Text messaging interventions are associated with reductions in HbA1c among patients with diabetes: a systematic review and meta-analysis

Abstract

Introduction

Achieving optimal glycemic control remains challenging for many patients with diabetes. Text message-based interventions offer a scalable approach to enhance management. This systematic review and meta-analysis evaluated the impact of texting interventions on glycemic control in adults with diabetes.

Research design and methods

We searched EMBASE, PubMed, and Cochrane CENTRAL for randomized controlled trials comparing texting interventions to standard care in high-income countries. The primary outcome was the between-group difference in hemoglobin A1c (HbA1c) change from baseline. Risk of bias and overall quality of evidence were assessed using the Cochrane and Grading of Recommendations Assessment, Development, and Evaluation tools respectively. Results were pooled using an inverse variance random-effects model. Heterogeneity was evaluated using the I² statistic.

Results

Over 3 months of follow-up (14 trials, n=1,460 intervention, n=1,487 control), texting interventions were associated with a 0.29-unit greater reduction in percent HbA1c over control (95% CI 0.14 to 0.45, p=0.0001, I²=57%). At 6 months (20 trials, n=2,332 intervention, n=2,371 control), texting was associated with 0.19-unit greater HbA1c reduction (95% CI 0.07 to 0.30, p=0.001 I²=45%). At 12 months (seven trials, n=2,038), there was a non-significant benefit associated with texting. Among studies with a mean baseline HbA1c ≥8.6%, texting was associated with 0.48- and 0.36-unit greater HbA1c reductions at 3 (p=0.004) and 6 (p=0.004) months, respectively. Subgroups were not significantly different.

Conclusion

Text messaging interventions are associated with modest improvements in glycemic control over 3–6 months, particularly in patients with poorer baseline HbA1c. These effects may be meaningful at scale and support texting as a potential adjunct to routine diabetes care. Benefits appear to diminish by 12 months, underscoring the need for high-quality trials focused on long-term impact and intervention optimization.

PROSPERO registration number

CRD42023416462.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Texting has been studied as a scalable adjunct to usual diabetes care. Prior reviews with relatively few randomized controlled trials hint at modest hemoglobin A1c gains over usual care but with heterogeneous results and uncertain durability.

WHAT THIS STUDY ADDS

• Our meta-analysis of 29 trials indicates that texting intervention benefits are most evident in patients with relatively poorer baseline control and diminish by 1 year. Texting-only interventions may be less impactful compared with multicomponent interventions.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These results support targeted, short-term use of texting as an adjunct to usual care—particularly for patients with poorer baseline control—while motivating more high-quality trials to do the following: define the beneficial components of texting (eg, higher message frequency, interactivity), optimize multicomponent approaches, and test maintenance strategies to sustain improvements.

Introduction

Diabetes is a leading cause of death in developed countries.¹ Despite substantial advances in pharmacotherapy and educational initiatives, many individuals with diabetes remain unable to achieve optimal glycemic targets. Barriers include socioeconomic determinants of health and fragmented care delivery. As emphasized in the 2025 American Diabetes Association Standards of Care, episodic clinical visits alone are often insufficient to achieve sustained behavioral change or glycemic control.² Mobile health technologies, particularly text messaging interventions, are emerging as promising tools to bridge care gaps and complement in-person evaluations due to their scalability and asynchronous nature.³ These interventions leverage existing cellular network infrastructure and take advantage of a behavior that is deeply ingrained in daily life across socioeconomic groups.⁴

There remains a gap in knowledge regarding whether text messaging interventions can help patients with diabetes achieve better glycemic control. Current systematic reviews on this topic are limited by heterogeneous study populations, variable follow-up durations, and an incomplete capture of the contemporary evidence base. Many reviews also included a relatively small number of studies and did not investigate heterogeneity or outcomes across key subgroups.

Accordingly, we conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) evaluating text messaging interventions for glycemic control in adults with diabetes. Our goal was to capture the maximum number of eligible trials through broad inclusion criteria, thereby enabling a more precise overall effect estimate. We also aimed to stratify outcomes by follow-up duration to assess whether effects persist over time and check for gaps in long-term evidence. Finally, we sought to investigate key subgroups to determine whether certain populations derive greater benefit from these interventions.

