Internet-Delivered Health Interventions That Work: Systematic Review of Meta-Analyses and Evaluation of Website Availability

Introduction

Background

The World Health Organization recognizes that implementation of population-based strategies to improve health is critical [1,2]. Likewise, the Institute of Medicine’s list of suggestions for action includes the implementation of population-based strategies to improve health [3]. The need for population approaches to solve health problems was recently reviewed by David Hunter as he stated, “As countries struggle to transform their health systems to cope with rising demand, aging populations, and largely avoidable lifestyle related illnesses within limited budgets, policy makers are desperate for the right kind of evidence” [4]. With such broad goals in mind, it is surprising that evidence-based mechanisms are not yet fully engaged so that Internet-delivered health interventions can be exploited to achieve these goals.

Although there are numerous websites that contain health-related information, the ability of the consumer—or the patient—to find scientifically robust (ie, evidence-based) health interventions is not fully known. Data from the Pew Research Center indicates that 72% of adults who use the Internet have searched for health-related information in the previous year (based on 2012 survey data) [5]. Furthermore, there is insufficient information to assist the public in deciphering which sites contain useful information that could help them stay healthy, ameliorate risky behaviors, recognize early disease, or assist with treatment of their existing disorders.

The central question is, “Which Internet-delivered health interventions actually work?” For scientists, the answer to this question can be addressed by evaluating the results from randomized controlled trials (RCTs). In fact, the efficacy of some Internet-delivered interventions has already been assessed by investigators. Yet, there is not yet a fully formed mechanism to link these results with the individuals who may wish to use this information.

To expedite this process, there are necessary preparatory steps before implementation. Our translational model is shown in Figure 1 and illustrates the steps. Many RCTs and meta-analyses of RCTs of Internet-delivered health-related interventions are already published and, therefore, some evidence is available. We now, through this report, present the results from the evaluation step of dissemination. That is, we conducted a systematic review of published meta-analyses of RCTs on Internet-delivered health-related interventions. We evaluated this evidence and generated a list of evidence-based websites currently available for use. We were especially interested in Internet-delivered therapies that do not require real-time interaction with a therapist or other health care provider. That is, the application was housed on the Internet for general use by the public.

‎View this figure

Aims of the Study

The aims of this study were to (1) discover the range of health-related topics that were addressed through Internet-delivered interventions, (2) generate a list of current websites used in the trials which demonstrated a health benefit, and (3) identify gaps in the research that may have hindered dissemination.

Methods

Inclusion Criteria

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were used [6]. To be eligible for inclusion, studies were required to be meta-analyses of RCTs with at least one intervention that was Internet-delivered and reported a health-related outcome. Within each meta-analysis, we required that there be at least one self-guided intervention without therapist or clinician support. For this review, we excluded group-based interventions (ie, trials that enrolled groups of people to experience the intervention together). There were no restrictions on the types of individuals in the trials or the type of health outcomes.

Search Strategy

A comprehensive search strategy was developed with a biomedical research librarian and was undertaken to identify articles for inclusion. The following electronic databases were searched: MEDLINE via Ovid, PsycINFO, Embase, Cochrane Database of Systematic Reviews, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL). Search strategies utilized a combination of keywords and MeSH headings (Table 1). The last date of the search was June 14, 2016. There were no language restrictions on the search.

Screening of Articles

Three authors (KL, RK, and MR) independently reviewed the title and abstract of each record to determine eligibility. Any disagreements regarding inclusion or exclusion were resolved by a discussion between two authors (MR and KL). Full papers of the selected title and abstracts were reviewed independently by three authors (KL, RK, and MR) and disagreements regarding inclusion or exclusion were resolved by a discussion between two authors (KL and MR). All RCTs within each meta-analysis were screened for eligibility (self-guided Internet-based health-related intervention).

