User‐Centered Design to Develop Automated Short Message Service Messaging for Patient Support of Rheumatic Heart Disease Secondary Antibiotic Prophylaxis

Abstract

Background

The Active Community Case Management Platform is a cloud‐based technology developed to facilitate rheumatic heart disease case management by health care providers. This study aimed to design and pilot an automated short message service (SMS) intervention to support secondary prophylaxis adherence.

Methods and Results

We developed a concise library of messages to support secondary antibiotic prophylaxis. The SMS intervention used TextIT, an interface that enables users to send out interactive SMS messages at scale. The message bank was piloted in a cohort of 50 patients with rheumatic heart disease randomized into 2 groups. Group 1 received standard support (nurse‐led/Active Community Case Management Platform). Group 2 received standard support plus SMS intervention for 2 months in the Lira and Gulu districts of Northern Uganda. We collected qualitative data on participants' experiences and assessed treatment adherence. Using a sequential user‐centered process consisting of 4 phases (phases 1–4), we developed a message bank (n=43) deployed during our pilot study. There were no between‐group differences in treatment adherence or acceptance. Interviews of participants indicated that the intervention was viewed positively. A total of 75% of SMS recipients responded to the messages, and 25% called the study staff to acknowledge receipt of text messages.

Conclusions

This study has successfully developed a bank of SMS messages to support secondary antibiotic prophylaxis adherence. We have demonstrated the feasibility and acceptability of SMS technology in rheumatic heart disease care management. Future work will include integrating automated SMS into the Active Community Case Management Platform and a larger study of integrated SMS to reduce health care worker burden for patient support and improve adherence to secondary antibiotic prophylaxis.

Nonstandard Abbreviations and Acronyms

ACT

Active Community Case Management

ADD‐RHD

Active Case Detection and Decentralized Dynamic Registry to Improve the Uptake of Rheumatic Heart Disease Secondary Prevention

RHD

rheumatic heart disease

SAP

secondary antibiotic prophylaxis

SMS

short message service

Clinical Perspective.

What Is New?

• Our findings suggest that short message service messaging interventions can be effectively incorporated into the clinical workflow to provide patient‐facing support and impact secondary antibiotic prophylaxis adherence in Uganda.

• User‐centered methods allowed us to ascertain that the patient's feedback and needs were adequately represented in the intervention.

What Are the Clinical Implications?

• Secondary antibiotic prophylaxis has been shown to prevent progression of valvular heart disease and improve health outcomes in rheumatic heart disease.

• Using short message service messages to provide reminders shows promise in improving secondary antibiotic prophylaxis adherence in low‐income settings.

Rheumatic heart disease (RHD) is a common cardiovascular disorder that affects >40 million children and young adults worldwide. ¹ , ² , ³ Currently, the only intervention proven to prevent the progression of valvular heart disease and reduce cardiovascular complications is secondary antibiotic prophylaxis (SAP), delivered as every 3‐ to 4‐week intramuscular penicillin. ¹ , ² , ³ Studies have shown that 80% adherence is the benchmark for SAP to be effective. ⁴ Meeting and maintaining this protection threshold has been challenging globally. ⁴

The Active Case Detection and Decentralized Dynamic Registry to Improve the Uptake of Rheumatic Heart Disease Secondary Prevention (ADD‐RHD) Center, under the umbrella of the American Heart Association Strategically Focused Health Technologies and Innovation Research Network, aimed to develop and refine a technology‐based approach to increasing SAP uptake at scale in real‐world settings. To support decentralized delivery of SAP, ADD‐RHD developed a cloud‐based application, Active Community Case Management (ACT) Platform. ACT is a provider‐facing application geared toward helping primary care nurses efficiently manage RHD cases, connecting lower‐level providers to specialists, and providing dynamic tools for monitoring patient adherence, essential medication stocks, and quality of care delivery. In the current iteration, primary care nurses can generate a list of patients who are due and overdue for SAP as well as visualize and sort by adherence but must manually call or text patients for SAP reminder support. Patient‐facing supports are needed to both ease health care providers' burden by automating patient reminders and to provide more consistent support to patients to improve SAP adherence.

