Systematic literature review of Internet interventions across health behaviors

Purpose: This review examines Internet interventions aiming to change health behaviors in the general population. Methods: Internet health interventions in the USA published between January 2005 and December 2013 were identified through Medline and CINAHL. Keywords used were (Internet or e-health or social media or web) paired with (intervention or program*). A total of 38 articles met all criteria and were reviewed. Results: Studies were analyzed by targeted health behavior interventions: tobacco (5), alcohol (4), weight loss (7), physical activity (PA) (7), nutrition (2), PA and nutrition combined (5), HIV or sexual health (4), and chronic diseases (4). Interventions ranged from one session to 24  weeks (average 6–12 weeks). Common strategies used, including web-based information, tailored feedback, weekly e-mails, goal setting, and self-assessment. Social cognitive theory and the transtheoretical models were the most commonly used frameworks. Recruitment strategies were typically media based varied by settings and populations. Except for the tobacco interventions, the majority studies yielded significant outcomes. Conclusion: This review provides updates and synthesized knowledge on the design and consistent effectiveness of Internet interventions across health behaviors. Results have implications for public health and healthcare professionals, as they play a key role in developing and delivering health promotion interventions as well as in assisting the communities and clients serviced obtaining evidence-based health information.


Introduction
The Internet is a powerful global communication medium that provides instantaneous information across geographical, cultural, language, and time spheres. Increased Internet access worldwide has resulted in a great number of health researchers and practitioners using it as a platform for the delivery of public health interventions. In recent years, several research studies have evaluated Internet health interventions and their intention to change health behaviors in the general public. Although there has been limited research to determine the efficacy of health-related information on the Internet, a systematic review of the early years of scientific evidence using e-health interventions, before 2003 and during 2002-2009, indicates significant positive effects on health behaviors in diverse populations and across a wide range of health conditions (Neuhauser & Kreps, 2010;Risk & Dzenowagis, 2001).
Most of the existing Internet study reviews focus on specific topical issues, with the vast majority focusing on nutrition, dietary, and physical activity (PA) behaviors, and on weightloss interventions (Hamel, Robbins, & Wilbur, 2011;Neuhauser & Kreps, 2010). Current reviews on Internet health interventions also tend to focus on specific components, such as the evaluation of the delivery methods, dose of treatment, attrition and/or user satisfaction, rather than providing a more comprehensive review on the interventions' integrated components in their findings. The limited integration of intervention components in findings reported in existing studies can make it difficult to interpret results or determine what characteristics or components of the Internet health interventions are the key to achieving behavioral changes in measured outcomes (Hamel et al., 2011). It is necessary that studies on Internet health interventions analyze the existing literature in a way to better understand and improve the feasibility, usability, and efficacy of such interventions (Risk & Dzenowagis, 2001).
An examination of Internet health intervention studies on a variety of topics and their existing components allows for comparisons and contrasts across a variety of health conditions or outcomes. Evaluating e-health interventions from this perspective could provide new insights for researchers and practitioners to better develop, adapt, or adopt Internet-based health interventions. To the best of our knowledge, there is currently no review on Internet health intervention studies conducted in the USA that examines across age, gender, and health conditions, without focusing on a specific intervention component or measurement outcome. The insights gained from this literature review that crosses demographic categories will provide a deeper understanding for effective Internet health intervention development.
For the purposes of this study, the term Internet health intervention was operationally defined as systematic intervention programs, usually addressing one or more determinants of health delivered primarily via the Internet and interfacing with an end user (Bennett & Galsgow, 2009). We included only behavioral health interventions where the major component was delivered via the Internet to distinguish its effects from other interventions provided primarily by other media, such as print material, CD-ROM, videos, and phones (MMS and SMS). This review focuses primarily on the effectiveness of Internet health interventions, employing a more comprehensive approach by not limiting the scope of the research to a particular health condition. It also evaluates a multitude of considerations for using the Internet as a platform of health intervention development and delivery, such as their sampling and recruitment strategies, target populations, theoretical framework or models for behavior change, program delivery, and outcomes of the interventions. Throughout this review we compared and contrasted strategies used by various Internet health interventions, identified trends and themes as they emerge, and discussed the value of using behavior change theoretical frameworks, recruitment strategies, differences in target populations and settings, and any enhancements associated with supplemental strategies as they relate to the health behavior or outcomes.
The aim of our study is to provide a comprehensive narrative review of the literature published on Internet health interventions between January 2005 and December 2013 and to identify patterns from the studies reviewed as well as bridge the gap of current knowledge to inform the development of future research. We provide a detailed analysis from a recent sample of Internet health interventions to examine intervention characteristics, determine the evidence of their effectiveness as well as identify strategies to improve health outcomes.

