Messages focused on reminding participants to attend their prescheduled HIV care appointment or reminded participants to schedule a new appointment.

This comprised message reminders to take their HIV medications. Messages varied from highlighting the importance of medications to, keep your body strong and healthy to reaching out to caseworkers to ask for help with medication adherence.

Participants chose to receive a message on safe sex practices or tips and mantras to avoid substance use.

Messages focused on areas participants may need help with when leaving the correctional system, such as finding housing, employment, etc. This category was adapted from the formative work to include specific resources found in Washington, DC, eg, providing the actual phone number of the office that helps returning citizens find employment within the message.

CARE+ Corrections study staff worked with an SMS vendor (Dimagi, Cambridge, MA) to create an SMS platform for automated text messaging. The initial SMS platform delivered messages from all 4 content areas in a single message thread at a preset frequency and in a single communication. Before study initiation, study staff members conducted pilot testing of the platform and concluded that greater flexibility in the frequency of messaging was required. Staff members indicated that receiving all messages in a single thread led to message fatigue and content was often overlooked, given multiple messages needed to be read at the same time. In response to this feedback, the SMS vendor was able to adapt the platform to allow participants to choose the frequency of messaging (daily versus weekly), timing of messaging (eg, am or pm), and ability to change messaging content in each category (Table 1), with slight variations in 2 message content categories: HIV Appointment Reminder and Barriers to Care.

Frequency for the HIV appointment reminder message depended on the date of the participant’s appointment and the Barriers to Care message was only available once per week to participants during the first month of the intervention or, if reincarcerated during the study period, 1 month following reentry into the community.

Most participants (52/57, 91%) chose to use a cell phone provided by the study (Table 2); of those participants, nearly two-thirds (35/52, 67%) required a replacement phone during the follow-up period. Most replacement phones (30/35, 86%) were provided because of reported loss or theft of the phone. Only 14% (5/35) of replacement phones were provided because of a faulty phone and did not count toward the participant’s one-replacement-phone policy. At the end of the study, participants were asked to return their study phones; however, more than 90% (47/52) of participants kept the phone provided by the study (they reported not having their phone at the final visit).

As nearly two-third of the phones provided by the study had to be replaced because of loss or theft, many were left without service while waiting for a replacement phone. In addition, reincarceration led to service disruptions. To avoid intervention interruptions because of phone issues, study staff had regular clinic hours in a known location throughout the duration of the study and participants knew to drop-in regarding any issues with phones. Furthermore, participants knew which community partners were affiliated with the CARE+ study, and study staff would receive calls from participants at these community partner locations to schedule appointments for phone replacements.

We used a pooled minutes cell phone plan, in which all phone lines shared available minutes, to account for some participants using more minutes and others less. Using this model, the study never went over its total allotted monthly minutes. Several participants exceeded their monthly minute allotments, and if the amount was significant, study staff would call the participant and review the participant’s cell phone plan. In 2 instances, participants used smartphone services that incurred additional fees (eg, downloading apps and calling 411 for information). In both cases, study staff worked with participants to call one of the community partners instead for information and inform participants that downloading apps on the study phone was not allowed. In addition, study staff worked with the phone carrier to limit the downloading of apps on study phones.

As the median age of CARE+ Corrections participants was 42 years, study staff recognized that cell phone training for smartphones would be required for some study participants. During the initial session, study staff assessed participants’ knowledge and ability in using the smartphones provided by the study and found that the majority of participants required 1-on-1 training before the initiation of the intervention. Study staff developed a 5- to 10-min cell phone training module that was delivered during the baseline visit (see Multimedia Appendix 2). The training module was interactive and included instructions on how to turn the phone on and off, entering information for contacts, opportunities to practice sending SMS text messages and making telephone calls, and instructions for using the talk-to-text option. We provided ongoing support at all follow-up appointments, including reviewing the training again at in-person visits.

The most popular message chosen at baseline for the HIV appointment reminder message (see Table 3) was, Hey how you feeling? Don’t forget to give a call and make your appointment (19/57, 33%), followed by Don’t forget your appointment – it’s important (11/57, 19%) and You’re worth it – remember your clinic appointment (8/57, 14%).

