This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Online social networks are popular components of behavior-change websites. Research has identified the participation of certain network members who assume leadership roles by providing support, advice, and direction to other members. In the literature, these individuals have been variously defined as
To analyze data from two large but distinct Web-assisted tobacco interventions (WATI) to help gain insight into superuser demographic characteristics and how they use social networks.
We extracted cross-sectional data sets containing posting behaviors and demographic characteristics from a free, publicly funded program (the Canadian Cancer Society’s Smokers’ Helpline Online: SHO), and a free, privately run program (StopSmokingCenter.net: SSC).
Within the reporting period (SHO: June 26, 2008 to October 12, 2010; SSC: May 17, 2007 to October 12, 2010), 21,128 individuals registered for the SHO and 11,418 registered for the SSC. Within the same period, 1670 (7.90%) registrants made at least one post in the SHO social network, and 1627 (14.25%) registrants made at least one post in the SSC social network. SHO and SSC superusers accounted for 0.4% (n = 95) and 1.1% (n = 124) of all registrants, and 5.7% (95/1670) and 7.62% (124/1627) of all social network participants, and contributed to 34.78% (29,422/84,599) and 46.22% (61,820/133,753) of social network content, respectively. Despite vast differences in promotion and group management rules, and contrary to the beliefs of group moderators, there were no statistically significant differences in demographic characteristics between the two superuser groups.
To our knowledge, this is the first study that compared demographic characteristics and posting behavior from two separate eHealth social networks. Despite vast differences in promotional efforts and management styles, both WATI attracted superusers with similar characteristics. As superusers drive network traffic, organizations promoting or supporting WATI should dedicate resources to encourage superuser participation. Further research regarding member dynamics and optimization of social networks for health care purposes is required.
Tobacco-related illnesses are the leading cause of death in North America, yet 46 million (21%) Americans [
In traditional treatment for tobacco dependence, extensive evidence proves the effectiveness of behavioral interventions such as brief or intensive advice, individual or group counseling, tailored self-help, and telephone quitlines [
As of June 30, 2010, it is estimated that 29% of the world’s population has access to the Internet [
Following this trend, Web-assisted tobacco interventions (WATI) are proving to be efficacious [
In a 2010 study, Cobb et al [
Social network members who assume leadership roles by providing direct support, advice, and direction are defined by Cobb et al as
While the potential to help individuals through legitimate and validated eHealth interventions is exciting, the explosive growth of Internet access parallels the ever-increasing number of websites and Internet protocols (or IPs or URLs, commonly referred to as
One method for legitimate eHealth interventions to attract users is through directories such as Beacon [
An alternative grassroots method that can attract general users is through social networking and the retention of superusers. As they expand the size of a network and facilitate discussion, superusers are valuable assets for eHealth social networks to recruit and retain.
For organizations implementing and managing online social networks, identifying and retaining superusers could contribute to the natural development and growth of website traffic and promote adherence. The challenge is determining how to identify superusers, how to attract them, and how to promote their retention. The first step in this process is to begin to understand who superusers are.
As an initial step designed to investigate superuser characteristics, this observational study analyzed data from two large social networks for smoking cessation: the Canadian Cancer Society’s Smokers’ Helpline Online (SHO) (http://smokershelpline.ca;
SmokersHelpline.ca version 6.0 home page.
SmokersHelpline.ca version 6.0 support group home page.
StopSmokingCenter.net version 7.1 home page.
StopSmokingCenter.net version 7.1 support group home page.
As a publicly funded program with an internationally recognized and respected brand, SHO undertakes substantial promotional campaigns extending to Internet, television, radio, and print advertising, attracting many smokers. The SHO social network is also used to promote offline activities, such as The Ontario Driven to Quit Challenge. Also, SHO program health educators are required to rigorously monitor content posted by members. An unlimited number of individuals may register for the SHO program. The SHO is a licensed version of the SSC that is modified and updated by the Canadian Cancer Society.
Contrarily, the SSC is not a recognized brand, has never been advertised or promoted, and is maintained as a social enterprise. The organization that operates the SSC (Evolution Health System Inc, San Francisco, CA, USA) does not actively promote the program, nor does it optimize search engine rankings or actively seek links from other websites. Therefore, members using the SSC are a self-seeking, naturalistic Internet population. Although users must endorse a user agreement to participate in the SSC social network, unlike in the SHO, health educators control very little content posted by members.