Methods
Medline and CINAHL, two major electronic databases, were used to locate intervention studies that used computer-based information and communication technology to deliver a specific health intervention. The key search terms (Internet or e-health or social media or web) were paired with the search terms (intervention or program*) in the title field. Internet health intervention studies in the USA published between January 2005 and December 2013 were included for the purpose of this review. This time period were decided to reflect a reasonable time-period coverage during the rapidly changing Internet era to provide a timely update while not duplicating existing review efforts as well as aimed to paint a good picture of the current state of Internet health studies. The last search for this systematic review was performed on 15 January 2014.

Study selection
Inclusion criteria were original research conducted in the USA, published in English between January 2005 and December 2013, and contained the above search-terms in the title. Further inclusion criteria included research design using at least one comparison or control group (CG), network provided by the Internet during the studies' intervention or program, use of a web-based computer program as a significant part of the intervention, and health behaviorrelated outcome(s) as a primary part of the evaluation. The bibliographies of retrieved articles were also scanned for additional references.

Data extraction
The following information was systematically extracted by two reviewers; author/s and year of publication, target group and demographics, recruitment methods, targeted health behaviors, setting or location, theoretical frameworks, research design and follow-up, characteristics of the intervention (intervention type/components and delivery strategies), and effectiveness of the study. During the process of completing the final review, the information was compared and contrasted, which can be found in Tables 1 and 2.

Results
Literature searches yielded 913 studies. After duplicate records were removed, 740 unique records remained. Two reviewers examined all retrieved journal articles using the above criteria to determine relevance for inclusion. Titles and abstracts of these studies were screened and 693 were excluded on the basis that they were not directly Internet-based, not intervention studies, not addressing any health behaviors, or not conducted in the USA. The full text of 47 relevant articles were reviewed and 9 articles were excluded at this stage due to either health behavior change not being a primary outcome or the study did not including at least one comparison or CG. Finally, 38 articles were identified to meet all of our inclusion criteria and were downloaded into the Endnote electronic reference manager ( Figure 1).
Overall, interventions ranged in length from one 60-90 minute session to multiple sessions lasting 24 weeks, with the overall number of sessions lasting on average between 6 and 12 weeks. Most interventions comprise of modules or activity components ranging from 1 to 73 sessions. Below, the researchers discussed review analyses of studies by different targeted health behaviors on the extracted variables described under the methods.

