Site statistics are available to the site administrator online from LinkMedica's back end based on Web site log files. Total number of registered users at the end of each month and number of unique diary users each month (number of users entering diary values at least once per month) were extracted and plotted against time for visual inspection.

Two user surveys were carried out. The first was an online pop-up survey targeted at all users visiting the site from April 29 to May 30, 2002. When visiting LinkMedica during this period, the user was presented with a pop-up window asking the user if he or she was willing to participate in an online survey. Except for the introductory text, a yes and a no button were the only elements on the pop-up form. If the user pressed the no button, the window was closed. If the user pressed the yes button, he or she was redirected to the survey form. Regardless of the button pressed, a cookie was set on the user's hard disk to prevent the pop-up window appearing on subsequent visit. If the pop-up window was closed by other means-eg, by clicking the cross in the upper right corner-no cookie was set. No attempts were made to prevent users from submitting more than one survey form by filling in the form on different computers or by deleting the cookie from their own hard disk. We assumed, however, that the risk of a significant number of users doing this was negligible.

The response rate was not monitored, but the number of submissions was compared with the number of unique visitors (by IP-address) in May 2002.

The survey questionnaire contained 17 questions. In this article we present the results of 7 selected questions: "Your age?"; "Your gender?" (male, female); "What is your background?" (patient, relative, health care professional); "What is your primary reason for visiting LinkMedica?" (seeking information, seeking advice, asthma diary); "How often do you visit LinkMedica at present?" (daily, weekly, monthly, less than monthly); "How often do you intend to visit LinkMedica in the future?" (daily, weekly, monthly, less than monthly); "How do you rate the quality of LinkMedica?" (very good, good, poor, don't know). The last question was asked for each subsection of LinkMedica (diary, knowledge centre, forum).

Complete results from the full questionnaire are available in an internal AstraZeneca report, which is available free (in Danish) to anyone interested.

Males and females were compared with respect to age distribution, reason for visiting LinkMedica (information, advice, diary), and user's background (patient, relative, health-care provider).

The second survey, a mailed questionnaire, was sent to all health-care providers that- according to AstraZeneca's customer database-had received a user name and password for LinkMedica. The questionnaire was in two sections. The first section of 4 questions was intended for all respondents. The second section of 15 questions was intended for those who, in their own opinion, had ample experience in using LinkMedica. Only the results from the first section are presented in this article. The full report (in Danish) is available free to anyone interested. The questions from the first section were: "Your profession?" (physician, nurse, secretary, other); "Have you heard of LinkMedica?" (yes; no); "Do you think that there is a need for Internet tools like LinkMedica in medical practice?" (yes, no); "Do you ever use LinkMedica in collaboration with your patients?" (Yes-frequently, Yes-sometimes, I have looked at it-but did not find it useful, No, No-but I would like to try).

The questionnaire results allowed us to select persons representing different types of LinkMedica users for interviews; respondents of both surveys were asked to provide name, address, and phone number if they were interested in being interviewed.

In-depth qualitative interviews were conducted with users in order to get an understanding of the patients' approach to their illness and their use of the Internet and the doctors' approach to the use of Web-based monitoring systems.

A total of 15 users were selected from respondents who were willing to be interviewed for semi-structured interviews: 8 patients, 2 mothers of children with asthma, and 5 GPs. They were selected to represent: 1) male and female users, 2) users of AMC and of KC and Forum, 3) frequent visitors and occasional visitors, and 4) health-care providers, patients, and relatives.

The interviews were designed to address three issues:

The perspective of the qualitative method is to understand the world as inter-subjective-to understand the world from the point of view of those who live in the world. The purpose of this interpretative approach is to understand social phenomena-to understand the lived experience and the complex world this experience takes place in [14,15].

All interviews were conducted as semi-structured qualitative interviews according to Kvale's criteria for conducting and analyzing qualitative interviews [16]. They were taped and transcribed. Each interview was broken down into thematic units and these were compared across interviews. As the qualitative method provides insight into the inter-subjective world, it is not possible to quantify the data: they are interpreted as themes from the lived experience of those interviewed.

