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.
Training for Australian general practice, or family medicine, can be isolating, with registrars (residents or trainees) moving between rural and urban environments, and between hospital and community clinic posts. Virtual communities of practice (VCoPs), groups of people sharing knowledge about their domain of practice online and face-to-face, may have a role in overcoming the isolation associated with general practice training.
This study explored whether Australian general practice registrars and their supervisors (trainers) would be able to use, and would be interested in using, a VCoP in the form of a private online network for work and training purposes. It also sought to understand the facilitators and barriers to intention to use such a community, and considers whether any of these factors may be modifiable.
A survey was developed assessing computer, Internet, and social media access and usage, confidence, perceived usefulness, and barriers, facilitators, and intentions to use a private online network for training purposes. The survey was sent by email link to all 139 registrars and 224 supervisors in one of Australia’s 17 general practice training regions. Complete and usable responses were received from 131 participants (response rate=0.4).
Most respondents had access to broadband at home (125/131, 95.4%) and at work (130/131, 99.2%). Registrars were more likely to spend more than 2 hours on the Internet (
General practice registrars and supervisors are interested in using a private online network, or VCoP, for work and training purposes. Important considerations are the extent to which concerns such as privacy and usefulness may be overcome by training and support to offset some other concerns, such as time barriers. Participants at an early stage in their training are more receptive to using an online network. More senior registrars and supervisors may benefit from more training and promotion of the online network to improve their receptiveness.
Training for general practice, or family medicine, in Australia is a postgraduate specialty program. After graduation from medical school, doctors spend a minimum of 1 year in the hospital system. To become a general practitioner, they must join a 3-year general practice training program run by one of 17 regional training providers across Australia. This program consists of 1 hospital year and 2 supervised general practice years. During these 3 years, trainees are required to work in a number of different rural and urban general practice locations, with at least 6 months located in a rural area. These locations are often small practices with a limited number of medical colleagues on-site, in contrast to the large hospitals with many colleagues that characterize early medical training.
As a result of these features, general practice training can be isolating [
The types of isolation experienced can be categorized as structural, personal, and professional [
The general practice workforce in Australia is under pressure [
A recent literature review proposed a role for virtual communities of practice (VCoP) in overcoming isolation, particularly professional isolation, through improved knowledge sharing [
This study explored whether Australian general practice trainees and their supervisors would be able to use, and would be interested in using, a VCoP of this type for work and training purposes. It also sought to understand the facilitators and barriers to intention to use, such as community, and considered whether any of these factors could be modified.
The sampling frame for the current study included all general practice trainees and supervisors in a large regional training provider in Australia in May 2010. In ascending order, the training levels are basic registrar, advanced registrar, subsequent registrar, supervisor, and educator. The training provider, Coast City Country General Practice Training (CCCGPT), provides general practice training across a wide geographic area, including the urban centers of Canberra in the Australian Capital Territory and Wollongong in New South Wales, alongside large regional and small rural centers spread across approximately 160,000 square kilometers.
Surveys were sent to all trainers and trainees on the CCCGPT database via an email link to SurveyMonkey [
Ethics approval was obtained from the University Human Research Ethics Committee.
There is a lack of literature on VCoPs in general practice training [
The instrument was piloted among a group of general practitioners, general practice trainees, and health researchers. Afterwards, a group discussion among pilot participants led to the amendment of wording and several response options alterations, to improve clarity and better reflect GP work.
The final survey consisted of 26 questions, including categorical and Likert response items (see
Data were analyzed using SPSS version 19 (IBM Corp, Armonk, NY, USA). Respondents were categorized as registrar or supervisor for comparisons between groups. The
Factor analysis using varimax rotation was used to determine which Likert items grouped naturally in questions with multiple Likert items for constructs such as computer confidence (questions 10 and 11) and usefulness (question 22). Factors were included if their eigenvalues were >1.0. The Cronbach alpha test for reliability was used to determine the degree of agreement between the Likert items. Cronbach alpha was >.8 for both items, higher than the recommended threshold of .70.