Methods

This systematic review and meta-analysis was conducted in accordance with the 2020 Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines and registered in PROSPERO (International Prospective Register of Systematic Reviews) (CRD42023416462).⁵

Literature search

We conducted a systematic search of PubMed, EMBASE and the Cochrane Central Register of Controlled Trials for relevant papers published from January 2000 to December 2023. The start date of January 2000 was selected as the cut-off, given that RCTs on texting for health promotion were published after this. Medical subject headings and keywords for searches focused on the following three key topics: text messaging, diabetes, and RCTs. Additionally, references from selected articles were manually reviewed to identify relevant citations. The detailed search strategies for each database are provided in online supplemental table 1A–C.

Eligibility criteria

The following inclusion criteria were used to screen for eligible studies: (1) Study type: RCTs; (2) Population: adults aged ≥18 years with type 1 or type 2 diabetes; (3) Intervention: automated text-based or video-based messaging delivered directly to patients via SMS or applications to a phone or tablet device, where the messages promoted glycemic control via awareness of blood glucose trends or encouraging healthy behaviors (eg, physical activity, healthy eating, medication adherence, weight loss); (4) Outcome: mean difference in hemoglobin A1c (HbA1c) between baseline and follow-up in intervention and control groups; (5) Setting: high-income countries as defined by the World Bank (see online supplemental table 2 for a full list of eligible countries)⁶; (6) Language: study results available in English (text messages could be in any language). To minimize unintentional exclusions, these criteria were manually applied during abstract and full-text screening.

Studies were excluded if messages were only delivered to non-mobile or non-tablet devices (eg, pagers, glucometers, or computer websites requiring login) or if text message generation involved a purely manual process. Compared with automated messages, manual text message generation represents a more labor-intensive and costly intervention with potentially different effects. Studies were also excluded if participants were selected based on a history of cardiovascular events (eg, myocardial infarction, stroke, transient ischemic attack) or if participants had pre-diabetes or gestational diabetes (current or prior) without a diagnosis of diabetes.

Screening and data extraction

Titles, abstracts, and full texts were screened independently with the Rayyan citation manager by two authors (NP and GSK) in accordance with the predefined eligibility criteria. Data were manually extracted (without automation tools) using full texts to gather study characteristics (first author, year of publication, country, sample size, and type 1 or type 2 diabetes), participant characteristics (age, sex, race and ethnicity, and baseline HbA1c), intervention details (frequency and duration of texting, degree of personalization, presence of bidirectional messaging, and other intervention components), and outcomes (time points and results). Discrepancies in screening eligibility or data collection were resolved via a consensus meeting between NP and GSK. When necessary, consultation was sought by the corresponding author DTK. Where available, within-group means and SDs of HbA1c changes from baseline to follow-up were extracted for intervention and control groups. If a study reported HbA1c results at multiple time points, all were included in the analysis. Results were converted to units of percent HbA1c. For studies reporting only baseline and follow-up HbA1c values, the change in HbA1c between time points was calculated using a correlation coefficient of 0.5, which has been done in similar studies.⁷

Outcomes

The primary outcome of our study was the between-group difference in HbA1c change from baseline to follow-up.

Risk of bias

Two authors (NP and GSK) independently assessed the quality of included studies using V.2 of the Cochrane risk-of-bias tool for RCTs.⁸ This tool evaluates potential biases in the following key areas: the randomization process, deviations from the intended interventions (effect of assignment to intervention), missing outcome data, outcome measurement, and selection of reported results. To enhance clarity, we used the online robvis (Risk-Of-Bias VISualization) package to visually summarize risk of bias findings.⁹

Statistical analysis

Statistical analyses were performed using the meta-analytic functions within Cochrane’s Review Manager (RevMan) V.5.4 and JASP. The between-group difference in HbA1c change from baseline to follow-up was calculated using an inverse variance random effects model. Pooled estimates were expressed as weighted mean differences with 95% CIs. Forest plots were created based on this data. Statistical heterogeneity was assessed using the I² statistic and Cochran’s Q. Funnel plots and Egger’s regression test were done through the open-source program JASP.¹⁰

We conducted two unplanned subgroup analyses. Initially, our a priori protocol intended to stratify studies based on text message frequency. This approach proved unfeasible due to substantial variability in frequency across studies and the absence of a clear cut-off to define a high versus low frequency subgroup. Furthermore, there was an insufficient number of studies per time point to conduct meta-regression with frequency as a continuous covariate. However, we were able to identify alternative subgroups with strong theoretical justification. The first subgroup analysis divided studies based on the study-level mean HbA1c at baseline (≥8.6% vs <8.6%), a cut-off which differentiated participants with poor glycemic control. According to guidelines from the American Diabetes Association and Diabetes Canada, an HbA1c level above 8.5% in elderly individuals is associated with increased risks of long-term harm.¹¹ We hypothesized that text message interventions would have greater benefits for individuals with poorer baseline glycemic control.

