JOURNAL HOME CME HOME THIS MONTH PAST ISSUES ETOC COLLECTIONS AUTHORS REVIEWERS EDITORIAL BOARD FEEDBACK RSS HELP
		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Avidan, A. Articles by Sprung, C. L. Search for Related Content PubMed PubMed Citation Articles by Avidan, A. Articles by Sprung, C. L.

---

ECONOMICS, EDUCATION, AND HEALTH SYSTEMS RESEARCH

An Internet Web Site as a Data Collection Platform for Multicenter Research

Department of Anesthesiology and Critical Care Medicine, Hadassah - Hebrew University Medical Center, Jerusalem, Ein Karem, POB 12000, Jerusalem, 91120, Israel

	Abstract

Top Abstract Introduction Methods Results Discussion Conclusions References

 
The Internet can be an effective alternative for data collection for multicenter studies. It has major advantages over the "classical" method of using paper and traditional ("snail") mail. We developed an Internet site and implemented collection of data for a multicenter study of ethical decision-making. The Web site was built with Microsoft FrontPage as the authoring tool. Database management was performed with Microsoft Access. Security issues were the major concerns for the web design. Thirty-seven European centers enrolled 4248 patients during 1.5 yr using the Internet site. The use of this Internet site for data acquisition was highly effective, and the investigators were able to use the system without training. Overall satisfaction of the investigators was high. After security issues, interactivity and user-friendly design are the main cornerstones for the success of such a system.

	Introduction

Top Abstract Introduction Methods Results Discussion Conclusions References

 
Data collection and validation are critical cornerstones in every research project. With the tremendous developments in computer technology it seems logical to move from traditional media of data communications and storage (paper and "snail" mail) to electronic systems. The Internet (World Wide Web and e-mail) provides what appears to be the ideal platform to collect and transmit research data. However, using an Internet platform for data collection has some drawbacks. Security is the most critical issue, especially for personal and confidential data that may be collected as part of medical investigations. Furthermore, building and maintaining of an Internet site for data collection may be expensive (1) and requires personnel or consultants with extensive knowledge and experience in various programming languages and software platforms.

The challenge was to collect data for a multicenter study of ethics and decision-making in intensive care units (2). The study was expected to continue for 2 years and include between 3000 and 5000 patients. Realizing that it could be impossible to cope with vast amounts of data using classic data acquisition methods, we built an Internet site to facilitate data entry and communications with the participating centers.

This article examines the problems and issues faced during the design, implementation and operation of this Internet data collection scheme. The aim was to identify problems that arose during this endeavor and propose solutions for future use of the Internet for data collection.

	Methods

Top Abstract Introduction Methods Results Discussion Conclusions References

 
The data collection system should allow easy data entry for 30 to 50 centers located in Europe. It was anticipated that 3000 to 5000 patients would be enrolled over a 1- to 2-year period. For each patient, the investigators were required to fill out a questionnaire containing 19 detailed multipart questions with 194 different data items. The data were to be sent automatically to the coordinating center located in Jerusalem, Israel, where they were to be reviewed by the coordinating center. Queries for missing or incomplete data were to be sent to the investigators for their corrections and additions. The system was also to be used as a communication platform to inform the investigators of developments, progress, and potential problems with the project. Confidentiality of patient data had to be insured, whereas coordinators required the ability to distinguish individual patients. At the end of the study, an anonymous survey of the participating centers was to be performed to assess their satisfaction of this type of data collection system.

Building a Web site for data entry requires client-sided (e.g., a browser program on the user’s computer) and server-sided (on the computer where the site is hosted and connected to the Internet) programming technologies (for example Java-Script, Perl, Active Server Pages). Microsoft FrontPage (Versions 97 and 2000, Microsoft, Redmond, WA), a web authoring tool with built-in scripting functions, was chosen for building the Web site. Client-sided built-in JavaScript functions were used for basic data validation and to assure that certain required fields are filled in before transmitting the data. The server-sided scripting functions were used for data forwarding by e-mail to the coordinating center.

