Investigation of Long Term User Engagement in Disease Specific Social Networks

Aim:

To investigate patterns of user engagement associated with online, disease specific social networks. Specific areas of investigation include:

1. How long does a user stay engaged?
2. How does membership grow over time?
3. Are engagement patterns similar for earlier/later registrants?

Background:

Numerous commentators have predicted that the next big healthcare developments will come not from pharmaceuticals or medical devices, but from disease specific social networks that allow patients to track key health metrics, share insights and generate wisdom from crowds.

Since longevity of user engagement can be used as a proxy for user benefit (i.e. patients only engage if they are deriving a commensurate reward for their efforts), this study has been designed to gather key statistics from a well recognized disease specific social network in order to better assess the validity of these predictions.

Methods:

A random sample of patients was selected from members of the Multiple Sclerosis (MS) community at “Patients Like Me” (PLM), a highly awarded social network dedicated to supporting patients with life changing diseases. PLM’s MS community was first launched in 2007, and was chosen due to its length of existence and its ability to represent user dynamics in a more mature community.

151 patients were randomly selected from 2265 PLM MS community members with a publicly available profile. Sampling was conducted by sorting users in order of most recently registered to least recently registered. A randomly generated seed between 1 and 15 was used to sample the first patient from this sorted list, with every 15th patient thereafter being sampled systematically. This sampling methodology was undertaken to ensure users were consistently sampled across the lifespan of the community. The date of last login and the date of last update for each sampled user were identified from their publicly available profile pages.

Patients that registered with PLM more than 2 years ago with a last login date more than 1, 3, 6, 12 and 24 months later than their registration date were flagged as being active at 1, 3, 6, 12 and 24 months respectively. Patients that registered more than 12 months ago were analyzed in a similar way for activity at 1, 3, 6 and 12 months. The same process was repeated for patients registered with PLM more than 6, 3 and 1 months.

Results:

Table 1: New Registrants PMS MS / Qtr

	# in Sample	Est. Growth
Q4 09	8	900
Q3 09	13	1463
Q2 09	12	1350
Q1 09	15	1688
Q4 08	24	2700
Q3 08	19	2138
Q2 08	12	1350
Q1 08	18	2025

Table 2: Engagement Activity / All Users

Months	1	3	6	12	24
Users Count	145	142	136	108	30
Engaged Users	89	69	55	39	8
Disengaged	39%	51%	60%	64%	73%

Table 3: Engagement Activity / User Registered in Past 12m

Months	1	3	6
User Count	37	34	28
Engaged Users	14	9	5
Disengaged	62%	74%	82%

Discussion:

Recognizing that (1) using last login date may over-estimate length of user engagement since login does not always mean contribution of new data or interaction with other users, and (2) users with public profiles may not be representative of the overall PLM MS population, these results nonetheless provide a number of potential insights into how users engage with disease specific social networks:

a) Disease based social networks may struggle to obtain viral growth dynamics

After more than two years in operation, Table 1 suggests that PLM’s MS community appears to be – at best – maintaining a linear state of user growth and, at worst, experiencing growth rate deceleration. These statistics present in stark contrast with published statistics for social networks like Facebook and Twitter that have accelerated growth dramatically after their first year of operation. These figures suggest that either (a) users are not sufficiently connected to enough eligible non-users to mobilize them for participation in PLM MS in significant numbers, or (b) users do not derive sufficient benefit / engage sufficiently with the social network to warrant their mobilizing friends to join.

This implies that the marginal cost of patient recruitment for disease specific social networks may not diminish over time as the user base grows. This increases the pressure on networks like PLM to  generate increased revenue per capita as the membership grows, rather than hoping to achieve profitability through diminishing patient acquisition costs.

b) Most users do not engage with disease specific social networks for long periods of time.

As highlighted in Table 2, although users who join PLM MS are assumed to be more engaged in personal health management than those who do not, more than one third of users still disengage from the network within the first month of joining. By 3-months, 50% of users have disengaged. By 6-months, 60% have disengaged – although attrition thereafter seems to get slower with 27% of patients still logging into the network 2 years after registering.

These figures appear to indicate that patients engage early with the PML MS community, or not at all. Those users who disengage early may have tried to fill a short term information need, but been unable to derive sufficient ongoing benefit to sustain their engagement. Further profiling of this “hyper engaged” subset may help social networks like PLM maximize the commercial value of their user base and/or direct patient recruitment efforts towards the acquisition of more engaged users. These figures also sound a note of caution, however, for organizations seeking to leverage social networks to communicate with more patients since more than two-thirds of published user numbers may no longer be active in the community.

c) Later adopters are less likely to develop sustained engagement

Interestingly, of patients that registered with PML MS in the most recent 12-months, 62% of users had disengaged within 1 months, and 82% of users had disengaged by 6 months – significantly higher disengagement rates than 39% and 60% respectively for the overall PLM MS population.

These statistics suggest the disease specific social networks like PLM may be at risk of generating lower user engagement rates over time, which – when coupled with static or declining growth rates – may cause challenges in maintaining engaged user numbers over time. Possible explanations for this engagement behavior may be due (a) changes to the way PLM has approached recruitment of patients over time, (b) personal involvement of early members in the development of the PLM MS community – resulting in greater initial commitment to the service, or (c) the possibility that early members have, through their online behavior, created a culture that may not be attractive or engaging for all users.

CONCLUSION

Despite high expectations for the growth of patient-centric disease specific social networks, indicative metrics derived from PLM’s MS community present three significant challenges to their growth and cast doubt over the level of benefit experienced by most users. These challenges include (a) lack of visible viral growth within existing communities, (b) early disengagement by most community members, and (c) diminished patterns of engagement amongst newer community members. Whereas there is no doubt that some users can derive significant benefit and support from a disease-specific online community, these challenges will almost certainly require substantial investigation and intervention before disease specific social networks can take on a more significant role in the clinical landscape.

Advertisement