Introduction This protocol describes a study testing the efficacy of interpersonal psychotherapy (IPT) for major depressive disorder following perinatal loss (early and late fetal death and early neonatal death). Perinatal loss is associated with elevated risk of major depressive disorder and post-traumatic stress disorder (PTSD). Perinatal loss conveys specific treatment needs. The trial will be the first fully powered randomised trial of treatment for any psychiatric disorder following perinatal loss.
Methods and analysis A sample of 274 women in Flint and Detroit areas in Michigan who experience a major depressive episode following a perinatal loss will be randomised to group IPT for perinatal loss or to group coping with depression. We anticipate that 50% of the sample will have co-occurring PTSD. Assessments occur at baseline, mid-treatment (8 weeks), post-treatment (16 weeks) and follow-up (28 weeks). Clinical outcomes include time to recovery from major depressive episode (primary), depressive symptoms, PTSD symptoms and time to recovery from PTSD. Additional outcomes include social support, social role functioning (including parental functioning for those with living children), well-being, grief (including complicated grief and fault beliefs) and fear of subsequent pregnancies. Social support and grief are hypothesised mediators of IPT effects on time to recovery from major depressive episode.
Ethics and dissemination The trial was approved by Michigan State University’s Biomedical Institutional Review Board. It has a data and safety monitoring board and has been submitted to the community-based organisation partners community ethics review board. Written operating procedures outline methods for protecting confidentiality, monitoring and recording adverse events, and safeguarding participants. We will share study results with research and clinical communities, community organisations through which we recruited, and will offer results to study participants. Deidentified datasets will be available through the National Institute of Mental Health Data Archive and to qualified investigators on request.
Trial registration number NCT04629599.
- fetal death
- perinatal death
- major depressive disorder
- posttraumatic stress disorder
- interpersonal psychotherapy
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- fetal death
- perinatal death
- major depressive disorder
- posttraumatic stress disorder
- interpersonal psychotherapy
Strengths and limitations of this study
This study addresses a clinical need, and will provide an evidence base for treating a n understudied population whose distress has historically been minimised.
The trial will have strong representation from disparities populations (especially African-American women and socioeconomically disadvantaged women) who experience higher rates of perinatal loss, increasing its significance.
Rigour and reproducibility are ensured by the randomised design, clear inclusion criteria, use of well-established research, recruitment and retention methods, use of reliable and valid measures, the use of raters blinded to treatment condition, and transparent power and statistical analyses.
Intervention strengths include clearly distinct treatment conditions, use of manualised treatment protocols and fidelity assessments, and team members with decades of clinical experience responding to perinatal loss.
Challenges may include recruitment during a global pandemic in communities with a higher levels of research mistrust and more mental health stigma than in the pilot trial.
About 650 000 women in the USA experience perinatal loss each year (including early and late fetal death and the death of a liveborn neonate within the first 28 days).1 2 Rates of major depressive disorder (MDD) after perinatal loss are higher than after giving birth to a living infant, and are three times the rates among matched samples of community women.3–20 Rates of post-traumatic stress disorder (PTSD) after perinatal loss are up to seven times the rates of PTSD among mothers of living infants,13 and elevated PTSD symptoms can occur for years after the loss.21 Fetal or neonatal death triples rates of suicide and of hospitalisation for suicide attempts.22 23
Not only do women with perinatal loss have higher rates of MDD and PTSD than mothers of living infants, their needs also differ from needs following many other kinds of bereavement. In perinatal loss, the fact of bereavement is compounded by the physical experience of miscarriage or of delivering a baby that has already died.24 Many women experiencing perinatal loss grieve in secret, as others may not know about the loss. Even if others do know, there are few social norms that guide how others can or should support the bereaved, making support less likely and often less helpful (eg, ‘you can just have another one’).12 25 Reasons for loss are often unclear and many women blame themselves.
