Version of July 5, 1999

Portions of this paper have appeared in the publications of the Department of Health and Human Services Science Panel for Interactive Health Communications.

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Introduction

This paper reviews the published literature on impact of interactive health communications on demand for health services.  First a few definitions before we start.  By health communications we mean health education as well as patient provider interactions during a visit.  Thus we include in our review data about effectiveness of online health education as well as online visits.  Even though telemedicine is an interactive health communication, we do not include it in this review because a number of organizations have reviewed the impact of telemedicine separately.[1] We also exclude from our review health education efforts that are not interactive, e.g. books or home base electronic monitoring. [2]  These are important applications of technology but do not constitute examples of interactive communication with health care consumers.  Our definition of interactive health communications is not limited to complex technology. We include Interactive Voice Response (IVR) telephone technology, kiosk based services, computer programs on disk as well as Internet services.  We focus the study on consumer / patient oriented Interactive Health Communications as opposed to interventions directed at providers. 

Why a new review?

In recent years, there have been several excellent reviews of the literature on online communications.[3]  Naturally, one should ask why should there be yet another review.  Most reviews to date have reported several empirical findings, highlighting the importance of IHC by including use of interactive health communications by different groups of patients, and in some occasions the surprising positive impact of these communications on the health of the patients.  They have shown that Interactive Health Communication applications can affect cost and quality of care.  However, these reviews have not fit the empirical data into a cohesive theoretical model so that results can be understood and generalized.  Without a theory, all empirical findings stand on their own merit and do not build to a larger understanding of the phenomena being studied.

In addition, a growing concern with the quality of information has led to calls from different groups to rate and evaluate interactive health communications.[4]  But evaluation of different products is time consuming and expensive.  Providers and purchasers would like to know what they could learn from existing empirical studies.  They want to know if they can generalize from a group of empirical studies to their patients.  They want to know what types of patients will benefit from which types of interactive health communications and under what conditions.  The problem is especially acute for purchasers who are beginning to pay for these services.  In January of 1999, the Health Care Financing Administration will begin to pay for telemedicine consults to certain regions.  For these purchasers the issue of evaluating the impact of interactive health communications is no longer a theoretical possibility.  It is a significant preoccupation.

Providers of online health services and developers of interactive health communications are also perplexed.  They want to understand how to produce products that are more likely both to be used by patients and to have a positive impact on them.  For these developers it is not enough to say that interactive health communications are effective.  They need to look inside the black box and understand why is it effective so that they can copy and learn from the experience of others in the field.  In short they need a theoretical model that could explain the data.  Without a theory grounded in empirical studies, it is not possible for anyone (patients, providers, purchasers, and developers) to understand the growing literature on interactive health communications. 

This paper reviews the existing studies and posits a mechanism for how interactive health communications affect people's lives.  It both summarizes the empirical studies and  speculates about the causal links.  Thus, this paper tries to fit the reported data into a theory, which can be further refined and improved with additional empirical data. 

Why a new theory?

There are numerous well-established theories about behavior change and about health communications. Glanz and colleagues[5] review numerous theoretical models for behavior change, including:

• ·        Four theories for individual change: Health Belief Model,[6]^,[7] Stages of Change Model,[8] Reasoned Action,[9] Stress and Coping Model of change.[10]

• ·        Three theories of interpersonal health behavior: Social Cognitive Theory,[11] Social Support theory,[12]^,[13] and Patient-Provider communication.[14]

• ·        Four theories of community and group intervention models: Community Organization,[15] Diffusion of Innovations,[16] Organization change,[17] and Communication Theory.[18]^,[19]

•          In addition, other scientists have combined several existing theories into more broad set of models for behavior change.[20]-[21][22][23]

Given such a crowded field of theories about behavior change, it is natural to ask why there is a need for yet another one?  We have three reasons for trying to propose a new theory.  First, none of the theories proposed can explain some of the contradictory recent findings in the field.  For example, we will shortly review studies that show IHC applications both increase and decrease clinic visits. We find it difficult to explain many empirical findings with existing theoretical models and believe that a new and more detailed model can do a better job.

