Health behaviour interventions: Disadvantaged young women: Systematic review of evidence: Maternal nutrition
The nutritional status of mothers is an important determinant of the growth and well-being of their offspring(Reference Barker1). Maternal diet and nutrition influence growth and development of the baby during pregnancy and in postnatal life, and growth and development at these stages will influence risk of chronic disease later in life.
Women from disadvantaged backgrounds are more likely to have poor diets. Studies of women taking part in the Southampton Women's Survey, a population-based study of women aged 20–34 years(Reference Robinson, Crozier and Borland2), have demonstrated that women of lower educational attainment have less-healthy dietary patterns(Reference Robinson, Marriott and Poole3). There is also evidence that infants born to mothers of lower educational attainment are less likely to have healthy diets; of mothers participating in the Southampton Women's Survey who became pregnant, those with less-healthy diets were also found to be less likely to follow guidance on infant feeding(Reference Robinson, Marriott and Poole3).
Associations between maternal nutrition and infant growth and development suggest that improving the diets of women of child-bearing age from disadvantaged backgrounds might be an important component of public health strategies aimed at improving diet and reducing the burden of chronic disease in their offspring, as well as improving the health, nutrition and well-being of women themselves.
The development of an intervention to improve young women's nutrition needs to be informed by a systematic collation of evidence; the first step in designing a complex public health intervention(Reference Campbell, Fitzpatrick and Haines4). The present paper describes the collation of evidence from systematic reviews of interventions that aim to change health behaviours. Women from disadvantaged groups are at greater risk of a number of unhealthy behaviours; they are more likely to smoke and less likely to choose to breast-feed their babies(Reference Acheson5). The reasons underlying the choices women make about these different health behaviours are likely to be similar and to relate directly to their social and family circumstances. Interventions and strategies that have been effective in relation to these health behaviours will therefore have relevance when designing interventions to improve diet. The aim of the present review was to look for evidence relating to the features of interventions that are effective in achieving behaviour change. This process represents the first phase in the collation of evidence to inform the design of an intervention to improve the diet of young women of child-bearing age from disadvantaged backgrounds.
The methods for review of systematic reviews set out by the Centre for Reviews and Dissemination, University of York, UK(6) were followed.
Completed systematic reviews of the effectiveness of interventions that aimed to change health behaviours were included. The focus was on reviews in which the findings were based at least in part on studies of women of child-bearing age. However, studies that included children were considered if the interventions required change in health behaviour by family groups including the mothers. An a priori list of health behaviours was drawn up: diet; physical activity; smoking; breast-feeding. Systematic reviews that focused on interventions targeted at more than one of these health behaviours were also eligible for inclusion. The focus was on reviews of primary studies carried out in developed country settings, since our interest was in intervening to improve the diets of women living in the UK.
Recognised database sources of systematic reviews were searched, including the Cochrane Database of Systematic Reviews, the Database of Abstracts of Reviews of Effectiveness (at the University of York) and the Health Technology Assessment Database. Medline was searched from 1966 to January 2008.
The application of inclusion criteria and data extraction was carried out by a single reviewer (J.B.). However, the assessment of each review for evidence of effectiveness was independently validated by two further reviewers (H.M.I. and C.C.). Where the reviewers assessments were discordant, further assessment was carried out by the first reviewer and then disagreements resolved through consensus. The quality of systematic reviews was assessed according to Centre for Reviews and Dissemination guidance(6). A narrative approach to synthesis of findings was taken.
Searches of Medline resulted in the identification of 1847 potentially-relevant abstracts, which were assessed for relevance to the listed topic areas. This procedure led to the identification of sixty-one reviews that were assessed in more detail. In addition, seventy-two abstracts were identified as a result of searches of the Cochrane Database of Systematic Reviews, the Database of Abstracts of Reviews of Effectiveness and the Health Technology Assessment Database. Overall, fourteen systematic reviews met review inclusion criteria. Of the fourteen reviews, five related to diet, three to breast-feeding, one to physical activity, four to smoking and one to both diet and smoking. The methods and main findings of the fourteen systematic reviews are described in Table 1.
RCT, randomised controlled trials; CRD, Centre for Reviews and Dissemination; RR, relative risk.