A total of 3 participants created custom message content at baseline, such as, Dr. [name] on [date] (see Multimedia Appendix 3 for other custom messaging). The most popular message time chosen at baseline was 8:00 am (20/57, 35%), followed by 12:00 pm (12/57, 21%). During follow-up, 28% (16/57) and 14% (8/57) of participants made changes to message content and time, respectively. The most popular message content changes during follow-up were to choose, Don’t forget your appointment – it’s important (5/17, 29%) and Your health comes first – go to your appointment (5/17, 29%). The most popular message time changes during follow-up were to choose 8:00 am (3/9, 33%) and 10:00 am (3/9, 33%).

The most popular message chosen at baseline was Don’t forget your skittles! (31/57, 54%), followed by, Meds keep your body strong and healthy (10/57, 18%) and Hey, take your vitamins! (7/57, 12%; Table 4).

A participant created a custom message at baseline and wanted us to include a smiley face within the message, Hey [name] don't forget those mones! :) The most popular message times chosen at baseline were 10:00 am (17/57, 30%) and 8:00 am (15/57, 26%). At baseline, most participants chose daily message frequency (36/57, 63%). During follow-up, 21% (12/57), 18% (10/57), and 14% (8/57) of participants made changes to message content, time, and frequency, respectively. The most popular message content change during follow-up was to create custom message content (3/12, 25%). The most popular message frequency change during follow-up was from weekly to daily (7/8, 88%).

The most popular message chosen at baseline was, Stay strong. Stay clean (18/57, 32%), followed by Safe sex is important. Use a condom (8/57, 14%) and Be smart. Use a condom (7/57, 12%; Table 5).

A total of 2 participants created custom content at baseline (see Multimedia Appendix 3). For 1 participant, the custom message of Keep your eyes on your own work was a saying his father used often to help the participant remember to stay focused on what matters.

When the messages were categorized as substance use prevention, safe sex, or custom content, most participants (31/57, 54%) chose substance use content. The most popular message times chosen at baseline were 8:00 am (13/57, 23%) and 8:00 pm (12/57, 21%). At baseline, most participants chose weekly message frequency (31/57, 54%). During follow-up, 26% (15/57), 28% (16/57), and 18% (10/57) of participants made changes to message content, time, and frequency, respectively. The most popular message time change during follow-up was to 10:00 am (7/18, 39%). The most popular message frequency change during follow-up was from weekly to daily (9/13, 69%).

The most popular messages chosen at baseline for the Barriers to Care messaging (Table 6) were Holla at your case manager, they’re here to help (12/57, 21%), Hey! Stay linked to your clinic so you can get your meds and care (9/57, 16%), Get help for your housing: call (XXX) XXX-XXXX [Local CBO that helps with housing] (9/57, 16%), and Check on job and training programs today (9/57, 16%).

More participants chose to create custom content for this message category compared with the other categories (see Multimedia Appendix 3). For example, a custom message, Hey don’t forget your parole appointment on [date], highlighted the importance of incarceration-related priorities and other custom messages focused on positivity, such as Keep hope alive!, We love you! and Stay positive! The most popular message times chosen at baseline were 10:00 am (17/57, 30%) and 8:00 am (16/57, 28%).

The study team saw huge improvements to the CARE+ Corrections intervention after pilot testing the intervention. Recognizing the significance of customization (ie, timing and frequency) of messaging to avoid message fatigue, the study team was able to make changes to the SMS platform before intervention implementation to provide study participants more flexibility.

Adapting the SMS text message library to accurately reflect the common experiences for the population was essential. For example, study participants were more likely to choose a substance use message versus the safe sex message under the Prevention Reminder category. This reflected what has been previously observed among incarcerated persons facing a 3- to 8-fold increased risk of drug-related death, 1 to 2 weeks following release compared with 3 to 12 weeks following release [35]. Future interventions should review the literature, engage with community-based organizations (CBOs), and conduct formative research with the population to inform message development to identify unique barriers and resources for the population of interest.