The site was hosted at Verio (http://www.verio.com), one of the world’s largest web hosting companies.

Data were imported to the database (Microsoft Access 97) and reviewed and validated manually by the study-coordinating center. Investigators were contacted by e-mail to request correction or clarification of data. The investigators were to send corrections or clarifications through the Web site or by e-mail.

	Results

Top Abstract Introduction Methods Results Discussion Conclusions References

 
A Web site for data entry for a multicenter study of ethical decision-making was developed and implemented. Thirty-seven centers located in 17 European countries participated in the investigation. For 1.5 yr, data on 4248 patients were collected.

Data entry was performed directly onto the Web site through different web pages with 19 detailed questions. These pages incorporated 194 different data entry fields, such as text form fields, combo boxes (predefined selection lists), radio buttons (only one option of several can be chosen), and check boxes (checked or unchecked) (Fig. 1).

View larger version (37K): [in this window] [in a new window]   Figure 1. Data entry fields. Single line text form field; combo box (opened); radio button; check box.

The data entered into the combo box, the radio button, and the check box were coded and only the code was sent, whereas with the free text field the actual text was sent.

FrontPage has built-in JavaScript functions for basic client-sided data validation. These functions checked that the right data type was entered in the free text field (e.g., number and not text in the patient’s age field) and to assure that certain mandatory fields (e.g., patient number) were filled in before the data could be sent. A message box informed the user of incorrect or missing data. The number of free text fields was kept as small as possible so that efforts were directed at using combo boxes with predefined values to reduce improper data entry and to improve the Web site’s ease of use.

Security was an important consideration. Entrance to the web site was username/password protected and the site was not published in any Internet search engine. The patients were identified only with a unique number. The key identifying the patients was kept with the centers. After data entry, the built-in server-sided FrontPage functions generated an e-mail message containing the data and sends it to an account on the research server at Verio. Those e-mails were periodically automatically downloaded to the research computer at the Hadassah University Hospital, which was connected to the Internet through a firewall program protecting the Hadassah local area computer network.

After downloading to the research computer, the e-mails with the patient data were saved as text files and imported to the database (Microsoft Access 97). The data were then reviewed and validated manually by members of the research group. Also, data sent as free text (e.g., patient diagnosis), was coded manually for later statistical analysis. The centers were contacted by e-mail to clarify questionable or missing data. The corrected and completed data could be returned by the investigators to the coordinating center either by e-mail or via the Web site.

Newsletters were published regularly on the Web site informing the investigators about study related issues. On the FAQ (Frequently Asked Questions) page answers to common problems were provided.

At the end of the study, 24 (65%) centers answered an anonymous survey we performed to evaluate the design and usability of the Web site (10 = best to 1 = worst) (Table 1). Overall satisfaction was high (median 8, interquartile range 8–9).

	Discussion

Top Abstract Introduction Methods Results Discussion Conclusions References

 
This report is the first documentation of the successful design and implementation of a large-scale web-based data collection system for a multicenter study in medicine implemented with an easy-to-learn, easy-to-use, and commercially available inexpensive web-authoring tool (Microsoft FrontPage). Other reports noted prototype systems only (3,4) or databases with only several hundred enrolled patients (5,6). One study included a larger number of patients (7), but no detailed information is available about the technical aspects of the design of that Internet site (8).

The overall performance of this data collection system was as planned. The participating investigators used the Web site successfully with minimal training. Acceptance was high and data from large number of patients were entered. Newsletters published regularly on the Web site maintained communication between the coordinating center and the investigators.

In the following in-depth analysis of this data collection system, we discuss a number of important issues that should be of concern when planning and implementing such a system.

