Introduction

Up to 20% of pregnant women are affected by mental health disorders such as depression and anxiety, during pregnancy or in the first year after giving birth (Bauer et al. 2014; PwC Consulting 2019). This has important implications for the mother’s mental health, infant attachment and wider family relationships, the mother’s partner or other children within the family unit (PwC Consulting 2019). Additionally, there are impacts on productivity through direct and indirect means (including healthcare costs), with total costs estimating up to $877 million dollars during the first year after birth in Australia (PwC Consulting 2019). Identifying and addressing mental health concerns in a timely manner with prompt and appropriate referral services for pregnant and postpartum women is vital.

There are several recommended validated measures for screening for perinatal psychological wellbeing in order to facilitate prompt referral and management for women at increased risk. Common validated screening tools include the Edinburgh Postnatal Depression Scale (EPDS) (Cox et al. 1987), Antenatal Risk Questionnaire (ANRQ) (Austin et al. 2013), Patient Health Questionnaire (PHQ-9) (Kroenke et al. 2001); Whooley Questions (Whooley et al. 1997) or General Anxiety Disorder Assessment (GAD-7) (Spitzer et al. 2006). Implementation of screening tools varies depending on the measures used and how they are implemented.

Perinatal mental health screening has primarily been undertaken as a clinical assessment or using paper-and pen-based assessments for validated measures, often conducted in clinics or during home visits. Barriers to perinatal mental health screening include limited mental health education and training for midwives and obstetricians, shortage of resources, time constraints, patient/provider interaction, and systems level issues such as cost and location (Kim et al. 2010; Byatt et al. 2012). In addition, pen and paper-based assessments are often time consuming and prone to scorer error between 13.4% and 28.9% (Matthey et al. 2012), and may involve time delays for processing reports within a clinic setting.

Digital screening for mental health in pregnancy and postpartum may provide a way to save time, reduce scorer error and increase referral and treatment for mental health issues. Digital health is increasingly being incorporated in health services across the world, can facilitate sharing of health information between patients and health professionals and across health systems and can support decision making with built in algorithms and local care pathways (Bernabe-Ortiz et al. 2008; Paperny et al. 1990; Quispel et al. 2012). Digital screening as defined in this systematic review is the use of valid and reliable screening tools such as the EPDS (Cox et al. 1987) or ANRQ (Austin et al. 2013) used in digital or electronic format (e.g., mobile phone, tablet, laptop, desktop computer, through mobile applications or web link) completed by women in pregnancy and postpartum (up to 36 months).

To the authors’ knowledge, there are no systematic reviews that explore digital screening for mental health in pregnancy and postpartum. Therefore, this review aims to determine if digital screening for mental health in pregnancy and postpartum is acceptable, feasible, and more effective than standard care (e.g., paper-based psychological assessments; no screening). Effective screening accurately detects symptoms of mental health conditions in pregnancy and postpartum (or accurately identifies women at elevated likelihood of currently experiencing a mental health condition), leading to an appropriate referral for further assessment being made. In practice, this usually means screening for depression and anxiety as recommended in clinical guidelines as the most common mental health conditions in the perinatal period. Feasibility results in quicker administration time, increased screening capacity, reduced scoring error, generated individual tailored clinical and patient reports, prompted referrals for the treatment of depression and anxiety and technology accessible and easy to use. Acceptability and feasibility will be determined by information reported by both women and healthcare professionals (HCPs) and effectiveness will be determined by good internal consistency (e.g., Cronbach’s α), comparison groups and cross-cultural considerations. For this review, HCPs refers to professions involved in women’s perinatal care, including doctors, midwives, obstetricians, nurses, psychologists, and psychiatrists.

This systematic review also aims to determine what the barriers (e.g., challenges) and enablers (e.g., facilitators) are to implement digital screening for mental health in pregnancy and postpartum and provide recommendations for best practice digital perinatal mental health screening.

Method

Search strategy and selection criteria

A study protocol was registered with PROSPERO (CRD42020198372) (https://www.crd.york.ac.uk/prospero/#recordDetails). The search located research literature within the last 21 years (from 2000 to 2021) on digital screening for mental health in pregnancy and postpartum. This time period was chosen to reflect the introduction of digital technology around the world. This review placed no restrictions on language. In total, eight databases were searched: OVID MEDLINE, PsycINFO, SCOPUS, CINAHL, Embase, Web of Science, Joanna Briggs Database and All EMB reviews incorporating Cochrane Database of Systematic Reviews (OVID). The search strategy included terms for digital health, screening, mental health, pregnancy and postpartum, including but not limited to mobile technologies, self-report, psychiatric disorder, peripartum period and postpartum period. A full list of search terms is available in Supplementary File 1). The end date of the search was 23rd July 2021.

