This study employed a convergent mixed methods design [28], combining a quasi-experimental quantitative comparison with qualitative inquiry. The quantitative arm assessed changes in maternal knowledge, health practices, satisfaction, and pregnancy outcomes between intervention and control groups. Recognizing the limitations of a relatively small sample size, the qualitative arm was designed to complement quantitative findings, provide explanatory depth, and explore mechanisms of practice change. Quantitative and qualitative results were analyzed separately and integrated during the interpretation stage to enable a comprehensive evaluation.

The study was conducted between January and December 2024 in a district-level maternal and child health hospital in Beijing. Women attending routine ANC were invited to participate in the quantitative arm. Eligibility criteria included singleton pregnancy (24‐34 wk gestation), ability to attend antenatal classes, access to a smartphone for WeChat use, and provision of informed consent. Exclusion criteria included severe pregnancy complications (eg, preeclampsia, placenta previa requiring hospitalization), preexisting metabolic or cardiovascular conditions, and cognitive or communication difficulties. A total of 40 women were enrolled, with 20 attending an experiential antenatal class only (control group) and 20 attending the same class followed by a 7-day online logging program (intervention group). As part of routine study management, participants in each arm were also connected via a closed WeChat group. The qualitative arm was nested within this cohort. Twenty women, drawn purposively from both groups to ensure variation in age, parity, and engagement levels, were invited for in-depth interviews (10 intervention and 10 control). Participants were anonymized using alphanumeric codes (DK01-DK10 for intervention; WDK01-WDK10 for control).

The intervention comprised 2 components. First, all participants attended a 90-minute session delivered by obstetricians, midwives, and nutritionists. This session combined lecture-style teaching with demonstration, role-play, group discussions, and practice exercises, and covered key topics, such as pregnancy nutrition, gestational weight management, prevention of gestational diabetes and anemia, and safe physical activity. The control group received this standardized experiential class only. The intervention group was enrolled in a WeChat-based miniprogram for 7 consecutive days, during which they were asked to log daily information on diet; exercise; and, where relevant, blood glucose monitoring. Health care providers monitored entries, provided reminders, and offered personalized feedback through the miniprogram, ensuring continuity of support beyond the classroom. The overall intervention workflow is shown in Figure 1. Thus, the 2 groups differed only in the presence of the digital logging component.

In addition, separate closed WeChat groups were established for the intervention and control arms to facilitate routine communication, study management, and responses to participants’ questions. Group access was restricted to enrolled participants and required manual approval by the group administrator (a research team member; the same administrator managed both groups). There was no fixed discussion theme; however, participants were informed that irrelevant content or commercial advertisements were not permitted. The study team proactively posted logging reminders and the miniprogram entry link in the intervention group only, whereas the control group did not receive proactive messages. In both groups, staff responded to questions raised by participants. Personalized feedback on the logged content was delivered to individual participants within the miniprogram rather than posted in the group chat.

Outcomes assessed corresponded to the quantitative result domains that include maternal knowledge, health practices, service experience, and selected clinical outcomes. Knowledge, health practices, and service experience were assessed once at a standardized 7-day postclass follow-up for both groups. During the 7-day period, participants in the intervention arm logged diet, exercise, and (where applicable) glucose readings via a WeChat miniprogram with provider feedback, whereas controls received the class without any digital component. A structured knowledge–attitude–practice questionnaire [29] was administered on day 7 (±1 d). The knowledge section comprised 7 items on pregnancy nutrition, weight management, gestational diabetes mellitus prevention, and anemia recognition, yielding a total score of 0 to 7 (1 point per correct response). Health practice indicators were coded as binary according to prespecified thresholds: dietary management ≥5 days/week, physical activity ≥3 times/week, regular supplement use (daily or every other day), and sharing health information with family (occasionally/often). Service experience was measured on a 5-point Likert scale (clarity of communication, respect for needs, personalization, and overall satisfaction). Clinical outcomes (labor duration, delivery mode, perineal trauma, postpartum hemorrhage, gestational complications, and neonatal outcomes such as birth weight, asphyxia, and malformations) were abstracted from medical records.

