Supporting parenting to address social inequalities in health: a synthesis of systematic reviews

Background In 2009, the World Health Organization’s Commission on Social Determinants of Health set out its recommendations for action, which included establishing equity from early childhood onwards by enabling all children and their mothers to benefit from a comprehensive package of quality programmes. In order to address social inequalities in health, it is recommended that action be taken from early childhood, and actions providing support for parenting are an effective lever in this respect. The aim of this review of systematic reviews is to analyse, on the one hand, the components and characteristics of effective interventions in parenting support and, on the other, the extent to which the reviews took into account social inequalities in health. Methods A total of 796 reviews were selected from peer-reviewed journals published between 2009 and 2016 in French or English. Of these, 21 reviews responding to the AMSTAR and selected ROBIS criteria were retained. These were analysed in relation to the consideration they gave to social inequalities in health according to PRISMA-equity. Results The reviews confirmed that parenting support programmes improved infants’ sleep, increased mothers’ self-esteem and reduced mothers’ anger, anxiety and stress levels. The mainly authors noted that the contexts in which the interventions had taken place were described either scantly or not at all, making it difficult to evaluate them. Only half of the reviews had addressed the question of social inequalities in health. In particular, there had been little research conducted on the relational aspect and the social link. Conclusion In terms of addressing social inequalities in perinatal health, the approach remains both modest and reductive. Understanding how, for whom and in what conditions interventions operate is one way of optimising their results. Further research is needed to study the interactions between the interventions and their contexts.


Background
The social determinants of health are one of the principal causes of health inequalities, that is the unjust and sizeable discrepancies recorded between social or geographical groups [1][2][3][4][5][6][7]. In 2009, the World Health Organization's (WHO) Commission on Social Determinants of Health set out its recommendations for action, which included establishing equity from early childhood onwards by enabling all children and their mothers to benefit from a comprehensive package of quality programmes [8].
Early childhood is a key period in the genesis and reproduction of Social Inequalities in Health (SIH). Epigenetic studies and life course epidemiology confirm the link between life circumstances in early childhood and health in adulthood [9][10][11][12][13]. Social inequalities have an effect on health, most notably by creating biological modifications throughout the life course. In particular, development is negatively influenced by antenatal and neonatal stress [14][15][16][17][18][19][20].
Clearly, the perinatal period is particularly sensitive. Promoting the health of pregnant women and new mothers is therefore essential, and parenting support is one of the principal strategies that can really help [13,[21][22][23].
Many publications show that parenting support is an effective lever in promoting health in mothers and their newborns, and they are the subject of many reviews. However, have these studies taken into account SIH? The aim of this review of systematic reviews is to analyse, on the one hand, the components and characteristics of effective interventions in parenting support and, on the other, the extent to which the reviews have taken into account SIH.

Study procedure
The screening, eligibility and inclusion stages were conducted and presented according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses model (PRISMA) [24,25].
The study population was pregnant women, parents of newborns and newborns (from birth to 3 years old, preschool). The interventions involved promoting perinatal health, and more specifically, programmes and schemes offering parenting supportfrom the standpoint of obstetrics, paediatrics, psychology, sociology, education and public health, in the broad sense of the term.
The comparators were not relevant to this study. All the results (whether favourable or not) were collated and evaluated according to their effects on SIH, their psychosocial effects and their effects on perceived health. Intervention durations were taken into consideration.
A systematic search was conducted using Cochrane, PubMed and PsycINFO, which are the principal scientific databases in the field of health. The keywords referenced in the MeSH (Medical Subject Heading) were based on perinatal health promotion, SIH, the social determinants of health, parenting support and health programmes relating to parenting support (Table 1). Keywords were selected in collaboration with Céline Aubert, documentalist at the Faculté de Médecine in Nancy.
Each of the standardised search formulas was reproduced for all the databases, based on titles, abstracts and keywords, for journals published in French or English between January 2009 and September 2016.
The research protocol was written but not published. It is available on request from the authors.

