Two-way text message interventions and healthcare outcomes in Africa: Systematic review of randomized trials with meta-analyses on appointment attendance and medicine adherence.

BACKGROUND: The growing access to mobile phones in Africa has led to an increase in mobile health interventions, including an increasing number of two-way text message interventions. However, their effect on healthcare outcomes in an African context is uncertain. This systematic review aims to landscape randomized trials involving two-way text message interventions and estimate their effect on healthcare outcomes.
METHODS: We searched Medline, Embase, Cochrane Central Register of Controlled Trials, The Global Health Library (up to 12 August 2021) and trial registries (up to 24 April 2020). Published and unpublished trials conducted in Africa comparing two-way text message interventions with standard care and/or one-way text message interventions were included. Trials that reported dichotomous effect estimates on healthcare appointment attendance and/or medicine adherence were assessed for risk of bias and included in meta-analyses. Results of other outcomes were reported descriptively.
RESULTS: We included 31 trials (28,563 participants) all set in Sub-Saharan Africa with a wide range of clinical conditions. Overall, ten different trials were included in the primary meta-analyses, and two of these had data on both medicine adherence and appointment attendance. An additional two trials were included in sensitivity analyses. Of the 12 included trials, three were judged as overall low risk of bias and nine as overall high risk of bias trials. Two-way text messages did not improve appointment attendance, RR: 1.03; 95% CI: 0.95-1.12, I2 = 53% (5 trials, 4374 participants) but improved medicine adherence compared to standard care, RR: 1.14, 95% CI: 1.07-1.21, I2 = 8% (6 trials, 2783 participants).
CONCLUSION: Two-way text messages seemingly improve medicine adherence but has an uncertain effect on appointment attendance compared to standard care. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42020175810.

Introduction

For the past 30 years, information and communication technologies have transformed the world [1]. In recent years mobile phone access has expanded immensely across Africa and other low- and middle-income countries (LMICs). According to the World Health Organization (WHO), more people have access to a mobile phone than to clean running water in sub-Saharan Africa [2]. In 2017, approximately 8 out of 10 people in sub-Saharan Africa owned a mobile phone with an increasing amount of these being smartphones [3]. The growing access to mobile phones has led to new possibilities in relation to health, i.e., mobile health (mHealth), which comprises a wide range of communication channels such as mobile phone surveys, mobile phone calls or Short Message Service (SMS)/text messages between clients and health practitioners [2, 4]. One-way text messages are messages, which the receiver cannot respond to, while two-way messages are a form of interactive communication where the text message receiver can respond to the message or in other ways interact with the sender, e.g., by participating in a text message quiz, responding to the text message, calling the sender or receiving a phone call from the sender [5].

A systematic review and meta-analysis from 2015 –which included trials conducted in various settings–suggested that interactive communication between a client and a healthcare provider may lead to greater support, motivation, and safety for the patient. The meta-analysis found a higher intervention effect of two-way messages compared to one-way text messages on medicine adherence [6]. Further, a series of Cochrane reviews published between 2012 and 2017 assessed the effect of various text message interventions on different health issues [7-12], but the reviews did not differentiate between one-way and two-way text message interventions in their analyses, hence, the effect of one-way versus two-way text messages interventions was unclear. Additionally, most trials included in these reviews were conducted in high-income countries and only two reviews included trials from Africa [10, 11]. However, the geographical setting may be an important factor to account for in relation to the effect of text message interventions as digital literacy, network infrastructure, and cultural/social acceptance of mHealth interventions may differ across regions [2]. A systematic review and meta-analysis from 2019 [5] landscaped randomized controlled trials (RCTs) of one-way text message interventions in Africa and reported an effect on healthcare appointment attendance, OR: 2.03; 95% CI: 1.40–2.95 whilst the effect was uncertain on medicine adherence, RR 1.10; 95% CI: 0.98–1.23. Research shows that linkage to care is a challenge in many resource-limited settings [13], and it is plausible that two-way text messages–which allow for interactive communication with health care providers–may be a more effective health care tool than one-way messages. Appointment attendance and medicine adherence are outcomes frequently used to measure linkage to care across clinical settings, yet the effect of two-way messages on these outcomes in African context is unclear. Therefore, this systematic review aims to landscape randomized trials in Africa involving two-way text message interventions compared to standard care or one-way text messages and analyze their effect on appointment attendance and medicine adherence. By conducting a meta-analysis on the same outcomes as the previous systematic review of one-way messages in Africa [5], this review will provide an even greater understanding of how various types of text message interventions work in an African context. Further, in line with the previous review, we also describe other health outcomes to provide a full overview of two-way message trials set in Africa.

Materials and methods

Protocol registration

This systematic review is based on a protocol reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocol (PRISMA-P) guidelines [14] (S1 File) and registered prior to review conduct (PROSPERO CRD42020175810).

Eligibility criteria

We included randomized trials of male and female healthcare clients, partners of or guardians for healthcare clients, e.g., parents for child patients. Trials targeting health personnel and healthcare providers were excluded. Trials comparing two-way text message interventions with standard care and/or one-way text messages were included. We defined two-way text messages as a text message the receiver could react or respond to, with the use of a mobile phone, in any way. If trials were multi-arm, e.g., included one-way and two-way interventions as well as standard care, these were included and data for each arm were analyzed separately. If co-interventions (e.g., written material) were received by participants in both the intervention and control arms, this was considered to be part of standard care, and such trials were included. We included published and unpublished randomized trials in any language conducted in Africa, including pilot and cluster trials. International multicenter trials that had African sites were included if separate results for the African sites were reported or could be accessed upon request. Trials reporting all healthcare outcomes were included if they met all other eligibility criteria.

Information sources and search strategy

We searched Medline, Embase, Cochrane Central Register of Controlled Trials and The Global Health Library until 12 August 2021. The search strategy was adapted from a strategy used in our previous systematic review of one-way text message interventions in Africa [5]. We tailored the search strategy for each database with the support of an information specialist. The search string included search terms such as “randomized controlled trial”, “Africa”, “SMS”, “mobile phone”, “two-way text messages” and “two-way SMS” (S2 File). ClinicalTrials.gov, Pan African Clinical Trial Registry (PACTR) and The International Standard Randomised Controlled Trial Number registry (ISRCTN) were searched up to 24 April 2020 for additional trials, including ongoing and unpublished trials. Reference lists of relevant systematic reviews and of the included trials were searched for additional trials. Finally, we searched the UN and World Bank reports for additional trials (up to 18 March 2020). In August 2021 the status of ongoing trials was checked for published results.

Unique records were uploaded to Covidence (www.covidence.org). Two authors (ESØ, LSL) independently screened titles and abstracts of all retrieved records. In the title-abstract screening, trials were only excluded if the title and abstract obviously stated that interventions consisted of one-way messages as according to our definition. Trials selected for full text evaluation were independently assessed by two authors (ESØ, LSL) for inclusion. Two authors (ESØ, LSL) extracted data, and each author extracted data from half of the included articles into a standardized Excel data extraction sheet and verified each other´s data extraction for trial outcomes. Disagreements in relation to trial inclusion and data extraction were resolved through discussion, and if consensus could not be reached an arbiter (DSL) made the final decision. Extracted data included: first author, title, publication year, name of journal/registry, objective, inclusion/ and exclusion criteria, randomization method, trial period, sample size, description of experimental and control interventions, primary target group, pre-test and theoretical foundation of intervention, length of follow-up, trial outcomes, results for primary trial outcome and results for appointment attendance and medicine adherence. Two authors (ESØ, LSL) contacted trial authors for missing data, including data from unpublished trials.

Risk of bias assessment

Two authors (ESØ, LSL) independently assessed all trials that were eligible for meta-analysis (trials reporting the outcomes “appointment attendance” and/or “medicine adherence” for risk of bias using the Cochrane Risk of Bias Tool [15]. If consensus could not be reached, and arbiter (DSL) made the final decision. The items assessed were random sequence generation and allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other biases. As part of the other biases assessment “The Beall’s list of potential predatory publishers” and the “List of Predatory journals” [16, 17] were searched, and trials that were published in a potential predatory journal were judged as high risk of other biases. For cluster randomized trials, risk of bias related to recruitment and baseline imbalance were also assessed [15]. The items were judged to have low, high, or unclear risk of bias. However, attrition bias was not assessed for trials that only reported “appointment attendance” as an outcome as incomplete data is an integral part of this outcome, i.e., non-attendance is equivalent to loss to follow-up. If trials were judged to have low risk of selection, detection and reporting bias, they were judged to have overall low risk of bias. If not, they were judged to have overall high risk of bias.

Data analysis

Two-way text message interventions were expected to be used in various types of populations and settings and for multiple types of outcomes. Therefore, we conducted an overall descriptive analysis of all trials and restricted meta-analyses to the outcomes “appointment attendance” and “medicine adherence”, which we regarded to be uniform. For our descriptive analysis, we reported unadjusted trial results for the primary outcome, unless the trial only reported adjusted results. If no quantitative estimates were reported, but binary data were available, risk ratios (RRs) and 95% confidence intervals were calculated.