For the qualitative data, semistructured interviews were conducted 2 to 8 weeks after the intervention. The interview guide explored (1) emotional responses to the program, (2) perceived changes in knowledge and behavior, (3) experiences of peer and provider interaction, and (4) views on the applicability of learned strategies in daily life. Interviews lasted 10 to 25 minutes, were audio-recorded, transcribed verbatim, and anonymized prior to analysis.

Quantitative data were analyzed using SPSS version 25.0 (IBM Corp). Descriptive statistics were used to summarize baseline characteristics and outcome measures, with continuous variables presented as means and SDs and categorical variables expressed as frequencies and percentages. Group comparisons were explored using χ² or Fisher exact tests for categorical data and independent-samples t tests for continuous data where appropriate. However, given the modest sample size, emphasis was placed on reporting effect sizes and 95% CIs to indicate the magnitude [30] and direction of differences rather than relying solely on P values for statistical significance. All quantitative findings were interpreted as preliminary trends and considered alongside qualitative results to provide a more comprehensive understanding of the intervention’s impact.

Qualitative data from semistructured interviews were analyzed thematically following the Braun and Clarke 6-step approach [31]. Transcripts were read repeatedly for familiarization, coded independently by 2 researchers, and iteratively developed into themes through discussion and refinement. Discrepancies were resolved by consensus, and representative quotations were selected to illustrate key points. Rigor was enhanced through member checking with a subset of participants and triangulation with class records and log data. Integration of quantitative and qualitative findings occurred at the interpretation stage: while the quantitative results identified patterns of change in knowledge, behaviors, and outcomes, the qualitative analysis provided insight into the mechanisms through which these changes occurred, thereby adding contextual depth and explanatory power.

Integration of findings was conducted at the interpretation stage using a convergent mixed methods approach. Quantitative results identified measurable changes in knowledge, behaviors, satisfaction, and selected clinical outcomes, whereas qualitative narratives offered insights into the processes and mechanisms underlying these changes. The 2 strands were brought together to assess complementarity, convergence, and divergence: quantitative trends pointing to improved dietary adherence, supplement use, and self-monitoring were supported by qualitative accounts of enhanced confidence, emotional reassurance, and peer support; observed differences in maternal outcomes were contextualized by participants’ descriptions of how new practices were adopted and sustained. In this way, the integration of datasets provided a more comprehensive understanding of both the effects and the pathways of the intervention than either method could yield alone.

To ensure the validity and trustworthiness of the findings, several strategies were adopted across both components of the study. For the quantitative analysis, descriptive statistics and effect sizes were reported in addition to P values, with 95% CIs used to illustrate the precision of estimates. Given the limited sample size, findings were interpreted cautiously as preliminary trends rather than definitive effects. For the qualitative component, rigor was enhanced through independent dual coding, iterative theme development, and resolution of discrepancies by consensus. Member checking was undertaken with a subset of participants to confirm the resonance of interpretations with their lived experiences, and triangulation with classroom records and digital log data was applied to strengthen confirmability. Reflexive notes and coding logs were maintained to document analytic decisions, ensuring transparency and dependability. Finally, the integration of quantitative and qualitative findings followed a structured convergent mixed methods framework, which enabled the systematic comparison of results and enhanced the overall credibility of conclusions by drawing on multiple sources of evidence.

The study protocol was reviewed and approved by the Ethics Committee of Beijing Daxing District Maternal and Child Health Hospital (approval number: LLSC2024014). All participants provided written informed consent prior to participation. Participants did not receive any financial compensation, reimbursement, or other incentives for participation. The study complied with the principles of the Declaration of Helsinki. Data confidentiality was safeguarded through anonymization and the secure storage of both quantitative and qualitative materials.

For the digital component, the WeChat miniprogram collected only study-relevant behavioral logs (diet, exercise, and glucose values, where applicable) and did not require participants to enter national identification numbers or unrelated personal data. Access to the administrative interface was restricted to authorized study staff via password-protected accounts. Exported datasets were deidentified prior to analysis and stored on encrypted, access-controlled institutional computers. Any participant quotations used in reporting were anonymized.

As communication also occurred within closed WeChat groups, residual privacy risks inherent to social messaging environments could not be fully eliminated, including participant-initiated screenshots or secondary sharing beyond the research team’s control. To mitigate this, group membership was restricted to enrolled participants with manual administrator approval, and participants were informed at enrollment not to disclose identifiable or highly personalized information in the group chat.