To actively participate in each social network, registration is required. However, actively participating in the social networks is voluntary; in the study period only 7.90% (n = 1670) of SHO registrants and 14.25% (n = 1627) of SSC registrants made at least one post in the social network (see
Total program registrations and social network participation in Smokers’ Helpline Online (SHO) and StopSmokingCenter.net (SSC) Web-assisted tobacco interventions.
Characteristic | SHO | SSC |
Study period | June 26, 2008–October 12, 2010 | May 17, 2007–October 12, 2010 |
Duration (days) | 839 | 1245 |
Number of registrants | 21,128 | 11,418 |
Members with at least one post in the social network, n (%) | 1670 (7.90%) | 1627 (14.25%) |
Every post made in each support group is published immediately. However, to ensure compliance with program rules and regulations, trained health educators review and approve each post through WebTriage (Evolution Health System Inc, San Francisco, CA, USA), a software application designed to facilitate the rapid approval, editing, or deletion of each post.
SHO health educators are paid employees who also manage the Canadian Cancer Society’s telephone quitline. SSC health educators are paid employees of Evolution Health Systems Inc. In addition to training in the use of WebTriage, all health educators receive training in cognitive behavioral therapy, stages of change, motivational interviewing, life coaching, data protection, and user privacy. During high-volume seasons, SSC health educators assist SHO health educators with moderating the SHO social network (WebTriage).
For over 5 years, SHO and SSC social network managers have met together on a bimonthly basis to discuss social network management policies, specific incidences within each support group, moderating strategies designed to increase social network traffic, and the management of superuser populations.
On a quarterly basis, SHO and SSC social network managers meet to review cumulative statistical reports, which include the number of program registrants, number of social network posts, traffic statistics (page views, visits, and unique visitors), and basic demographic data collected at registration.
At the onset of this study, we conducted informal qualitative, in-person interviews with health educators and their management teams. In these interviews, we asked interviewees to reflect on bimonthly and quarterly meetings, each program’s promotional and operational differences, and each network’s tone and social environment. In these interviews all health educators felt that each program attracted superusers with heterogeneous characteristics, and that in each social network posting behavior and demographics would be correlated.
However, health educator and social network managers indicated that, based on their years of experience, they could not generalize superusers into a single category, and that three distinct types of superusers existed: (1) superusers who cumulatively authored the greatest number of posts (
All study participants consented to the use of their anonymous data for research purposes. Data collection procedures adhered to international privacy guidelines [
Registration to both SHO and SSC is anonymous and free of any commitment or fees. From June 26, 2008 to October 12, 2010 (839 calendar days), 1670 members posted at least once in the SHO social network, and there were 84,599 posts made in 7916 threads. From May 17, 2007 to October 12, 2010 (1245 calendar days), 1627 members posted at least once in the SSC social network, and there were 133,753 posts made in 10,967 threads (see
Total social network activity in Smokers’ Helpline Online (SHO) and StopSmokingCenter.net (SSC) Web-assisted tobacco interventions.
Characteristic | SHO | SSC |
Study period | June 26, 2008–October 12, 2010 | May 17, 2007–October 12, 2010 |
Duration (days) | 839 | 1245 |
Number of members with at least one post in the social network | 1670 | 1627 |
Number of posts in the social network | 84,599 | 133,753 |
Number of message threads | 7916 | 10,967 |
Number of posts in shortest thread | 1 | 1 |
Number of posts in longest thread | 73 | 87 |
Users completed a baseline questionnaire upon program registration. Questions were based on the North American Quitline Consortium Minimal Data Set [
In accordance with observations by Cobb et al [
In each social network, we combined the top 100 posters (ranked according to their total number of posts), thread starters (ranked according to the total number of threads they started), and clique members (ranked according to the number of threads they participated in) in a single database (n = 300). Duplicate entries were removed, leaving a sample of 219 unique superusers. We conducted a power analysis [
At registration, SHO and SSC used identical formats and response options for the following demographic variables: gender, age, cigarettes per day, past quit attempts, cohabitant smokers, years smoked, and past nicotine replacement therapy usage. Both programs collected other demographic data, but question formats or response options differed significantly, and thus we excluded those data from analysis.
Descriptive statistics pertaining to general posting behavior and demographic characteristics of both superuser groups are presented first. Next, we conducted sets of univariate logistic regressions to detect differences in demographic characteristics between SHO and SSC superusers. Finally, Pearson correlations were computed to detect relationships between behavior and demographic characteristics in each superuser subgroup. All analyses were performed using SPSS for Windows version 18.0 (IBM Corporation, Somers, NY, USA). The significance level was set at
Posting trends in smokershelpline.ca.
Posting trends in StopSmokingCenter.net.
