Public health interventions in midwifery: a systematic review of systematic reviews

It is important to consider both the type of evidence included in reviews i.e. was the review restricted to randomised trials only or were other types of studies included and also assess how well the review was conducted methodologically. As such, a two stage process was employed: initially the level of evidence was graded and secondly, the methodological quality was assessed. Recognised frameworks were used to support this process [13, 14]. In the hierarchy of evidence, randomised controlled trials are perceived as the gold standard and as the aim of this paper is to present high quality evidence, an evidence grade was given to each review based on the Scottish Intercollegiate Guidelines Network [13] framework in order to distinguish between different levels of evidence. This framework grades the associated risk of bias based on the level of evidence in a hierarchal manner from a grade of 1++ (meta analysis and RCT evidence) through to 4 (expert opinion), as outlined in Table 1. The SIGN framework was modified as this review was restricted to systematic reviews and therefore reviews could only be graded as 1++, 1+, 1- or 2++. This paper only presents evidence which was graded 1- or above; any review graded below 1- was not deemed eligible for inclusion. Following selection of the type of evidence, the second stage focused on the methodology of eligible reviews. Clarke [15] suggests the successful interpretation of results from systematic reviews should consider the methodological conduct of the review. The methodological quality of included reviews was assessed and rated as low, medium or high quality. Appraisal of methodological quality was based on Smith et al.[14], which contains similar elements to other tools used to assess review quality, for example, the AMSTAR tool [16]. Reviews were graded as high quality if they included evidence of a search strategy, selection and inclusion criterion, assessment of publication bias and assessment of heterogeneity. Reviews were rated as medium quality if no evidence of assessment of heterogeneity or publication bias was provided and low quality reviews were those which provided evidence of a search strategy only. Effectiveness of interventions was evaluated using a similar approach to van Sluijs et al.[17]. A differentiation was made between reviews which reported a statistically significant difference (P<0.05), therefore referred to as effective and those which reported no difference in effect between control and intervention group and are referred to as inconclusive or not effective (as appropriate). This paper focuses specifically on interventions which are evidenced by a statistically significant meta analysis or where the intervention is supported by a generally positive trend of results when a meta analysis was not possible. Reviews have been included where a small number of studies reported statistically significant positive effect of the intervention however the wider interpretation of these results is limited. As outlined previously, the aim of the original review was to identify any public health intervention relevant to midwifery. However for the purpose of this paper the focus was to report on public health interventions relating to midwifery that demonstrated a statistically significant effect in favour of the intervention (referred to subsequently as effective interventions for the sake of brevity). Reviews graded 1- or above and of high methodological quality which reported evidence of no effect, are not discussed in this paper. However, they have been summarised in Table 2[18‐23]. In the case of any disagreement regarding grading of evidence, quality appraisal of reviews or effectiveness of the intervention, consensus was reached by discussion between all three authors.

There were no high quality reviews that reported on effective interventions in the pre conceptual period.