Tobacco prevention
A total of five studies reviewed focused on tobacco use, with sample sizes ranging from 139 to 2523 tobacco users (Patten et al., 2006;Severson, Gordon, Danaher, & Akers, 2008). One intervention targeted adolescents (Patten et al., 2006), whereas the four remaining interventions focused on adult smokers (18+ years old). The most common recruitment methods were flyers, RCT; tailored expert system condition used balance -6 week self-help program; opportunity to enroll "buddy" who were sent email messages and encouraged to provide informal support; materials consisted of an initial guide followed by tailored action plan delivered at 1, 3, and 6 weeks into prog.
Weight loss SCT No differences in percent weight loss between intervention or CG at three months (−2.6 ± 3.8%podcast vs. −2.6 ± 3.5%podcast and mobile; p > .05 between groups) or six months (−2.7 ± 5.1%podcast vs. −2.7 ± 5.6%podcast and mobile; p > .05 between groups); twitter posts decreased between the 0-3 months period to the 3-6 months period; adjusting for demographic variables Twitter was a good predictor of % weight loss at six months, each post correlated with −0.5% weight loss PA and nutrition Bosak et al. (2009) PA Bandura self-efficacy Sig. improvements found in IG HDL cholesterol (z = -2.024; p = .04) and self-efficacy for PA (z = -1.970; p = .04).  Sig. pre-to-post differences in FJV consumption (p = .002), PAyesterday (p = .001), PAusually (p = .001), and FJV self-efficacy (p = .003) Winnet et al. (2007) Nutrition and PA SCT Increased FV at post-test IG (∼1.5 servings) vs. CG (∼0.5 servings); p = .005, increased fiber 3.0 g vs. 1.5 g; p = .006, increased steps daily 1500 vs. 400 steps; p = .05, decreased weight ∼−0.3 kg vs. +0.6 kg; p = .03 Milan and White (2010) Nutritionfolic acid intake TTM Sig. greater proportion of college women in IG-stage tailored intervention (32.6%) vs. CG non-tailored education (19.9%) taking MV @ post-test than pre-test; p = 0.015 Thompson et al. (2009) Fruit juice (FJ) and low-fat vegetable consumption SCT Significant increases in FJ consumption (p = .003), FJ home availability (p = .009), and low-fat vegetable consumption self-efficacy (p = .004) among IG immediately, but not maintained six months later HIV and chronic diseases Bowen et al. (2007) HIV risk reduction in MSM SCT HIV AIDS knowledge, self-efficacy and outcome expectancies increased after participating in the intervention and changes were maintained at one week follow-up Bull et al. (2008) Promotion of condom   Joseph et al. (2007) Asthma management TTM and HBM At 12 months IG students reported, fewer symptom-days and nights, schooldays missed, restricted activity days and hospitalizations compared with CG; adjusted relative risk and 95% CI were 0.5 (0.4-0.8); 0.4 (0.2-0.8); 0.3 (0.1-0.7); 0.5 (0.3-0.8); and 0.2 (0.2-0.9), respectively (p < .05) Smith et al. (2012) Depression, stroke survivors and their care givers NA Findings suggest that the intervention reduced the outcome of depression for the care givers at statistically significant level (40% of CGs in the IG reported a 50% decrease from baseline scores); depression was not reduced for stroke survivors at significant levels they did show a change in the predicted direction (depression reduction did not appear until one month after treatment) Notes: NS, non-significant/no significant; sig., significant; OR, odds ratio; CI, confidence interval; HDL, high density lipoprotein; SE, standard error; IMB, information-motivationbehavioral.
press releases, advertising in newspapers, magazines, and Internet sites, such as Google.com. Only one study utilized emails and electronic newsletters, as it was conducted in a worksite setting (Swartz et al., 2006). Approximately half of the smoking studies used the social cognitive theory (SCT) as a theoretical framework and all the studies were randomized control trials. All studies focused on smoking cessation and/or prevention of relapse, and only one study focused on smokeless tobacco cessation (Severson et al., 2008). All interventions required multiple access attempts to Internet content. Activities and information were provided using between 5 and 38 modules or sessions (Patten et al., 2006;Swartz et al., 2006). Modules were based primarily on smoking cessation and relapse prevention and provided personal quitting assistance, information on barriers, benefits, and/or strategies and nicotine replacement therapy. Some studies provided individual counseling in addition to Internet use, and one study used a virtual counselor as an "automated approximation of live smoking cessation counselor" (Swartz et al., 2006). The interventions used peer support forums, expert support options, video testimonials, or links to other Internet sites. Most of the interventions reviewed offered personalized messages for participants in order to target a specific behavior. For example, in one study, participants were reminded or encouraged to continue using the intervention even if they did not log on in a week (Japuntich et al., 2006). Some other studies did not permit staff to have any contact with participants during web intervention.
Two studies found higher quit or cessation rates in intervention groups (IGs) than in CGs (Severson et al., 2008;Swartz et al., 2006). Three studies found no significant differences on quit-rates between the IG and CG at the end of the intervention, nor at follow-ups (Japuntich et al., 2006;McKay et al., 2008;Patten et al., 2006).

