The Internet was born in the 1960s and its applications were initially limited by the military uses for which it was originally conceived. The original “Internet” consisted of a cooperative network of four university computers in the United States (Stanford Research Institute; University of California, Los Angeles [UCLA]; University of California, Santa Barbara; and University of Utah) [1]. The development of a protocol for information distribution in 1990 by Tim Berners-Lee paved the way for the emergence on the Internet of applications with broader public appeal [2]. Fifteen years after its inception, the World Wide Web has become a nearly indispensable tool in education, government, business, news media and, most important for the purposes of this paper, medicine and research [3]. Originally designed as an emergency communications network, the medium evolved from a communications tool for academics and the military to a medium used for education, government, business, news media, entertainment, medicine, and research. The Internet has grown at a phenomenal rate; with over 100 thousand domains or hosts in 1993 it currently has over 250 million [4]. It is the first unrestricted uncensored broadcast medium, and under ideal circumstances (namely, the right location, low traffic volumes and the right service provider), it can be very cost-effective, because unlike the telephone system, there is no charge for long-distance service.

A clinical trial can be defined as any form of planned experiment involving patients [5]. The goal of a trial is to discover or verify the safety and effectiveness of interventions designed to promote wellness and prevent, diagnose, treat and provide prognosis information about disease [6]. The essence of a trial is comparison [7]. The comparison is between a group of patients who receivedtreatment with the intervention in question and a group of patients who receivedplacebo or another standard treatment. The modern clinical trial has evolved to include several features in order to provide reliable and valid results. A good trial addresses a specific clinical question for which there is equipoise (an uncertainty as to whether any of the treatments is to be preferred over the others). It uses a predefined patient population, a well-defined intervention in comparison with an appropriate control, predefined outcomes, and a methodology that involves getting informed consent from participants. Further, a trial involves appropriate blinding, randomization, and analysis. The inclusion of a control group, as opposed to historical data, is to ensure that any observed differences are due to the treatment under investigation and not another prognostic factor [5]. The purpose of randomization is to balance the treatment groups for both known and unknown prognostic factors such that any observed differences in outcome are more likely to be due to differences between the treatments in question [8]. Hence, randomization helps to prevent patient selection bias. The purpose of blinding (patients, investigators, and analysts) is to prevent outcome assessment bias.

Figure 1

Cumulative number of randomized control trials (RCTs) versus online RCTs (based on Medline and Old Medline searches from 1950) on a logarithmic scale over time

An RCT is conceptually simple, but to plan a protocol for a study, obtain funding, recruit patients, conduct the trial, and analyze the data collected require considerable resources. The initial clinical trials evaluating antibiotic therapy for communicable diseases had the advantage of large treatment effects–Hill's trial on streptomycin therapy demonstrated a 74% risk reduction for mortality [12]. Today, most interventions investigated in superiority trials are expected to have a more modest benefit, perhaps a 10% to 20% risk reduction for an important outcome [6]. In order to investigate these more modest treatment effects it is necessary for modern trials to be carefully designed so that both systematic and random error are minimized, as differences of this magnitude cannot be detected reliably against a background noise of chance or other influences. Systematic error is minimized with a well-designed protocol that avoids bias, and random error is avoided by studying a large enough sample size [13]. Sample size is of particular importance in the conduct of equivalence trials. Equivalence trials, in contrast to superiority trials, are designed to establish no difference in efficacy between two interventions. However, in order to show equal efficacy, equivalence trials usually will require a 10% larger sample size in comparison with conventional superiority trials [14]. In order to achieve a sufficient sample size in a reasonable time, many trials recruit patients from multiple centers across several geographical entities (eg, cities, countries) [6]. These multicenter trials require infrastructure which is accomplished with a central coordinating center that usually handles the recruitment of study centers, the randomization of patients, any necessary laboratory analysis of patient samples, data collection, data analysis, and quality control [15].

A well-designed RCT begins with the identification of a medically important question [16]. Before undertaking a new trial it is important to know what research has been done on the question in the past. To identify previous trials and systematic reviews, the Internet can be used to search online databases. Medline, EMBASE, and the Cochrane Library are online resources that can be used to quickly identify both systematic reviews and clinical trials [17]. Medline can be accessed free of charge using PubMed, but both EMBASE and the Cochrane Library require registration and an access fee [10-12]. Once relevant citations are found, most of the full text articles can be obtained by accessing the journal's home page. Members of academic institutions can often access electronic journals free of charge from their homes or offices by accessing websites via a proxy server, most often the institution's library home page [18]. Ongoing and some completed trials can be located from online trial registries in both the United States and Europe [19,20]. Online searches are useful in identifying published studies but researchers interested in exhaustive searches on a subject will have to supplement them with conventional hand searching of relevant article reference lists and by contacting experts in the area [21].