Selection of articles
The documents retained were all narrative and systematic reviews published in French or English in peer-reviewed journals between 2009 and 2016. The year 2009 was chosen as the start date because it was the year the WHO published its report from the Commission on Social Determinants of Health [8].
Interventions relating to prematurity, a specific pathology or to those concerning school-age children were excluded, as were reviews of reviews. Duplicate articles were also removed. The reason for each individual rejection was given in a note. A pilot study was carried out in May 2016 using a sample of the articles in order to validate the protocol (relevance of the search formula, selection criteria, evaluative consistency of the different team members).
The full texts of the articles selected were subject to a methodical, standardised interpretation responding to each of the AMSTAR (Assessing the Methodological Quality of Systematic Reviews) items [26]. The reviews that did not meet these criteria of methodological quality were excluded.
The selected reviews were all included whether or not their results were favourable so as to exclude the possibility of any bias resulting from selective publication of results. The grey literature and the recommendations published by France's Haute Autorité de Santé and the UK's National Institute for Health and Clinical Excellence were subject to an exploratory study but were not retained for this review since they did not correspond to the study aims. Table 2 shows the PICOTS criteria for the methodological search.

Literature analysis
The whole process was carried out by one person under the supervision of three public health and social sciences experts, who all agreed on the interpretation. Differences of opinion were discussed until a consensus was reached, and these were all documented.
The systematic reviews included in this review were analysed in two stages based on the full texts alone, in other words without referring to the original articles collated by the systematic reviews. Firstly, the data describing the characteristics of the interventions and programmes were compiled in a summary table. Secondly, the reviews were analysed for the consideration they gave to SIH using the PRISMA-equity tool [27].
This tool is an extension of the PRISMA guidelines and was developed by the Campbell and Cochrane Equity Methods Group. Its aim is to take into account the notion of equity in reviews. Each of the elements of the systematic reviews (i.e. the title, abstract, introduction, method, results and discussion) was broken down and analysed to determine how the authors addressed equity.
A total of 796 publications were obtained from the literature search. Once the articles that were duplicated across the databases had been taken out and those about prematurity, maternal and infantile pathologies and children aged over three had been excluded, only 32 articles remained. The next stage allowed us to exclude articles that did not meet the AMSTAR criteria (for   (Table 3). Finally, following a detailed analysis, 21 articles were included in the corpus (Fig. 1). All the documents constituting the corpus were read twice. In cases of uncertainty, a second expert was consulted. In cases of disagreement between the first two experts, a third was consulted.
AP, FA, LFH carried out the documentary research protocol for the selection. For the analysis, AP read all the manuscripts twice, FA validated and, in cases of uncertainty, LFH had the final say. AL and LFH reviewed the manuscript.

Summary of evidence-based knowledge
The data produced by the review were recorded in a summary table showing the type of intervention studied in each systematic review, the review's aims, and the levels of evidence such as they were analysed and described by the authors and the principal characteristics of the programmes [37][38][39][40][41] (Table 4).
Characteristics of the reviews Disciplines of the journals publishing the reviews The articles were drawn from journals in a variety of disciplines. A third (n = 7) of the publications came from public health journals. Cochrane published six reviews on the topic. The remaining publications came from journals in the field of child psychology (n = 3), paediatrics (n = 2), midwifery (n = 1), social work (n = 1) and nursing care research (n = 1).

Authors' countries of affiliation in which the studies were conducted
The reviews were mainly carried out by American and British authors and generally concerned interventions in the United States and the United Kingdom. There was only brief reference made to other European countries in eight of the reviews. These comprised a Finnish review, which listed only two European programmes in the 98 studies they examined [28], a Dutch review, which compared the Netherlands to the United States and Canada [29], two Cochrane reviews mentioning some studies conducted in Germany and the Netherlands [21,30], two reviews that referenced Ireland [31,32], one review mentioning interventions conducted in Switzerland [31] and another that compared Sweden with the United Kingdom [32].