Second, none of the theoretical models pay close attention to the networked nature of many interactive health communications.  In many of these IHC applications, patients communicate with each other.  When the network is small, few patients join and use the network.  But when the network is large and it includes people they care about, many patients join and use the network.  For example, the use of email may depend on who can be reached by email. When you can reach the people you care about, you will use email more frequently.  Thus, email may be more likely to be used in larger networks than in smaller ones.  Several of the existing theoretical models, e.g. health belief model or stages of change model, ignore or are silent about the role of networks in creating social norms and habits.  Other theories, such as social support or innovation diffusion, explicitly refer to networks, but see people being influenced by networks of people and not necessarily by communication networks.

Finally, our last reason for providing yet another model is because existing theories are silent on a number of practical questions that to us are central to interactive health communications.  For example, these theories do not answer if teenagers will use interactive health communications and under what circumstances -- what about older patients.  Which group will use it more?  They do not address when in the life cycle of illness would interactive health communications have the largest impact on patients. They do not address whether there is an optimal size to online communities of patients, after which more participants will not be associated with better outcomes.  Note that many of these questions may not have been studied to date.  The point is that these questions are not raised by existing theories. Yet these questions, and many like them, are central to effective organization of interactive health communications.  Hence, we would like to propose a new model that not only integrates existing empirical data but can help us raise new questions.

Methods of review

We searched the Medline database for the following terms:  

The abstracts of the articles we found are at http://gunston.gmu.edu/healthscience/722/frsession2.htm.    In addition, we contacted several listservs and asked members of these list serves to identify for us published or unpublished studies they had underway.  We also called authors and personally talked to members of the Science Panel on Interactive Health Communications asking them to identify their doctoral students and recent publications in this area.

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Impact on satisfaction with care

Several studies report that patients are satisfied with their experience of interactive health communications.[24]^,[25] This is not surprising, as patients who are not satisfied with these services are likely not to use them and therefore will not show in studies of satisfaction of users of these services.  A better picture of satisfaction with interactive health services emerges when we examine patients’ use of these services. The average use of IHC applications is high,[26] especially for electronic support groups. For example, Brennan and colleagues report that during one-year of study caregivers to persons with Alzheimer's disease used electronic support groups twice a week each time for an average of 13 minutes.[27]   Similarly, Alemi and colleagues report that cocaine using pregnant patients used electronic services over 7 months period on an average of 3.22 times per week.[28]   Extensive use of IHC applications is one indication of patients' satisfaction with these services.

While patients' use and satisfaction of IHC applications is well understood, the impact of these innovations on overall satisfaction with health care system is not understood.  Data suggest that when patients have access to both online and face-to-face counseling, they prefer online counseling.  In particular, in an unpublished study of recovering patients who had access to both online and outpatient substance abuse treatment, 30% showed for outpatient and 87% showed for online treatment.[29] In a randomized study of recovering mothers of new infants were 8 times more likely to use electronic support groups than face-to-face groups.[30]    Similar data have been reported for adult children of alcoholics preferring online counseling and support groups to face-to-face groups.[31]  These studies suggest that use of -- and by inference satisfaction with -- face-to-face treatment may go down when online treatment is available.   Other data suggest the opposite. 

One study reported that online services improved the relationship between the face-to-face counselor and the patient.[32]  The theory we propose speculates that these contradictory findings are possible because of the extent of integration of online services with face-to-face services.  When online and face-to-face visits are closely integrated (e.g. both visits are to the same clinician) then online services may increase satisfaction with face-to-face services.  When online and face-to-face visits are not fully integrated, then online services may reduce satisfaction with face-to-face visits.

Integration of online and face-to-face services is also related to the attitude that providers have towards online services.  Providers' satisfaction with online services is not well documented. One of the few studies surveyed 325 members of American Association of Diabetes about their preferences for different methods of education including books, videotapes, computer instructions and audiotapes. [33]   Providers were least enthusiastic about computer instructions.  Providers' negative reaction to interactive health communication is surprising in light of two other facts: 

1.  Patients prefer interactive health communications to other forms of health communication,[34] and sometimes to visits.[35]

