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.
Understanding how people participate in and contribute to online health communities (OHCs) is useful knowledge in multiple domains. It is helpful for community managers in developing strategies for building community, for organizations in disseminating information about health interventions, and for researchers in understanding the social dynamics of peer support.
We sought to determine if any patterns were apparent in the nature of user participation across online health communities.
The current study involved a systematic review of all studies that have investigated the nature of participation in an online health community and have provided a quantifiable method for categorizing a person based on their participation style. A systematic search yielded 20 papers.
Participatory styles were classified as either multidimensional (based on multiple metrics) or unidimensional (based on one metric). With respect to the multidimensional category, a total of 41 different participation styles were identified ranging from Influential Users who were leaders on the board to Topic-Focused Responders who focused on a specific topic and tended to respond to rather than initiate posts. However, there was little overlap in participation styles identified both across OHCs for different health conditions and within OHCs for specific health conditions. Five of the 41 styles emerged in more than one study (Hubs, Authorities, Facilitators, Prime Givers, and Discussants), but the remainder were reported in only one study. The focus of the unidimensional studies was on level of engagement and particularly on high-engaged users. Eight different metrics were used to evaluate level of engagement with the greatest focus on frequency of posts.
With the exception of high-engaged users based on high post frequency, the current review found little evidence for consistent participatory styles across different health communities. However, this area of research is in its infancy, with most of the studies included in the review being published in the last 2 years. Nevertheless, the review delivers a nomenclature for OHC participation styles and metrics and discusses important methodological issues that will provide a basis for future comparative research in the area. Further studies are required to systematically investigate a range of participatory styles, to investigate their association with different types of online health communities and to determine the contribution of different participatory styles within and across online health communities.
Participation rates of people in online communities are known to be highly variable with some people contributing much more than others. Across all types of online communities, the variability in degree of user participation consistently follows a pattern [
Although post frequency may constitute a simple indicator of engagement, from post frequency alone it is not possible to ascertain exactly what ways a person contributes. Post frequency does not indicate whether a person starts new discussions, welcomes newcomers, is available at critical times in the day when people are most likely to need support, or is knowledgeable about certain topics. In order to ascertain whether people contribute these different kinds of value, it is necessary to measure their participation based on various other metrics.
There may be value for those who are involved in the development of an OHC to identify users who contribute particular types of value to the OHC. This points to the need for multiple metrics to define user contributions. For example, in a qualitative paper on building and sustaining OHCs, Young described how certain core members were vital to the development and sustainability of an OHC [
For a variety of reasons, including time constraints and size of the community, not all community managers are able to have a strong qualitative understanding of the roles of particular individuals in their OHC. However, community managers would potentially benefit from a simple operationalization of user participation in terms of metrics that are automatically collected in the log data of the OHC software. This would help them to identify the core members and various other users who contribute in different ways so that they may apply the community building techniques recommended by Young [
OHCs also provide an opportune setting for interventions that encourage certain positive health behaviors [
Finally, there is scientific value in investigating the ways in which different people participate in OHCs across multiple contexts. There may be patterns in the way in which people participate that can be found across multiple different OHCs. These patterns may help us learn more about the social dynamics of OHCs and the way that people seek help and provide it to others.
User profiling by categorizing participation styles is conducted in studies of online communities more broadly. There are some roles such as “newbies” and “celebrities” that may be found in any online community, but most others are likely to be specific to the type of community [
This study seeks to advance this area by conducting a systematic review of all studies that provide replicable, quantifiable criteria for categorizing the nature of participation in an OHC. We aimed to document all participation styles that had been identified to date and the OHCs from which they came. Our objective was to determine if any patterns were apparent in the nature of user participation across OHCs for different health conditions or within each.
A systematic review was conducted to identify articles that investigated participation styles in an online health community. For the current purposes, an online health community was defined as any Internet-based platform designed to enable people to communicate about health issues. A participation style was defined as any type of engagement with an OHC that can be measured quantitatively. This does not include simply the presence or absence of participation (ie, posters and lurkers), as this has been well documented elsewhere [
Three databases (PubMed, PsycINFO, and Cochrane) were searched for all articles prior to December 2014. Adapted search terms from Eysenbach et al [
In addition, papers from relevant journals and conference proceedings in the computer and information science field published since 2005 (including the
A total of 7457 articles were screened. Of these, 3150 were retrieved from the database search and 4307 were from the additional journals and conference proceedings. A total of 82 duplicate articles were identified and removed. Relevant articles were selected through a multistage process (
Study identification flow diagram: PubMed (PM), PsychINFO (PI), Cochrane (C), Internet Interventions (II), International Conference on Healthcare Informatics (ICHI), American Medical Informatics Association Annual Symposium (AMIA), Journal of the American Medical Informatics Association (JAMIA), Journal of the Association for Information Science and Technology (JASIST).
