Pregnancy represents a major life transition that is frequently accompanied by heightened psychological stress, including pregnancy-specific stress related to concerns about maternal health, fetal well-being, and childbirth. Recent evidence indicates that pregnancy-specific stress is prevalent even among low-risk populations and tends to increase with perceived pregnancy risk and advancing gestational age [1]. Elevated stress during pregnancy has been consistently associated with adverse maternal mental health outcomes as well as negative neonatal and child health outcomes, underscoring the importance of timely and effective stress management during the antenatal period [2-4].

A range of nonpharmacological interventions has been developed to address stress during pregnancy, and their effectiveness has been demonstrated across multiple studies. These include psychoeducational programs, structured psychotherapies such as cognitive behavioral therapy and interpersonal psychotherapy, and mind–body approaches including mindfulness and yoga [5-8]. Traditionally, such interventions have been delivered through face-to-face sessions. While effective, in-person delivery can be constrained by barriers related to access, cost, time, geographic distance, and the availability of trained professionals—limitations that may be particularly salient during pregnancy.

Digital health interventions (DHIs) have emerged as a scalable and low-threshold approach for delivering stress-management strategies, with growing evidence supporting their effectiveness in the general population. Meta-analytic evidence indicates that app-based stress-management interventions yield small but statistically significant improvements across self-reported, physiological, and neuroendocrine stress outcomes [9], and similar modest reductions in perceived stress have been reported in randomized trials of mental health smartphone apps [10]. These findings suggest that digitally delivered interventions can effectively support stress reduction when designed and implemented appropriately.

However, evidence specific to pregnancy remains more limited and less conclusive. Much of the existing literature on DHIs during pregnancy has focused on broader perinatal mental health outcomes—particularly depression and anxiety—rather than stress as a primary target. For example, reviews of digital mindfulness and nurse-led eHealth interventions have reported consistent benefits for depressive and anxiety symptoms but mixed or inconclusive effects on stress [11,12], while telemedicine-based psychological interventions often combine antenatal and postpartum populations and rarely prioritize stress outcomes [13]. An umbrella review of DHIs for perinatal women suggested overall benefits for psychological outcomes, including stress; however, its findings were pooled across heterogeneous populations, intervention types, outcomes (stress, depression, and anxiety), and perinatal stages, with only a small subset of reviews specifically addressing stress during pregnancy [14]. Collectively, these syntheses highlight that stress-focused evidence in pregnant populations remains fragmented and underdeveloped.

At the same time, DHIs offer distinct methodological and practical advantages that warrant more nuanced evaluation. Digital interventions are inherently multicomponent, comprising combinations of delivery methods (eg, mobile apps, web-based platforms, and telemedicine), functional features (eg, self-monitoring, automated feedback, and reminders), and intervention contents (eg, psychoeducation, mindfulness, relaxation, and cognitive behavioral skills). Behavioral science frameworks emphasize that technology functions primarily as a delivery vehicle rather than the therapeutic agent itself [15], and empirical assessments of stress-management apps similarly distinguish delivery modalities, functional features, and therapeutic content as separable but interacting components [16]. Despite this, prior reviews of digital mental health interventions during pregnancy have rarely synthesized evidence across these dimensions, limiting understanding of which combinations of delivery methods, functions, and contents are most relevant for stress reduction in pregnant women.

Taken together, the existing literature highlights a persistent gap in evidence regarding digital interventions specifically designed to reduce stress during pregnancy. Although DHIs for perinatal mental health have received growing attention, most prior reviews have focused primarily on depression and anxiety or have pooled multiple psychological outcomes, with stress often insufficiently examined as a primary outcome. As a result, the effectiveness of digital interventions targeting stress during pregnancy remains unclear.

Moreover, previous syntheses have largely categorized digital interventions by delivery platform, providing limited insight into how specific intervention components contribute to stress reduction. DHIs are inherently multicomponent, integrating delivery methods, functional features, and therapeutic content; however, the relative importance of these components for managing pregnancy-related stress has not been systematically examined.

To address these gaps, the present systematic review and meta-analysis provide an updated and focused synthesis of digital stress-management interventions during pregnancy. By centering stress reduction as the primary outcome and examining intervention effectiveness through the lens of delivery methods and intervention content, this review aims to clarify how digital interventions can be optimally designed and implemented to support stress management in real-world antenatal care settings.

Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidance [17] and PRISMA-S [18], this systematic review aimed to (1) identify and describe digital stress-management interventions for pregnant women (Population), including delivery platforms and key components (Intervention); (2) evaluate the effectiveness of these interventions compared with usual care, wait-list, or active comparators (Comparator) on validated stress outcomes (Outcome); and (3) summarize intervention strategies and delivery methods associated with engagement and stress reduction.

We conducted a systematic review and meta-analysis to evaluate digitally delivered interventions for reducing stress in pregnant women. The review adhered to the PRISMA guidelines [17] and PRISMA-S (PRISMA literature search extension) [18] (Multimedia Appendix 1).

Eligibility criteria were predefined using the PICOS (population, intervention, comparison, outcome, and study design) framework. Participants were pregnant women. Interventions included any digitally delivered program primarily aimed at reducing or managing stress during pregnancy (antenatal period), delivered via mobile apps, web-based platforms, telemedicine, or other technology-enabled modalities. Studies in which the intervention was initiated postpartum were excluded. Comparators included usual care, waitlist/no intervention, or nondigital interventions. Outcomes included stress assessed using validated instruments; when the Depression Anxiety Stress Scales (DASS) was used, only the stress subscale was extracted. Study designs included RCTs and quasi-experimental/nonrandomized intervention studies with extractable pre-post outcome data.

Exclusion criteria included (1) postpartum-only interventions (interventions initiated after delivery), (2) interventions not primarily targeting stress management (eg, programs targeting depression, weight control, smoking cessation, or other conditions in which stress was only a secondary outcome without a stress-focused rationale), (3) insufficient outcome data to estimate effects (no extractable pre-post data or between-group comparisons), and (4) conference abstracts or trial registry entries without sufficient methodological detail or outcome data for extraction and risk-of-bias assessment.

The initial systematic search was conducted across 4 databases (CINAHL, Cochrane Library, Embase, and PubMed) from database inception to September 2023. In response to reviewer comments and to ensure the currency of the evidence base, the search was updated by re-running the same search strategy in all databases for records published from September 2023 through November 30, 2025. The updated search was executed on December 10, 2025, by JYL and SHP, and all newly retrieved records were screened and incorporated into the final synthesis as appropriate. Searches were conducted with the assistance of 2 medical librarians, and the final search strategies for each database are reported in Multimedia Appendix 2. Each database was searched independently using its native platform (PubMed, Embase.com, CINAHL via EBSCOhost, and the Cochrane Library). Databases were not searched simultaneously on a single platform. In addition to database searching, reference lists of included studies were manually screened to identify potentially relevant articles using citation searching methods. We did not contact study authors, experts, or intervention developers to obtain additional data, nor were other information sources such as trial registries or gray literature repositories searched.

Electronic search strategies were developed in consultation with 2 medical librarians, who proposed candidate keywords and controlled vocabulary; the research team reviewed and refined the strategies, and final searches were executed by the study team with librarian support. All search strategies were newly developed for this review and were not adapted from or reused based on search strategies reported in previous literature reviews. The strategy combined controlled vocabulary (MeSH [Medical Subject Headings], Emtree, and CINAHL Subject Headings) and keywords related to pregnancy, digital health, and stress. Pregnancy-related terms included “pregnan*,” “pregnancy,” “antenatal,” and “prenatal” (and related indexing terms). Digital health terms included “digital health,” “mHealth,” “eHealth,” “telemedicine,” “mobile app,” and “smartphone.” Stress-related terms included “stress” and relevant intervention terms (eg, relaxation, mindfulness, cognitive behavioral). Animal studies were excluded, and results were limited to English-language publications. Full search strategies for each database are provided in Multimedia Appendix 2.

All records were imported into EndNote and deduplicated. Three reviewers (JIK, JYL, and SHP) independently screened titles and abstracts to identify potentially eligible studies. Full texts were retrieved for records deemed relevant or uncertain; when full-text articles were difficult to obtain, a medical librarian assisted with document retrieval. Two reviewers (JIK and SHP) independently assessed full-text eligibility and cross-checked decisions, with disagreements resolved through discussion (and adjudication by a third reviewer when necessary). For the updated search, a second screening pass was completed on December 12, 2025, and all newly retrieved articles were screened and assessed for eligibility using the same procedures as the initial review, with eligible studies incorporated into the final synthesis.

Data was extracted using a piloted structured form capturing intervention purpose, participant characteristics, timing and duration, intervention strategies/components and digital functions, delivery mode, comparator, sample size, stress measures, and outcome values. When outcomes were measured repeatedly, the assessment closest to intervention completion was extracted; follow-up outcomes were extracted separately where available. Two reviewers independently entered data into an Excel file and verified consistency. Discrepancies were resolved by consensus among 3 reviewers to ensure accuracy and consistent terminology.

