Why Are Health Care Interventions Delivered Over the Internet? A Systematic Review of the Published Literature

Background: As Internet use grows, health interventions are increasingly being delivered online. Pioneering researchers are using the networking potential of the Internet, and several of them have evaluated these interventions. Objective: The objective was to review the reasons why health interventions have been delivered on the Internet and to reflect on the work of the pioneers in this field in order to inform future research. Methods: We conducted a qualitative systematic review of peer-reviewed evaluations of health interventions delivered to a known client/patient group using networked features of the Internet. Papers were reviewed for the reasons given for using the Internet, and these reasons were categorized. Results: We included studies evaluating 28 interventions plus 9 interventions that were evaluated in pilot studies. The interventions were aimed at a range of health conditions. Reasons for Internet delivery included low cost and resource implications due to the nature of the technology; reducing cost and increasing convenience for users; reduction of health service costs; overcoming isolation of users; the need for timely information; stigma reduction; and increased user and supplier control of the intervention.


Introduction
The Internet is still a relatively new medium for seeking and delivering health care, although this use is increasing rapidly [1,2] and includes health information seeking [3], Internet-based peer support groups [4], online health consultations [5], and delivery of health interventions [6]. Some pioneer researchers have published studies that evaluate health interventions delivered directly to the users via the Internet for their ability to improve the health status of their users. In this paper we review their work, focusing on the reasons why these authors chose to use the Internet for delivery of a health care intervention. Our aim was to consolidate the findings from these early research papers to inform the development of future research. We include only health interventions in which the networking provided by the Internet is a component of the intervention. This is to distinguish them from other media such as print material, CD-ROM, and video. We reflect on the drivers to using the Internet for the delivery of health care. This paper does not review the outcomes of the interventions.

Identification of Studies
The initial identification of studies used five sources: three existing systematic reviews of eHealth interventions [7][8][9], a hand search of JMIR (vol 1(1) to vol 8(1)), and our own previous qualitative review of the literature concerning the Internet and consumer health information [10]. This latter review involved collation and identification of relevant literature through systematic searches of electronic bibliographic databases covering health and social sciences literature (1990 to December 2003, including Medline, HMIC, CINAHL, Sociological Abstracts, Sociofile, and Web of Science). We used search terms such as "Internet," "electronic mail," "computer communication networks," and "health information," "communication," or "health informatics." Two investigators reviewed the list to identify potentially relevant articles. We worked in pairs, reviewing the search results to identify relevant intervention studies. We did not set out to identify every published eHealth intervention paper, but aimed to search the majority of the available literature in a systematic way for a meaningful overview of the field.

Inclusion and Exclusion Criteria
Inclusion and exclusion criteria were applied to the studies identified from the three sources described above. We included only peer-reviewed full research papers. We defined intervention studies as the use of information and communication technology to deliver a specific health intervention to a client or patient group. The study had to include a health-related outcome as part of evaluating the intervention, and the intervention had to include use of the networking provided by the Internet. Networked features included the use of email to contact a therapist, the use of chat rooms or bulletin boards by client/patient groups, or the transfer of personal health data via the Web between a health care site and personal network access (eg, between a clinic and patient's home). Studies with no networked features, such as computer-based decision support systems delivered from a CD or interventions where there was no use of the Internet beyond delivery (ie, they could have been delivered by a CD), were excluded. A further key characteristic of the Internet is its accessibility via a networked computer anywhere and anytime. Hence, we excluded studies in which access to the intervention was provided only in the clinical setting as use of the intervention is restricted in place and time. It is also possible that effectiveness may be influenced by the clinical setting.
Our review focused on the use of the Internet for delivery of the interventions and therefore did not include non-Internet based telemedicine studies. The focus was on specific interventions for specific health problems, so we excluded interventions involving the provision of general Internet access such as home computers, Internet kiosks, or training in use of the Internet even if the outcomes included health related measures. We only included interventions in which the individuals using them were known to the health care professional or organization delivering the intervention to be sure that the participants were using their real identity and responding in a genuine way to the intervention. This cannot be ensured for a study that recruits participants solely via the Web, with no direct contact between investigators and participants. We excluded studies that solely involved the placing of health information on the Web for public access, even when there was opportunity for interaction or feedback.

