Evaluation of a community-based intervention to ... - Semantic Scholar

9 downloads 861 Views 125KB Size Report
pregnant women, support of breastfeeding mothers and local awareness-raising activities. The study subjects were all women attending the antenatal booking ...
Journal of Public Health Medicine

Vol. 22, No. 2, pp. 138–145 Printed in Great Britain

Evaluation of a community-based intervention to increase breastfeeding prevalence Rhona J. McInnes, Janet G. Love and David H. Stone

Abstract Background The aim of the study was to determine whether peer counselling in the antenatal and post-natal period would increase the prevalence and duration of breastfeeding among low-income women in Glasgow. Methods The study was designed as a quasi-experimental evaluation of a community-based breastfeeding promotion programme. The intervention comprised peer counselling of pregnant women, support of breastfeeding mothers and local awareness-raising activities. The study subjects were all women attending the antenatal booking clinic resident in either the intervention or control area. Data were collected by means of four self-completion questionnaires, two administered antenatally and two post-natally. The proportions intending to breastfeed or breastfeeding in the intervention and control groups were compared at each assessment and over time. Results Of the 995 women enrolled in the study, data were available for analysis on 919 (92 per cent) to 6 weeks postnatally. At booking, 18 per cent of the intervention group and 21 per cent of the control group stated an intention to breastfeed. At delivery, the proportions initiating breastfeeding were 23 per cent of the intervention subjects and 20 per cent of the controls, and by 6 weeks post-natally, the proportion providing any breast milk had declined to 10 per cent of the intervention group and 8 per cent of the control group. Using multivariate analysis to adjust for factors (such as socio-economic status) known to influence breastfeeding, the breastfeeding prevalence was significantly higher in the intervention group relative to the controls at delivery [odds ratio (OR) 2.0; 95 per cent confidence interval (CI) 1.2–3.1, p = 0.006]. By 6 weeks post-natally the difference between the two groups was not statistically significant (OR 1.8; 95 per cent CI 1.0–3.4, p = 0.07). Conclusions As the impact of the intervention was not sustained even for the modest duration of 6 weeks postnatally, it would be premature to justify widespread use of peer support programmes to increase the prevalence of breastfeeding in socially disadvantaged communities.

United Kingdom4 or 12 months in the United States.5 According to the World Health Organization, around 95 per cent of women are physiologically capable of breastfeeding.6 In Scotland in the early 1990s, only about 50 per cent of mothers attempted to breastfeed7 and in some peripheral housing estates in Glasgow (Scotland’s largest city) the prevalence of breastfeeding at the end of the first week of life was around 10 per cent.8 The clear inverse relationship between social class and breastfeeding presumably accounts for this phenomenon. Generally, older, well-educated, non-smoking women from affluent backgrounds are more likely to want to breastfeed than younger, poorly educated smoking women from disadvantaged areas.9 In 1993, the health promotion department of Greater Glasgow Health Board (GGHB) took a strategic decision to attempt to increase the exceptionally low prevalence of breastfeeding in many of the more disadvantaged areas of the city. The present study was one outcome of that initiative. Conceived as a form of action research, the project was predicated on the notion that an innovative, community-based approach to breastfeeding promotion should be developed, implemented and evaluated. The starting point was the recognition that the National Health Service had made a relatively limited impact on infant feeding practices. The research evidence suggested that doctors and midwives had little influence on the choice of feeding method. Most women appeared to have decided before any contact with health professionals in pregnancy.7,10 Moreover, there was a rapid decline in the prevalence of breastfeeding over the first 2 weeks of the infant’s life despite intensive contact with health professionals. Given the strong feelings that most mothers have about infant feeding, the powerful effects of role models and peer groups, and the failure of the mainstream health care system to achieve substantial changes in breastfeeding behaviour, we

Keywords: infant feeding, health promotion, evaluation 1

Paediatric Epidemiology and Community Health (PEACH) Unit, Department of Child Health, University of Glasgow, Yorkhill Hospital, Glasgow G3 8SJ.

