Depression and anxiety are among the most common morbidities in pregnancy, with prevalence rates of 13%-29% [1-3], and are the leading causes of maternal mortality in Western countries [4]. Without early screening and treatment, 50% to 70% of women with prenatal anxiety or depression symptoms [5] will experience persistent symptoms through their child’s early years [6,7]. Recent findings from longitudinal birth cohorts also reveal that chronicity of depression starting in pregnancy, whether severe or subclinical, is emerging as a major risk factor in child development and mental health [8-11].

In the absence of routine screening, mental health problems are severely underdetected and undertreated in perinatal settings [12,13]. This is, in part, due to the fact that the majority of pregnant and postpartum women do not volunteer information about their mental health without being prompted by a perinatal care provider [14-18]. In our previous studies, we reported that while 67% of pregnant women did not raise concerns about their mental health with their physician or obstetrician, nurse, or midwife, 97% indicated they were comfortable with provider-initiated screening [19]. Our studies have identified other common barriers that deter women from self-identifying mental health problems, including “false” reassurance that they have received from friends or family, not knowing whether their symptoms are “normal” or not during pregnancy, and stigma-induced concerns such as not wanting their care provider to see them as depressed or anxious and not wanting to be seen as a bad mother [19]. However, 79% of pregnant women reported they would disclose mental health concerns if asked as part of their routine prenatal care [19,20]. Clearly, routine, provider-initiated mental health screening plays an important role in women’s willingness to disclose prenatal mental health problems.

Studies also show that underdetection of mental health problems in the perinatal population is due to the lack of standardized screening, where up to 80% of cases remain unidentified by perinatal care providers who used unstandardized approaches for detection of mental health disorders (eg, without a validated tool) [21,22]. However, recent studies reveal that routine prenatal and postnatal screening enhances detection and increases the likelihood that women with a positive screening result for depression will link to mental health services [23,24]. Thus, antenatal mental health screening holds benefits for enhanced detection, increased linkage to services, and improved clinical outcomes.

Although current evidence and international guidelines from the United Kingdom [25], Australia [26], and the United States [27,28] support the need for antenatal screening as a key intervention for interrupting the cycle of perinatal mental disorders and their negative impact on maternal and child well-being, serious challenges exist for its implementation. Despite high acceptance by women [29-31] and providers [32-35], only 20% of North American perinatal care providers conduct proactive screening as part of prenatal care [36] and less than 15% of pregnant or postpartum women receive the help they need [37]. A systematic review conducted by our team (manuscript in preparation) found that substantial personal and system-based barriers to routine screening exist for health care providers, including lack of time to screen, lack of accurate assessment tools and knowing how to interpret them, lack of defined referral processes, the absence of connections with mental health services, and frustration with the lack of availability of timely services for their pregnant patients [38-40].

Taken together, this body of research underscores the need to develop screening processes that are feasible for and acceptable to both women and service providers, are designed to overcome barriers to implementation, and are cost-effective and clinically useful. Indeed, the most effective perinatal mental health screening and management programs are those characterized by screening processes that are incorporated into routine care with designated systems of referral and treatment that are initiated immediately after screening [26,41,42].

E-screening has potential to be an effective, low-resource screening approach that can be feasibly embedded in a variety of perinatal settings [43,44]. Importantly, it has potential to overcome the personal and system-based barriers to screening identified by pregnant women and health care providers, and it can screen for sensitive issues such as intimate partner violence [45-47] and postpartum depression [48]. It can increase efficiency of mental health care by reallocating scarce human resources to where they are most needed—in-depth follow-up assessment, referral, and treatment. It is well-suited for busy clinical settings, can be personalized to patient needs, offers audio or video options for low literacy, provides real-time data [43,44], achieves similar or greater rates of disclosure compared with interviews, and is preferred by patients because of its anonymity [44,45,49,50]. However, to date no studies have evaluated the feasibility or acceptability of e-screening as an approach to routine screening in pregnant women.