The starting point of the interviews was the patients' relation to their disease, and both doctors' and patients' strategies for information seeking and use of the Internet. Themes that appeared during the first interviews were pursued in later interviews.

In February 2003, a total of 7653 diary users were registered on LinkMedica. The number of unique diary users at that time was 138 per month.

Figure 2 shows that the growth rate of user numbers falls into three phases: In the first 3 months after launch, more than 2000 new users registered. During the next year and a half approximately 4000 users registered, and during the last year there was a growth of approximately 50 new users per month.

The trend in unique diary users is more uneven. After a peak of 100 diary users in May 2000, the number decreased until November 2000, when only 9 users kept an online diary. During the next year the number of diary users increased (with a decrease during holiday seasons) to a maximum of 307 in January 2002. This increase occurred at the same time as AstraZeneca started marketing LinkMedica. After this, the number of unique diary users was steady for a period of about 4 months. However, in February 2003, the diary user number had declined to 138 per month.

Figure 2

Number of unique users that have entered their diary per month (bars) and total number of registered users (line)

Between April 29 and May 30, 2002, 85 users responded to the online pop-up survey. Compared to a total of 3689 unique visitors in May 2002, this gives an estimated response rate of 2.3%. Of these 59 (69%) were patients, 12 (14%) were mothers of children with asthma and 8 (9%) were health-care providers (see Table 1). Inter-quartile age range was 29 to 43 years.

Because of a programming error in the survey, gender was not recorded for 13 users. Of those remaining, two thirds were female. Mean age was 36 years for females and 41 years for males. This difference was not statistically significant (p=0.12). The age distribution for males and females respectively is shown in Figure 3.

User background and reason for visiting LinkMedica differed somewhat between males and females. Most females were seeking information and advice (49%), while the majority of males gave the asthma diary as primary reason (56%) (Table 2). Only 1 of 7 health-care providers was female. This difference between male and female users with respect to their background (patient, relative, health-care provider) and their reason for visiting LinkMedica (information, advice, diary) was statistically significant (p<0.05, chi square test).

Seventy-two percent reported that they visited LinkMedica at least once a month, and 92% reported that they expected to visit LinkMedica at least monthly in the future.

When asked how they perceived the quality of LinkMedica, the majority of users answered that the quality was good or very good. However, for each main section a rather large percentage of users answered that they did not know (Table 3).

Figure 3

Age distribution of users

Out of 489 questionnaires mailed to health-care providers that-according to AstraZeneca's customer database-had been given user name and password for LinkMedica, 131 were returned (response rate 26.8%). Among the respondents there were 127 (97%) physicians and 4 (3%) nurses. Fifty-one (39%) used LinkMedica intermittently or frequently for their asthma patients. Questionnaire results are summarized in Table 4 and Table 5.

A total of 15 one-to-one in-depth interviews, each lasting approximately 60 minutes, were conducted.

Surprisingly, we found that the users having most difficulties with LinkMedica were the doctors, whom we expected to be confident computer and Internet users. None of the patient users expressed any difficulties whatsoever using LinkMedica. The main complaint from doctors was that the login procedure is complicated and time consuming. The doctors have difficulties managing different user names and passwords for different services. Even though changing user name and password to something easy to remember is straightforward in LinkMedica, none of the doctors did this.

From training sessions with GPs, it is our personal experience that doctors find it difficult to use more than one application at a time. In general, doctors are satisfied with their electronic patient record system available from their desktop, but they do not feel confident in "windows juggling." This prevents them from having LinkMedica at hand whenever an asthma patient comes to the clinic.

