Figure 1 presents an example PRISMA (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) flow diagram for child SEWB and the other flow diagrams are presented in Additional file 2. Nineteen unique studies were identified across the four searches. Six studies primarily focused on child SEWB [36-41]. Shaw et al [36] was the only one of these six studies to not also assess parental mental health. Four studies primarily focused on infant sleep outcomes [42-45]: of which, two also assessed parental mental health and child SEWB [42, 43] and one also assessed parental mental health [45]. Five studies focused on home learning environment and reported on no other outcome areas [46-50]. Four studies focused on parental mental health and reported on no other outcome areas [51-55].
<Insert Figure 1 here>
A summary of study characteristics for each of the priority area outcomes is presented in Table 2. Although a small number of studies for each priority area, there were some observations: (i) Child SEWB studies were predominantly targeted, low risk of bias, and delivered by healthcare staff, (ii) Home learning environment studies were all universal, without group components and predominantly delivered in healthcare settings, (iii) Infant sleep studies were predominantly single-session and delivered by researchers, and (iv) parental mental health studies were all universal, and often single -session and delivered by healthcare staff.
Table 2. Summary of study characteristics for each priority area
|
|
Child Social & Emotional Wellbeing
(n=6)
|
Home Learning environment
(n=5)
|
Infant sleep
(n=4)
|
Parental mental health
(n=4)
|
Approach
|
|
|
|
|
|
Universal
|
1
|
5
|
2
|
4
|
|
Selected/indicated
|
5
|
0
|
2
|
0
|
Risk of bias
|
|
|
|
|
|
High
|
1
|
1
|
1
|
1
|
|
Medium
|
0
|
2
|
1
|
2
|
|
Low
|
5
|
2
|
2
|
1
|
Group based component
|
|
|
|
|
|
Yes
|
3
|
0
|
2
|
2
|
|
No
|
3
|
5
|
2
|
2
|
Number of sessions
|
|
|
|
|
|
1
|
2
|
1
|
3
|
3
|
|
2
|
1
|
2
|
1
|
0
|
|
3
|
3
|
1
|
0
|
0
|
|
4
|
0
|
1
|
0
|
1
|
Setting
|
|
|
|
|
|
Family home
|
2
|
1
|
1
|
2
|
|
Health-related
|
4
|
4
|
3
|
2
|
Fields of intervention provider
|
|
|
|
|
|
Health
|
5
|
3
|
2
|
3
|
|
Social
|
2
|
1
|
0
|
1
|
|
Research
|
0
|
0
|
2
|
0
|
|
Other
|
2
|
2
|
0
|
0
|
Table 3a highlights how individual study characteristics are associated with effectiveness whereas Table 3b highlights how indicators of engagement from families is associated with effectiveness. Individual details of the studies are presented in Additional file 3.
<Insert Table 3a and 3b here>
- Child social and emotional wellbeing
Of the eight studies that report outcomes relating to child SEWB, six were considered to primarily target child SEWB [36-41] whereas two primarily focused on infant sleep in studies recruiting families that presented with infant sleep problems [42, 43]. Studies examining improvements for child SEWB were mostly well-conducted with 7 of 8 fulfilling all or most of the NICE checklist criteria (Hayes et al [38] being the exception). The outcome measures selected were comparable across studies (five of the studies used the Child Behaviour Checklist). Despite the robust study designs, the interventions themselves varied considerably in the format they were delivered (e.g. group/individual, home visit/health centre).
From these studies, there is evidence that populations with identified risk factors can benefit from brief interventions that target child SEWB. Specifically, interventions that focused on motivational interviewing and examining family context to identify appropriate needs had benefits two years later [36, 37]. Of the two studies primarily targeting improving sleep, Gradisar et al [42] examined children of comparable ages to those in the other studies whereas Hiscock et al [43] recruited a younger sample of infants but as their interventions focused on sleep it is not unexpected that child SEWB remained unchanged.
There was little evidence of the effectiveness of universal interventions. Hiscock et al [39] was both the only (i) universal intervention and (ii) one of two studies targeting child SEWB that did not demonstrate a benefit. A structurally similar group-based intervention also held in maternal child health centres in Melbourne, Australia showed significant improvement in child SEWB [38]. Hiscock’s study [39] received a higher quality appraisal than Hayes’s study [38], but an alternative explanation may be that Hayes et al’s sample had self-referred so may have been more engaged or motivated.
- Infant sleep
Three of 4 studies tested infant sleep interventions in indicated/selected populations, with Gradisar et al [42] asking participants to self-refer if their child was experiencing a sleep problem while Hiscock et al’s studies [43, 45] both recruited patients who had been screened for a sleep problem through routine health visits.
The interventions were all essentially single session but differed in the approach taken. There was evidence of effective child behavioural interventions [42, 43] but weak evidence for interventions using parent education alone [44, 45]. Child behavioural interventions may be the ‘best bet’ approach as these interventions were supported by two studies of high methodological quality. Both these studies permitted parents to choose one of two interventions. Interestingly, Gradisar et al [42] showed that two interventions improved different sleep outcomes (e.g. one reduced number of awakenings whereas the other increased total sleep time).