Risk of bias was assessed using RoB 2 (a revised Cochrane risk-of-bias tool for randomized trials) for randomized trials [19] and ROBINS-I (risk of bias in nonrandomized studies of interventions) for quasi-experimental/nonrandomized studies [20]. Two reviewers (JIK and SHP) independently rated each study, and disagreements were resolved through discussion with adjudication by a third reviewer [JYL] when needed. RoB 2 judgments were made across 5 domains and summarized as low risk, some concerns, or high risk. ROBINS-I judgments were made across 7 domains and summarized as low, moderate, serious, or critical risk of bias. Risk-of-bias assessments informed sensitivity analyses and the certainty of evidence (GRADE [Grading of Recommendations, Assessment, Development and Evaluation]).

The primary outcome was the change in stress from baseline to postintervention (closest to intervention completion). When DASS was used, only the stress subscale was extracted. Continuous outcomes were synthesized as standardized mean differences (SMDs; Hedges g) and reported with 95% CIs [21], with negative values indicating lower stress in the intervention group. When studies did not report means and SDs directly, effect estimates were derived from available statistics (eg, F values, odds ratios, chi-square values) using standard conversion methods.

Study characteristics (population, intervention components and digital functions, delivery mode, stress measures, and effects) were summarized narratively. For quantitative synthesis, meta-analyses were conducted using random-effects models to reflect expected between-study variability in true effects. Analyses were performed in in R (version 4.5.2; R Foundation for Statistical Computing, Vienna, Austria) using the Hartung–Knapp–Sidik–Jonkman method [22]. Heterogeneity was quantified using τ² and I², with the Cochran Q test reported where appropriate. To distinguish the average pooled effect from the expected distribution of effects across settings, we reported 95% prediction intervals for the main meta-analyses (but not for subgroup analyses). CIs describe uncertainty around the pooled mean effect, whereas prediction intervals indicate the plausible range of effects in a new setting. Prespecified subgroup analyses (eg, intervention strategies/components and delivery modes) and, when sufficient studies were available, meta-regression were used to explore potential sources of heterogeneity. Sensitivity analyses excluded studies at high or serious risk of bias and tested alternative assumptions for imputed parameters where applicable. Small-study effects were assessed using funnel plots and Egger’s regression test when ≥10 studies were available for a given main analysis.

Two reviewers (JIK and SHP) independently assessed certainty of evidence using GRADE, with disagreements resolved through consensus and third-reviewer adjudication. Certainty was evaluated across risk of bias, inconsistency, indirectness, imprecision, and publication bias. Downgrading decisions were documented in the Summary of Findings table and were informed by RoB 2/ROBINS-I judgments, heterogeneity (including the extent to which prediction intervals indicated variability across settings), imprecision (width of CIs and information size), and evidence of small-study effects (funnel/Egger).

The study selection process is summarized in Figure 1. Studies were excluded if they targeted health conditions other than stress (eg, depression or anxiety, weight control, or smoking cessation) without a stress-focused intervention rationale, or if stress was measured only as a secondary outcome. Studies describing intervention development without reporting evaluable effectiveness outcomes were also excluded. Citation searching identified additional records, which were assessed at the full-text level but ultimately did not meet the inclusion criteria. Reasons for exclusion at each stage were documented and are presented in the PRISMA flow diagram.

Following the updated search and eligibility assessment of newly retrieved records, 4 additional studies were included. In total, 19 studies (15 from the initial search and 4 from the updated search) were included in the final review.

Included studies were published between 2016 and 2025 [23-41]. Among them, 7 studies [24,26,28,31,33,34,38] were conducted in the United States (37%), followed by studies from Iran (n=2) [29,37], China (n=2) [25,34], Korea (n=2) [24,32], and other countries. Overall, there were 12 RCTs (63%) and 7 quasi-experimental studies (37%; Table 1).

The results of the risk of bias assessment are presented in Figure 2 [23-41]. Among the 12 RCTs assessed with RoB 2, one study was judged at high risk of bias overall, driven by concerns in the randomization process (Domain 1). Two additional RCTs were rated as having some concerns overall, reflecting incomplete reporting or concerns in specific domains, while the remaining RCTs were judged at low risk of bias across domains.