The 95 SHO and 124 SSC superusers accounted for 0.4% and 1.1% of total program registrants and 5.7% and 7.6% of all active social network members. They were responsible for 34.78% (n = 29,422) and 46.22% (n = 62,820) of all social network posts (see
Analysis of demographic characteristics collected at registration (means) and their standard deviations revealed, from the perspective of marketing and moderating techniques, slightly distinct populations (see
Superuser social network activity in Smokers’ Helpline Online (SHO) and StopSmokingCenter.net (SSC) Web-assisted tobacco interventions.
Characteristic | SHO | SSC |
Total number of superusers | 95 | 124 |
Percentage of total registrants | 0.4% | 1.1% |
Percentage of active social network members | 5.7% | 7.6% |
Total number of posts in social network | 84,599 | 133,753 |
Posts in social network by superusers, n (%) | 29,422 (34.78%) | 61,820 (46.22%) |
Superuser demographic characteristics in Smokers’ Helpline Online (SHO) versus StopSmokingCenter.net (SSC) Web-assisted tobacco interventions.
Characteristic | SHO | SSC |
Gender (female), n (%) | 79 (83%) | 91 (73%) |
Age (years), mean (SD) | 46.0 (11.5) | 48.6 (9.5) |
Cigarettes smoked per day, mean (SD) | 20.3 (8.5) | 24.6 (11.2) |
Number of past quit attempts, mean (SD) | 3.5 (3.2) | 3.5 (3.1) |
Number of cohabitant smokers, mean (median) | 1.1 (1) | 0.5 (0) |
Smoking duration (years), mean (SD) | 17.3 (5.0) | 27.0 (10.8) |
Past or current nicotine replacement therapy usage, n (%) | 50 (52%) | 33 (26%) |
A multivariate logistic regression revealed no statistically significant differences in demographic characteristics between the two populations. Gender, age, cigarettes per day, past quit attempts, cohabitant smokers, years smoked, and past or current use of nicotine replacement therapy were all nonsignificant.
To gain further understanding of superuser and superuser-subset characteristics, we determined Pearson correlations to investigate associations between posting behavior and five key demographic characteristics (age, cigarettes smoked per day, quit attempts, cohabitant smokers, and years smoked).
In the SHO analysis (see
Bivariate correlations between Smokers’ Helpline Online superuser type and demographic characteristics.
Variable | All superusers |
Posters |
Thread starters |
Clique members |
Age | .32 (.002) | .22 (.08) | .17 (.18) | .23 (.07) |
Cigarettes per day | .05 (.66) | .06 (.66) | –.02 (.90) | .06 (.66) |
Quit attempts | –.07 (.52) | –.07 (.58) | .03 (.83) | –.06 (.62) |
Cohabitant smokers | .03 (.74) | .16 (.21) | .18 (.16) | .20 (.12) |
Years smoked | .09 (.37) | .13 (.27) | .14 (.17) | .27 (.03) |
In the SSC analysis (see
Bivariate correlations between StopSmokingCenter.net superuser type and demographic characteristics.
Variable | All superusers |
Posters |
Thread starters |
Clique members |
Age | .03 (.73) | .07 (.46) | .03 (.76) | .07 (.52) |
Cigarettes per day | .12 (.18) | .13 (.21) | .07 (.52) | .18 (.08) |
Quit attempts | –.25 (.01) | –.24 (.02) | –.16 (.11) | –.23 (.02) |
Cohabitant smokers | –.08 (.40) | –.02 (.90) | .02 (.82) | –.04 (.71) |
Years smoked | .24 (.01) | .25 (.01) | .18 (.07) | .25 (.01) |
At first glance, based on general demographic characteristics outlined in
However, contrary to our original hypothesis, and to information gleaned from informal qualitative in-person interviews with health educators, differences in marketing and the operation of both social networks, tone of discussions, and rules of conduct, superusers in both social networks had similar demographic characteristics.
We were also somewhat surprised to find minimal correlations between posting behavior and demographic characteristics in all eight of the superuser groups analyzed. For example, a Health Canada general population survey found correlations in demographic characteristics and behavior. Although there was little difference in quit rates between Canadian men and women, 29% of smokers aged 20–24 years had quit versus 71% of those aged 45+ years, former male smokers reported an average of 3.2 quit attempts before quitting for good (versus 2.7 quit attempts for females), and at the time of quitting, former smokers reported smoking 18.1 cigarettes per day [
Based on the Health Canada general population survey, one might expect to find strong correlations between frequency of posts and age, past quit attempts, or number of cigarettes smoked per day among superusers. Instead, we found only weak relationships between posting behavior, quit attempts, and years smoked, but only in SSC superusers, posters, and clique members.