The objective of this study was to compare the views of pregnant women randomized to an intervention group and a paper-based screening control group on the feasibility (eg, disclosure of concerns about their mental health, specific features of screening) and acceptability (eg, women’s preference) of e-screening. A secondary objective was to identify factors associated with women’s preferences for e-screening and their ability to disclose mental health concerns.

The study was a parallel-group, randomized controlled superiority trial (Figure 1). The protocol has been previously published (ClinicalTrials.gov identifier: NCT01899534) [51]. Approval for this study was granted by the Human Research Ethics Board at the University of Alberta.

Pregnant women were eligible if they were (1) able to speak or read English; (2) willing to be randomized to e-screening; and (3) willing to participate in a follow-up diagnostic interview within 1 week of recruitment. Because the Web-based screening tool was intended to be completed unassisted, it was designed for use by women with varying degrees of computer literacy.

Women were recruited from 2 community-based family physician–led maternity clinics, a high-risk antenatal unit in a tertiary care center, and 2 community hospital-based prenatal classes in Edmonton, Alberta. The high-risk antenatal unit is located in the inner city and serves a demographically diverse population. One maternity clinic was located in a middle-class suburb, while the other was located in a neighborhood with a high proportion of new immigrant families. The hospital-based prenatal classes served women from across the urban center. Thus, the recruitment strategy aimed to include participants with diverse demographic and obstetric characteristics. None of the care providers at the recruitment sites had specialized training in mental health care. Additional details on the recruitment sites are available in the study’s published protocol [51].

Trained research assistants at each site used a standardized script to invite women to participate in the study. Once women completed the consent form electronically on a tablet computer, the computer program designed by the Women and Children’s Health Research Institute (WCHRI) automatically randomized them (1:1) to the intervention or control group. Thus, the research assistant was blinded to group allocation. Full details on recruitment and consent procedures are found in the trial protocol [51].

The e-screening intervention was a full psychosocial assessment including (1) the Antenatal Psychosocial Health Assessment (ALPHA) [21,35] and (2) the Edinburgh Postnatal Depression Scale (EPDS) [52]. This is aligned with international guidelines recommending that screening comprise an evaluation of risks for mental health problems (ALPHA) as well as current symptoms (EPDS) [26,42].

The ALPHA asks women questions on the topics of family life, stressors, feelings about the pregnancy, substance use, abuse, and family of origin. Items are rated as yes or no or 6-point Likert-scale options and are completed in 5 minutes. On the basis of a review of responses in each category, providers subjectively judge whether women are at low, some, or high psychosocial risk. The ALPHA has undergone extensive face and content validity testing, can readily be implemented in routine prenatal care, and pregnant women find it acceptable [35,53].

The 10-item EPDS is one of the most widely used screening instruments for detecting antenatal and postnatal depression symptoms within the previous 7 days [52]. A score of 13 or higher (range 0-30) is a well-established cutoff for clinically significant depression symptoms consistent with meeting the criteria for a major depressive episode [52]. Original psychometric testing resulted in a sensitivity of 85%, specificity of 77%, positive predictive value of 83%, split-half reliability of .8, and Cronbach alpha of .87 [52]. A screenshot of the Web-based version of the EPDS is provided in Multimedia Appendix 1.

Women randomized to the intervention group accessed the intervention for free. They completed the assessment on a single occasion and had no further access to the intervention beyond that time. Research assistants recruiting the women provided technical assistance as needed to assist women to get started on the tablet computer, although the link was labeled and readily available on the home screen. No specialized training was offered to the intervention group because completion of the Web-based e-screening tool was intended to be a largely self-sufficient, self-report option for clinical settings. No study participants were aware which group was the intervention of interest.

Women in the control group completed paper-based versions of the ALPHA and EPDS, followed by the Web-based baseline questionnaire.