The final set of articles included any study that (1) quantitatively investigated ways that people participate in an online health community, and (2) categorized users based on any quantifiable metric that can be used to show they have engaged with the community.
Studies that converted written content to quantitative data by a means that was computerized (eg, machine learning algorithm) were included, but studies that relied on human interpretation of written content to create quantitative data were not. This ensured that the methods identified could be accurately replicated and would be scalable to large OHCs. For similar reasons, studies that used self-report data from surveys were not included. This meant that only studies reporting data that had been automatically logged by the OHC software or that had been extracted by programs that crawl publicly available data were included in this systematic review. Protocol papers, articles not written in English, and papers on OHCs solely for health practitioners were not included.
After applying these criteria, a set of 15 papers were included. The reference lists of included papers and those that cited them (as per Google Scholar) were hand searched. This yielded an additional 5 papers, resulting in a final set of 20 included papers.
The included papers were coded by 1 rater (BC). Each participation style identified by a paper was listed. Three attributes of each participation style were coded: (1) the name used by the authors to describe the participation style, for example, “superuser,” (2) the metrics used to quantitatively describe their style of participation, for example, frequency of posts, and (3) the inclusion criteria used to determine who was categorized as having that participation style, for example, the top 1% of users whose frequency of posts was greatest were deemed to be superusers.
Across the final set of 20 papers, users were categorized into participation styles a total of 74 times, of which 28 were duplicates. These duplicates included participation styles that had been assigned different names by different studies but used the same metrics and same inclusion criteria (or very similar) to define them. By merging all these redundant categorizations into the same participation style, we determined that 44 participation styles had been identified in OHCs to date.
There were 41 participation styles in the multidimensional category (13 activity based, 11 network based, and 17 content based). In all instances where a unidimensional participation style was identified, the studies divided the users into no more than 3 groups that we have summarized as high, medium, and low engagement. There were 8 different metrics used in the high engagement category (5 activity based, 3 network based), 3 in the medium category (2 activity based, 1 network based), and 4 in the low category (3 activity based, 1 network based).
The results of each subcategory of participation style (content based, network based, and activity based) are described in turn for the 41 multidimensional participation styles. Following this, the results of the unidimensional participation styles are described together for each of the 3 participation styles identified.
Summary of online health community characteristics.
Online health community name | Year of study | Health condition | Country | Sample size, n |
SOL-Cancer Forum | 2007 [ |
Cancer | Not reported | 84 |
Cancer Survivors Network | 2014 [ |
Cancer | United States | 27,173 |
Cancer Compass | 2011 [ |
Cancer | United States | 7991 |
WebChoice | 2013 [ |
Cancer (breast and prostate) | Norway | 103 |
Breastcancer.org | 2014 [ |
Cancer (breast) | United States | 49,552 |
Cancer Compass | 2010 [ |
Cancer (melanoma) | United States | 851 |
Five unnamed forums in English and Spanish | 2013 [ |
Diabetes | Not reported | >140,000 |
BlueBoard | 2014 [ |
Mental health | Australia | 2932 |
DepressionCenter | 2014 [ |
Mental health (depression) | Not reported | 5151 |
PanicCenter | 2014 [ |
Mental health (panic disorder) | Not reported | 11,372 |
AlcoholHelpCenter | 2014 [ |
Mental health (problem drinking) | Not reported | 2597 |
PTT.CC—Psychosis Support Group | 2009 [ |
Mental health (psychosis) | Taiwan | 438 |
SharpTalk | 2011 [ |
Mental health (self-harm) | United Kingdom | 77 |
Deutsche Multiple Sklerose Gesellschaft | 2014 [ |
Multiple sclerosis | Germany | 1169 |
The Canadian Cancer Society’s Smokers’ Helpline Online | 2012 [ |
Smoking | Canada | 1670 |
QuitBlogs | 2014 [ |
Smoking | New Zealand | 3448 |
Alt.Support.Stop-Smoking | 2014 [ |
Smoking | Not reported | 8236 |
QuitPlan | 2008 [ |
Smoking | United States | 233 |
QuitNet | 2010 [ |
Smoking | United States | 7569 |
2013 [ |
Not reported | |||
StopSmokingCenter | 2012 [ |
Smoking | United States | 1627 |
2014 [ |
44,870 | |||
#HCSMCA | 2013 [ |
Social innovation in health care | Canada | 486 |
Summary of participation styles including name, metrics, and inclusion criteria.