Aim of the reviews
All but three systematic reviews aimed to assess the effectiveness of the programmes by evaluating their impact over a relatively short period on the mothers' and infants' wellbeing, the mothers' mental health and the mothers' adjustments to the behaviour of their babies and very young children. In three of the systematic reviews, the aim was to analyse the methods and concepts used in research relating to parenting support. One of these listed the main research areas, the methods used and the groups targeted [28]. The second studied the socioeconomic disadvantage characteristics used in research studies [29], and the third analysed the concept of preparing for parenthood [33].

Principal programmes studied
Generally, the reviews focused on the practices of the professionals; and the programmes they studied mainly involved improving parents' knowledge and skills. Three reviews looked at parental needs. One of these was a meta-analysis focusing specifically on preparation for parenthood [31]. The second examined the needs of fathers [34], and the third looked at the needs of mothers feeding their children with substitute milk [35].

Principal characteristics of the interventions Recipient populations
The reviews focused on different periods and populations: a specific stage in the parents' history: pregnancy (n = 3), the first week after giving birth (n = 1) a broad timeframe from birth to three years (from birth to two months (n = 1), up to six months (n = 2), up to one year (n = 1), up to three years (n = 5) socially disadvantaged families (n = 2) teenage parents (n = 1) fathers only [34] (all the others concerned mothers, parents or the couple as an entity) the general population through an analysis of prevention campaigns [36].

Main results and conclusions
All the reviews provided either convincing or promising data about the positive impact on mothers' and children's wellbeing of the support programmes and actions at and following birth. The programmes were found to increase the mothers' self-esteem, reduce their anger, anxiety and stress levels, improve the infants' sleep and promote the infant's language development.

Three distinct levels of evidence
In accordance with the authors' evaluations, as described and synthesised in their reviews, we grouped the evidence into three categories (this was chosen with reference to similar studies [37][38][39][40][41]): 1. proven evidence (n = 10): this corresponded to interventions for which the review authors      concluded that the evidence was strong and unequivocal. The majority of the reviews was made up of interventions whose evidence was proven. 2. promising (n = 5): this concerned intervention results based on either insufficient statistical power (linked to the fact that the samples were too small) or the fact that it was impossible to draw long-term conclusions from the results [21,30,36,42]. 3. lack of evidence (n = 3): according to the review authors, it was difficult to evaluate the quality of evidence in the studies they analysed because either the methodological rigour was considered poor [43], it could not be objectivised [44] or there was no description of the intervention's content, implementation or results [45].
It should be noted that the two systematic reviews on research methods and the review that sought to define the concept of preparing for parenthood escaped this categorisation.
The most effective programmes Behavioural and cognitive support through therapy groups and telephone support significantly improved parents' psychosocial health [21]. Psychoeducation showed considerable potential for improving a couple's adjustment to parenthood and their awareness of and receptiveness to their newborn [46].
The most effective programmes were those begun before the birth and those in which the parents were able to actively participate [31].
Some of the reviews evaluated interventions promoting the mental health of parents during the first few years. They concluded that, for only a small financial outlay, the benefits for the mothers and babies were substantial [47].
The promising programmes One review that was carried out on eight studies concluded that parental support programmes improved parents' psychosocial health and promoted parent-child interactions. However, a lack of statistical power precluded any formal conclusion on teenage parents [30]. Based on an analysis of 48 studies in 2014, the same lead author produced proven evidence from among the general population. She concluded that group training programmes for mothers aimed at improving their children's emotional and behavioural adjustment led to a short-term reduction in stress and anger, but she added that the data was not conclusive on the long-term effects [48].
Postnatal parental education programmes seeking to optimise infants' health and parent-child relationships resulted in an improvement in sleep, in the mothers' knowledge and in the infants' sense of security. However, larger samples would have allowed a more formal conclusion to be drawn [23].
The programmes that did not prove their effectiveness The programmes carried out in the home setting to promote socially disadvantaged families' cognitive and socio-emotional development showed questionable results [44,49]. Massage programmes aimed at boosting     the couple's relationship influences their parenting behaviours _ the mental and physical health of infants did not succeed in proving they were effective [43].
Two key areas for improvement in parenting support programmes Strengthen the quality of interventions On the whole, the parental support actions viewed parents as an entity and did not differentiate the fathers' issues from those of the mothers [43]. Moreover, there were cultural and institutional obstacles to the fathers' involvement in parental education programmes, and these need to be taken into account at the design and implementation stages of programmes aimed at this particular group [34]. Parents should have the opportunity to actively participate in interventions, and these should begin in the early prenatal stage and continue into the postnatal period. Furthermore, health promotion stakeholders seem to assume a lack of parental skills. Parental knowledge should be evaluated prior to the intervention [45]. Finally, the value of peer groups is an under-researched area [30] .
Develop complementary research Longitudinal studies were extremely under-represented. Nevertheless, a paper included in the Cochrane database claimed that long-term programmes focusing on teenage parents had long-term positive influences on parent-child relationships [30]. Longitudinal studies could be conducted by carrying out interviews at different points during pregnancy or after the birth [28].
A number of authors stated that the level of evidence of an intervention's effectiveness was difficult to evaluate due to a lack of any description given of the intervention's design, content, implementation or results [43][44][45]. Understanding how and in what context interventions were carried out appears to be essential for the evaluative framework [28].
There are few elements that define the concept of preparing for parenthood [33].
The roles of advertising and information networks could be pertinent research topics, but they remain unexplored as yet [42].
There has been little development of the relational aspect in the research on social support [42,50].
A systematic review of European studies conducted between 1999 and 2013 concluded that an evaluation of interventions must be carried out in a wider range of countries than just the United Kingdom. For example, France, Germany and Italy were not represented at all in the literature [32], thus creating an "Anglo-American bias" [28].