Alcohol prevention
Four studies reviewed focused on alcohol; all had sample sizes less than 300 participants, with the exception of one that had 1137 (Bingham et al., 2010). All participants were young adults or college aged. Only one college-based study used a convenience sample from college courses (Moore et al., 2005) and the other two recruited primarily via email invitations (Bingham et al., 2010;Neighbors et al., 2009). The one non-college-based study recruited their sample through Human Resource departments in various companies (Doumas & Hannah, 2008). The common goal in all the studies was to reduce alcohol use but each study focused on different aspects. For example, one study focused on binge drinking, whereas another focused on reducing 21st birthday drinking (Moore et al., 2005;Neighbors et al., 2009). Three out of four studies used randomized control designs.
Theoretical frameworks to guide the studies design and methodology included the motivational enhancement model, social norm, and a hybrid model based on four traditional health behavior theories (health belief model [HBM], theory of planned behavior, transtheoretical model [TTM], and precaution process adoption model) (Bingham et al., 2010;Doumas & Hannah, 2008). Different behavior change strategies were used for each intervention. Personalized feedback about drinking behavior was provided in each intervention with the exception of one that used newsletters delivered in print and web versions (Moore et al., 2005). Interactive interventions included online sessions tailored to the participants' alcohol-related risks, allowed participants to print blood alcohol concentration (BAC) chart based on gender and weight, and provided graphed feedback for a visual display of drinking levels related to financial cost (Bingham et al., 2010;Neighbors et al., 2009). All studies reviewed on alcohol Internet interventions demonstrated effectiveness on decreasing drinking in IG, except one (Moore et al., 2005).

Weight loss
A total of seven studies focused on weight loss. Six of the studies had samples ranging between 57 and 192, whereas one study sampled 2862 participants (Rothert et al., 2006). All the studies targeted overweight and/or obese persons except for one that targeted African-American (AA) girls with at least one obese parent and allowed parents to participate in the study (Williamson et al., 2005). Most of the studies recruited participants via newspaper ads, and some included supplemental recruitment strategies, such as community talks, flyers, announcements on websites and listservs, utilization of integrated health care system network for clinician referrals, and letters to members in diabetes and cardiovascular disease (CVD) registries. The primary objectives across all the studies were to reduce weight and behavior modifications.
All the studies followed a randomized controlled trial (RCT) design. However, no specific theoretical framework was identified or used for most studies. The intervention strategies consisted of primarily behavioral modification treatment, mostly via interactive internet counseling, weekly online chats led by trained therapists, and weekly feedback given by either the computer program or tailored messages from a counselor (Gold et al., 2007;Jones et al., 2008;Tate et al., 2006;Webber et al., 2008;Williamson et al., 2005). All the interventions focused on providing a tailored action plan for the participants in the IG. For example, one designed an online therapistled weight maintenance program (Gold et al., 2007), and another one used email counseling, with all groups meeting at least once for a face-to-face session (Tate et al., 2006). Only one intervention used social media to influence weight loss, using a weight-loss counselor-posted messages daily to reinforce the podcasts and stimulate discussions (Turner-McGrievy & Tate, 2013). When compared to CGs, most of the IGs lost more body fat, body weight, dietary fat intake, and maintained higher weight loss at 12 months (Gold et al., 2007) except for two studies in which the use of an Internet motivation-based intervention did not result in any weight loss (Webber et al., 2008).