In addition, as our study population was older, adapting the implementation strategy to meet the needs of older participants was essential to effective implementation of the SMS text message plan. On the basis of previous literature [10,36,37], we recognized that additional cell phone training would be required for some study participants and developed a simple in-person training (see Multimedia Appendix 2). This was consistent with previous literature, in which a 2016 review of designing, implementing, and evaluating mHealth solutions among older adults highlighted the importance of ease of use of the mobile platform and an understanding of technical literacy of the user [38]. Furthermore, in-person contact helped to facilitate familiarizing the older population with mobile platforms. In our study, participants received 1-on-1 support with the initial setup of the CARE+ SMS program during their baseline visit. Another option for future mHealth interventions among older populations would be to use YouTube videos to further build rapport among study participants and build mobile skills [39].

The Android smartphones provided by the study were highly desirable; however, participants’ chaotic personal environments impacted smartphone retention. Most CARE+ participants opted to receive the phone provided by the study; however, most required a replacement phone and almost all kept their phone at the end of the study. In contrast, another STTR study reported 100% of users discarded the inexpensive, older model flip phone [10]. Future mHealth interventions, if possible, should continue to provide smartphones but consider budgeting more funds for cell phone replacement, given the likelihood of phone replacements. If the budget to provide a cell phone is not available and/or sustainable, an idea for future interventions would be to use Web-based telephone services (ie, Google Voice and WhatsApp) to continue sending SMS text messages via email when cell phone service is turned off, as was used in another STTR site [10].

Leveraging community partnerships proved very useful to avoid cell phone service interruptions because of lost or faulty phones. In addition to having regular office/study hours at the CARE+ community site, CARE+ staff identified contacts at multiple popular CBOs to provide participants with options for study engagement. Participants knew which community partners were affiliated with CARE+ and study staff would receive calls from participants at these community partner locations to set up appointments for phone replacements. This also proved useful for study retention purposes.

Identifying economic solutions with the cell phone carrier reduced economic burden on study budget. Using the pooled minutes approach provided the much-needed flexibility for CARE+ study participants, as some used very little of the suggested minutes allotted per month whereas others went over consistently. Future interventions should consider the pooled minutes approach to avoid overage issues and work with carriers to turn off specific apps and phone options that could incur monthly charges. Furthermore, incorporating cost-effectiveness analyses using templates [40], such as the one used by Reback et al [41] to evaluate a substance use and HIV risk reduction SMS intervention, will provide useful information for the feasibility of mHealth interventions at the population level.

Previous research has indicated that SMS interventions with 1 or more daily messages demonstrated smaller effects than interventions that only sent messages weekly [42]; however, CARE+ participant preferred more frequent SMS text messages. At baseline, participants were more likely to choose to receive the message every day instead of once per week. Furthermore, during follow-up, a large proportion of participants changed their message frequency option from weekly to daily.

Few chose to create their own message. We believe this could reflect the success and importance of the formative work, adequately reflecting the messages they wanted to receive. However, even with few selecting to create their own message, providing this as an option is important, as interventions that allow for message customization are more effective at promoting adherence to ART than those that send uniform messages to all participants [42].

For those who did create custom messages (see Multimedia Appendix 3), they provided us insights into missed messaging opportunities for future SMS interventions for persons with a history of incarceration. For example, a custom message reminded the participant of the participant’s parole appointment and thus highlighted the potential of adding community supervision—related reminder options to the Barriers to Care content area. Highlighting CJ status within the message was in contrast to our initial study goals of avoiding mentioning either HIV or CJ status. Future research is needed to explore the breadth and effectiveness of CJ-focused messaging.

Participants who were enrolled in the jail chose different Barriers to Care messaging compared with those participants enrolled in the community—highlighting the differences in barriers encountered at 2 different time points (immediate release versus up to 6 months before release from jail). The popularity of messages about seeking housing and job training programs among jail enrollees versus community-enrolled participants could be because of the fact that the latter may have had more time in the community to address these needs. This is supported by the literature, with persons immediately released from correctional facilities reporting transitional challenges, such as not knowing how to find shelter and feeling dumped into the city, unsure where to spend their first night [43]. Immediate needs following reentry can also vary by gender, with men reporting finding a job and education as most important immediately, whereas women identified shelter and substance abuse as their top priorities [44]. Future interventions should consider the timing of release from a correctional setting and choose Barriers to Care messaging that reflects the timing of release.