For most mothers, there is no time of greater risk to their physical and mental health than during pregnancy, childbirth and the first few months of their child’s life – often referred to as the perinatal period (Oates Reference Oates2003; Ronsmans & Graham Reference Ronsmans and Graham2006; Souza et al. Reference Souza, Tunçalp and Vogel2014). To support these women, specialist perinatal mental health services have emerged in recent decades (Galbally et al. Reference Galbally, Blankley and Power2013), and these continue to grow and develop (Howard & Khalifeh Reference Howard and Khalifeh2020). However, owing to (a) the relative infancy of this field of mental health research and (b) the complexities associated with carrying out medical research in pregnant women, evidence on effective perinatal interventions remains relatively limited.
Complex post-traumatic stress disorder, or cPTSD, was introduced into ICD-11 in 2018 (Box 1). Although study of this condition is still emerging, it is considered likely that people with a history of cPTSD are more likely to develop or have exacerbations of mental illness in the perinatal period (Madora & Vaughn Reference Madora and Vaughn2022). It is also considered likely that pregnancy, childbirth and caring for an infant might serve as triggers or reminders of past traumas and ‘reactivate’ or worsen cPTSD symptoms (Seng et al. Reference Seng, Rauch and Resnick2010; Muzik et al. Reference Muzik, McGinnis and Bocknek2016).
ICD-11 classification of complex post-traumatic stress disorder
Complex post-traumatic stress disorder (cPTSD) is classified in ICD-11 as a ‘disorder specifically associated with stress’. Individuals with this condition meet the diagnostic criteria for PTSD, but experience additional disturbances in self-organisation. Essential features therefore include:
exposure to an ‘extremely threatening or horrific’ event or series of events;
the presence of all of the core features of PTSD lasting for at least several weeks, namely: re-experiencing (commonly referred to as ‘flashbacks’), avoidance and hypervigilance;
difficulties with affect regulation – such as dissociating in stressful circumstances, reckless or self-destructive behaviour;
persistent negative beliefs about themselves, often related to their traumatic experience;
relationship difficulties – such as difficulty maintaining relationships.
These symptoms must cause significant impairment to the person’s ability to function, or effective functioning must require ‘significant additional effort’.
Unlike in PTSD, where the condition is thought to develop as the consequence of a single ‘traumatic event’, a person with cPTSD has usually experienced a prolonged series of traumatic events – therefore, the most common antecedent trauma in someone with this diagnosis is childhood maltreatment.
It is specified that symptoms of cPTSD are generally more severe and persistent in comparison with those of PTSD.
(Adapted from World Health Organization 2025: 6B40 Post traumatic stress disorder)
It has also been reported that trauma experienced by a parent (mother or father) can indirectly affect subsequent generations, via a range of means (Yehuda et al. Reference Yehuda, Halligan and Grossman2001): this is often referred to as intergenerational or transgenerational trauma. It is possible that interventions for parents with symptoms of cPTSD may therefore benefit the infant too.
Given the specific difficulties that potentially face parents with cPTSD, as well as their children, the review in this month’s Cochrane Corner used an ambitious design to evaluate what effective interventions are available to support these families in the perinatal period (Jones et al. Reference Jones, Freijah and Brennan2023).
The Cochrane Review
Summary
Jones and colleagues (Reference Jones, Freijah and Brennan2003) conducted meta-analyses of 15 randomised controlled trials (RCTs), considering the effectiveness of interventions during the perinatal period designed to support parents who were experiencing symptoms of cPTSD and/or had a history of maltreatment in childhood. They evaluated the effect of 17 interventions involving a total of 1925 participants.
The authors posed a meaningful clinical research question according to the Population, Intervention(s), Comparison and Outcome(s) (PICO) model. The literature search, data analysis and synthesis appear to have been conducted according to Cochrane recommendations (Higgins et al. Reference Higgins, Thomas and Chandler2024). In this Round the Corner, we therefore focus only on some key aspects of the methodology and results, as outlined below.