2. Randomized studies (reviewed here) show that interactive health communications are effective in changing behaviors.      

 Thus, the providers' reported negative attitudes are held despite patients' preferences and clinical studies of effectiveness.  One explanation of providers' attitudes is that they have not been exposed to the type and quality of interactive health education reviewed here.  These providers’ attitudes may change as they have more opportunities to examine interactive health communications and as the impacts of these innovations are more widely discussed.  But there might be something else at work here.  Providers' negative attitudes may be a function of practice changes that follow use of IHC.  Effective implementation of IHC requires not only substitution of books and pamphlets with IHC applications but also changes in the way clinicians interact with patients.    When information can be tailored to the patients' condition, when some components of care can be done ahead of visit, or when follow up can be accomplished without a visit, then the very nature of visits change.  It is possible that providers' poor evaluation of IHC may be a function of difficulties they face in integrating these technologies into their practices.[36]

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Impact on providers practice patterns

Interactive health communications reviewed in this paper focus on consumers and patients.  But when consumers change, providers also change their behavior. Thus, the impact of IHC on patients' behavior may be indirectly through changing providers' practice patterns. The patient may interact with IHC applications but the results of the interactions are shared with the provider who may change his/her advice to the patient.  Good examples of such applications are a number of studies of shared medical decision making and informed consent.  These studies have shown that multimedia applications can be used to assess patients' preferences.[37] [38]  Little data is available on effectiveness of shared decision making applications in changing practice patterns. 

Other examples of how IHC applications may change practice patterns and thus influence the health of patients are applications that focus on reminding patients.  Overwhelming data show that reminders can change clinicians practice patterns.[39]  With some exceptions,[40] when computers call and remind parents to visit a clinic, on time immunization and vaccination rates improve.[41]^{, [42], [43], [44], [45]} These data suggest that IHC applications can affect health outcomes by encouraging patients to show for their scheduled visit.  

Another way practice patterns might be affected by IHC applications is through computerized history taking.  Studies show that are more likely to be truthful to a computer than to a clinician.[46]  For example, Locke and colleagues found that patients donating blood are more likely to report their HIV risk factors to a computer than to a clinician.[47] These studies suggest that interactive health communications may solicit more accurate information.  If this is true, it may be reasonable to assume that the new information may change the clinician’s decision and the course of treatment.

Interactive health communications have also impacted practice patterns by increasing home interventions. For example, when clinicians educate patients by telephone about Cardiac Pulmonary resuscitations (CPR) they improve survival rates after emergency calls.[48] These studies show that interactive health communications can play a significant role of extending clinicians’ services to the home setting and change the practice and outcome of care.

The impact of patients' interactions with IHC on practice patterns was made clear in a recent dissertation.  In this study, the computer interviewed patients before their visits and faxed its findings to the medical record room.  The computer findings were put into the record and were made available to attending clinicians.  Clinicians not only were satisfied with this service, but also thought that the reminders had changed their practice.  Independent verification of the results also showed that clinicians were detecting 15% more alcoholics than the clinic’s average.  Thus they had indeed changed their practices.[49] These data further confirm that patients' interaction with health communications can change clinicians’ practice patterns and through these changes have an impact on the patients' treatment and possibly health. 

One way Interactive Health Communications may make provider practices more effective is through enabling patients to review their records.  Limited evidence exists that patients review of portions of their own records may enhance the records accuracy and change patients' behavior.[50][51]  In a recent survey 71% of patients wished to have access to their medical record, while 3% did not.[52] 

Impact on patients' life styles

With push technology, more voice and better graphics, online services are increasingly looking and feeling like established mass media such as broadcast television and cable services.  Several studies have shown that mass media could be used to change the behavior of large number of people in a community. For example, 26 hours of mass media promotion of healthy behaviors led to 16% reduction in cardiovascular risks across the community. [53] Similar impact was reported for community wide reduction of alcohol use.[54]    Even though mass media are not interactive communications, to the extent that online communications are evolving into mass media, they may be effective in bringing about wide spread behavior change. 

A number of other investigators have directly examined the impact of interactive health communications on behavior change.  These authors report that IHC can change health behaviors.[55]^,[56] But not all have been successful in bringing about behavior change. [57]^{,[58], [59], [60], [61]} Furthermore, the impact of IHC on health behavior is not always sustained.  For example, Lando and colleagues found that computer instruction improved quit-rates for 6 months but not for 18 months post baseline.[62]  When are IHC applications likely to have a sustained impact on behavior change?