Six systematic reviews of interventions to improve diet were included in the review. One related specifically to dietary knowledge and attitudes in women of child-bearing age(Reference Van Teijlingen, Wilson and Barry7), three to fruit and vegetable consumption in a range of age-groups(Reference Ciliska, Miles and O'Brien8–Reference Pomerleau, Lock and Knai10) and one to healthy eating in the general population(Reference Roe, Hunt and Bradshaw11). The sixth review related to a range of health behaviours including diet and smoking(Reference Mullen, Simons-Morton and Ramirez12). Although one of the reviews related to fruit and vegetable consumption in children and adolescents aged 5–18 years, it was included because many of the interventions were targeted at families rather than just children(Reference Knai, Pomerleau and Lock9). Much of the evidence considered by these six reviews came from intervention studies set in the USA. Three of the reviews drew specific conclusions relating to disadvantaged groups(Reference Van Teijlingen, Wilson and Barry7, Reference Ciliska, Miles and O'Brien8, Reference Pomerleau, Lock and Knai10). All six reviews were of good quality. The review that focused on interventions to improve the dietary knowledge and attitudes of women of child-bearing age identified only nine controlled trials of relevance, five of non-pregnant women and four of pregnant women, and highlighted the lack of good-quality research in this area(Reference Van Teijlingen, Wilson and Barry7). The findings of this review suggested that interventions that had an educational component and that aimed to support and empower women can improve nutrition knowledge and behaviour in both the general population and low-income groups of women. The 1999 systematic review of community interventions to increase fruit and vegetable consumption in children and adults showed that the most effective interventions used clear messages about the benefits of fruit and vegetables and multiple strategies to enforce the messages and were delivered to families over a longer period than just one or two contacts(Reference Ciliska, Miles and O'Brien8). This review also suggested that peer educators may be an effective means of delivering intervention to low-income families. The 2006 review examined the effectiveness of interventions targeted at children aged 5–18 years(Reference Knai, Pomerleau and Lock9). While the majority of the intervention studies considered in the review was based on children in a school setting, one of the features associated with effectiveness was family involvement in the intervention. The review of interventions to increase fruit and vegetable consumption in adults suggested that personal counselling or education were effective in increasing fruit and vegetable consumption in low-income and other population groups(Reference Pomerleau, Lock and Knai10). The fifth review focused on interventions to promote healthy eating in the general population(Reference Roe, Hunt and Bradshaw11). While this review suggested that interventions employing behavioural strategies such as goal setting and self-monitoring and involving personal contact with those delivering the intervention were effective, the precise relevance of this review to women of child-bearing age from disadvantaged backgrounds was not entirely clear, as findings were not reported according to gender or socio-economic status. However, such findings in general-population samples of women were considered relevant to the questions posed. The sixth review included sixty-four studies of patient education and counselling interventions delivered to disease-free populations in a range of clinical settings as part of disease-prevention programmes(Reference Mullen, Simons-Morton and Ramirez12). Although disease-free, many of the participants in trials were selected because they were at greater risk of disease; for example, many of the intervention studies focusing on diet were of individuals with raised serum cholesterol. Overall, patient education and counselling were effective in changing health behaviours; behavioural techniques, particularly self-monitoring and the use of a range of communication materials (such as media and personal communication), were most effective for improving diet and reducing smoking.
Three systematic reviews of interventions to promote and prolong breast-feeding were identified. Two of the reviews related to interventions to promote the initiation of breast-feeding(Reference Dyson, McCormick and Renfrew13, Reference Fairbank, O'Meara and Renfrew14). Both had rigorous methods; the first was a Cochrane review(Reference Dyson, McCormick and Renfrew13) and the second was a Health Technology Assessment review(Reference Fairbank, O'Meara and Renfrew14). Many of the studies considered in these reviews were of women from disadvantaged backgrounds. The Cochrane review was based on randomised controlled trials only, whereas the HTA review included non-randomised controlled studies and before-and-after studies. Both reviews suggested that one-to-one education is effective in improving breast-feeding initiation rates. In addition, the HTA review findings suggested that peer support may be effective in improving initiation in women from disadvantaged backgrounds; this evidence came from two controlled studies(Reference Fairbank, O'Meara and Renfrew14). The third systematic review was a Cochrane review of interventions to provide support for mothers in order to prolong the duration of breast-feeding(Reference Britton, McCormick and Renfrew15). This review, updated in 2006, showed that professional support is effective in prolonging any breast-feeding but its effect on exclusive breast-feeding is unclear. Additional lay support was effective in prolonging exclusive breast-feeding while its effect on duration of any breast-feeding was unclear.