A confidence scale was constructed using all items from questions 10 and 11; the summated data were used as an independent variable in further analysis. The Pearson product moment correlation (
Survey content and question type.
Question content | Question type | Question number (categorical options or Likert items) |
Demographic | Categorical | 1 (2), 2 (2),3 (1), 4 (2), 5 (2) |
Access and usage | Categorical | 6 (2), 7 (2), 8 (6), 9 (7) |
Confidence | Likert items | 10 (4), 11 (7) |
Social networking usage | Categorical | 12 (2), 13 (9), 14 (11), 15 (2), 16 (9),17 (2), 18 (1), 19 (2), 20 (5), 21 (8) |
Usefulness | Likert items | 22 (14) |
Usefulness | Categorical | 27 (6) |
Barriers | Categorical | 23 (8), 24 (8) |
Intention to use | Likert items | 25 (2), 26 (2) |
Of the 131 respondents, gender was evenly split (males: 66/131, 50.4%; females: 65/131, 49.6%). Registrars accounted for 61.8% (81/131) of respondents and the remainder were supervisors. The response rate among trainees was higher than supervisors (registrar: 81/139, 58%; supervisor: 50/224, 22%). The mean age of the sample was 41.5 years (range 23-66 years, SD 10.369), with a significant difference between ages of trainees and supervisors (trainees: mean 35.9, SD 7.21; supervisors mean 51.0, SD 7.21,
Over half (75/131) of respondents were from rural settings, whereas the remainder worked in a general (nonrural) setting, with no significant differences between training stage and rurality or age and rurality.
Almost all general practice trainees and supervisors had access to broadband Internet at home (125/131, 95.4%) and at work (130/131, 99.2%). However, usage was found to be significantly different between registrars and supervisors, with 20.0% (10/50) of supervisors compared to 33.3% (27/81) of registrars spending more than 2 hours per day on the Internet (
Registrars were significantly more likely to use social networking sites for nonwork purposes (registrars: 41/81, 50.6%; supervisors: 14/50, 28%,
Out of all online social media activities, registrars and supervisors were most likely to watch online videos (registrars: 63/81, 77.8%; supervisors: 27/50, 54.0%), followed by reading discussions (registrars: 53/81, 65.4%; supervisors: 25/50, 50.0%). They were least likely to construct a wiki (registrars: 3/81, 3.7%; supervisors: 0/50, 0.0%). Video watching was significantly correlated with age, with younger users watching more video (
Factor analysis was performed on the 4 general computer confidence items, revealing only 1 factor, which was labeled
Confidence using discussion boards, wikis, blogs, online communities, chat, online video, and Twitter was assessed on a 5-point Likert scale for each of the 7 items. Confidence among supervisors was low to moderate, from a mean of 2.32 (SD 0.91) to a mean of 2.98 (SD 1.29), and was significantly lower than among registrars for all applications except Twitter, which was low for both groups (see
Factor analysis was performed on the 7 social media confidence items, revealing only 1 factor which was labeled
Cronbach alpha for the items in the confidence scale including all 11 items was .92. The inter-item correlations ranged between 0.21 and 0.78 indicating that there were no redundant items.
Using a 5-point Likert scale, 13 items were asked regarding perceived usefulness of social networks, regardless of whether the respondent currently used social networks, for aspects such as training purposes, keeping in touch with other trainees, job networking, and social support (
The question “keeping in touch with other registrars” was the only item to show a significant difference between registrars and supervisors (
A number of barriers to using social networks for work were described. The main concerns were worries about privacy (registrar: 61/81, 75.3%; supervisor 30/50, 60.0%) and insufficient time (registrar: 41/81, 50.6%; supervisor: 36/50, 72.0%; see
An important aim of the survey was to assess whether doctors would use a social network for training purposes. Respondents were asked whether they would use a private network or an open network, such as Facebook, for work purposes or social purposes.