Following recruitment, women completed the Web-based e-screening or paper-based version of the screening tools independently (in the recruitment setting) as well as a Web-based baseline questionnaire including (1) demographics (age, parity, marital status, education, income, ethnicity, country of birth, and length of time in Canada); (2) obstetric history (current and past, including use of fertility treatments); (3) mental health history (diagnoses, treatment); (4) level of comfort with computer technology; (5) quality of relationship with perinatal care provider; (6) level of social support, experience of talking with doctor, nurse, or midwife about emotional health; and (7) adverse childhood experiences (using ACE questionnaire [54]). With the exception of the ACE questionnaire, questions were drawn from those used in the All Our Babies Study [55,56] and the Canadian Maternity Experiences Survey [57]. The baseline questionnaire also included Renker and Tonkin’s Computer Assessment Evaluation (CAE), which is a feasibility and acceptability assessment. Logos belonging to institutional affiliations were visible on questionnaires and consent forms, as well as in the description of the primary investigator. Women in both groups were telephoned by a trained research assistant (blinded to group allocation), 1 week after recruitment, to complete a MINI International Neuropsychiatric Interview (MINI, version 6.0.0) [58].

No data were stored on the tablets. When women submitted their information, it was sent to a secure server housed in the Faculty of Medicine & Dentistry at the University of Alberta. The Web-based e-screening tool and questionnaire were built using an existing infrastructure offered by Checkbox Survey software provided by the WCHRI Clinical Research Informatics Core. The Web-based e-screening tool was tested for usability and navigability through focus groups as part of a concurrent randomized controlled trial, as described in that trial’s published protocol [59]. No changes to the tool were made after initial testing. Data transfer between the tablet and server was encrypted. Data imported to statistical databases for analysis were not identifiable.

If women met the criteria for a mood or anxiety disorder on the MINI or if they scored 13 or more on the EPDS, the research assistant made a referral for the women (with their permission) to the hospital-based reproductive mental health support program.

Because no data were available to guide estimation of a minimal clinically important difference in true cases detected through e-screening, we used a confidence interval approach [60]. On the basis of high levels of acceptability and disclosure reported using computer screening [46,47], we established that e-screening would be feasible if, in the intervention group, 85% of women indicated that they were able to tell the truth on all questions on the psychosocial assessment (question 7 of the computer assessment evaluation, CAE). For our other outcomes, we based the sample size calculation on 85% of women scoring 4-8 on the Risk subscale of the Disclosure Expectations Scale and 85% scoring 16-20 on the Utility subscale of the Disclosure Expectations Scale. Using a margin of error of .05 and 25% as an estimated loss to follow-up, we calculated that 261 women per group (N=522) were required (see protocol [51] for calculation). At a final sample size of 636, the study is sufficiently powered to detect differences in the outcomes between groups if they exist.

We measured women’s views on the feasibility and acceptability of e-screening with an adapted 9-item version of Renker and Tonkin’s assessment of the feasibility of computerized screening for interpersonal violence [46,47]. Of the 9 items, 2 items are related to acceptability of e-screening (I did or would like to use a tablet to answer these questions about emotional health; I would prefer answering questions about emotional health on the tablet compared to a paper questionnaire) and the remaining 7 items measure broad areas of feasibility, including disclosure (I was able to tell the truth on all the questions about emotional health; I did not like answering the questions about emotional health on the tablet or paper because it was hard to answer how I really felt), features of screening (privacy; ease of understanding questions; impersonal; time to completion), and comparison with paper-based screening. Women in the intervention group answered questions rating their experience of e-screening (eg, I liked answering questions about emotional health on a tablet because it felt private). Those in the control group answered questions to assess their views on e-screening if they had the opportunity (eg, I would like answering questions about emotional health on a tablet because it would feel private). Responses were rated on a 5-point Likert scale (1=strongly disagree to 5=strongly agree).

Intention-to-treat analysis was conducted for all research questions. We used descriptive data (frequencies and 95% CIs; means and SDs) to describe the sample. Baseline differences in groups were compared using independent t tests (means) and chi-square tests (%) to determine the extent to which randomization was successful. Statistical significance for all analyses and final models was set at P<.05. We used chi-square tests to compare proportions of women in each group responding affirmatively to questions on feasibility and acceptability.