Mobile telephone use and direct messaging have shown potential to improve self‐efficacy and healthy behaviors across many disease states. ⁵ , ⁶ Less research on the efficacy of text messaging interventions has been done in low‐resource environments, but a recent systematic review, including 12 short message service (SMS) trials from sub‐Saharan Africa, showed an overall improvement in medication adherence for participants in an SMS intervention arm compared with those receiving standard of care. ⁵

Therefore, the decision was made to use a user‐centered design process to develop motivational and informational text messages to increase SAP adherence in Uganda. Here, we present the user‐centered design of the messaging bank and the subsequent feasibility pilot of the 2‐way messaging intervention using RapidPro, ⁷ an open‐source software designed by the United Nations Children's Fund (UNICEF) that allows for at‐scale mobile mass communication in low‐income settings.

Methods

All supporting data and methods used in this study are available on request from Dr Ndate Fall (ndate.fall@cchmc.org).

Design of the Message Bank

We used the theory‐informed user‐centered design framework, which integrates a user‐centered design approach with crowdsourcing and study team expertise. ⁸ The iterative process is broken into phases, including message drafting, feasibility testing, pilot testing, crowdsourcing, expert review, and final message selection (Figure 1).

Figure 1. User‐centered design method.

Study Settings

The study took place between October 2022 and May 2023 in the Gulu and Lira districts in Northern Uganda, where the ACT Platform was being tested. The Uganda Heart Institute and Cincinnati Children's Hospital Ethics Boards and the Ugandan National Council of Science and Technology approved the study. All participants signed written informed consent before the initiation of research activities. The message development stage of the project took place in the Gulu district, and the intervention was subsequently piloted in both communities.

Text Message Bank Design

All messages were drafted in English and Acholi, the primary language of Northern Uganda.

Phase 0: Message Drafting

Grounded in evidence‐informed interventions, ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ a cohort of our US and Uganda research team members (n=11) and selected community stakeholders (3 local primary care providers, 3 village health team members, and 4 patients with RHD from the Uganda RHD registry) were tasked with developing the initial library of messages (n=100) based on the health behavior change constructs from social cognitive theory. ⁹ Three major messaging types were developed, including those that focused on self‐efficacy, emphasized the benefit of SAP, and provided a reminder of an upcoming SAP appointment.

Phase 1: Feasibility Testing

Phase 1 (September 2022) focused on testing patients' reactions to the messages. Patients with RHD (5 male and 5 female patients) from the National RHD Registry between the ages of 15 and 32 years were recruited. A card‐sorting task was used to elicit feedback from the participants. ¹⁴ Participants were asked to sort a subset of 36 randomly selected messages in either English or Acholi, based on their preference of liked, disliked, or neutral, and to explain why they chose that category. The card‐sorting activity was modified to suit the study settings and population. Messages were written at a fifth‐grade education level, and accommodations were made for limited literacy or language barriers by reading cards aloud (15% participants) and encouraging oral feedback after each card was read. On completion of this phase, messages were modified to better incorporate content around SAP and RHD.

Phase 2: Pilot Testing

This phase of the message bank design was completed in November 2022. Ten newly enrolled (≤6 months) patients with RHD (50:50 male/female ratio) from the National RHD Registry were enrolled. A subset of 36 messages randomly selected from phase 1 was used in phase 2. The selected messages were evenly representative of the theoretical framework categorization (self‐efficacy, benefit, and reminder). Furthermore, poorly performing messages from phase 1 were also included to provide variability. The TextIt interface was used to send participants a daily automated 2‐text message SMS. Study staff called daily at a prearranged time to gather qualitative feedback on the messages and the ease of use of the technology.

Phase 3: Crowdsourcing

Phase 3 took place in December 2022. In this phase, 100 participants were recruited from the Uganda National RHD Registry and tasked with theoretical construct characterization of all messages in the draft library (phase 0). Participants were asked to read the messages and determine to which theoretical category they belonged (self‐efficacy, benefit, and reminder) ⁹ based on detailed explanations provided by study staff. The task was set up as a multiple‐choice response with the 3 categories as the possible answers. Congruency (design intent matched participant categorization) scores were calculated for each participant and each message. Accommodations were made for limited literacy by reading the messages and options aloud and eliciting feedback after each question.