Consideration given to SIH in the reviews
To analyse the consideration given to SIH in efficacy analyses, each of the elements of the reviews (i.e. the title, abstract, introduction, method, results and discussion) were broken down and analysed. The results of this analysis are summarised in Table 5.
A fragmented view of the SIH issue in the reviews studied Only half of the reviews (n = 10) addressed the issue of SIH. For the most part, the notions of equity related to their results and conclusions [28,33,34,50,51]. A fifth (n = 4) clearly integrated SIH into their analysis strategy, according to the PRISMA-equity criteria [29,32,42,49].
All of the reviews that tackled the SIH question presented their results and conclusions in relation to equity [21,29,32,42,44,49]. Three of these clearly explained their results as supporting a reduction in inequalities [29,32,42], and one mentioned an increase in inequalities for some interventions [49]. A return to the original studies that were the subject of these reviews revealed that one programme, Sure Start, presented results that only seemed to benefit the most advantaged socioeconomic gradients [49].
In the studies that examined the link between children's environment and their health, there was no evidence-based for the socioeconomic characteristics used in the study protocols. However, there was "a general consensus that the combination of income, employment and education is the best way to measure household socioeconomic status" [29].
One review noted that none of the studies directly addressed housing quality as a daily living condition [32]. Very little attention was paid to childcare services in the studies (including only three referenced in Europe [32]).
Relational aspects and social links were also found to have received little research attention even though they increase the feeling of control and help prevent or reduce the risks of depression, stress and anxiety [50]. Research on social links must include a recognition of inequality because the quantity and quality of resources and how they are used can vary enormously depending on the socioeconomic class [50].
The vast majority of studies carried out dealt with at-risk populations, with little attention paid to universal services [28]. Only one review focused on community interventions [42]. Moreover, very few reviews (only three out of the 21 studies) were constructed with the aim of reducing SIH in childhood. Two of these focused on the perinatal period [32,42], and the third looked at the period from childhood through to adolescence [49].
An analysis of the reviews revealed effective strategies for addressing SIH There was consensus on the need to support and guide parents during the perinatal period in tackling SIH [28,32,42,49].  Diversifying approaches shared between state, school and neighbourhood organisations leads to better effectiveness [50]. On the one hand, parenting does not just involve the parents [33,45]; it also concerns their environment, including the professionals and organisations involved and their social contexts more generally. On the other, parenting is unique to each individual. It is affected by cultural and societal expectations as well as lifestyles [50]. Proportionate universalism is a solution that was promoted by a number of reviews, most notably those of Morrison et al. (2014) and Welsh et al. (2015) [32,49]. Michael Marmot (president of the WHO's Commission on Social Determinants of Health) explained the concept as follows: "To reduce the steepness of the social gradient in health, actions must be universal, but with a scale and intensity that is proportionate to the level of disadvantage. We call this proportionate universalism" [52]. Consequently, there is a need to explore the interventions' mechanisms to understand which interventions are (in)effective and in what contexts [21,28,45].
One review clearly addressed the importance of varying the interventions in accordance with neighbourhood organisations and services [50]. Morrison et al. (2014) showed that, in Europe, the most effective interventions (those with the best results and high levels of evidence) were the educational programmes that started at the beginning of pregnancy and included home visits from specialist personnel [32].