Analyses

The purpose of the analyses was to combine the results from across all the meta-analyses so that the results could be summarized and the Internet programs could be located. Individual RCTs within each meta-analysis were grouped by topic: Substance Abuse, Mental Health, Diet and Physical Activity, Disease Management, Disease Prevention, and Childhood Health Problems. A health benefit was defined as a statistically significant improvement in any health-related outcome within an RCT; all trials assessed outcomes through inferential statistics with alpha set at .05, 2-tailed. The concurrent control groups did not receive the Internet-delivered intervention (generally a wait list) unless specifically stated. Measures of efficacy were calculated when data were available within each RCT; for binary outcomes, number needed to treat (NNT) was calculated when absolute measures were reported. For outcomes measured using continuous scales, mean changes were listed (intervention relative to control).

Each RCT was reviewed for the name of the intervention and the website that housed the intervention. Functional websites of such evidence-based interventions (demonstrating a health benefit) were located on November 18, 2016. We defined functional website as those sites which housed the program which was tested in the RCTs and was available for general use.

Results

There were 363 records identified through the search (Figure 2) which yielded 304 records after removing the duplicates. The abstracts were reviewed and 162 were excluded because they did not meet the eligibility criteria. Full-text articles were reviewed for the 142 remaining articles and 36 were excluded due to noninvolvement of the Internet, 15 were excluded due to therapist or clinician support only, and 20 were excluded due to either protocol only, no health outcomes, group-based interventions only, not a meta-analysis, or did not include RCTs. There were 71 meta-analyses of Internet-based interventions that met eligibility criteria and were included in this study [7-76].

Within the 71 meta-analyses, there were 1733 studies. Of these studies, there were 268 unique RCTs that were self-help Internet-based interventions; and of the 268 studies, there were 57 trials demonstrating a health benefit with a functional website [77-138]. The topics covered are listed in Table 2.

Internet self-help for substance abuse was the most frequent topic in RCTs, with alcohol having the greatest number of trials. Of the 72 trials on alcohol use, there were 8 with functioning websites. Tobacco use was also a common subject for interventions with 7 websites (from the 30 trials reviewed). Mental health interventions were available, including anxiety (5 websites on generalized anxiety disorder, 3 on phobias, 2 on panic disorders, 2 on post-traumatic distress disorder, and 1 on obsessive compulsive disorder [OCD]) and depression (4 websites). There were 46 RCTs reviewed for diet and physical activity interventions, and 13 of those yielded a functioning website. There were fewer RCTs on disease management, with insomnia and chronic pain yielding 13 and 10 trials, respectively. Within the meta-analyses, there were a few RCTs specifically on cardiovascular risk factors (blood pressure, cholesterol, and hyperlipidemia) and several on sexual health (sexually transmitted disease, sexual dysfunction, unintended pregnancy). However, there was only one RCT on self-help regarding cancer prevention (for skin cancer). Finally, there were two trials targeted to parents of children with health problems—one on behavioral problems and one on encopresis.

In most instances, the Internet-delivered interventions were offered only to the study participants in the context of the RCT; websites to deliver the intervention after the conclusion of the study were not available. For example, Student Bodies was an efficacious Internet-delivered program for eating disorders in girls but the program was not available for general use [139-141]. In the area of sexual health, there were several efficacious Internet-delivered programs regarding sexually transmitted diseases [142-144]. However, these sites were only available for study participants during the course of the research study. Overall, in only 21.3% (57/268) of instances, there was a functional website for the interventions after the conclusion of the trial.

We compiled a list of websites of the Internet-delivered interventions providing a health benefit and these are shown in Table 3 with the name of the program, hyperlink to the site, cost, and the languages utilized for delivery of the program.

‎View this figure

Substance Abuse

From this review, we found that there are eight currently available evidence-based websites on alcohol use. All but three were conducted using college or university students; some are specifically targeted to universities, offering a suite of programs regarding substance abuse, health, and wellness. The most common techniques utilized in these interventions (Table 4) were personalized and normative feedback, as well as goal setting. Some sites included more tailored feedback and interactive journaling. The health benefit observed in the trials was generally a reduction in alcohol consumption, although some trials showed a reduction in the consequences of heavy drinking such as impairment in control and fewer embarrassing actions. The length of the programs varied—some being rather brief screening tools and others encompassing 6 months of structured activities [77-91]. The freely available websites for alcohol emanated from various European countries. Screen shots of each home page of the websites are given in Multimedia Appendix 1.