Reviews (n=4) related to screening reported on interventions relating to ultrasound [26, 27], lower genital tract infection screening [28] and the use of decision making aids [29]. Bricker et al.[26] conducted a large Health Technology Assessment review on the clinical and cost effectiveness and women’s views of USS. The review comprised of three systematic reviews on routine ultrasound in early pregnancy, routine ultrasound in late pregnancy and routine Doppler ultrasound in pregnancy which were published in the Cochrane database around the time of Bricker et al.[26] however, all have since been updated or revised in the Cochrane database, one of which has been included in this paper. The final conclusions of Bricker et al.[26] indicated that a two stage regimen of USS in pregnancy, one in early pregnancy (booking USS) and a second anomaly USS around 20 weeks, was recommended. Whitworth et al.[27] reviewed the use of ultrasound for fetal assessment in early pregnancy and concluded that it reduces failure to detect multiple pregnancy (RR 0.07 95% CI 0.03-0.17) and accuracy of gestational dating may reduce the number of inductions of labour for post term gestation (RR 0.59; 95% CI 0.42-0.83). The authors also reported there was no reduction in adverse outcomes or health service use by mothers or infants and long term follow up did not indicate detrimental effect on children’s physical or mental development. The impact of antenatal screening for lower genital tract infection for preventing preterm delivery was reviewed by Sangkomkamhang et al. [28]. The review included one large RCT (n=4155), which indicated that preterm birth before 37 weeks was significantly lower in a group of women randomised to a screening programme before 20 weeks’ gestation (RR 0.55; 95% CI 0.41-0.75). The review provides evidence to suggest there may be some benefit to introducing a universal screening programme for lower genital tract infection; however the results are based on the findings of one study. O’Connor et al.[29] conducted a review on the use of decision aids for people facing screening decisions. The meta analysis indicated that the use of decision aids, such as leaflets or DVD’s are better than usual care and resulted in: greater knowledge (MD 15.2 out of 100; 95%CI 11.7 to 18.7), perception of risk (RR 0.6; 95% CI 0.5 to 0.8), lower decisional conflict related to feeling uninformed (MD −8.3 of 100; 95% CI −11.9 to −4.8), lower decisional conflict related to personal values (MD −6.4; 95% CI −10.0 to −2.7), reduced the proportion of people who were passive in decision making (RR 0.6; 95% CI 0.5-0.8) and reduced the proportion of people who remained undecided post intervention (RR 0.5; 95% CI 0.3-0.8). Although the results suggest decision aids are effective, the effect size was not consistent across studies and only three of the included studies related directly to antenatal screening.

Eight reviews [30‐37] considered supplementation during pregnancy including iron, micronutrients, folic acid, calcium and Long Chain-Poly Unsaturated Fatty Acids (LC-PUFA’s). Two reviews [30, 31] focused on folic acid supplementation, both of which concurred that the risk of neural tube defect was significantly reduced with supplementation: Blencowe et al., [30]; 70% reduction; 95% CI 35-86 and Lumley et al., [31]; RR 0.28; 95% CI 0.13-0.58. Iron supplementation during pregnancy was reviewed by Pena-Rosas and Viteri [32] who included 49 trials relating to the prevention of iron deficiency or anaemia at term. The authors concluded that daily iron supplementation was associated with increased haemoglobin before birth (MD 6.00; 95% CI 2.75-9.25) and reduced risk of anaemia at term (RR 0.46; 95% CI 0.29- 0.72) based on meta analyses of high quality trials only. Shah et al.[33] reviewed multi-micronutrient supplementation on pregnancy outcomes and reported there was a reduction in the risk of low birth weight amongst women given micronutrient supplementation (12 studies, RR 0.81; 95% CI 0.73-0.91) and iron-folic acid supplementation (RR 0.83; 95% CI 0.74-0.93) compared to placebo. The mean birth weight was higher (11 studies; WMD 54g; 95% CI 36-72g) in infants born to mothers who had micronutrient supplementation compared to iron-folic acid supplementation (no difference with placebo).

Calcium supplementation was the focus of three reviews [34‐36]. Hofmeyr et al.[34] reported a reduction in pre-eclampsia (RR 0.68; 95% CI 0.57-0.81) and fewer babies born <2500g (RR 0.83; 95% CI 0.71-0.98). However the benefits seen were from small trials and not observed in the largest trial included. Hofmeyr et al.[35] reported that with supplementation a reduction in blood pressure (RR 0.7; 95% CI 0.57-0.86), pre-eclampsia (RR 0.48; 95% CI 0.33-0.69) and maternal death/morbidity (RR 0.80; 95% CI 0.65-0.97) was noted and advocated research to investigate calcium supplementation at community level. The most recent review [36] conducted by several of the same authors as Hofymeyr et al.[34] on calcium supplementation concluded that there was a reduced risk of increased blood pressure (RR 0.65; 95% CI 0.53-0.81) and preeclampsia (RR 0.45; 95% CI 0.31-0.65). The effect was greatest for high risk women (RR 0.22; 95% CI 0.12-0.42) and women with low baseline calcium (RR 0.36; 95% CI 0.20-0.65). Maternal death or serious morbidity was reduced (RR 0.80; 95% CI 0.65-0.97) although this was mostly in low risk women and women with low calcium and there was no effect on preterm births, stillbirth or death before discharge. Horvath et al.[37] reviewed the effect of advising high-risk pregnant women to take LC-PUFA supplementation on a number of pregnancy outcomes. The authors found a significantly lower rate of PTD <34 wks (RR 0.39; 95% CI 0.18-0.84) although this result was based on two trials (n=291). There was no effect on duration of pregnancy, PTD <37 wks, infant birth weight or the occurrence of IUGR. Although significant, the authors concluded that there was not enough evidence to recommend routine use of LC-PUFA supplements by high-risk women and that further research involving larger sample sizes was needed.