One possible reason why some IHC applications have not had a lasting impact on behavior change could be that information alone is not sufficient for behavior change. Patient education may be more effective by combining it with role-playing and support through electronic bulletin boards.  When IHC applications have included role-playing and support groups, they have been more successful in bringing about behavior change.  Gustafson and colleagues report successful examples of combining patient education with online support.[63] Alemi and colleagues report that computer role-playing helped teenagers better assimilate health messages.[64] 

Others have suggested that patient education can be made more effective by tailoring the information to key issues and characteristics of the patient.[65]  For example, smokers who received a letter tailored to their circumstances were more likely to quit than those who received a general message.[66]   Similar results were obtained for patients’ trying to diet reduce fat intake.[67]^,[68]

Another possible way to improve patient education is to do more of it. Given that patients face multiple sources of information, computerized health education is more likely to be effective when the effort is frequent and sustained over time.  Brief interventions may loose their effectiveness over time. Growing evidence suggests that there must be a minimal level of interaction before the impact of IHC on behavior can be measured.  In one study, for example, no beneficial impact was measured unless patients had used the system for at least 3 times a week over a seven-month period.[69]  In another study benefits were observed for patients with even higher use patterns. [70] Though both of these two studies focussed on patients who self selected to use the services, these studies suggest that impact of IHC may be more pronounced when patients use the service at least 3 times a week.  The problem of dose-response has been around in studies of drug effects; and it may be possible that a similar problem may also be present in studies of effectiveness of IHC applications.

IHC can also lead to poor health outcomes.  Many investigators have raised concern that patients may be misled by electronic sources of health information.  In one study for example, 89% of medical information was provided by non-health professionals.[71]  There are numerous reports of misinformation on the Internet.[72]  Recently, the Science Panel for Interactive Health Communications of the US Department of Health and Human Services has called for voluntary reporting of evaluation studies associated with Interactive Health Communications.[73]^,[74]  Despite the large number of warnings of adverse health outcomes and occasional well-publicized reports of adverse effects, there are few studies documenting a pattern of negative impact associated with interactive health communications.  One exception is the area of video games.  "Playing video games is associated with a variety of physical effects including increased metabolic and heart rate, seizures, and tendinitis. Aggressive behavior may result from playing video games, especially among younger children."[75]  .Studies of interactive games have not shown a link to violence, though video games reduce pro-social behavior.[76]  In one study, the use of Internet was  associated with depression.  In this study communication patterns among members of 73 households were examined over time.  Greater use of the Internet was associated with declines in face-to-face communications with family members and increases in depression and loneliness.[77] The increased association between computer use and depression is a concern.  Randomized clinical studies are needed to clarify what is a reasonable dose of exposure to IHC that will lead to healthy behavior change without the side effects of depression and loneliness.

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Impact on use of health services

Several studies have examined the impact of interactive health communications on cost of care and utilization of services.   It has been known for sometime that health education can reduce unnecessary visits.  For example, Fries and colleagues had shown that a book on self-care could reduce demand for care with no adverse health effects.  They have adapted this book and presented the content of it in an IHC format.   Their earlier success with behavior change suggests that the online version of the book may also be successful.[78] 

Direct evidence that computer-based health education can reduce unnecessary visits comes from a study by Robinson.  This study compared two randomly assigned groups of graduate and under graduate students who did or did not receive computerized health education.  The group that received the computerized health education had 22.5% lower medical visits than the group that did not receive the IHC information and continued with their usual source of information. [79]

Of particular interest is a study at University of Wisconsin which combined computer health education and computer support groups.  These investigators provided 200 HIV patients with several computer services, including a computer bulletin board -- where patients could post written messages to a public forum.  Patients were randomly assigned to control and experimental groups.  Only the experimental group had access to the computer services.  Experimental patients used the computer forum to communicate their experiences with the health care system, and their difficulties in communicating with their family and friends.  Among the various computer services provided to the experimental patients, computer mediated social support was the most frequently used service.  Investigators evaluated the project after 3 months and 6 months.  They found that patients with access to the computer, as compared to control patients, were more likely to report higher quality of life in several dimensions including social support and cognitive functioning.  The experimental patients had fewer office visit (dentists, primary provider and alternative care providers) and shorter time per visit to the primary care provider, HIV provider or the mental health provider.  The experimental patients were also less likely to be admitted to a hospital and more likely to have a short stay.  In summary, experimental patients had lower total health care cost than control patients did.  These data confirmed the importance of electronic communities in bringing about behavior change and showed that use of electronic support groups could lead to drastic reductions in cost of care.[80]