Despite recognition that MDD (with or without co-occurring PTSD) following perinatal loss causes impairment and that treatment as usual is often inadequate,26–30 our previous pilot work31 created and tested the first manual for treating any psychiatric disorder after perinatal loss. The manual is structured and applies interpersonal psychotherapy (IPT) principles to MDD following perinatal loss in a group format. Previous IPT and other treatment manuals for perinatal depression32–35 focus on helping women adjust their relationships, identity, roles and routines to the demands of a new baby. This is inappropriate in the context of perinatal loss. Our manual applies IPT social support and communication principles to issues such as resolving conflicts over how to respond to the loss, grieving and requesting support in the absence of social norms about how to do so, and resolving questions of fault and role competence. It can be used by providers who do not know IPT. A randomised pilot trial of women experiencing MDD following perinatal loss established acceptability of proposed study procedures and identified high rates of co-occurring PTSD (54%) among study participants.31 Results favoured the new IPT manual for PTSD recovery, treatment satisfaction, depressive symptoms, grief and social support relative to coping with depression (CWD), a cognitive-behavioural-based intervention that does not focus on perinatal loss, interpersonal issues or social support.31
Based on those promising results, this study will be the first fully powered randomised trial of treatment for any psychiatric disorder following perinatal loss. We will compare our group IPT manual to a general group depression treatment (CWD) in a sample of 274 women experiencing MDD in the context of perinatal loss. The trial will test the hypotheses that:
IPT for perinatal loss will result in reduced time to recovery from MDD (primary), depressive symptoms and PTSD symptoms, relative to CWD. Among women meeting criteria for PTSD, IPT will result in reduced time to recovery from PTSD relative to CWD.
IPT for perinatal loss will result in increased social support, social role functioning (including parental functioning for women with living children) and well-being, and decreased grief and fear of subsequent pregnancies, relative to CWD.
Social support and grief will mediate the effects of IPT on time to MDD recovery.
This trial will provide an evidence base for treating a vulnerable and understudied population whose distress has historically been minimised. Given that poverty increases risk of perinatal loss,36–38 and doubles the risk of perinatal depression,39–46 and that rates of perinatal loss for African-American women are double those for white women,47–49 the location of the trial in Southeast Michigan, which includes Flint and Detroit (minority-majority cities with high rates of poverty), increases the trial’s significance.
Methods and analysis
Patient and public involvement
Research questions arose from a clinical need identified by provider colleagues. Patients provided feedback on treatments, measures and study procedures in the pilot trial.31 Local minority-led community-based organisations provided feedback on measures, procedures and recruitment methods appropriate for the Flint and Detroit areas. The study team is embedded in Flint and Detroit. Team members have lived experience. The trial has been submitted for voluntary review by the Flint-based community-based organisation partners community ethics review board for additional community feedback.50
Rationale for design
Given that no other treatment exists for women who experience perinatal-loss related MDD that could be used as a comparator condition, we chose to use a general depression treatment, the CWD course, as a control condition. We chose CWD because it is the group treatment with the most empirical support for treating MDD51 52 and because it is distinct from IPT. IPT addresses MDD through emotional exploration, work on relationships, communication, grief and social support. CWD addresses MDD by changing thinking and behaviour; it focuses on skills for reducing depression in general and does not have perinatal-loss specific components. Our IPT treatment differs from CWD in its focus on exploring reactions to the loss, addressing loss-related interpersonal challenges and improving loss-related social support and grief-specific coping. The trial’s secondary outcomes (social support, social functioning, grief) assess hypothesised differences between treatments. As desired, our pilot study found differences between conditions in the hypothesised mechanisms of social support and grief and in terms of in-session activities.31
Manualised treatments are attention-matched (12 groups of 90 min each, 1 individual pregroup session and 1 booster session). Every 4 weeks, both treatments allow new women to enter the group and women completing 12 weeks to leave the group.
Participants in the IPT condition receive 12 group sessions and 2 individual (pregroup and 1-month booster) sessions as outlined in the structured manual (see table 1 and pilot trial).31 The individual sessions prepare patients to use the group effectively, to keep group members focused on their treatment goals and to maintain treatment gains. In addition, 3 of the 12 group sessions invite women to include their partners or other support people to bolster the woman’s social support system and to reduce conflicts over how to react to the loss. Relationship distress is common following perinatal loss.12 53
Group sessions are semistructured, and each woman covers the four group topics listed in table 1 three times over her 12 group sessions, approaching each topic from a different stage in the mourning process. New women are allowed to enter group every four sessions. This allows remaining women to see their own progress and encourages new women through example and peer counselling.
