The Office of the National Coordinator for Health Information Technology Beacon Community program, established under the HITECH act, gives grants to local communities to explore health IT innovations. Three Beacon Communities that have been piloting text-message-based diabetes interventions published a paper in the Journal of Medical Internet Research detailing some challenges and best practices for implementing a mobile health intervention.
Voxiva powered all three interventions. Two of the communities, the Crescent City Community in New Orleans, Louisiana and the Southeast Michigan Community in Detroit, used txt4health, a 14-week program aimed at individuals at risk for diabetes. The other, the Utah Beacon Community in Salt Lake City, used Voxiva’s Care4Life platform, a 26-week program for adults already diagnosed with Type 2 diabetes.
Researchers from all three Beacon sites shared the following six lessons for future texting interventions.
1. “Identify community partners that are willing to engage with and support the program, and leverage their resources and community presence to design the program strategy and reach the target audience.”
Because Utah’s pilot was a randomized clinical trial, all participants had to meet certain requirements that made it hard to go outside of clinical settings to find them. The Michigan and Crescent City groups, however, reached out through local partnerships and various outreach channels, including television, radio, and social media.
“The Crescent City and Southeast Michigan Beacons were challenged in reaching a typically hard-to-reach target audience,” the study authors write. “… They advertised txt4health in public transit, bus shelters, laundromats, barbershops, salons, and other settings frequented by the target audience, which allowed them to optimize outreach while conserving limited resources.”
The Crescent City team also leveraged a large retail pharmacy, as well as churches and university student groups, to get the word out about the intervention.
2. “To the extent possible, design outreach, enrollment, and message content around the needs and perspectives of end users to increase program enrollment and engagement.”
The Utah site opted not to modify the Care4Life platform due to the clinically validated nature of the content. But the other two adapted the content of the messages based on the location of the program.
“For example,” the study authors write,” in Southeast Michigan diabetes is often referred to as ‘sugar’. By incorporating this term into the text messages, the txt4health team hoped participants would perceive the program as more accessible and relatable and would thus engage with it more actively.”
The Michigan team also took into account local safety concerns about Detroit neighborhoods, and designed the message to always suggest an indoor alternative to outdoor events.
3. “Anticipate that traditional marketing tactics may be insufficient to drive enrollment, and plan and budget additional staff time and resources for in-person engagement with the target audience to help drive enrollment.”
Enrollment proved a challenge because many members of the target audience were older people with older phones. Users of feature phones without a full keyboard found sometimes found enrollment via texting too taxing, the studies found. Others did not have unlimited text messaging, and opted out of enrollment for that reason. And though the interventions offered online enrollment, many users also had trouble with that. In a post-study phone survey of members of the Utah pilot, 35 percent reported limited or no access to a computer and 38 percent reported having trouble using a web browser.
To address these concerns, study sites set up onsite training and enrollment stations to get people signed up and show them how to participate in the programs, even teaching them how to use text messaging.
“While surveys indicated that the traditional marketing tactics such as advertisements and brochures increased community awareness of the programs, direct in-person engagement drove enrollment to a much greater extent,” the authors write. However, they added, the addition of in-person programs made the interventions more cost- and labor-intensive then originally planned.
4. “To the extent feasible, bring care providers into the process—even if it means developing work-around solutions—to help them understand and promote mHealth as a tool to enhance patient care.”
All three programs found that working with providers improved both adoption and the credibility of the process, but a number of factors made bringing providers into the process more difficult. One was that appointment times, at 8-10 minutes, were too short for doctors to spend time introducing a text messaging initiative. Another was that, under fee for service payment models, there was often no structure for doctors to get paid for introducing a texting initiative. Some of the programs circumvented this difficulty by introducing the intervention during patient check-in and check-out rather than the doctor visit.
Another barrier was the inability to incorporate data from txt4Health or Care4Life into providers’ electronic health records. The paper suggests that data from the text messaging services could be useful at the point of care, and could further incentivize provider support, but only if EHR integration and interoperability are perfected.
5. “When planning the evaluation strategy, decide at the outset which aspects of the program will be critical to measure and which will not, to determine what stakeholders ‘need to know’ versus what would be ‘nice to know’.”
The Beacon Communities encountered unexpected biases and incomplete data sources when looking to evaluate their program. For instance, the Utah group had hoped to provide data about participants’ HbA1c levels, but they ultimately couldn’t, because so many patients missed their scheduled follow-up appointments. The Crescent City community used random digit dialing to ascertain the effect of the intervention on the overall community, but those survey results aren’t trustworthy because the team only called landlines — to ask about a mobile phone-based intervention, effectively eliminating mobile-only users from their sample.
In general, the authors said, the Beacon Community model presents a challenge to good outcomes data, because other interventions were happening simultaneously in these same communities. “This context makes it difficult to control for external factors and tease out the impact of—or attribute observed outcomes to—the specific mHealth intervention,” they wrote.
6. “Share lessons learned with others to allow them to benefit from your experience.”
“A primary take-away from the Beacon Community experiences with txt4health and Care4Life is that mHealth technology itself is not a ‘silver bullet’,” the authors conclude. “As is increasingly evident in the adoption of many other health IT tools, the full value of mHealth will be realized only when attitudes, behaviors, and health care delivery also change.”