The promise and practice of technology – Implementing electronic development systems

Current Drug Discovery

current_logoBy Dr Michael Rosenberg, President & CEO, Health Decisions

December 2001 – Although information technology has been promoted as the tool that will transform pharmaceutical development, advances to date have been incremental, for the most part having little if any impact on development timelines. This lack of effect can be partly explained by increasing demands for data, which to a large extent have offset the marginal improvements in data handling. In some cases the use of technology has actually impaired progress, and in a manner unrelated to the difficulties associated with introduction of change.

However, times are changing and maximising and protecting one’s product in the market is becoming more important. Regulatory agencies, particularly the US Food and Drug Administration (FDA), have become increasingly concerned about the safety of newly marketed products – a concern that has been magnified with highly publicised criticism levied by consumer groups after the withdrawal of products, such as troglitazone. A typical pre-approval clinical development programme generally results in a database of 2,000 to 4,000 patients. However, this can only accurately detect adverse reactions more frequent than one in 1,000. As a result, companies are being encouraged or indeed required to sponsor post-marketing surveillance studies of their product using scientifically rigorous methodology. In constructing expensive trials that purely examine safety, companies may be unable to indulge in those other post-marketing trials that help develop the product’s market platform. We propose that the integration of web-based clinical trial systems into pre- and post-approval trials can help deliver on both these objectives.

In clinical trials, technology has been most visible in the area of data capture, most recently with electronic data capture (EDC), but not long ago in the form of faxback systems. Both approaches are intended to quickly collect data from the site. Faxback systems met with limited success because the systems did not work well from the investigating sites’ perspective and even when data could be quickly faxed back, the same bottlenecks that previously existed with data entry remained. Asa result, faxback systems have largely faded from use. Today, however, the same claims about being able to accelerate development are being made about EDC systems.

Many companies have had poor experiences with other ‘technology solutions’ and consequently are wary of adopting EDC. But the need for these systems stems from recognizing the increasing complexity of development and the competitive advantages of improvements. This article discusses the broad issue of technology and competitiveness, including reasons for the limited impact of the systems to date, and explores some of the changes that will be required for technology to have a meaningful impact.

The need for technology
There can be little question that pharmaceutical development requires the effective management of thousands or even millions of data points, making computerized systems mandatory. There is nothing complicated about the principles involved, but the execution is enormously challenging. Not only are there many thousands of data points, but also a complex set of regulations under which data are collected, processed, stored, and analyzed, including the requirement that every point be traceable to a source based on interaction with a patient, laboratory assessment, or other verifiable means.

Other industries face even more daunting challenges in data handling. The banking industry, for example, may handle thousands of transactions per second, with a requirement for absolute accuracy. Complex manufacturing operations such as assembly of automobiles require immediate status knowledge of intricate supply chains, assembly progress, and fault management. From the broad perspective of international commerce, the pharmaceutical industry makes very little use of modern capabilities for information collection, processing, and process control.

At the heart of the challenge facing the pharmaceutical industry is the need to transform data into effective information and subsequently translating this into knowledge with which to make effective and timely strategic decisions. At a tactical level, a process that utilizes a technology platform to provide ongoing access to information and knowledge can improve the efficiency of clinical trials management, facilitate adjustments and refinements to patient recruitment, and provide support to individual sites. At a strategic level, ongoing access allows more effective green/red light decisions, providing a framework for planning and investment decisions, and helping to reduce development time leading ultimately to a faster and greater return on investment.

Technology’s role is to provide more data, of better quality, earlier in the process than would otherwise be available. When this is combined with an experienced team of drug developers and dedicated project management teams, then the claim that technology can transform drug development can be truly substantiated.

Product development: Competitive landscape
Competitive pressures will inevitably require the pharmaceutical industry to move beyond the current piecemeal approach to technology. Drug discovery is no longer restricted to large multinational companies with huge staff and laboratory resources; small startups can now purchasea roomful of gene sequencers for a comparatively modest investment. The same transformation will occur with product development. One could argue that many recent mergers were borne of weakness rather than strength, specifically due to difficulties discovering and developing drugs. Large companies may have a cushion for inefficiencies, but biotechs must juggle limited resources with the need to move products quickly through testing. But whether you run a multinational or a startup, there is little question that the efficiency of the development process will define your viability as a company. With combinatorial chemistry, genomics, and a host of small companies eyeing the market, we are now faced with an abundance of drug candidates. The only way to commercialize these is through clinical testing, and one of the key elements involved in improving the efficiency of clinical testing is better application of information technology.