Other studies have confirmed the importance of electronic support groups in reducing cost of care.  In one study, [81] voice-based electronic support groups were compared to face-to-face support groups.  Over time, the groups that met online were eight times more likely to meet.  Furthermore, subjects who used the electronic bulletin boards were less likely to come in for a visit.  Reduced visits did not lead to poor health status.  These data suggest that electronic support groups may be reducing inappropriate use of health services.

Other forms of interactive health communications can also reduce cost of care.  In a very successful study, Wasson and colleagues replaced office follow up visits with three scheduled telephone calls.  Over a 2-year period, the telephone care group had 28% less cost per patient than the usual care group.  The savings of 28% on costs are large and practically significant for many health care operations. [82] Other studies confirm the potential of telephone based delivery of health services. [83]

Interactive health communications do not always lead to lower utilization of services. Some studies of telephone follow up have shown no effect.[84]  Other studies of IHC have shown the opposite effect: Several studies reviewed earlier have shown that IHC could increase (not reduce) office visits. In one study, substance abuse recovering patients, who used interactive health education more than 3 times a week, were 1.5 times more likely to remain in outpatient substance abuse treatment.[85] In another study, patients receiving calls on medication and exercise were more likely to visit a dietitian.[86] These data suggest that the impact of IHC applications on cost of care depend on the content of the messages being communicated.  If messages encourage a visit, then more visits are likely.  If messages are neutral (as in electronic support groups) or discourage visits (as in case of messages encouraging self-care) then IHC applications may reduce visits.

Before leaving this section it is important to make two points about the impact of IHC on cost of care or utilization of services.  Four randomized clinical studies by independent groups of investigators and using different modes of communication on different types of patients report that more than 20% of visits have been avoided.[87]^{,[88],[89],[90]}  This is not a small reduction in visits and is practically significant for many delivery systems, who are searching for ways to reduce cost of care while improving quality of services.  If the savings can be generalized to other settings and diseases, these studies suggest a technological fix for the health care cost crisis. 

The second point that we need to make about cost reductions is that these are reductions in direct cost of care and not necessarily reduction in total burden of illness as felt by the patient, the employer or other members of the society.  Many IHC applications encourage patients to participate in their own care.  Increased self-care reduce may reduce visits or hospitalizations but it clearly increases the burden to the patient and to the patients' caregivers.   

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Factors that affect use and impact of IHC

Studies show that a handful of individuals use IHC applications extensively, many use it moderately and some do not use it at all or use it sparingly.[91]  Uneven use of these services raises the question of who is must likely to use and benefit from IHC applications?  The variables predictive of use of the system can be classified into the following groups:

Individual characteristics

These characteristics include individual's prior experiences,[92] individual's media preferences and style,[93] attitudes towards the technology and attitudes of the individual's reference group towards the technology,[94] individual's general willingness to communicate,[95] and individual's reading capability (for text based health communications).[96] 

Studies that worked with narrow band of age differences have not found any age difference.  Other studies with larger range for age have found a definitive influence for age: Older people are less likely[97] and teenagers are more likely to use IHC applications.[98]  

Task appropriateness 

Early investigators indicated that interactive health communications (in particular text-based electronic bulletin boards) were not suitable for tasks in which group members need to arrive at consensus or need to express their emotion.[99]^,[100] These authors speculated that because of social anonymity and lack of status participants may make strong and uninhibited comments,[101] thus participants may antagonize as opposed to support each other.  More detailed studies have shown that frequent users of bulletin boards have different preferences than the occasional users.  In particular, frequent users are more willing to use the system to express their emotions and express support for others.[102]^,[103]  Recent studies have shown that the majority (54.6%) of comments left by patients on voice bulletin board was for positive emotional support of each other.  The remainder (39%) was task oriented comments and a small minority was negative emotions (6.4%) with no instance of overt disagreement or "flaming."[104]  These studies contradict earlier studies and suggest that interactive health communications are especially suited for discussion of emotional tasks.  Because of the anonymity offered by interactive health communications, these interventions may be especially well suited for addressing highly emotional and socially controversial issues such as sexual dysfunction, HIV-AIDS infection, drug abuse, or suicide.  