Once a research question is formulated and the literature in the field is reviewed, the Internet has tools to aid with the task of protocol development. The US National Cancer Institute maintains a website that has suggested templates for phase I – III studies, guidelines for dealing with various patient groups, as well as guidelines for formulating informed consent documents [22]. The University of California, San Francisco, School of Medicine maintains a website devoted to clinical research tools [23]. The site includes templates for study subject screening and data collection, data and safety monitoring, financial tracking, study budget, and checklists for protocol feasibility and study management. If the local expertise is not available to help with the development of the trial protocol, companies advertising online offer experienced teams of medical experts, biostatisticians, and clinical research specialists to help clients design clinical trials [24]. One of the key steps in the generation of a trial protocol is calculating the required sample size; online tools exist to perform this calculation [25].

A difficult hurdle is obtaining funding to conducta clinical trial. The Canadian Institute for Health Research, the National Institutes of Health in the United States, and the Medical Research Council in the United Kingdom maintain websites that contain advice to applicants and online submission forms for specific grants [26-28].

There are many reasons for a multicenter clinical trial to have a website [29]. A study website can be used for the following tasks: providing information to potential participants, study subjects, and investigators; listing contact information; and centralizing data handling for patient registration, randomization and data collection. Detailed information about the trial can be displayed, and the entire protocol (apart from any confidential aspects) can be made available. A secure (password protected) section of the website can be used as a powerful means of communication for trial personnel (investigators, monitors, sponsors and committee members). Today electronic mail is the standard for communication amongst members of a trial group; it is faster than conventional mail, cheaper than using long-distance telephone service, and provides an archive record of the communications. A directory on the website of the investigators, committees, sponsors, and monitors with their email addresses can help improve communications. A directory of participating centers and regional coordinators would also be helpful. A news section of the website can provide a progress report concerning the trial status and advertise upcoming meetings. A ”Frequently Asked Questions” section can provide investigators with answers to common questions regarding the study protocol, and a download page can be a means of distributing study materials (protocol, case report forms, informed consent forms) to participating study centers.

The Internet also plays an increasing role for informing the general public about ongoing trials that are recruiting patients. Prior to the emergence of the Internet most patients were recruited for clinical trials through their physicians or perhaps through mass media advertising [30]. This system depends on individual physicians keeping up-to-date with a large range of clinical trials--an impossible task. The US Food and Drug Modernization Act of 1997 required the Department of Health and Human Services to establish a registry of clinical trials for both the government and the private sector [31]. As a result a new trial registry was launched and the home page banner reads “linking patients to medical research.”[19] The site was launched in February 2000 and currently contains approximately 11300 clinical studies sponsored by the National Institutes of Health, other US government agencies, and the pharmaceutical industry in over 90 countries. People who access the site can find trials by searching by disease condition or funding source. The website also provides information for people considering participating in a trial, including basic information on clinical trials. Several other commercial websites have been launched with the business idea of linking patients with clinical trials [32-34]. It is important for potential participants to be cautious because financial incentives used to recruit patients may interfere with ethical informed consent [30].

Once a patient indicates interest in participating in a particular trial, he is then screened for eligibility, and provided with the information necessary for informed consent and a consent form for signature. The necessary data for enrollment into the study is then collected. A study website can provide detailed information about the clinical trial presented in terms which the general public can understand. An online questionnaire canscreen for potential participants, and eligible patients, who elect to participate, canbe directed to the enrollment page and consent forms, made available for downloading from the website. This paradigm necessitates that potential participants have access to the Internet and that they be reasonably familiar with computers. To access the Internet, potential participants would require a personal computer, a Web browser, and access to the Internet via an Internet service provider [35]. Given these requirements, this method of patient recruitment could lead to selection bias. Surveys conducted on the demographics of Internet users show that the average user is young, white, employed, well-educated, with a higher social-economic status, and suburban [36]. Those who lack the resources for online access (for example, those with a disability that prevents access, or those who are socially disconnected or lack knowledge about Internet access points in the community) would be less likely to use the Internet and would therefore be underrepresented; whereas professionals working in the computer or telecommunications industries would likely be overrepresented.

Traditionally, study participants have signed consent documents by hand, but new legislation in both the United States and Canada has given legal weight to digital signatures for the purpose of facilitating electronic commerce [37,38]. A digital signature is a unique string that special software creates by applying a mathematical function and an encryption key to a message or file [39]. The unique string confirms both the file author's identity and the maintenance of the integrity of the file during its transmission. If accepted as ethical and legal for clinical trials, digital signatures would save the step of mailing hand-signed consent forms to the coordinating center. Regardless of the method used to obtain consent, it is important that the study participants are appropriately informed of the potential risks and benefits of the trial intervention, and of their rights regarding their electronic information. It is necessary to offer patients the option of not having their information handled electronically (for those that refuse)and to give them the option to request removal of their electronic information from the electronic environment [40]. In terms of informed consent, an argument couldbe made that all eligible patients should speak to a study representative (in person or on the phone) in order to ensure that the complexity of the study and confidentiality issues are clearly communicated and understood prior to proceeding with registration into the clinical trial. In-person contact with all trial participants wouldhelp with verification of the baseline data collection and help guard against people who might attempt to pose as a patient for mischievous reasons.