Introduction The benefits of breastfeeding for both infant and mother1–3 are well established. To optimize the health benefits, breastfeeding has been recommended for a period of at least 4 months in the

2

Robertson Centre for Biostatistics, University of Glasgow, Glasgow.

Rhona J. McInnes,1 Research Midwife Janet G. Love,2 Research Fellow David H. Stone,1 Director Address correspondence to Dr Stone.

q Faculty of Public Health Medicine 2000

139

BREASTFE EDING PROMOTION

hypothesized that a community-based intervention involving local peer counselling and support held the most promise for increasing breastfeeding prevalence among low-income women. A breastfeeding promotion initiative was therefore launched in a socio-economically disadvantaged urban setting in the West of Scotland with known low breastfeeding rates. The model of peer counselling and support has been used in several studies elsewhere. A study from Utah employed historical controls to evaluate peer counselling among Native American subjects and reported encouraging results albeit on small numbers.11 A randomized controlled trial of peer counselling in Mexico City reported a significant increase in breastfeeding exclusivity and duration in a sample of 130 women with a high rate of breastfeeding initiation.12 A peer counselling programme in Iowa provided peer support to lowincome women and demonstrated increased breastfeeding duration in the intervention group but had a low follow-up rate (around 50 per cent).13 In Britain, the La Leche League Peer Counselling Programme has become popular, despite the absence of rigorous evaluation. The aim of the project was to plan, implement and evaluate an innovative, community-based intervention designed to increase the prevalence and duration of breastfeeding in a disadvantaged area of Glasgow. This paper reports an analysis of the impact of the intervention on two key outcomes, breastfeeding intentions prenatally and infant feeding behaviour post-natally.

Table 1 Socio-economic (1991 Census) characteristics of intervention and control areas

Permanently sick Social class IV and V Male unemployment Local authority housing No car

% in intervention area

% in control area

27 38 41 80 83

23 37 34 81 76

Source: Greater Glasgow Health Board, Health Information Unit.

Materials and methods

Drumchapel and who were resident in the control area at the time of booking. There were no exclusions at this stage. Thereafter, women who aborted or miscarried, or whose babies were stillborn or died post-natally were included up to the point at which they could no longer participate. Subjects were excluded if they moved from the intervention to the control area after having been initially in the intervention group, as were subjects who moved post-natally in the opposite direction. Subjects were also excluded who moved out of either area and would deliver in neither of the maternity hospitals. In total, 995 women were recruited to the study, 474 from the intervention area and 521 from the control area. (Sample size calculations had indicated that at least 389 women would be required in each study group, assuming a required study power of 80 per cent, a significance level of 0.05, and breastfeeding prevalence at 6 weeks post-natally of 12 per cent in the intervention group and 6 per cent in the control group).

The study areas

The intervention

The study was designed as a quasi-experimental communitybased controlled trial in which the intervention would be delivered to a selected target community (Easterhouse) and withheld from a comparable control community (Drumchapel). These communities were located in the northeast and northwest of Glasgow respectively and were geographically separated by several miles. The designation of intervention or control status was arbitrary. Both were known to have a very low incidence of breastfeeding, were culturally and socially similar, and were served by separate maternity hospitals. There were in fact some minor socio-demographic and health differences between the two study areas (Table 1) with the intervention area being slightly more disadvantaged than the control community.

The intervention was conceived as a breastfeeding promotion programme comprising personal peer counselling of pregnant women, post-natal support and local awareness raising activities over a period of 2 years. The detailed content of the programme was developed in 1993–1994 following a review of the literature and a concurrent series of discussions with professionals and members of the public. The latter comprised a small number of pregnant women in the selected target area who were approached informally in the antenatal booking clinic to obtain an impression of their views on infant feeding and their likely receptivity to a peer support programme. The response was generally positive. Thereafter, contact was established with seven local mothers who fulfilled the following criteria: they lived within the intervention area, they had at least one child under 5 years of age, they had breastfed a baby for at least 3 months, and they were interested in helping other mothers to breastfeed. These women then attended a training programme specially designed to suit the needs of the intervention and the trainees. It aimed to provide them with sufficient information and skills to enable them to promote breastfeeding and to support breastfeeding mothers. After completing the training, the women, who became known as