The second subgroup analysis compared studies in which texting was the sole intervention component to studies in which texting was combined with other interventions (eg, glucometer, website, telephone calls). This distinction sought to determine whether any observed benefits of texting interventions could be attributed to texting itself.

RevMan was used to conduct a meta-analysis within each subgroup and explore potential sources of heterogeneity.

Overall quality of evidence

The quality of the evidence regarding the effect on glycemic control at 3, 6, and 12 months was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria.¹² GRADE evaluates several factors that can reduce the quality of evidence including the type of evidence, risk of bias (assessed using the Cochrane risk-of-bias tool),⁸ consistency across studies, directness of evidence, precision of the estimate, and risk of publication bias. Due to pitfalls in relying on a single method to assess publication bias, both visual inspection of funnel plots and Egger’s regression test were used.¹³

Sensitivity analysis

We also performed an additional analysis excluding studies at high risk of bias based on the Cochrane RoB 2 tool to evaluate the robustness of our findings.

Results

A total of 3,501 records were identified through the literature search, which was reduced to 2,324 after duplicates were removed. Full-text screening resulted in the inclusion of 29 studies after seven were excluded due to insufficient reporting of results (figure 1). The seven excluded studies can be provided in a file on request.

Figure 1. Study selection PRISMA flow diagram. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Characteristics of included studies

Table 1 summarizes the studies included in the meta-analysis.14,42 Additional information on study country and intervention versus control protocols is included in online supplemental table 3. The 29 studies included in the meta-analysis totaled 6,552 participants. The mean study-level age was 56 years, with 52% of participants being male. Over half of the trials had at least 100 participants. The mean study-level baseline HbA1c was 8.5% for the intervention groups and 8.4% for control groups. Intervention duration varied from 3 months to 2 years. Text message frequency ranged from 1 text every 2 weeks to 14 texts per week. About one-third of studies provided automated feedback tailored to participants’ glucose data.

Table 1. Characteristics of included studies in the meta-analysis.

Study identifiers		Participants and baseline characteristics					Intervention details
First author	Sample size, n	Mean age, y	Male, %	Mean baseline HbA1c, percent A1c	T2DM, %	Texts/week*	Texting duration	Personalized or fixed texting†, bidirectional or unidirectional texting‡
Agboola14	126	51	48	I: 9.0, C: 8.4§	100	14	6 mos	Personalized, bidirectional
Arora15	128	51	36	I: 10.2, C: 10.0	100	14	6 mos	Fixed, unidirectional
Azelton16	45	NA	33	I: 8.2, C: 8.3	100	7	3 mos	Personalized, bidirectional
Bauer17	69	62	48	I: 8.0, C: 7.7	91	14	6 mos	Fixed, unidirectional
Bell18	65	58	55	I: 9.6, C: 9.0	92	7	6 mos	Fixed, unidirectional
Boels19	230	59	60	I: 8.1, C: 8.3	100	2–6	6 mos	Personalized, unidirectional
Cheung20	642	60	65	I: 8.5, C: 8.5	100	4	6 mos	Personalized, unidirectional
Choudhry21	488	60	55	I: 9.8, C: 9.5	NA	0–7	12 mos	Personalized, unidirectional
Dobson22	366	47	52	I: 10.1, C: 9.8	65	1	3–9 mos	Personalized, bidirectional
Faridi23	30	56	37	I: 6.4, C: 6.5	100	NA	3 mos	Personalized, unidirectional
Fortmann24	126	48	26	I: 9.5, C: 9.6	100	14–21	6 mos	Fixed, bidirectional
Gautier25	499	62	58	I: 7.1, C: 6.9	100	7	3 mos	Personalized, unidirectional
Gimbel26	229	63	62	I: 7.5, C: 7.6	100	3	12 mos	Personalized, unidirectional
Huo27	502	60	82	I: 6.9, C: 7.1	NA	6	6 mos	Personalized, bidirectional
Jeong28	226	53	67	I: 8.2, C: 8.4	100	3–6	6 mos	Personalized, unidirectional
Kim29	100	48	50	I: 9.8, C: 9.8	100	7	3 mos	Personalized, unidirectional
Kim30	110	57	44	I: 8.4, C: 8.5	100	14	10 wks	Fixed, unidirectional
Leong31	181	59	69	I: 6.9, C: 6.7	100	2–3	3 mos	Personalized, unidirectional
Lim32	103	68	41	I: 7.8, C: 7.9	100	8	6 mos	Personalized, unidirectional
Lim33	100	65	75	I: 8.1, C: 7.9	100	14	6 mos	Personalized, unidirectional
Middleton34	40	33	50	I: 7.2, C: 7.3	100	2	12 mos	Personalized, bidirectional
Quinn35	100	53	51	I: 9.3, C: 9.2	100	NA	12 mos	Personalized, bidirectional
Ramallo-Farina36	1123	56	51	I: 7.3, C: 7.3	100	2	2 mos	Personalized, unidirectional
Saslow37	44	52	25	I: 8.4, C: 8.4	100	5	12 mos	Fixed, unidirectional
Spierling Bagsic38	172	58	52	I: 9.7, C: 9.4	100	6–13	6 mos	Personalized, bidirectional
Sugita39	41	56	70	I: 9.6, C: 9.5	100	2	6 mos	Fixed, unidirectional
Waller40	395	62	51	I: 8.2, C: 8.2	100	4–7	6 mos	Personalized, unidirectional
Xu41	65	55	31	I: 9.8, C: 9.2	100	2–21	6 mos	Personalized, bidirectional
Zamanillo-Campos42	207	62	65	I: 9.0, C: 9.1	100	5	3 mos	Personalized, unidirectional