Discussion
To reiterate our main results, the reviews focused, for the most part, on the practices of the professionals. The programmes studied were mainly concerned with improving the mothers' knowledge and skills. All the reviews that described a proven or promising level of evidence showed that the parenting support programmes increased the mothers' self-esteem and reduced their anger, anxiety and stress levels and that they improved the infants' sleep.
Most authors concluded that the quality of the interventions could be improved by developing complementary research examining the interventions' content, implementation and results. These authors noted that the contexts in which the interventions took place were given little consideration since they were described either only scantly or not at all, making their evaluation difficult.
Only half of the reviews addressed the question of SIH. Most notably, the relational aspects and social links were found to have received little research attention even though they increase the feeling of control and help prevent or reduce the risks of depression, stress, anxiety and other mental illnesses.

Parenthood and perinatology: Semantic notions
Generally speaking, despite the fact the keywords "parent" and "parenting" were used, the population targeted by the reviews was mainly mothers. Issues relating to fathers or to the couple as a parental entity received little research attention. The methodological difficulty of understanding mother-father-child interactions was often mentioned by authors, although some did refer to the obvious impact of the interrelations of each family member [23,28,34,46].
The notions of parenting and parenting support were found to have multiple meanings and were difficult to define because they depended on many elements. In fact, these notions had different frames of reference depending on the discipline, policy or social history in question [33,50]. The Collins English Dictionary (online) defines "parenthood" as "the state of being a parent". The definition therefore refers as much to the father as to the mother.
The time immediately before and after a birth, that is from pregnancy to the first months of the child's life, has received little attention in the studies we have listed. Indeed, most of them focused either on the period preceding the birth or on the infant a few months old [31,46].

SIH and perinatal health SIH and vulnerable populations
While the actions, interventions and programmes providing support to parents were described as effective levers in addressing SIH, very few were designed with equity in health as their primary objective.
When authors did address SIH, their reviews mainly targeted population categories such as teenage parents and culturally or economically disadvantaged families. This strategy represents only one reductive vision of SIH and does not, in any way, take into account social gradients. The more socioeconomically disadvantaged an individual's situation is, the worse their health will be.
This can be observed across the whole social spectrum. Every level in the socioeconomic stratification of the population is affected by SIH [5,6]. Only one review of all those studied demonstrated the pertinence and effectiveness of a strategy constructed on proportionate universalism [49,53].
A small number of the reviews (those constructed with equity in health as a clear objective) made reference to an approach that goes beyond the notions of high-or low-risk categorisation according to individual living conditions. One of these examined the value of considering neighbourhood in terms of approximating the notion of socioeconomic deprivation [29].