Meta-analyses were done using Review Manager 5.4.1 [18]. Due to the anticipated clinical and methodological heterogeneity, we planned to calculate pooled RRs and estimate 95% confidence intervals (CIs) using a random-effects model with the Mantel-Haenszel method for dichotomous data, for both appointment attendance and medicine adherence. However, two trials assessing appointment attendance [19, 20], and one trial assessing medicine adherence [21], were randomized at cluster level, and we therefore used the inverse variance method for our analyses. We only included trials with results adjusted for clustering in our primary analysis. Heterogeneity was assessed using I2 [15]. We conducted separate analyses comparing two-way text messages to one-way text messages. Subgroup analyses were performed comparing overall low risk of bias trials with high risk of bias trials, clinical areas and types of interventions, e.g. if the receiver could interact with a health care professional (HCP) through a phone call or text message. Further, sensitivity analyses were performed using the fixed-effect models and excluding the cluster randomized trials [19-21]. In addition, one of the three cluster trials did not adjust for clustering [21] and was excluded from our primary analysis but included in an additional sensitivity analysis. We intended to exclude trials published in potential predatory journals in a sensitivity analysis, but all trials were published in legitimate journals. Trials where we were unable to extract the required dichotomous event and participant data or trials that only reported continuous outcomes were excluded from the meta-analysis and analyzed descriptively. If adherence was measured in multiple ways in a trial, the most objective outcome measure was chosen for the meta-analysis, e.g. pill box openings were prioritized over self-reported adherence. However, dichotomous self-reported outcomes were used if it was the only outcome measure apart from a continuous outcome.

Results

Study selection

We identified 2624 records in our database search. After removal of duplicates there were 2150 unique records of which 2068 were excluded based on title- or abstracts (Fig 1). Eighty-two records were eligible for full text evaluation and a total of 29 trials were eligible to be included in the review [19-47], however, one multicenter trial was excluded because of no individual data on the African settings [42]. Further, searching other sources led to the inclusion of 3 additional trials [48-50]. In total, we included 31 trials, of which 20 were published [19–35, 43–45] and 11 were unpublished; six of the unpublished trials were finished [36–38, 41, 48, 49] and 5 had an unknown status [39, 40, 46, 47, 50].

Fig 1

Flow diagram.

From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7): e1000097. doi:10.1371/journal.pmed1000097. For more information, visit www.prisma-statement.org.

Study characteristics

Trials were published between 2010 and 2021 and included a total of 16,924 participants (median: 846 participants per trial) [19–35, 43–45] (Table 1). For the eleven unpublished trials, ten trials planned to enroll a total of 11,639 participants (median: 1164 participants per trial) [36–41, 46–50] (Table 1), and one remaining trial did not report the planned number of participants to be enrolled [49]. Eighteen corresponding authors were contacted for missing data including data from unpublished trials, and ten replied [28, 33–35, 37, 40–42, 48, 50]. Relevant queries were answered, but none of the respondents provided us with missing data or unpublished results.

Table 1

Two-way text message trials set in Africa.

Publication year, author	Country	Clinical area	Trial size (n)	Female participants	Age (mean)	Follow-up length (weeks)	Primary outcome	Type of two-way intervention	Effect of two-way intervention on primary outcome compared to control
								Type of one-way intervention	Effect of two-way intervention on primary outcome compared to one-way intervention
Finished trials, published
2021, Abiodun [43]	Nigeria	HIV	212	48%	16.6	20	Medicine adherence	Appointment and medicine adherence reminder + text message response option	RR: 1.2 [95% CI: 0.9 to 1.6](VAS score ≥95% adherence)
								Appointment reminders (control group)	-
2021, Kinuthia [44]	Kenya	HIV	824	100%	27	24	(1)Appointment attendance (2)Medicine adherence	Reminders + Educational + text message response option	aRRAppointment attendance: 1.0 [95% CI: 1.0 to 1.0] ¶
									aRRMedicine adherence: 0.8 [95% CI: 0.5 to 1.2]¶
								Reminders + Educational	aRRAppointment attendance: 1.0 [95% CI: 1.0 to 1.0] ¶
									aRRMedicine adherence: 0.8 [95% CI: 0.5 to 1.2]¶
2021, Sumari-de Boer [45]	Tanzania	HIV	245	71%	41,2	48	Medicine adherence	Reminders + text message response option	RR††: 1.1 [95% CI: 0.8 to 1.4]
2020, Ampt [35]	Kenya	Unintended pregnancies among sex workers	882	100%	25.4	52	Incidence of unintended pregnancy	Educational and motivational with text message response option and on-demand additional messages	HR*: 1.0 [95% CI: 0.7 to 1.4]
2019, Feldacker [22]	Zimbabwe	HIV-related circumcision	722	0%	24	2	Adverse event rate	Supportive with text message response option	Rateintervention: 6/320; Ratecontrol: 3/359
									RD†: 1.04% (p = 0.32)
2019, Harrington [32]	Kenya	Post-partum contraception	260	100%y	22.8	38	Post-partum contraceptive use	Educational with text message response option	RR‡: 1.2 [95% CI: 1.0 to 1.5]
2019, Odeny [19]	Kenya	HIV and post-partum care	2515 (20 clusters)	100% HIV-positive mothers	27	8	(1) Infant HIV testing; (2) post-partum appointment attendance	Supportive + educational + reminders + text message/phone call option	RRHIV-testing: 1.1 [95% CI: 1.0 to 1.1]
									RRpost-partum attendance: 1.2 [95% CI: 1.0 to 1.3]
2018, Unger [25]	Kenya	Antenatal care	298	100% Mothers	23	24	Delivery at health facility	Individual tailored educational + motivational + quiz	RR: 1.0 [95% CI: 1.0 to 1.0]
								Individual tailored educational + motivational	-
2018, Van der Kop [26]	Kenya	HIV	700	60%	33.7	55	Appointment attendance	Supportive with text message response option	RR: 1.0 [95% CI: 0.9 to 1.1]
2017, Linnemayr [27]	Uganda	HIV	332	61%	18.3	48	Medicine adherence	Supportive with text message response option	Mean adherence‡ 2-way intervention vs. control = 0.61 vs. 0.67, (p = 0.15)
								Supportive	-
2017, Rokicki [33]	Ghana	Sexual reproductive health	756 (38 clusters)	100%	17.7	15	Reproductive health knowledge	Educational + quiz where correct answer is sent after responding to the quiz	24% higher score than control group [95% CI: 19 to 28]
								Educational	13% higher score than one-way intervention [95% CI: 8 to 18]
2017, Van Olmen [34]	Congo Cambodia Philippines	Diabetes type 1 and 2	1471 Congo: 506	66%	60	104	Diabetes control (Hb1Ac < 7%)	Educational + supportive + text message/phone call option	RD§diabetes control, all countries = 2,7% (p = 0.4)
									RD§diabetes control, Congo = 7,5%
2016, Bobrow [23]	South Africa	Hypertension	1372	72%	54.3	52	Change in mean systolic blood pressure	Educational + motivational + reminders + phone call option	RD: -1.6 mmHg [95% CI: −3.7 to—0.7]
								Educational + motivational + reminders	-
2016, Kassaye [21]	Kenya	HIV	550 (26 clusters)	100% Mothers	25.6	6	Mothers medicine adherence	Educational + motivational + supportive + reminders + text message/phone call option	aRR¶: 1.0 [95% CI: 0.9 to 1.2]
2016, Leiby [28]	Zambia	HIV-related circumcision	1652	0%	23 [median]	26	Appointment attendance	Educational + text message option [Conventional two-way intervention]	OR ||: 1.1 [95% CI: 0.8 to 1.7]
								Educational + text message option [Tailored two-way intervention]	-
2014, Lund [20]	Tanzania	Antenatal care	2550 (24 clusters)	100% Mothers	27 [median]	6 (post-partum)	Appointment attendance	Educational + reminders + phone call voucher	ORappointment attendance:1.5 [95% CI: 0.8 to 3.0]
2014, Modrek [29]	Nigeria	Malaria	457	46%	39	4 days	Medicine adherence	Reminder + phone call option	OR: 2.1 [95% CI: 1.4 to 3.0]
2014, Odeny [24]	Kenya	HIV	388	100% Mothers	29 [median]	8	(1) Infant HIV testing; (2) post-partum appointment attendance	Supportive + educational + reminders + phone call option	RRHIV-testing: 1.1 [95% CI: 1.0 to 1.2]
									RRpost-partum attendance: 1.7 [95% CI: 1.0 to 2.7]
2012, Mbuagbaw [30]	Cameroon	HIV	200	74%	40.2	26	Medicine adherence	Motivational + reminder + phone call option	RR: 1.1 [95% CI: 0.9 to 1.3]
2010, Lester [31]	Kenya	HIV	538	65%	36.7	52	Medicine adherence	Supportive with text message response option	RR: 1.2 [95% CI: 1.0–1.4]
Finished trials, unpublished
2018, Gonsalves [38]	Kenya Peru	Sexual and reproductive health	1395	-	-	8	Sexual and reproductive health knowledge	Educational + quiz + access to previous educational domains [Two-way intervention A]	-
								Educational + quiz [Two-way intervention B]	-
2018, Odeny [37]	Kenya	HIV	1338	100% Mothers	-	12	Medicine adherence	Educational + supportive + call back option	-
								Community mentor mother support	-
								Educational + supportive + call back option + community mentor mother support	-
2016, Lippman** [41]	South Africa	HIV	752 (18 clusters)	39%	41,5	12	Medicine adherence	Educational + reminders + phone call option	OR: 1 [95% CI: 0.6 to1.5]
								Peer navigators	OR: 1.5 [0.9 to 2.5]
2016, Awiti [36]	Kenya	HIV	600 (planned)	100% Mothers	-	30	Medicine adherence	Supportive with text message response option	-
2015, NCT02627365** [48]	Kenya	HIV	119	100%	-	6	Medicine adherence	Educational + motivational + response option	-
2010, NCT01157442 [49]	Kenya	HIV	-	100%	-	12	Medicine adherence	Reminders + text message response option	-
Trials with unknown status, unpublished
2020, Unger [46]	Kenya	Antenatal care	5000 (planned)	100%	-	28–36 weeks gestation to 6 weeks post partum	Neonatal mortality	Educational + text message response option	-
2019, NCT04038060 [50]	South Africa	HIV	350 (planned)	100%	-	12	Medicine adherence	Supportive text message + counselling or drug level feedback	-
								Supportive WhatsApp group + counselling or drug level feedback	-
2019, Tickell [47]	Kenya	Malnutrition	1200 (planned)	-	-	6	Time to diagnosis of acute malnutrition	Reminders + text message response option	-
2017, Drake [39]	Kenya	HIV	825 (planned)	100% Mothers	-	24	Medicine adherence	Educational + motivational reminders + quiz	-
								Educational + motivational reminders	-
2017, Zunza [40]	South Africa	HIV	60 (planned)	100% Mothers	-	24	Breastfeeding adherence	Motivational + reminders + phone call option + motivational interviews	-