^*Where text messages were tapered over time, the initial frequency is shown.

^†Personalized texting was defined as texting tailored to an individual participant, which ranged from including a participant’s name in a text message to tailoring messages based on a participant’s motivational status, recent blood glucose values, etc.

^‡Bidirectional texting was defined as responses to a participant’s text message (eg, an automated text sent in response to a participant’s blood glucose value).

^§Study groups were significantly different.

C, control; HbA1c, hemoglobin A1c; I, intervention; mos, months; NA, not available; T2DM, type 2 diabetes mellitus; wks, weeks; y, years.

Systematic review of included trials

In most studies, texting interventions were associated with greater HbA1c reductions than control. At 3 months, 12 of 14 trials favored texting interventions, with 6 showing significant reductions over control of 0.2 to 1.2 units. At 6 months, 13 of 20 trials favored texting interventions, with 5 showing significant reductions over control of 0.3 to 0.8 units. At 12 months, two of five trials showed significant benefit. Intervention design, population characteristics, and methodological quality varied. A consolidated summary of pooled HbA1c effects at 3, 6, and 12 months is provided in table 2.

Table 2. Pooled effects of text-messaging interventions on HbA1c at 3, 6, and 12 months, including post hoc subgroup analyses by baseline HbA1c and by intervention components. Values are pooled mean differences in percent HbA1c and represent the additional reduction in HbA1c with texting over control, positive values favor texting.

Meta-analysis	3 months (95% CI; p value; I²)	6 months (95% CI; p value; I²)	12 months (95% CI; p value; I²)
Original	0.29 (0.14 to 0.45; p=0.0001; 57%)	0.19 (0.07 to 0.30; p=0.001; 45%)	0.16 (−0.16 to 0.47; p=0.33; 68%)
Sensitivity analysis excluding high-risk studies	0.33 (0.11 to 0.55; p=0.003; 56%)	0.14 (0.01 to 0.26; p=0.04; 41%)	0.12 (−0.34 to 0.59; p=0.6; 71%)

Subgroup analysis: baseline A1c
Baseline A1c ≥8.6%	0.48 (0.15 to 0.80; p=0.004; 60%)	0.36 (0.12 to 0.61; p=0.004; 25%)	N/A
Baseline A1c <8.6%	0.20 (0.05 to 0.34; p=0.007; 43%)	0.13 (0.01 to 0.26; p=0.04; 53%)	N/A

Subgroup analysis: intervention components
Mixed intervention (texting+other)	0.35 (0.14 to 0.56; p=0.001; 54%)	0.27 (0.09 to 0.46; p=0.008; 59%)	N/A
Texting alone	0.24 (−0.03 to 0.50; p=0.08; 68%)	0.12 (0.00 to 0.25; p=0.06; 18%)	N/A

HbA1c, hemoglobin A1c; N/A, not available.