Studies were included if they met the following criteria: (1) participants were (a) women of birthing age or had given birth or were currently pregnant or (b) Healthcare Professionals; (2) mental health screening was conducted using digital technology (e.g., tablets, mobile phones, online survey link, computers) and validated tools (e.g., EPDS); (3) have comparison groups including no screening, paper-based screening, clinical assessment only, non-validated symptom assessment, psychosocial assessment without symptom assessment, or no comparison group; (4) outcomes included barriers and enablers to digital screening within the Theoretical Domains Framework (Cane et al. 2012); acceptability, feasibility, effectiveness, efficiency, cost and sustainability of digital screening; symptoms of anxiety or depression; presence of psychosocial risk factors, with the proportion of women meeting threshold scores to be considered at risk; mean or median scores; (5) study type included systematic reviews (with or without meta-analysis) with a quality or risk of bias assessment; longitudinal cohort studies; cross-sectional studies; case control studies; qualitative studies; evaluations; medical records audits; administrative data; randomised control trials and before and after studies and (6) included all languages, information after the year 2000 and no sample size limit.

The use of paper-based psychological assessments or clinician administered assessments uploaded into the Electronic Medical Record (EMR) or Electronic Health Record (EHR) were not considered as digital screening for this systematic review. Clinical decision support systems, algorithms and machine learning were only included if they involved the use of a psychological assessment in digital format.

Studies were excluded if they met the following criteria:

  1. (1)

    The study explored only men’s or father’s experiences

  2. (2)

    The study only included women with post-traumatic stress disorder or other pre-existing mental health issues

  3. (3)

    Used non-validated tools

  4. (4)

    Had family violence as a sole outcome measure

  5. (5)

    Were a conference abstract, commentary, editorial, narrative review, position statement or a non-research letter

Metric definitions

Acceptable intervention

– determined as reported by women and HCP’s.

Feasible intervention

– determined by quicker administration time, increased screening capacity, reduced scorer error, generated individual tailored clinical and patient reports and prompted referrals for the treatment of depression and anxiety, technology accessible and easy to use.

Effective intervention

– determined by accurately detecting symptoms of depression and anxiety in pregnancy and postpartum (or accurately identifies women at an elevated likelihood of currently experiencing depression and anxiety), leading to an appropriate referral being made. Will be determined by good internal consistency (Cronbach’s α), comparison groups and cross-cultural considerations.

Study selection

One author (JC) independently assessed the title, abstract, keywords and full-text of every article retrieved against the defined selection criteria. Two authors (RN & MS) shared the role of second reviewer of all studies that met criteria. Any disagreement at both the title and abstract review and full-text review stage was resolved by discussion with the second reviewers to achieve 100% consensus.

Data extraction

The study characteristics for the 34 full-text articles included author and year, country and setting, ethnicity, study population, sample size, age of participants, research objectives, recruitment strategy, key inclusion and exclusion criteria of the studies, digital mode, methodological and theoretical approach, method and duration of data collection, data analysis, study design, rate of attrition, study findings, outcomes and power calculations, preliminary TDF domain and quotes from qualitative and mixed-methods studies. Data extraction information was recorded on one Excel spreadsheet. Five authors from the included studies were contacted for additional information and added to the data extraction.

Risk of bias assessment

One author (JC) independently assessed risk of bias using assessment templates suitable for the included studies, including the Critical Appraisal Skills Programme template for qualitative studies (CASP 2018), MCHRI risk of bias templates for quantitative studies (MCHRI 2013; MCHRI 2014) and the Mixed Methods Appraisal Tool (MMAT) risk of bias template for the mixed methods and remainder of the studies (e.g., Quantitative – Descriptive pre-post-test) (Hong et al. 2018). The risk of bias templates assessed the studies’ internal and external validity such as the use of appropriate study design, inclusion and exclusion criteria, reporting bias, confounding, sufficient power analyses and any conflicts of interest. Using a descriptive approach, the studies were given a rating of low, moderate, high or unclear risk of bias. Twenty percent of the studies (7 articles) were reviewed by author MM. Authors JC and MM discussed the risk of bias and evaluation methods used until 100% consensus was reached.