Coping with depression
CWD is a structured, manualised54 psychoeducational group treatment for MDD. The CWD course is cognitive behavioural. The problems shown by depressed individuals are viewed as behavioural, with cognitive patterns that can be unlearned or relearned. Its effectiveness is comparable to other forms of psychotherapy in depression.52 The course content teaches skills including relaxation, cognitive skills and behavioural activation. The CWD pregroup and booster sessions are the pregroup and booster sessions from the published CWD manual.54 To ensure that the CWD intervention was distinct from IPT, we excluded the two sessions on social skills and emphasised pleasant activities that were individual rather than social. Consistent with standard CWD, we focused on addressing depression rather than discussing grief or perinatal loss. We expanded other CWD material (eg, relaxation practice) to replace sessions on social skills. In our studies, the 12 CWD sessions covered: an introduction to social learning rationale of depression; learning to relax; relaxation in everyday situations; pleasant activities and depression; formulating a pleasant activities plan; constructive thinking; planning for constructive thinking; and maintaining gains (see published manual54 and pilot trial31 for additional details).
Participants will be 274 women who are experiencing MDD in the context of perinatal loss who (1) meet Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-555) criteria for MDD; (2) have experienced a perinatal loss (including early and late fetal death, death of a liveborn neonate within the first 28 days and medically recommended termination) within the last 1–12 months; (3) are 18–50 years old; (4) speak and understand English well enough to understand questionnaires when they are read aloud; (5) can provide the name and contact information of at least two locator persons and (6) have access to a telephone. Exclusion criteria are: (1) onset of current major depressive episode prior to news of difficulties with the pregnancy or health risk to the infant (women with prior episodes are included); (2) current or past diagnosis of bipolar disorder, schizophrenia or other psychotic disorder; (3) primary diagnosis of current substance use disorder; (4) acute suicidal or homicidal risk; (5) beginning or changing dose of antidepressant medication or psychotherapy in the previous 12 weeks) and (6) any IPT or cognitive-behavioural treatment in the previous 12 weeks. Women in stable concurrent psychotherapy who are included are asked to suspend this treatment during the active study treatment phase. PTSD is not an inclusion criterion. However, based on our pilot,31 we anticipate that slightly more than half the sample will meet criteria for PTSD at study enrolment.
Therapist training and supervision
We trained eight study therapists (four in IPT and four in CWD) who are MSWs or clinical or counselling psychologists. Therapists are recruited from their respective communities to moonlight as clinicians in this proposed study. Therapists are provided with the detailed treatment manuals. Training for both conditions includes education and role plays. Therapists in both conditions will be monitored for adherence/competence throughout the study and retrained as needed.
Supervision involves review of therapists’ audiotaped sessions and a weekly 1-hour small-group telephone meeting for feedback and case discussion. Treatment sessions are audiorecorded using digital audio recorders or a Health Insurance Portability and Accountability Act (HIPAA)-compliant version of Zoom. Study therapists remotely upload the recordings to the study’s secure research server, where the supervisors can remotely access them.
Women are randomly assigned to IPT or CWD in a 1:1 ratio. We stratify randomisation on (1) whether women have been taking antidepressant medications or attending other psychotherapy (stability of dose is an inclusion criterion) and (2) type of perinatal loss (miscarriage, stillbirth, neonatal death). Randomisation sequences were created by the study statistician. Assignment is concealed in an envelope that research assistants (RAs) open at randomisation.
Participants are recruited from counties in Southeast Michigan using a broad outreach strategy. We partner with regional health systems, community-based organisations in Flint and Detroit, and a regional Medicaid system in study recruitment. Recruitment also includes flyers and referrals from: (1) local birthing centres, emergency departments, OBGYN offices and federally qualified health centres; (2) hotlines, support groups, family nurse partnerships; (3) funeral homes; (4) churches, daycare centres, other places where women and mothers congregate (WIC offices, Medicaid offices, etc); (5) bus ads and (6) online venues. We began recruitment on 1 September 2021 and plan to end on 1 March 2025.