*HR: Hazard ratio;

†RD: Risk difference;

§Only p-value reported/no p-value reported;

¶Only adjusted results reported;

||OR; Odds Ratio

**Reported results from Clinicaltrials.gov/ISRCTN

††RR calculated based on 90% pharmacy refill (median), stated in table 4 in [45]

The mean age among trial participants ranged from 17 to 54 years. The primary target group for the text-message interventions were guardians of infant patients [19, 20, 24, 25, 36, 37, 39, 40, 46, 47] where the mother was the receiver of the text message intervention, whilst the remaining trials solely targeted men [22, 28] or women [32, 33, 35, 44, 46, 48–50] or both men and women [23, 26, 27, 29–31, 34, 38, 41, 43, 45]. All trials were set in Sub-Saharan Africa with vast majority being set in Kenya (n = 17) followed by South Africa (n = 4) (Fig 2). Participants were enrolled from a wide range of settings, e.g., local health facilities, public hospitals, HIV clinics, schools, and privately-owned pharmacies. The clinical areas in trials were HIV (n = 20), reproductive health (n = 4), antenatal health (n = 4), non-communicable diseases (n = 2), and malaria (n = 1). Ten trials reported that the text message content was developed based on health behavioral theories [19, 24, 28, 30, 32, 34, 35, 37–39], eight reported that the content had been pre-tested or developed in consultation with experts, clinical staff and/or potential participants [21, 23, 25, 26, 31, 39, 44, 46] and thirteen trials did not report either [20, 22, 27, 29, 33, 36, 41, 43, 45, 47–50]. The content of the two-way text messages varied from supportive, educational, and motivational messages to reminders and quizzes with a two-way component of either replying to the message via text message or phone call or by requesting phone calls or additional information from the sender (Table 1).

Fig 2

Overview of two-way text message trials in Africa according to clinical area and trial status.

When looking across all trials, ten trials included appointment attendance as an outcome [19, 20, 24–26, 28, 31, 39, 41, 44] and sixteen trials included medicine adherence as an outcome [21, 22, 25, 27, 29–31, 36, 37, 39, 41, 43–45, 48, 49]. Medicine adherence was measured in multiple ways, including self-reported adherence, pill box openings and clinical outcomes affected by adherence such as change in mean systolic blood pressure, proportion of patients with HbA1c < 7%, or suppressed HIV viral load [21, 25, 28, 29, 32, 34, 35, 38, 40, 42, 44]. Other trials had reproductive health knowledge [31, 39], adverse events [20], neonatal mortality [46], malnutrition [47], or unintended pregnancy as outcomes [33] (Table 1).

Descriptive analyses of trials not included in meta-analysis

Eight trials reported other outcomes or reported results on medicine adherence and/or appointment attendance in a format that did not allow us to include the data in meta-analyses. Therefore, these outcomes were only analyzed descriptively.

Overall, results varied among the trials. Four trials reported outcomes related to reproductive health; One trial found that two-way text messages increased postpartum contraceptive use compared to standard care (RR: 1.2; 95% CI: 1.0–1.5) [32]. One three-armed cluster randomized trial investigated the effect on reproductive knowledge among adolescent girls in Ghana. The two-way text message group had 24% (95% CI: 19%-28%) higher questionnaire scores than the control group and 13% (95% CI: 8–18%) higher than the one-way text message group [33]. Another trial from Kenya assessed the effect of one-way and two-way text messages on facility delivery, exclusive breastfeeding, and contraceptive use. Compared to standard care, there was no statistically significant difference between intervention groups in facility delivery and contraceptive use. However, both one-way and two-way text messages improved early breastfeeding compared to standard care (week 10: 79%control versus 93%one-way text versus 96% two-way text message) [25]. Further, one cluster randomized trial investigated the effect of two-way text messages that promoted contraception use and compared these to a control group receiving a sham text message (i.e. text messages with nutrition-focused-content) on the incidence of unintended pregnancies among sex workers in Kenya. They found no difference in the incidence of unintended pregnancy over 12 months of follow-up in the intervention group compared to the control group (hazard ratio (HR): 0.98; 95% CI: 0.7 to 1.4) [35].”

Four trials reported outcomes related to HIV. One trial investigated the effect of two-way text messages (intervention) compared to routine in-person visits (standard care) for post-operative follow-up among male circumcision patients Zimbabwe and did not find any significant difference in adverse events between groups (risk difference (RD): 1.04% (p = 0.32)) [22]. Another trial reported that a two-way text-message intervention did not improve HIV medicine adherence (measured as continuous outcome) in Uganda compared to controls (mean proportion of pills takentwo-way: 0.64 versus meancontrol: 0.67 (p = 0.15) [27]. Further, two trials assessed HIV medicine adherence and appointment attendance dichotomously but did not report the required patient and event data for be included in the meta-analyses. One of these trials was an unpublished trial set in South Africa (results recorded in clinicaltrials.gov) found a marginally effect in the two-way text message group compared to the control group (OR: 1.9; 95% CI: 0.9 to 4.2) [41]. The other trial was 3-arm trial on prevention of mother-to-child transmission of HIV in Kenya [44]. The trial found that two-way text messages did not improve on-time clinic appointment attendance (aRRtwo-way compared to control: 1.01; 95% CI: 0.98 to 1.04) and medicine adherence (measured as viral load non-suppression at any time) compared to a one-way text message group or a control group (aRRtwo-way compared to control: 0.80; 95% CI:0.52 to 1.23) [44].

A total of twelve trials were assessed for risk of bias as they were included in either the primary meta-analyses or sensitivity analyses; ten trials were included in the primary meta-analyses and two additional trials was included in a sensitivity or subgroup analysis. Three trials [26, 30, 43] were judged as overall low risk of bias, and nine as overall high risk of bias [19–21, 23, 24, 28, 29, 31, 45] (Fig 3, S5 File). Six out of twelve trials had unclear risk of selection bias due to their method of randomization or allocation concealment being inadequately described [19–21, 23, 28, 29]. Further, three trials had high risk of reporting bias as the information in trial registries were recorded after the trial had started or because the trial did not report all pre-specified outcomes [21, 28, 31].

Fig 3

Risk of bias assessment*.

*Empty cell: No risk of bias assessment—Attrition bias was not assessed for trials that only reported “appointment attendance” as an outcome as incomplete data is an integral part of this outcome.

Meta-analysis

Primary analyses

Sixteen published trials reported results on either medicine adherence and/or appointment attendance and of these, ten trials were included in the primary meta-analyses. Further, one cluster trial did not adjust for clustering and were excluded from the primary analysis on appointment attendance, though included in a subsequent sensitivity analysis [20], and one trial only compared two-way to one-way messages and was included in another sensitivity analysis [43]. One trial was excluded from the meta-analysis as it only reported continuous data on “medicine adherence” [32] and three trials did not report results in a format that allowed inclusion in meta-analysis [27, 41, 44]. Among the included trials, one trial had a 3-arm intervention, where two arms had two different types of two-way text messages and one control arm. In the meta-analysis, we used the results from the tailored two-way text message arm, where the messages targeted the participants self-reported stages of change according to a theory of change model [28]. Further, one trial reported data on both appointment attendance and medicine adherence and were included in both analyses [23].

Five trials (6627 participants) were included in our primary analysis on appointment attendance [19, 23, 24, 26, 28]. Overall, two-way text messages did not improve appointment attendance compared with standard care on appointment attendance, RR: 1.03; 95% CI: 0.95–1.12, I2 = 53% (Fig 4) (S3 File).

Fig 4

Effect of two-way text messages versus standard care on appointment attendance.