Among the 7 nonrandomized studies assessed with ROBINS-I, 2 studies were judged at serious risk of bias overall, primarily due to deviations from intended interventions (Domain 4). One study was judged at moderate risk of bias, and the remaining studies were judged at low risk across ROBINS-I domains. No study was judged at critical risk of bias.

The meta-analysis results of these studies were as follows.

Twelve of the 19 included effect sizes were derived from RCTs. Using the GRADE, the certainty of evidence for stress reduction was rated as moderate, downgraded by one level for risk of bias (1/12 RCTs rated overall high risk, and 2/12 rated some concerns in RoB 2), with no serious concerns identified for inconsistency, indirectness, imprecision, or publication bias.

The certainty of evidence was rated as moderate, downgraded by one level for risk of bias. The pooled effect estimate for RCTs was SMD −0.39 (95% CI −0.52 to −0.26), reflecting a moderate level of certainty in the stress-reducing effect of digital interventions (Multimedia Appendix 3).

Guided by our objectives to characterize antenatal DHIs for stress management, evaluate their effectiveness, and summarize intervention strategies associated with engagement, this systematic review and meta-analysis found that digitally delivered stress-management interventions are generally effective in reducing stress among pregnant women compared with routine care or control conditions. These findings support the growing role of DHIs as a feasible and scalable approach for addressing pregnancy-related stress. Consistent with an umbrella review of digital interventions for perinatal psychological outcomes, digital approaches appear capable of reducing stress symptoms overall, although prior syntheses often pooled stress with other mental health outcomes [14]. They are partially aligned with a recent systematic review and meta-analysis of digitally delivered mindfulness interventions in pregnant women, which reported clear benefits for depression and anxiety and more variable effects for stress [11]. Our findings extend this evidence by focusing specifically on stress as a primary outcome during pregnancy.

When interventions were classified by dominant strategy, mindfulness-based approaches tended to show the most consistent signals of stress reduction. This pattern aligns with earlier reviews of perinatal mindfulness interventions and with emerging trials of digital mindfulness programs tailored to pregnancy [13,42]. Recent randomized trials suggest that mobile-delivered mindfulness programs can reduce depression, underscoring the potential value of skills-based, self-directed content when engagement is maintained [43]. Education-focused programs were also beneficial, particularly when paired with supportive elements such as coaching, peer interaction, or structured follow-up, suggesting that informational content alone may be insufficient for sustained stress reduction. In contrast, gamified interventions were evaluated in only one study and showed mixed results, with improvements detected in a biological stress indicator but not in self-reported stress, underscoring the need for further research using multiple outcome measures. Importantly, although subgroup-specific point estimates varied across strategies, the formal test for between-subgroup differences was not statistically significant. This indicates insufficient meta-analytic evidence to conclude that any single strategy is superior for stress reduction during pregnancy. Given the limited number of studies within some subgroups and the resulting low statistical power to detect moderator effects, these subgroup patterns should be interpreted cautiously and viewed as exploratory rather than confirmatory.

Beyond intervention strategy, how these programs were delivered also appeared to shape feasibility and engagement. Across studies, most interventions were delivered via mobile apps, with fewer web-based platforms and telemedicine approaches, reflecting contemporary trends in digital health delivery for pregnant populations. The predominance of mobile app–based interventions observed in this review mirrors widespread smartphone use among pregnant women and supports the feasibility of delivering brief, scalable stress-management content within routine antenatal care [16,44]. The increasing dominance of mobile-delivered interventions in more recent studies highlights a shift toward app-based platforms that can leverage a wider range of interactive functions—such as continuous monitoring, automated feedback, and reminders—which may enhance engagement and responsiveness compared with more static web-based programs. While earlier meta-analyses of internet-delivered psychological interventions primarily emphasized benefits for depression and anxiety outcomes [45], our findings suggest that web-based interventions can also contribute to stress reduction during pregnancy, albeit based on a smaller number of trials.

Beyond delivery mode alone, consistent patterns emerged in how interventions were constructed. Most DHIs combined mobile or web-based delivery with self-guided use, supported by functional features such as self-monitoring, automated feedback, reminders, or asynchronous communication, and delivered therapeutic content including psychoeducation, mindfulness practices, relaxation techniques, or cognitive behavioral therapy–based skills. Multi-component designs were the norm rather than the exception, indicating that digital stress-management interventions during pregnancy rarely rely on a single active ingredient. Compared with earlier reviews that grouped interventions primarily by delivery methods or broad perinatal mental health outcomes [13,46], the present review adopts a strategy-oriented perspective that provides a more clinically meaningful framework for understanding how digital interventions may be designed to reduce stress during pregnancy. These findings highlight the need for future research that directly compares single-strategy interventions (such as mindfulness alone or psychoeducation alone) with multi-component interventions, in order to clarify whether additive or synergistic effects contribute to stress reduction in this population.

