Implementation of free-draft text messaging to enhance care retention and satisfaction for persons living with HIV infection

Abstract

Eligible persons with HIV infection can receive client-centered case management to coordinate medical and social services. Novel mobile health interventions could improve effective case management and retention in care, an important goal to help end the HIV epidemic. Using a hybrid type I effectiveness-implementation design, we assessed whether access to bidirectional, free-draft secure text messaging with a case manager and clinic pharmacist could improve client satisfaction and care retention in a Southern academic HIV clinic. Sixty-four clients enrolled between November 2019 and March 2020, had a median age of 39 years, and were mostly male, single, and African-American. Heavy app users texted over 100 times (n=6) over the course of the 12-month intervention while others never texted (n=12). App usage peaked during months of clinic closure due to COVID-19. Most participants reported high satisfaction with the app and planned continued usage after study completion. Changes in clinic retention and virologic suppression rates were not observed, a result confounded by practice changes due to COVID-19. High usage and satisfaction of free-draft text messaging in case-managed HIV clients supports inclusion of this communication option in routine HIV clinical care.

Introduction

A disproportionately high percentage of persons living with HIV infection who are not linked and retained in care in the United States reside in South-eastern states, including South Carolina (1). Barriers to linkage and retention include a lack of trained physicians in rural areas, distance to care and inadequate access to transportation services (2, 3), socioeconomic barriers (4), and stigma (5). Retention in care allows access to highly effective antiretroviral medications that suppress HIV replication and prevent progression to immunodeficiency (6). Viral suppression also eliminates the risk of HIV transmission to sexual partners, which is a particularly important public health measure to control the epidemic as most new HIV transmissions in the US occur from persons with known infection who are out of care and not virologically suppressed (7–9). Thus, novel strategies to facilitate linkage and care retention are urgently needed to help end the HIV epidemic.

Mobile health technologies (mHealth) such as text messaging, video calling, telehealth, and virtual support groups have potential to address and help overcome linkage and retention barriers. Most people in the United States own a cell phone or smartphone irrespective of their socioeconomic status, including persons with HIV infection (10). Adoption of novel mHealth services on existing client devices could support inexpensive, convenient, and accessible approaches to improve linkage and care retention. Prior mHealth interventions that promoted medication adherence were of interest to providers and clients and demonstrated predominantly positive outcomes (11–19). The capacity to adapt impactful mHealth interventions into standard clinical practice requires further study, however, as privacy concerns or stigma could limit client interest (20). In addition, many mHealth studies have focused on scripted messages, sampled messages of support, reminders about adherence to anti-retroviral therapy (ART), or appointment reminders (16, 18, 19, 21). There has been more limited research examining use of bidirectional, free-draft text messaging, in which the composition of messages is directly composed by users rather than chosen from pre-selected choices, and prior studies did not all fully evaluate patient and provider satisfaction to assess acceptability and feasibility (22, 23).

The Ryan White funded Infectious Diseases Clinic at the Medical University of South Carolina (MUSC) provides comprehensive outpatient care for approximately 1300 persons living with HIV infection each year. Clients reside in rural and urban South Carolina and have varied transportation needs and access to internet services. Approximately 20% of clients are enrolled in case management services at any given time and receive client-centered coordination of medical and social services, which has been shown to positively affect clinical outcomes (23, 24). In a prior analysis, we conducted semi-structured interviews to understand mHealth acceptability and preferences and found that clients and providers in our clinic both perceived text messaging to be a highly acceptable, appropriate, and feasible way of communication to facilitate care (25). In the current study, we tested the hypothesis that access to bidirectional, free-draft text messaging between clients and their case manager and clinic pharmacist would improve client satisfaction and retention in care.

Materials and Methods

Study Design and Setting

We employed a hybrid type-I effectiveness-implementation study design to assess how access to bi-directional, free-draft texting using a secure app impacts client satisfaction and retention in care. Secondary goals were to assess intervention implementation, uptake, acceptability, and feasibility. Questionnaires were developed to measure participant perceptions and satisfaction pre- and post-intervention. Clinical data within the electronic medical record (EMR) were used to measure viral suppression and clinic retention in the study cohort relative to the entire clinic.