The study population Having obtained their informed consent, study subjects were recruited by the project clerical officer at antenatal booking clinics over the study period (November 1994–October 1996). The intervention group comprised all women attending the clinic in Easterhouse (or the local maternity hospital) and who were residents of the intervention area at the time of booking. The control group comprised all women attending the clinic in

140

JOURNAL OF PUBL IC HEALTH MEDICINE

‘helpers’, were the collective vehicle for delivering the intervention. This consisted of an offer to intervention group women in early pregnancy of a minimum of four contacts – two antenatally and two post-natally – to discuss infant feeding options, to emphasize the benefits of breastfeeding, and to provide specific, practical and individually tailored information about and support for breastfeeding to those mothers requesting it. The first antenatal visit was offered in the second trimester and the second in the third trimester; these visits were designed to help the prospective mother make an informed choice about feeding options. Visits were offered to all the subjects in the intervention group (474) regardless of feeding intention. A total of 471 (99.4 per cent) subjects accepted and 334 (70.5 per cent) actually received at least one antenatal visit by a helper. Post-natal visits were offered to all mothers who initiated breastfeeding at birth (n = 105). These mothers were contacted as soon after delivery as possible to enable the helpers determine the degree of postnatal support required. The extent and timing of post-natal support for each subject was flexible to enable the helpers to respond to the individual needs of each mother. Of the breastfeeding initiators, 80 (76.2 per cent) actually received at least one post-natal visit and an average of four visits each (range of 1–15). The helpers also visited 47 mothers who did not initiate breastfeeding despite declaring an antenatal intention to do so, in an attempt to determine why breastfeeding had not been initiated. All intervention group mothers were encouraged to contact the helpers if they wanted extra support. Leaflets containing breastfeeding information and the helper’s contact telephone number were either given directly to mothers or their relatives, or posted to subjects declining visits. All seven helpers stayed with the project for the first 2 years. Thereafter, competing demands on their time eventually reduced this number to four. The impact of any differences in commitment between the helpers appeared minimal, as an analysis of study subjects grouped according to their helpers revealed no significant variation in the prevalence or duration of breastfeeding between helpers.

Evaluation Data were collected from all the study subjects from booking to 6 weeks post-natally (Fig. 1) by means of four self-completion questionnaires, two administered antenatally and two postnatally. The face validity, acceptability and reliability of the questionnaires had been assessed during piloting in the planning stage of the study and the findings incorporated into the design of the final versions. The antenatal questionnaires were completed by the study subjects (with the help of the project clerical officer) in the clinics at booking, usually 8–12 weeks gestation (questionnaire I), and at the return visit around 28–32 weeks (questionnaire II). These were concerned with feeding intention, influences on choice of feeding method and previous feeding behaviour. The post-natal questionnaires were

Figure 1 Flow of study participants from recruitment to 6 weeks post-partum.

designed to obtain data on actual feeding behaviour after delivery. The first one was distributed by midwives in the postnatal wards and completed around the time of discharge, usually within the first 48 hours (questionnaire III). The second was distributed by health visitors at home at around 6 weeks post-natally (questionnaire IV). The professionals who distributed the questionnaires assisted in their completion and validated the responses on feeding behaviour. Case notes were also examined to obtain supplementary information when questionnaire responses were incomplete.