SIH and norms
The fact that the majority of actions were carried out by professionals for the benefit of parents raises the implicit question of a ranking of knowledge [32,45,51]. In this sense, traditional types of knowledge, such as group knowledge or that passed on from mother to daughter, tend to be devalued. It is interesting, nevertheless, to note that the notion of childcare emerged as a result of the drive for healthcare control, which was based on a medical discourse that was both injunctive and normative [54]. However, it seems that programmes and actions could be made more effective if they were constructed within a diagnostic approach that is shared between parents, professionals and institutions [28,31,34,49].
The mobilisation of peer groups and the active participation of recipient populations were described as strategies that would increase the effectiveness of actions. The representations of the different stakeholders, whether in terms of their expectations or their needs, still appear to be negligible in programme development. The controversial results from the home-based programmes for socially disadvantaged families [36,47] could be explained in part by these divergent viewpoints.

SIH and the social determinants of health Take action early
This review of reviews has identified the most effective actions on health. Therapy groups and telephone support significantly improve parents' psychosocial health. Psychoeducation programmes that are begun before birth and which allow parents the opportunity to actively participate are the most effective [29,41].

Act with the families
Facilitators or difficulties in setting up the actions were never described. This review of reviews has also revealed, in particular, an assumed lack of knowledge among parents and even a lack of analysis of the relational modalities between the professionals and the parents [21,31,35,45,48]. Collaboration involves the need for a common language between professionals from different disciplines, parents with multiple issues with different objectives. This review has shown, however, that studies focusing on the ability to listen and the quality of the professionals-mothers-parents-newborns relationship are under-represented [35,44]. Highlighting and theorising these interrelations represents an approach to understanding how interventions function, which enables them to be made more effective [55][56][57].

Act in interaction with the context
The ability of each person to take action to improve their health depends much on the social context in which they live, think and work. Health promotion does not just concern health services but also the social determinants of health that make up this context [54][55][56]. Social determinants have been widely shown to have a strong influence on health [2,5,6,53].
Proposing programmes and actions to tackle SIH therefore calls for a global, dynamic approach to its social determinants. An analysis of the relationships between the social, economic, cultural and environmental contexts in which these projects are rolled out would allow us to understand the functioning of these interventions.

Theorise to act
Almost all the reviews highlighted the lack of description given to the intervention development processes. A number of authors noted that the interventions were evaluated on their effectiveness but that insufficient information had been given about the interventions' settings or their populations [28,34,44]. This lack of contextual elements raises the question of the transferability of the programmes or actions described in the reviews analysed. For example, many programmes focused on parents' knowledge and skills but neglected to first define the specific needs of the populations targeted. It is therefore difficult to demonstrate their effectiveness and to envisage their adaptation to other contexts.

Limitations of this review
While many public health journals value publications from the human and social sciences, this synthesis of knowledge was carried out using only bibliographic health databases. As a result, it has undoubtedly overlooked certain elements within the sociological approaches.
The methodological choice to synthesise knowledge using articles from scientific journals precluded us from analysing public health policies or the social policies that these programmes and actions were interacting with.
It should be noted that the reviews we studied were written in English, which could have led to a possible selection bias. If this was the case, the bias remained limited since the majority of scientific publications are written in English.
Because the reviews focused on high-income countries, the results of this study cannot be extrapolated to low income countries, where the situations may be different.

Conclusions
This focus on the current knowledge concerning action on social inequalities in perinatal health shows that the approach remains both modest and reductive. On the one hand, very few authors have considered the notion of equity in health, and, on the other, the vision of SIH remains limited. Parental programmes focused, for the most part, only on the mothers, and the actions tended to target the most disadvantaged populations, with no consideration of the social gradients of health. This review shows that the majority of the publications came from English-speaking countries. Currently, to our knowledge, no study has been carried out in Europe on parenting support as a means of addressing SIH among mothers and their newborns.
Parenting support interventions are complex. Their effects are variable, and they result from multiple actions, with the different stakeholders interacting in and with a particular dynamic environment. In future research, the methodological challenge will be to understand how, for whom and in what conditions interventions function.