Data entry should be as easy as possible using predefined data fields (list fields, radio-button fields, check boxes). Free text files should be avoided to prevent incorrect data entry and to reduce the amount of data that has to be coded manually. This design appeared to reduce the amount of incorrect or questionable data entry.

Data validation may be accomplished in four different ways:

1. One-way data flow with remote validation (Fig. 2).
   
   Once data are entered into the web pages and sent to an end-point (e.g., database, spreadsheet, or e-mail) they are checked and validated remotely by the owner of the end-point. Any feedback is sent to the user with time delay because there is no immediate interaction between the database and the user. This is the easiest way to set up a data entry Web site, but it lacks automatic and immediate interactivity.
   
2. One-way data flow with client-sided validation (Fig. 3).
   
   Validation of the entered data can be performed on the client (browser) side. This is mostly done with JavaScript, a program language developed to run on browsers. The JavaScript code is sent with the HTML code (hypertext markup language, the computer language that defines the way web pages are displayed on a computer) of the web page to the user. It can react to false data entry or missing data with a message box informing the user. The advantage is that basic data check is performed before they are being sent. Yet, client-sided data validation has some major disadvantages (other than the need of programming knowledge). Because the JavaScript code is sent with the HTML code to the browser, the web pages are larger and this may lead to a longer download time, particularly for users with a slow Internet connection. In addition, anyone can see the programming code, a feature not always in the interest of the web site designer.
   
3. Two-way data flow with server-sided validation (Fig. 4).
   
   After the data are entered and sent to the server they are validated and checked by a program residing on that server. Depending on the validation result, a new web page is generated by the web server and sent to the user with, for example, a request to reenter incorrect data or to confirm that all data have been correctly entered. With this technique, there is no extra load on the web pages and the programming code is not visible to the user. The main disadvantage (other than the need for extensive programming knowledge) is that the data first have to be sent to the server and a response has to be sent back to the user, the so-called "client-server-client round trip." This may lead to a higher traffic workload, which can be annoying, especially to the user with a slow Internet connection.
   
4. Two-way data flow with server and client-sided validation (Fig. 5).
   
   The combination of the server and client-side interaction seems to be the ideal solution from the point of user friendliness. Simple data validation is performed using client-side scripting, and more complex tasks are performed on the server side. Yet its implementation is time consuming and complex.
   

View larger version (10K): [in this window] [in a new window]   Figure 2. One way data flow with remote data validation.

View larger version (16K): [in this window] [in a new window]   Figure 3. One way data flow with client-sided data validation and server-sided data storage.

View larger version (18K): [in this window] [in a new window]   Figure 4. Two way data flow with server-sided data validation.

View larger version (19K): [in this window] [in a new window]   Figure 5. Two way data flow with client-sided data validation.

Remote-sided databases lack interactivity and are less user-friendly compared to server-sided databases. On the other hand, security is easier to handle on a remote-sided database (i.e., a computer behind a firewall program or even without direct connection to the Internet) than server-sided ones (i.e., a computer connected to the Internet). Server-sided databases may be prone to hacking, which is less possible with a remote-sided database. Backing up data is easier and safer for remote-sided than server-sided databases. When backing-up a server-sided database to a local backup system the data have to be transported over the Internet (with encryption technology). Depending on the speed of the Internet connection, this can be time consuming.

It is possible to set up one’s own server, which resides physically in the coordinator’s location. This allows the study coordinator to have more control over a server-sided database. But this approach requires considerable experience and investment in hardware, software, and manpower.

Microsoft FrontPage allows building and designing Web site without any knowledge in HTML. With its "What you see is what you get" approach, the user can easily design a webpage and Microsoft FrontPage immediately creates the corresponding HTML code. Furthermore, no knowledge in any programming language is needed to use the built-in client and server-sided scripting functions for data validation and handling. Microsoft FrontPage is an inexpensive, easy-to-learn, and easy-to-use web-authoring tool.