PA and nutrition
There were seven studies focused primarily on increasing PA, and additionally five studies also examined related factors, such as nutrition and binge eating. Two studies were categorized as nutrition only interventions, with one focused on folic acid intake and the other targeted fruit juice and vegetable (FJV) consumptions (Milan & White, 2010;Thompson et al., 2009). Sample sizes ranged from 22 to 368 participants in studies targeting PA only (Bosak et al., 2009;Carr et al., 2008;Dunton & Robertson, 2008;Irvine et al., 2013;Marcus et al., 2007;Ornes & Ransdell, 2007). Recruitment methods varied from newspaper advertisements to email solicitations, clinic database selection, and researcher presentations in a variety of settings. In the studies targeting both nutrition and PA, sample sizes were generally larger, ranging from 80 to 1071 participants Winett, Anderson, Wojcik, Winnett, & Bowden, 2007). Recruitment methods were also varied and included emails from management in work settings, voluntary sign-ups in the university setting, and pulpit announcements and bulletins in the church setting (Cook et al., 2007;Franko et al., 2008;Winett et al., 2007). Both types of studies targeted a variety of populations, such as AA girls, university and high school students, and company employees. The studies that examined additional factors related to PA were more likely to target increasing motivation to change health behaviors, increase knowledge, and improve social support and attitudes towards exercise, rather than just an outcome of increased PA and healthier eating. Most of the studies used a RCT design and were based on the SCT. Additional theories guiding the interventions were the TTM and HBM.
Most of the intervention strategies to increase primarily PA included behavior-specific goalsetting approaches, and all the interventions allowed participants to monitor or record their PA progress and receive tailored individualized feedback. Despite fairly similar intervention strategies were used, information was disseminated in a variety of ways. One study provided new content on the website each week including links for evidence-based websites (Bosak et al., 2009), whereas another administered weekly emails with links to interactive tailored web pages to the IG (Dunton & Robertson, 2008), and yet another study delivered the intervention in four weeks through nine modules via WebCT (Ornes & Ransdell, 2007), the final study included a social media campaign using Facebook and the Internet Support for Health Associations Promoting Exercising (INSHAPE) to encourage participation from the participants (Cavallo et al., 2012).
Likewise, the studies focusing on additional factors related to PA also included goal setting as a key feature of the intervention programs, allowed the participants to self-monitor their progress and report it for feedback. Most modules and new topics were introduced weekly, and interactive components included group discussions, assessment pages, resources, and motivational messages. Innovative strategies that were not common in all the groups included the use of comics and role modeling on the intervention website, and the provision of motivational support from the church pulpit announcement and bulletins (Thompson et al., , 2009Winett et al., 2007).
All the interventions were successful in the studies focusing primarily on PA, except one (Marcus et al., 2007). PA was increased in three IGs, in addition to decreased waist circumference and coronary risk ratio (Carr et al., 2008). The cost of development and delivery of Internet health interventions was also found to be lower than consultation and follow-up in the clinic (Carr et al., 2008). All the studies examining nutrition alone, nutrition and other factors in addition to PA reported increases in healthy dietary behaviors and PA levels in the IGs; except for one study that found no differences on PA or stress (Cook et al., 2007).