Primary outcome: effect of texting interventions on HbA1c at 3, 6, and 12 months

At 3 months (14 trials, n=2,947), texting interventions were associated with a 0.29-unit greater reduction in percent HbA1c over control (95% CI 0.14 to 0.45; p=0.0001). At 6 months (20 trials, n=4,703), texting interventions were associated with a 0.19-unit greater HbA1c reduction (95% CI 0.07 to 0.30; p=0.001). At 12 months (seven trials, n=2,038), texting interventions were associated with a 0.16-unit greater HbA1c reduction, but this was non-significant (95% CI −0.16 to 0.47; p=0.33). These results are shown in figure 2A–C.

Figure 2. Meta-analysis of mean difference in HbA1c from baseline to follow-up between texting and control groups. Results are depicted in NGSP units. Depicted are forest plots at 3 months (A), 6 months (B), and 12 months (C). Green squares represent the difference in mean HbA1c reduction, with the area of the square proportional to the inverse variance of the estimate. Horizontal lines denote 95% CIs for point estimates. The solid vertical line denotes a null effect. Diamonds represent the pooled difference in mean HbA1c changes derived under the random effects model. HbA1c, hemoglobin A1c. NGSP, National Glycohemoglobin Standardization Program.

Subgroup analysis based on baseline HbA1c

At 3 months (figure 3A), among seven studies with baseline HbA1c ≥8.6% (n=873), texting interventions were associated with a 0.48-unit greater HbA1c reduction compared with control (95% CI 0.15 to 0.80; p=0.004). In seven studies with baseline HbA1c <8.6%, texting interventions were associated with a 0.20-unit additional drop in HbA1c (95% CI 0.05 to 0.34; p=0007.).

Figure 3. Subgroup analyses of 3 months (A) and 6 months (B) HbA1c outcomes between texting and control groups based on participant baseline percent HbA1c. Subgroup analysis of 3 months (C) and 6 months (D) HbA1c outcomes based on whether intervention was texting alone or texting with other interventions. Horizontal lines denote 95% CIs for point estimates. The solid vertical line denotes a null effect. Diamonds represent the pooled difference in mean HbA1c changes derived under the random effects model. HbA1c, hemoglobin A1c.

At 6 months (figure 3B), texting interventions in nine studies with baseline HbA1c ≥8.6% (n=964) showed a 0.36-unit greater reduction in HbA1c (95% CI 0.12 to 0.61; p=0.004). In 11 studies with baseline HbA1c <8.6%, texting interventions were associated with a 0.13-unit greater reduction in HbA1c (95% CI 0.01 to 0.26; p=0.04).

No significant difference emerged between subgroups at 3 or 6 months. Subgroup analysis at 12 months was not completed due to small sample size.

Subgroup analysis based on intervention components

At 3 months (figure 3C), five studies using texting alone (n=1,193) showed a 0.24-unit greater reduction in HbA1c versus control (95% CI −0.03 to 0.50; p=0.08). In nine studies incorporating texting in multicomponent interventions (n=1,754), the additional HbA1c reduction over control was 0.35 units (95% CI 0.14 to 0.56; p=0.001). Subgroup differences were not significant.

Figure 3D depicts the subgroup analysis for 6 months of follow-up. Subgroup analysis at 12 months was not completed due to small sample size.

Publication bias

At 3 months, Egger’s test indicated significant asymmetry (p=0.014), and visual inspection revealed skew in the funnel plot (online supplemental figure 4A). A Trim and Fill analysis (online supplemental figure 5A and B) imputed five studies on the right side of the funnel, with effect sizes ranging from 0.27 to 0.81 in favor of the control group. These adjusted the pooled mean difference from a 0.29 percentage point greater reduction with texting interventions to a 0.22-point greater reduction (95% CI 0.07 to 0.37). At 6 months, Egger’s test was not statistically significant (p=0.078), and the funnel plot (online supplemental figure 4B) appeared more symmetric. At 12 months, Egger’s test was also non-significant (p=0.427), and the funnel plot (online supplemental figure 4C) revealed no obvious asymmetry, though the small number of studies (n=7) and wide variation in SEs limit interpretability.

Overall quality of the evidence

An initial GRADE score of 4 was assigned to outcomes, indicating a high certainty that texting interventions have a non-zero impact on HbA1c at 3, 6, and 12 months. No downgrades were applied for publication bias. However, concerns about allocation concealment (either absent or not clearly reported) lowered the certainty to moderate across all time points (online supplemental figure 6). Inconsistency further downgraded certainty to low at 6 and 12 months, and imprecision reduced it to very low at 12 months.