Analysis
When there were several papers concerning the same intervention (eg, a pilot study followed by a full evaluation), we grouped these papers together and treated them as one study. For each study, all the reasons given for delivering the health care intervention on the Internet were listed. These were the reasons the authors of the papers gave for choosing the Internet as the mode of delivery, rather than post hoc reasoning given in the discussion of the study results. We then categorized the reasons; one study could be categorized in a number of different groups. Again, we worked in pairs, comparing results and resolving any discrepancies through further examination of the papers and discussion among team members.

Types of Interventions
We found full evaluations of 28 interventions and a further 9 interventions for which only pilot work had been published (Multimedia Appendix). All the papers were from Europe, North America, or Australia. The interventions were aimed at a wide range of conditions, including cancer (3 studies), HIV/AIDS (3 studies), diabetes (3 studies), mental health (1 study), eating disorders (2 studies), and back pain (1 study). Some targeted health promotion issues such as smoking cessation (1 study), physical activity (1 study), and obesity (3 studies). Other interventions aimed to support caregivers, for example caregivers of people with Alzheimer's disease (3 studies), stroke patient caregivers (1 study), new or young mothers (2 studies), and parents of children in intensive care (1 study). One intervention aimed at supporting rural women with chronic illness. One study reported the delivery of cognitive behaviour therapy (CBT) for a number of disorders, including headache, tinnitus, and panic disorders; two other studies reported CBT delivery for depression and one for post-traumatic stress disorder. Three interventions offered education and/or communication with specialist nurses for cardiac patients. Three interventions were specifically for young people or children: one for pain self management by children in hospital, one for those with cystic fibrosis, and one for the management of encopresis.

Reasons for Internet Delivery of Interventions
The reasons cited for using the Internet to deliver health interventions included the unique advantages of the Internet technology, reducing cost and increasing convenience for users, reducing health service costs, reaching isolated or stigmatized groups, timeliness of access to the Internet, need for user or supplier control of the intervention, and research-related reasons (Textbox).
Not all the studies in the early research papers mentioned the reasons for use of the Internet. Therefore, in the following analysis, the papers referenced are the papers for which the reason for Internet use was mentioned. • Potential for reinforcing the problems the intervention was designed to help • May overcome isolation of time, mobility, and geography, but may be no substitute for face-to-face contact

Elements of future evaluations:
• Incorporate the cost not just to the health service, but also to users and their social networks • Be alert to unintended effects of Internet delivery of health interventions, and include a comparison with more traditional modes of delivery

Unique Advantages of the Internet Technology
There were 13 interventions studied [6,11,16,20,24,32,35,37,41,45,52,57,60,67,[71][72][73] for which the reason for Internet use was connected with the nature of the technology: reaching many people with just one posting, easy storage of large amounts of information, ease of updating information, providing personalized feedback, and the possibilities of broadband and video transmission. Two of these 13 studies [57,16] expressly valued the Internet for its ability to reach a maximum number of people at minimum cost. All these studies also give other reasons for Internet use.

Reducing Cost and Increasing Convenience for Users
Reducing cost and increasing convenience for the user was given as a reason for delivery over the Internet in 20 of the interventions studied [20-23,29,32,39,43-48,50,52,53,56-58, 60-63,65,66,75]. These studies targeted a range of health issues. Various aspects of increased convenience to the user were mentioned, including saving the user time, requiring less effort from the user, being more accessible, and not requiring the user to attend a particular facility. One US study [43], reporting an intervention for women with breast cancer, stated users' lack of money for a second opinion as one of the reasons for Internet delivery. Two studies advocated use of the Internet as it may reduce the loss of users from their maintenance programs for obesity [46,47].