TDF framework

The Theoretical Domains Framework (TDF) (Cane et al. 2012) provides an integrative theoretical framework for the evaluation of behaviour change and implementation across disciplines within the healthcare industry. It comprises of 14 key domains (e.g., Knowledge) with 83 constructs. The findings of the systematic review were mapped to the TDF domains and constructs to identify the barriers and enablers digital screening has for mental health in pregnancy and postpartum for both women and HCP’s.

The TDF framework is a valid framework that explores both individual and organisational aspects of implementation research and is effective in providing theory informed and evidence-based support for healthcare interventions (e.g., Michie et al. 2008; Cane et al. 2012; Francis et al. 2012; French et al. 2012).

It has been used previously to evaluate perinatal mental health screening (Nithianandan et al. 2016) (Table 1 and 2).

Table 1 Summary of the study characteristics of the 34 included full-text articles in systematic review
Full size table
Table 2 Summary table of effectiveness, feasibility and acceptability of digital screening in pregnancy and postpartum
Full size table

 Data extraction using TDF

Step 1: Data extraction

  • First review author (JC) independently identified and extracted information from the included qualitative and mixed-methods studies about women’s and HCPs perceptions and experiences of digital screening for mental health. Extracted data from 12 of those studies were recorded in an Excel spreadsheet (one spreadsheet per study). Each data point was categorised as either (1) raw data (e.g., participant’s quotations from qualitative studies); (2) analysed data from the results sections (e.g., thematic analysis) or (3) interpretive descriptions and summaries from results.

    Information from the included studies with a qualitative component were in the form of single quotes, several quotes or paragraphs deemed appropriate within a particular author’s key theme regarding digital screening. Data extracted from the studies included key themes and sub-themes in regards to digital screening. Specifically, it included the author’s interpretation or description of the key theme (verbatim author interpretation), the quote (with page number of article) and if the extracted data was considered a barrier, enabler or both to digital screening.

Step 2: Data coding

  • Extracted data was deductively coded (i.e., mapped) to the TDF by author JC according to the TDF domain and construct that they were determined to represent. A TDF Coding Manual developed by authors JC and JB assisted with identifying the key domains, constructs, barriers and enablers to digital screening for mental health in pregnancy and postpartum at a theoretical level.

  • Step 3: Data checking

    A second author (MS) independently reviewed all of the included qualitative and mixed-methods studies to determine author agreement and consensus on the TDF coding using the TDF Coding Manual. Any disagreement or uncertainty was discussed until 100% consensus was reached.

  • Step 4: Presentation of findings

    Results from the TDF domain coding of the included full-text articles are presented in Table 3. The TDF domains and constructs were counted in frequencies to reflect their importance within those categories and key themes and may include single or multiple TDF domains and constructs (Atkins et al. 2017).

Step 5: Recommendations

  • Recommendations for the implementation of digital screening for mental health in pregnancy and postpartum were developed using Michie et al.’s (2008) matrix which maps theoretical domains (e.g., behavioural determinants) to effective behaviour change using an expert consensus, combined with the TDF domains (Cane et al. 2012). The authors used their multi-disciplinary clinical and research experience to provide recommendations most relevant to the settings found to address the barriers and enablers identified by the research presented in the systematic review (Table 4).

Table 3 TDF mapping of key themes regarding digital screening for mental health in pregnancy and postpartum
Full size table
Table 4 Best practice recommendations for implementation of digital screening for mental health in pregnancy and postpartum
Full size table

Results

Database searching retrieved 2,288 relevant articles. These studies were imported into Endnote reference management software (Thomson Reuters 2020) and filtered for duplication. Studies were then transferred into Covidence systematic review software for screening (Veritas Health Information, 2020), where further duplicate articles were removed, leaving 2,118 articles to be screened. There were 1,878 studies excluded after title and abstract screening and 206 studies removed after full-text screening, resulting in a final number of 34 papers (PRISMA Flowchart, Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram of systematic review

Full size image

Results

Table 1 displays a simplified summary of the study characteristics of the 34 included full-text articles of the systematic review recorded from the data extraction Excel spreadsheet. Table 2 displays a summary table of effectiveness, feasibility and acceptability. Table 3 displays the mapping of the articles to the TDF (Cane et al. 2012). Table 4 displays the best practice recommendations for the implementation of digital screening for mental health in pregnancy and postpartum, with examples to support both HCPs and women across different healthcare settings.