Method
The study population
Jones et al note that cPTSD is a relatively new diagnosis and therefore to include only evidence relating to participants with a formal diagnosis of this condition would make the review unsustainable. They also therefore included interventions targeted at participants with a history of childhood maltreatment, on the basis that this is such a common antecedent of cPTSD, as well as participants with a history of removal from their own parents, on the basis that this is likely to have followed child abuse.
The authors chose to include a wide range of parents – biological, step, surrogate, kinship carers and parents without children currently in their care – and they included fathers as well as mothers.
The interventions and comparison
The review included interventions taking place in the perinatal period (defined here as from pregnancy up until 2 years after childbirth) designed to support parents experiencing symptoms of cPTSD or who had experienced childhood maltreatment (or both). The review authors recognised the likelihood that the search would identify a wide range of interventions, across a range of settings and delivered by a range of people. They therefore developed a logic model (Box 2) to identify how interventions could be categorised for analysis. Intervention categories included psychological therapies, parenting or parent–infant or relationship-focused interventions, service system approaches and more. Comparison was provided in a range of forms across the studies, including usual care, waiting list and no care.
Logic models
A logic model is a diagram, which usually looks like a flowchart or a map. It is produced with the intention of hypothesising how a complex intervention works to produce its outcomes. Logic modelling has been used for many decades (Kaplan & Garrett Reference Kaplan and Garrett2005) and is often used to design, plan or understand complex health and social care pathways and systems (Anderson et al. Reference Anderson, Petticrew and Rehfuess2011). More recently logic models have been used in systematic reviews for a range of purposes, including assessing and classifying healthcare interventions and their effects (Baxter et al. Reference Baxter, Blank and Buckley Woods2014).
The outcomes
The review authors chose two main categories for primary outcome measures: parental psychological or socio-emotional well-being (to include trauma-related symptoms; substance use; self-harm; and other measures); and parenting capacity (to include parent–child relationships; and parenting skills). There were no primary outcome measures relating primarily to the child or to the wider family. A range of secondary outcome measures were also used, including (but not limited to): child developmental outcomes; child emotional and behavioural outcomes; and acceptability of the intervention to the parents.
Results
The literature search resulted in the inclusion of 15 studies – 14 individual RCTs and 1 cluster RCT – involving a total of 1925 randomised parents. In 12 of the studies the participants had reported experiencing childhood maltreatment, in 1 they were experiencing PTSD symptoms and in 2 they reported both. Most participants were biological mothers (only 3 studies included fathers, representing a total of 29 participants).
All 15 studies took place in North America. Studies recruited parents from a wide range of locations (obstetric clinics; government databases; community parenting programmes) and using a variety of methods.
Seven studies assessed ‘parenting interventions’ and eight studies assessed psychological interventions. Several categories of intervention which had been identified in the logic model did not feature at all (including pharmacological and biomedical therapies). The nature, duration and location of interventions varied widely, as did the person/people delivering them.
Outcome measurement also varied, and the measurements used for analysis were taken from the first recorded point of follow-up assessment. The most common outcome was psychological well-being (10 studies), followed by parenting skills (5 studies) and trauma-related symptoms (5 studies). Some of the outcomes included in the study design (e.g. parental self-harm) were not reported in any of the studies.
The review authors used the GRADE framework (Box 3) and deemed that the evidence presented by all 15 studies was of low or very low quality. A key concern was risk of bias, especially relating to the randomisation process and to missing data resulting from high drop-out rates.
GRADE
GRADE stands for Grading of Recommendations Assessment, Development and Evaluations. It is a tool used by Cochrane reviewers, as well as others, to rate the quality of evidence. There are four levels of evidence within the GRADE framework – very low, low, moderate and high. A ‘high’ rating would mean that the person applying the GRADE tool had explored all of the potential problems with the quality of the evidence and decided that they can have confidence that the effect estimate is close to the ‘true effect’. The person evaluating the evidence uses five ‘domains’ (risk of bias; imprecision; inconsistency; indirectness; and publication bias) to determine the final rating. Because this process involves a person making judgements, the rating given will be subjective. In Cochrane Reviews, GRADE ratings can usually be found in a table called ‘summary of findings’.