Phase 4: Expert Rating

After discarding poorly performing messages from phase 3 and considering feedback from phases 1 and 2, 3 research team members (NF, JP, FO) rated the remaining messages (n=62) independently using a 5‐point Likert scale, ranging from 1 (poor quality) to 5 (high quality). Each member had access to all the messages and data from the previous 3 phases.

Phase 5: Final Selection

A final bank of messages (n=43) was selected on the basis of team ratings and participant feedback.

Feasibility Pilot of Bidirectional Text Messaging

Following message bank development, a short, 2‐month pilot was conducted to understand the feasibility and patient acceptance of delivering SAP support through a 2‐way text messaging platform. A convenience sample of 50 participants was selected, and 10 (5 from each arm) underwent more in‐depth interviews. None of the participants included in phase 0 (message drafting) through phase 3 (crowdsourcing) was approached for participation in the feasibility study.

Automated 2‐Way Messaging

RapidPro ⁷ is an open‐source software designed by UNICEF that allows users to design and scale mobile mass communication without relying on a software developer's services. TextIt interface ran RapidPro ⁷ and allowed us to bypass the complexity of building the system. Flows in RapidPro ⁷ are linked steps that enable users to define interactions. In the example shown (Figure 2), we sent a reminder to the patient that asked for a response. The patient's answer triggered an automated response. The systems allow users to schedule delivery dates and times and personalized messages to specific contacts. This feature will enable us to customize reminder messages and automate them to send on a particular schedule.

Figure 2. Example of a TextIt short message service flow.

Pilot Design (SMS Intervention)

A convenience cohort of patients with RHD (n=50) from the Uganda RHD Registry receiving care through the ACT Platform in the Gulu and Lira districts were recruited for participation in the pilot and randomized to either (arm 1) continue with routine nurse‐led support (SAP calendar, intermittent telephone‐based support) or (arm 2) receive routine nurse‐led support plus enhanced support through the SMS intervention (biweekly motivational messaging and monthly adherence support message). Potential participants were excluded if they were aged <13 years, did not have access to a telephone capable of sending and receiving an SMS, or were not prescribed SAP. Participants completed the Chronic Treatment Acceptance (ACCEPT) questionnaire ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ at enrollment and 2 months after the SMS intervention deployment to assess their acceptance of SAP. ACCEPT is a 25‐item questionnaire to evaluate patients' long‐term treatment acceptance. It measures the following 5 domains, scored 0 to 100 (0, lowest acceptance; and 100, highest acceptance): inconvenience, long‐term treatment, constraints, adverse effects, and effectiveness.

In addition, a subset of participants from both arms of the study completed semistructured interviews at 2 months focused on capturing their experiences around receiving SAP and, for arm 2 participants, their perception of the SMS intervention.

An interview guide with open‐ended questions was used to encourage participants to relate their experiences. Purposive sampling was used to select a cohort of participants from the pilot at 2 months. Recruitment ended when data saturation was reached, and no new information was discovered. Interviews lasted ≈30 minutes, were recorded, translated from Acholi into English, and transcribed. During the verification process, the data were deidentified, final transcripts were stored on an encrypted server, and all recordings were destroyed.

Qualitative Data Collection and Analysis

The goal of the mixed‐methods analysis was to fully describe and compare the participants' perceptions on their prophylaxis support and its effects on sustaining long‐term adherence to RHD prophylaxis. Quantitative data were summarized per standard methods. Verbatim transcriptions of interviews were entered into NVivo, a secure, cloud‐based qualitative analysis program. An iterative approach facilitated by NVivo was used to code the transcripts. Line‐by‐line coding was used to identify a portion of text or meaningful phrases in the transcript concerning the research question and assigned a code word or phrase in the first few transcripts. Members of the research team met to discuss the codes and associated passages; any disagreements were discussed, and once consensus was reached, a codebook was created in NVivo. The remaining transcripts were uploaded to NVivo, and coding was conducted using codes from the initial codebook. Emergent themes were developed through this process.