Organizing strategies 

The use and satisfaction with use seem to depend, in part, on the rules under which an interactive health communication is organized.  For example, a group facilitator, who listens and proposes a synthesis, can affect the outcome of electronic conversations in electronic bulletin boards.[105]  An electronic bulletin board that has a moderator and organizing principles governing the conversation is likely to lead to more user satisfaction.[106]^,[107]   

Furthermore, the way information transfer is organized may matter.  Patients can be led through leading questions to arrive at new information about themselves and their environments.  In contrast, patients can also be advised in a didactic fashion using authoritative sources.  Counselors have for some time argued that asking leading questions is more likely to overcome the patients' resistance to changing behaviors.[108]  In IHC applications that need to overcome clients' resistance to behavior change may also benefit from asking leading questions as opposed to providing didactic advice.  Examining the response to leading questions has the additional benefit of clarifying whether the client has understood the question or advice given.  Since in IHC non-verbal cues are absent, this may be an important step in advising clients.  Additional research is needed on how organization of information might affect the use and impact of IHC.

Studies show that communications (e.g. email use) is more prevalent in more de-centralized organizations and that the type and structure of organizations affect the rate of communication and the nature of messages sent.[109] Therefore, it seems reasonable to extrapolate that interactive health communications are more likely to flourish within groups and organizations that are more decentralized and less hierarchical.   These data suggest the hypothesis that unstructured support groups may have more utilization than moderated support groups.

Size and composition of population on the network 

The benefits that one can derive from use of interactive health communications depend on who is interacting with whom.  Two studies have shown that size and composition of networks might affect use.[110]^,[111]  If just one person is on the system, there is no benefit at all as no communication can occur.  If all the population with some shared characteristic are on the system, then the reasons for using the system are more universal and the habit of turning to the system more reoccurring and reinforced.  Adoption of the technology seems to depend on it being successful in reaching a critical mass of the target population.

The composition of, and not just the sheer size of, the group may also affect use of IHC applications.  IHC applications generally address a specific need or disease.  If patients have diverse disease then some of their needs may go unmet and the average use of IHC application may suffer.  For example, cancer patients may find little to communicate with rape victims and vice versa. We speculate that the use of IHC applications increases when networks include patients that are more similar in illness thus are more likely to have common concerns.

Need for information and nature and stage of illness 

Data presented in an earlier section suggests that IHC applications are likely to improve patients' compliance with treatment and lead to life style changes.  If this is true, then IHC applications are most likely to impact cost and quality of care in diseases in which patients compliance and behavior are a barrier to effective treatment.

We further speculate that IHC applications are most likely to be effective during early part of the illness, when patients work through denial, anger and acceptance of illness.  This is a period in time when patient is most uncomfortable with the disease and information and support is likely to have the largest impact on the patients' behavior and compliance with treatment advice.  As time goes on, patients' needs for support and information change and therefore the likelihood of impact of IHC changes.  Little data confirm or refute this speculation.  To our knowledge no one has studies how stage of illness affects use of IHC.  One exception is a study patients with HIV symptoms.  In this study persons with mild HIV symptoms used electronic discussion groups but did not benefit as much as the persons with more severe symptoms.[112]  This study confirms that information needs change as the disease progresses.  But it refutes the notion that patients are most in need of information and support in early stages of the disease.  Additional studies are needed to understand the relationship between stages of disease and use of IHC applications.

Presentation of information and impact of the medium 

As reviewed earlier tailoring the information to the needs of the patients[113] and presenting the information in ways that are vivid improves the effectiveness of information.[114] The effect of medium of presentation (voice, text, visual, etc.) has not been fully studied.  In this review we have presented diverse IHC applications including voice and text bulletin boards.  The differential impact of voice, text and visual messages is not known.    Given the ongoing transformation of online communication from text based to multi-media, the potential impact of the presentation should be examined more carefully.