Risk of bias assessment

Of the 34 included studies, nine were determined to be low risk, eight were low/moderate risk and seventeen articles with moderate risk of bias (Table 1). A total of eleven studies employed a qualitative study design, twenty-one studies employed a quantitative design and two studies used a mixed-methods design (e.g., qualitative and quantitative). Of the quantitative studies, thirteen used a cohort study design (e.g., longitudinal and the participants were selected on the presence or absence of a risk factor), three were cross-sectional design, three were randomised controlled trials, one was a non-randomised controlled trial and one used a pre-test post-test study design.

Of the qualitative studies, four were determined to be low risk and seven were low/moderate risk of bias. The studies were adequately descriptive in nature, with clear outcomes and justifications for the research methodology. Several had no or limited information (e.g., Gordon et al. 2016; Doherty et al. 2018) on whether their assumptions had been adequately explored (e.g., consideration of own role, potential bias and influence during the study), however, this was not considered to be a major flaw given the information the articles reported.

Of the 21 quantitative studies, five were determined to be low risk, one was low/moderate risk of bias and 15 were moderate risk of bias. Limited information was provided in some studies both in regards to the blinding of outcome assessors to the exposure and to what percentage of individuals were not included in their results. Most of the quantitative studies followed cohorts of women over an extended period of time.

Of the two studies that were of mixed-methods study design, both were determined to be of moderate risk of bias. Limitations included inconsistencies in the reporting of the quantitative and qualitative components of the studies (Guevara et al. 2016; Diez-Canseco et al. 2018).

Key

EDS:

Edinburgh Depression Scale.

EDS-10:

Edinburgh Depression Scale (10 questions).

EPDS:

Edinburgh Postnatal Depression Scale (EPDS-2; EPDS-3; EPDS-5; EPDS-7).

EPDS-10:

Edinburgh Postnatal Depression Scale (10 questions).

EPDS (online):

Online version of EPDS (e.g., completed online using the Internet and a Laptop).

EPDS-lifetime version:

Edinburgh Postnatal Depression Scale (21 questions).

BrightSelf:

BrightSelf mHealth Application.

EMA:

Ecological Momentary Assessment (5 questions—Mood, Sleep, Worry, Enjoyment and Energy).

SRQ (WHO):

Self-Reporting Questionnaire – World Health Organisation.

mHealth:

mobile health.

Ginger.io Application:

mobile phone application (downloaded to smartphone).

PHQ-2:

Patient Health Questionnaire (2 questions).

PHQ-9:

Patient Health Questionnaire (9 questions).

GAD-2:

General Anxiety Disorder (2 questions).

GAD-7:

General Anxiety Disorder (7 questions).

EMR:

Electronic Medical Record.

EHR:

Electronic Health Record.

Global Health Scale:

Assess the participant´s mental and physical health as well as pain, fatigue, social connections, and overall health and quality of life (10 questions).

AUC:

Area Under the ROC Curve.

ROC Curve:

Receiver Operating Characteristic Curve.

Whooley Questions (2 questions).

Arroll Question (1 Question).

STAI:

State-Trait Anxiety Inventory (20-items).

STAI-S:

State-Trait Anxiety Inventory (state subscale – 20 items).

STAI-T:

State-Trait Anxiety Inventory (trait subscale – 20 items).

PRAQ-R/PRAQ-R2:

Pregnancy-related Anxiety Questionnaire revised (10-items).

PPAQ:

Pregnancy Physical Activity Questionnaire (32 activities).

SBIRT:

Screening, Brief Intervention, Referral to Treatment (SBIRT) framework.

AAS-2:

Abuse Assessment Screen (2 items).

NIDA Quick Screen:

National Institute on Drug Abuse (Quick Screen).

AUDIT-C:

Can help identify patients with alcohol misuse (3-question screen).

Pre-Pregnancy BMI:

Pre-pregnancy Body Mass Index.

GWG:

Gestational Weight Gain.

GTT:

Glucose Tolerance Test.

Godin Shepard:

Godin Shephard Leisure-Time Physical Activity Questionnaire.

Insomnia Severity Index:

Insomnia Severity Index outcome measure.

Tablet-based:

Using a i-Pad; iPad computer.

iOS App:

Mobile Operating System (Apple Inc.)

PMD:

Perinatal Mental Disorders.

PPD:

Post-partum Depression.

iCOPE:

Digital Screening Platform (Centre for Perinatal Excellence).

EPQ-N:

Eysenck Personality Questionnaire – Neuroticism (12-items).

EPQ-R:

Eysenck Personality Questionnaire – Revised (48-items).

eDPPMobile Application developed by Jiménez-Serrano et al. (2015).