Results

Phase 0: Message Drafting

The research team and stakeholders within the community developed an initial draft bank consisting of 127 messages framed around reinforcement and self‐efficacy. ⁹ The resulting library exceeded our initial goal of generating a minimum of 100 messages. Messages were designed around 3 categories, including motivation to receive SAP (n=51), benefit of SAP (n=37), and reminders of SAP appointments (n=39). Table 1 provides an exemplar for each subset of messages. Reminder messages were designed to reinforce positive behavior around SAP adherence. In contrast, the benefit and motivation messages were drafted to boost patients' self‐efficacy and ability to maintain their prescribed treatment plan successfully.

Table 1.

Examples of Messages by Theoretical Category

Theoretical category	Messages
Self‐efficacy (motivation)	Together let's end this disease RHD from our community.
	My health is my priority. I will not miss my injections.
	I am in charge of my health; I can't take it lightly. I am going for my injection.
Reminder	Injection date is approaching. Please go for your injection.
	Tomorrow is your injection day, please don't miss.
	Remember to go get your injection. 2 days to go.
Benefits	The injection is responsible for improvement of the health condition.
	I am not feeling sore throat anymore; my injection is helping me.
	I am feeling healthy; my injection is helping me.

RHD indicates rheumatic heart disease.

Phase 1: Feasibility Testing

A representative subset of 36 randomly selected messages (12 motivational, 13 reminder, and 11 benefit) was assessed during phase 1 (Table S1). The card‐sorting activity found that participants preferred the motivational messages crafted around benefits and self‐efficacy concepts to the reminder messages. Overall, 36% of benefits and motivational messages were liked by 80% of the participants compared with 15% of the reminder messages. Of the 36 messages assessed in this phase, 9 (25%) were liked by >80% of the study cohort, whereas 14 (50%) were favored by at least half of the participants. One motivational message was unanimously liked, whereas only 1 message received a dislike rating from 50% of the participants. The unanimously liked message, “I believe one day I will heal, getting my injection is my priority,” focused on the potential benefit of the injection, whereas the disliked message, “The journey still continues, keep it up,” emphasized the long‐term aspect of RHD treatment. The most well‐received benefits messages highlight the advantages of antibiotic prophylaxis, such as improved well‐being, and access to supportive health care providers. The motivational messages crafted to empower patients to take charge of their own health and messages that deepen the community role in addressing RHD received the most positive feedback from the participants. The qualitative feedback indicated that the messages were clear and easy to understand in English and Acholi.

Phase 2: Pilot Testing

Response to research team telephone calls was nearly 100%. One participant did not respond to daily telephone calls the first few days of the pilot, but by day 5, the response rate was on par with the rest of the cohort. All participants reported liking the messages and the technology. No one had any difficulties interacting with the technology on their telephone. Furthermore, the fact that participants did not incur financial charges for using this technology was reported as a positive factor. Participants related positive feelings around getting reminders. One participant nicely summed it up:

It will help a lot because human beings have a lot of things to do; therefore, a person can easily forget. So, I feel that the reminder messages will be helpful in ensuring that a person receives the injection on time.

Despite receiving frequent SMS during the pilot, the overwhelming majority reported positive feelings. It was suggested that ideal timing would be twice a month and preferably a few days to a week before an injection due date. Only 1 participant reported that the frequency of messages and telephone calls during the pilot was tiresome, and the daily follow‐up calls were tedious. Participants most liked the motivational messages. They also reported that although they liked the messages, a telephone call from their health care provider would make them feel cared for and seen.

Phase 3: Crowdsourcing

Following the card‐sorting activity, 37% (±0.11%) of the messages were categorized appropriately into the theoretical category (Table 2). The best‐performing category was reminders, with 74% of the message correctly characterized, whereas the messages crafted around benefits fared the worst, with a 15% correct categorization rate. Messages misclassified at a rate >25% were discarded and not included in the remaining evaluation steps. This resulted in discarding >50% of the benefits and self‐efficacy messages.

Table 2.