Six trials (3362 participants) were included in our primary analysis on medicine adherence [21, 23, 29–31, 45]. Overall, two-way text messages improved medicine adherence compared to standard care, RR: 1.14; 95% CI: 1.07–1.21, I2 = 8% (Fig 5).

Fig 5

Effect of two-way text messages versus standard care on medicine adherence.

Subgroup analyses

When comparing high risk to low risk of bias trials for appointment attendance, we found that high risk of bias trials reported a seemingly higher effect of two-way interventions compared to the single low risk of bias trial, though the subgroup difference was not statistically significant; RRhigh risk: 1.10; 95% CI: 0.95–1.28 versus RRlow risk: 0.98; 95% CI: 0.91–1.05 (interaction test p = 0.15) (Fig C in S4 File). Similarly, for medicine adherence, we found a seemingly increased intervention effect when comparing high risk of bias trials with the single low risk of bias trial, though the difference in effect was not statistically significant: RRhigh risk: 1.16, 95% CI: 1.08–1.23 versus RRlow risk: 1.02; 95% CI: 0.85–1.22 (interaction test: p = 0.21) (Fig D in S4 File).

When stratifying the primary analyses into different clinical areas, eight out of ten trials concerned HIV care [19, 21, 24, 26, 28, 30, 31, 45] whilst the remaining two trails concerned hypertension [23] and malaria [21]. When comparing the various clinical areas, there was no significant differences for neither appointment attendance (RRhypertension: 0.98, 95% CI 0.93 to 1.04 versus RRHIV: 1.10, 95% CI 0.95 to 1.29) or medicine adherence (RRhypertension: 1.18, 95% CI 1.02 to 1.36 versus RRHIV: 1.09, 95% CI 1.01 to 1.19 versus RRmalaria: 1.22, 95% CI 1.08 to 1.37) (Fig E, F in S4 File).

When comparing the different types of two-way text messages, we found a seemingly higher intervention effect on appointment attendance in trials where participants had the possibility to get in contact with a HCP through a phone call, though the difference in effect was not significant, RRHCP contact: 1.27; 95% CI: 0.87–1.83 versus RRHCP no contact: 0.99; 95% CI: 0.94–1.03 (interaction test p = 0.17) (Fig G in S4 File). For medicine adherence, it appeared that the text option was more effective than the phone call option, RRHCP contact: 1.03; 95% CI: 0.93–1.15 versus RRNo HCP contact: 1.19; 95% CI: 1.11–1.29 (interaction test: p = 0.03) (Fig H in S4 File).

Sensitivity analyses

One cluster trial only reported results unadjusted for clustering [20] and was included in a subsequent sensitivity analysis for appointment attendance. This analysis showed an increased intervention effect compared to our primary analysis but with an increase in heterogeneity, RR: 1.15; 95% CI: 0.99–1.33, I2 = 89% (Fig I in S4 File). When excluding both cluster trials in a sensitivity analysis for appointment attendance, the intervention effect of two-way text messages on appointment attendance did not change from our primary analysis, RRcluster excluded: 1.01; 95% CI 0.93–1.10, I2 = 50% (Fig J in S4 File). For medicine adherence, we found that when excluding the single cluster trial the intervention effect was still statistically significant, RRcluster excluded: 1.17; 95% CI: 1.09–1.25, I2 = 0% (Fig K in S4 File).

When using a fixed effect model in our sensitivity analysis for appointment attendance and medicine adherence, the estimate was quite similar to the results from the random effects model, RRfixed, appointment atttedance: 1.00; 95% CI: 0.96–1.05, I2 = 53% (Fig L in S4 File); RRfixed, medicine adherence: 1.14; 95% CI: 1.07–1.21, I2 = 8% (Fig M in S4 File).

Only one trial compared two-way messages to one-way messages for appointment attendance [23] and two trials compared two-way messages to one-way messages for medicine adherence [23, 43]. There was a tendency for one-way messages to be more effective than two-way messages for medicine adherence, RR: 0.89; 95% CI: 0.80–1.00, Fig B in S4 File).

Discussion

Main findings

In this systematic review and meta-analysis of two-way text message trials set in Africa, we identified a total of 31 trials; 20 published and 11 unpublished trials, of which 10 were eligible to be included in our primary meta-analyses as they reported binary data on appointment attendance or medicine adherence. Our analyses showed an effect of two-way messages on medicine adherence and an uncertain effect on appointment attendance, but most trials had high risk of bias. Our sensitivity and subgroup analyses did not change the overall conclusions of our primary analyses. However, when comparing the effect estimates for both appointment attendance and medicine adherence in relation to risk of bias, we found a seemingly higher intervention effect in trials with high risk of bias compared to low risk of bias trials though the subgroup differences were not statistically significant. Further, most trials included in the meta-analyses concerned HIV (n = 8/10), hence, the findings mainly relate to this field. Overall, our descriptive analysis showed that two-way text messages slightly improved some clinical outcomes including reproductive health knowledge, post-partum contraceptive use, and early breastfeeding.

Strength and limitations

To our knowledge, this is the first systematic review and meta-analysis that specifically investigates the effect of two-way text message interventions in an African setting, where the receiver can interact with the sender. Our analyses demonstrate that the effect of two-way text messages varies within different outcomes, clinical areas, and types of interventions and provides an in-depth perspective on effect of such mHealth interventions in an African context. A strength of this review is that it is based on a predefined protocol and that we conducted a comprehensive search that involved both published and unpublished trials. However, the review also has several limitations. A two-way text message intervention is a rather simple, low-cost intervention that overall appears effective for medicine adherence in an African context. However, participants must have access to a mobile phone and be able to read and send text messages or have family members who are able to. The illiteracy rate among adults in Sub-Saharan Africa was 35% in 2019, and such individuals may not benefit from the interventions to the same extent as literates [51]. Nevertheless, it is not given that people deemed illiterate in a classic reading and writing context cannot benefit from text message interventions as they may still be capable of understanding text messages. Furthermore, we only identified trials from 10 out of 54 countries in Africa, and the majority (17 of 31) were set in Kenya. This may limit the generalizability of our results to the whole of Africa as countries are heterogenous in relation to culture, health care systems, and socio-economic status. Furthermore, there are huge domestic differences within African countries. Future researchers should take demography and national geography into account to get a better understanding of who two-way text messages benefits the most and how to develop a mHealth tool that reaches as many as possible.

We included all types of health outcomes in this review and conducted meta-analyses on the outcomes “appointment attendance” and “medicine adherence” whilst other outcomes were described descriptively. The majority of included trials reported results on either medicine adherence or appointment attendance, yet this still only entailed that 10 trials were included our meta-analyses as the other were reported in a format that did not allow for inclusion. Further, eight out of ten trials included in the meta-analyses concerned “HIV”. This limits the generalizability of our findings beyond the field of HIV and reduce the potential impact of this review on public health practice. As evidence from African mHealth trials within different clinical areas is growing, we recommend that reviews like this one are regularly undertaken. This will allow for additional meta-analyses within different clinical areas, which will provide a more nuanced insight into the effect of text message trials across clinical fields and increase the impact on public health practice.

Only two out of ten trials included in our primary meta-analyses were judged to be in overall low risk of bias [26, 30] whilst eight had overall high risk of bias [19, 21, 23, 24, 28, 29, 31, 45]. Hence, the quality of trials was generally low, and this may have biased the effect estimates which was also somewhat indicated in our subgroup analysis. Further, we only focused on intervention effects in relation to two health outcomes of interest and we did not focus on other aspects of mHealth interventions, such as the participants perspectives. As suggested by the WHO guide “monitoring and evaluating digital health interventions” and the model for assessment of telemedical applications (MAST), an evaluation of not only the clinical effectiveness but also other dimensions, which the technology affects, may provide a more nuanced understanding of how technological interventions truly work [51, 52].

Finally, another limitation of this review is that two-way text messages can be considered a heterogenous type of intervention, and one could question whether it is fair to pool them as one type of intervention as done in our primary meta-analyses. It is plausible that an intervention where you achieve direct contact with a health care professional and an intervention where you have the option to answer a quiz, may have a different effect. We tried to address the issue of comparability in our subgroup analyses. However, for future reviews, researchers should aim to segregate the different types of two-way text message interventions to get a better understanding of the true impact of various types of two-way text messages.

Comparison of findings

HIV was the most prevalent clinical condition in in our meta-analyses, and in line with our findings a systematic review of one-way text message interventions in Africa found a fairly similar effect on HIV medicine adherence (RR: 1.18, 95% CI: 1.02 to 1.37) [5]. This indicates that different types of text message interventions–both those that allow for interaction with the sender and those that do not–may improve HIV medicine adherence in an African context. This may be due the interventions lowering the threshold for contact with health care personnel and clinics and simultaneously works as a supportive and reminder tool. A systematic review and meta-analysis from 2019, which investigated the effectiveness of mobile phone interventions and HIV medicine adherence, found a somewhat similar effect for text message interventions, though not significant and with high heterogeneity (RR: 1.25; 95% CI: 0.97 to 1.61; I2 = 74%) [53]. The trials in that review were conducted globally and included high-, low- and middle-income countries. It is plausible that the effect of text messages on HIV medicine adherence may differ across various settings, and more literature comparing the effect of mHealth interventions in high- and low-income countries are needed.