In this sense, lack of practice and confidence hinders the use of LinkMedica. But surprisingly, only doctors seem to have this problem. This observation could be the result of a selection bias, leaving only computer literate patients for survey and interview. Although this may be partially true, we do not believe this fully explains why patients apparently have fewer problems than doctors accessing LinkMedica. It may be that doctors have higher demands and are more critical because of time constraints in their work. The fact that doctors on average are older, have less experience with the Internet, and seem to use the Internet less than patients for general information seeking may also play a role.

Easy access to the Internet is critical for users' experience with LinkMedica. In 2002, 76% of Danes had access to the Internet either from work or home, and 56% used the Internet at least weekly (38% daily) [18]. Thus, in Denmark, Internet availability is hardly a barrier. But speed of connection may be. In the surveys, we did not ask how the users connected to the Internet. But in the interviews, some users complained that the connection and logon time through an analog modem was an obstacle. Although our impression is that LinkMedica loads faster than many other Web sites with comparable content, booting the computer and dialing up with a modem may take several minutes. In comparison, it usually takes less than 20 seconds to fill in the diary and receive the feedback message. As broadband connections become more available and affordable in the future, we expect these problems to diminish.

In our opinion, an interesting finding from this study is how much users' everyday lives interact with LinkMedica. As an example, take the doctors' time schedules. They have about 10 minutes per patient. This, together with the fact that most doctors do not feel confident in using the Internet, or even their computers, has an enormous impact on how the doctors look at LinkMedica. The end result is that not a single one of the doctors that we interviewed used LinkMedica on a regular basis, despite their positive attitude. Furthermore, the doctors were mistaken in their views of who would use LinkMedica. According to the doctors, the typical LinkMedica user would be a young, sporty man, when in fact, the typical user is a mother around 40 years of age.

The following is another example of how everyday life influences the use of LinkMedica: Typically, the diary users measure their peak flow in the morning and write down the value. They collect the values for a week and type them in later, during the weekend. This way, feedback messages are received days after the condition that triggered them, and the whole idea of immediate dose adjustments due to changes in symptoms or peak flow is lost.

We are convinced that this problem has something to do with the Internet still being separate from the rest of people's everyday lives. The Internet is something you actively connect to, not something that is just there like the telephone or the television. In this sense, LinkMedica and other Internet-based disease management systems are ahead of their time, and we would expect these programs to gain popularity and usability as the Internet gets more integrated into our everyday lives. Currently, we are investigating other means of connecting to LinkMedica, eg, short messaging system (SMS) and general packet radio service (GPRS). These technologies have the advantage of being closely integrated into people's everyday lives and are immediately available to anyone, anywhere without the inconvenience of having to start the computer or wait for slow dial-up connections.

A number of patient subtypes were identified: controllers, neglecters, professional mothers and emotional mothers. The doctors came in two groups: users (or more correctly, former users) and potential users, who had no prior experience with LinkMedica but who were interested in trying it. During the interviews, it became obvious that these highly different user types have very different requirements and expectations of LinkMedica.

Different parts of LinkMedica (AMC, KC, Forum) are considered important depending on the user's perspective: inside-out or outside-in. The doctor and the controller, together with the professional mother, expect the diary (AMC) to function without a hitch. News, discussion forums, and ask-the-expert sections are merely distracting elements preventing them from having fast access to the diary. If they ask for information, they want it to be as complete as possible, enabling them to decide for themselves how to act. They prefer evidence-based articles and expert opinions to news and advice from other users. The neglecter and the emotional mother, on the other hand, seek information only when they need it. They expect concise information and concrete advice to help them in their current situation. They usually avoid scientific articles and expert opinions unless they are directly applicable to their current needs. They are not interested in monitoring their disease using the diary.

Creating a Web site that seeks to satisfy such conflicting needs and user perspectives may not be a good idea. For future Web projects like LinkMedica, we suggest that the target users be defined clearly from the very beginning. For existing Web sites of complex nature like LinkMedica, it may be worthwhile considering a split into several more focused sites, which may, of course, be interlinked.