Categorization Results

Category	No.	Correct categorization, mean %
Self‐efficacy	51	24
Reminders	39	74
Benefits	37	15

Phase 4: Expert Rating

After discarding poor‐performing messages, 3 team members (NF, JP, FO) independently rated the remaining (n=62) using a 5‐point Likert scale. The raters considered data from phases 1 to 3 during this process. The average rating was 4 (±0.6), and 43 (69%) messages with an average rating >3.5 were selected for the final message bank.

Phase 5: Final Selection

Feedback from all phases was combined to select the best messages from the subset that passed phase 4 (n=43). Sixty‐five messages were discarded following phases 1 and 3, which included 1 disliked message from phase 1 and 64 messages that were misclassified at a rate >25% during phase 3. An additional 19 messages (rating, <3.5) were discarded following phase 4. After the process was completed, the final bank consisted of 43 messages: reminder (n=31), self‐efficacy (n=8), and benefit (n=4).

Two‐Way Messaging Intervention Pilot (SMS Intervention)

Table 3 details the demographics of the pilot participants who had similar demographics to active RHD registrants across Uganda. Biweekly motivational messages and monthly SAP reminders were automatically sent to each participant 3 days before injection due date. For the duration of the pilot, a total of 188 messages were sent, and 188 were received by participants. A total of 80% (n=152) of participants responded back to the SMS, and the remaining 20% (n=36) contacted the study team via telephone to indicate receipt of reminder and confirm injection date. Mean ACCEPT scores for arm 1 participants at enrollment and postpilot ranged from 43 to 61 at enrollment and from 52 to 63 at 2 months post‐SMS intervention. Arm 2 scores ranged from 43 to 60 at enrollment and from 45 to 59 at 2 months postintervention. The mean changes in score for arm 1 and arm 2 of the intervention at enrollment and 2 months following the deployment of the SMS intervention are 3.66 and 3.64, respectively.

Table 3.

Participant Demographics

Variable	SMS pilot		RHD registry (active patients)
Age, y	Mean	24	22
	Range	13–54	4–70
Female sex, %		66	61.5
Time since diagnosis, y	Mean	6.8	9.4
	Range	2–23	1–23
RHD category at diagnosis, %	ARF	20	31
	Latent borderline RHD	6	4
	Latent definite RHD	12	12
	Clinical RHD	62	54

ARF indicates acute rheumatic fever; RHD, rheumatic heart disease; and SMS, short message service.

During the interviews (F.O.), participants from arm 1 (n=5) and arm 2 (n=5) were asked to share their experiences around getting their antibiotic prophylaxis each month. Arm 2 included an additional line of questioning to elicit feedback on the SMS intervention. Following analysis of the data, emergent themes included the following: (1) benefits of the injection, (2) barriers to getting the injection, and (3) benefits of the SMS intervention.

Benefits of the Injection

Every participant, regardless of which arm of the pilot they were randomized to, indicated a significant improvement in their overall well‐being and symptoms since starting the monthly injections. Participant 259 relayed to the team, “Before starting the monthly injections, I used to get a lot of pain around my chest and difficulties breathing, especially after fetching water, but now I don't feel those pains even If I have done a lot of work. Therefore, there is a great improvement in my health status.” A notable subtheme identified during analysis was “this disease is curable.” Most interview participants believed that antibiotic prophylaxis would cure their RHD. Some participants indicated that their health care providers gave them this information during clinic visits, whereas others shared that some SMS messages led them to this belief.

Barriers to Getting the Injection

Most respondents pointed to distance and transportation costs as the most significant barrier to getting their injections on time. The second most notable barrier in this cohort of patients was related to work or school schedules. A few participants indicated that the injection‐related adverse effects created a significant barrier for them. And 1 participant cited medication shortages at the clinic and paying out of pocket, which was challenging.