Clients enrolled in case management services in the MUSC Infectious Diseases clinic were screened for eligibility and offered participation during routine encounters with their case managers. Eligibility included being age 18 years or older, having an HIV diagnosis, and possession of a cellular device capable of text messaging. Clients were not eligible if they were currently in jail or prison or were not fluent in English. Case managers and the clinic pharmacist, hereafter referred to as providers, were also offered study enrollment. Clients and providers provided written informed consent and the study was approved by the Institutional Review Board at MUSC.

Study procedures and data security

Client participants completed a baseline questionnaire at the time of enrollment and were provided a $20 gift card. Client participants completed the free registration for Qliq (https://www.qliqsoft.com) (26), the secure text message app used in the study, with a study coordinator, and downloaded the app on their mobile device or computer. The app is accessed on mobile devices or computers by entering in a user-determined security code and does not interface with the medical record. Participants could access Qliq and communicate with their case manager and/or the clinic pharmacist as frequently as desired over the 12-month intervention period (Supplemental File 1 is the operational information sheet provided to study participants). After 12 months, participants were asked to complete a follow-up questionnaire and were compensated with a $20 gift card. Provider participants completed questionnaires at baseline and quarterly and were not compensated for participation.

Provider phones were protected by password and standard security measures (mobile device management), which allows remote wiping of data in the event of a lost or stolen device. Qliq usage data used for analysis was stored on an MUSC network server and coded with a key stored in a separate location. Extracted data was deidentified and assigned a unique identifying variable for analysis to protect sensitive protected health information. Participants could remotely delete their Qliq account if their mobile device was lost or stolen. Qliq requires clients to log-in using a passcode and is HIPAA-compliant (26).

Study questionnaires

Clients completed a self-administered questionnaire after enrollment using a secure database REDcap (Research Electronic Data Capture). Questionnaire topics assessed sociodemographic characteristics, satisfaction with care, personal health status and medical care history, utilization of clinic services, intervention feasibility (access to and comfort level with mobile technology), and interest in mHealth interventions (See Supplemental File 2 for Study Questionnaire). Questions about mHealth elicited descriptions of texting and video calling behavior, privacy concerns, comfort with texting and video calling, comfort with smartphone technology, and app usage. Responses related to comfort level and use of mobile technology were scaled on a 5-point Likert scale (ranging from “very interested” to “not interested at all”) (27).

Patient satisfaction was quantitated using a validated measure that includes two questions: 1) “Overall, how do you feel about the care you got at this clinic in the past 12 months?” and 2) “Would you recommend this clinic to other patients with HIV?” (28). Each question was based on a scale of 0–10 and responses were averaged to derive an overall score. Provider satisfaction was assessed by questionnaire with Likert-scale questions related to satisfaction with the intervention, impacts on burden, workload, client communication, and efficiency and allowed open-ended feedback (Supplemental File 3).

Data analysis

The study was initially designed to analyze text usage and compare retention data 18 months prior to and 12 months after the date of enrollment for each individual participant. The study was powered to detect a 10% change in pre- to post-satisfaction and required 75 participants. Given the profound impact of COVID-19 on clinic practices, the approach to analysis was modified for clinical outcomes to anchor analysis of the data on the date of clinic closure due to COVID-19 (March 16^th, 2020). Thus, irrespective of the date each participant enrolled (all subjects enrolled prior to clinic closure), retention and viral suppression were compared 18 months prior to March 16^th, 2020 and 12 months after. In an attempt to disentangle intervention effects from COVID-19 effects on clinical outcomes, data for enrolled participants was also compared to the entire clinic population, excluding the study cohort, over the same time period. While this comparison is imperfect (i.e., case-managed clients differ from the general clinic population in a number of ways), it allowed for a general comparison across both time and population (exposed vs. unexposed to the intervention). For outcomes related to care satisfaction and app usage, the original pre- to post-comparison (i.e., enrollment to 12-months post-enrollment) was maintained. Of note, on May 11^th, 2020 the clinic reopened for in-person visits, after which time clients could elect to have visits conducted either in-person or via telehealth, and MUSC labs remained open throughout the course of the pandemic.