A single systematic review of interventions to promote physical activity was identified, which was a Cochrane review of seventeen randomised controlled trials to assess the effectiveness of interventions promoting physical activity in adults aged ≥16 years(Reference Foster, Hillsdon and Thorogood16). Interventions giving professional guidance followed by continued support had short-term positive effects on physical activity levels, but there was little evidence of an effect on longer-term outcomes because the majority of studies considered in the review finished after 1 year of follow-up.
Four Cochrane systematic reviews of smoking cessation were identified(Reference Lumely, Oliver and Chamberlain17–Reference Secker-Walker, Gnich and Platt20). The first related to interventions during pregnancy(Reference Lumely, Oliver and Chamberlain17). Pooled data from forty-eight trials of a range of interventions were considered and overall were effective at reducing smoking (relative risk 0·94 (95% CI 0·93, 0·95)). Of the forty-eight trials the largest group was cognitive behavioural therapy interventions; their effect size was similar to that for the overall group. The smallest group of randomised controlled trials (n 2) was of social support and financial reward interventions targeted at women from disadvantaged groups. The effect size in this group was larger than overall (RR 0·77 (95% CI 0·72, 0·82)). The other three systematic reviews focused on smoking cessation interventions in the general population(Reference Lancaster and Stead18–Reference Secker-Walker, Gnich and Platt20). In the first review evidence from twenty-one randomised controlled trials of approximately 7000 individuals (the precise number of participants was not reported in this systematic review) demonstrated that individual counselling is effective in reducing the prevalence of smoking(Reference Lancaster and Stead18). The review of the influence of group programmes on smoking cessation found that they were more effective than self-help approaches(Reference Stead and Lancaster19). However, there was insufficient evidence to assess whether group programmes were more effective than individual counselling. Both forms of counselling (individual and group) have an educational component in that they involve explanation of the health risks of smoking and the benefits of smoking cessation. The final review explored the effectiveness of community interventions targeted at whole populations(Reference Secker-Walker, Gnich and Platt20). Only four of the thirty-seven studies included in the review used random assignment of communities. Overall, community interventions had a small effect on smoking prevalence, which was true even for the trials of most rigorous design.
Aspects of intervention design associated with effectiveness
Four aspects of intervention design were identified that were effective at changing more than one of the health behaviours considered in the present review: the use of an educational component; provision of continued support after the initial intervention; social support from peers or lay health workers; family involvement with the intervention. Table 2 summarises evidence of effectiveness in relation to these four aspects of intervention design for each of the included systematic reviews. Reported findings are based on independent assessments carried out by three reviewers, as outlined in the methods. Assessments were concordant at initial assessment in eight of the fourteen systematic reviews. In the remaining six reviews consensus was reached following further assessment and discussion of the review findings. All fourteen reviews suggested that an educational component involving some explanation of the health risks associated with a particular behaviour and the benefits of change was effective in changing health behaviours. This outcome applied to all the health behaviours considered. Seven of the fourteen reviews(Reference Ciliska, Miles and O'Brien8, Reference Knai, Pomerleau and Lock9, Reference Roe, Hunt and Bradshaw11, Reference Dyson, McCormick and Renfrew13, Reference Fairbank, O'Meara and Renfrew14, Reference Foster, Hillsdon and Thorogood16, Reference Stead and Lancaster19) suggested that continued support after the initial intervention of months rather than weeks duration was a feature of effective interventions, particularly in relation to diet and to breast-feeding, although the evidence in relation to physical activity was weaker(Reference Foster, Hillsdon and Thorogood16), and it appeared that prolonged support may be ineffective in relation to smoking cessation(Reference Lumely, Oliver and Chamberlain17, Reference Lancaster and Stead18, Reference Secker-Walker, Gnich and Platt20). Evidence for the importance of family involvement came mainly from the reviews of diet, with three of the five reviews suggesting that this feature was an important component of effective interventions(Reference Ciliska, Miles and O'Brien8, Reference Knai, Pomerleau and Lock9, Reference Roe, Hunt and Bradshaw11). Evidence of the importance of social support from peers or lay workers came from two of the five reviews relating to diet(Reference Van Teijlingen, Wilson and Barry7, Reference Knai, Pomerleau and Lock9) and from two of the three reviews of breast-feeding(Reference Fairbank, O'Meara and Renfrew14, Reference Britton, McCormick and Renfrew15). There was weaker evidence on the importance of social support from one of the four reviews of smoking, which related specifically to smoking cessation in pregnancy(Reference Lumely, Oliver and Chamberlain17).
