A single-arm pretest-posttest intervention group was used to test various outcomes (see Measures section). Due to our community partner expressing ethical concerns about including a control group without an intervention, it was not possible to establish an untreated control group. Therefore, a pretest-posttest design was chosen. This was also recommended by our community partner—the Fire and Rescue Department (ie, where this study was conducted). A total of 40 people enrolled in this study.

The home of the community paramedicine program, which was where this study was conducted, receives about 1000 hypoglycemia-related EMS calls on an annual basis. All patients who called 911 due to hypoglycemia-related events were screened based on the following inclusion criteria: (1) residents in the service area of the fire district, (2) individuals aged ≥18 years, (3) individuals receiving intravenous 50% dextrose (intravenous treatment for hypoglycemia provided by EMS personnel), and (4) individuals who are not enrolled in any diabetes-related educational programs. All successfully screened participants were provided with an informed consent form. Participants were only enrolled in this study after they provided consent. Any resident who met the first 3 inclusion criteria based on EMS records (prescreening) was contacted via telephone by EMS personnel to assess their interest in participating in this study. If they were interested and were successfully screened for the last inclusion criterion, the resident was scheduled for an in-person consenting and baseline data collection session. This was done sequentially until 40 participants were enrolled in this study.

Unlike paramedics, emergency medical technicians in fire departments are not allowed to administer glucagon in prehospital settings [28]. They usually respond to hypoglycemic events by using intravenous 50% dextrose. Therefore, we did not include glucagon administration as part of the inclusion criteria.

The intervention included an initial in-person visit (week 1), which was followed by 3 weekly telecoaching calls (weeks 2-4), 1 biweekly call (week 6), and a final in-person visit in week 8 to perform posttest data collection (Figure 1). Anecdotal evidence from our community partner and related literature [29] indicated that having a lack of blood sugar and other vital sign monitoring devices and not visiting primary care physicians (to reevaluate prescriptions) often contributed to hypoglycemia episodes. To encounter these scenarios, health and wellness kits that included blood glucose meters, a blood pressure monitor, and oral dextrose gels were provided as an incentive to the participants.

During weeks 1 and 6 of this study, EMS personnel visited the homes of the recruited participants. During the week 1 visit, the EMS personnel used tablet computers to educate the participants. Their education involved the retrieval of tailored multimedia content, which was shown as a part of the verbal education provided on topics related to hypoglycemia. Based on an assessment of participants’ diabetes literacy, numeracy [30], and knowledge [31], appropriately tailored content that matched participants’ needs from the Diabetes Literacy and Numeracy Education Toolkit (DLNET) was provided by the telehealth platform. The DLNET is a comprehensive platform that was designed to facilitate diabetes education for patients with diabetes, especially for those with low health literacy. The DLNET provides 24 interactive modules that consist of diabetes care topics, such as blood glucose monitoring, exercise, and dietary instructions.

Student volunteers and EMS personnel used a telehealth dashboard, which was designed for this study, to coach and monitor participants over the phone during weeks 2, 3, 4, and 6. These calls focused on the active monitoring of hypoglycemia-related symptoms and the reinforcement of any education that participants received during week 1.

The telehealth platform was built by repurposing, refining, and customizing a proven technical infrastructure that is currently being used by multiple projects (Figure 2). The platform used an Apache server (Apache Software Foundation) that runs a CakePHP back end that is connected to an Angular/Ionic framework–based front end. This enabled the telehealth platform to be delivered as both a web application and a hybrid mobile app with very minimal changes.

The telehealth dashboard automatically scheduled all of the recurring coaching calls for times that were convenient to the participants and in line with the intervention protocol. The community paramedicine personnel received alerts when it was time to call a participant and were able to mark the success or failure of completing the calls. When the calls were not successful, the calls could be rescheduled. It was also possible to perform weekly data collection and take notes during calls.

All data pertaining to the exploratory measures were collected during the in-person visits conducted in week 1 and week 8 of this study. The data were directly entered into the tablet computer that was carried by the EMS personnel.

Participant attrition, session attendance, and overall instrument completion were recorded and analyzed by using descriptive statistics. We tested the pre-post exploratory measures by using simple parametric tests (one-tailed Student t test) after testing the normality of the data obtained. All statistical analyses were conducted by using SAS 9.4 (SAS Institute), and statistical significance was set at P<.05.

The findings from this study fell into 2 categories. First, we focused on the feasibility-related aspects of this study. Second, we focused on the exploratory outcomes. The lessons learned and challenges in implementing this study are presented in the Discussion section.

A total of 40 participants enrolled in this study. The mean age of participants was 67.13 years. The average age of males (n=18) was 69.33 years, and the average age of females (n=22) was 65.32 years.