The web site implemented in the current report had one-way data flow and combined client-sided and remote data validation. Although FrontPage has built-in functions for server-sided database handling, for security reasons we preferred to choose a remote-sided database approach and to have the database reside physically on a computer at Hadassah.

The lack of interactivity between the user and the database resulted in a large workload, as for every incorrect or missing piece of data that slipped through the initial client-sided check, the investigators had to be contacted by e-mail and the corrected data had to be entered manually. To reduce the workload, the use of free-text fields should be minimized or even avoided.

After the data were entered through the Web site, server-sided FrontPage functions (so called "FrontPage Server Extensions") sent the data by e-mail to an account on the server. Although the e-mails were then downloaded automatically to the research computer at Hadassah, the rest of the data importing process (opening the e-mail, saving it as a text file, and importing the data into the database) was done manually. This was a very time consuming process, which can now be done automatically using various commercially available add-in programs for automatic e-mail handling.

During implementation and use of the system, several problems were encountered. Among the main problems was the web site’s lack of interactive features.

1. Data entry could be done only once the patient reached the end of the study protocol because there was no possibility of successive data entry. The time span of patient observation ranged from hours to several months so the data had to first be collected using hard copy versions of the various forms.
   
2. If data entered on the Web site were lost before being sent to the server, they had to be reentered. There was no possibility of saving a web page, including the data entered on the client-side, for later use.
   
3. Investigators did not receive immediate confirmation of their data’s safe arrival.
   
4. Several investigators complained that some of their data "got lost" in cyberspace. Unfortunately, this was impossible to verify. It is possible that data got lost on the way from the investigator’s computer to the server when poor or slow Internet connections were used. But it is impossible to exclude incorrect data processing by the server-side FrontPage extension. These data had to be re-entered.
   
5. Investigators had no ability to review the data sent via the web site.
   

Security, next to user friendliness, is the most important issue when building a web site for data entry (1,9). Although there are several techniques to ensure data safety (e.g., usernames, passwords, data encryption, firewall programs), no system is foolproof. Besides technical issues, other precautions have to be considered, e.g., patient data should be depersonalized using a coding system to prevent patient identification by unauthorized persons.

For data safety reasons we decided to use a remote-sided database concept. Our database resided on a personal computer connected directly to the Internet and protected by a firewall program, and it was therefore "invisible" to the Internet community and potential hackers. The down side of this security feature is lack of interactivity between the person entering data and the database. This led to a large workload for the coordinating center, which had to manage and validate incoming data. With the technologies for data encryption available today, secure database management can be done over the Internet, probably as safely as with older paper systems (1).

Transport of the data (from the client to the server and from the server by e-mail to the coordination center in Jerusalem) was done over the Internet in an unsecured manner (without data encryption). It was assumed that the chance of somebody hacking these e-mails (which contained almost only numbers without any information about the nature of the data or details identifying the patients) and breaking confidentiality was very remote.

The results of the anonymous survey of investigators showed a very good acceptance of this Web site design (Table 1). Only the interactivity of the Web site was rated lower (median 6, interquartile range 5–8) than the overall satisfaction (median 8, interquartile range 8–9). As discussed above, the lack of almost any interactivity is the major drawback in choosing a one-way data flow system with remote data validation. Unfortunately only 65% of the centers replied to the survey, so it is possible that some nonresponders disliked this design. Oral feedback received at several meetings with the investigators was generally very positive.

	Conclusions

Top Abstract Introduction Methods Results Discussion Conclusions References

 
The Internet is an excellent tool for collecting and transmitting data for multicenter studies. It is possible to build a Web site for data entry without much programming experience using commercially available programs. Such a Web site allowed rapid, inexpensive and easy transfer of data between researchers and the coordinating center. It also permitted easy communications between the coordinating center and the investigators. Many technical and practical lessons were learned from this experience so that further systems may be more efficient and reliable.