We had planned to offer baseline assessments at women’s homes or our offices, group sessions in community locations convenient for participants (as we did in the pilot trial),31 and to conduct follow-up assessments by telephone. However, due to the COVID-19 pandemic, we are currently conducting all assessments and treatment sessions by Zoom, with an option to go back to holding groups in person in the future.
We employ techniques we have found helpful in achieving low attrition rates in previous studies.31 56–58 These include study staff’s strong relationships with participants, efforts to value and appreciate the women’s participation in the study, and frequent personal contact. We are flexible about follow-up appointment scheduling and train research staff to be culturally sensitive. Follow-up assessments take place by Zoom or phone, with well-established safety procedures for emergencies.56 We maintain a list of two other people who know where the participant resides. We conduct treatment groups at different times and locations to make attendance easier. If a woman misses a treatment appointment, the therapist calls her to check in and problem-solve barriers to attendance. Finally, participant fees for follow-up assessments help facilitate retention.
Assessments take place at baseline, mid-treatment (8 weeks), post-treatment (16 weeks) and follow-up (28 weeks; see table 2). Assessments are conducted by RAs trained and certified in interviewer administered instruments and blind to treatment assignment. Interviewers and senior staff meet regularly to review assessment tapes, address questions and monitor inter-rater reliability. Data quality is maintained through clerical and clinical checks after data are entered and through regular examination of distributions, missing data and outliers.
The Structured Clinical Interview for DSM-559 is used to establish study eligibility. During the follow-up, the Longitudinal Interval Follow-up Examination (LIFE),60 61 a standardised retrospective calendar-based interview, is used to assess MDD and PTSD recovery. The LIFE uses Psychiatric Status Ratings to categorise DSM-5 symptoms on a scale of 1 (asymptomatic) to 6 (incapacitated) for each week. A PSR of 5 or 6 indicates the participant meets full diagnostic criteria, 3 or 4 indicates subthreshold disorder, and 1 or 2 indicates the participant is not in episode. For survival analyses, recovery is defined as eight consecutive weeks of a PSR of 1–262 at any time between baseline and the 28-week follow-up. Women who do not have at least eight consecutive weeks of PSR of 1–2 during this time are considered ‘not recovered’. The LIFE is the gold-standard way of determining onset and offset of psychiatric disorder.62 We also use the LIFE to track participation in psychotherapeutic and psychopharmacologic treatment at baseline and follow-up. We assess partner violence using the Women’s Experience with Battering screen.63 Battered women (scores of 20+) are included in the study and provided with partner violence resources.
Depressive symptoms are assessed using the Quick Inventory of Depressive Symptoms (QIDS), Self-Report version.64
PTSD symptoms are assessed using the Life Events Checklist and PTSD Checklist for DSM-5.65–68 We also assess whether PTSD symptoms are related to the perinatal loss.
Social support and social functioning
We use the 12-item Multidimensional Scale for Perceived Social Support (MSPSS)69 to assess overall social support. We use a validated adaptation of the Relationship Assessment Scale70 to assess satisfaction with an important significant other (partner or other support person of the woman’s choosing) relationship. We assess social functioning using the Short version of the Social Adjustment Scale-Self-Report (SAS-SR).71 Because depression can affect parenting, we will analyse the SAS-SR total score as well as its parental functioning subscale.
Well-being (including life satisfaction, purpose and meaning) is measured by using the 23-item National Institutes of Health (NIH) Neuro-Quality of Life (Neuro-QoL) scale for positive affect and well-being.72
Grief symptoms are measured using the Perinatal Bereavement Grief Scale (PBGS).4 73 Complicated grief is measured using the Inventory of Complicated Grief (ICG).74 A few items on the ICG were reworded to refer to perinatal loss. Given that many women present with unwarranted beliefs about what caused the loss, we assess deservingness and guilt as grief outcomes using a 7-item scale about loss beliefs (the Loss Beliefs Scale). This scale includes items such as ‘I think what happened was my fault’ and ‘The miscarriage, stillbirth, or baby’s death was caused by something about me’. This perinatal loss specific scale was created after reviewing the literature on beliefs about deserved bad outcomes.75–77
Fear of subsequent pregnancies is assessed by 7-point Likert items (from 1=‘strongly disagree’ to 7=‘strongly agree’): (1) ‘I am afraid to become pregnant again’; (2) ‘I look forward to becoming pregnant again’; (3) ‘I plan to become pregnant again’; (4) ‘I worry about what might happen if I get pregnant again,’ (5) ‘I do not want to be pregnant again.’