Descriptive statistics including frequencies, percentages, and measures of central tendency (i.e., mean and median) were used to describe the study population and characteristics related to intervention feasibility. Logistic regression using a penalized likelihood measure was used to assess associations with interest in mHealth interventions with demographic variables (29).

Analysis of pre- and post-intervention patient questionnaires utilized paired t-tests for continuous variables with categorical data transformed into (1 = Very Poor, 2= Poor, 3 = Okay, 4 = Good, 5 = Very Good, 6 = Excellent). Cross-tabulations and chi-square tests were used to assess relationship between app usage (dichotomized at the median, <=13 texts sent as “low” use vs. >13 texts sent as “high” use) and reported satisfaction. Qualitative, open-ended responses from providers and clients were analyzed using content analysis (30), with free text responses extracted by one team member and categorized into themes following review and discussion among the study team.

To assess app usage, text messages with the subject matter removed were downloaded from the Qliq server and de-identified for analysis. Texts were sorted based on participant ID, staff member ID, and delivery status. As the study had rolling participant enrollment, analysis of messages for each participant was measured for 12 months from the date of enrollment.

To assess clinical outcomes, including retention in care, clinic visit completion data were extracted from the EMR and completed clinic visits and “no show” clinic visits were quantitated in the study cohort and the remaining clinic population (n=917, excluding the study cohort) who had available data for pre- and post-intervention time periods (9/9/18–3/15/20 and 3/16/20–3/15/21, respectively). HIV viral loads were extracted from the EMR and designated as suppressed (VL <200 copies/ml) or unsuppressed (VL>200 copies/ml) for the same time frames.

Results

Participant demographics and feasibility

Between 11/4/19 and 3/10/20, 84 clients were approached and 64 enrolled in the study and completed a pre-intervention questionnaire (Supplemental Figure 1). Sixty participants completed the study, and 4 participants withdrew prematurely due to moving out of state or death. The median cohort age was 39 years (range 19–65) and most participants were under 45 years (64%) (Table I). A majority of participants were male (57%), African American (81%), and single (60%). Nearly half of participants did not report education beyond high school (47%), over half were employed either part or full-time (59%), and 17% reported an annual income over $25,000.

Table I.

Participant Demographics (n = 64)

Variable	Categories	Frequency	Percent
Sex	Male	36	57
	Female	26	41
	Transgender (M to F)	1	2
Age	18 to 30 years old	19	30
(Median: 39, Range: 19 – 65)	31 to 50 years old	29	45
	> 50 years old	16	25
Years since HIV Diagnosis	< 1 year ago	5	8
	1 to 5 years ago	18	29
	> 5 years ago	39	63
Ethnicity	African American	52	81
	White	8	13
	Native American or Other	4	6
Sexual Orientation	Heterosexual (Straight)	30	48
	Homosexual (Gay)	21	34
	Bisexual	10	16
	Transgender	1	2
Relationship Status	Single, not in a committed relationship	38	60
	Unmarried, in a committed relationship	15	24
	Married	3	5
	Divorced, separated, or widowed	7	11
Employment Status	Full-time	24	39
	Less than full-time	12	20
	Not employed, seeking employment	12	20
	Not employed, not seeking employment	13	21
Insurance Status	Uninsured (care through Ryan White)	7	12
	Affordable Care Act insurance	17	30
	Private health insurance	7	12
	Medicare	8	14
	Medicaid	10	18
	Other	7	12
Annual Income	< $10,000	22	42
	$10,000 - $25,000	21	40
	> $25,000	9	17
Highest Level of Education Completed	High school or less	29	47
	Some college or associated degree	23	38
	College graduate or higher	9	15
Residence	Urban	27	42
	Rural	13	20
	Peri-Urban	10	16
	Don’t Know	14	22
Distance to Clinic	< 30 minutes	35	55
	> 30 minutes	29	45