ML:

Machine Learning.

PR:

Pattern Recognition.

ANN:

artificial neural network.

SVM:

support vector machines.

ICC:

Intraclass correlation coefficients.

WHO-5 Wellbeing Index:

World Health Organisation – Wellbeing Index (5-questions).

Cambridge Worry Scale:

16-item questionnaire assessing worry.

IVR:

Interactive Voice Response (technology).

ALPHA:

Antenatal Psychosocial Health Assessment (15 risk factors).

MINI:

Mini International Neuropsychiatric Interview.

ITT:

Intention to Treat Analysis.

Questback:

Anonymous Online Questionnaire administered by Questback (http://www.questback.com).

IPTW:

inverse probability of treatment weighting.

CI:

Confidence Interval.

NHS:

National Health Service.

NICE:

National Institute for Health and Care Excellence.

Snap Mobile App:

Snap Mobile Application (for Apple iOSTM running on Apple iPad Air and Apple iPad mini tablet computers. Responses were stored in Snap WebHost.

RCT:

Randomised Controlled Trial.

ANCOVA:

Analysis of Covariance.

Promote W:

Electronic Psychosocial Screening and Referral Tool (standardised screening tools).

WHO-QOL:

World Health Organisation’s WHOQOL-BREF Scale (26-item version of the WHOQOL-100 assessment).

Patient Navigator:

Person who assists patient in a clinic environment to navigate the health care system and needs of patients.

GyPsy:

GyPsy (in Dutch), derived from Gynecology and Psychiatry).

PDA:

Personal Digital Assistant (self-report screening; hand-held computer).

HM-Web & HM-App:

HappyMom Web & App versions.

ERQ:

Emotion Regulation Questionnaire (10-items).

CAE:

Cuestionario de Afrontamiento del Estrés = Stress Coping Questionnaire (42-items) (Martinez-Borba et al. 2019).

QLI:

Quality of Life Index (33-items).

SRSS:

Social Readjustment Rating Scale (43-items).

ODK:

Open Data Kit Software (android application).

MUAC:

Mid upper arm circumference.

HemoCue Hb 301 system:

Haemoglobin concentration.

Household Food Insecurity Access Scale:

Household food insecurity.

IPV (HITS assessment):

Intimate Partner Violence (Hurt, Insult, Threaten and Scream).

MSSS:

Maternity Social Support Scale.

GHS:

Global Health Scale.

WAST:

Women’s Abuse Screening Test (5-items).

HFIAS:

Household Food Insecurity Access Scale (9 questions).

MPAS:

Maternal Postnatal Attachment Scale (19 items).

Overall summary

Twelve studies examined the effectiveness of digital screening (Studies 3,6,9,10,15, 18,19,21,23,28,30,31). Eight studies explored the acceptability and feasibility of digital screening for mental health in pregnancy and postpartum (Studies 14,16,22,25,27,29,32,34). Some studies explored effectiveness, feasibility and acceptability (e.g., Study 23; Kingston et al. 2017). The remaining 14 studies explored the lived experiences of women and healthcare professionals, the prevalence of depression and anxiety among women, mental health care use and referral for services, the development of digital screening tools and the implementation and effectiveness of digital screening within healthcare systems (Table 1).

Method and type of digital screening

Overall, the total sample size across the 34 included studies included 32,859 participants. Screening methods varied with completion primarily via a tablet, computer or mobile phone application. The EPDS (Cox et al. 1987) was the main psychological assessment used to assess depression and anxiety symptoms, with 20 of the 34 included studies using some form of the EPDS as their primary psychological assessment (e.g., EPDS-10; EPDS-lifetime version; validated for specific cultures). The PHQ-2 (Löwe et al. 2005) or PHQ-9 (Kroenke et al. 2001) was also commonly used (n = 11), often in conjunction with the EPDS or other assessment measure as they were found to be more feasible and easier to implement.

Many researchers developed their own applications, with adaptive features, with some applications focused on the personal subjective experience of women and encouraged support-seeking behaviours, such as prioritising the midwife-client relationship (BrightSelf-App; Diez-Canseco et al. (2018). Others focused on the ability to screen the patients for depression and other concerns (MatCHAT; Wright et al. 2020), and undertake risk prediction (eDPP Predictor; Marcano-Belisario et al. (2017). Some had the ability to screen women in multiple languages and produce both clinical and patient reports (iCOPE; Highet et al. 2019) and one enabled responses to vocal prompts using Interactive Voice Response (IVR) technology (Kim et al. 2007). Other applications were able to screen and manage symptoms of perinatal depression and promote wellness during pregnancy (VeedaMom; Dyurich & Oliver (2020) and some provided screening and advice (GyPsy Screen and Advice; Quispel et al. (2012) & iCOPE (Highet et al. 2019).