It is an interesting observation that diary users (controllers) were the ones least likely to follow the advice in the feedback messages. As mentioned previously, not a single interviewee who received an alert message instructing him or her to increase the dose of steroid followed this advice. This observation calls into question the whole idea of the diary design as it is today. If the users most likely to use the diary are also the users least likely to follow specific advice from feedback messages, it might be worthwhile reconsidering the format and the content of the feedback messages.

The idea of having an electronic asthma diary with an "intelligent" feedback system was to support and educate asthma patients in self-management. In this respect, the diary may serve as a daily consultation with a virtual asthma expert. This study taught us, however, that an important difference between virtual and real experts is that patients do not readily accept advice from a virtual expert if this advice conflicts with the patient's own previous experience and attitudes in general, eg, that they are opposed to the use of steroids. The lesson learned is that no matter how intelligent and how well supported by acknowledged experts, a computer system cannot replace real face-to-face contact between doctor and patient. In future versions of LinkMedica and similar systems, we suggest that the very detailed (and complicated) feedback system be replaced by a simple "traffic light" approach. For example: "If your asthma diary says green, all is well. If it says yellow two days in a row or more, or red on a single day, contact your asthma doctor."

From classical behavioral psychology we know that in order to reinforce certain behavior, the latency time between behavior and reward must be short. This phenomenon has great implications for treatment of asthma.

In general, it takes time to achieve asthma control. The effect of inhaled steroids on symptoms and exacerbations resulting from previously inadequate treatment may be delayed weeks or months after start of treatment. Furthermore, relapse after cessation of treatment may also be delayed. Since the reward (improved health) is delayed, it can be difficult for the patient to understand why taking regular medication is important. This latency problem is probably one of the reasons that inhaled steroids are being used much less than recommended [19] and also a main reason why patients tend to stop using LinkMedica after a short time.

As with medical treatment, self-management using an electronic asthma diary may take some time to prove its value to the patient. The immediate advantage of measuring peak flow, turning on the computer, connecting to the Internet, logging on LinkMedica, and entering diary values is simply not big enough for the patient to continue doing this for longer periods. Even if a patient has experienced improved asthma control from using LinkMedica, the advantage of continuing use may not (from the patient's viewpoint) justify the inconvenience.

In principle, there are two ways to solve this problem: by improving the accessibility of the diary, which could be achieved by, for example, improved Internet connections, mobile phones, or wireless devices (peak flow meters, electronic dispensing devices, etc.); or by increasing the immediate advantage of diary entry. There may be several ways to achieve the latter. Further studies need to be carried out to reveal whether it is the technology or the lack of immediate advantages that creates obstacles. User studies and explorative design methods [20] can help to clarify the users' needs.

As an example, teenagers would probably adhere more to the diary if they were permitted a small number of free SMS messages for their mobile phones after each diary entry, while others would be encouraged by reimbursement of a portion of their medicine costs linked to how often they entered their diary values. The decrease in unscheduled doctor visits, days off work, etc. resulting from the improved health of users would probably offset the cost of such reimbursement programs. Rewards may also be "virtual": small games where points earned from filling in the diary give access to new game levels might encourage children (and some adults) to adhere to the diary. General information on the user's health-eg, weekly or monthly messages with overall information about the user's asthma, whether it has improved or worsened, and what to do about it-might also prove useful.

Given 3 years of experience developing and marketing LinkMedica together with the results of this study, we are able to suggest improvements that would make new users more likely to hold onto their Web-based asthma diaries for longer periods:

In general, LinkMedica is regarded as a very reliable and advantageous system by both patients and doctors. However, only a few users are using LinkMedica as intended, and most users, patients as well as doctors, stop using the diary after a short time. There are several reasons for this, the main reason being that the Internet in general and LinkMedica in particular are still not integrated into people's everyday lives. Consequently, if LinkMedica is to become more popular, it needs to be adapted to the conditions of the users, so it becomes a natural and integrated part of their everyday lives.