The need for speed

Clinical trials of investigational new drugs have always been both time-consuming and very costly. Proving the safety and efficacy of new treatments requires years of testing and billions of dollars. And, while testing is under way, the patent clock is ticking such that the patent protection duration may be reduced by half or even two thirds before regulatory approval is granted.

Increasingly careful regulatory oversight, and consumer expectations have driven the industry to clinical trials that require ever more time and expense. The standard practices for management of clinical trials, however, date from an era of handwritten Case Report Forms (CRFs) and manual entry of data for statistical analysis.

Consider the standard process. Clinical investigators fill in 10s of thousands of paper CRFs, each costing several dollars just to print. These CRFs are then transferred to the sponsor. The data are entered manually into computer databases and then re-entered for verification. About five percent of the entries are questionable and generate a so-called ‘query’, each of which requires a follow-up phone call or even a site visit by a monitor, to resolve. Every change must leave an audit trail to help ensure data integrity. Typically, each CRF page costs US$20-30 to verify.

This process continues, typically, for three to six months after trial completion until all questionable entries are resolved. Often, a questionable entry may involve a datum recorded many months before any question of its integrity was raised. If the datum in question involves a qualitative assessment such as ‘level of pain’ or ‘degree of anxiety’, then resolving an old query may become nothing more than an educated guess.

The patent clock looms in front of all clinical development. When patent time expires, competitors are free to market the same chemical entity, while having incurred only a minute fraction of the clinical expense borne by the patent holder. Clearly, sponsors have a powerful interest in conducting trials faster. If speed costs more money, it may be worthwhile nevertheless.

‘Haste makes waste’ or ‘he who hesitates is lost’

Speed is often expensive. Consider fuel economy, overnight mail or rush orders at Photomat. But consider as well jumbo jets, e-mail and digital cameras. The newest jets get about 100 passenger miles per gallon at 500 knots, e-mail is nearly instantaneous and free and a digital camera can print pictures instantly for the cost of the paper. (Ironically, the time to print a digital picture is less than the time needed to develop an old Polaroid ‘instant’ picture).

A new technology can change the paradigm. Often, it requires a change in thinking, it may shift expenses from labor to capital, and most critically, it may require new skills. Dale Earnhardt, Jr. drives a car at 200 miles per hour better than just about anyone else, but most would rather that a different person was piloting their jumbo jet. Postal carriers are reliable and hardworking, but no use at all in connecting a cable modem.

For e-mail to be useful and ‘free’, an enormous amount of capital for file servers, high speed transmission lines and software was expended. E-mail is free only after a few trillion dollars was spent on global infrastructure and PCs. What technology in clinical development can have the same magnitude of impact as do PCs, e-mail and instant cameras?

EDC to the rescue

Electronic data capture (EDC) provides a parallel in clinical development. EDC replaces paper CRFs with a computer program containing heuristic edit checks – an eCRF. Investigators still enter clinical data, but now into a computer that has been programmed to detect the most obvious errors. Common errors, such as violation of inclusion criteria, values grossly out of range, inconsistent or missing data are easily detected by software. The investigator is compelled immediately to correct the error or to verify the value out of range. No query raised, no phone call, no site visit by a monitor, no manual entry of the investigator’s handwriting, no data re-entry.

The digital record is transmitted by telecommunications to a central server for statistical analysis. So there are no mountains of paper or paper lost and no FedEx charges. If investigators are encouraged to complete and transmit the eCRF in a timely way, resolution of the few remaining queries can occur while a patient’s visit is still fresh in the investigator’s mind. There are therefore far fewer ‘educated guesses’. It is possible, indeed practical, and even preferable to lock a database in days or hours after the last bit of clinical information is available. Data management becomes at least one decimal order of magnitude faster, and entails a decimal order of magnitude fewer queries, than the legacy paper process.

How much?

The tool needed for EDC is a software system that contains edit checks, transmits data and stores the resultant database on a server accessible by authorized parties for further processing. A number of vendors license turnkey packages for a large fee, while alternatively, a sponsor can write its own software or acquire rights to an existing system. Vendor licensing fees typically amount to several dollars per page, whereas writing an EDC application would cost a few million dollars and require one and a half to two years.

At two companies, we chose to acquire rights to an existing system for a one-time fee. Free from further vendor fees, the cost of data management was reduced to about US$5 per page. Queries have been reduced 93 percent. Most significantly, the time to lock databases has been reduced, on average, from 10+ weeks to four days. At these two companies, more than 700 studies have been successfully conducted in this fashion.

EDC is a solution that is much faster, better and, in our embodiment, much cheaper than the paper alternatives.

Why now?

EDC is not new. Some companies began using EDC as far back as the early 1980s, even before the dawn of the internet. Bayer USA used EDC exclusively since 1995. EDC appears to be catching on, not because it is now possible, but because a small number of sponsors have shown that clinical development can be accelerated while actually reducing cost. In a trendy industry, it becomes increasingly difficult to explain to shareholders why a sponsor has failed to deploy modern practices that are clearly faster, better and cheaper.

Why not?

Order-of-magnitude change requires a re-focus from labor to capital – from working harder to working smarter. New skills will assuredly be required but also new organizational structure. Technology is a tool, not a solution. The internet is a convenience, not a requirement. Early adopters of EDC used FedEx and, later, 1200 baud RS-232 telephone lines. The difference in these communications media is merely quantitative, not fundamental. The ‘baud rate’ of a floppy disk in a FedEx envelope is about 7 bits/second. This is awfully slow compared to a cable modem, but still much faster than dispatching a clinical trial monitor to an investigator’s site to resolve a query. Using this metaphor, most trial sponsors still using paper CRFs ‘transmit’ their clinical data at a rate slower than that of a floppy disk in a FedEx envelope.