Adding to the urgency of improving the process is the increasing demand for data being placed on new drugs. The need to increase the number of patients and clinical studies together with the number of data points per clinical study all add to the burden on the drug development process. According to a 2000 study by PhRMA, the number of clinical studies in an average NDA increased from 30 in 1981-84 to 68 in 1994-95 and close to 200 in 2001. The FDA approved 27 new drugs in 2000 compared to 35 in 1999 and 30 in 1998. Moreover, the mean approval time for the FDA to approve these drugs was 17.6 months in 1999 versus 12.6 months in 1999 and 11.7 months in 1998, respectively. In other words, the FDA is requiring more information, taking more time, and approving fewer drugs.

Technology’s role in improving the process
Since technology’s greatest contribution is to improve processes and allow a decrease in time to market, its effective leveraging requires a critical appraisal of current processes. Probably the single most important reason for the limited impact of technology to date, which focuses on data collection, is that it represents an attempt to speed existing processes rather than allowing more efficient processes to be implemented. This is illustrated by the fact that the vast majority of current research continues to use paper as a means of recording data, and EDC, an attempt to improve that process, has not decreased development times. Enhancing the flow of information, study execution and strategic decision-making starts with three basic steps:

The ability to know what is going on in the field in a timely manner
The ability to summarize and share this information
The ability to respond by taking corrective actions.
This cycle is repeated, so the effect of corrective actions is measured by continued incoming information from the field. Traditionally, however, the inability of the pharmaceutical industry to accomplish the first step precludes either of the latter steps. For example, according to a recent report by CMR International, the industry average for processing data is about 30 days. Those who deal with data management know that this period is frequently much longer, especially at the beginning of a trial when the need is most critical.

Adverse reactions

Industry’s current emphasis on data collection arises from the vision of reducing the time to generate clean data. However, even this well-circumscribed step requires a number of complementary elements to be in place. Most notably, the software must work well at the site – reaction to many of these products is lukewarm at best. Complaints that software interferes with patient flow by anchoring staff to one location, that it is not intuitive, and that it interrupts process flow through inability to skip questions or suspend the program when interruptions occur, all interfere with a carefully choreographed flow of patients.

Medical staff may dislike a new system because it requires that they record answers on a form, then later transcribe the answers to the computer, making them data entry clerks for which they may be illprepared, as well as expensive and inefficient labor. Distaste for data entry and interference with daily activities means substantial delays between the time a patient is seen and when data are transcribed, undermining the intended benefit of immediate feedback and data cleaning. Finally, many EDC systems are all-ornothing: all the sites use them or none do, and it is rarely the case that the level of sophistication in all sites is sufficient for smooth flow, especially if the study is multinational.

One approach we have taken to this problem is machine-readable paper forms (which can be used in conjunction with EDC, with some sites using EDC and others paper). Like many companies with early experience in EDC, we found poor acceptance in the field and consequently mostly use forms that can be overnighted for validation and query generation. This approach allows us to validate and release queries for as many as 8000 pages a day, and our average time from receipt to query release is 4 hours. Used in conjunction with a web-based system of managing queries, this provides the ability to respond quickly while the case is still fresh in the minds of the investigators. Many sites mention that this system is actually faster in returning queries to the site than EDC systems as well as being easier to use in day-to-day activities.

Integrating drug development
Maximizing the use of technology means stepping back and looking at the full course of actions and processes involved in drug development and making sure all the elements are tightly integrated and work well together. Included in these are:

Data collection
Data validation and query generation
Query management
Status reports
Regulatory submissions
Achieving the three requisite steps for information flow noted earlier requires a considerably broader outlook than simply data collection and at the same time provides the real opportunity for technology. Consider the requirement for summarizing and sharing information: basically this involves transforming a mountain of data into a more digestible synopsis that reflects progress. The internet is a wonderful medium that allows around-the-clock access throughout the world. Web interfaces can query study databases in real time and return immediate information. Reports can be filtered for different audiences, perspectives, and needs. Sites, for example, may wish to examine trends in number or types of queries generated, while sponsors may focus on screening failure rates, enrollment, and study milestones.

Study monitoring and query management should also be integrated. With current systems, study difficulties are not apparent until after the study is completed. This in part reflects the fact that we send monitors out without any real sense of what is occurring with the data – having them go down a checklist without providing them with the information needed to manage a study. Monitoring can be improved by having the sites deal directly with queries, freeing monitors from this arduous task and allowing them to focus on site management. How many patients fail screening? How many queries are outstanding, and how many are overdue? How does this site compare to others, particularly where competitive enrollment is active? The ability of sites to deal directly with queries generated on recently seen patients rather than having to wait for periodic monitoring visits both improves the process from the sites’ perspective and provides immediate feedback that reduces queries to a rate as low as one every 20- 50 pages.

