This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Mental Health, is properly cited. The complete bibliographic information, a link to the original publication on https://mental.jmir.org/, as well as this copyright and license information must be included.
The COVID-19 pandemic has placed strains on communities. During this public health crisis, health systems have created remote methods of monitoring symptom progression and delivering care virtually.
Using an SMS text message-based system, we sought to build and test a remote model to explore community needs, connect individuals to curated resources, and facilitate community health worker intervention when needed during the pandemic. The primary aims of this pilot study were to establish the feasibility (ie, engagement with the text line) and acceptability (ie, participant ratings of resources and service) of delivering automated well-being resources via smartphone technology.
Eligible patients (aged 18 years or older, having a cell phone with SMS text messaging capability, and recently visited the emergency department) were identified using the electronic health record. The patients were consented to enroll and begin receiving COVID-19–related information and links to community resources. We collected open-ended and close-ended resource and mood ratings. We calculated the frequencies and conducted a thematic review of the open-ended responses.
In 7 weeks, 356 participants were enrolled; 13,917 messages were exchanged including 333 resource ratings (mean 4) and 673 well-being scores (mean 6.8). We received and coded 386 open-ended responses, most of which elaborated upon their self-reported mood score (29%). Overall, 77% (n=274) of our participants rated the platform as a service they would highly recommend to a family member or friend.
This approach is designed to broaden the reach of health systems, tailor to community needs in real time, and connect at-risk individuals with robust community health support.
The COVID-19 pandemic continues to disrupt the health and well-being of communities. An immediate call to address clinical issues has been followed by a call to public health and community action [
Health care institutions are woven into the fabric of communities and serve as a critical hub of information, resources, and care. These systems will continue to play an integral part in supporting communities throughout the prolonged response to, and recovery from, the pandemic [
An emerging and critical health care professional is the community health worker (CHW). The American Public Health Association defines CHWs as “frontline public health worker who is a trusted member of and/or has an unusually close understanding of the community served” [
This pilot study’s purpose was two-fold. First, to identify patients with an initial health system encounter (emergency department [ED] discharge) early in the pandemic (April-June 2020) and engage them via SMS text messaging to understand collective community and well-being needs. Second, to curate a community-driven repository of community resources and connect individuals experiencing emotional distress or those who express other social support needs to CHWs for urgent nonclinical needs.
We hypothesize that we will be able to connect high-risk individuals to CHWs for urgent social support and nonclinical needs. Second, we will be able to collect community-identified resources and better understand collective community and well-being needs.
Our team deployed a text message-based script, and using an automated text messaging platform, the patients were consented to enroll and begin receiving COVID-19–related information and links to community resources (eg, social services, resources for families with children, mental health and well-being, and virtual, hyperlocal cultural events). The text messaging script sought to engage participants and elicit feedback on all resource content in order to prioritize the content driven by community members [
The participants were identified using the electronic health record (EHR) and sent an initial message if they met the following inclusion criteria: adult (>18 years of age); mobile number in EHR; and recent ED discharge. Enrollment occurred between April through May 2020. We defined this study sample as high-risk due to their ED use during the COVID-19 pandemic in Spring 2020. The participants were recruited through a Health Insurance Portability and Accountability Act (HIPAA)-compliant technology platform (Mosio), to invite eligible participants and engage in 2-way text messaging. The participants received a text message about the pilot and were asked if they wanted to opt in with a simple “yes” reply. Potential participants who did not respond or replied “stop” were no longer contacted. The entire process of approaching, enrolling, and engaging participants was completed through text messaging.
The participants were asked to provide quantitative ratings on resource links, mood ratings, and open-ended narrative feedback. Participants were asked to provide quantitative ratings (1-5 scale, 5=very helpful) on resource links, mood ratings, and open-ended narrative feedback weekly. Participant feedback and ratings of the resources guided the content of the mobile web-based resource hub. This allowed the content to be curated and defined by community members. This created a mechanism to continuously tailor content and create a living, community-driven resource hub.