Benefits of the SMS Intervention

All participants in arm 2 shared that the experience was positive and helpful. They all preferred the monthly reminder messages over the rest. They indicated that these messages were helpful reminders of the upcoming injections and allowed them to plan accordingly. Participant 370 shared, “The reminder messages help me plan my work and travel in advance to avoid certain problems that might make me miss my injection like my boss not letting me leave.” Most arm 2 participants relayed to the research team that they would like the SMS intervention extended beyond the pilot because it has benefited them overall. None of the participants related problems or concerns interacting with the technology. They indicated SMS was more efficient than paper reminders or monthly telephone calls from clinic nurses because the messages are sent directly to mobile telephones. Participant feedback on the frequency and timing of the messages was highly positive. The frequency and timing were found to be appropriate by everyone, with 2 participants suggesting that reminders be sent 2 days before the injection due date.

Discussion

Using user‐centered design methods, ⁸ which incorporated feedback from stakeholders and, most specifically, patients with RHD, allowed us to develop and test a library of motivational and reminder messaging to promote secondary antibiotic prophylaxis adherence on an easy‐to‐use technology platform that could alleviate care management burden for health care workers. One of the challenges in this process, which became apparent during phase 3 (crowdsourcing) was effectively translating theoretical concepts into concrete, clear, and engaging messages. The benefits and self‐efficacy messages performed poorly, with a significant misclassification rate and discarded messages. These findings may be related to the language barrier and ability to explain these theoretical concepts effectively during the message‐crafting phase and subsequent crowdsourcing. The reminder messages performed well in almost all the design process steps, except during phase 1, where participants indicated a slight preference for the motivational messages. Despite the challenges around the motivational messages, participants in arm 2 of the pilot had an extremely positive response to the SMS intervention and were eager for the program to continue beyond the study end date.

Globally, inadequate adherence is a major barrier to successful SAP in RHD prevention programs. ¹⁹ , ²⁰ Research around chronic diseases has identified self‐management support as a key factor to improve patient knowledge and motivation and provide patients with tools to manage their own health. ⁶ One aspect of self‐management is the use of SMS reminders to improve medication and clinical visit adherence. Several studies in low‐income settings suggest that SMS programs are feasible options that have demonstrated some success in improving medication adherence and show some promise in supporting clinic attendance. ⁵ , ⁶

Another element that has been identified as a key barrier to RHD management is patient education. ¹⁹ , ²⁰ , ²¹ Access to accurate information about RHD is essential in allowing patients to make informed decisions and understand symptoms, treatment, and the implication of SAP nonadherence.

With this pilot, we have demonstrated the acceptability and feasibility of 2‐way messaging technologies in promoting SAP adherence in Northern Uganda. The technology was easy to use and well received by the study cohort; however, more work is needed to refine the motivational and educational messages and to examine effectiveness of 2‐way messaging to achieve ideal SAP adherence and thereby RHD care management.

Limitations

This study had several limitations. First, although we considered cultural context, idioms, and literacy level during the message drafting phase, categorization of messages as either motivational or benefit was challenging. It may be that these 2 categories of messaging could ultimately be combined, as hearing the benefit of SAP is also motivating to those who need to receive it. Future directions in this work include development and testing of additional motivational messages. Second, the pilot was largely designed to be a feasibility study, and we did not find a difference in treatment acceptance between groups. It will be important to study treatment acceptance further in the future, including among newly diagnosed patients with RHD to avoid bias among those who have already shown high retention in care. We also plan a larger trial powered to determine if there is a difference in adherence and patient‐ focused outcomes between those receiving SMS support and those who receive standard care. Finally, although the feedback from participants was overwhelmingly positive about the experience of the SMS pilot, we must be cautious that some of this positivity could relate to cultural context and researcher‐introduced biases.

Conclusions

In this study, we coproduced a library of 43 messages that can be used to support SAP adherence in Uganda. Furthermore, we demonstrated that an SMS platform, using the RapidPro and the TextIt interface, had high uptake and was considered positively by patients enrolled in the current RHD registry. Next steps in this work include integration of automated reminders and motivational messages into ACT and a larger study to determine the impact of an SMS SAP support program on patient retention, adherence, and satisfaction with care.

Sources of Funding

Dr Fall was supported by an American Heart Association Award (No. 963234). Additional funding support for the project was provided by the American Heart Association Award (No. 20SFRN35360185).

Disclosures

None.

Supporting information

Table S1