Respondents differed in their intentions to use private as compared with open networks. All respondents were significantly more likely to use a private network for work purposes compared to using an open network for work purposes (
To investigate which factors had an independently predictive value for the outcome “I would use a private network for work and training purposes,” a multivariate generalized linear regression model was developed using private work as the dependent variable. To inform this model, multiple correlations and
In the initial model, age was not independently predictive, whereas training level was predictive. Given that training level is related to age, the subcategories of training status were analyzed in the model.
The final model was significant (
Means and standard deviations for confidence using Internet-based applications and services.
Item and groupa | n | Mean | SD |
|
|
95% CI | ||
|
|
|
|
|
|
LL | UL | |
|
|
|
|
2.05 | .04 | 0.01 | 0.82 | |
|
Registrars | 81 | 3.40 | 1.02 |
|
|
|
|
|
Supervisors | 50 | 2.98 | 1.29 |
|
|
|
|
|
|
|
|
4.21 | <.001 | 0.44 | 1.21 | |
|
Registrars | 81 | 3.22 | 1.07 |
|
|
|
|
|
Supervisors | 50 | 2.60 | 1.11 |
|
|
|
|
|
|
|
|
2.68 | .008 | 0.14 | 0.91 | |
|
Registrars | 81 | 3.12 | 1.02 |
|
|
|
|
|
Supervisors | 50 | 2.60 | 1.20 |
|
|
|
|
|
|
|
|
4.17 | <.001 | 0.46 | 1.30 | |
|
Registrars | 81 | 3.48 | 1.22 |
|
|
|
|
|
Supervisors | 50 | 2.60 | 1.23 |
|
|
|
|
|
|
|
|
3.98 | <.001 | 0.40 | 1.27 | |
|
Registrars | 81 | 3.46 | 1.22 |
|
|
|
|
|
Supervisors | 50 | 2.62 | 1.24 |
|
|
|
|
|
|
|
|
3.60 | <.001 | 0.34 | 1.13 | |
|
Registrars | 81 | 3.69 | 1.01 |
|
|
|
|
|
Supervisors | 50 | 2.96 | 1.26 |
|
|
|
|
|
|
|
|
1.32 | .19 | –0.12 | 0.59 | |
|
Registrars | 81 | 2.56 | 1.04 |
|
|
|
|
|
Supervisors | 50 | 2.32 | 0.91 |
|
|
|
|
a Likert scale: 1=strongly disagree, 2=disagree, 3=neither agree nor disagree, 4=agree, 5=strongly agree.
Responses of registrars and supervisors about the usefulness of social networks.
Item and groupa | n | Mean | SD | |
|
|
|
|
|
|
Registrars | 80 | 3.60 | 1.01 |
|
Supervisors | 49 | 3.43 | 0.82 |
|
|
|
|
|
|
Registrars | 80 | 4.11 | 0.83 |
|
Supervisors | 48 | 3.69 | 0.55 |
|
|
|
|
|
|
Registrars | 79 | 3.37 | 1.12 |
|
Supervisors | 49 | 3.61 | 0.76 |
|
|
|
|
|
|
Registrars | 79 | 3.61 | 0.93 |
|
Supervisors | 49 | 3.63 | 0.57 |
|
|
|
|
|
|
Registrars | 80 | 3.61 | 0.95 |
|
Supervisors | 49 | 3.59 | 0.65 |
|
|
|
|
|
|
Registrars | 79 | 3.96 | 0.86 |
|
Supervisors | 49 | 3.78 | 0.65 |
|
|
|
|
|
|
Registrars | 80 | 3.60 | 0.99 |
|
Supervisors | 49 | 3.63 | 0.67 |
|
|
|
|
|
|
Registrars | 80 | 3.60 | 0.99 |
|
Supervisors | 49 | 3.63 | 0.10 |
|
|
|
|
|
|
Registrars | 80 | 3.40 | 1.06 |
|
Supervisors | 49 | 3.63 | 0.71 |
|
|
|||
|
Registrars | 79 | 3.58 | 1.01 |
|
Supervisors | 49 | 3.47 | 0.82 |
|
|
|
|
|
|
Registrars | 80 | 3.81 | 0.94 |
|
Supervisors | 49 | 3.63 | 0.67 |
|
|
|||
|
Registrars | 80 | 3.64 | 0.98 |
|
Supervisors | 49 | 3.65 | 0.72 |
|
|
|
|
|
|
Registrars | 80 | 3.86 | 0.92 |
|
Supervisors | 48 | 3.65 | 0.76 |
|
|
|
|
|
|
Registrars | 24 | 3.13 | 0.68 |
|
Supervisors | 14 | 3.50 | 0.76 |
a Likert scale: 1=strongly disagree, 2=disagree, 3=neither agree nor disagree, 4=agree, 5=strongly disagree.