When looking at one-way versus two-way text messages on medicine adherence we found a tendency for one-way text messages to more effective than two-way, though the finding was not significant and only based on two trials (RR: 0.89, 95% CI: 0.8–1.0). In contrast, a systematic review and meta-analysis from 2015 found a higher intervention effect of two-way text messages compared to one-way text messages on medicine adherence (RRtwo-way: 1.23 95% CI: 1.12–1.35; RRone-way:1.04 95% CI: 0.97–1.11, p = 0.007), yet this review did not have any geographical restrictions [6]. When comparing our findings to the systematic review and meta-analysis that landscaped the effect of one-way text messages trials in an African setting [5], the other review found that one-way text messages improved appointment attendance (OR: 2.03; 95% CI: 1.40 to 2.95) though not medicine adherence (RR: 1.10; 95% CI: 0.98 to 1.23), whilst we found an effect of two-way text messages on medicine adherence and an uncertain effect on appointment attendance. Yet we had few data in this review, which may have influenced our results and the apparent differences in results could be due to lack of statistical power. Therefore, more large high-quality trials are needed, preferably three arm trials, to test if there is a difference in effect between one-way and two-way text message interventions in an African context.

WHO´s strategy on digital health 2020–2024 emphasizes the importance of building evidence, developing, and disseminating knowledge on eHealth in order to help policymakers and public servants to understand the power and complexity of eHealth. Their vision is that appropriate digital health technologies is a key component to achieving health for all [54]. This review has shown that one element of digital health—in the form of two-way text messages–have potential to improve medicine adherence in an African context. Yet our findings and our comparison of findings to other reviews also shows that text message interventions are complex, and the effect of such interventions is not uniform and easy to interpret. Hence, to better understand the clear effect of text message interventions, it is important that future studies clearly define the type of text message intervention they are undertaking and compare its effect to other similar interventions. We recommend “The mHealth evidence reporting and assessment checklist” as a reporting guide for future trials, to support replication and improve the quality of mobile phone health intervention trials [4].

Conclusions

Two-way text messages seemingly improve medicine adherence but has an uncertain effect on appointment attendance and other healthcare outcomes compared to standard care. We recommend taking this into account when planning future digital health strategies in Africa. To better estimate the true impact of two-way text message interventions, we recommend that future trials and reviews also consider other aspects of technological interventions, such as feasibility, usability and sustainability as recommended by the WHO.

PRISMA checklist.

(PDF)

Click here for additional data file.

Database search strategy.

(PDF)

Click here for additional data file.

Dataset for meta-analysis.

(PDF)

Click here for additional data file.

Subgroup and sensitivity analyses.

(PDF)

Click here for additional data file.

(PDF)

Click here for additional data file.

4 Aug 2021

PONE-D-21-10098

Two-way text message interventions and healthcare outcomes in Africa: Systematic review of randomized trials with meta-analysis

PLOS ONE

Dear Dr. Linde,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 18 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Michelle L. Munro-Kramer, PhD, CNM, FNP-BC

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

Thank you for this comprehensive manuscript. Both reviewers have provided detailed comments that should be addressed on resubmission. Please ensure that all reference numbers and references are accurate before resubmission.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

3. Please update search and analysis to include studies published since March 2020

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

5. "Please upload a new copy of Figures 1,2,4 and 5 as the detail is not clear. Please follow the link for more information: https://blogs.plos.org/plos/2019/06/looking-good-tips-for-creating-your-plos-figures-graphics/" https://blogs.plos.org/plos/2019/06/looking-good-tips-for-creating-your-plos-figures-graphics/.

6. We note that Figure 2 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figure 2 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: 1. Overall

1.1. I find this article to be a robust and interesting manuscript. However, there is a major methodological question of the definition of a “two-way” text messaging intervention versus a one-way intervention. Several of the interventions included in this review - that meet the eligibility criteria the authors used - I would not necessarily describe as two-way. For example, Ref 33 provides an option to participants to text in for more electronic information (delivered electronically and via automation) but this is only an option and participants may or may not use it; many of the other included studies could be similarly described. Other included studies have what could be considered a weak two-way option where a simple quiz may be occasionally used. I find this a major weakness of this manuscript and the authors need to speak to this more extensively throughout the paper.

1.2 Furthermore, I think it would be a stronger methodology to focus on two-way text messaging interventions that provide the option to speak with a health care provider, rather than the option to electronically access additional information which is common in the studies included in this review. Perhaps this could be used to describe the type of two-way text messaging intervention, and then use this variable to test for differences in effects based on automated replies versus reply from a live health care provider.

2.0 Comments on Specific Sections of the Manuscript

2.1 Introduction

2.1.1 Page 3 Line 67, please elaborate on what you mean by “various types of text message interventions.” Also you refer to “type of intervention” and use this as a comparison criteria / potential moderator throughout the paper, so please describe what is meant here and how your classification scheme was developed for this analysis.

2.1.2 Page 4 Line 75, please elaborate on “other health outcomes”; furthermore, please justify why appointment and medication adherence were the two focal outcomes when there are so many relevant and important healthcare outcomes to look at

2.1.3 The introduction would benefit from discussion of why two-way text messaging may be better than one-way. Adding theory and evidence behind this hypothesis would strengthen the introduction, which is very short.

2.2 Methods

2.2.1 Page 4 Eligibility Criteria, was any outcome eligible for inclusion as long as it met your other eligibility criteria? Please clarify and explain this better and discuss this in your Introduction

2.2.2 Eligibility Criteria, Please describe how (and with what search terms) two-way text message interventions were included in your search strategy; often you cannot tell whether an SMS intervention is 1 or 2 way unless a careful review of the intervention/report is undertaken. This should also be mentioned as a limitation of your review, in that it’s likely 2-way interventions were missed and others might describe two-way interventions differently than the authors did for this manuscript.

2.2.3 Page 5 Line 122, were all trials assessed for bias by two people - so 100% double coding for risk of bias? This is not clear from the text.

2.3 Results

2.3.1 Page 12 Descriptive Analyses, i think this paragraph would benefit from a revised subtitle and explanation of what is being presented in this section. Initially it seemed like it would be trials not included in other analyses, but this is not the case. Please clarify and revise.

2.4 Discussion

2.4.1 Page 17 Line 351, I do not think that it is accurate to state that people who are deemed “illiterate” in some contexts would not benefit from text messages and interventions; we have found that people can be literate in some information contexts while “illiterate” in others - and that even people who are deemed “illiterate” in classic reading/writing literacy can have a high capability of understanding text messages.

2.4.2 Page 20 Line 410-411, I would suggest including a reference to the mERA checklist (BMJ 2016;352:i1174 http://dx.doi.org/10.1136/bmj.i1174) - this checklist contributes to standardization in reporting/comparisons between interventions, and ensuring that essential aspects of mHealth interventions are attended to.

2.5 Overall more description and discussion of two-way text messaging is needed throughout the paper.

Reviewer #2: This is a timely review of an important tool in the mobile health toolbox: two-way texting interventions for improved health outcomes. The paper is well written and well organized, as is expected from this highly qualified author group. It is also critical that these reviews, although not novel, are completed routinely as the evidence is growing and the technology changing rapidly.

However, the included studies are few and the outcomes limited, reducing the potential impact of this review on public health practice. There are many reviews of texting interventions – as the authors note and cite – including many focused on HIV, as is the majority of this one. Could more health topics be included (malnutrition, diabetes, vaccinations, etc)? Could the study assessment criteria extend beyond a more simple assessment of high or low bias to a more comprehensive and nuanced review of the study rigor?

Furthermore, the call for more better evidence has been made for over a decade. The authors know and could give more specific suggestions on what is needed. For example, the authors could strengthen the paper by making more detailed suggestions for what could create an improved evidence base, potentially including focal areas in the WHO guide, “Monitoring and evaluating digital health interventions: A practical guide to conducting research and assessment” (https://www.who.int/reproductivehealth/publications/mhealth/digital-health-interventions/en/ ). Whether referring to that guide or other evaluation criteria, the authors could make a larger contribution to the digital health community if the review would focus more on the weaknesses of the included studies (or the excluded studies), moving past a more generic call for “more high quality trials.” That would exponentially increase the impact of this paper.

Other comments:

1. The words used for searching seem too exclusive. Other than, “appointment attendance” and “medicine adherence”, what other words were used? Strings like “visit attendance,” “treatment adherence,” and “linkage to care” might have also brought in trials. Were these key words also considered? Texting, cell phones or digital health may have brought up other studies.

2. Lines 49-51 and 51-53: break into 2 sentences for clarity.

3. Many sentences in the intro would benefit from a grammatical review.

4. Line 85: males and females separate or also together?

5. Line 150: what about for non dichotomous outcomes, i.e. % of on-time visits/year, for example?

6. 154: what is I2 ? What is it or spell out.

7. Line 185: This does not make sense. How can infants be the target group for a mHealth innovation? Does this mean the guardian or parent of an infant?