One study assessed the relationship between the survey layout (e.g., paging or scrolling on the app) and screening with scrolling resulting in a slightly faster completion time (median = 4 min 46 s) than a paging layout (median = 5 min 33 s) (Marcano-Belisario et al. 2017). Another study compared platforms web (HappyMom-Web) or mobile (HappyMom App; HM-App; downloaded app) longitudinally over pregnancy and post-partum. A higher proportion of women responded at each time point to the HappyMom Web sample (27.3–51.1%), compared to the HappyMom-App sample (9.1–53.1%), possibly because involvement was supported by HCPs for the web-based program. However, whilst longitudinal retention was low for both it was slightly higher for the app (9.1%) compared to the web platform (4.6%) (Martinez-Borba et al. 2019).

Effectiveness of digital screening for mental health in pregnancy and postpartum

The 12 studies (3,6,9,10,15,18,19,21,23,28,30,31) that assessed the effectiveness of digital screening reported effectiveness in detecting and referring women for mental health treatment in pregnancy and postpartum, with good internal consistency (Cronbach’s α = 0.88–0.90, Quispel et al. 2012; Cronbach’s α = 0.81, Drake et al. 2014; Cronbach’s α = 0.89, Highet et al. 2019). It was also effective when compared to paper-based screening, measured with an adapted version of Renker and Tonkin’s tool of feasibility and acceptability (Kingston et al. 2017) and in different languages (Quispel et al. 2012; Tsai et al. 2014; Diez-Canseco et al. 2018; Fontein-Kuipers & Jomeen 2019; Kallem et al. 2019).

Acceptability and feasibility of digital screening for mental health in pregnancy and postpartum

There were eight studies that assessed the acceptability and feasibility of digital screening for mental health. Digital screening for mental health was acceptable and feasible (Martinez-Borba at el., 2019) to both women (Hahn et al. 2021; Kim et al. 2007; Marcano-Belisario et al. 2017; Poleshuck et al. 2015; Willey et al. 2020 and Wright et al. 2020) and HCPs (Guevara et al. 2016; Wright et al. 2020). Digital screening was found to be acceptable across cultures and countries (e.g., North America, United Kingdom, Spain, Australia), healthcare settings (e.g., Public Health, Community Health Clinics, Antenatal Clinics, Hospital Settings) and using various delivery options, suggesting generalisability of the results to the wider population (Table 1).

Research in a community maternal and child health setting found that completing the EPDS and Psychosocial Questions on a tablet enabled women to complete screening themselves in a timely manner, with reduced scorer error (e.g., reverse scoring of EPDS items; Matthey et al. 2012) and 100% accuracy. An automated tailored plain language report sent to women by SMS or email reported risks and directed relevant health information and available health services. This facilitated the health professional consultation and supported self-management at home. Clinical summaries prompted referrals when required and provided scores saving time for health professionals enabling more time for discussion with women (Highet et al. 2019). Screening of parents post-partum over 20-months in the USA by Guevara et al. (2016) utilised both paper-and-pencil and electronic versions of the PHQ-2 within EHR that incorporated electronic screening alerts and a check box for service referrals. The use of electronic alerts reminded clinicians when to screen patients, facilitated screening and included suggested language for explaining the results to parents. Use of alerts increased screening from 12.8% of eligible parents to 54.5% and interviews with clinicians identified that alerts were of benefit in reminding them when screening was due and that the electronic discussion points and automatic scoring of the depression tool facilitated screening (Guevara et al. 2016).

Kallem et al. (2019) found digital screening completed as part of routine care at the 2-month well child check beneficial and effective in identifying women at risk of mental health concerns. Mothers completed the screening via a tablet in the waiting room, with the results of the screen presented in the child’s EHR.

Table 2 displays a summary table of the effectiveness, feasibility and acceptability of digital screening in pregnancy and postpartum. Table 3 displays the TDF mapping of key themes regarding digital screening for mental health in pregnancy and postpartum.

Barriers and enablers to implementation for digital screening in pregnancy and postpartum

Results of the systematic review were mapped to the TDF (Cane et al. 2012) to identify barriers and enablers, as well as key themes (Table 3). The three main TDF domains identified included Social/professional role and identity, Emotion and Environmental context and resources.