- Home learning environment
The five studies measuring outcomes relating to cultivating a positive home learning environment all tested universal interventions that recruited families engaging with routine health visits [46-50]. All five interventions could be delivered within very short timeframes (e.g. waiting rooms, 5-minute time slots) or independent of practitioner involvement. However, the studies used different techniques (distribution of books/reading materials/play activities, and literacy promotion programs).
There is currently a paucity of high-quality evidence for brief interventions aiming to improve the home learning environment. Any positive evidence is undermined by methodological issues. Studies reporting positive intervention effects predominantly used non-validated tools devised for the purposes of testing the specific intervention. Goldfeld et al [47] was the only study not to report any improvement on any outcomes. This study had the highest quality rating and used a variety of validated outcome tools, as such the evidence is more robust and generalisable. Other methodological limitations include follow up time points limited to 6 months or less [46, 49, 50], and no data on the number of participants that were initially approached nor retention rates [48]. Goldfeld et al [47] had high retention rates at 4-year follow up and as such the findings are more indicative of the long-term impact (or lack) of the intervention.
- Parent mental health
Twelve studies reported on parental mental health outcomes. Of these, four interventions focus on parent mental health as their primary outcome [51-55], but three interventions primarily focus on infant sleep disorders [42, 43, 45] and five primarily focus on child SEWB [36, 37, 39-41].
Many of the intervention approaches such as individual counselling and psychoeducational programs were delivered in subtly different formats throughout the different trials. Therefore, it is not possible to definitively recommend one implementation method over another. All four interventions targeting parental mental health demonstrated positive results. Interventions targeting parental mental health were all delivered by a nurse and therefore should be adaptable to most universal child health and development programs. All studies apart from Glavin et al [54, 55] were conducted through existing services in Australia so it is unclear whether they would be applicable within similar contexts. Glavin et al’s counselling intervention was the only intervention modelled on the principle of ‘proportionate universalism’; those from a universal base with increased need received more sessions or referral to additional services.
The group intervention tested in Fisher et al’s studies [51, 53] recruited couples. Further adaption and testing would be required to implement these interventions either with a single parent or a single parent and supportive other. In the study which did not target couples by Giallo et al [52], the follow up time was limited but findings suggest that self-directed intervention alone is not as beneficial as with telephone support.
The evidence is predominantly negative when the intervention primarily addresses other outcome areas. Among the child SEWB studies, Dishion et al [37] reported improvements in parental mental health and child SEWB, whilst Hiscock et al [39], Dittman et al [40], and Hiscock et al [41] demonstrated no improvements in parental mental health. Interestingly, Hayes et al [38] reported improvements in child SEWB and parental depression, anxiety and stress but the wait-list control group only reported improvements in depression when they received the intervention. Among the sleep studies, intervention groups in both of Hiscock et al’s studies [43, 45] showed greater improvements in depression. Yet, only Hiscock et al [43] showed an effect on infant sleep outcomes. The inverse was observed by Gradisar et al [42] as while infant sleep was improved, parental mental health was unaffected.
Whilst this review aimed to assess interventions directed to both maternal and paternal populations, no brief intervention studies were identified that addressed the mental health of fathers. All other studies represented preventative interventions used to mitigate the risk of mothers developing mental illness in the post-partum period.
The evidence suggests that a classic model of services structured on a fixed number of repeated sessions with mothers is not necessary to improve mental health outcomes and that brief interventions can be effective. Consideration should be made to the theoretical underpinnings of interventions to identify the causative links between mental health improvement and intervention components.
Can these interventions be delivered through a UCHS platform?
Brief interventions should theoretically be acceptable to both families and healthcare practitioners and entail less resources to deliver. From the evidence reviewed we derived data to examine recruitment, adherence and retention rates; providing an indication of the acceptability of these interventions to families to complement the review of effectiveness. Details on indices of engagement are presented in Table 3b.
Uptake
In the 12 studies testing universal interventions, the proportion of participants completing baseline assessments varied across studies from 32.9 to 95.6%, with two studies not providing details on the numbers approached. Eight studies reported the number of participants who explicitly refused to participate. Of these, the refusal rates coming into the studies ranged between 8.8% to 15.6% for the four that focused on parental mental health [51-55], 26.6% for Hiscock et al’s study targeting infant sleep [45], 11.2% for Hiscock et al’s study targeting child SEWB [39], and 10.7% to 35.1% for studies targeting home learning environment [47, 50]. These low refusal rates suggest that most interventions did appeal to parents. Mental health interventions that could be perceived as stigmatizing were also taken up well by the families.