HIV and chronic diseases
Four studies focused on HIV prevention or sexual health, and four studies focused on other chronic diseases: diabetes management, colon cancer screening, asthma management, and depression in stroke survivors and their caregivers (Bond et al., 2007;Chan & Vernon, 2008;Joseph et al., 2007;Smith et al., 2012). The HIV prevention studies recruited participants online, in addition to clinic-based referrals (Bull et al., 2008) and face-to-face recruitment (Bowen et al., 2007). Two studies targeted men who have sex with other men (MSMs) and enrolled a maximum of 112 participants (Bowen et al., 2007;Carpenter et al., 2009), whereas one targeted youth regardless of sexual orientation and enrolled approximately 4000 in their study (Bull et al., 2008). For the other chronic disease studies, the number of participants did not exceed 408.
Most of the studies were RCTs. HIV prevention studies used a combination of strategies based on constructs from primarily three theoretical frameworks: the SCT, information-motivationbehavioral skills model, and motivational interviewing (MI). Among the chronic disease studies, three did not use any particular theoretical framework while one used both the TTM and HBM (Joseph et al., 2007).
HIV prevention studies used a range of 2-7 modules; these were tailored for the specific population and included questions related to HIV risk, pictures and audio role model stories, interactive graphic conversations between two men of different HIV status, and motivation, information and, skills training (Bowen et al., 2007;Bull et al., 2008;Carpenter et al., 2009). Only one study mentioned providing tailored feedback throughout the intervention (Carpenter et al., 2009). The sexual health intervention focused on consistent condom use using Facebook to deliver information to the AA and Latino youth. Results showed condom use remained stable in the IG and declined in the CG (Bull et al., 2012).
The diabetes management intervention focused its strategies primarily on goal-setting and problem-solving skills (Bond et al., 2007). The asthma management study used approximately five tailored educational modules to make use of feedback from previous sessions and correspond to participants' stage of readiness (Joseph et al., 2007). The intervention focused on colon cancer screening study was the only one to include involvement from a primary care physician and record biomarkers (Chan & Vernon, 2008). The intervention on depression management in stroke survivors and their caregivers provided caregivers with knowledge, resources, and skills to reduce distress and provide best care to their stroke survivors (Smith et al., 2012).
Most of these Internet health interventions demonstrated significant effectiveness in increased HIV/AIDS knowledge and self-efficacy, decreased number of unprotected sexual acts, reduced HbA1c, weight and cholesterol levels, and fewer asthma symptoms and related hospitalizations (Bond et al., 2007;Bowen et al., 2007;Joseph et al., 2007;Lustria, Cortese, Noar, & Glueckauf, 2009). Only one study was considered ineffective as the public access arm of the study yielded a higher return rate of the fecal occult blood test (FOBT) for colon screening from CG than IG participants (33%) (Chan & Vernon, 2008).

Social media
Of the articles reviewed, only three included social media interventions; two used Facebook and one used Twitter. For each of these studies, the researchers use a trained professional that facilitated discussions on the social media outlet. Cavallo et al. (2012)  Twitter and podcast to promote weight loss. The moderators in these campaigns were trained counselors for the specific health behavior that the program focused on.