For tobacco use, there are several free evidence-based websites for reducing or quitting smoking. Four of these sites were available in English, 4 in Spanish, and 1 in Norwegian; the Stop-tobacco program was available in multiple languages. Often they were supported with governmental or public health funding such as smokefree in the United States, Stop-tobacco in Switzerland, and Quit in Norway. In general, the primary outcome was greater abstinence rates of smoking which were achieved through structured cognitive behavioral techniques, including motivational materials, personalized and tailored advice, goal setting, feedback mechanisms, and self-monitoring.

There were a few RCTs on curbing cannabis use which reported a reduction in the frequency or quantity of use. The mechanisms used to achieve the health benefits were similar to those used for tobacco use, relying heavily on cognitive behavioral approaches (listed in Table 4). One free Internet-delivered intervention Reduce Your Use comes from Australia and another was developed in Germany (Quit the Shit).

Mental Health

There were 4 functional websites that were intended to help individuals with depression. MoodGYM and BluePages are generally used together, the first for the delivery of cognitive behavioral therapy and the following as an adjunct. MoodGYM was more effective when the entire program was completed—not brief interventions [145]. MoodGYM contains five sequential modules which are completed at the pace of each user. There were a few commercial sites as well— Deprexis (9-week program) and Color Your Life (8 weeks with a 9th-week booster). Both of these programs led to a reduction in the symptoms of depression. Cognitive behavioral therapy was the mechanism utilized in each of these interventions which was delivered in a modular, stepwise manner over several months. It is important to note that one of the Internet-delivered programs for treating insomnia (Insomnie) also led to a decrease in depression and anxiety.

There were several websites that addressed generalized anxiety, most delivering cognitive behavioral therapy and one delivering mindfulness therapy. Mental Health Online and This Way Up both emanate from Australia. Stress and Mood Management is from the Center for Workforce Health and is a commercial program targeting workers. On the site, a suite of programs is offered on various health-related topics. My Student Body, likewise, offers a suite of programs, one being on Stress targeted to colleges and universities. Cognitive behavioral therapy was the mechanism used in all of the anxiety interventions except for one which used mindfulness. Mindfulness treatment was available in several languages which reduced anxiety, depression, and the severity of insomnia.

Post-traumatic stress disorder (PTSD) was addressed at two partner websites—one focusing on various trauma recoveries and the other directed to individuals who experienced disasters. It was tested in an RCT for hurricane survivors [112] and is available in English, Spanish, and Mandarin. The Internet sites for PTSD used various coping strategies and behaviors which were based on social cognitive theory. The modules included social support, self-talk, relaxation, trauma triggers, unhelpful coping, and professional help.

There are functional websites for the treatment of specific phobias and panic disorders as well. Both Mental Health Online and This Way Up address these disorders through cognitive behavioral therapy. In addition, This Way Up offers a 10-week program for individuals with OCD which was effective in reducing the symptoms of OCD, as well as reducing distress and depression in those with OCD.

Talk to Me was developed by Spanish psychologists to treat fear of public speaking and is available for use. This 2-month program using cognitive behavioral techniques was effective in decreasing the fear and avoidance of public speaking. This same group developed Without Fear which is an Internet-delivered program for fear of small animals (spiders, cockroaches, mice).

Diet and Physical Activity

There were several websites on diet and physical activity interventions. In general, these programs included interactive components with goal setting and personalized feedback. Often self-monitoring and tracking of progress were included. Some of the commercial websites were found to be efficacious in terms of reducing body mass index or weight. The Biggest Loser Club was efficacious (decreased weight, body mass index, waist circumference) with a 12-week program [123]. Additional support (periodic reminders) did not improve the basic Internet program [123].