8. Line 213: This specific RCT was testing whether post-operative follow-up by two-way text messages was as safe as in-person visits for follow-up to identify and report adverse events. The intent was to use two-way messaging as a form of telehealth – comparing adverse events between the intervention arm as compared to standard care – trying to reduce workload of nurses. Two-way texting actually identified /more/ adverse events (a positive for quality care) as the men were provided with reassurance on wound care and encouraged to return to care if they had a problem. Two-way texting was not non-inferior (the texting actually improved reporting), and the lack of significant difference between the arms is a positive outcome for potential workload reduction and quality care. The original sentence starting on line 213, “Another trial investigated the effect of two-way text message intervention on circumcision harms in Zimbabwe and did not find any differences compared with standard care (risk difference (RD): 1.04% (p=0.32) [20],” is incorrect and misleading. Texting was not related to harms but on identifying potential harms. It should be replaced with something like, “Another trial investigated the effect of two-way text messages (intervention) as compared to routine in-person reviews (standard care) for post-operative follow-up among male circumcision patients in Zimbabwe and did not find any significant difference in adverse events between arms (risk difference (RD): 1.04% (p=0.32) [20].

However, although this study outcome [20] was not among those studies included in the primary outcomes of the overall meta-analysis, misunderstanding these study outcomes does give pause as to the veracity of the other findings reported in the paper. I am not going to review the content of each included paper, but the authors should be confident that the content of the meta-analysis and the study summaries are correct.

9. Lines 225-227, what is the outcome of the study? The sentence tells only what the study investigated.

10. Line 289: so all HIV studies were compared to the one hypertension study? What would a significant finding tell the reader anyway if almost all trials were for HIV?

11. Line 298. What is meant by the effect of the call back versus text back option mean? Is this comparing whether the clients sent back an SMS or called back to confirm attendance? To confirm attendance intent? Please clarify.

12. Line 300: please clarify again what is meant by call back vs. SMS. This is critical to readers’ understanding of what these results may mean for future interventions. Does this mean giving clients a chance to call back was more effective for medicine adherence?

13. Lines 334-338 lack clarity, “compared to standard care, two-way text messages slightly improved diabetes control [32], reproductive health knowledge [31, 39] post-partum contraceptive use [30], early breastfeeding [23] though not adverse events [22], unintended pregnancy [33] or HIV medicine adherence [42].

I /think/ that citation [22] here should be [20]. First, this makes me request that you review all your citations within the paper. Second, if [22] is meant to be [20], please rephrase to, ….”early breastfeeding [23] and adverse event ascertainment [20] although not unintended pregnancy [33] or HIV medicine adherence [42].” However, isn’t this metareview about HIV medicine adherence? Why wasn’t [42] included above?

14. Line 359: the restricted outcomes to only 2 (appointment attendance or medicine adherence) is also a large limitation. What other outcomes are the intended impact of two-way texting interventions? Testing uptake? Linkage to care? Self-monitoring? Smoking or alcohol harms reduction? Intimate partner violence? Improved nutrition or malnutrition identification?

15. Line 365: text back options are not well explained. What does this mean? I thought that the previous lines (298-302) noted the impact of the call back? This is unclear, muddling a potential impact of this type of review on actual intervention development.

16. Line 402: Spell out WHO the first time

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

4 Jan 2022

Journal Requirements

Comment Response

#1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Response #1: Thank you for pinpointing this. We have gone through all documents to ensure they are aligned with PLOS ONE’s requirements.

#2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Response #2: We have reviewed our reference list and updated it with new references, which were found during our updated literature search.

3. Please update search and analysis to include studies published since March 2020

Response #3: Thank you for this comment. We have updated search and included three new trials to the review:

Abiodun O, Ladi-Akinyemi B, Olu-Abiodun O, Sotunsa J, Bamidele F, Adepoju A, et al. A Single-Blind, Parallel Design RCT to Assess the Effectiveness of SMS Reminders in Improving ART Adherence Among Adolescents Living with HIV (STARTA Trial). J Adolesc Health. 2021;68:728-36. doi: 10.1016/j.jadohealth.2020.11.016.

Kinuthia J, Ronen K, Unger JA, Jiang W, Matemo D, Perrier T, et al. SMS messaging to improve retention and viral suppression in prevention of mother-to-child HIV transmission (PMTCT) programs in Kenya: A 3-arm randomized clinical trial. PLoS Med. 2021;18:e1003650. doi: 10.1371/journal.pmed.1003650.

Sumari-de Boer IM, Ngowi KM, Sonda TB, Pima FM, Masika Bpharm L, Sprangers MAG, et al. Effect of Digital Adherence Tools on Adherence to Antiretroviral Treatment Among Adults Living With HIV in Kilimanjaro, Tanzania: A Randomized Controlled Trial. J Acquir Immune Defic Syndr. 2021;87:1136–44. doi: 10.1097/QAI.0000000000002695,

#4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

Response #4: Thank you for pinpointing this error. There is no restriction for data availability. We have attached an anonymized data set as “Supporting Information File 6” and updated our data availability statement.

#5. Please upload a new copy of Figures 1,2,4 and 5 as the detail is not clear

Response #5: We have uploaded new copies of figures 1-5.

#6. We note that Figure 2 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). …

Response #6: Thank you for pinpointing this issue. We have contacted MapChart and been granted permission to use the map and have attached the signed permission form as documentationBelow the figure it now states:

“Reprinted from mapchart.net under a CC BY license, with permission from Minas Giannekas, original copyright 2021.”

Reviewer 1

Comment

#1.1. I find this article to be a robust and interesting manuscript. However, there is a major methodological question of the definition of a “two-way” text messaging intervention versus a one-way intervention. Several of the interventions included in this review - that meet the eligibility criteria the authors used - I would not necessarily describe as two-way. For example, Ref 33 provides an option to participants to text in for more electronic information (delivered electronically and via automation) but this is only an option and participants may or may not use it; many of the other included studies could be similarly described. Other included studies have what could be considered a weak two-way option where a simple quiz may be occasionally used. I find this a major weakness of this manuscript and the authors need to speak to this more extensively throughout the paper.

Response #1.1: We thank the reviewer for the constructive comment. We agree with the reviewer that we have a broad definition of two-way messages, which could clutter the true effect of the stronger two-way text message options. Yet we still find that the distinction between having the option to interact with the sender through two-way options compared to only receiving one-directional messages is important clarify. The reasons for this broad definition was that 1) we wanted to provide an overview of the field 2) we were afraid that more narrow inclusion criteria could results in loss of power in our meta-analyses. We agree that we have not clarified the limitation of using broad inclusion criteria clearly in the manuscript and added the following paragraph to the discussion:

“Finally, another limitation of this review is that two-way text messages can be considered a heterogenous type of intervention, and one could question whether it is fair to pool them as one type of intervention as done in our primary meta-analyses. It is plausible that an intervention where you achieve direct contact with a health care professional and an intervention where you have the option to answer a quiz, may have a different effect. We tried to address the issue of comparability in our subgroup analyses. However, for future reviews, researchers should aim to segregate the different types of two-way text message interventions to get a better understanding of the true impact of various types of two-way text messages.”

#1.2 Furthermore, I think it would be a stronger methodology to focus on two-way text messaging interventions that provide the option to speak with a health care provider, rather than the option to electronically access additional information which is common in the studies included in this review. Perhaps this could be used to describe the type of two-way text messaging intervention, and then use this variable to test for differences in effects based on automated replies versus reply from a live health care provider.

Response #1.2: We thank the reviewer for this suggestion. We have rephrased the description of our subgroup analyses; however, we have not changed our meta-analyses as they were conducted according to our pre-defined protocol. The methodology section now states:

“Subgroup analyses were performed comparing overall low risk of bias trials with high risk of bias trials, clinical areas and types of interventions, e.g. if the receiver could interact with a health care professional (HCP) through a phone call or text message.”

#2.1: Introduction: 2.1.1 Page 3 Line 67, please elaborate on what you mean by “various types of text message interventions.” Also you refer to “type of intervention” and use this as a comparison criteria / potential moderator throughout the paper, so please describe what is meant here and how your classification scheme was developed for this analysis.

Response #2.1.1: We thank the reviewer for this comment. Further down in our introduction we state what we mean by the different interventions. We hope this answers the reviewer’s question:

“Further, a series of Cochrane reviews published between 2012 and 2017 assessed the effect of various text message interventions on different health issues [7-12], but the reviews did not differentiate between one-way and two-way text message interventions in their analyses, hence, the effect of one-way versus two-way text messages interventions was unclear.

As stated in our response to comment #1.2, we have clarified what we mean by “type of intervention”, and stated in our methodology section that it means whether the receiver could interact with a health care professional.

#2.1.2 Page 4 Line 75, please elaborate on “other health outcomes”; furthermore, please justify why appointment and medication adherence were the two focal outcomes when there are so many relevant and important healthcare outcomes to look at

Response #2.12: We thank the reviewer for this comment and have elaborated on our choice of outcomes. The following is now stated in the introduction:

“Research shows that linkage to care is a challenge in many resource-limited settings [13], and it is plausible that two-way text messages – which allow for interactive communication with health care providers – may be a more effective health care tool than one-way messages. Appointment attendance and medicine adherence are outcomes frequently used to measure linkage to care across clinical settings, yet the effect of two-way messages on these outcomes in African context is unclear. Therefore, this systematic review aims to landscape randomized trials in Africa involving two-way text message interventions compared to standard care or one-way text messages and analyze their effect on appointment attendance and medicine adherence. By conducting a meta-analysis on the same outcomes as the previous systematic review of one-way messages in Africa [5], this review will provide an even greater understanding of how various types of text message interventions work in an African context. Further, in line with the previous review, we also describe other health outcomes to provide a full overview of two-way message trials set in Africa.”