Reducing Health Service Costs
By using Internet delivery, 14 of the interventions studied [11,12,14,15,24,[28][29][30]39,41,42,46,48,57,58,60,65,67,68,[72][73][74][75] aimed to reduce costs to health services or address a lack of provision. Of these, two studies, one on linking parents with their low-birth-weight babies in intensive care [41], and the other on the management of encopresis [60], specified reduction in health service cost as a reason for Internet use. The cost of service provision was also given as a reason by a number of other studies, but with slightly different emphases. One study saw the Internet as a cost-effective way of delivering an intervention to encourage physical activity in a broad range of people in many places [58]. Five of the interventions studied gave a lack of health service resources as their reason, two citing a lack of practitioners in CBT [11,12,42] and the others a lack of support for caregivers of those with Alzheimer's disease [24,39] and AIDS [28][29][30]. All the above studies discussed a general lack of these services. None of the studies gave a specific localized service failure as the reason for Internet delivery, but two mentioned service failure more generally. One study [60] gave the lack of physicians trained in the treatment of encopresis as the reason for Internet delivery, while another study [67] aimed to reduce barriers to nutrition education due to general practitioners' lack of skills and time. The authors of one other study [48] argued for Internet delivery because patients with diabetes have been found to have poor control despite specialist care, and their control may become even worse after the devolution of diabetes care to primary care services.
In six of the interventions studied, researchers wanted to avoid the cost to the health service of providing the intervention face-to-face, including psychological interventions for the treatment of depression [42], eating disorders [71][72][73][74][75],obesity , lack of physical activity [57,58], and a range of conditions (headache, tinnitus, panic attacks, and insomnia) [11][12][13].The authors of the latter intervention studied also argued that Internet delivery increases access to an otherwise costly therapy [14,15]. Another study [19] explicitly addressed inequalities of health care, suggesting that Internet delivery helped to overcome inequalities of access to health services.

Reaching Isolated Groups
Dimensions of isolation were given as reasons for using the Internet in 13 of the interventions studied [18,19,21,25-30, 33,34,37,43,49,52,53,75]; 5 stated geographical isolation as their reason. These studies included interventions for rural women with chronic disease [33] or diabetes [34], an intervention for women with breast cancer [43], a cognitive behavioral program for eating disorders [75], an intervention for people with post-traumatic stress disorder [53], and an intervention for children with cystic fibrosis [52]. The authors of the latter study also mentioned the advantage of providing peer support without the risk of cross infection that can be life threatening for children with cystic fibrosis.
In other studies, the cause of the isolation was not geographical. One [28][29][30] stated the isolation of people living with HIV/AIDS as a reason for Internet delivery. Several studies cited the isolation of caregivers who were unable to easily go out, such as those living with an Alzheimer's sufferer [21,[25][26][27] and young mothers with children at home [37]. The physical immobility of individuals, including fatigue and disability, was mentioned in interventions focused on people with HIV/AIDS [28] or breast cancer [43] and on children with cystic fibrosis [52].
Several studies mentioned that Internet delivery enabled users to be in contact with people with similar health issues and so receive support. The implication was that this would be unlikely to happen otherwise as the condition was rare or restricting, for example, children in pain [49], children with cystic fibrosis [52], young mothers [37], people living with AIDS [28][29][30]43], and people with type 2 diabetes [18,19].

Reaching Stigmatized Groups
The researchers of 11 interventions saw the Internet as a way of reaching people suffering from conditions that caused them to feel embarrassed or stigmatized [18,19,22,24,28,30,31,37,43,45,46,57,60,[73][74][75]. The anonymity of Internet delivery was a reason for using the Internet in the following interventions: an intervention for mental health problems [31], in which the authors considered stigma to be a problem; an intervention for people living with type 2 diabetes [18,19], in which the authors suggested that anonymity prevented people from being judged on the basis of their appearance; two interventions to improve the self-care of people living with AIDS [28,30,45]; an intervention for young women at risk of eating disorders [73][74][75]; and a support intervention for young mothers [37]. Three studies suggested that Internet delivery avoids embarrassment about the health issue for which the intervention was used. One of these was for breast cancer [43], the second referred to embarrassment about failure to lose weight in an obesity intervention [46], and the third was aimed at child encopresis [60]. One study of a support system for caregivers of patients with Alzheimer's disease [22,24] and one study of an intervention to encourage physical activity [57] suggested that Internet delivery encourages openness of communication.