Integration with service delivery 

Because clients have multiple sources of information, providing a consistent source of information across different mediums helps.[115]  In fact, the use of interactive health communications is substantially improved when clinicians recommend it.[116]  Furthermore, health messages are more likely to be followed when they are also emphasized by the clinicians.[117]^,[118] In part this increased use is associated with increased exposure; Kiosk based IHC are used more when placed in visible and high-traffic public areas.[119]

We speculate that use of IHC applications may increase when these services are built into health care operations.  For example, when telephone nurse advice is made available at the time of call for appointment, the service may be used more heavily. Like in retail, where use of a shop depends in part on its location, use of IHC applications may also depend on its location in the process of care.  The more IHC is integrated with service delivery the more one can expect patients to use these services.  The more patients can use the online service to conduct transactions (e.g. schedule a visit, re-order prescriptions, etc.) the more they are likely to be exposed to the health education messages available at the site.  Investigators should further examine the hypothesis that use of IHC services may be related to its integration with other clinic practices.

Technology driven use

In some IHC applications, use of services is encouraged by the system initiating the contact with the patient as opposed to waiting for the patient to access the application.  For example, the interactive voice response (IVR) systems such as computer reminders or computerized voice bulletin board discussed earlier can call patients at home.  Thus these systems do not always have to wait for the patient to initiate the contact.  These systems can call and prompt non-users as a method of encouraging their use of the system.[120]

Introduction Why a new review Why a new theory Methods of review Impact on satisfaction with care Impact on practice patterns Impact on patients' lifestyles Impact on use of services Factors that affect use Summary References

Summary

The review showed that Interactive Health Communications increase satisfaction with care, reduce cost of services and change providers' practice patterns.  Perhaps the most surprising finding was that a number of randomized clinical studies have proven that IHC applications may reduce utilization of health services by more than 20%.  This is a large and significant change in the delivery of health services.  We are surprised by the magnitude of the savings reported in randomized clinical studies without adverse effects on health outcomes.  It points to the need to expediently use the lessons learned to date from IHC applications and organize new delivery models.   

In addition, this review organized the variables that affect use and impact of IHC in a coherent model.  Figure 1 depicts the proposed model of how IHC applications affect outcomes of care.  While the model was in part derived from the literature reviewed in this paper, it represents our hypotheses about how IHC works.  It includes much speculation about how various variables are related to each other. 

The model proposes that a number of variables (including individual characteristics, stage of disease, organizing principles, etc.) affect the use of IHC applications.  The use of IHC applications in return affects availability of tailored consistent information plus support and role playing to the patients.  These two variables in combination affect adoption of behavior change and compliance with treatment.  The model emphasizes the importance of support and role playing in improving compliance or bringing about behavior change.  Finally the model proposes that better compliance and healthier behavior affect the patients' health status and utilization of health services.

One implication of the relationships articulated in the model is that IHC is less likely to be effective for diseases in which compliance and life styles are not important problems.  It also suggests that designers of IHC can be more effective if they were to focus on issues related to compliance and behavior change.

Of particular interest is the proposed role of integration with health services.  The model suggests that integration affects many variables.  First, integration with health services leads to increased use of IHC applications.  Second, integration improves accuracy of information available to providers and thus changes their practice patterns.  Finally, third integration with face-to-face services affects how IHC applications impact satisfaction with face-to-face care.  When integrated, IHC applications lead to increased satisfaction with face-to-face care.  When not, it leads to reduced satisfaction and substitution of IHC applications for face-to-face care.  The model sees a central role for how integration affects care outcomes.  Few IHC applications are fully integrated with delivery systems, thus the model presents a number of different hypotheses that can be further investigated as IHC applications are integrated with services.

The model also sees a pivotal role for how patients can use IHC applications to better inform providers.  Such improved accuracy of communication is seen to affect providers' practice patterns and in turn the outcomes of care.  Growing use of email by clinicians may be one easy way that IHC applications may affect information available to providers for treatment decision making.

We hope that investigators actively modify and improve the proposed model of how IHC affects outcome of care and that any evaluation of IHC applications be based on a theoretical model so that studies add to our understanding of how IHC works.