++, Strong evidence to suggest that this feature of interventions is important in achieving behaviour change; +, some limited evidence that this feature of interventions is important in achieving behaviour change; −, limited evidence that this feature is ineffective; NA, effectiveness of feature not assessed as part of this systematic review.
* Explanation of health risks associated with a health behaviour and of the benefits of change.
† Ongoing support after the initial intervention over a period of months rather than weeks.
‡ Family involvement; active involvement of one or more family member in the intervention.
§ Involvement of peers or lay workers in the delivery of the intervention.
Main findings of the study
The evidence relating to interventions that might bring about behaviour change in women of child-bearing age is relatively limited, particularly for interventions to improve diet and nutrition in this group. Despite the paucity of evidence in some areas, there were some consistent findings across the systematic reviews assessed and four intervention features were identified that were effective in relation to more than one of the health behaviours considered. Interventions with an educational component were effective at bringing about improvements in behaviour for all the health behaviours considered, suggesting their potential to improve diet, increase and prolong breast-feeding, increase physical activity and induce smoking cessation. Prolonged support was important for diet, breast-feeding and physical activity but was of more limited relevance for smoking because of the evidence that brief interventions for smoking cessation can be effective. Evidence for the importance of family involvement came mainly from the diet reviews, and social support from peers and lay workers was of most importance for changing diet and breast-feeding behaviour.
Comparison with other research
There are similarities between the present findings and those of a more recent review of the content and effectiveness of behaviour-change interventions in low-income groups(Reference Michie, Jochelson and Markham21). This review considered thirteen studies of diet, smoking and physical activity interventions delivered to men and women aged ≥18 years. Consistent with the present findings, the conclusions of this review are that providing information about health behaviours, together with goal setting may be effective in low-income groups.
Two recent reviews of evidence carried out to inform the development of National Institute for Health and Clinical Excellence public health programme guidance(22, 23) have relevance to the present review. While the main focus of the National Institute for Health and Clinical Excellence review of maternal and child nutrition was on specific nutrients, including folic acid and n-3 fatty acids, and on the avoidance of alcohol during pregnancy, a small number of intervention studies aimed at improving nutrition knowledge were considered(22). Consistent with the present findings, the National Institute for Health and Clinical Excellence concludes that interventions with an educational component are effective at improving nutrition knowledge, although the evidence that was considered was related to pregnant women rather than all women of child-bearing age. The National Institute for Health and Clinical Excellence has also issued recent programme guidance on behaviour change(23). It identifies evidence of good quality showing that nutrition counselling interventions delivered to primary-care populations of men and women can change eating habits. However, it was considered that insufficient evidence was available to draw any conclusions about dietary interventions in women of child-bearing age.
Strengths and weaknesses
The methods set out by the Centre for Reviews and Dissemination for reviews of systematic reviews(6) were followed. A priori inclusion criteria were developed and recognised sources of completed systematic reviews were searched for relevant systematic reviews. While screening of abstracts and data extraction were carried out by a single reviewer, the assessment of review evidence was independently validated by two further reviewers. The majority of the systematic reviews included in the present review had rigorous methods and were of high quality; many of them were Cochrane reviews. However, most reviews did not comprehensively search the ‘grey literature’ or report on the findings of unpublished studies. Thus, publication bias is possible. Most of the reviews highlighted the lack of good-quality primary studies and identified important gaps in the literature that need to be addressed through further research. Most reviews also found that there was little evidence of cost-effectiveness in the studies considered. Despite these limitations, there are some consistent findings across a number of the systematic reviews that seem relevant to the design of an intervention study to change the health behaviours of women from disadvantaged backgrounds.
Consistent evidence was found of intervention features associated with effectiveness in bringing about change in a number of health behaviours. The findings of the present review suggest that interventions to change the health behaviour of women of child-bearing age from disadvantaged backgrounds will require an educational approach delivered in person by professionals or peers and should provide continued support after the initial intervention. There is also some evidence to suggest that social support from peers and family involvement in the intervention may be important features of interventions that aim to bring about dietary behaviour change. These findings are of relevance to the design of an intervention to improve diet in this group of women.
There is no conflict of interest. This study was funded by the University of Southampton and the Medical Research Council. All authors were responsible for the design of the review. J.B. carried out the review with assistance from H.M.I. and C.C. The first draft of the paper was produced by J.B. All authors commented on the manuscript and contributed to the critical revision of the manuscript.