the IVR setup process could be managed with pick-lists, drop-down menus and other conveniences that (he stresses) do not require any professional software experience. Much of his code can be reused.
And his IVR system can be set up by ordinary clinical operations people, folks who could not write a line of code to save their lives. Says Tourtellotte: ”Business users set up the data and administer it using switches and knobs—and not programming.” The elements that can be configured with such parameters include the basic study design, drug kits, visit schedules, dosage rules, formulations and dispensing schedules.
What about the rest of the 20 percent of the requirements? Such stuff may be coded by the customer, or by Tourtellotte’s programmers. It’s here that he feels the Trident software shines. “No one else has figured out how to architect custom-pluggable code,” says Tourtellotte. He says even with custom coding for the last 20 percent of the process, it will still be far faster than existing techniques from other suppliers. Indeed, he goes so far as to say the present IVR approach is past its prime. “It’s a dinosaur,” says Tourtellotte. “It needs to be extinct.”
He’s often asked if the system supports voice, i.e., whether a phone keypad could be used to randomize patients at sites that have bad internet access. The answer is: yes. But he sees an eventual transition away from the phone, thanks to user preferences and the rise of electronic data capture.
“People hate the phone,” says Tourtellotte. “I spend a lot of time railing against voice, and then I say, ‘we’re going to do it anyway.’” He says few protocols in the age of electronic data capture really require voice; most sites are online and prefer to randomize patients using the web.
But Tourtellotte does appreciate that a few sites in the Borneo jungle or the Siberian steppes may not have broadband internet access, or have ever seen a computer. On such occasions, Tourtellotte believes, there may be alternate methods to bring in the data: perhaps just dedicating a person or help desk. But he’s as resigned to voice as suppliers of electronic data capture are to paper.
Adaptive, Minus Hype
It’s not necessary or obligatory to combine tcVisualize, his drug-supply simulation system, and Trident, the new randomization tool. But Tourtellotte is convinced there are significant opportunities for efficiency for companies taking that approach.
The idea is that systems for drug supply and randomization should be in sync given that that is the case as a trial unfolds. When Tourtellotte hit Trident’s “play” button, the spectacle was as far from Microsoft Excel as Microsoft Excel is from an abacus. It’s very clear what supplies are where. “I’ve got an instant global picture of my trial. I can simulate forward given my actual state,” says Tourtellotte. “The more data you have, the better your forecast is. If you’re going to change something, you can add that to your forecast.”
The technology could facilitate adaptive designs, Tourtellotte says, though he quickly concedes there are not many such protocols under way. Says Tourtellotte: “There is all this hype around it. I don’t want to slam adaptive trials. Bad designs are flooding the FDA. There is some risk that the adaptive paradigm could fail. There are some adaptive designs that are phenomenally great. I want to help it get there.” His firm is working with Cytel, as we wrote here a while back.d9A2t49mkex