Name | Metrics | Inclusion criteria | ||
|
||||
|
|
|||
|
|
Influential user [ |
69 activity, network, and content features including influential responding replies | Machine learning classifier (relying initially on expert judgement to identify exemplars) |
|
|
Leader [ |
68 activity, network and content features | Machine learning classifier (relying initially on expert judgement to identify exemplars) |
|
|
Opinion leader [ |
Word vectors, degree | Latent semantic analysis and high degree |
|
|
Information providers [ |
Social support type | High information support |
|
|
Community builders [ |
Social support type | High companionship support |
|
|
Emotional support providers [ |
Social support type | High emotional support |
|
|
Information seekers [ |
Social support type | High information support seeking |
|
|
Emotional support seekers [ |
Social support type | High emotional support seeking |
|
|
Information enthusiasts [ |
Social support type | High information support seeking, high information support |
|
|
All-around contributors [ |
Social support type | No particular metric stands out |
|
|
Balanced source user [ |
Source of information | Cited information from a range of sources |
|
|
Social media fan [ |
Source of information | High social media |
|
|
Organization follower [ |
Source of information | High organizations |
|
|
Homepage promoter [ |
Source of information | High static informational websites |
|
|
Seeker of health care [ |
Source of information | High health practitioners |
|
|
User of uncommon sources [ |
Source of information | High uncommon sources |
|
|
Sophisticated contributor [ |
Word count, source of information | High word count, high academic references |
|
|
|||
|
|
Key player [ |
Degree (nonredundant) | Key Player 1.4 software |
|
|
Hub [ |
Out-degree, in-degree | Hyperlink-induced topic search algorithm |
|
|
Authority [ |
Out-degree, in-degree | Hyperlink-induced topic search algorithm |
|
|
Facilitator [ |
Out-degree, in-degree | Hyperlink-induced topic search algorithm |
|
|
Trusted user [ |
Out-degree, in-degree | PageRank algorithm |
|
|
Help-seeker [ |
Out-degree, in-degree | Low in-degree, high out-degree (within the scope of the edge between 2 users) |
|
|
Star [ |
Out-degree, in-degree | Top ranked individual (outlier) |
|
|
Prime givers [ |
Out-degree, in-degree | Very high out-degree, high in-degree |
|
|
Serious members [ |
Out-degree, in-degree | Moderate out-degree, moderate in-degree |
|
|
Moderate users [ |
Out-degree, in-degree | Low out-degree, low in-degree |
|
|
Takers [ |
Out-degree, in-degree | No out-degree, low in-degree |
|
|
|||
|
|
Caretaker [ |
Time logged in, episodes, reading, posting, thread initiation | High time logged in, low episodes, high reading, low posting, low thread initiation |
|
|
Here for you [ |
Thread initiation, posting, forum | Low thread initiation, high posting in support forum |
|
|
Butterfly [ |
Time logged in, episodes, posting, forum | High time logged in, high episodes, high posting in support forum |
|
|
Crisis-oriented individual [ |
Posting, forum | High posting in support forum |
|
|
Discussant [ |
Thread initiation, posting, forum | High thread initiation, high posting in discussion forum |
|
|
Average user [ |
Thread initiation, posting, forum, topic, days active, word count, source of information | No particular metric stands out |
|
|
Highly active relational poster [ |
Posts per day, thread participation, thread initiation | High posts per day, high thread participation, low thread initiation |
|
|
Topic-focused responder [ |
Thread initiation, posting, topic, days active | Low thread initiation, low posts per day, high fraction of topic-related posts, low days active |
|
|
Topic-spammer [ |
Posting, days active, word count, topic, source of information | Low days active, high posting, low word count, high fraction of topic-related posts, low references |
|
|
Long-term high-activity users [ |
Days active, posting | High days active, high posting |
|
|
Short-term high-activity users [ |
Days active, posting | Low days active, high posting |
|
|
Short-term low-activity users [ |
Days active, posting | Low days active, low posting |
|
|
Long-term low-activity users [ |
Days active, posting | High days active, low posting |
|
||||
|
|
|||
|
|
High-engaged user | Posting | >2 posts [ |
Reading | >5 posts [ |
|||
Time logged in | Top 33.3% of users [ |
|||
Thread initiation | Top 100 users [ |
|||
Thread participation | Top 100 users [ |
|||
Network based |
|
|||
Friendship | Mutual friend nomination between 2 users and >4 interactions between them [ |
|||
In-degree | Top 10 users [ |
|||
Out-degree | Top 10 users [ |
|||
|
|
Moderate-engaged user | Posting | 2-10 percentile (9%) of users [ |
Time logged in | Middle 33.3% of users [ |
|||
Network based |
|
|||
Friendship | Friend nomination of another user and >0 interactions with them [ |
|||
|
|
Low-engaged user | Posting | 1-2 post [ |
Reading | 1-5 posts [ |
|||
Time logged in | Bottom 33.3% of registered users [ |
|||
Network based | ||||
Friendship | Any interactions with another user [ |
A description of the mtrics used to classify participation styles.