Interpretation of these findings should consider both between-study variability and study quality. Although between-study variability did not indicate extreme inconsistency, the prediction interval suggests that beneficial effects are likely across future settings, albeit with meaningful variation in magnitude. Reporting prediction intervals alongside CIs helps distinguish the average effect from the distribution of effects expected in real-world implementations [19]. Subgroup findings should therefore be interpreted in the context of overall consistency rather than as definitive evidence of differential effectiveness across strategies [47]. Finally, assessments of small-study effects should be interpreted cautiously, as funnel plot asymmetry may reflect mechanisms beyond publication bias, including true heterogeneity or methodological differences across studies [48].

These results suggest that DHIs can serve as a practical adjunct to routine antenatal care by extending access to evidence-based stress-management skills, particularly for women who face barriers to in-person services. Interventions that combine structured mindfulness practice or targeted education with functional support (eg, reminders, progress feedback, brief coaching, or moderated peer support) may be especially useful for maintaining adherence and reinforcing skills [16]. Implementation may be facilitated by integrating these tools into prenatal education pathways and ensuring appropriate guidance for women with elevated distress who may need stepped-up care. Recent trials suggest that mobile-delivered mindfulness may improve depressive symptoms during pregnancy, but sustained engagement and equitable access remain key implementation challenges [43]. From a design perspective, the most promising programs provided clear goals, brief and repeatable programs, and opportunities for personalization. Future DHIs could also explore safe integration of passive sensing or wearable-enabled feedback to tailor content in real time, while ensuring data privacy and minimizing user burden [47,48]. Recent trials also highlight the practical importance of reporting engagement metrics (eg, module completion, frequency of practice, and prompt response) so that implementation decisions can be based not only on efficacy but also on real-world use patterns [43].

Several limitations should be noted. The number of trials within some strategy subgroups was small, limiting precision and the ability to detect differential effects. Our restriction to English-language publications and the use of only 4 databases may have missed relevant studies, and most included trials were conducted in high-income countries, which may limit generalizability to low- and middle-income contexts. Outcomes were primarily self-reported, raising the possibility of reporting bias. In addition, incomplete reporting of intervention details and adherence in some trials constrained the interpretation of which components drove benefit. Because the review was not prospectively registered, transparency may be lower than in preregistered reviews; however, we sought to mitigate this limitation by adhering to PRISMA guidance and applying the GRADE approach to characterize certainty of evidence. Moreover, given the rapid evolution of digital perinatal interventions, studies published after our search window may influence future pooled estimates; ongoing evidence surveillance or living-review approaches may therefore be warranted [14,43].

Overall, this systematic review and meta-analysis provide novel evidence that digitally delivered stress-management interventions can meaningfully reduce stress during pregnancy, highlighting the potential of digital health approaches as a complementary component of antenatal care. Unlike previous reviews that primarily categorized interventions by delivery platform or aggregated stress with other psychological outcomes, this study offers an innovative strategy-based synthesis that disentangles how specific intervention contents and supportive features contribute to stress reduction in pregnant women. By distinguishing intervention strategies rather than technologies alone, the findings advance understanding of what works within digital stress-management programs during pregnancy.

Across studies, the most consistent benefits were observed for mindfulness-based approaches and for educational interventions combined with supportive features such as coaching or peer interaction, underscoring the importance of active skill-building and engagement rather than information provision alone. Importantly, while digital tools are not intended to replace clinician-delivered care, the results indicate that they can extend access to evidence-based coping strategies and provide scalable, low-intensity support that fits within real-world antenatal workflows. As mobile app–based interventions continue to expand, their ability to integrate multiple functions—such as self-monitoring, feedback, and reminders—positions them as particularly feasible and adaptable tools for routine maternity care.

By clarifying the effectiveness of digital stress-management interventions specifically during pregnancy and by framing interventions according to their strategic components, this review contributes actionable insights for clinicians, researchers, and health system planners. The findings support the integration of digital stress-management programs as an adjunct to standard antenatal services, with the potential to improve reach, equity, and continuity of psychosocial support for pregnant women in diverse care settings.