Perceived difficulties in using online social networks for professional purposes.
Difficulty | Registrars, n (%) |
Supervisors, n (%) |
Worried about privacy | 61 (75.3) | 30 (60.0) |
Insufficient time | 41 (50.6) | 36 (72.0) |
Worried about security | 39 (48.1) | 19 (38.0) |
Not sure how to use them | 22 (27.2) | 20 (40.0) |
Not interested | 12 (14.8) | 17 (34.0) |
Technical Issues | 23 (28.4) | 9 (18.0) |
Lack of other colleagues known to use them | 27 (33.3) | 22 (44.0) |
Other | 4 (4.9) | 4 (8.0) |
Private versus open network usage among registrars and supervisors.
Item and group | Open |
Private |
|
|
|
|
|
|
|
|
All | 2.09 (0.97) | 3.57 (0.93) | <.001 |
|
Registrars | 2.2 (0.99) | 3.85 (0.77) | <.001 |
|
Supervisors | 1.9 (0.90) | 3.16 (0.97) | <.001 |
|
|
|
|
|
|
Registrars | 3.21 (1.30) | 3.19 (1.10) | .85 |
|
Supervisors | 2.40 (1.35) | 2.62 (1.05) | .25 |
Factors correlated with the intention to use a private network for work or training purposes.
Factor | Significance ( |
Training level: supervisor or registrar | <.001 |
Rural versus urban | .42 |
Age | .01 |
Confidence (computer + social) | .03 |
Usefulness | .03 |
Concern about privacy | .11 |
Concern about time | .004 |
Concern about security | .82 |
Not sure how to use | .61 |
Uses Facebook | .24 |
Gender | .07 |
Intention to use a private network for work purposes.
Factor | Beta | SE |
|
|
95% CI | Effect sizea | |
|
|
|
|
|
LL | UL |
|
Privacy | –0.382 | 0.166 | –2.296 | .02 | –0.711 | –0.052 | 0.046 |
Time | 0.561 | 0.149 | 3.765 | <.001 | 0.266 | 0.856 | 0.115 |
Confidence: social and computer | 0.211 | 0.111 | 1.901 | .06 | –0.009 | 0.431 | 0.032 |
Age | 0.008 | 0.010 | 0.763 | .45 | –0.012 | 0.028 | 0.005 |
Usefulness | 0.318 | 0.095 | 3.327 | .001 | 0.128 | 0.507 | 0.092 |
Basic registrar | 1.371 | 0.346 | 3.963 | <.001 | 0.685 | 2.056 | 0.126 |
Advanced registrar | 0.998 | 0.390 | 2.558 | .01 | 0.225 | 1.771 | 0.057 |
Subsequent registrar | 0.884 | 0.346 | 2.550 | .01 | 0.197 | 1.570 | 0.056 |
Supervisor | 0.693 | 0.298 | 2.321 | .02 | 0.101 | 1.284 | 0.047 |
Medical educator | 0a |
|
|
|
|
|
|
a Measured by partial eta squared.