All 64 participants had access to a mobile phone and most had access to a smartphone (89%) (Table II). Nearly all participants described “texting all the time without limits” (91%), and most did not have privacy concerns that would limit their use of texting (84%). Almost all participants expressed some level of interest in text communication with their case manager (96%) or pharmacist (93%), but no sociodemographic factors were associated with these perceptions (data not shown). There was also interest in virtual support groups with other clinic clients via an online chat group (61% interested). Age over 50 years (p=0.03) and having a college degree or higher level of education (p=0.02) were associated with less interest in virtual support groups. Participants who had to travel >30 minutes to get to clinic were significantly more likely to be interested in a virtual support group (p = 0.01). Additional feasibility and mHealth interest responses are shown in Table II.

Table II.

Pre-Intervention Responses Regarding Feasibility of Mobile Health Interventions (n = 64)

Variable	Categories	Frequency	Percent
MyChart Use (Patient Care Portal within the EHR)	Yes	44	72
	No	17	28
Access to a Cell Phone	Smartphone	57	89
	Basic cell phone (non-smartphone)	7	11
Description of Texting Behavior	Text all the time without limits	58	91
	Would like to text more, limited by cost	2	3
	Do not like texts, prefer calls	4	6
Privacy Concerns over Texting	Don’t care at all	21	33
	Thought about it, but would still text	32	51
	Concerned enough to limit texting	8	13
	Concerned enough to not text because of it	2	3
Comfort with Texting	Love it, no problems	43	68
	It’s okay, use it with specific people	15	24
	Can handle it, but prefer calling	5	8
Interest in communicating with case manager by text?	Not interested	1	2
	Neutral	1	2
	Somewhat interested	4	6
	Very interested	58	90
Interest in communicating with pharmacist by text?	Not	2	4
	Neutral	2	3
	Somewhat interested	10	16
	Very interested	48	77
Comfort with Smartphone	Love it, cannot stand to be away from it	41	65
	It’s okay, use it for calls can certain things	16	25
	Can handle it, use it only for calls	3	5
	Don’t own a smartphone	3	5
App Usage	All the time, like to try new apps	42	67
	Yes, but have fewer than 25 on phone	11	17
	Use a couple apps, only known ones	6	9
	Don’t use or don’t know much about apps	3	5
	Don’t own a smartphone	1	2
Interest in online chat group with other clinic patients?	Not interested	14	22
	Neutral	11	17
	Somewhat interested	11	17
	Very interested	28	44

Participant-based clinic satisfaction and intervention acceptability

Of 64 participants who enrolled, 41 completed both pre- and post-intervention questionnaires. Participant satisfaction with the clinic and self-reported medication adherence were already high pre-intervention and did not change significantly over the course of the study (Table III). Of note, 33% of post-intervention questionnaire respondents agreed that app usage made their health better, 45% felt they were more involved in their care, and 53% agreed texting helped monitor their condition (data not shown). Most participants who completed the post-intervention questionnaire were satisfied with the app (73%), agreed the app was easy to use (68%), and planned to continue using the app following study completion (51%) (data not shown).

Table III:

Pre- and Post-Intervention (n = 41) Comparisons

	Pre-Intervention Score (SD)	Post-Intervention Score (SD)	p-Value for paired analysis a
On a scale of 1–10 how satisfied are you with your care?b	Paired: 9.56 (1.31) Unpaired: 9.67 (1.31)	Paired: 9.53 (1.53)	0.87
On a scale of 1–10, how likely are you to recommend this clinic?c	Paired:9.26 (2.25) Unpaired: 9.77 (0.61)	Paired: 9.43 (1.66)	0.61
On a scale of 1–10, how happy are you with your case management?d	Paired: 9.57 (1.61) Unpaired: 9.95 (0.21)	Paired: 9.77 (0.93)	0.55
How would you rate your communication with the pharmacist?e*	Paired: 4.95 (1.15) Unpaired: 5.13 (1.04)	Paired: 5.13 (1.06)	0.39
How would you rate your overall health?f*	Paired: 4.53 (1.12) Unpaired: 4.87 (1.14)	Paired: 4.60 (1.05)	0.65
In the past 4 weeks, how would you rate your ability to take all your HIV medicines as your doctor prescribed?g*	Paired: 5.09 (1.18) Unpaired: 5.43 (0.99)	Paired: 5.04 (1.28)	0.81

Paired data are from participants completing both questionnaires. Unpaired data are from participants who only completed the pre-intervention questionnaire.

^ap-values are for paired responses only

^b(n = 36

^cn= 41

^dn= 37

^en= 39

^fn= 41

^gn= 41).

^*Categorical responses were translated to numeric values (1 = Very Poor, 2= Poor, 3 = Okay, 4 = Good, 5 = Very Good, 6 = Excellent) and averaged.

Participants were given the opportunity to leave free response comments on the post-intervention questionnaire to assess their attitudes towards use of the app. One participant reported, “I love it!”. One concern raised was limited internet access, with one participant stating that “[they] live in the country and so sometimes the service is limited.”

Provider perceptions and experiences with intervention implementation

Four providers enrolled in the study, including three case managers and one clinic pharmacist. Providers were generally enthusiastic about use of the app and this perception did not change during the study, as assessed by deidentified quarterly questionnaires (Supplemental Table I). Over time, most providers agreed text messaging helped improve efficiency and made a positive difference in clients’ lives. Perceived benefits noted by providers included increased efficiency through “minimiz[ing] a lot of ‘phone tag’,” and being able to reach clients who otherwise would not answer a phone call. At baseline, there was concern of receiving “too many texts” and concerns that text messaging would be duplicative of existing modes of communication, such as online messaging available through the online patient portal, “MyChart”. At the end of the study, however, no providers indicated that app usage increased their workload, created a burden, or that they received too many texts. During the intervention, providers reported some logistical challenges of app usage, such as participants losing their passcode and questions as to whether using standard text messaging would be easier. Use of the app was noted to be particularly helpful during the onset of the COVID-19 pandemic. When asked specifically about COVID-19, one case manager said “the text messaging has increased due to the COVID-19 pandemic. More clients have started to take advantage of the program. It is beneficial and easier to answer questions they have a hard time asking verbally.” For providers, an unexpected benefit to the app during the pandemic was using it to communicate with one another, not just with clients, particularly in the setting of remote work.

App Usage

Of 64 participants who enrolled, 6 were heavy text users (>100 texts), 9 texted between 50 and 100 times, and 12 participants did not send a single text (Figure 1A). Texting usage by month was highest early in the intervention and during the period of clinic closure during COVID-19, and subsequently decreased over the course of the study period (Figure 1B). In the qualitative responses provided by responses, this phenomenon was anecdotally noted by a provider in the final quarter, “I no longer feel the texting app has the usage by patients that it once did.” When data were normalized by month of enrollment for all participants, texting usage was more uniform (Figure 1C), suggesting the decline observed in Figure 1B may have been influenced by participants who enrolled later and used the app less frequently. Participants who reported intention to text after completing the 12-month intervention texted more frequently than participants who reported no further plans to text or who did not answer this question (Figure 1D). There was no correlation between texting frequency and MyChart usage, gender, urban vs rural residence, age, education level, race, or length diagnosed with HIV infection (data not shown).