	Footnotes

 
The present study was funded as part of the European Concerted Action project ETHICUS "End of Life Decision Making and Life Ending Procedures in European Intensive Care Units" funded by the European Commission (Contract No. PL 963733), in part by grant no. 4226 from the Chief Scientist’s Office of the Ministry of Health, Israel, in part by the European Society of Intensive Care Medicine, and in part by the Walter F. and Alice Gorham Foundation, Inc. The authors are solely responsible for the publication, it does not represent the opinion of the European Commission and the Commission is not responsible for any use of the published data.

Accepted for publication July 26, 2004.

Address correspondence to Alexander Avidan, MD, Department of Anesthesiology and Critical Care Medicine, Hadassah - Hebrew University Medical Center, Jerusalem, Ein Karem, POB 12000, Jerusalem, 91120, Israel. Address e-mail to alex{at}avidan.co.il.

	References

Top Abstract Introduction Methods Results Discussion Conclusions References

 

1. Marshall WW, Haley RW. Use of a secure Internet Web site for collaborative medical research. JAMA 2000;284:1843–9.[Abstract/Free Full Text]
2. Sprung CL, Cohen SL, Sjokvist P, et al. End-of-life practices in European intensive care units: the Ethicus Study. JAMA 2003;290:790–7.[Abstract/Free Full Text]
3. Sippel H, Eich HP, Ohmann C. Data collection in multicenter clinical trials via Internet: a generic system in Java. Medinfo 1998;9:93–7.
4. Mezzana P, Madonna Terracina FS, Valeriani M. Use of a web site in a multicenter plastic surgery trial: a new option for data acquisition. Plast Reconstr Surg 2002;109:1658–61.[ISI][Medline]
5. Dorman K, Saade GR, Smith H, Moise KJ Jr. Use of the World Wide Web in research: randomization in a multicenter clinical trial of treatment for twin-twin transfusion syndrome. Obstet Gynecol 2000;96:636–9.[Abstract/Free Full Text]
6. Siegel RM, Kiely M, Bien JP, et al. Treatment of otitis media with observation and a safety-net antibiotic prescription. Pediatrics 2003;112:527–31.[Abstract/Free Full Text]
7. Marks R, Bristol H, Conlon M, Pepine CJ. Enhancing clinical trials on the internet: lessons from INVEST. Clin Cardiol 2001;24:V17–23.[Medline]
8. Pepine CJ, Handberg-Thurmond E, Marks RG, et al. Rationale and design of the International Verapamil SR/Trandolapril Study (INVEST): an Internet-based randomized trial in coronary artery disease patients with hypertension. J Am Coll Cardiol 1998;32:1228–37.[Abstract/Free Full Text]
9. Norifusa M. Internet security: difficulties and solutions. Int J Med Inf 1998;49:69–74.

This article has been cited by other articles:

				B. M. Ilfeld, V. J. Loland, J. C. Gerancher, A. N. Wadhwa, E. M. Renehan, D. I. Sessler, J. J. Shuster, D. W. Theriaque, R. C. Maldonado, E. R. Mariano, et al. The Effects of Varying Local Anesthetic Concentration and Volume on Continuous Popliteal Sciatic Nerve Blocks: A Dual-Center, Randomized, Controlled Study Anesth. Analg., August 1, 2008; 107(2): 701 - 707. [Abstract] [Full Text] [PDF]

				J. H. van Oostrom Web-Based Data Collection: Security Is Only as Good as the Weakest Link Anesth. Analg., December 1, 2005; 101(6): 1888 - 1888. [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Avidan, A. Articles by Sprung, C. L. Search for Related Content PubMed PubMed Citation Articles by Avidan, A. Articles by Sprung, C. L.

Anesthesia & Analgesia® is published for the International Anesthesia Research Society® by Lippincott Williams & Wilkins with the assistance of Stanford University Libraries' HighWire Press®. Copyright 2006 by the International Anesthesia Research Society. Online ISSN: 1526-7598   Print ISSN: 0003-2999