Social/professional role and identity

Social/professional role and identity refers to HCPs ability to do their job effectively, including the requirements of their job and the belief that digital screening is a part of their role, which enables them to implement it effectively. The most prominent constructs included professional role, professional confidence and professional boundaries. Key barriers were the ability to which the HCP’s thought that digital screening was part of their role and what it consisted of on a daily basis (Pineros-Leano et al. 2015), the level of confidence that the HCPs or the women had in their ability to complete digital screening effectively (Doherty et al. 2020) and if the women felt that their HCP’s were acting appropriately within their professional boundaries (Johnsen et al. 2018).

Emotion

Emotion refers to the use of digital screening as a tool for women to help express their emotions during pregnancy and postpartum, which was considered a key enabler. The most prominent constructs were affect, anxiety, depression, fear and stress. This was reflected in the ability for a digital screening platform to encourage women to identify, label and express their emotions effectively during the pregnancy and postpartum period (Barry et al. 2017; Dyurich and Oliver 2020; Willey et al. 2020) and seek further knowledge (e.g., watching videos) and social support.

Environmental context and resources

Environmental context and resources were key barriers to digital screening. They highlighted the importance of the environment in which a woman completed the digital screening and the resources provided by the healthcare professionals and organisations. The most prominent constructs were resources/material resources and person and environment interaction. This was reflected in the importance of the availability and accessibility of technology (e.g., computer, tablet, mobile phone) (Pineros-Leano et al. 2015; Gance-Cleveland et al. 2019), room (i.e., available or separate), available staff, finances (i.e., organisations or women), organisational support and workload pressure to complete many digital screening assessments (Diez-Canseco et al. 2018).

Discussion

This systematic review explored the acceptability, feasibility and effectiveness of digital screening for mental health in pregnancy and postpartum, as well as barriers and enablers to the implementation and best practice recommendations for future clinical practice.

Acceptability, feasibility and effectiveness

The review found good evidence that digital screening for mental health in pregnancy and postpartum is acceptable and effective for women and HCPs and is feasible to undertake in clinical practice, providing a better alternative to standard care (e.g., paper-based screening; Kingston et al. 2017), across a variety of cultures and healthcare settings. Valid and reliable screening measures, with the EPDS (Cox et al. 1987) being the primary assessment measure of choice were able to be completed using digital platforms with accuracy by both women or HCP’s. Digital screening provided quicker administration time, increased screening capacity, reduced scoring error, generated clinical and patient reports and prompted referrals for the treatment of depression and anxiety (Highet et al 2019). The choice of user interface (app or web-based) may influence the implementation and uptake of the digital screening. However, these studies were for women completing screening at home at multiple time points and the relevance for screening in a clinical context may not apply (Martinez-Borba et al. 2019).

Barriers to implementation of digital screening in pregnancy and postpartum and best practice recommendations for future clinical practice

Barriers to the implementation of digital screening included skills and social/professional role and identity of HCPs. This related to their role in screening and identifying women with anxiety or depressions but also their role in using digital platforms. It is important to support HCPs to increase their knowledge of digital screening, through education, training, clarity around the scope of their practice and time constraints (Bayrampour et al. 2018) as well as supporting HCPs less literate in technology and for non-regular staff unfamiliar with the technology. However, research has found digital screening and assessment is favourable and comfortable among midwives and women in general (Schmied et al. 2020) and particularly during the COVID-19 pandemic (Martin-Key et al. 2021). HCPs should reassure women regarding their beliefs about the consequences of completing digital screening, such as outcome expectancies and anticipated regret through information provision.

Environmental context and resources can also provide barriers to the implementation of digital screening, with the main area of concern being the resources/material resources and the person and environment interaction. Key barriers at an organisational level include the lack of available technology and increased workload for HCPs. Women who completed the digital screening did not find many barriers to technological issues, however, issues that were encountered were overcome with assistance from staff at healthcare facilities. Women from Culturally and Linguistically Diverse (CALD) backgrounds (e.g., people who come from different countries across the world) experienced some difficulty in responding to questions on the digital screening platforms, with feelings of being uncomfortable, uncertainty of questions and embarrassment with question content (Willey et al. 2020).