Analysis To compare outcomes in the intervention and control areas, the feeding categories breast, bottle, breast plus bottle and any breastmilk were examined. For the primary analyses, the breast and breast and bottle categories were combined into a breast category, and bottle, undecided and other categories were combined into a non-breast category. This enabled us to compare over time the proportions in the two groups intending to provide or providing any breast milk for their babies. Univariate analyses were carried out at each assessment using x2 tests, and confidence intervals for the ratios of the proportions breastfeeding in the two areas. Supplementary multivariate logistic regression analyses were carried out to enable us to assess the magnitude of difference in feeding practice at delivery and at 6 weeks post-natally, adjusting for a number of confounding factors. These confounding factors are listed in Table 2. Factors included were those that had been previously shown to be associated with feeding outcome. One

141

BREASTFE EDING PROMOTION

Table 2 Factors included in the multivariate analyses for breastfeeding initiation, and practice at 6 weeks Factor Breastfeeding intention at booking visit Deprivation category Caesarean section Admission of the baby to special care Previous breastfeeding experience Attendance at breastfeeding workshops Attendance at antenatal classes Receiving milk tokens Regular smoker in the last 12 months Living with partner Gestation at booking Premature delivery Low birthweight (< 2.5 kg) Primigravid Age at booking visit Study area

of the key confounding factors was socio-economic status – a minority (13 per cent) of control subjects were residing in postcode sectors other than the most deprived (Depcat 7, as described by Carstairs and Morris14), whereas all intervention subjects resided in Depcat 7 areas. To evaluate changes in intention or practice relative to the previous visit, the ‘net gain’ in the proportion intending to breastfeed or breastfeeding was calculated for each group as the proportion who changed from not intending to breastfeed at the earlier visit to intending to breastfeed or breastfeeding at the later visit, discounting the proportion whose intention or behaviour changed in the opposite direction. These sequential net changes in the intervention and control groups were compared using 95 per cent confidence intervals for the differences in net gains.

Results The data are presented on an ‘intention to treat’ basis. Of the 995 women enrolled in the study at booking, data were available for analysis on 919 (92.4 per cent) at 6 weeks postnatally although not all of these completed the intermediate

Breast/Non-breast 4 or 7 for our study areas Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No #16 weeks/ >16 weeks Yes/No Yes/No Yes/No Intervention/Control

questionnaires (Fig. 1). The response rates to each of the four questionnaires were consistently high and comparable between the two groups (Table 3). The attrition of the study population was due mainly to study subjects being excluded because of migration or adverse reproductive outcomes rather than to non-response. Table 4 shows that the intervention and control groups were similar with respect to a number of key socio-demographic variables at the time of booking. Table 5 summarizes the responses to enquiries about feeding intention in questionnaires I (booking) and II (return), feeding behaviour at delivery, and in questionnaires III (discharge) and IV (6 weeks post-natally). At booking, a significantly greater proportion of those in the intervention group were unsure of their feeding intentions compared with the control group (23.2 versus 10.9 per cent, respectively). A lower proportion of intervention than control subjects at booking stated an intention to bottle feed. However, of those who had formulated their feeding intentions at this stage, the proportions intending to breastfeed were almost identical (86/364, or 23.6 per cent, in the intervention group versus 111/464, or 23.9 per cent, in the control group). By the return visit, there was an increase in the intervention

Table 3 Response to the four questionnaires (values are numbers, with percentages given in parentheses) Intervention group

Control group

..........................................

........................................

Questionnaire

Eligible

Response

Eligible

Response

First: booking Second: return Third: delivery Fourth: 6 weeks

474 (100) 456 (100) 449 (100) 449 (100)

474 (100) 399 (87.5) 449 (100) 447 (99.6)

521 (100) 484 (100) 477 (100) 477 (100)

521 (100) 430 (88.8) 477 (100) 472 (98.9)

142

JOURNAL OF PUBL IC HEALTH MEDICINE

Table 4 Comparison of frequency of selected personal and social characteristics of intervention and control groups at booking

Mean age (years) at booking (range) Number primiparous* Mean number of children per household Number with previous breastfeeding experience* Number smoking regularly* Number receiving milk tokens*† Number living with partner*

Intervention group (n = 474)

Control group (n = 521)

25 (15–39) 206 (44) 1.7 47 (10) 293 (62) 188 (40) 304 (64)

25 (15–41) 211 (41) 1.8 64 (12) 326 (63) 214 (41) 326 (63)

*Values in parentheses are percentages. †Eligibility depends on whether or not receiving income support.