Treatment acceptability is measured using the Client Satisfaction Questionnaire-Revised (CSQ-8-R).78
We will use the IPT and CWD adherence and competence scales developed in the pilot trial31 to rate fidelity using audio recordings. As in the pilot trial, raters will also assess the percent of time in each group session spent discussing the perinatal loss and discussing loss-related communication strategies.31
Primary analyses will be intenon to treat. We will examine dose-response effects in secondary analyses. Primary tests will be two sided with p=0.05. Descriptive statistics will include effect sizes and measures of clinical significance.79 Primary and secondary outcomes and all hypotheses are stated a priori, therefore, 0.05 level of significance will be used. Per Kraemer,80 we will not test for differences between conditions due to randomisation as those differences are due to chance alone, rendering p values meaningless. Covariates for analyses are specified a priori based on subject matter expertise. No interim analyses are planned.
Attrition analysis and missing data
We will compare characteristics of those who drop out by trial arm and compare those who complete the study with those who do not to assess generalisability of findings. For the primary outcome of time to MDD recovery, unobserved time to MDD resolution for the drop-outs will be treated as censored. For secondary outcomes, regression techniques below allow for missing at random (MAR) mechanism.81 If patterns of missing data indicate potential not MAR mechanisms, then models describing missing mechanisms will be considered (eg, pattern-mixture models),82 83 and sensitivity analyses will be employed.
For survival analyses, the proportional hazard assumption will be evaluated. If it holds, then survival analyses will use Cox regression. If not, time-varying effects will be investigated, and the model will be modified to include an interaction of relevant covariates with a function of time variable. Stratified models will also be considered.
Linear mixed effects (LME) models will be used to test differences between trial arms for continuous outcomes. All participants with at least one completed postbaseline assessment will be included. We expect these scores to follow normal distributions. However, generalised LME (GLME) modelling will be used if outcome is not normally distributed and cannot be normalised using transformations.
(1) Using survival analysis, with initial QIDS score as a covariate, we will test the hypothesis that IPT, relative to CWD, will result in reduced time to recovery from the major depressive episode (primary). (2) Using LME or GLME with baseline QIDS score as a covariate, we will test the hypothesis that IPT, relative to CWD, will result in reduced depressive symptoms (QIDS scores) across post-baseline assessments. (3) Using LME or GLME with baseline PCL score as a covariate, we will test the hypothesis that IPT, relative to CWD, will result in reduced PTSD symptoms (PCL score) across postbaseline assessments. (4) Using survival analysis, with initial PCL score as a covariate, we will test the hypothesis that IPT, relative to CWD, will result in reduced time to recovery from PTSD.
Using baseline scores as covariates, we will separately test the hypotheses that IPT for perinatal loss will result in increased social support, social role functioning and well-being, and decreased grief and fear of subsequent pregnancies, relative to CWD, using the LME or GLME modelling described above. Specifically, controlling for baseline values, we will use separate analyses to test the effects of IPT vs CWD on MSPSS, Relationship Assessment Scale, SAS-SR total, SAS-SR parenting subscale, Neuro-QoL, PBGS, ICG, Loss Belief Scale and Fear of Subsequent Pregnancy scores. We will also compare conditions on CSQ-8-R Treatment Acceptability.
We will test the hypotheses that social support (MSPSS scores) and grief (PBGS scores) will mediate the effects of IPT on time to MDD recovery. To test for mediation, trial arm will be treated as the independent variable and each potential mediator (one at a time) will be tested for their effect on the outcome variable at weeks 8, 16 and 28, with the baseline value of that outcome treated as a covariate. We will use a bias corrected bootstrapping analytic strategy84 85 based on 5000 bootstrap samples to estimate CIs around the indirect effect of study group on the outcome variable, through the mediator.
We will calculate differential effect estimates and test the interactions between several participant characteristics and the intervention effect in predicting time to recovery from MDD. Characteristics to be examined include race, ethnicity, having living children, type of perinatal loss, stable use of other antidepressant or psychosocial treatment at baseline, having PTSD at baseline, having a partner, number of past depressive episodes and time since loss. Given our emphasis on meeting the needs of minority women, we will also evaluate whether mediation relationships or intervention effects on secondary outcomes differ by race.