Creation or evolution?

One risk attendant to a new technology is to regard the exception as the rule, the cracked tooth as the whole smile. Once a new technology shows what is possible, it is tempting to idealize an environment in which the fullest capabilities of electronic information management and self-organizing conceptual integration become automatic. At best, these daydreams become goals. At worst, they become requirements standing firmly in the path of real progress. The ideal environment usually requires replacing the entire suite of legacy systems, often in one stroke. This is the ‘creationist’ approach. The creationists can dispense with the petty obligations of the day to day work that must continue. To be successful, however, the entire suite of new systems must be developed before the requirements change and they must be deployed without bringing the sponsor to a stop.

This ideology would say that the Wright Brothers were failures because their airplane was subsonic and lacked a pressurized cabin. Anyone who used EBay in its early years grew accustomed to crashes so frequent that EBay was forced to extend automatically the auction duration whenever a crash occurred within one hour of its scheduled end. One supposes that entrepreneurs should make their businesses perfect before they make them public.

The alternative is an ‘evolutionary’ approach. Day to day business must go on, but requirements and people change. A ‘creationist’ project will long have been abandoned before any useful result emerges. The evolutionary approach finds the low-hanging fruit and deploys a solution for one part of the overall environment. In all likelihood, like EBay, the first versions will have bugs and will not do everything that everyone wanted. Many of the bugs and limitations will not even be evident until a system sees widespread use. Adaptations over time will continue to improve performance and refine a system’s menu of features.

It is essential not to regard the goal as a requirement. The goal is to be perfect. The requirement is to be no worse than the alternatives but superior in one or more respects. “Fast” is better than “better.” “Better” is the enemy of “good enough.” Bells and whistles merely introduce more opportunities for bugs and impose greater burdens of training and support. Eventually, the evolutionary approach will bring a sponsor closer to perfection. The creationist approach will bring the sponsor to its knees.

The literature is replete with examples of companies that failed to adapt and individuals who obstructed beneficial change that threatened their own livelihood. Resistance to change is to be expected. The job of management is to push it through anyway. There is no escaping a fundamental truth – to do something for less means needing fewer people to do it. To do something for a lot less means not only fewer people but different skills for those who remain.

What’s next?

Some of the time and expense of clinical trials is unavoidable. A dosing regimen lasting six months will still require half a year. Weeks and months to clean a database, to perform statistical analysis, and to prepare New Drug Applications is a legacy of a bygone era. Data Management used to be the long pole in the tent. Now, other steps have become conspicuous in their consumption of time and money. Electronic submission of New Drug Applications is much, much easier when there are no longer millions of pages of paper records to paginate, photocopy and verify for legibility. With EDC, an e-submission is less work than a paper submission. With paper CRFs, e-submissions represent even more work than paper submissions.

In principle, data from an ongoing clinical trial could be used to alter the course of a trial for greater efficiency and safety. These so-called ‘adaptive designs’ have long been possible only in theory. As a practical matter, the elapsed time to clean databases and to prepare statistical analyses is simply too long to provide any benefit. The trial would be over before any guidance could be provided to alter its course. With EDC, they become practical. Adaptive designs could fundamentally alter clinical trials. Dynamic patient sample size re-estimation becomes feasible. ‘Failing faster’, or succeeding faster, can become routine. Abandonment of unproductive study arms could slash clinical costs and reduce fruitless efforts.

Adaptive clinical trials are simply not feasible if months after the last patient’s last visit are required to reach a clinical decision. All but the lengthiest trials will be finished before any interim results are known. The adaptive trial will have become nothing more than a conventional trial with the added burden of a statistical penalty for looking at the data prematurely.

Speed is key and EDC makes this speed possible and e-submissions simple. Speed enables adaptive designs. No one can argue credibly that the archaic paper process is cheaper or leads to better data integrity. It is simply obsolete. Flogging paper CRFs harder will not achieve the benefits of EDC any more than giving Dale Earnhardt a bigger engine would let him drive coast to coast in six hours.

EDC is the tool for speed. To be useful, it must be deployed by knowledgeable managers. Their senior executives need the spine to deal with the disruptions that accompany a paradigm shift. Disruptive technologies are, after all, disruptive.

Speed, quality and efficiency can stand high on a solid foundation of EDC – e-submissions six weeks after LPLV – adaptive designs terminating halfway through. Sponsors successfully deploying these methods will be months or years ahead of laggards and at far less cost. Recalcitrants will be remembered alongside stenographers and blacksmiths as curiosities of a past era. Their stock shares will become mere relics of yet more companies that failed to adapt.

About the author:

Dr Sylva Collins is a leading authority in management of biostatistics, data management and clinical systems development, with over 20 years experience in the pharma industry, including the preparation of more than two dozen successful NDAs.

In her current position at Novartis, Vice President, Advanced Clinical Systems, Collins is responsible for exploring opportunities for fundamental changes in clinical development by leveraging new technologies, systems and integrating processes to make drug development ‘faster, better, and cheaper’.

In her previous position as Vice President and Global Head of Electronic Data Management, Collins led the deployment of EDC. She received the Novartis Excellence award in innovation for her efforts in deploying EDC.

Collins serves as a member of the CDISC Board of Directors, the PhRMa Taskforces on Adaptive Designs and Improving Efficiency of Clinical Research.