Three reviews [38‐40] considered different types of supportive interventions for women during pregnancy. These ranged from using midwifery models of care to provision of emotional support to reduce the risk of preterm delivery or low birth weight infants. Hatem et al.[38] reviewed midwife led models of care versus other models of care and concluded that the majority of women should be offered midwifery led care. Women who had midwife led models of care were less likely to experience antenatal hospitalisation (RR 0.90; 95% CI 0.81-0.99), use of regional analgesia (RR 0.81; 95% CI 0.73-0.91), episiotomy (RR 0.82; 95% CI 0.77-0.88) and instrumental delivery (RR 0.86; 95% CI 0.78 to 0.96) and were more likely to experience no intrapartum analgesia/anaesthesia (RR 1.16; 95% CI 1.05-1.29), vaginal delivery (RR 1.04; 95% CI 1.02-1.06), to feel in control during childbirth (RR 1.74; 95% CI 1.32-2.30), attendance at birth by a known midwife (RR 7.84; 95% CI 4.15-14.81) and initiate breastfeeding (RR 1.35; 95% CI 1.03 to 1.76). In addition, women who were randomised to receive midwife-led care were less likely to experience fetal loss before 24 weeks’ gestation (RR 0.79; 95% CI 0.65-0.97). There was no difference between groups for birth by caesarean section (RR 0.96; 95% CI 0.87-1.06) and no statistically significant differences in fetal loss/neonatal death of at least 24 weeks (RR 1.01; 95% CI 0.67-1.53) or fetal/neonatal death overall (RR 0.83; 95% CI 0.70-1.00) and their babies were more likely to have a shorter length of hospital stay (mean difference in days: -2.00; 95% CI −2.15 to −1.85). Hodnett & Fredericks [39] assessed the value of emotional support to women who were judged, by a health professional, to be at increased risk of preterm delivery or having a low birth weight baby. No significant effect was detected for either outcome, however, women receiving support interventions were significantly less likely to undergo a caesarean section (RR 0.88; 95% CI 0.79-0.98) and were more likely to terminate their pregnancy (RR 2.96; 95% CI 1.42-6.17). There was also a trend towards improvement in maternal psychosocial outcomes although this was not significant. Denis & Kingston [40] reviewed the effect of telephone support during pregnancy and early postpartum period specifically on smoking, preterm birth, low birth weight, breast feeding and postpartum depression. The authors report a positive effect on breast feeding (3 trials; n=618; RR=1.18; 95% CI 1.05-1.33), low birth weight (3 trials; n=2,027; RR=0.78; 95% CI 0.63-0.97) and postpartum depression at 4 weeks (RR 0.24; 95% CI 0.06-1.00) and 8 weeks (RR 0.30; 95% CI 0.10-0.92), although all were from small numbers of trials and the finding on postpartum depression was from one pilot trial including 42 women.

