Apps and connected devices are sufficient for chronic condition management
Many virtual care companies rely largely or entirely on remote monitoring devices and apps for tracking health and behavior across users and health conditions.
Hypothesis: Based on our review of clinical literature and initial internal insights, we hypothesized that interactions with coaches and peer groups would achieve better member engagement and outcomes than apps and devices alone.
Data: Across Omada programs,
we found that:
Members who engage with their Omada care team and community in the first week of the program are ~94% more likely to achieve their target health outcomes in the Omada program.1
Members are more likely to churn—specifically, to stop logging into Omada after the first month if they engage only with connected devices.2
Members who message their coach or specialist experience two times more weight loss.3
Omada insight: Apps and devices are not enough. Trusted relationships with healthcare coaches and professionals, especially at the outset of a virtual care program, are crucial for purposeful engagement and sustained outcomes across conditions.
Learning loopback: As a result of this insight, we strengthened resources for our care teams and propietary care delivery platform, and we offloaded device troubleshooting to our support team so that coaches and specialists could focus on building relationships with members.
Education is sufficient for behavior change
Many virtual care solutions depend on self-serve educational content to motivate behavior change among people with chronic conditions
Hypothesis: We hypothesized that information is necessary but not sufficient for behavior change and health improvement, and that those with chronic conditions need help finding the intrinsic motivation to put information into action.
Data: Analysis of our prediabetes program found that the following factors, ranked in order of importance, predicted weight loss outcomes at four months:4
engagement with coaches
engagement with peers
in-app meal tracking
remote monitoring via connected devices
interaction with health lessons
Omada insight: Among Omada members with prediabetes, health education content was the least likely intervention to predict weight loss outcomes. Again, coach and community engagement had the greatest relative effects.
Learning loopback: In response to this insight, we removed lesson completion from our milestone-based pricing model in favor of coach and community interaction, along with other, more diverse interactions that align incentives with purposeful engagement and longterm health outcomes. Value-based care arrangements have always been at our core, and we apply insights and innovations beyond our products to our pricing models as well.
Reactive health coaching is sufficient for sustained behavior change and outcomes
Many virtual care providers utilize health coaching only for acute needs—for instance, a one-time conversation during a low blood sugar event. Coaches are assigned based on availability, not on prior relationship with or knowledge of the member who needs support.
Hypothesis: We hypothesized that reactive health coaching leads to inferior engagement and outcomes as compared to proactive health coaching and ongoing rapport.
Data:We conducted prospective studies on protocols for giving food feedback to members after they track their meals in the app. We found that timely, proactive feedback from Omada coaches led to a 10–15% increase in meal tracking retention, which directly caused 0.5% more weight loss in four months, compared to a control group that received reactive coaching.5 Proactive feedback included personalized guidance on possible meal substitutions, recipe recommendations, callouts if the member included photos, and probing questions if the coach needed clarification on a meal log.
Omada insight:Proactive, dynamic coaching leads to deeper support and engagement, contributing to more sustained behavior change and better outcomes.
Learning loopback:We do more proactive coaching across our entire population, and we use artificial intelligence (AI) to assist our coaches and specialists in starting relevant conversations with members. Omada’s care teams now initiate greater than half of messaging behavior on our care platform.
AI, automated nudges, and gamification are sufficient for generating sustained engagement
Many virtual care companies lean on automated, algorithmic communication with users to encourage behavior change because it’s affordable and scalable.
Hypothesis: We hypothesized that these methods would not be as effective as live care teams at achieving behavior change and improved health outcomes
Data: We ran a series of internal analyses comparing automated nudges and coach led encouragement. Here’s what we found:
Automated nudges resulted in a 30% increase in meal tracking year over year compared to a control group that did not receive the intervention, but they showed no corresponding change in health outcomes, specifically, weight loss at four months.⁶
In another study, members who received coach-led encouragement and feedback had 10% more weigh-ins than members who received the same feedback via automated nudges.⁷
Coaches are 10% better at re-engaging disengaged members than nudges.⁸
Omada insight: While automated nudge tactics do increase short-term engagement, they don’t seem to improve short- or long term health outcomes for people with chronic conditions. Instead, members who interact with live care teams are more likely to purposefully engage with providers and experience sustained behavior change and health improvements.