Sensitivity analysis

In a sensitivity analysis (online supplemental figure 7A and B) excluding five high-risk-of-bias studies, texting interventions were associated with a 0.33 percentage point greater reduction in HbA1c at 3 months compared with control (95% CI 0.11 to 0.55; p=0.003; I²=56%). At 6 months, excluding five high-risk-of-bias studies yielded a 0.14 percentage point greater reduction in HbA1c with texting interventions compared with control (95% CI 0.01 to 0.26; p=0.04; I²=41%). At 12 months, sensitivity analysis did not alter results (table 2).

Discussion

Principal findings

Across 29 RCTs, texting interventions were associated with modest, statistically significant reductions in HbA1c at 3 and 6 months versus usual care, with no clear benefit and high heterogeneity by 12 months. Effects were larger in studies with higher baseline HbA1c, highlighting baseline control as a key effect modifier. Signals were also greater when texting was delivered as part of multicomponent programs rather than as a stand-alone intervention, although there was significant heterogeneity in cointervention design. Findings were robust to sensitivity analyses excluding RCTs with higher risk of bias. At 3 months, Trim and Fill analysis attenuated but did not remove the benefit. We interpret the attenuation cautiously given that the original funnel contained both small and null-effect studies (consistent with small-study variation rather than selective publication). Overall, texting interventions appear to offer small, short-term improvements in glycemic control—most evident among those with poorer baseline control—while the durability of effect and the added value of specific components require higher-quality, longer-term trials.

Clinical relevance of findings

Pharmacotherapies typically achieve reductions in HbA1c of 0.5–1.5 units, depending on the medication.⁴³ Although the incremental benefit associated with texting interventions relative to control was below this threshold, it approached 0.5% in participants with poorer baseline control. Furthermore, modest reductions of <0.5% in HbA1c at a population level have been associated with significant reductions in harmful outcomes.⁴⁴ Further investigation and intervention optimization is warranted to determine whether texting interventions can achieve clinically meaningful HbA1c reductions over standard care, and to understand its impact at the population level.

Comparison with prior literature and implications for future research

Our findings are consistent with prior meta-analyses ranging from 9 to 13 studies that have reported reductions in HbA1c with texting interventions of 0.37 to 0.61 units over control.745,49 By including more trials, our meta-analysis extends previous work by highlighting the time-limited effects of texting interventions on HbA1c, the lack of high-quality evidence on long-term outcomes, and a potentially greater impact among individuals with poorer baseline glycemic control. Taken together, our results position texting interventions as a potential scalable adjunct to routine care for short-term HbA1c improvement—particularly in those starting from poorer control—while highlighting a need for more long-term, mechanism-focused trials.

Limitations

Our findings should be interpreted in the context of its limitations. We intentionally excluded studies from low- and middle-income countries, and this a priori restriction enhanced internal validity at the cost of external validity. Low- and middle-income countries differ from high-income countries in key contextual factors such as mobile technology access and infrastructure, digital health literacy, and diabetes medication use.⁵⁰ Pooling studies from these heterogeneous contexts could have introduced substantial confounding. Therefore, the present findings are best interpreted as applying to well-resourced health systems; complementary reviews focused on low- and middle-income countries are needed to determine whether texting yields comparable benefit.

Additionally, the external validity of our findings is limited by the relatively young population (study-level mean age 53 years) and the predominance of type 2 diabetes. Excluding studies published in English resulted in exclusion of only one study (figure 1). High statistical heterogeneity was observed at 12 months (I²=78%), limiting confidence in the long-term pooled estimate. While we attempted to explore sources of variability using subgroup analyses, our ability to do so was limited at 12 months due to the small number of contributing studies per covariate. Our ability to investigate differences in intervention design (eg, texting frequency, personalization) was limited at all time points by significant variability and inconsistency in reporting across studies. Furthermore, we were unable to explore cost-effectiveness, patient satisfaction, and privacy concerns related to text messaging interventions.

Conclusion

In summary, text messaging interventions are associated with short-term improvements in glycemic control among patients with diabetes, particularly those with elevated baseline HbA1c. In the future, these interventions may serve as a practical complement to in-person care, though higher-quality, longer-term studies are needed to refine and sustain their impact.

Data availability statement

Data sharing not applicable as no datasets generated and/or analyzed for this study.