Educational interventions in the antenatal period were the focus of four systematic reviews [41‐44] that considered education about pelvic floor muscle training (PFMT) and promotion of smoking cessation in pregnancy Lumley et al.[41] reviewed the effect of interventions for promoting smoking cessation and included 72 studies of which 56 were RCT’s. Interventions to encourage cessation of smoking had a significant effect on the number of women smoking; 6 out of every 100 stopped, and a reduction in the number of cigarettes smoked by women was also evident. There was a significant reduction of smoking in late pregnancy (RR 0.94; 95% CI 0.93-0.96), reduction in LBW (RR 0.83; 95% CI 0.73 -0.95), preterm birth (RR 0.86; 95% CI 0.74-0.98) and an increase in mean birth weight (53.91g; 95% CI 10.44g - 95.38g). Naughton et al., [42] reviewed the use of self help interventions for smoking and reported greater likelihood of quitting compared to usual care (13.2% v 4.9%; OR 1.83; 95% CI 1.23-2.73). The cost effectiveness of this method was also emphasised, however, further research is necessary to determine the intensity level of the intervention to maximise effectiveness. Hay-Smith et al.[43] and Lemos et al.[44] reviewed pelvic floor muscle training and concluded that for primigravida women PFMT was effective. Hay-Smith et al.[43] reported that women without prior incontinence were less likely to report incontinence in late pregnancy (RR 0.44; 95% CI 0.30-0.65) and up to 6 months postpartum (RR 0.71; 95% CI 0.52-0.97) similar to Lemos et al.[44] who reported significantly reduced development of urinary incontinence from 6 weeks to 3 months after delivery (OR 0.45; 95% CI 0.3-0.66; 4x RCT; n=675). Pregnant women with persistent incontinence 3 months after delivery and received PMFT were less likely to report urinary incontinence at 12 months post delivery (RR 0.79; 95% CI 0.70-0.90) and less likely to report faecal incontinence at 12 months (RR 0.52; 95%CI 0.31-0.87) [43].

One review by Dennis & Creedy [45] considered interventions to prevent postnatal depression and all but one involved an intervention from a health professional. The authors reported that preliminary evidence suggests that intensive postnatal nursing home visits with at risk mothers assisted prevention of postpartum depression (RR 0.67; 95%CI 0.51-0.89).

Cluett & Burns [46] reviewed immersion in water for labour or birth (n=11) and reported from a meta analysis of 6 RCT’s. There was evidence to indicate that immersion in water for the first stage of labour significantly reduced the rate of epidural, spinal, paracervical analgesia and anaesthetic analgesia (478/1254 versus 529/1245; OR 0.82; 95% CI 0.70-0.98; p 0.025). However further research is required on other outcomes where there was no difference identified including assisted vaginal deliveries, C/S, perineal trauma, maternal infection, Apgar score < 7 at 5 mins, neonatal unit admissions or neonatal infection rates. Rabe et al.[47] reviewed delayed umbilical cord clamping and indicated from a meta analysis that there are benefits for both term and preterm infants. A delay of 30–120 seconds of cord clamping reduced the need for transfusions (RR 2.01; 95% CI 1.24-3.27, p=0.0049) and intraventricular haemorrhage (RR 1.74; 95% CI 1.08-2.81, p=0.022) in infants born <37 weeks [47]. Although the short term benefits are clear, further longitudinal work is needed to clarify the long term benefits.

The birth setting was the subject of four reviews although all were on different aspects. Hodnett et al.[48] reviewed the evidence regarding alternative versus conventional institutional settings for birth, which did not include any trials conducted in free standing birth centres. The review reported that for women allocated to the intervention (alternative setting) there was a significant increased likelihood of no analgesia/anaesthesia (RR 1.17; 95% CI 1.01-1.35), spontaneous vaginal delivery (RR 1.04; 95% CI 1.02-1.06), very positive views of care (RR 1.96; 95% CI 1.78-2.15), breastfeeding rates at 6–8 weeks (RR 1.04; 95% CI 1.02-1.06) and decreased episiotomy rate (RR 0.83; 95% CI 0.77-0.90). There was no effect on serious perinatal or maternal morbidity or mortality. Continuous support during childbirth was reviewed by Hodnett et al.[49]. The intervention involved one to one support during labour and found increased likelihood of shorter labour (WMD −0.43 hours; 95% CI −0.83 to −0.04), spontaneous vaginal delivery (RR 1.07; 95% CI 1.04 to 1.12) and were less likely to have intrapartum analgesia (RR 0.89; 95% CI 0.82- 0.96) or report dissatisfaction with childbirth experience (RR 0.73; 95% CI 0.65- 0.83). The authors only reported on outcomes where at least four trials were included in the meta analysis and highlighted that, generally, continuous intrapartum support was associated with greater benefits when it was not a member of hospital staff, when it began in early labour and in settings where epidural was not routinely available. Hodnett et al.[49] concluded that continuous support should be the norm rather than the exception for all women and further research is required as to the effectiveness of doula or lay support.