#2.1.3 The introduction would benefit from discussion of why two-way text messaging may be better than one-way. Adding theory and evidence behind this hypothesis would strengthen the introduction, which is very short.

Response #2.1.3: Thank for this suggestion. We have taken this comment into consideration and updated our introduction. It now states:

“A systematic review and meta-analysis from 2015 – which included trials conducted in various settings – suggested that interactive communication between a client and a healthcare provider may lead to greater support, motivation, and safety for the patient. The meta-analysis found a higher intervention effect of two-way messages compared to one-way text messages on medicine adherence [6].”

#2.2 Methods: 2.2.1 Page 4 Eligibility Criteria, was any outcome eligible for inclusion as long as it met your other eligibility criteria? Please clarify and explain this better and discuss this in your Introduction

Response #2.2.1: Trials reporting all health care outcomes were included if they met all other eligibility criteria. This has now been clarified in the manuscript. As stated in our response to comment #2.1.2, the following sentence has been added to the introduction:

“…we also describe other health outcomes to provide a full overview of the effect of two-way messages in an African context.”

#2.2.2 Eligibility Criteria, Please describe how (and with what search terms) two-way text message interventions were included in your search strategy; often you cannot tell whether an SMS intervention is 1 or 2 way unless a careful review of the intervention/report is undertaken. This should also be mentioned as a limitation of your review, in that it’s likely 2-way interventions were missed and others might describe two-way interventions differently than the authors did for this manuscript.

Response #2.2.2: We thank the reviewer for pinpointing this issue. As described in the introduction we defined two-way text messages as a text message the receiver could react or respond to, with the use of a mobile phone, in any way. We have now also included this definition in our methodology section our “Eligibility Criteria”, which now states,

“Trials comparing two-way text message interventions with standard care and/or one-way text messages were included. We defined two-way text messages as a text message the receiver could react or respond to, with the use of a mobile phone, in any way.”

First, as can be seen from our full search strings in the S1 file, our search strategy was not only focused on “two-way messages” but on text messaging in general, e.g. they also contained search terms that were broad terms such as “SMS” or “mobile phone”. Second, in the title and abstract screening we only excluded trials that obviously stated interventions consisting of one-way messages according to our definition. All other text message trials, where the type of intervention was uncertain, were read in full text to clarify the type of intervention. Based on the reviewer’s comments we have now clarified this and the “Information sources and search strategy” section now states:

“The search string included search terms such as “randomized controlled trial”, “Africa”, “SMS”, “mobile phone”, “two-way text messages” and “two-way SMS” (S1 File).”

AND

“In the title-abstract screening, trials were only excluded if the title and abstract obviously stated that interventions consisted of one-way messages as according to our definition.”

#2.2.3 Page 5 Line 122, were all trials assessed for bias by two people - so 100% double coding for risk of bias? This is not clear from the text.

Response #2.2.3: All trials were assessed for risk of bias independently by two authors. We have added “all trials” to the sentence to clarify this,

“Two authors (ESØ, LSL) independently assessed all trials that were eligible for meta-analysis (trials reporting the outcomes “appointment attendance” and/or “medicine adherence”) for risk of bias using the Cochrane Risk of Bias Tool [15]. If consensus could not be reached, and arbiter (DSL) made the final decision.”

#2.3 Results: 2.3.1: Page 12 Descriptive Analyses, i think this paragraph would benefit from a revised subtitle and explanation of what is being presented in this section. Initially it seemed like it would be trials not included in other analyses, but this is not the case. Please clarify and revise.

#Response 2.3.1: We thank the reviewer for this suggestion and agree that it needs revision. We have revised the heading and it now states,

“Descriptive analyses of trials not included in meta-analysis”

Further, we have added the following sentence to the descriptive analyses section,

“Eight trials reported other outcomes or reported results on medicine adherence and/or appointment attendance in a format that did not allow us to include the data in meta-analyses. Therefore, these outcomes were only analyzed descriptively.”

#2.4 Discussion: 2.4.1: Page 17 Line 351, I do not think that it is accurate to state that people who are deemed “illiterate” in some contexts would not benefit from text messages and interventions; we have found that people can be literate in some information contexts while “illiterate” in others - and that even people who are deemed “illiterate” in classic reading/writing literacy can have a high capability of understanding text messages.

Response #2.4.1: We thank the reviewer for this important comment. We have revised the discussion so it now states:

“The illiteracy rate among adults in Sub-Saharan Africa was 35% in 2019, and such individuals may not benefit from the interventions to the same extent as literates [51]. Nevertheless, it is not given that people deemed illiterate in a classic reading and writing context cannot benefit from text message interventions as they may still be capable of understanding text messages.”

#2.4.2 Page 20 Line 410-411, I would suggest including a reference to the mERA checklist (BMJ 2016;352:i1174 http://dx.doi.org/10.1136/bmj.i1174) - this checklist contributes to standardization in reporting/comparisons between interventions, and ensuring that essential aspects of mHealth interventions are attended to.

Response #2.4.2: We thank the reviewer for this suggestion. We have included the reference in our manuscript. The discussion now states:

“We recommend “The mHealth evidence reporting and assessment checklist” as a reporting guide for future trials, to support replication and improve the quality of mobile phone health intervention trials [55].”

#2.5 Overall more description and discussion of two-way text messaging is needed throughout the paper.

Response #2.5: We thank the reviewer for this comment. We have revised the manuscript thoroughly according to the reviewers’ constructive comments and two-way messaging is both described and discussed more in-depth in the revised manuscript.

Reviewer 2

Comment

#A. This is a timely review of an important tool in the mobile health toolbox: two-way texting interventions for improved health outcomes. The paper is well written and well organized, as is expected from this highly qualified author group. It is also critical that these reviews, although not novel, are completed routinely as the evidence is growing and the technology changing rapidly.

However, the included studies are few and the outcomes limited, reducing the potential impact of this review on public health practice. There are many reviews of texting interventions – as the authors note and cite – including many focused on HIV, as is the majority of this one. Could more health topics be included (malnutrition, diabetes, vaccinations, etc)?

Response #A: We thank the reviewer very much for the supportive comment. We did not exclude any health outcomes in the paper but included all two-way text message trials set in Africa. Yet, we only conducted meta-analyses on trials that had the outcomes “medicine adherence” and “appointment attendance” (dichotomously measured), and trials that did not have such outcomes were assessed descriptively. The trials eligible for meta-analyses concerned “HIV”, “malaria” and “hypertension”. Effects of the text messages on these health topics were assessed in subgroup analyses according to our protocol. Health topics such as “diabetes” or “reproductive health knowledge” were analyzed descriptively (table 1).

We agree that the number of two-way text message trials set in Africa are limited, which reduces the potential impact of this review on public health practice. We have revised the discussion, so it now states:

“We included all types of health outcomes in this review and conducted meta-analyses on the outcomes “appointment attendance” and “medicine adherence” whilst other outcomes were described descriptively. The majority of included trials reported results on either medicine adherence or appointment attendance, yet this still only entailed that 10 trials were included our meta-analyses as the other were reported in a format that did not allow for inclusion. Further, eight out of ten trials included in the meta-analyses concerned “HIV”. This limits the generalizability of our findings beyond the field of HIV and reduce the potential impact of this review on public health practice.”

#B. Could the study assessment criteria extend beyond a more simple assessment of high or low bias to a more comprehensive and nuanced review of the study rigor? Furthermore, the call for more better evidence has been made for over a decade. The authors know and could give more specific suggestions on what is needed. For example, the authors could strengthen the paper by making more detailed suggestions for what could create an improved evidence base, potentially including focal areas in the WHO guide, “Monitoring and evaluating digital health interventions: A practical guide to conducting research and assessment” (https://www.who.int/reproductivehealth/publications/mhealth/digital-health-interventions/en/ ). Whether referring to that guide or other evaluation criteria, the authors could make a larger contribution to the digital health community if the review would focus more on the weaknesses of the included studies (or the excluded studies), moving past a more generic call for “more high quality trials.” That would exponentially increase the impact of this paper.

Response #B: We thank the reviewer for these constructive comments. We agree that the recommendation of better evidence is outdated and have now deleted it from our recommendations and have revised the discussion in line with the reviewer’s comments:

“Further, we only focused on intervention effects in relation to two health outcomes of interest and we did not focus on other aspects of mHealth interventions, such as the participants perspectives. As suggested by the WHO guide “monitoring and evaluating digital health interventions” and the model for assessment of telemedical applications (MAST), an evaluation of not only the clinical effectiveness but also other dimensions, which the technology affects, may provide a more nuanced understanding of how technological interventions truly work [51, 52].”

The conclusion now states:

“Two-way text messages seemingly improve medicine adherence but has an uncertain effect on appointment attendance and other healthcare outcomes compared to standard care. We recommend taking this into account when planning future digital health strategies in Africa. To better estimate the true impact of two-way text message interventions, we recommend that future trials and reviews also consider other aspects of technological interventions, such as feasibility, usability and sustainability as recommended by the WHO.”