Metric | Description | |
Activity-based metrics: measure the individual actions taken by users in an OHC | ||
|
Posting | Number of posts a person has made in the OHC |
|
Time logged in | Amount of time a person has spent accessing the OHC |
|
Reading | Number of posts that a person has read |
|
Thread initiation | Number of times a person has created a thread |
|
Episodes | Number of times a person has accessed the OHC |
|
Days active | Number of days between a person’s first and last post |
|
Forum | Number of posts a person has made in a particular subforum of the OHC, eg, support or discussion |
|
Thread participation | Number of different threads a person has posted in |
Network-based metrics: measure the relationship and interactions between users | ||
|
Degree (in/out) | The number of people a person has communicated with. Where it is possible to tell who the source of the communication was and to whom it was directed, the number of people a person has made outgoing communication with is called the “out-degree” and the number of people that a person has received communication from is called the “in-degree.” When it is not possible to tell the direction, the communication is counted for both people as a measure of degree. Degree is considered to be a measure of a user’s centrality in a network [ |
|
Friendship | The extent to which a person is connected with at least one other person in the OHC as defined by 3 thresholds: Low—any interactions with another user; Moderate—friend nomination of another user and >0 interactions with them; and High—mutual friend nomination between 2 users and >4 interactions between them. |
Content-based metrics: measure the nature of the content within posts | ||
|
Word vectors | A representation of the proportion of words in a message that fit a certain topic. |
|
Influential Responding Replies | Number of posts a person has made that have influenced the sentiment of the thread initiator |
|
Social support type | Number of posts a person has made that either provide or seek information support, emotional support, or companionship |
|
Topic | Number of posts a person has made which included subject matter on a specific topic |
|
Source of information | Number of citations a person has made from a particular source |
|
Word count | Number of words in a post |
Zhao et al [
Myeni et al [
Wang et al [
Sudau et al [
A sophisticated contributor is a user whose posts are longer than those of the average user participation style and contain more references. In contrast to the activities of most users, these references are more often to scientific publications than to social media sources. Sudau et al [
Cobb et al [
Hubs and authorities are concepts borrowed from the computer science literature on the Web. Hubs and authorities are identified using the hyperlink-induced topic search (HITS) algorithm [
Similar to the HITS algorithm, the PageRank algorithm [
In a relationship between 2 people where one communicates with the other much more often, the person who instigates more communication (higher out-degree than in-degree) is labeled a Help-Seeker. Chomutare et al [
The earliest recorded participation styles were identified by Bambina [
Jones et al [
One user in a sample of 77 people was considered to take the “here for you” participation style by Jones et al [
Another user in the Jones et al [
Six users of the Jones et al [
A discussant is a user who is mainly focused on discussion about health-related topics as opposed to providing or receiving support. They initiate a high number of threads in the discussion section of the OHC and participate actively in them. This participation style was identified by both Jones et al [
A user type that is not distinctly based on any metric, the average user category was identified by the application of a second k-means clustering algorithm conducted by Sudau et al [
These are the most active users of an OHC by post frequency. Sudau et al [
A user whose activity is concentrated on a specific topic, the topic-focused responder is distinct from a discussant in that they do not post as much and do not initiate as many threads. Sudau et al [
This is a user who is active for a very short period, that is, only a few days. In that time, they contribute a high number of posts on a specific topic in the discussion forum. However, these are not particularly sophisticated posts, rather they are short and lack references. This participation style was identified by Sudau et al [
Stearns et al [
All but one of the studies [
All 8 studies that made a unidimensional categorization [
Frequency of posting was the most commonly used metric used by 6 of the 8 studies [
Thread initiation and thread participation (together with posting frequency) were used by one study [
In-degree and out-degree were employed by 2 studies to classify users as highly engaged [
Other metrics employed to classify users as highly engaged included reading [
Four studies classified users as moderately engaged based on 3 different metrics. Two were based on posting frequency [
