This systematic review was conducted following the guidelines of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [27]. The study protocol was registered in PROSPERO (the International Prospective Register of Systematic Reviews) as CRD42021258541.

For this study, 4 English databases (PubMed, the Cochrane Library, CINAHL, and PsycINFO) and 2 Chinese databases (CNKI and Wanfang) were electronically searched to identify randomized controlled trials (RCTs) on the effectiveness of telehealth interventions for women with PPD from inception to February 28, 2021. Advanced searches were performed using a combination of two groups of terms according to the syntax rules of each database: (1) telehealth-related terms, including telehealth, telemedicine, telecommunication, telephone, remote consultation, information technology, mobile health, m-Health, e-Health, internet, web-based, social media, application, and software, and (2) PPD-related terms, including postpartum depression, PPD, post-partum depression, postnatal depression, post-natal depression, maternal depression, postpartum mental disorder, and puerperal disorder. In addition, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform, and the Chinese Clinical Trial Register were searched for unpublished trials relevant to this review.

To be eligible, RCTs had to meet the following criteria: (1) target adult women with EPDS scores ≥9 points; (2) use telehealth interventions including mobile phones, apps, websites, or other remote technologies compared with routine care in the control group (participants in the control group were inaccessible to any telehealth technologies, but were free to receive routine care including any offline treatment at public hospitals or maternal and child health care centers); (3) assess the primary outcome of depression symptoms using EPDS, and secondary outcomes including any improvement of social support, loneliness, and anxiety using any scale; and (4) be published in English or in Chinese.

Studies were excluded for the following reasons: (1) study included women with severe physical illnesses, a history of mental illnesses and medication treatments, drug and/or alcohol abuse, or suicidal tendency; (2) study included women whose infants had adverse neonatal outcomes such as 5-minute Apgar score <5 points, assisted ventilation for more than 6 hours, neonatal seizure, birth injury, or neonatal death; and (3) study was an RCT protocol or duplicate.

The software Endnote X9 (Clarivate Analytics) was used to import all the references and remove duplicates. The remaining studies were assessed against the inclusion and exclusion criteria by two independent reviewers (LZ and JC). Study selection was conducted in a stepwise manner. First, titles and abstracts of all studies were independently screened for potential eligibility. Any disagreements were discussed until consensus was reached. Second, the full papers of all included abstracts were independently assessed. Any discrepancies that arose during the assessment were resolved by a third reviewer (RHX).

The risk of bias was assessed according to the guidelines provided by the Cochrane risk-of-bias tool for randomized trials (version 2.0) [28]. Risk ratings of “low risk,” “unclear,” and “high risk” were assigned to each type of bias based on the presence of selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Any disagreements with respect to study quality were resolved by a third reviewer (RHX).

The data were extracted independently by two reviewers (LZ and JC) at the same time and disagreements were resolved by consensus. Data extracted from each relevant trial included author, the year of publication, country, participant, inclusion criterion (EPDS scores), sample size, telehealth technology, specific therapy, follow-up time, and outcomes (primary and secondary outcomes) with measure scales. Any disagreements between the two reviewers were resolved by a third reviewer (RHX).

RevMan 5.4 software (The Cochrane Collaboration) was applied in the meta-analysis of the data. The effect estimate was expressed as means and standard deviations (SD) for continuous data. The standardized mean differences (SMD) with their corresponding 95% CI were applied to combine studies that measured the same outcome with different scales. If the same scales were used to evaluate one outcome, mean difference (MD) with its 95% CI could be employed. MD was derived from inverse variance methods. If the SD was not reported, it was computed from standard error (SE) following the Cochrane Handbook [28]: . Heterogeneity was assessed by I², with significant statistical heterogeneity as I²>50%. The fixed-effect model was used to estimate one true effect in cases without significant heterogeneity (I²<50%), whereas the random-effect model was employed to estimate the effects in cases with significant heterogeneity between studies (I²>50%). When there was significant heterogeneity, data would be double-checked and then subgroup analysis or sensitivity analysis was performed to explore the sources of heterogeneity. A P value <.05 was considered statistically significant.

The PRISMA flow diagram for this study is shown in Figure 1. The search strategies yielded 1001 potentially relevant citations from the 6 electronic databases searched. After excluding duplications and screening titles and abstracts for eligibility, 31 studies were retained for full-text evaluation. Of them, 22 studies were excluded according to the inclusion and exclusion criteria. A total of 9 studies [29-37] were included in this systematic review.

This systematic review included 9 RCTs [29-37] with a total of 1958 participants. The 9 studies included were conducted in Singapore, Canada, Portugal, Iran, China, and the United Kingdom and were published between 2003 and 2020. Participants were postpartum women aged ≥18 years. All recruited participants had baseline EPDS scores ≥9 points. The sample sizes ranged from 42 to 910. The 9 included studies applied telehealth technologies including telephones [29-31,34,36,37], apps [33,37], and websites [32,35,36]. Of them, 3 studies [29,30,37] used peer support therapy, 5 studies [32-36] used cognitive behavioral therapy and behavioral activation therapy, and 1 study [31] used interpersonal therapy. The follow-up period of these 9 studies [29-37] ranged from 4 weeks to 36 weeks after completion of telehealth interventions. Characteristics of the 9 studies included are presented in Multimedia Appendix 1.