One review considered interventions aimed at reducing caesarean section rates [50]. Chaillet & Dumont [50] reported from a meta analysis that regular audit, detailed feedback regarding aspects of caesarean section performance (responsibility for decision making, rates, review of cases in clinical practice and multi faceted strategy approaches, such as development of guidelines, education of health professionals and women about vaginal birth after caesarean section (VBAC) were effective for reducing the caesarean section rate (RR 0.81; 95% CI 0.75-0.87). Details of relative risk for each type of strategy are included in Table 3.

Reviews on this topic generally related to either support or promotion of breastfeeding. Britton et al.[51] reviewed the evidence in relation to support for breastfeeding mothers and key findings indicated that all forms of extra support for any breastfeeding (exclusive or partial) increased the duration of breastfeeding (RR 0.91; 95% CI 0.86-0.96) and the effect was greater for exclusive breastfeeding (RR 0.81; 95% CI 0.74-0.89). These findings were supported by Chung et al.[52] and Sikorski et al.[53]. Breastfeeding interventions included in both Britton et al.[51] and Chung et al.[52] involved formal or structured breastfeeding education, informal breastfeeding education or breastfeeding support either lay or professional. Chung et al.[52] from a meta analysis of 34 studies reported that breastfeeding interventions were effective in relation to increasing short term (1-3mths) and long-term (6-8mths) exclusive breastfeeding (RR 1.28; 95% CI 1.11-1.48 and RR 1.44; 95% CI 1.13-1.84) although statistically significant heterogeneity was noted for short term exclusive breast feeding (I² =55%; p= 0.006). The authors also highlighted an increased rate (22%) of any (RR 1.22; 95%CI 1.08-1.37) and exclusive (RR 1.65; 95%CI 1.03-2.63) short term breastfeeding with interventions that included a component of lay support. Sikorski et al.[53] reviewed additional support versus standard care and concluded that additional professional support was more beneficial than standard care for duration of any breastfeeding (RR 0.89, 95% CI 0.81-0.97; 10xRCT; n=19,696) and additional lay support was effective in reducing the cessation of exclusive breastfeeding (RR 0.66; 95% CI 0.49-0.89; 5xRCT; n=2530). Effect sizes for interventions with an antenatal education element (RR 0.85; 95% CI 0.70-1.04) were not statistically significant, while those with a postnatal element alone were (RR 0.80; 95% CI 0.80-0.96). Four trials using WHO/UNICEF training showed significant benefit in prolonging exclusive breastfeeding (RR 0.70; 95% CI 0.53-0.93), but were highly heterogeneous. The authors highlight the need to assess support in different settings especially with low rates, conduct economic analyses and use qualitative research to explore specific elements of support. Dyson et al.[54] focused on breastfeeding initiation rates and indicated from a meta analysis of five studies (n=582) that breastfeeding education had a significant effect on increasing initiation rates (RR 1.57, 95% CI 1.15-2.15, p=0.005) compared to standard care in low income groups although substantial statistical heterogeneity was noted (I²=53.4%). Early skin to skin contact was reviewed by Moore et al.[55] who reported statistically significant effects of early skin to skin on breastfeeding at one to four months post birth (OR 1.82, 95% CI 1.08-3.07) and breastfeeding duration (WMD 42.55, 95% CI −1.69 -86.79). In this review, data from more than two trials were only available for a small number of outcomes (8/64). Ahmed & Sands [56] reviewed breast feeding interventions. While the authors were unable to conduct a meta analysis they found from individual trials, statistically significant results relating to kangaroo care, peer counselling, in home breast milk measurement, and post discharge lactation support for improving breast feeding outcomes.