Dietary Approaches to Stop Hypertension (DASH) for Health is another evidence-based program with a focus on diet and physical activity. For those who completed 12 months of use, overweight or obese individuals lost weight (mean decrease of 4 pounds) [126]. Overall, people with hypertension lowered their systolic blood pressure by an average of 7 mmHg [126]. It also led to increased consumption of fruits and vegetables and lower consumption of carbonated beverages. The program included weekly education, motivation, and mechanisms for self-monitoring with progress reports [126].

The Center for Workforce Health includes a suite of programs, some of which specifically address diet and physical activity. The RCTs captured in this review indicated that the two modules entitled Stress and Mood Management and Food Smart showed health benefits. There was a reduction in stress after the completion of Stress and Mood Management (3-month program), and after the Food Smart program, there was improvement in dietary self-efficacy, dietary attitudes, and dietary stage of change [108,122].

My Student Body also contains a suite of Internet programs. Our review indicated that the packages for Nutrition, Alcohol use, and Stress were efficacious in various RCTs. These are now combined and sold commercially, generally to colleges and universities. My Student Body—Nutrition specifically increased the intake of fruit and vegetables.

Healthy Living Check, Active Living Every Day, and My Path to Healthy Life addressed both diet and physical activity. Completion of these programs led to various health benefits including a reduction in the intake of saturated fat, reduction in waist circumference, and an increase in physical activity. My Path to Healthy Life was targeted to adults with diabetes mellitus [121], and completion of this program led to a decrease in fat intake and an increase in physical activity.

Some health interventions were paired with other interventions. ExecuPrev (LeadWell LiveWell) paired a leadership intervention with cardiovascular disease prevention. This program decreased waist circumference in women, although it did not affect body mass index overall [125].

Disease Management

All of the meta-analyses of RCTs on insomnia indicated that Internet-based cognitive behavioral therapy for insomnia was efficacious. In general, the therapy was delivered over several months through a series of modules and included self-monitoring through sleep diaries. The content of the therapy often included sleep information, sleep hygiene, relaxation, stimulus control, sleep restriction, and various cognitive techniques such as restructuring, paradox, mindfulness, imagery, putting day to rest, and thought stopping. Only some of the applications, however, are currently available. Insomnie was developed in the Netherlands and is available in Dutch. SHUTi, Sleepio, and RESTORE are available commercially in English. Generally, completion of these programs takes weeks to months, with specified activities required during each step of the program. The main health benefits were improvement in sleep efficiency and sleep quality, with a decrease in the severity of insomnia.

There were two evidence-based sites for chronic pain: painACTION and the Chronic Pain Management Program. The free site painACTION offers programs in back pain, migraines, neuropathic pain, and pain due to cancer or arthritis. painACTION is a 4-week course followed by 5 monthly boosters and includes self-management education in which problem solving skills were taught to reach specific goals [133,134]. The 6-week Chronic Pain Management Program covered four domains: cognitive (thinking better), behavioral (doing more), social (relating better), and emotional (feeling better). Both pain-related sites utilize various cognitive behavioral approaches with self-management strategies and interactive elements. These programs led to a reduction in the intensity or severity of pain, as well as a reduction in stress, anxiety, and depression.

Disease Prevention

The Blood Pressure Action Plan (now called the Heart and Stroke Foundation Health e-Support program) resulted in lower systolic blood pressure, lower pulse pressure, and lower total cholesterol in those individuals who completed the 4-month program [136]. DASH for Health also lowered systolic blood pressure [126]. These programs involved setting priorities and included self-monitoring, progress reports, and tailored advice.

Childhood Health Problems

There were two efficacious programs on childhood health problems. Triple P Online (TPOL) assists parents in addressing behavioral programs in children through teaching positive parenting skills. It was shown to decrease problematic child behavior, dysfunctional parenting styles, parental anger, and to improve parent’s confidence. It has been studied quite extensively and is used in 25 countries throughout the world (with availability in English, Spanish, Dutch, and German). The techniques used included goal setting, evaluation, self-efficacy, personal agency skills, with video-based modeling, experiential learning, prompting, and customizable output. The other site for childhood problems is entitled U Can POOP Too which addresses encopresis. It has been shown to reduce fecal soiling and improve toileting skills through various cognitive behavioral approaches including reinforcement and modeling of behaviors and actions.