We summarized the findings for risk of bias in Figure 2 and Multimedia Appendix 2. Among the 9 RCTs [29-37], 5 studies [30-33,37] used a computer-generated random sequence, 1 study [34] used a random number table, 2 studies [35,36] used a minimization algorithm including a stochastic element, and 1 study [29] only mentioned the word “random,” with no details of the randomization method used. Allocation concealment was done in 5 studies [29,34-37]. Furthermore, 1 study [37] implemented the blinding of participants and personnel and outcome assessment, 4 studies [32,33,35,36] were unclear about the blinding of outcome assessment, and 4 studies [30,31,33,34] had high risk of bias on implementing the blinding of participants and personnel. All 9 studies included had low risk of bias on incomplete outcome data, selective reporting, or other bias.

In relation to the primary outcome (ie, depressive symptoms improvement), the pooled estimates were consistent when excluding one study at a time, indicating that the meta-analysis results of this study were stable and reliable. A series of subgroup analyses were performed to uncover more information about the heterogeneity (as shown in Table 1). Subgroup analyses were performed according to the severity of PPD, telehealth technology, specific therapy, and follow-up time. However, considerable heterogeneity (I²≥71%) existed in each subgroup analysis.

Our systematic review summarized the effectiveness of telehealth interventions on PPD and associated maternal mental health problems in the postpartum period (defined as ≤12 months after childbirth). We found that telehealth interventions could significantly improve depression and anxiety symptoms, although their effectiveness in improving social support and reducing loneliness was less certain.

In this study, we demonstrated that maternal depression scores were significantly lower in the telehealth group compared to the control group. Previous studies [30,34,38] were inconsistent regarding whether telehealth could improve maternal depression in the long term. One study [38] reported that the EPDS scores of women in the telehealth group increased at 24 weeks after interventions. Another study [30] discovered no significant difference in EPDS scores at 24 weeks after interventions between the two groups. However, one study [34] reported that EPDS scores at 24 weeks after interventions were significantly lower in the telehealth group. The synthesized result of this systematic review suggested that telehealth was effective in reducing EPDS scores at 24 weeks after the interventions, which was consistent with an earlier systematic review [39] comprised of 7 RCTs with a total of 1106 participants.

In this study, we also summarized the effectiveness of telehealth interventions on mental health issues associated with PPD including social support, loneliness, and anxiety. However, telehealth interventions were not significantly effective at improving social support and loneliness in women affected by PPD. This may be related to the use of different assessment tools and the small number of included studies in the literature reporting on these outcomes. In an internet-based peer therapy project in Singapore [37], the authors found that the degree of loneliness in depressed women was reduced while social support was increased after an online peer support intervention that used telephones and apps. On the other hand, the symptoms of anxiety were significantly reduced in the telehealth group, updating the results in the study which found no evidence on the effectiveness of telehealth for anxiety [40]. Coexistence of PPD and anxiety was common [41] and was mainly related to negative life events experienced by mothers with inadequate social support and increased childcare burden [42]. Results of this systematic review suggested that the potential for telehealth to improve mental health care for either PPD or anxiety is being increasingly recognized by affected women and health care providers [43]. Many women have already used publicly available online apps to access informational support, to consult a team of specialists, or to seek and find resources to alleviate PPD and anxiety [44].

This review has multiple strengths. The results of subgroup and sensitivity analyses suggested the findings are robust. More than half of the included trials were of high quality, with a relatively high degree of evidence that telehealth interventions could be effective in PPD treatment. Furthermore, the studies included were conducted in both developed and developing countries, expanding generalizability. Finally, this review analyzed not only the effectiveness of telehealth interventions on PPD, but also on social support, loneliness, and anxiety.

However, there are several limitations in this review. First, there could be selection bias in the original studies as most women participated in the studies on a voluntary basis and were recruited online. Second, most of the included studies relied on self-report measure scales at either recruitment or follow-up, which may lead to inflated estimates of effect sizes. Third, the findings of this meta-analysis were limited by major heterogeneity. There was methodological heterogeneity among the studies included in terms of the severity of PPD, telehealth technology, specific therapy, and follow-up time. In addition, the secondary outcomes in the 9 RCTs included were assessed using different scales.

Through telehealth services, women could have access to the relevant knowledge of psychological interventions anytime and anywhere. The anonymity of chat rooms in telehealth services could help protect women’s privacy, providing a new treatment option for women who do not want to receive a face-to-face treatment due to social stigma.

This review highlights some directions for future research, including increasing research attention on antenatal and peripartum depression and determining the applicability of telehealth interventions for adolescent mothers who may be more comfortable with novel technologies. Consideration of an intervention including the mother’s partner may be an important approach in future research, especially in cases where triggering events (such as adverse infant outcomes, major negative life events) may affect both the woman and her partner. Future studies should also aim to enhance methodological quality through consistent design and execution on such aspects of the study including telehealth technology, follow-up time, and severity of PPD.

Telehealth interventions could effectively reduce the symptoms of depression and anxiety in women with PPD. However, better designed and more rigorous large-scale RCTs targeting specific therapies are needed to further explore the potential of telehealth interventions for PPD.