The Timeliness of Access to the Internet
Several interventions [12,21,22,25,[28][29][30]37,43] mentioned the need for timely information and advice as a reason for Internet delivery, including interventions to support Alzheimer's caregivers [21,22,25], people with AIDS [28][29][30], young mothers [37], those with breast cancer [43], headache sufferers [12], and an intervention encouraging physical activity [57,58]. The suggestion was that people need information or advice at a time of crisis, for example, when their child is ill or when they are making a decision such as a change in treatment or their own behavior. The continuous access provided by the Internet was seen as helpful in these situations.

User Control of the Intervention
Many authors advocated use of the Internet because users could take control of the intervention [11,21,32,37,45,48,58,60,63,67,72,74], tailoring the information they received to their own needs. This included interventions for Alzheimer's caregivers [21], those with HIV/AIDS [45], a glucose modeling tool for type 1 diabetes [48], and an intervention promoting physical activity [58]. Other studies advocated use of the Internet because users could use the intervention at their own pace. These included CBT for depression [32] or tinnitus [11], an educational program for cardiac patients [63], peer support interventions providing young mothers with support [37] or facilitating weight loss [67], an intervention for those at risk of eating disorders [72,74], and an intervention for encopresis [60].

Supplier Control of the Intervention
For some interventions that delivered CBT as a self-help program, the Internet was seen as a potentially appropriate mode of delivery for such a structured, evidence-based intervention [12][13][14][15]32,42,57,58,66,[71][72][73][74]. One author stated that Internet delivery was superior to professional psychologists in delivering structured and standardized interventions [42]. However, in delivering these structured programs, the studies supplemented the standardized intervention through individualized email feedback, tailored information, online peer support, or a combination of the three.

Research-Related Reasons
Almost all authors justified the evaluation of Internet-delivered interventions by saying that they need evaluating or adapting for specific populations. Six studies give this as their only reason [38,40,54,55,59,61]. Most studies give examples of successful Internet-delivered interventions to support their own research. However, one study gives, as its only reason for Internet-delivered intervention, that the intervention or a similar intervention had been useful in other studies [40]. One study [54] questions whether face-to-face and online support groups for those with breast cancer would work together, and gives this question as the only reason for delivering the intervention via the Internet.
A few studies did not give a research-related reason for evaluating an Internet-delivered intervention. These studies were descriptive accounts of an intervention [33,48,51] or were evaluating the use of an Internet-delivered intervention that was in response to a specific health service-related problem [60].

Other Reasons
The following reasons, alongside others mentioned above, were also given for delivery of an intervention via the Internet: • poor information received by patients from health professionals [48] • novelty [57,58] • attractiveness of the Internet to young people and children [51] • online communication as one of the main forms of communication used by young people [60]