Learning loopback:Based on this insight, our Product Design team divested from primarily automated nudges for retention and doubled down on further improvements in relationship driven interactions.
Care delivery platform:We created a care platform product team whose primary goal is to supercharge the provider-to-member relationship. This team has transformed our care platform from the basic inbox it once was to the member-centric interface it is today. It surfaces unique information and key moments in a member’s journey for our care teams, making it easier to build rapport and
personalize communication and care plans.
Notification control: We gave users more granular control for notifications so that they could decide which nudges they received, rather than leading with nudges in all instances.
Public trust in the U.S. healthcare system is at an all-time low
Over the last four decades, Gallup polling has found that confidence in the medical system has dropped from 80% in 1975 to 36% in 2019.9 Moreover, a 2006 study suggested that distrust in the healthcare system is strongly associated with self-reported fair or poor health.10 Despite low trust, Omada’s competitors in the virtual care space rely on automated nudges and reactive, one-off outreach from clinicians and coaches with whom the member has little or no prior relationship.
Hypothesis: We hypothesized that people with chronic health issues trust healthcare providers with whom they have established rapport and relationships, and that this trust can lead to better health outcomes.
Data: In a retrospective analysis of members in our diabetes, prediabetes, and hypertension programs, the Omada Insights Lab found that rapport between members and coaches correlates strongly with target health outcomes.11 Omada members send a message of gratitude to their care team approximately once every two minutes.
Omada insight: Relationships between patients and individual healthcare providers can counteract lack of trust in the healthcare system as a whole. Long-term relationships with care providers establish trust, facilitate behavior change, and improve health outcomes. This is consistent with third-party clinical studies showing that patients who had higher trust in their individual healthcare provider reported healthier behaviors, fewer symptoms, higher quality of life, and greater satisfaction with their treatments.12
Learning loopback: We invest continually in relationship-driven care. We train coaches and specialists to build rapport with members so that they have a solid foundation of trust on which to facilitate the lifestyle improvement necessary to manage chronic conditions like diabetes. We also measure rapport quantitatively to ensure that we’re oriented toward building high-trust relationships within Omada’s care delivery platform
We can use the same engagement strategies and care plans for people with different chronic conditions
Many virtual care companies apply the same or similar methodology to engage people with different chronic conditions and use one-size-fits-all plans to manage their care.
Hypothesis: We hypothesized that people with different conditions have different underlying motivations, needs, and ways they manage their health, so how we engage members and create care plans should be tailored to each condition to maximize health outcomes.
Data: In a correlational study, we found that members in Omada for Prevention who interacted with devices, completed lessons, and engaged in any tracking or social activity in at least nine out of 16 weeks averaged two times more weight loss than those who didn’t. By contrast, among members in Omada for Diabetes, these measures of engagement were not correlated with their target health outcome, hemoglobin A1c reduction.13
Omada insight:Engagement strategies that work for one condition won’t necessarily work for another—even when health issues are ostensibly similar. In order to get the best outcomes for people with different chronic conditions, we need to give members guidance, support, and care plans that are personalized to their unique health circumstances.
Learning loopback:As a result of this insight, we updated and customized our engagement metrics and care plans for the Omada for Diabetes program to more effectively drive their particular target health outcomes.
Engagement: We leaned into tracking data from connected blood glucose monitors and continuous glucose monitors, as well as condition-specific surveys to assess medication adherence and diabetes distress.
Care plans: We added more SMART goals14 that a member can set related to diabetes-specific challenges such as remote monitoring and medication tracking. We also enhanced data sharing with members’ primary care providers due to the importance of care coordination.
Programs that treat a single chronic condition can manage that condition most effectively
Conditions like diabetes, back pain, and depression are all high drivers of cost. Many employers and health plans partner with single-point solutions that address only one major chronic condition, or they cobble together several apps or services to address each one individually.
Hypothesis: We hypothesized that single-point solutions miss or ignore comorbidities related to the chronic health condition they aim to manage.
Data: Based on surveys of our population through October 2020, the Omada Insights Lab found that 70–80% of participants in Omada for Diabetes and Omada for Hypertension have multiple conditions. Among members with diabetes in particular, 85% have elevated body mass index (BMI), 47% have hypertension, and 21% have mild depression symptoms.
