Marc D. Kohli, MD, Indiana University; Darel Heitkamp, MD; John Bradshaw, MD

Aggregating heterogeneous information resources with a wiki is useful for resident education.

wi·ki (wĭk'ē)
n. pl. wi·kis
A collaborative website whose content can be edited by anyone who has access to it.

[Originally an abbreviation of WikiWikiWeb, software developed by American computer programmer Howard G. Cunningham (born 1949) : Hawaiian wikiwiki, quick + web.][1]

Wikis are becoming an increasingly important tool for collaboration, as evidenced by several recent publications on the subject.[2-11] To our knowledge, no one has described using a wiki as a collaborative tool for resident education.

Educating a radiology resident is an amazingly complex procedure that requires hours of self-study, an enriching didactic environment, as well as extensive clinical experience. Residents are increasingly expected to track information from multiple sources. At our institution we have three separate PACS, three different electronic medical record systems, an e-mail system, a department web page, and several different hospital web pages. Prior to implementation of the wiki, policies were disseminated via e-mail, and would be archived by each user. This lead to fragmentation of policies and a data management challenge. Additionally, some information, particularly dictation templates, were stored locally on workstations scattered throughout the enterprise and were not widely accessible.

Due to the near infinite flexibility and collaborative nature of wikis, they provide an ideal environment to combine references from several heterogeneous systems into an “one-stop-shop.” Additionally, because anyone can edit a wiki, users are empowered to change out-of-date information, which improves the quality and timeliness of information when compared to having a single central editor. While the validity and accuracy of material contained within wikis can be controversial, even Encyclopedia Britannica recently announced that they would begin using a modified Wikipedia model.[12]

The ideal wiki would be diverse in the content that it could support, including file attachments and images. The ability to quickly evaluate recent changes and to have an easy mechanism for correction is also important. The ideal wiki should be easy for an administrator to set up and maintain, be easy for users to edit, and be able to leverage existing authentication structures. This last piece is incredibly important, as we all have growing lists of passwords that are increasingly difficult to manage.

After an extensive review of available wiki packages, we decided that dokuwiki (http://wiki.splitbrain.org/wiki:dokuwiki) best fit our needs. One of the main features that led to the selection of dokuwiki is that it uses plain text files to store pages rather than a separate database server. The lack of a database provides two main benefits to a small website - easy backup and fewer software packages to track and update. Another important feature of dokuwiki is that it is open source and easily integrated into our existing university-supported authentication back-end. This integration eliminates a barrier to entry by leveraging our user’s existing credentials. The login integration was the only custom development that was required.

User-based authentication allowed us to host sensitive content, including resident phone numbers and addresses. Fortunately, the dokuwiki authentication structure also allows authentication based on ip address, so that many pages are accessible without a login, by virtue of using a computer within the hospital. We have several user categories including: resident, fellow, faculty, and staff.

Dokuwiki is written in the PHP programming language, which is widely used on the Internet and can run on a variety of systems including Windows, Linux, and Mac OS X. Our initial server was a repurposed desktop computer - Pentium II processor with 256 MB of RAM and a 20 GB hard drive running Red Hat Enterprise Linux 6, Apache, and PHP. After a 6 month trial period, the wiki was moved to a department server - Quad-Core Xeon 2.8GHz server with 4GB of RAM, 60 GB hard drive running Windows Server 2003 and Apache. The move was primarily performed to integrate the wiki into the department backup routine. The only noticeable change in performance between the two servers was the speed with which search queries could be returned.

In order to assess the wiki’s effect on the residency program we designed a survey to measure resident use, as well as participation in editing and maintenance. Survey questions are listed in Table 1. The survey was distributed over e-mail, and responses were tabulated and reviewed.

Table 1 – Survey Overview

Fifty-one out of sixty residents responded to the survey (85%). All responding residents correctly identified the wiki URL and reported having visited it at least once. On average, residents reported visiting the wiki 5.6 times per week, with one resident reporting 20 times per week. Most residents (78%) reported knowing how to edit an existing page. However, far fewer reported knowing how to add a new page (38%). Despite such a large percentage of residents reporting how to edit, only 35 (69%) had edited content. Due to the flexible nature of a wiki, it is important to stay up-to-date by occasionally scrolling through the list of recent changes. Only 43% of residents reported knowing how to find recent changes. A majority of residents planned on adding content in the future (71%). Residents were also asked to report what information they accessed most frequently. Resident responses were aggregated and compared with server logs (Figure 1). On average, residents thought the wiki was very useful (Leikert scale) and reported that it replaced a variety of sources, including word of mouth (Table 2).

Figure 1 – What do you access most survey response vs. server logs

Table 2 - What information sources did the wiki replace?

We attribute much of the success of the wiki to the quality of the content. During the early days of our wiki, there was a push from the chief residents to include high-yield content, including rotation and call schedules, phone number listings, and general on-call help. All the top URLs and survey responses were part of the original data available on the wiki, and remain the most popular pages.

Another attribute that has been critical to the wiki’s success has been the presence of administrative oversight. In our department, the two chief residents maintain the wiki with daily surveillance (looking for spurious edits), and adding new content. To date, we have not had a single episode of malicious editing.

Overall, our residents see the wiki as a valuable tool, replacing several other sources of information, including e-mail and word-of-mouth. We hope that this technology can be applied in other residency environments and welcome questions regarding our implementation.