A total of 40 pregnant women completed the quantitative component of the study, with 20 assigned to the intervention group (experiential class plus online logging) and 20 to the control group (experiential class only).

Thematic analysis of postintervention interviews revealed 3 interrelated mechanisms of change through which the experiential class and online logging intervention shaped women’s experiences and practices: (1) empowerment and self-efficacy, (2) practice and persistence, and (3) systematic and environmental support.

This mixed methods study evaluated the feasibility and short-term, preliminary effects of integrating experiential antenatal classes with a 7-day digital logging component in a district-level maternal and child health institution setting in Beijing. We observed short-term improvements in maternal knowledge, selected self-reported practices, and service satisfaction, alongside underpowered and nonsignificant clinical trends (such as reduced labor duration and postpartum blood loss). Thematic analysis further illuminated a chain of mechanisms of change: self-efficacy was enhanced collectively by knowledge from experiential learning and interpersonal interaction and, in turn, translated into daily health practices under the influence of external supports or barriers. Together, these results provide new evidence that bridging participatory education with digital accompaniment can address persistent gaps in maternal health promotion.

The quantitative analysis demonstrated a significant incremental effect provided by experiential classes and online logging on the knowledge and practices of participants in the intervention group. The further enhancement of these 2 indicators could be associated with each other. Knowledge gains are driven by feedback loops enabled by online logging practices. For example, women posting photos of meals or asking doctors questions via WeChat allow doctors or peers to correct misconceptions or address individual gaps. This is consistent with recent literature. For instance, He et al [32] found that WeChat‐based prenatal education’s real-time communication between patients and clinicians enables clarification of misunderstandings and strengthens comprehension. Wu et al [33] demonstrated that caregiver feeding knowledge was improved not just through message delivery but through caregiver queries and interactions in online groups. Furthermore, regular practices facilitated by online logging can be seen as an extension of experiential learning. Concrete action and reflection can deepen understanding and embed skills in daily routines, helping knowledge “stick” in practice [16].

The intervention group’s superior behavioral outcomes suggest that the digital logging component reinforced continuity of support. However, the results from similar digital interventions are mixed: some trials show strong improvements in adherence and dietary or gestational weight outcomes, whereas others report minimal or no practice changes beyond standard care [34-36]. Our qualitative findings suggest another contributing mechanism: digital logging made shared goals more explicit, and therefore, participants reported feeling socially motivated. It aligns with the concept of collective efficacy, where group norms and peer visibility contribute to sustained practices. Studies in participatory interventions show that when group members believe in shared capability and observe others’ progress, adherence is more likely [37,38].

Although differing in scale, participants in both control and intervention groups reported that gaining knowledge through antenatal education helped reduce nervousness and fear. In our quantitative results, both groups rated high satisfaction with services. This finding resonates with other studies of interactive or group-based antenatal education (small classes, discussions, etc), such as Augar et al [39] and Brixbal et al [40]. There is also evidence that good satisfaction contributes to continuity of care and positive health outcomes [41,42]. These mechanisms are consistent with contemporary ANC guidance that places women’s experience, communication quality, and continuity of support at the center of care [11].

Additionally, participants in our study reported that both external and environmental conditions inhibited and facilitated practice change. This is consistent with qualitative evidence. For example, Escañuela Sánchez et al [43] show that even when women have awareness and motivation, social norms, stigma, and poor interaction with health care professionals often block effective weight management practices. Similarly, Jhaveri et al [44] in Uttar Pradesh find that structural opportunity barriers, such as lack of transport, limited food availability, and financial constraints, prevent pregnant women from translating knowledge into consistent nutrition, ANC attendance, or supplement intake. These findings reinforce that didactic changes (knowledge alone) are insufficient and practice changes depend critically on environmental support and opportunity.