Discussion
Findings from the current review study indicated that, overall, Internet or web-based interventions (WIs) produce favorable results and are effective in producing and increasing targeted health or behavioral outcomes. Common trends for particular methods to recruit participants in the Internet health intervention studies throughout the review typically included some form of media, such as newspaper advertisements and announcements, press releases, television advertisements, posters and flyers, and internet-based recruitment (banners and/or paid advertisements on websites). Existing successful studies have commonly used a combination of methods for recruitment instead of relying solely on one method, clinical referrals for instance. The review indicates that specific settings and/or target populations use certain recruitment strategies more often. For example, school-based interventions typically recruit in classes and school clinics used flyers, emails, and student rallies, whereas those based in the community are more inclined to use newspaper, radio and television media, billboards, flyers, and even pulpit announcements. Financial incentives were most often used as a strategy to increase retention, and they ranged from $5 to raffle prizes of $500. Most studies offered $10-30 to participate in either each session or at the completion of the study, as an incentive. Intervention time length and number of modules do not appear to be highly correlated with the success rate of interventions; i.e. shorter interventions or those with fewer modules do not necessarily result in less desirable outcome measures than long interventions or those with more modules. Instead, intervention components, such as modules, structure, and interaction nature of the intervention, seemed having a greater impact on the measured outcomes. For instance, it was evident that the involvement of professional or a trained counselor to provide professional advice during online interventions was often associated with positive significant intervention effects on the target population, especially in the case of smoking cessation, weight loss, and social media-related interventions. This is not surprising as the interaction of experts with participants is expected to be more substantial through the provision of direct feedback and monitoring (Dijkstra & de Vries, 1999).
Common strategies identified, including the use of web-based information, tailored feedback, weekly emails, goal setting, and self-assessment. A handful of studies also used innovative intervention strategies, such as the use of familiar/relatable avatars, humor in a birthday card message, individual graphed feedback and self-directed programs. Our review consistently showed that Internet health interventions using a combination of approaches or methods often resulted in effective intervention outcomes.
The SCT and the TTM were the most commonly used theoretical foundations for all the interventions, particularly within the smoking cessation and PA/nutrition intervention studies, with the SCT used in more than half of the studies. Most of the interventions are grounded in theoretical frameworks, except that the majority of the weight-loss interventions did not identify specific theoretical frameworks. Much research underscores the importance of using theory as a basis for behavioral interventions. Theoretical models such as the HBM, Theory of Planned Behavior, and SCT are among those most often used in health education, therefore serve as an excellent frame of reference for Internet health interventions. Our review supports the use and application of theories as a foundation for behavior change interventions would be expected to increase the likelihood of effectiveness for the intervention (Hamel et al., 2011). Additionally, the literature on Internet health interventions indicates that the use of behavioral theories could facilitate program exposure (Crutzen et al., 2011). Although most of the successful alcohol internet intervention studies reviewed did not specify theoretical frameworks, majority of these studies did indicate using tailored messages or tailored action plan, and/or Internet counseling/chats led by trained therapists. In one systematic review of computer-tailored behavioral interventions, researchers also noted that the utilization of tailoring based on population characteristics, health foci, behaviors, and situations are necessary to maximizing the effectiveness of computer-assisted tailored programs (Risk & Dzenowagis, 2001).
All in all, the majority of interventions reviewed yielded significant improvement on health behavioral outcomes. This review indicates that using multi-dimensional approaches, such as tailored intervention components and personalized feedback, is associated with favorable intervention effects. Almost all the interventions across health issues showed significant impact on health outcomes, except the tobacco Internet health interventions. Most of the tobacco Internet interventions yielded non-significant results for cessation and quit-rates, although the results were mostly noted as promising. It is not clear why these interventions did not show effectiveness. Some recommendations proposed are that longer follow-up periods be implemented to determine long-term impacts of programs, and that sample sizes be increased to increase statistical power to possibly detect any long-term significant differences (Patten et al., 2006;Swartz et al., 2006).
This review study has a few limitations. The relationship between the components of the intervention and actual outcome measured although associated may not be causative. Our review was limited to only interventions delivered to target groups in the USA and may not be representative of the acceptability and utility of internet health interventions overall worldwide. To make the scope of this review manageable, two major electronic databases were searched to retrieve published Internet health intervention studies (from 2005 to 2013); i.e. CINAHL and Medline; thus our results and analyses should not be considered an exhaustive analysis. Furthermore, our review might suffer risk of bias commonly exists in systematic review studies, including the publication bias, selective reporting within studies, and/or incomplete retrieval of identified research and reporting bias. Finally, due to the diversity of the type of programs as well as the process and/ or outcome variables used, it was unfeasible to conduct meaningful meta-analyses. For studies that examined some similar health conditions, not all studies examined or analyzed the same outcome variables. Although it would also be nice to be able to analyze different population groups, many studies had mixed age or gender participants. In addition, existing Internet intervention programs reviewed commonly used multiple strategies and generally did not examine or provide sufficient information on the effectiveness of specific strategies. Our review provided thoughtful narrative analyses on some of these areas the best we can, given available data and information provided in existing studies.
Nevertheless, this review clearly shows that a broad range of successful Internet health interventions has been conducted and they all share some common themes across their topics related to their success. The researchers conducting this review made every effort to include all relevant evidence pertaining to this topic published in English in the USA. The comprehensiveness of this review allows a broader review scope and an enhanced ability to consider and discuss themes as they occurred across various health conditions and behaviors. It serves as an important update on the rapid progress of interventions developed in the recent years and provides synthesized knowledge on the design and effectiveness of delivering Internet health interventions.
Overall, the results lend support to the use of the Internet or Web as a viable alternative to more traditional health interventions as a feasible means for health promotion and disease prevention. Health practitioners are encouraged to consider incorporating the use of technology into more traditional intervention approaches to enhance the overall outcomes among their target population. Further research is needed to evaluate participants' satisfaction with the Internet health interventions compared to more traditional approaches. Additionally, future studies are encouraged to examine what dose or exposure to Internet health interventions is optimum to achieve outcomes across various health behaviors or conditions.