For the primary outcome of time to MDD recovery (censored at 28 weeks), assuming 1:1 allocation, power 0.80, two-sided tests at p=0.05, and observed estimated HR=1.79, the required number of events is 94 (number of MDEs resolved). By week 28, there were 17 events in the preliminary data (17 women had MDD resolved, 11 in the IPT condition and 6 in the CWD condition) out of 45 participants, so the rate of events was 0.38. Thus to have 94 events, total N=246 is required. Given that 90% of pilot trial participants completed at least one follow-up assessment,31 we increased the sample size to 274. Given the observed HR of 5.85 for PTSD in the pilot trial, power for PTSD recovery should be greater than 80%. Tests of mediation in aim 3 will have greater power than the primary outcome because of reduction in error variance when controlling for the mediator.
Continuous (secondary) outcomes
Assuming an unadjusted d=0.32 and a correlation of 0.6 between follow-up measures and 0.3 with baseline (as observed in the preliminary data), the adjusted effect size would be 0.40, and only n=200 participants would be needed before attrition. If study attrition is higher, the study still has >80% power for secondary outcomes (table 3).
Ethics and dissemination
The trial was approved by Michigan State University’s Biomedical Institutional Review Board (FWA 00004556). A three-member external data and safety monitoring board reviews data and safety of study participants. Trial safety procedures are codified through checklists and a written manual of operating procedures.
Informed consent and confidentiality
When potential participants contact the study, study RAs meet with them privately (electronically or in person). RAs explain risks, benefits and the voluntary nature of the study and obtain participants’ signed informed consent.
Confidentiality is protected by research staff trained to manage sensitive clinical issues. Participants are informed about the limits of confidentiality concerning suicidal intent, homicidal intent and suspected child or elder abuse. Computer files are available only to authorised personnel, with no names or obvious identifying information stored in data files. Confidentiality of recordings of study assessments and treatment sessions is protected through: (1) use of encrypted audiorecorders and HIPAA-compliant videoconferencing; (2) labelling recordings with study IDs rather than names; (3) storing recordings on a secured computer server designed to hold and protect research data and (4) limiting access to recordings.
Participant safety is monitored during study assessments and during study treatment sessions. Prespecified adverse events are recorded and monitored using a structured system created within Research Electronic Data Capture (REDCap) with alerts for follow-up actions.56
During assessments, participants who score 2 or above (any active suicide ideation) on the QIDS suicide item are transferred via warm handoff to a national suicide hotline contracted for this trial. The suicide hotline assesses suicidality and emergent treatment needs, provides follow-up and securely transmits a written disposition. This procedure has worked well in previous trials.56
If a participant develops manic or psychotic symptoms, or if her QIDS score increases by 5+ points from baseline, she is evaluated by an independent clinician to see if she needs to be referred for other treatment and/or removed from the trial.
Women in either condition whose MDD has not remitted by the end of the study (week 28) are referred for other treatment.
Dissemination policy and access to data
Dissemination activities will include academic papers, presentations to clinical communities, community reports and talks with organisations through which we recruited, and offering to share final study results with study participants. De-identified datasets will be available to qualified investigators on request.
Patient consent for publication
Contributors JEJ: conceptualisation (lead), funding acquisition (lead), investigation (lead), intervention development (equal), safety protocols (lead), writing—original draft preparation (lead), writing—review and editing (lead). CZ: conceptualisation (supporting), funding acquisition (supporting), clinical supervision (equal), safety protocols (supporting), writing—review and editing (supporting). ABP: funding acquisition (supporting), intervention development (equal), clinical supervision (equal). AS: conceptualisation (supporting), funding acquisition (supporting), data curation, formal analysis. KDK: funding acquisition (supporting), recruitment strategy (supporting). SL, BT, CH and MC: recruitment strategies (lead), development of operating procedures (supporting), choice of measures (supporting).
Funding This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) grant number (R01 HD100471). NICHD had no role in the design of the study and will have no role in the implementation of the study, analysis of the data, or writing of the manuscript.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.