The six studies that recruited selected/indicated populations either (i) proactively screened participants through routine health visits or directly contacting families by telephone or (ii) advertised the intervention and relied upon participants self-referring. The percentage of participants refusing screening ranged from 3.4% to 27.8%. The percentage of participants defined as ineligible after screening ranged from 19.8 to 47.2. It was difficult to determine numbers ineligible and numbers refusing to participate and therefore the extent that the service appeals to patients. Furthermore, there were few details in selected/indicated populations regarding the time and resources for screening against the proportion of patients ultimately eligible.
Risk factors for non-participation
Twelve of the 19 studies identified in this review stated that sufficient language to complete the assessments was an explicit inclusion criterion. However, any service rolled out on a universal platform would have to explicitly encourage participation from culturally and ethnically diverse populations as many of these populations are at a higher risk of poor parental and child outcomes. Not being a native speaker is a recognised risk factor for not receiving appropriate healthcare resources [56]. Consequently, the interventions may not be generalizable for culturally diverse populations. In addition, several studies highlighted that participation was associated with stress and mood variables [36, 52], indices of social deprivation and socio-economic status [39, 41, 43, 45], levels of education [41, 45, 51, 53], or non-native resident/speaker [45, 52]. This review highlights that socio-economic factors were a barrier to engagement and adherence; even when interventions have been designed to be brief and provided a financial incentive.
Adherence
While examining uptake and the risk factors for non-participation provide an indication of the initial appeal of the intervention, measures of adherence to the intervention (i.e. completed all aspects) indicate how well interventions engage with and are accepted by families. Even within these brief interventions the number of parents that attended all sessions of the intervention were limited. If brief interventions have been appropriately designed, each session should be designed to impart the maximal amount of information within a limited timeframe. As such, missing a single session may mean that an individual misses vital intervention content that could improve the treatment effect. For example, Fisher et al [53] found a significantly lower prevalence of mental health diagnoses in those that received the full intervention compared to the group who received usual care, whereas receiving only the partial intervention was not associated with a reduction in prevalence of mental health diagnoses. In addition, the variable rates of attendance for interventions with a limited number of sessions highlights that interventions with a higher number of sessions may have increasing difficulty to retain participants. This is seen even in interventions that recruited participants actively seeking help [38, 42].
Retention
Encouragingly, retention rates were routinely high across studies irrespective of timepoint. Only two studies reported retention rates lower than 70% [46, 54, 55]. Of the targeted interventions, the only two studies with retention rates below 70% were the two studies that recruited through self-referral. Gradisar et al [42] showed a 53.5% retention rate at an interim assessment but managed to gain 100% follow up at 12 months. However, Hayes et al [38] exhibited less than 60% retention at less than six months. As this study had high attrition between self-referral and a baseline assessment, it suggests that the parallel triage service may have been a serious confounder.
Synthesis of evidence: ‘Best bet’ interventions
A combination of critical assessment of effectiveness data, indicators of acceptability, and assessments of quality (bias) across all studies was performed to identify potential ‘best bet interventions’ for adoption into UCHS. Studies with a combination of ‘Long’/’Medium’ follow up, ‘Low’/’Medium’ risk of bias, and green-coded effectiveness data (Table 2) were critiqued against potential implementation issues to determine whether recommendable in the context of UCHS.
There were two “best bet” interventions identified for potential use in universal services [45, 54, 55]. While Hiscock et al’s [45] child behavioural intervention did not elicit a benefit on sleep outcomes, the intervention was effective at reducing levels of parental depression. As the intervention itself entailed few resources and a single group session we would advocate the use of this intervention for new parents to improve maternal mental health; although there was evidence that those of a lower socio-economic status may be less likely to engage in the intervention. Future research should aim to measure the cost-effectiveness of each part of the program (e.g. DVD, self-help material, group session). We would also recommend Glavin et al’s [54, 55] intervention based on triage for mental health symptoms in all mothers. The intervention was associated with benefits in parent mental health at scale and over a long follow up period. More importantly this intervention was upskilling existing staff to provide additional support as part of universal care making it far more sustainable. The only main limitation is the quasi-experimental approach in which this was tested but as this was a pragmatic trial it is perhaps more reflective of how the intervention would work once implemented in a real-world context. While Christakis et al’s intervention [46] was effective, a fuller understanding of the mechanistic theory underlying the intervention’s benefit is needed along with a longer term follow up that demonstrates the cost-effectiveness of providing the toys used in the intervention.
Of the targeted interventions, we recommend Hiscock et al’s [43] intervention as it effected long term change on both sleep disorders and parental mental health and is feasibly delivered through health centres. In contrast, while Shaw et al [36], Dishion et al [37], and Gradisar et al [42] all demonstrated that their respective interventions were effective at long term change, the feasibility of delivery via existing UCHS has yet to be established as these studies primarily used research staff for delivery. In both Shaw et al [36] and Dishion et al [37], participants were financially reimbursed for assessments, which is not feasible for most UCHS; and the same intervention was shown ineffective in a study by Hiscock et al [41]. In addition, Gradisar et al [42] had a relatively small sample size that were predominantly in a marriage-like relationship, had education qualifications, and were middle- to high-income earners so has not been tested at scale in families from wider socio-demographic backgrounds.