The participants self-reported well-being on a weekly basis on a numeric scale (0-10, with 0 representing the worst mood and 10 as the best mood) [
Penn Med with You participant flow.
Automated text messaging engagement.
Tailored messaging schematic for well-being.
Community-tailored resource hub.
This study was approved by the University of Pennsylvania Quality Improvement Institutional Review Board. A protocol for identifying risk of emotional and psychological harm was established. We built automated notifications to hover over self-reported well-being scores and free text input. We built automated responses for any messaging that the system could not interpret as an answer and keyword phrases that triggered immediate human intervention (eg, “death,” “kill,” “suicide,” “hang”). This automated messaging directed the participants to seek medical care if needed and directed them to a website for a comprehensive list of other resources. We ultimately established a human mechanism in place to identify the participant, connect with them via telephone, and route them to the appropriate professional.
Our team is multidisciplinary, spanning from clinician and nonclinical members. It includes Penn Medicine Center for Digital Health, Center for Healthcare Innovation, department chair and faculty from Emergency Medicine, medical students from the Perelman School of Medicine, data analysts, graphic design students from the University of Pennsylvania, the text messaging platform, and the Individualized Management for Patient-Centered Targets team. The latter is a nationally recognized community health worker program that hires and trains trusted neighborhood residents to become CHWs to carry out culturally appropriate outreach activities, social support, patient advocacy, and health system navigation, with the goal of improving health in underserved populations [
Descriptive statistics were used to summarize demographic variables. Authors LS and RO categorized the open-ended responses; they reviewed 50 responses to identify common themes and applied a thematic analysis approach [
The 409 individuals who opted into the “With You” program were asked to self-report their well-being. A total of 286 (70%) individuals responded with a well-being rating. The majority of participants were Black (275/409, 73.5%) and female (264/409, 70.1%) (
Participant demographics.
Characteristics | Valuesa | ||
Age (years), mean (SD) | 46.9 (16.8) | ||
Female, n (%) | 264 (70.1) | ||
|
|
||
|
Black | 275 (73.5) | |
|
White | 69 (18) | |
|
Asian | 14 (3) | |
|
Hispanic Latino | 14 (3) | |
|
Other | 16 (4) | |
|
|
||
|
Hospital A | 225 (60.2) | |
|
Hospital B | 149 (39.8) |
aOf the total study sample (n=383), demographic data were missing for 9 (2.3%) participants.
Over the course of 7 weeks, we exchanged 13,917 text messages, elicited 673 self-reported well-being ratings (mean 6.8), and received 333 resource ratings (mean 4) (
Initial community engagement and mood ratings.
We received 368 open-ended participant responses (
I am in a good space. Praying for this attack on America to end so that we can get back to a normal existence. On a scale from 1-10 my mood is a 10.
or
My mood is good I say it about 9 thank you for being concerned.
Additionally, after describing one’s mood, most participants (125/368, 34%) would include pleasantries such as the following:
thank you for looking out.
or
I appreciate the text messages, keep them coming!
Qualitative codes demonstrating themes.
Codes | Operational definition | Frequency, n (%) |
Administrative | Question or concern related to text line logistics | 55 (15) |
Health | Physical and mental health-related question | 53 (14) |
Mood | Elaborates on mood score response | 135 (36.6) |
Pleasantries | Responds to automated message | 125 (34) |
Overall, 77% (164/213) of our participants rated the platform as a service they would highly recommend to a family member or friend. Moreover, 67% (143/213) of unstructured feedback on the platform and content were positive. Through open-ended questions, the participants made the following remarks:
The line is good…Enjoying the information you’re directing to me.
Keep the texting line active. Folks like myself appreciate a caring text and health information/resources. Thank you so much. Be safe….
Thanks, you for crisis info your help is right on time at times I just got to talk to someone thank you once again.