#1. The words used for searching seem too exclusive. Other than, “appointment attendance” and “medicine adherence”, what other words were used? Strings like “visit attendance,” “treatment adherence,” and “linkage to care” might have also brought in trials. Were these key words also considered? Texting, cell phones or digital health may have brought up other studies.

Response #1: We thank the reviewer for this comment. In our review, we included all health outcomes, hence, our search strings did not have any limitations in relation to appointment attendance and medicine adherence, which can be seen in the S1 file. We have clarified this further under “eligibility criteria”. It now states:

“Trials reporting all health care outcomes were included as long as it met all other eligibility criteria.”

#2. Lines 49-51 and 51-53: break into 2 sentences for clarity.

Response #2: We thank the reviewer for this direct point. We have changed the sentence. It now states:

“For the past 30 years, information and communication technologies have transformed the world [1]. In recent years mobile phone access has expanded immensely across Africa and other low- and middle-income countries (LMICs). According to the World Health Organization (WHO), more people have access to a mobile phone than to clean running water in sub-Saharan Africa [2]. In 2017, approximately 8 out of 10 people in sub-Saharan Africa owned a mobile phone with an increasing amount of these being smartphones [3].”

#3. Many sentences in the intro would benefit from a grammatical review.

Response #3: We thank the reviewer for this point. We have now gone through the manuscript and corrected grammatical errors.

#4. Line 85: males and females separate or also together?

Response #4:We thank the reviewer for pinpointing. We have now clarified it in the manuscript. It now states:

“We included randomized trials of male and female healthcare clients, partners, or guardians for healthcare clients, e.g., parents for child patients.”

#5. Line 150: what about for non dichotomous outcomes, i.e. % of on-time visits/year, for example?

Response #5: We thank the reviewer for this comment. The one trial that reported continuous outcomes on medicine adherence and appointment attendance was analyzed descriptively. We have clarified this in the manuscript. The methodology section now states,

“Trials that reported dichotomous outcome data in a format that did not allow inclusion in meta-analysis or trials that only reported continuous outcomes were analyzed descriptively”

#6. 154: what is I2 ? What is it or spell out.

Response #6: We thank the reviewer for this point. We have now made a reference to the methods behind the I2 statistic:

“Heterogeneity were assessed using I2 [15]”

#7. Line 185: This does not make sense. How can infants be the target group for a mHealth innovation? Does this mean the guardian or parent of an infant?

Response #7: We thank the reviewer for this question. We have rephrased the sentence. It now states,

“The primary target group for the text-message interventions were guardians of infant patients [19, 20, 24, 25, 36, 37, 39, 40, 46, 47] where the mother was the receiver of the text message intervention…”

#8. Line 213: This specific RCT was testing whether post-operative follow-up by two-way text messages was as safe as in-person visits for follow-up to identify and report adverse events. The intent was to use two-way messaging as a form of telehealth – comparing adverse events between the intervention arm as compared to standard care – trying to reduce workload of nurses. Two-way texting actually identified /more/ adverse events (a positive for quality care) as the men were provided with reassurance on wound care and encouraged to return to care if they had a problem. Two-way texting was not non-inferior (the texting actually improved reporting), and the lack of significant difference between the arms is a positive outcome for potential workload reduction and quality care. The original sentence starting on line 213, “Another trial investigated the effect of two-way text message intervention on circumcision harms in Zimbabwe and did not find any differences compared with standard care (risk difference (RD): 1.04% (p=0.32) [20],” is incorrect and misleading. Texting was not related to harms but on identifying potential harms. It should be replaced with something like, “Another trial investigated the effect of two-way text messages (intervention) as compared to routine in-person reviews (standard care) for post-operative follow-up among male circumcision patients in Zimbabwe and did not find any significant difference in adverse events between arms (risk difference (RD): 1.04% (p=0.32) [20]. However, although this study outcome [20] was not among those studies included in the primary outcomes of the overall meta-analysis, misunderstanding these study outcomes does give pause as to the veracity of the other findings reported in the paper. I am not going to review the content of each included paper, but the authors should be confident that the content of the meta-analysis and the study summaries are correct.

Response #8: We thank the reviewer for pinpointing this error and apologize for not being clear in our conclusions of this trial. We have revised the paragraph as suggested,

“One trial investigated the effect of two-way text messages (intervention) compared to routine in-person visits (standard care) for post-operative follow-up among male circumcision patients Zimbabwe and did not find any significant difference in adverse events between groups (risk difference (RD): 1.04% (p=0.32)) [22].”

#9. Lines 225-227, what is the outcome of the study? The sentence tells only what the study investigated.

Response #9: We thank the reviewer for this point. We now have described the outcome more clearly,

“Further, one cluster randomized trial investigated the effect of two-way text messages that promoted contraception use and compared these to a control group receiving a sham text message (i.e. text messages with nutrition-focused-content) on the incidence of unintended pregnancies among sex workers in Kenya. They found no difference in the incidence of unintended pregnancy over 12 months of follow-up in the intervention group compared to the control group (hazard ratio (HR): 0.98; 95% CI: 0.7 to 1.4) [35].”

#10. Line 289: so all HIV studies were compared to the one hypertension study? What would a significant finding tell the reader anyway if almost all trials were for HIV?

Response #10: We thank the reviewer for this valid point. We conducted the subgroup analyses according to our protocol where we stratified the analysis across various clinical areas. The majority of trials included in our meta-analysis were HIV trials, which limits the generalizability of our findings. We have clarified this in our results and discussion sections.

Results:

“When stratifying the primary analyses into different clinical areas, 8 out of ten trials concerned HIV care [19, 21, 24, 26, 28, 30, 31, 45] whilst the remaining two trails concerned hypertension [23] and malaria [21].”

AND

Discussion:

“Further, eight out of ten trials included in the meta-analyses concerned “HIV”. This limits the generalizability of our findings beyond the field of HIV and reduce the potential impact of this review on public health practice.”

#11. Line 298. What is meant by the effect of the call back versus text back option mean? Is this comparing whether the clients sent back an SMS or called back to confirm attendance? To confirm attendance intent? Please clarify.

12. Line 300: please clarify again what is meant by call back vs. SMS. This is critical to readers’ understanding of what these results may mean for future interventions. Does this mean giving clients a chance to call back was more effective for medicine adherence?

Response #11+12: We thank the reviewer for these comments. It is clear to us that our classification of two-way text messages into “call back” and “text back” options causes confusion, as also pointed out by reviewer #1. The description now states,

“Subgroup analyses were performed comparing overall low risk of bias trials with high risk of bias trials, clinical areas and types of interventions, e.g. if the receiver could interact with a health care professional (HCP) through a phone call or text message”

#13. Lines 334-338 lack clarity, “compared to standard care, two-way text messages slightly improved diabetes control [32], reproductive health knowledge [31, 39] post-partum contraceptive use [30], early breastfeeding [23] though not adverse events [22], unintended pregnancy [33] or HIV medicine adherence [42]. I think that citation [22] here should be [20]. First, this makes me request that you review all your citations within the paper. Second, if [22] is meant to be [20], please rephrase to, ….”early breastfeeding [23] and adverse event ascertainment [20] although not unintended pregnancy [33] or HIV medicine adherence [42].” However, isn’t this metareview about HIV medicine adherence? Why wasn’t [42] included above?

Response #13: We thank the reviewer for pinpointing this error. We have gone through all the references and made sure that they are correct. It is correct that reference 42 (now 41) included the relevant outcome. However, the trial results were reported in a format that did not allow inclusion in our meta-analysis. This is also stated under “Meta-Analysis: Primary analyses”:

“One trial was excluded from the meta-analysis as it only reported continuous data on “medicine adherence” [32] and three trials did not report results in a format that allowed inclusion in meta-analysis [27, 41, 44].”

#14. Line 359: the restricted outcomes to only 2 (appointment attendance or medicine adherence) is also a large limitation. What other outcomes are the intended impact of two-way texting interventions? Testing uptake? Linkage to care? Self-monitoring? Smoking or alcohol harms reduction? Intimate partner violence? Improved nutrition or malnutrition identification?

Response #14: We thank the reviewer for this point. The issue was also raised by reviewer #1 (comment 2.1.2). Please see our response to reviewer #1’s comment.

#15. Line 365: text back options are not well explained. What does this mean? I thought that the previous lines (298-302) noted the impact of the call back? This is unclear, muddling a potential impact of this type of review on actual intervention development.

Response #15: We thank the reviewer for this comment, which is well in line with the reviewer’s previous comments #11 and #12. We have now rephrased our description of these messages as described above.

#16. Line 402: Spell out WHO the first time

Response #16: This has now been done.

Submitted filename: Response note.docx

Click here for additional data file.

28 Mar 2022

Two-way text message interventions and healthcare outcome in Africa: Systematic review of randomized trials with meta-analyses on appointment attendance and medicine adherence

PONE-D-21-10098R1

Dear Dr. Linde,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Michelle L. Munro-Kramer, PhD, CNM, FNP-BC

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Thank you for addressing all previous comments. Congratulations on the acceptance of this manuscript.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: The paper has been well revised. I do not have any additional comments to the authors. Now, meeting character counts.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No