Figure 1:

Variable app usage by participants and providers over the course of the study. (A) Shown are the number of participants who sent the indicated number of texts over 12-months intervention (n=64). (B) Graphed are the average number of texts sent per participant per month between November 2019 and March 2021 (n=64). The number of participants enrolled during each calendar month is indicated numerically. Dates of clinic closure due to COVID-19 are indicated. (C) Graphed are the average number of texts sent per participant per month normalized for each subject by the time of enrollment (n=64). The number of participants enrolled during each month of the intervention is indicated numerically. (D) Shown are the total number of texts sent over the course of the intervention in subjects based on whether they plan future use of the app after completion of the 12-month intervention. p-value is by unpaired t-test. (E) Shown are the average number of texts sent and received per provider per participant. CM=case manager and PharmD=clinic pharmacist. (n=5)

Similarly, use of the app varied across providers. Several case managers used the app frequently, while another case manager and the clinic pharmacist used the app infrequently (Figure 1E). The pharmacist was much more likely to use the app to communicate with case managers rather than directly with patients (data not shown). For communication with clients, the pharmacist found that a phone call was typically preferred for follow-up rather than a text message due to topic complexity.

Clinical outcomes

Quantitative data within the EMR were extracted to assess clinic visit adherence and virologic suppression pre- and post-clinic closure due to COVID-19 in the study cohort relative to the overall clinic. As noted in the methods section, this analytic approach was undertaken to describe and separate intervention effects from effects on clinical outcomes caused by the dramatic changes in clinic practices that occurred as a result of COVID-19. The study cohort had more completed visits pre-COVID clinic closure relative to the rest of the clinic, suggesting participants receiving case management services may have or require more frequent clnical encounters (Figure 2A). Completed visit frequency declined slightly post-COVID clinic closure in the study cohort (Figure 2A), as might be expected across the institution due to changes in clinic practices and transition of case managers to remote work. “No-show” visits were higher post-COVID closure in the study cohort relative to the rest of the clinic (Figure 2B), suggesting use of the app did not influence no-show rates. There was a decrease in the number of suppressed viral loads post-COVID in both the clinic and study cohorts (Figure 2C), suggesting all patients had labs drawn less frequently post-COVID clinic closure, which was likely true for all patients cared for across the institution outside of our clinic. While the study cohort had a higher number of un-suppressed viral loads pre-COVID clinic closure relative to the clinic cohort (Figure 2D), there was no difference post-COVID clinic closure, suggesting the text messaging intervention could have helped with medication adherence.

Figure 2:

Clinic retention and viral suppression metrics in the study cohort (n=64) relative to the rest of the clinic (n=917). Shown are the number of completed clinic visits (A), “no-show” clinic visits (B), suppressed viral loads (C), and detectable viral loads per 12 months per participant. Data are grouped by the study cohort (“cohort”) relative to the rest of the clinic (“clinic”) and for the 18 months prior to clinic closure due to COVID-19 (“Pre”) and the following 12 months (“Post”). Pre vs. post analyses within groups were performed by paired t-test. Pre- vs pre and post vs post analyses between groups were performed by unpaired t-test.

Discussion

In this study, we identified high interest and feasibility of using a bidirectional, free-draft texting program in a subset of case-managed participants and their case managers in the MUSC HIV Clinic. All case-managed participants had access to a cellphone and most had a smartphone even though few reported an annual income over $25,000. Furthermore, feasibility did not appear to be limited by privacy concerns associated with the use of mobile phones, as most participants did not have enough concerns to limit their texting behavior. Interest and participation in the text messaging intervention with a case manager or pharmacist was high across all sociodemographic factors. Results related to the high acceptability of mHealth interventions found in this study are consistent with other studies conducted among people living with HIV in the United States (31, 32).

Actual usage of the text messaging app varied widely across clients and providers. While this finding is not unexpected and potentially speaks to individual preferences, barriers to access, and differing perceptions of utility, it is important nonetheless as a reminder that mHealth interventions may not have widespread utility and be enthusiastically embraced by all. However, as evidenced by the substantial amount of app usage among some participants, having the ability to text message with a case manager and/or pharmacist was useful for a subset of participants, and the majority of participants who completed the post-intervention questionnaire wanted to continue using the app after study completion. This finding affirms the utility of having text messaging as an available tool for client communication.