One review focused on improving maternal mental health and considered postnatal psychological and psychosocial interventions [57]. Dennis & Hodnett [57] reported that any psychosocial or psychological intervention compared to usual postpartum care was associated with a reduction in the likelihood of continued depression from their review of nine trials. Examples of psychosocial and psychological interventions reviewed included non-directive counselling, supportive interactions, delivered via telephone, home or clinic visits, or individual or group sessions in the postpartum period by a health professional or lay person, cognitive behavioural therapy and interpersonal psychotherapy.

One review considered support for women in relation to weight reduction in the post partum period [58] focusing on the effect of diet or exercise or both for reducing weight after childbirth. They found that women who took part in a diet (1 trial; n=45; WMD −1.70 kg; 95% CI −2.08 to −1.32; z=8.73; p<0.00001), and women on a diet plus exercise programme (4 trials; n=169; WMD −2.89 kg; 95% CI −4.83 to −0.95; z=2.92; p<=0.00049), lost significantly more weight than women in the usual care. The authors also noted that there was no adverse effect on breastfeeding, although cautioned that further research is necessary to confirm this finding. Three reviews considered extra support for vulnerable groups of women in the form of home visiting or parenting interventions [59‐61]. Corcoran & Pillai [59] reviewed rates in repeat pregnancy following the introduction of hospital-based programmes providing education and counselling to a sample of adolescent mothers. They found that although there was a 50% reduction in the odds of repeat pregnancy compared to comparison-control conditions at 19months (OR 0.474; 95% CI 0.322-0.695), the effect had dissipated by 31 months. All studies were US based and the majority conducted in low income groups (74%) and African Americans (60%). Two reviews focused on parenting interventions [60, 61]. Pinquart and Teubert [60] reported small effects on parenting, parental stress, child abuse, health promoting behaviour, cognitive, social development, motor development, child mental health, parental mental health & couple adjustment from parenting education interventions. Vanderveen et al.[61] demonstrated an overall positive effect on neurodevelopment from early parental interventions (all involved teaching or enhancing parental skills) lasting up to 36 months. Meta analysis of twelve studies indicated higher cognitive scores at 12 months (WMD 5.57; 95% CI 2.29-8.86; p=0.0009), at 24 months (7 studies; WMD 7.59; 95% CI 5.01-14.31; p=0.0003) and at 36 months (2 studies; WMD 9.66; 95% CI 5.01-14.31; p=0.0001), but not at 5 yrs (3 studies p=0.24). The authors suggest further research is needed to clarify the most effective interventions and the long term effect.

The review identified many gaps in systematic review literature relating to core midwifery practice, which potentially could impact on public health population goals. The UK Department of Health, Public Health Strategy [66] emphasizes the importance of improving maternal health and the subsequent impact on reducing infant mortality and premature births and yet this review identified limited systematic review evidence to support the implementation of midwifery interventions that could impact on perinatal morbidity and mortality. The review also highlighted it was difficult to accurately assess the potential public health impact in terms of effectiveness as some interventions were not well evaluated, evidenced by the large number of inconclusive reviews and reviews demonstrating no effect. The review of reviews identified some interventions that were effective but were limited in terms of methodological quality of included studies, for example, small numbers and design flaws, thus demonstrating the need for robust research and evaluation. One example of this is systematic review evidence in relation to weight management or obesity; a topic of growing concern to maternity care providers and yet the evidence from systematic reviews is limited in terms of quality. The systematic reviews included in the original review generally indicated that additional support related to diet or exercise for women in the postnatal period was effective, however, only one review was of a high quality. Another example of this is the evidence around home visiting for vulnerable groups of women in the postnatal period. While a significant body of research, including longitudinal studies has been published on parenting interventions indicating generally positive effects [67, 68] the evidence from this current systematic review of reviews is mixed. Current early years governmental policy in the UK focuses on giving children the best start in life and various interventions have been, or are currently being rolled out, for example, SureStart and the Family Nurse Partnership, however the longer term impact on women and families remains to be seen. Logic models highlight the causal linkage between inputs, outputs and outcomes (24). This is illustrated very clearly in relation to support for parents in the form of parenting interventions (input) which can result in the short term outcome of increasing support for women to improve health and lifestyle; optimize lifestyle and child development beyond the immediate perinatal period (medium term) and in the long term improve family health and wellbeing for this generation and those to come.