group – but not in the control group – in the proportion intending to breastfeed. The proportions who intended to bottle feed and who were undecided were similar at the return visit. After delivery, the proportions breastfeeding exclusively declined rapidly in both groups until 6 weeks post-natally when only a small minority of the intervention group (8.3 per cent) and the control group (5.3 per cent) were exclusively breastfeeding. Table 6 shows the proportions in each group intending to provide or providing any breast milk for their babies. Although the proportions were similar in the two groups, there was a discernible tendency for a progressively higher proportion of the intervention group to intend to provide or to provide breast milk compared with the control group, as reflected by the rising prevalence ratio (0.85 at booking to 1.26 at 6 weeks). By 6 weeks post-natally, the proportion providing any breast milk had declined to 9.6 per cent in the intervention group and 7.6 per cent in the control group.

None of the univariate analyses of breastfeeding prevalence detected any statistically significant differences between the two study areas. However, there were a number of potential confounding factors. In particular, the areas differed with respect to socio-economic status. Multivariate analyses correcting for the various confounding factors were conducted with respect to two key breastfeeding outcomes: breastfeeding initiation, and breastfeeding practice at 6 weeks post-natally. These analyses indicated that, after adjusting for the confounding factors, intervention subjects were significantly more likely to initiate breastfeeding at delivery when compared with the controls: intervention group versus control group odds ratio (OR) 1.95, 95 per cent confidence interval (CI) 1.22–3.14, p = 0.006. By 6 weeks, however, the difference between the intervention and the control group was less and did not quite reach statistical significance: OR 1.80, 95 per cent CI 0.96– 3.41, p = 0.07.

Table 5 Responses of study subjects to questions about feeding intention (booking and return visits) and behaviour (postnatally) (numbers, with percentages given in parentheses) Feeding category

Booking

Breast exclusively Intervention group Control group

86 (18.1) 111 (21.3)

Return

94 (23.6) 91 (21.2)

Delivery

105 (23.4) 94 (19.7)

Breast þ bottle Intervention group Control group Bottle exclusively Intervention group Control group

278 (58.6) 353 (67.8)

276 (69.2) 305 (70.9)

Undecided/other Intervention group Control group

110 (23.2) 57 (10.9)

29 (7.3) 34 (7.9)

Total Intervention group Control group

474 (100) 521 (100)

399 (100) 430 (100)

344 (76.6) 383 (80.3)

449 (100) 477 (100)

Discharge

6 weeks

66 (14.7) 65 (13.6)

37 (8.3) 25 (5.3)

16 (3.6) 4 (0.8)

6 (1.3) 11 (2.3)

365 (81.3) 408 (85.5)

401 (89.7) 436 (92.4)

2 (0.4) 0 (0)

3 (0.7) 0 (0)

449 (100) 477 (100)

447 (100) 472 (100)

At booking and return, those intending to breastfeed are shown in the exclusive breastfeeding category. At delivery, those who initiated breastfeeding are shown in the exclusive breastfeeding category.

143

BREASTFE EDING PROMOTION

Table 6 Summary of proportions intending to provide or providing any breastmilk Booking

Return

Delivery

Discharge

6 weeks

86/474 (18.1) 111/521 (21.3)

94/399 (23.6) 91/430 (21.2)

105/449 (23.4) 94/477 (19.7)

82/449 (18.3) 69/477 (14.5)

43/447 (9.6) 36/472 (7.6)

0.85 (0.66, 1.10) 0.21

1.11 (0.86, 1.44) 0.41

1.19 (0.93, 1.52) 0.17

1.26 (0.94, 1.69) 0.12

1.26 (0.83, 1.93) 0.28

Subjects providing any breastmilk Intervention group* Control group*

Prevalence ratio† Point estimate CI p value

At booking and return, those intending to breastfeed are shown in the breastfeeding category. At delivery, those who initiated breastfeeding are shown in the breastfeeding category. *Values in parentheses are percentages. †Calculated by dividing the percentage in the intervention group by the percentage in the control group.