Measures of Efficacy

The principal measure of efficacy was NNT; these were calculated for binary outcomes and are shown in Table 5. For continuous outcomes, changes in mean differences between the intervention and control groups are shown. Overall for substance abuse (alcohol, tobacco, and cannabis), the effect was moderate with NNTs of 9-26 for avoidance or reduction in use over a short-term period (up to 6 months). For mental health problems, the degrees of effect were commonly reported using conventional scales within each field. In general, the effects were moderate with a decrease in depressive symptoms, anxiety, or stress. There were two interventions (Deprexis and Mental Health Online for panic disorders) which demonstrated particular efficacy (ie, low NNTs of 4 and 2, respectively). The efficacy of the interventions for diet and physical activity, although significant, was modest (eg, 2.1 kg mean weight reduction compared with a 0.4 kg increase in controls). There were several efficacious interventions for insomnia; the severity of symptoms, in general, decreased moderately. For example, SHUTi showed an 8-point relative reduction in severity on the Insomnia Severity Index. In addition, Sleepio also demonstrated an increase in daytime performance (2.5 points on a 5-point scale). The intervention painACTION was particularly efficacious, with an NNT of 4 for back pain and a NNT of 3 for migraine headaches. The interventions targeted to parents of small children were also very efficacious. Triple P Online had an NNT of 3 for clinical improvement in behavioral problems in children and U Can POOP Too had an NNT of 4 for prevention of fecal accidents.

Discussion

Principal Findings

In this systematic review, we developed a list of Internet health-related programs that demonstrated an evidence-based health benefit. The majority of programs dealt with substance abuse, mental health, or diet and physical activity. In addition, there were Internet programs dealing with disease management such as insomnia and chronic pain, as well as evidence-based Internet therapies for childhood health problems. There were some interventions with considerable efficacy (NNT<5); these included painACTION, Mental Health Online for panic disorders, Deprexis, Triple P Online, and U Can POOP Too.

There were several characteristics of successful Internet-delivered health interventions. First, most of the programs were rather intensive; they required assignments and engagement by the user over the course of weeks to months. For a number of the programs, not only were there interactive elements that prompted personalized feedback and self-monitoring, but also there were assignments that required the user to implement actions when they were not on the Internet such as tracking their sleeping habits via a diary, recording their eating habits throughout the day, or conducting physical activities throughout the week. In all of the therapies, educational materials were presented but these were often adjuncts to the main therapeutic approaches—not the principal tactic. Often the interventions followed cognitive behavioral strategies that were well-grounded in the psychological literature. Thus, most of the successful interventions were not truncated bits of information delivered in a short period of time. They were well-thought out progressive modules of engagement with multilayers of targeted approaches. Many also encouraged individuals to seek professional assistance if further help was needed.

Perhaps the most desirable aspect of having Internet evidence-based programs is the sheer magnitude of the audience. There were 3.5 billion Internet users in the world by December 2016 with a steady increase over the past decade [146]. By providing evidence-based programs, the potential to ameliorate some health problems or behaviors is enormous—even if the completion rates are rather low. The challenge is to determine whether these types of programs work equally well when translated into other languages and delivered to people with different social and cultural backgrounds. More information is needed regarding the triggers of personal readiness to use such programs and what factors appear to serve as enablers to use.

We found that 25 Internet programs were free to the public although some require registration. The availability of free health information removes a key barrier to the public, particularly individuals with lower incomes. The Pew Research Center found that 26% of Internet users who wanted health information were asked to pay, but only 2% of them actually paid for the information [5]. Request for payment resulted in lower-income individuals giving up the search, whereas wealthier individuals sought other avenues for the information [5]. This is of consequence because uninsured and poor individuals tend to have disproportionately higher rates of some health-related behaviors that such programs may help to abate [147].