To our knowledge, this is the first study that compared demographic characteristics and posting behavior of the most active participants (superusers) from two moderated social networks designed to assist with smoking cessation. A particular strength is that both programs have relatively few barriers to enrollment in comparison with entry barriers that are typical of Internet-based clinical trials [
Another strength is the applicability of findings. Results from this study will influence SHO promotional efforts and operations. Web promotion is generally directed at specific and targeted demographics, such as those reported in the Health Canada general population survey. However, to attract superusers, who are not easily defined, promotional efforts should be broadened. Second, as a results of this study management will modify the training programs of SHO and SSC health educators. Health educators will be introduced to strategies that encourage the participation of superusers once identified within the network, and work is underway to develop computer algorithms that will assist health educators with the early identification of superusers and superuser subtypes.
It is important to note that we have used the term superuser and the three superuser subgroups (posters, thread starters, and clique members) only for the purpose of clarifying observations within this specific investigation. These participatory patterns have not been validated, and further research is required to determine whether they are observable across other types of WATI or health care-based social networks.
To more thoroughly understand superuser characteristics, future studies should compare superuser populations versus nonsuperuser populations (or social network members who create limited posts) and those who register with WATI but do not actively participate in the social network (lurkers). Future research studies should also incorporate the analysis of additional demographic and psychographic characteristics such as occupation, level of education, Goldberg Depression Scale score, Fagerström Test for Nicotine Dependence score, or frequency of Internet usage. These data may be collected at registration; however, in our experience, questions presented to users at registration are most often regarded as intrusive. To maximize use of Web-based programs in a naturalistic setting, registration questions should be kept to an absolute minimum, and other creative means of data collection should be used within interactive program content.
It is also important to note that this study focused only on smokers, and future studies should examine social network behavior and demographic characteristics from superusers, superuser subsets, moderate posters, and lurkers from other condition areas such as depression, panic disorder, problem drinking, self-harm, or healthy weight. This type of research may offer insight into general user characteristics for those who are attracted to Internet programs, and may be used in the development of predictive algorithms.
The results of this study indicate that superusers play powerful roles within social network traffic. While SHO and SSC superusers accounted for only 0.4% and 1.1% of total program registrants, they were responsible for 34.78% and 46.22% of social network content. The results of this study support the previous research of Cobb et al, Selby et al, Cunningham et al, and Jones et al, indicating that different types of superusers (and superuser subgroups) frequently exist.
Significant and well-recognized barriers to treatment include social stigmas that prevent treatment seeking [
The results of this study suggest that further research in this fast-growing field is required, and that there is potential to maximize the impact of social networks that promote wellness. However, to fully understand the unique mechanisms of Internet-based behavior change, collaboration and knowledge transfer between researchers, nonprofit organizations, and private organizations is recommended.
Smokers’ Helpline Online
StopSmokingCenter.net
Web-assisted tobacco interventions
Smokers’ Helpline Online (SHO), one component of the Canadian Cancer Society Smokers’ Helpline, is sponsored by the Canadian Cancer Society and is available at http://www.smokershelpline.ca. StopSmokingCenter.net (SSC) is sponsored by Evolution Health Systems Inc. The authors would like to thank Dr Robert Mann, Senior Scientist at the Centre for Addiction and Mental Health and Associate Professor at the Dalla Lana School of Public Health, University of Toronto, and Dr Walid Hejazi, Professor of Quantitative Reasoning for Management and Associate Professor of International Business at the Rotman School of Management, University of Toronto, for their comments on the study design and an early draft of this paper.
Trevor van Mierlo is the CEO of Evolution Health Systems Inc. and the owner of StopSmokingCenter.net and other eHealth software platforms. Peter Selby received funds from Schering Canada to provide buprenorphine training, and received honoraria for consultant work, grant funding, advisory board, and/or lectureships from Johnson & Johnson Consumer Health Care Canada; Pfizer Inc, Canada; Sanofi-Synthelabo, Canada; GSK, Canada; Genpharm and Prempharm, Canada; CTI; Evolution Health Systems Inc., Canada; Health Canada; Smoke-Free Ontario; and Canadian Institutes of Health Research. Funding was in compliance with the Canadian Medical Association and the Canadian Psychiatric Association guidelines and recommendations for interaction with the pharmaceutical industry. Sharon Lee is employed by the Canadian Cancer Society. Rachel Fournier is employed by Evolution Health Systems Inc. Sabrina Voci has no interests to declare. None of the authors received any tobacco industry funds.