The purpose of this study was to assess whether general practice registrars and supervisors in Australia would use a VCoP in the form of a private online network for training purposes and what factors are important in this decision. The results demonstrate that doctors in this sample have the access and interest needed to use a VCoP. High levels of access to computers and the Internet were coupled with overall high computer confidence. Although computer confidence was high, confidence using social media tools was lower and varied significantly between registrars and supervisors, and between applications. Confidence was also found to be related to training stage and age, but given that training stage and age are related, it was interesting to see in the regression that training stage became significant but age did not. This is in-line with previous findings that age is not a significant predictor of physicians’ use of social media [
Confidence was found to correlate with intention to use an online community, but did not reach significance in the generalized linear regression. This may be because confidence overlaps with training stage and, thus, it is the training stage that is the greatest predictor with confidence of secondary importance. However, confidence may still be worth considering when in the implementation of a virtual community. A study from the United Kingdom showed high levels of interest in social media among British doctors, but low levels of usage, with the authors concluding training as a potential gap [
In spite of good levels of access and confidence, overall use of social media for work purposes was low. This is in contrast to a recent study in the United States that showed a high uptake of social media tools, in particular physician-only communities, with 52% of respondents using online communities, such as Sermo or Ozmosis [
Perceived usefulness is another important predictor of use of an online community in this study. Initially it was thought that respondents’ levels of perceived usefulness and intention to use an online community could be covariate, but this was not the case and usefulness was an independent predictor of intention to use an online community. This is in keeping with findings of 2 studies of use and intention to use social media among health care professionals, and previous studies on technology acceptance [
Finally, barriers are important to address. In this study, time and concerns about privacy were important negative predictors of use, but concerns about security were not significant. This may have been because of a lack of understanding of the difference between privacy and security, or a lack of concern about security, or a higher value being placed on personal or patient confidentiality than computer security. In contrast to these possible concerns, in the Canadian stroke study, participants did not express particular concern regarding patient confidentiality in online exchanges [
Ease of use of a network is another important consideration [
The findings from this study can be looked at in terms of the proposed Health VCoP framework presented in the recent literature review of VCoPs in general practice training [
1. Facilitation
Facilitators promote engagement and maintain community standards
2. Champion and support
The network needs to have an initial stakeholder champion, with stakeholder support
3. Objectives and goals
Clear objectives provide members with responsibilities and motivates them to contribute more actively
4. A broad church
Consider involving different overlapping, but not competing, professional groups, different organizations, and external experts. However, make sure the church is not too broad
5. Supportive environment
Health VCoPs should promote a supportive and positive culture that is both safe for members and encouraging of participation
6. Measurement, benchmarking, and feedback
Health VCoPs should consider measurement as a factor in their design, including benchmarking and feedback
7. Technology and community
Online CoPs should ensure ease of use and access, along with asynchronous communication. Other options including chat and meetings can also be considered, along with the need for training
Communities are more likely to share knowledge when there is a mixture of online and face-to-face meetings, members self-select, and both passive and active users are encouraged
There are a number of limitations in this study. One limitation is that users self-selected to answer a survey on computing and social media by clicking a link in an email to an online survey. The resulting self-selection bias may therefore overreport computer confidence across the whole general practice registrar and supervisor population in the chosen training region. However, it should be noted that the levels of user confidence reported in this study are in keeping with, if not lower, than that found in other recent research [
General practice training can be isolating in Australia. Registrars move from a hospital environment with many colleagues, often in large urban centers, to small practices in urban and rural areas with fewer colleagues. The resulting structural, professional, and social isolation is one of the problems that can lead registrars to consider reducing working hours and moving away from rural work. The Australian general practice workforce is already under pressure, and if isolation can be addressed, this has positive implications for quality of primary care delivery and retention of a rural workforce.
Virtual communities of practice are an effective means of overcoming professional isolation in the business sector and show promise in the health sector. They can overcome isolation by providing a vehicle for knowledge sharing and social interaction. This study shows that general practice registrars and supervisors, in particular registrars, have the access, confidence, and interest to use a VCoP for work and training purposes. The main drivers for use appear to be perceived usefulness and a more junior training stage, with a suggestion that current computer and social media confidence is also beneficial. Barriers to use such networks include time and privacy.
These findings fit with some of the aspects of the Barnett et al [
community of practice
virtual community of practice
The authors would like to thank Coast City Country General Practice Training for funding support for this study. The assistance of Laura Robinson and Lance Barrie is also much appreciated.
Stephen Barnett is the Medical Director and part owner of E-Healthspace, an online community for Australian doctors.