System integration

The goal of an integrated system is the ability to produce clean data in a relatively short time. Queries, discrepancies, table shells, and a great deal of the work that normally goes on after a study has ended can be done ahead of time by assuring access to data (even while still blinded, if a pivotal study) as they are collected. By the time of the last patient’s visit, data should be cleaned, and databases should be locked within a week. Because table shells have similarly been set up, completing preliminary results should be a matter of days. These figures contrast with current industry norms of more than a month to complete data collection from the field and 4 months to lock a database.

Regulatory submissions are the final common pathway for much of the data, information, and knowledge collected over the course of years and often thousands of patients. Tying the study system into a framework to facilitate submissions is an important element of minimizing the interval between completion and regulatory submission. Technology now allows seamless integration of electronic clinical trial systems and document management systems. Coupled with dossier publishing software, the ideal aim of clarity and consistency of messages and supporting data can be achieved throughout the myriad documents that make up a regulatory submission. By using such technology throughout the development process, the regulatory submission itself can be reduced to a much more simplified task leading to faster time to submission following the end of pivotal clinical trials. This can be on the order of 3-4 months compared to the current industry average of 10-12 months.

Beyond technology
Achieving some of the timelines noted above depends on considerably more than technology. The most common reason for technology failing to have the expected impact is the failure of other components that need to be aligned. Development involves many inter-related components, of which data collection is a relatively small part. Companies must have an environment that optimizes the contribution made by technology and recognizes and rewards the people who use it well. Too often, technology is seen as a panacea that will allow a company to leap forward in its development efforts.

Any new development that changes established procedures requires, by defini¬tion, acceptance of a different environment in order to succeed. For technology to assist in moving drug development forward, a company must recognize the close integration of data management, project management, biostatistics, and quality management. Most pharmaceutical companies are compartmentalized, with these functions discretely defined. This sequential processing approach is based on a pen-and-paper model that does not allow one function to be initiated before the previous step is completed. This creates a bottleneck in information flow and makes it difficult to respond while incoming data are fresh enough to reflect what was actually occurring at the sites.

Getting it to work

Suppliers of technology systems, such as the EDC systems now marketed, may be frustrated by the poor results experienced by some of their customers. Indeed, they often find themselves in the awkward posi¬tion of having to provide a range of systems designed to work in different cor¬porate environments, but with no control over those environments. This places them in a dilemma: on one hand, what they can offer companies is the use of fragmented pieces, none of which by itself changes much and consequently fails to produce bottom-line results. On the other hand, the kind of change companies are seeking can only come from a system that broadly inte¬grates a number of different areas. It also requires an environment that can leverage, rather than interfere with, technology; this is often substantially different than the sit¬uation observed in the large pharmaceuti¬cal companies that are the primary customers for these purveyors. A simple technology such as data collection will not achieve full impact until the multiple com¬plimentary components that lie elsewhere can be aligned.

An extremely important realization of an integrated approach is that it changes the jobs of individuals, in some ways pro-foundly. For example, the easy availability of real-time information throughout a study team, from the leader to the monitor to the consultant, means that each member is empowered, and expected, to act on issues outside their immediate responsibilities. Many jobs change from an ‘assembly line’ role to one concerned with the overall flow of a process from start to finish. Companies need to recognize and reward activities that facilitate this important change; too often today’s companies have an environment that discourages any risk-taking. Managers know that not taking chances equates to stagnancy, and some feel this has occurred in an industry with traditionally high profit margins and high barriers to entry.

Another change may lie in the way companies work with developers of tech¬nology-based systems, CROs, and other outside groups. Traditionally seen as ‘handle-turners’, CROs have been regarded as vendors who supply a commodity (labor) that is interchangable and expend¬able. But innovation has traditionally been spawned by smaller companies, and those CROs that can effectively demonstrate an ability to provide more efficient clinical development linked to effective and proven technology will be able to provide a resource that many sponsors do not have. Wise sponsors will study such examples carefully, often by working with those com¬panies as codevelopers, in a manner that rewards the value they bring in the form of shorter timelines.

The future
Some analysts suggest that the pharmaceutical industry has yet to realize some of the most basic management principles that competitive pressures have imposed on other industries, notably manufacturing. An example can be found in the auto¬mobile industry, as described in James Womack’s book The Machine That Changed the World. The Japanese substituted mass manufacturing with lean production, resulting in a better, more cost-efficient product, higher productivity, and greater customer loyalty.

The hallmarks of lean production ¬teamwork, communication, and efficient use of resources – are equally applicable to any industry and require the same technol¬ogy foundation summarized here. And the results are remarkable: cars with one-third the defects, built in half the factory space using half the man-hours. Although pharma does not manufacture a similarly discrete product, the lessons are startling and clear, and portend the change our industry will have to undergo to remain competitive, as well as the tremendous opportunities for being among the first to achieve it.

Reprinted from Current Drug Discovery, December 2001.