Critical questions remain to better understand who will use and benefit from an mHealth intervention. In our study, we evaluated potential differences between those who used the intervention frequently and those who did not, but identified no differences related to sociodemographic and behavioral variables. We also assessed for differences between participants who completed the post-intervention assessment and those who did not, reasoning that attrition due to loss-to-follow-up might be synonymous with disengagement from the intervention itself (33), but again found no significant differences.

App usage among providers varied. This finding confirms formative work performed in preparation for study implementation that demonstrated differing perceptions of utility and expected burden of text messaging among providers (25), which presumably impacted subsequent enthusiasm and use. While we did not employ specific implementation strategies to encourage app use with providers, such as offering incentives or identifying staff “champions”, identifying strategies to engage providers will be important if text messaging is rolled-out on a larger scale. Alternatively, efforts can be made to ensure client communication preferences match those of their primary case manager. While the pharmacist participant used the app to text internally with case managers, in most cases they identified that a phone call was a preferred means of follow-up with clients due to topic complexity.

We found marginal evidence that allowing text messaging improved care satisfaction, but no evidence that clinic retention or viral suppression were impacted. However, it is important to interpret these results considering several limitations. The study was initially powered to detect a 10% change in pre- to post-satisfaction and required 75 participants. As recruitment was halted early due to the onset of the COVID-19 pandemic and only 64 participants enrolled, the study may have been underpowered to detect a significant difference in satisfaction. Additionally, satisfaction was high at baseline, resulting in a ceiling effect with little room for improvement. Regarding clinical outcomes, the COVID-19 pandemic caused substantial changes to care during the study, which limited our capacity to understand how access to the intervention impacted care outcomes. Another limitation is the comparison of enrolled case-managed patients to the entire clinic population to assess changes in clinical outcomes. These populations differ in several ways, and the clinic comparison group was not subjected to the same inclusion criteria used for the enrolled case-managed group (i.e., English fluency, ownership of a mobile phone). However, given the low socioeconomic status of case-management patients and the near ubiquity of cell phone ownership among this disadvantaged group, we do not anticipate these groups would differ significantly in their ability to use text messaging as a tool overall.

Additionally, the uneven usage of the app overtime, coupled with changes in use patterns presumably due to the disruptions caused by the onset of COVID-19, complicate our ability to attribute to findings to the intervention. However, given the fact that usage of text messaging peaked during the period when the clinic was temporarily closed due to COVID-19 suggests it proved useful in maintaining ties between clients and providers during a critical time.

The potential redundancy of having a secure text messaging app in addition to secure messaging available via the electronic patient portal was raised several times by providers. Communicating through the patient portal has the added benefit of seamlessly documenting patient-provider exchanges in the EMR, although apps like Qliq also have a growing capacity to link with EMR systems. However, in this study we found no association between use of the patient portal and use of Qliq, suggesting Qliq might be reaching a different segment of the client population. A prior study among patients with chronic conditions found low technology literacy, worries of overburdening providers, and uncertainty about which types of messages to send were barriers to using the securing messaging feature of the patient portal (34). More research is thus needed to understand differences in preferences for various types of electronic communication, as well as understanding the value and utility of integrating documentation of such communication into EMRs.

In conclusion, identifying high acceptability and feasibility for use of text messaging as a mode of communication between a subset of case-managed clients living with HIV and their case management team is encouraging, especially among populations in the Southeastern United States that are disproportionately impacted by HIV and face heightened structural and social barriers to care retention. This finding supports inclusion of this communication option in routine HIV clinical care. While these results are promising, we did not find evidence that having access to text messaging impacted care retention or virologic suppression, although the onset of the COVID-19 pandemic and its sequelae likely impacted our ability to determine effectiveness. Further research is needed to understand which clients may benefit the most from such communication and how communication channels can be enhanced to maximize benefits for clients and efficiency among clinic staff.

Supplementary Material

Supplementary Figure 1

Supplemental Figure 1: Flow and outcomes for participant enrollment.

Data availability:

Available from the corresponding author upon request.