For analyses related to factors associated with the outcomes of preference for e-screening and disclosure, we conducted bivariate analyses to identify independent factors that were significantly associated with each of the outcomes at P<.20, estimating unadjusted odds ratios and their 95% CIs. Those variables were entered in the 2 final multivariable models, where P<.05 defined factors that were significantly associated with the outcomes in the final models. All significant independent variables were entered into the multivariable models simultaneously, and each variable was controlled for by all other variables in the model. For the final models, we reported adjusted odds ratios and their 95% CIs.

Of the 675 eligible women who were approached between August 2013 and January 2015, a total of 636 (94.2%) women agreed to participate and were randomized to the intervention (n=305) or control (n=331) group. No women were deemed ineligible after the initial screening for eligibility. A total of 5 women withdrew from the study following group allocation, 3 in the intervention group and 2 in the control group. There were no statistically significant differences in relation to demographic variables between the 2 groups (Table 1).

The majority of pregnant women were 25 to 34 years of age, partnered (married, common-law, or living with a partner), were white, had incomes of Can $80,000 or more, had at least some postsecondary education, were pregnant with their first child, and were in their first trimester of pregnancy (80.3%, 494/615; see Table 1). More than a quarter of the participants had been diagnosed and treated for anxiety, depression, or another mental health concern before joining the study, and 18.0% (113/629) reported having 4 or more adverse childhood experiences. There were no statistically significant differences between the intervention and control groups in demographic characteristics, comfort with computers, tablets, or mobile phone previous mental health diagnoses or treatment, obstetric history, mean gestational age, or number of adverse childhood experiences. More than two-thirds of women (n=423) were recruited from community-based maternity clinics, 21.0% (131/624) were recruited from community-based prenatal classes, and 11.2% (70/624) were recruited from the high-risk antenatal unit. Missing data were 3.3% (615/636) or less for all variables, with the majority having less than 1.5% (10/636); thus, data imputation was not used.

The findings of this study suggest that e-screening is a feasible approach to mental health screening. Overall, women in both e-screening and paper-based screening groups indicated that they would be able to disclose their concerns about their mental health, with women in the e-screening group consistently reporting the features of e-screening as superior to paper-based screening. In the multivariable analyses, we found that women in the e-screening group were more likely to prefer e-screening compared with women in the paper-based screening group. However, none of the independent variables, including demographics, mental health history, mode of screening, or obstetric history, were significantly associated with women’s ability to be honest during screening.

In order to limit participant burden, we evaluated the e-screening and paper-based versions of the ALPHA and the EPDS. Although these 2 screening tools are widely used in perinatal clinical settings, evaluation of other tools is warranted (eg, Generalized Anxiety Disorder–7 scale, Whooley Questions, Antenatal Risk Questionnaire). Overall, women in our study tended to be well-educated, partnered, and affluent, which may limit the generalizability of the findings. However, in our multivariable analyses, these factors were not important influences in women’s perceptions of the acceptability and feasibility of mode of mental health screening. We also excluded women who did not speak English as the first step to trial e-screening. Indeed, one of the most important applications of e-screening may be the feature of having immigrant women answer questions in their preferred language and devising a computer-based algorithm to assess the scale score and interpret it automatically for the English-speaking provider. Future studies should evaluate this application of e-screening.

The final participation rate for the trial was 94.2% (636/675), suggesting that most women were eager to participate in a trial of screening for mental health concerns. This trial’s findings support the feasibility and acceptability of e-screening among pregnant women, suggesting that it is a viable service delivery option for mental health screening in busy, primary care settings. As an implementation consideration, clinic and hospital staff would require minimal training to support women in accessing the Web-based screening link on the tablet computer. Future studies should evaluate the effectiveness of e-screening on clinical outcomes, including follow-up assessment, linkage to services, and reduction of risk of mental health disorders.