Six studies classified users as low engaged based on 4 different metrics. Four were based on posting frequency [
This systematic review synthesized findings from studies that investigated the nature of participation in an OHC by categorizing users based on metrics of participation. The aim of this review was to identify the different ways in which users participate and contribute to OHCs, although we acknowledge that the resultant list of participation styles may not provide a comprehensive account of all possible styles. Our objective was to determine whether any patterns were apparent in the types of participation styles that were identified across and within different health conditions. With the exception of an overlap in engagement measured by posting frequency (which has been discussed elsewhere [
It was common for studies to use posting frequency as the sole means of classifying highly engaged users in an OHC. It was also common among these studies for researchers to regard these users as being particularly valuable to the OHC. However, it is not possible to know from post frequency alone in what way a person is contributing to an OHC. They might be contributing trivial or critical messages or their post might in other ways fail to support others. The rationale for the inference that high engagement is synonymous with high value may relate to another commonality across papers. The authors in question were also community managers of the OHCs that they were studying; therefore, they may have based their conclusions on reading content posted by these users. However, content analysis research is required to investigate whether posting frequency is a valid means of identifying generically valuable users.
Zhao et al [
Wang et al [
Similar to IRR, some participation styles described users who were useful in a particular way that would be potentially identifiable in any OHC, or for that matter, any social network. These were based on algorithms that used measures of centrality such as in-degree and out-degree. This includes authorities, hubs, facilitators, and trusted users. While these categories are quite useful, it should be noted that these algorithms are calculated in such a way that they introduce bias based on time elapsed such that users who participate earlier in the OHC receive higher scores [
Other more specific participation styles described users who have particular characteristics and may be found only in a subset of OHCs. This included, for example, the caretaker or the topic-spammer. The techniques used to identify these participation styles, k-means clustering algorithms and multivariate outliers, may not necessarily identify the same participation styles in other OHCs. However, they may be useful for identifying other particular or unique ways of participating in OHCs.
The scope of this study is quite broad. We included all studies that categorized a type of participation in an OHC despite the possibility that the culture and nature of participation in populations with different health conditions and with or without moderators could differ markedly. There was little overlap in the use of categorizations to define particular participation styles either in OHCs broadly or within specific health conditions. Thus, it is not possible to draw many specific conclusions at this early stage. A possible limitation and reason for this is that we may not have included all relevant studies, as our search terms may not have encompassed all the different terms used to describe participation styles at this early stage of research. Nevertheless, by synthesizing the findings of the included studies, this review provides a basis for future research to investigate the validity of styles identified to date by attempting to replicate findings for specific OHCs and exploring their validity across different OHCs. Future research should also investigate new participation styles not documented in this review.
Our systematic review identified a range of participation styles. Some of them may be generalizable to other OHCs. Others were more specific to particular OHCs but were identified by methods that could be used elsewhere. The findings of this review are intended to support the work of community managers in building community, organizations seeking to design targeted interventions and disseminate information through certain types of people in OHCs, and researchers seeking to understand the nature of peer support. We anticipate that this review will be useful for these groups in conducting investigations to determine the presence of participation styles that may be relevant to their work. However, it is too early to draw any conclusions about which OHCs would be most likely to contain users who have specific participation styles.
OHC concept search terms.
hyperlink-induced topic search
influential responding replies
online health community
BC is supported by an Australian Postgraduate Award. KA is supported by a Young and Well CRC PhD Scholarship. KG is supported by the Australian National Health and Medical Research Council (NHMRC) Research Fellowship .
None declared.