In order for midwives to utilise their potential in relation to public health it is important not only to consider the interventions that could be implemented but also take cognisance of wider strategies and policy relating to public health. The logic model (Figure 2), which was developed as a parallel process to the review, provides an overarching framework that should be used by midwives to visualise their contribution to public health. The model illustrates possible future roles but also facilitates recognition of the current contribution of midwives to improving the health of women and their families as part of their core role. An example of this is how vulnerable women (either social or medical) could be identified in the antenatal period by midwives and a supportive or educational intervention implemented which would result in improved outcomes in the short term i.e. reduced pre term birth or improved birth weight. A medium term outcome of this intervention would focus on optimising lifestyle beyond the perinatal period for example collaborating with health visiting services to provide education and support that would potentially have a longer term outcome of improved family health and well being. The review did not identify any systematic reviews which specifically focused on interventions relating to midwifery public health roles, highlighting a gap in review evidence. Biro [69] suggests it may be challenging for midwives to think beyond individual women but ultimately necessary in order to meet the challenge of public health to improve population health. Reframing routine midwifery activities in a public health context, identifying midwives as public practitioners and building on existing activities, such as collaboration, organisation of care and interagency working are essential to clearly define the relationship between midwifery and public health. An earlier, wider review on health-led parenting interventions in pregnancy and the first three years of life [8] suggested that many interventions, particularly in relation to supporting parenting, could be provided as part of routine care and that although the optimal time to start programmes was not clear, there was some consensus that those initiated in the antenatal period were more effective. Development of the public health role of the midwife will also require strategic thinking and support from planners and commissioners of maternity services to ensure that midwives can influence policy and effectively implement public health strategies. This will involve dedicating time and resources to develop local policies, providing training for midwives and building good relationships with other healthcare disciplines to work together.

There are a number of methodological challenges in using systematic review evidence which must be taken into account. It is difficult to summarise the evidence from systematic reviews as often there is significant diversity between interventions included in individual reviews or outcome measures used. In addition the results presented may be inconsistent between reviews or inconclusive, however, Smith et al.[14] suggest the strength of systematic reviews of reviews is that the best quality reviews can be highlighted in a single document. Systematic reviews are generally limited to published work and thus may be subject to publication bias. In addition, more recent, potentially conflicting, research may be available since the review was published or there may be effective interventions that have not been evaluated in a systematic review. A recent Cochrane overview of systematic reviews [70] highlighted that such reviews provide an accessible summary on the totality of the evidence in the area and minimised the need for referral to individual reviews, however suggested that readers may wish to do so for specific details. This review was similar in that it covered a broad scope of the evidence in relation to public health, providing a strategic overview while also providing a valuable resource for those who wish to consult individual reviews for additional specific details. In this paper, only high quality reviews (based on level of included evidence and methodology of review) reporting on effective interventions were included. While this provides reassurance regarding review findings, in that the conclusions are based on top level evidence, some interventions demonstrating effect may have been excluded because the review itself did not meet either the quality or level of evidence criteria for inclusion. In most cases this relates to areas worthy of future investigation, which need more robust evaluations. The search strategy utilised in the review was specifically focused on the public health role of the midwife and therefore incorporated key terms relative to key areas. However in doing so, some postnatal interventions, which extend beyond the role of the midwife, for example, parenting interventions that continue into early childhood may not have been included. In addition, due to the inclusion and exclusion criteria applied, it is possible that extensive broad reviews on particular topics have been excluded from this review due to the nature of evidence included within them, for example, the NICE Guideline on Antenatal and Postnatal Mental Health [9]. However, it is recognised these are valuable resources and contribute to wider understanding on specific subjects.