Omada insight:Assessing comorbidities and integrating care across related health conditions is crucial in the effective treatment of chronic disease, especially where chronic issues have similar underlying behavioral causes.
Learning loopback: We started with Omada for Prevention, but this insight prompted our expansion over the last five years to include treatment across four more chronic conditions—diabetes, musculoskeletal health, hypertension, and behavioral health—to allow us to synergize care across conditions to achieve better health outcomes.
de Groot M, Kushnick M, Doyle T, Merrill J, McGlynn M, Shubrook J, Schwartz F. “Depression Among Adults With Diabetes: Prevalence, Impact, and Treatment Options,” Diabetes Spectrum, Jan 2010; 23(1): 15-18. https://doi.org/10.2337/ diaspect.23.1.15.
Kaka B, Maharaj SS, Fatoye F. “Prevalence of musculoskeletal disorders in patients with diabetes mellitus: A systematic review and meta-analysis,” J Back Musculoskelet Rehabil, 2019;32(2):223-235. https://doi.org/10.3233/BMR-171086.
Sweet CC, et al. “Cost Savings and Reduced Health Care Utilization Associated with Participation in a Digital Diabetes Prevention Program in an Adult Workforce Population,” Journal of Health Economics and Outcomes Research, vol. 7,2 139-147, August 18, 2020. https://doi.org/10.36469/jheor.2020.14529.
Wilson-Anumudu F, Quan R, Castro Sweet C, Cerrada C, Juusola J, Turken M, Bradner Jasik C. “Early Insights From a Digitally Enhanced Diabetes SelfManagement Education and Support Program: Single-Arm Nonrandomized Trial,” JMIR Diabetes 2021;6(1):e25295. https://doi.org/10.2196/25295.
Learning loopback (n.):
The process of sharing a new insight across our interdisciplinary team of experts in order to apply it to future product design and care delivery; each insight leads to new hypotheses, which leads to further assessment and more insights, in a virtuous circle.
Virtual-first care (n.):
First-line medical care accessed through digital interactions where appropriate, guided by a clinician, and integrated into the wider healthcare ecosystem. Adapted from the Digital Medicine Society (DiMe), https://impact.dimesociety.org/v1c/.
Omada internal analysis, member population data 8/2020 - 3/2021, on our diabetes, prediabetes, and hypertension programs. Other Omada programs are still under evaluation.
Omada internal analysis, member population data 4/2018–4/2020. Members who engaged only with devices did not engage with coaches, meal tracking, or health lessons.
Omada internal analysis, member population data 1/2017–1/2020.
Omada internal analysis, member population data 1/2016-1/2020. Based on a feature importance analysis, a machine learning technique that outputs relative scores to explain which variables contribute most to the prediction of a target outcome.
Omada internal analysis, member population data 1/2017–1/2019, on our prediabetes program. Reactive coaching was defined as coaches primarily responding to member-initiated messages or providing no response to members. Proactive coaching was defined by contacting members proactively based on behavioral cues.
Omada internal analysis, member population data 1/2017–1/2020. Nudges for meal tracking included push notification reminders, group post reminders, and in-app modals to celebrate streaks and milestones, Nudges for weigh-ins included push notification reminders, badges for weighing in consistently, and push notifications to celebrate when a member weighed in.
Omada internal analysis, member population data 6/2017–3/2019. Nudges for meal tracking included push notification reminders, group post reminders, and in-app modals to celebrate streaks and milestones, Nudges for weigh-ins included push notification reminders, badges for weighing in consistently, and push notifications to celebrate when a member weighed in.
Omada internal analysis, member population data 6/2017–3/2019.
Gallup poll, “Confidence in Institutions,” accessed online June 2021.
Armstrong K et al. J Gen Intern Med, 2006.
Omada internal analysis, member population data 3/2019-3/2020.
Birkhäuer J et al. PLoS One, 2017.
Omada internal analysis, member population data from 2019 - 2020. Because this is a correlational study, it suggests a relationship but not necessarily direct cause and effect.
SMART goals are specific, measurable, achievable, relevant, and time-bound. Some examples of diabetes SMART goals include checking fasting blood glucose every morning or taking medications along every morning when leaving the house