The results could be summarised as showing no to possible small differences in the primary outcomes in intervention groups relative to comparison or control groups, and as being based on evidence of low or very low certainty in every case. One intervention (a financial empowerment education programme for low-income parents) increased depression slightly.
Discussion
Study design
The clinical research question
Jones et al’s meta-analysis is an ambitious project attempting to answer a clinical question relating to a diagnosis that does not appear to be attributed to any of the participants it includes. It seems that in 12 of the 15 included studies, participants were enrolled based on their history of childhood maltreatment rather than complex trauma and none of the studies appear to reference cPTSD in their design – this may be a function of the time at which the literature search was conducted (October 2021) relative to the introduction of cPTSD into the ICD (in a preliminary release of ICD-11 in June 2018). However, the significant resulting focus on childhood maltreatment as the inclusion criterion raises the question of whether the overlap between childhood maltreatment and cPTSD is as significant as the paper has assumed – childhood maltreatment is very common (the authors reference an estimated prevalence of 50%) whereas they refer to an estimated 3.8% of adults meeting the criteria for cPTSD. It stands to reason that there may be many participants in the studies who experienced childhood maltreatment but did not have symptoms of PTSD, in which case this meta-analysis may be answering a different clinical question to the one that is stated.
Population
The design of the study deliberately included both mothers and fathers, as well as a wide range of types of parent beyond simply biological parents, although most participants were biological mothers. The fact that 20% of the included studies featured fathers as participants but this only represented 29 of the 1925 participants included in the review (1.5%) indicates that this gender imbalance is mirrored in the studies themselves. This finding perhaps reflects the infrastructure available to fathers within health and social care systems in the countries where these studies took place (Singley & Edwards Reference Singley and Edwards2015).
Interventions
There are several further factors that potentially limit the external validity (Box 4) of the review’s findings. The interventions that the study included are wide-ranging and yet several categories of intervention (and outcome) anticipated in the review’s methodology are not represented at all. This highlights the fact that the available evidence is not adequate in volume or nature to answer the clinical question, in accordance with the study design. As well as indicating that more research might be required, this raises the question as to whether the hypothesised interventions and outcomes in the study are design priorities for decision-makers in health and social care systems, and where such interventions would sit within the system. Thinking about the UK as an example, the specific population described in the study (in particular, those included on the basis of past trauma rather than current symptoms) would be unlikely to meet criteria for secondary perinatal mental healthcare services, and, without clear evidence of benefit, the likelihood of investment in this area, particularly in publicly funded health and social care systems with already stretched resources, seems low.
External validity
External validity considers whether a study’s results might also hold true for a wider population, beyond those who were included in the study. In healthcare research, it can be useful for a clinician to know whether a study is considered to have good external validity, because it might help them to understand whether the results of the study might have relevance for their own patients.
Primary outcomes
Another potential problem with the design of this study is whether it is appropriate to answer this clinical question without including any primary outcomes that relate to the wider family – particularly to the child. It is possible, for instance, that these interventions, which seem to have limited impact on the parent’s own well-being in the perinatal period, may hold benefit for the child in the longer term.
Conclusions
There is no clear evidence arising from this Cochrane Review to support any specific intervention for parents with symptoms of cPTSD or a history of maltreatment in childhood. As Jones et al state, the results ought to be interpreted with extreme caution owing to the lack of methodological rigour in the included studies and the low certainty observed across the findings. It could also be argued that the scope of the review is so wide and heterogeneous – including a range of types of parent, a small but highly diverse range of interventions and people with a history of childhood maltreatment and/or symptoms of cPTSD – that it has been designed in such a way as to undermine its clinical utility and applicability.
Further research of good quality is required to establish whether there are any perinatal interventions that might be effective in improving perinatal outcomes for parents with cPTSD. Given the body of evidence indicating that parental history of trauma can have an impact on the infant, researchers ought to consider including outcomes relating to the children and/or wider family in any future study.
Data availability
Data availability is not applicable to this article as no new data were created or analysed in this study.
Acknowledgement
I thank Dr Lucy Caswell for her support and advice.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None.
eLetters
No eLetters have been published for this article.