Environmental context and resources are a pivotal component in the implementation of digital screening. Organisations play an important role in the effective implementation through the resources provided to complete digital screening, such as access to the digital technology used and availability of technology (e.g., iPads for women), technological support, choice of assessment measure, availability of assessment in different languages and formats (e.g., written & audio), choice of how the assessment measure is displayed, funding, how the referrals are recorded within the EHR/EMR healthcare systems and the use of electronic alerts to prompt clinicians to complete digital screening. An important consideration is the staffing within organisations and the workload required of HCPs to conduct digital screening with women if it is not self-completed. Consideration should be made by organisations as to which application they choose, any adaptations needed, requirements for local service users and any initial and ongoing costs.

Enablers to implementation of digital screening in pregnancy and postpartum and best practice recommendations for future clinical practice

Enablers to the implementation of digital screening include the knowledge it provides, timely self-completion, no scorer error, referral for social support, identification of emotions and the ability for women to self-monitor their own behaviours and emotions. Overall, women were able to complete digital screening effectively, with limited technological issues. They also found it particularly beneficial when the screening was available in their own language as it was more convenient, they were able to understand the questions more easily and were more truthful in their responses. Digital screening resultedin less embarrassment and improved privacy and supported equity among women and across cultures (Willey et al. 2020) or when completed by themselves through the use of Interactive Voice Response (IVR) technology (e.g., Kim et al. 2007), allowing women to self-enter their responses in a private clinic room.

For some women with decreased literacy, it was suggested that an audio format would further assist equity in access to screening (Willey et al. 2020). Women were receptive to being asked about their mental health state and the social support provided through referral either through digital screening or resources provided by HCPs. Digital screening allowed women to express their emotions, disclose mental health concerns, develop self-awareness and insight through self-monitoring (Dyurich & Oliver 2020). Recommendations to support women include providing them with information about digital screening, encouraging the development of realistic and achievable goals, providing appropriate support and referral pathways, adequate time for completion of digital screening and the provision of technical support if required. As best-practice guidance changes, it is possible that digital screening may be a more agile mode and adapt faster than paper-based screening. Clinical judgement is also used where indicated to assess for other conditions and the effectiveness of screening for other mental health conditions is beyond the scope of this paper.

Strengths and limitations

Limitations of the review involved the exclusion of particular study designs that may have been beneficial to include in the review, such as entirely algorithm-based digital screening. However, these were deemed not to be within the scope of this review, due to clinical decision support systems and machine learning. These were only included in the review if the psychological assessment was in digital format. Further, there were few studies that included women from CALD backgrounds (n = 15), limiting generalisability, as well as limited comparison groups due to small sample sizes and methodological approach chosen. Most of the digital based platforms used the EPDS. This is not surprising as it is currently the most widely used perinatal mental health screening measure, frequently recommended in clinical guidelines and translated and validated in a multitude of languages (Blackmore et al. 2022). However, there are some concerns with the use of the EPDS and its broad applicability such as for use with Indigenous women (Chan et al. 2021) and future studies of digital screening may need to explore other measures as the evidence base changes. Additionally, while a meta-analysis was originally planned, it was not feasible due to the small number of eligible studies. Finally, seventeen of the 34 included studies (50%) were at moderate risk of bias; while this is not a limitation of this review’s design, it does reflect a limitation of the existing evidence base and more high-quality studies are recommended. Strengths of the review included exploring research over a 21-year period in relation to digital screening for mental health in pregnancy and postpartum and theory-informed recommendations for both HCP’s and women.

Future directions

This review has identified key barriers and enablers to the implementation of digital screening and also provided recommendations for clinical practice. Future research and clinical practice should add to the literature by adapting current practice and implementing digital screening for pregnancy and postpartum in their specific healthcare settings worldwide (e.g., public, private or community), utilising the theory-informed best practice recommendations presented in this systematic review and the use of various language translations and formats. Development of new technologies (e.g., Fast Healthcare Interoperability Resources—FHIR) and mobile phone applications, including choice of layout and user interface, will be beneficial to the digital screening field for mental health in pregnancy and postpartum.

Conclusion

Digital screening provides an innovative, acceptable, feasible and effective method to screen women for mental health concerns such as depression and anxiety in the pregnancy and postpartum period. It is effective and acceptable to women and HCPs and feasible to implement in clinical care. Important enablers include support for women to understand the role and benefits of screening and provide technological assistance, as well as providing HCPs education and training about screening, how to use the digital technology and management for women at risk. Digital screening provides the opportunity for behavioural regulation through self-monitoring and empowering women to take an active role in their mental health care, referral and treatment. The provision of appropriate organisational resources and staffing is critical, enabling widespread usage, equity and access to mental health support for women around the world during the perinatal and postpartum period.