The method of dividing subjects into groups is called random allocation or randomization and is necessary to ensure that any baseline differences between groups are due to chance alone [41]. This prevents selection bias and ensures validity of certain statistical tests. Several methods of randomizing have been used over the years, including coins, dice, cards, lots, spinning wheels, random number tables, and random number generators on computers. For multicenter trials a central coordinating center often serves as the randomization center and participating centers access the randomization allocation by a 24-hour phone service. As an alternative to this service, which can be expensive, there are several online randomization programs (some free of charge and some commercial) that can generate random allocations [42]. Randomization.com is a cost-free online randomization program that generates simple lists of allocations that can then be printed [43]. Paradigm is a Web-based randomization package developed by the Netherlands Cancer Institute and the UK Medical Research Council; it is free of charge and guides through studies interactively [44].

Remote data entry to a central database is one of the more useful promises of conducting a clinical trial using the Internet. A single-center clinical trial can have data entry decentralized by having a two-tier (client-server) network system that involvesindividual application instances (thick clients) running on remote computers connected to a central database server [45].

In a multicenter trial, participating centers can be geographically separated by great distances across several cities and countries, making the traditional local area network unfeasible. A thin-client (less bandwidth intensive) Internet-based solution canbe used to connect study centers from all over the world. An Internet data entry solution has Web-browsers - thin clients running on remote computers with the application itself running in a central Web enterprise application server. The three tiers (client/investigator, Web application server, and database server) of an online trial system are illustrated in Figure 2. With this system the following processes occur: browser requests/submits data from/to the Web application server; the Web application server in turn executes incoming business logic and submits/requests data to/from the database server; the database server saves the submitted data and sends the requested data to the Web application server; and the Web application server then executes the outgoing business logic in the application, formats the resulting data into HTML,and sends it back to the browser as a Web page. With this Web system the traditional case report forms are translated into electronic forms in HTML [46].

HTML Web pages by themselves are static text documents that cannot accept data input [47]. It is necessary to incorporate an additional enterprise application between the Web server and the database server in order to facilitate data collection, increase the efficiency of database requests and offer additional functionality for interactive real-time data validation [46]. Real-time data validation (see below) can reduce transcription errors and avoids missing data; the data thus collected should be of higher quality. The step of double keying the data for quality assurance becomes redundant [48]. This additional functionality can be incorporated by running Java or .Net code in the application server, which allows for interactive behavior with each data field [46,49]. Java is a computer programming language developed by Sun Microsystems that allows small-application programs to be downloaded from a server to a client along with the data that each program processes [50]. More commonly Java writes server-side enterprise applications that interact with Web browsers and other enterprise applications through pure HTML and Extensible Markup Language (XML) over the Internet and corporate intranet, and XML Web services (small, discrete, building-block applications that connect to each other). Microsoft .NET is a set of Microsoft software technologies for software integration through the use of XML Web services as well as to other larger applications—via the Internet [51].

Figure 2

Example of a 3-tier architecture in an online clinical trial system

Clinical data comes directly from the patient, the medical record, or a laboratory test. In traditional paper-based clinical trials, data is recorded on paper case report formsand then transcribed into a computer. Electronic data collection through the Internet has a number of advantages including real-time data validation, time savings due to fewer steps in data collection, and reduced handling and storage costs due to the near-elimination of paper source documents. Real-time validation could alert a researcher to an invalid entry even as he is viewing the original data source. For example, a researcher recording systolic blood pressure (in mmHg) and entering a value of 1400 couldbe prompted immediately of an invalid entry, allowing for immediate correction. The disadvantage of this electronic approach is that the US Food and Drug Administration requires validation of clinical data from each trail and it is not clear how this can be done with electronic systems. In the past the computer hardware for mobile data collection was insufficient, and study data monitors have been reluctant to embrace a fully electronic data collection model [53].

After collection, data can be analyzed using online statistical tools. Simple Interactive Statistical Analysis (SISA) is one example of such Web service [25]. This Java program allows users to do statistical analysis directly on the Internet. Users select one of the procedure names, fill in a form, and click a button for immediate data analysis. Another website contains hundreds of links to free-of-charge online statistics books, tutorials, downloadable software, and related resources; immediate analysis of the results is available to the investigator [54].