To our knowledge, this is the first study to conduct a community paramedicine and mobile integrated health care intervention [34] with a specific focus on hypoglycemia by using a telehealth platform. Working with a community partner enabled us to establish a smooth enrollment process for the intervention. However, the low retention rate was an important problem during intervention delivery and data collection. These should be explored in the context of this study to better understand and interpret the intervention outcomes.

Residents of the area that was served by this project belong to the lowest quartile of health literacy scores [16]. Although we included 40 participants, relatively high dropout rates and low attendance rates were observed due to the participants’ low economic and educational statuses. As noted by our community health partner, most residents do not have a permanent phone number and use pay-as-you-go phones. Although participants were available for the baseline visit in week 1, they were not reachable for the final visit in week 8 for scheduling an in-person appointment. Tracking the participants via telephone was, by itself, challenging, and the lack of a permanent phone number made the process even more complicated. The results of this study show early promise, despite the challenging environment of the intervention. However, repeating this study in other socioeconomic settings or neighborhoods is needed to evaluate the intervention for its fullest potential. Additionally, other strategies for improving patient outreach and retention are needed to test this type of intervention.

This study helped us learn about several other aspects about the involvement of EMS personnel in community paramedicine programs. During our intervention, we experienced some challenges with completing the in-person visits. Due to safety concerns, EMS personnel carried out the visits at their convenience. Relying on EMS personnel hindered the collection of the data and the completion of this study. Additionally, because of the nature of this pilot study, we were unable to have exclusive staff join the EMS personnel during the in-person visits. Moreover, the EMS staff in this study changed several times during the intervention due to organizational reasons. Future studies should consider including exclusive staff to ensure protocol fidelity. Another solution is employing the temporarily injured employees of fire departments. Employees who are unable to actively return to fieldwork are often available in fire stations and are best suited for telehealth calls. Future research should consider using protocols for including injured EMS personnel who could actively participate as health coaches in studies.

Another valuable insight we learned from this intervention was about the impact that EMS personnel’s attire had on the participants’ confidence and trust. We observed that wearing a professional uniform favorably influenced participants’ trust and confidence in the paramedics. Participants also expressed the most confidence when the same EMS personnel attended to their emergency medical needs. Future studies that try to evaluate community paramedicine or telehealth need to ensure that such factors of trust are considered in the intervention design.

We designed a telehealth dashboard that the EMS personnel in this study could use to coach and monitor participants with hypoglycemia over the phone. At the end of the 8-week intervention period, nonsignificant improvements were found across various knowledge domains and subdomains (fear: P=.74; avoidance: P=.60; interference: P=.88) of the FH. The intervention resulted in significant improvements in participants’ self-efficacy in hypoglycemia self-management and improved their knowledge of the complications of diabetes, which was measured by using the SKILLD scale. The lack of significant FH-related results might have been due to the very low completion rate (retention), as only 13 out of the 40 participants had posttest data. To interpret the results of this study in the right context, the challenges faced and the lessons learned during this study must be considered.

Although hypoglycemia can usually be safely and cost-effectively treated by paramedics, EMS protocols have been developed independently. This has led to variations in protocol content and formats, which can result in varying standards of care. However, the clinical practices of paramedics and emergency care protocols should be evidence-based and reflect common standards of care, formats, and content [35]. A standardized protocol can be easily accessed through a telehealth platform and can be used to provide guidance and education to patients. Therefore, by using a platform with evidence-based content and a standardized protocol, our study established an example of a telehealth-supported community service that can equally benefit people in need.

Although we wanted to conduct a randomized feasibility study, the establishment of an untreated control group was not accepted by our community partner due to ethical concerns. Therefore, we chose a pre-post study design. Future studies should consider using designs that involve either simple randomization (by individuals) trials or cluster randomization (by fire districts) trials to understand our intervention’s broader impacts. Although this pilot study was limited to a sample size of 40, future studies should also consider having larger sample sizes. Given the high attrition and the challenging socioeconomic settings in which this study was conducted, the findings of this study cannot be generalized. Similar studies should be conducted across areas with different socioeconomic populations. Finally, we offered health kits that were comprised of blood glucose test kits, blood pressure cuffs, and dextrose gels. However, it should be noted that these kits could have had a confounding effect on the outcomes of this study, as there is evidence suggesting that the mere presence of self-monitoring equipment can influence diabetes-related outcomes [29].

Our study shows early promising results for a community-based hypoglycemia prevention intervention. However, our pilot study has several limitations. We comprehensively presented the challenges we faced and the lessons we learned throughout this study, and these should be considered when designing future studies.