The participants also provided constructive feedback and recommendations through the open-ended questions. This feedback helped guide our platform development. Some examples of unstructured and open feedback are as follows:
Live counselors would be awesome.
More resources for anxiety.
Text at different times of the day to check in.
As we enrolled more participants, we obtained additional feedback on resources and the platform itself as a means to institute continuous and real time quality improvement. For example, a participant requested information on better sleep in the qualitative resource feedback. The following week, we featured recommendations for healthy sleeping habits during COVID-19. The participants also reported feelings of isolation and disconnection; in response, we included content featuring how several neighborhoods were banding together to create outdoor socially distanced events for children.
After the pilot study concluded, the research team met twice virtually for approximately 30-45 minutes to elucidate feasibility and sustainability factors. Below are key factors for feasibility and sustainability:
Form an interdisciplinary team to frame out the engagement strategy and overarching method of connecting at-risk groups to urgent medical and social needs.
Design an approach that learns early and often to inform and adapt growth to needs.
Create a central, easy-to-access hub for social resources that currently exist. This includes content for those with low and high self-reported well-being.
Build a team and schedule to allow for intermittent but daily monitoring of responses for quality and safety assessment.
Recognize that content and resources will need to be tailored to the local community and environments and be open to change.
Partner with local champions and resources so as to not reinvent the wheel and lean on the expertise of existing networks.
This pilot study aimed to determine the acceptability and feasibility of a health system-driven community engagement for health and wellness line. This pilot study engaged discharged ED patients with a visit during the early phase of the pandemic via SMS text messaging to understand their well-being needs, curated a participant-driven repository of community resources, and proactively connected individuals with CHWs to support their well-being and connection to needed resources. In other health systems, CHWs were essential in addressing social determinants of health in vulnerable populations [
The text line also filled an important gap in mental health access. Earlier research found that rates of anxiety and depression are on the rise during the pandemic, and resources are needed [
This pilot study demonstrated that digital engagement through simple options such as SMS text messaging provides a means to engage and interact with communities through COVID-19 and other crises. Automated hovering [
This pilot study has limitations. First, it was limited by its enrollment and retention. Overall, 20% of the participants opted into the text messaging line, and 70% were retained. Despite the high rates for a pilot study, it is possible that this contributes to a selection bias. Second, the pilot study only assessed mood and resource ratings from participants who opted in; it did not have comparison or control groups. Our findings only represent community members who are engaged in care at 2 hospitals; these results do not represent those who receive care at other hospitals and most importantly, those who are not engaged in health care at all, which comprises some of the highest need populations. Third, we could not assess how many participants used the resources over the course of the pilot study. This is because the website was publicly available, and nonparticipants might have accessed it. Fourth, we defined our study sample as high-risk due to their ED use during the COVID-19 pandemic. We posit that there are other key variables such as chief complaint and housing stability that would be helpful in further classification. Lastly, we assessed feasibility of a health system implementing a community line; further research can delve into the acceptability of community perspective of the service. Our evaluation methods elicited feedback; however, future research can use frameworks such as the RE-AIM (reach, effectiveness, adoption, implementation, and maintenance) [
Our pilot study can be compared with prior work, such as the Thought Spot, a randomized controlled trial aimed at optimizing and evaluating a web-based and mobile-based platform designed to improve the ability of students to access mental health and substance use services [
This study suggests that health systems are well positioned to support community well-being. It is feasible and acceptable to proactively text health system patients and provide robust, wraparound support during a pandemic. The expanding use of digital technology offers an opportunity to engage community members throughout the stages of the COVID-19 pandemic.
community health worker
emergency department
electronic health record
Health Insurance Portability and Accountability Act
net promote score
reach, effectiveness, adoption, implementation, and maintenance
We would like to thank Hannah Jia and Haley McCalpin for their work monitoring the text line and Olenga Anabui, director of the Individualized Management for Patient-Centered Targets team.
None declared.