Security is a central issue when considering the Internet for sensitive information exchange. Both patients and study investigators need to be confident that the data entered on electronic forms and in email communications will not be intercepted by a sniffer. A sniffer is software that monitors network traffic and it is analogous to a telephone tap [48]. The database server itself needs to be protected from intrusion from unauthorized Internet users and from unauthorized intranet users (clients connected to the local area network) [39]. Lastly, the system needs to guard against spoofing (the practice of someone pretending to be someone else) [48]. Malicious Internet users could enter fictitious patient information and invalidate the trial results. Essentially, a secure Internet clinical trial system should ensure confidentiality (information is only disclosed to users authorized to access it), integrity (information is only entered or modified by users authorized to do so), and availability (information and other resources can be accessed only by authorized users) [55].

The underlying network protocol (TCP/IP) on the Internet contains no security layer [49]. To address the issue of secure Internet transmissions, Netscape designed a nonproprietary protocol for providing data security between application protocols (such as http, telnet, NNTP, or FTP) and TCP/IP [39]. The Secure Socket Layer (SSL) provides data encryption, server authentication, and optional client authentication for a TCP/IP connection. Encrypting a file changes it from readable text to a series of numbers that only parties that have the decryption key can interpret. The latest versions of Web browsers support 128-bit encryption, translating to a code that is almost impossible to break. A computer capable of 225 million instructions per second would take a dedicated year of processor time to break such an encryption code. Documents from secure servers can be identified from the location (URL) field. The letter “s” is added to the protocol (http:// becomes https://). Encryption is also available for email communications, and this is usually (depending on the software used) can be selected as an option in the preferences menu of the email application.

After secure transmissionand storage, the data needs protection from unauthorized access once it is stored on the central database server. For this purpose firewalls – hardware and/or software that sits between the database server and the Internet – are used [39]. Further, the database server needs to be placed in a secure location so that unauthorized users cannot access it.

With the confidentiality of the clinical data maintained by encrypted transmissions and firewalls, the integrity of the data can be maintained by user logins and passwords for data entry and editing [56]. More sophisticated user authentication is possible using digital signatures. Potential spoofers have to be screened out by the enrollment procedure. This can be accomplished by communicating with the primary caregivers of potential study participants.

Currently, most major medical journals are published online and individual articles, including the title and abstract, can be browsed; full text versions are often available for download. Freemedicaljournals.com is a website that contains links to over 900 medical journals with full text articles free of charge [57]. Several mainstream journals are included but some journals limit access to articles that have been published for greater than six months to a year. Open Access journals such as those listed in the Directory of Open Access Journals (eg. BioMed Central [58] or the Journal of Medical Internet Research) offer speedy peer review and rapid publication. Article Processing Fees need to be paid from the authors' research institution or grant to cover the expenses for the peer review process and the preparation for online publication. It is the main source of income to recover publication costs for Open Access journals since the articles can be viewed free of charge and no pay-per-view charges can be imposed.

The main advantage of online clinical trials is the ability to centralize study information and coordinate multiple trial processes in real time at a lower cost [75]. Multicentered trials are more manageable because a system can be scaled easily to many study centers around the world without special requirements for hardware or software. The only requirement for each participating center is a computer with a Web browser and Internet access. Site training, patient recruitment, randomization, data collection, site monitoring and patient safety can all be enhanced and simplified using a clinical trial system. Further advantages include fewer personnel for trial administration, reduced or neglible paper reporting, security and backup of the entire trial at a single location, optional updating and distribution of trial protocol and data collection forms from a single location, and simplified dissemination of results.

The key disadvantages of online trials are the real and perceived security threats that may inhibit both patients and study centers from participating. It is difficult to convince the average person of the efficiency of the abstract security measures used in Internet trials (firewalls, encrypted transmissions, password protection) compared with the conventional security measures used in traditional trials (locked file drawers). If participants are recruited through the Internet, this may lead to selection bias. Given the anonymous and transient nature of the Internet, it can be difficult for trial coordinators to assess the suitability of Internet resources that are not directly associated with well-known academic institutions. The transient and anonymous nature of the Internet is illustrated by the practice of citing the date of access for electronic resources and by the fact that many documents on the Internet do not have a documented author. If a trial relies on a third-party Internet resource, there is always the possibility that the third-party website ceases to exist prior to the completion of the study, leaving the coordinators to find an alternative resource to complete the trial. For example, finding another randomization site in the midst of a trial, which takesinto account previously allocated patients, would be problematic.

Other disadvantages of an online system includes system performance, lack of live support personnel, and the setup cost. The speed of the online system can be slowed significantly during peak Internet traffic and this can prolong every step of a study, from registration to data entry. The lack of a 24-hour call-in center can lead to the loss of some patients because some study centers may not be able to use online help to solve their difficulties with the study protocol or the registration and randomization steps. To set up and maintain an online clinical trial system requires experienced computer professionals. This might be too expensive for smaller trials where the administration budget is modest.