Discussion
We have reviewed many pioneering studies evaluating Internet use for the delivery of health care interventions and found a variety of reasons for delivering interventions through the Internet. All the interventions have been, or could be, delivered by other means. For example, support groups for isolated individuals can use more established means of communication such as telephones and post, and therapeutic programs can be delivered face-to-face. The key differences between non-Internet delivered interventions and those delivered via the Internet relate to time and place. For example, Internet support groups enable quick communication between many isolated individuals, and Internet-delivered therapeutic interventions can be taken up at any time and anywhere with Internet access.
Our literature search strategy was designed to systematically identify the majority of eHealth intervention studies meeting our inclusion criteria. However, as a qualitative analysis that aimed to explore the motivations for delivering such interventions online, it was not necessary to undertake an exhaustive search for every single eHealth study ever published in any language. This contrasts with the methodology of quantitative meta-analysis, which requires the identification of all possible studies to produce one summary result. We believe that our qualitative thematic approach met our objective and was both rigorous and repeatable. Qualitative methods of research synthesis are a relatively new area and can be very valuable in identifying lessons for future work, particularly as they do not focus solely on the results on previous studies, but also consider other factors such as the researchers' motivations. Our criteria for inclusion and exclusion of studies were designed to maintain the focus of the review on the added value from use of the Internet. Hence, they took account of the key characteristics of the Internet, particularly its networking potential and accessibility. Thus, our criteria differed from definitions of eHealth, for example, by excluding telemedicine [76] and general public access [77].
At this early stage of development, researchers should give careful thought to the reasons for using the Internet for any particular intervention. We should try to understand the unique advantages and disadvantages of Internet delivery of health care and in what circumstances Internet use could contribute most effectively to improving health. For example, why might speedier communication and flexibility of location enhance the effectiveness of the intervention? Answers may include, for example, overcoming inequalities of access to health services or encouraging openness of communication. However, to clarify the added contribution of Internet delivery over more traditional forms of delivery, evaluations should include a direct comparison between Internet-delivered interventions and those delivered by the most effective of available conventional means. Such evaluations will enable us to understand the effect of the real differences between the interventions. Few studies in our review undertook such a direct comparison.
Failing to undertake such a direct comparison may result in the failure to identify and quantify situations where face-to-face delivery is better than Internet delivery. For example, among the many studies of structured behavioral programs using Internet delivery, only one intervention [46,47] compared the benefits of this delivery method with time-intensive face-to-face therapy, and another compared it with a classroom-based intervention [70,72]. A systematic review comparing the effectiveness of Web-based and non-Web-based interventions [9] included, apart from the above two interventions, no other trials in which Web-based interventions had been compared to intensive face-to-face interventions. Undertaking an evaluation of Internet-delivered intervention without comparison may inappropriately encourage a reduction of the availability of the effective face-to-face intervention. This would work against the original motivation of the research to increase access to an effective intervention.
The design, delivery, and evaluation of an Internet-delivered intervention also need to consider the following questions: What may be the unintended harmful consequences of Internet delivery? What may be the negative effects of speedier communication and flexibility of location? For example, it is possible that providing low-cost Internet-based support for groups that are not currently provided with adequate support, such as caregivers of those with Alzheimer's disease, may reinforce the low priority of these groups for health and social services and thus increase their isolation. Providing an intervention via the Internet for individuals living with a health problem they feel is stigmatized could have the unintended consequence of the issue being less talked about outside the anonymity of the Internet and thus reinforcing the stigma (see Textbox). Although identifying such unintended consequences was not an aim of this study, it was notable that we did not identify any reports of such consequences in the papers reviewed.
Evaluations of Internet-delivered interventions should aim to ensure that they include both the benefits and potential harms of the mode of delivery for all those affected by it. For example, an economic evaluation should include not only the cost of the Internet intervention, but also costs to health services, specific services, users, and their social networks. The studies reviewed rarely included an evaluation of such indirect costs.
Although the Internet can overcome isolation of time, mobility, and geography, it may be a poor substitute for face-to-face contact with real people. The balance between use of the Internet and face-to-face contact should be carefully considered in each circumstance. This applies to structured interventions such as CBT as well as to more flexible interventions such as peer-to-peer support. In designing an evaluation, researchers should be aware that Internet-based contact may be providing something different than face-to-face contact and should seek to assess these potentially different effects (see Textbox).
A number of studies gave no reason for use of the Internet as the mode of delivery beyond stating that it exists and needs evaluating. Now that the field of Internet-delivered interventions is established, future researchers should carefully consider how the networking provided by Internet delivery may enhance the effect of an intervention. This should involve exploratory work and more explicit use of existing theory and modeling [78].
The pioneering researchers who undertook the studies reviewed in this paper were often looking to the Internet for a way to help resolve some of the current difficulties and dilemmas of health care. These included the provision of equal access to health care, limitations on resources for health care, changing roles of health professionals, and changing needs for particular skills.
Exploring the possible benefits of using the Internet to address these issues is important, but it is also important to make a meaningful comparison between using the Internet and using other more traditional ways of addressing the issues. Future research will hopefully shed more light on the benefits and disadvantages of Internet use particular to this mode of delivery.