The convergence of quantitative improvements and qualitative insights underscores the feasibility of scaling this model in urban ANC settings. The intervention required minimal additional resources beyond routine classes, leveraging an existing and widely adopted digital platform (WeChat) for continuity of care. As China continues to prioritize maternal and child health under the Healthy China 2030 strategy, integrating participatory education with digital reinforcement could serve as a cost-effective approach to optimize antenatal services, particularly in tertiary hospitals where patient volume and complexity are high. Furthermore, the qualitative findings identified emotional reassurance, trust, and peer connection as important dimensions of care that are not adequately captured by traditional clinical indicators but are essential for improving service quality and women’s lived experiences. However, we must acknowledge that scalability may differ substantially outside urban district hospitals. In rural or low-connectivity areas, limited internet access, lower smartphone penetration, and constrained staffing may reduce feasibility or require alternative modalities (eg, SMS text messaging–based prompts, offline-capable tools, or community health worker support). Implementation should be adapted to local digital infrastructure and population needs rather than assumed transferable from an urban, digitally literate sample.

To our knowledge, this is the first study integrating experiential antenatal classes with digital logging in maternal health promotion. The mixed methods design not only allowed for outcome measurement but also provided explanatory insights into mechanisms of change. By situating the intervention in a district-level maternal and child health hospital rather than a tertiary setting, the study reflects the real-world context where most women receive ANC, thereby enhancing external validity. The intervention itself was low cost and built on existing tools such as WeChat, making it adaptable and scalable in diverse settings. Moreover, outcomes were assessed across multiple dimensions, including knowledge, practices, satisfaction, and clinical indicators, broadening the evaluation scope and practical relevance of findings.

Nonetheless, limitations should be acknowledged. The modest sample size and quasi-experimental design without randomization limit statistical power and raise the possibility of residual confounding. In addition, knowledge, practices, and satisfaction were assessed only once post intervention, and the digital logging component was restricted to a short 7-day period, leaving the long-term sustainability uncertain. Furthermore, implementation relied on clinicians’ voluntary after-hours engagement to provide feedback, which may affect scalability under routine service conditions. When scaled, this model may require explicit workflow design (eg, protected time, standardized feedback templates, tiered escalation rules, or task-shifting to nurses/health educators) to avoid increasing clinician burden. Semiautomated reminders and rule-based feedback could preserve responsiveness while reducing manual workload but should be evaluated for safety and acceptability. Finally, participants were recruited from a single district-level hospital in Beijing, which may limit generalizability to other regions or lower-level facilities.

In addition, internal validity threats should be considered. First, the nonrandomized allocation and the small sample size increase the risk of selection bias and residual confounding (eg, baseline motivation, family support, and prior digital health engagement). Second, contamination is possible if participants shared information across groups. Third, key behavior outcomes relied on self-report and may be influenced by social desirability or Hawthorne effects. Future studies should use randomized (or cluster-randomized) designs with allocation concealment where feasible, incorporate objective behavioral indicators (eg, wearable activity data or clinic-verified glucose logs), and include longer follow-up periods to assess maintenance.

Equity considerations are also important. The digital component may preferentially benefit women with higher digital literacy, stable smartphone access, and supportive household environments. Women with limited autonomy, heavy caregiving/work burdens, or privacy concerns may be less able to log consistently or engage in provider messaging. Future implementation should include equity-oriented adaptations (eg, simplified logging, optional voice input, family-inclusive education when appropriate, and alternative low-tech reinforcement pathways).

Future studies can test longer logging periods (eg, 4‐12 wk) and/or periodic “booster” check-ins (eg, weekly prompts or trimester-based logging windows) to evaluate maintenance, burden, and engagement trajectories over time. Future research should evaluate this combined model in larger, multisite randomized controlled trials with longer-term follow-up to determine the sustainability of effects. Incorporating objective measures of practice changes, such as wearable physical activity trackers or biochemical indicators, would enhance validity beyond self-reported outcomes. Given that external and family environments strongly influence maternal practices, future studies should also explore family-inclusive education models. Moreover, investigating adaptation in resource-constrained settings will be essential for assessing scalability and equity. Finally, research into the mechanisms behind digital logging, such as accountability, social motivation, and feedback loops, could inform the design of more targeted and effective digital components.

This study provides preliminary evidence that integrating experiential antenatal classes with digital logging is a feasible, acceptable, and potentially effective model for promoting maternal health practices in China. The intervention improved knowledge, adherence, and satisfaction, with qualitative insights clarifying mechanisms of practice changes. While further research with larger samples is needed, the combined experiential-digital model offers a promising pathway for bridging the gap between knowledge and sustained health practices in ANC.