Therapies that are Internet-based offer an attractive option for certain types of conditions due to easy access and low cost. In some locations, there may be insufficient numbers of clinicians who provide specific therapies, such as cognitive behavioral therapy for insomnia. These programs may also resolve other access problems, such as long wait times or lack of transportation to services. Such programs may be a choice for the first-line of engagement and, if the problem is not resolved, further in-person visits could be arranged. Many of these successful programs provide links to additional resources and some have specific information for health care professionals. The substance abuse websites are particularly strong in providing such links.

Another desirable feature of Internet programs is the ability to reach individuals who shun public places and therefore, are less likely to seek face-to-face care. Our review indicates that there are evidence-based programs for several phobias including social phobias (including shyness), panic attacks, and OCD. Moreover, such programs may reduce the likelihood of social stigma which sometimes occurs when seeking traditional avenues for assistance. For individuals with such problems, Internet programs may have the potential to provide the first step to eventual engagement with medical and neighborhood communities.

Although we anticipated that the Internet would be a valuable location for programs related to sexual health, it was surprising not to find any current evidence-based websites on the prevention of sexually transmitted disease through this review. There were several evidence-based programs for HIV prevention and unintended pregnancy that yielded a health-related benefit, but the programs that were tested did not yield a functional site to continue the program after the completion of the trials.

One of the challenges of Internet-delivered therapies relates to the constraints of the modality itself. There may be problems for individuals with vision problems or those with specific functional disabilities. However, adaptive approaches may be possible to deliver audio programs for those who are blind and modifications may be available for those with specific motor-related disabilities. There are many case studies of computer technologies which have advanced the functional capabilities of those with various limitations; these include approaches which alter input devices, the use of assistive tools for processing, and restructuring the output [148,149].

Limitations

There are several limitations to this systematic review. There may be some evidence-based websites on health that were missed; we only included those RCTs that were part of a meta-analysis. Therefore, continuous updating will be necessary. This review is but the first step in this process; the development of mechanisms for continuous review is the next. Another limitation was that our focus was on self-help Internet programs. In this review, we did not include Internet-delivered health interventions that integrally involved clinicians, peer-to-peer therapies, or group therapies; an exhaustive review of each of these programs would be helpful for future research studies, so that the breadth of this field could be appreciated and any deficiencies identified. Moreover, this review is meant to initiate the process of dissemination of evidence-based websites and, therefore, additional steps will be necessary. We consider that this process will eventually become analogous to the procedures utilized during dissemination and implementation of conventional medical therapies. That is, RCTs are conducted and reviewed through meta-analysis. This is followed by professional guidelines and recommendations for use, based on the RCT evidence. This, then, typically yields studies which evaluate implementation in the wider population or within specific subgroups. Because these interventions are housed on the Internet, mechanisms for dissemination will involve Internet engagement but will likely require participation of public health professionals, policy makers, and providers of health care.

There are some precautions, however, when delivering Internet therapies directly to the public. Researchers understand that the demonstration of an overall benefit in an RCT relates to a group effect and that this does not necessarily indicate that every single person will receive a benefit. Therefore, part of the implementation process to the public should involve education regarding the limitations of evidence-based Web interventions. They do not guarantee a specific result; they only promise a greater likelihood of a benefit if the therapy is completed.

Conclusions

We identified several evidence-based health interventions that are currently available on the Internet. They include therapies related to substance abuse, mental health, diet and physical activity, disease management, disease prevention, and childhood health-related problems. Unfortunately, most of the Internet-delivered health interventions that were efficacious through RCTs were not available after the conclusion of the trials. The challenge is to find avenues through governments, organizations, universities, and interested corporations to host the evidence-based Internet programs and to notify the public of their locations. If this process is expanded, such therapies provide hope of a cost-effective mechanism to achieve healthier populations globally.