Table 7 follows the progress of the two groups from booking to 6 weeks post-natally within feeding intention and practice categories. In the intervention group, there was a net gain of 5.8 per cent in the proportion intending to breastfeed relative to booking, whereas in the control group there was a net gain of only 0.7 per cent. The difference between these net changes in the two groups was statistically significant ( p = 0.02). Between the return visit and the subsequent contacts, the net changes were negative (increasingly so after delivery) but with no large differences between the intervention and control groups.

Discussion The view that infant feeding attitudes and behaviour can be influenced by community-based peer support is widely held.15 Voluntary organizations such as the National Childbirth Trust and the La Leche League run a number of such programmes but no formal attempts to evaluate their impact on breastfeeding incidence and duration have been reported. To our knowledge, our study is the first controlled experiment to evaluate community-based breastfeeding promotion in a socially disadvantaged European population. Waterston16 has drawn attention to the inherent difficulties involved in mounting a randomized controlled trial (RCT) to evaluate community-based breastfeeding promotion efforts.

We considered designing the trial as an individualized RCT but rejected this option for two main reasons. First, we hypothesized that a collective, communal dimension is a necessary feature of a successful intervention as well as an individualistic one. Second, the avoidance of contamination of controls by intervention subjects would have been impossible in an individualized trial conducted within a single locality. In the event, the lay volunteers did indeed develop a collective identity that was strongly rooted in the local community, as reflected by their renting of office premises, the design of a logo for printing on T-shirts, and the holding of regular breastfeeding promotion events. There was no evidence that these activities distracted the helpers’ attention from their peer support role. A notable feature of our study was the extremely low attrition rate of the study population. The validity of the responses is difficult to assess but was optimized by careful piloting. Reporting bias of feeding intention at the booking clinic may explain the higher proportion of women in the intervention group who were undecided compared with the control group, as well as the lower proportion in the intervention group who stated an intention to bottle feed, because by the return visit both those differences had virtually disappeared. There was a consistent tendency in favour of breastfeeding in the intervention group relative to the control group.

Table 7 Sequential changes in proportions intending to provide or providing any breastmilk (values are per cent change)

Feeding category

Booking to return

Return to delivery

Delivery to discharge

Discharge to 6 weeks

Intervention group

þ5.8

¹1.0

¹5.1

¹8.3

Control group

þ0.7

¹1.9

¹5.2

¹5.9

þ0.1 (¹2.8, 3.0) 0.95

¹2.3 (¹5.7,1.0) 0.17

Difference between the sequential changes for the two groups Point estimate þ5.1 þ0.9 CI (0.8, 9.4) (¹3.0, 4.8) p value 0.02 0.66

144

JOURNAL OF PUBL IC HEALTH MEDICINE

Antenatally, a significantly greater net gain in intention to breastfeed between booking and return was seen in the intervention group, although after this point, no clear differences in the net gains were observed. It seems therefore that the temporal effect of the intervention was modest and was largely confined to the first study period (i.e. during pregnancy). However, multivariate analyses revealed a significantly higher frequency of initiation of breastfeeding in the intervention group than in the control group but that the impact did not extend to 6 weeks. There are a number of possible reasons why the intervention failed to show a sustained impact on feeding behaviour. First, only 70 per cent of intervention subjects received at least one antenatal visit from a helper, thereby diluting the impact of the intervention. The explanation for this disappointing uptake is unclear but it needs to be addressed in any future similar exercise. Second, the strong influence of social support networks (including partners and relatives) has been described by a number of researchers.17–19 The intervention may have been inadequate in addressing the influences of these networks. Third, the discretionary nature of post-natal support, whereby the frequency and content of the support was defined by the helper in negotiation with the subjects, may have undermined its effectiveness. A more structured intervention may be preferable to overcome the reluctance of many mothers to request support spontaneously.20–23 Finally, breastfeeding in a non-breastfeeding culture may be socially unacceptable and therefore likely to undermine maternal confidence, a factor known to be associated with reduced duration of breastfeeding.24–26 In summary, we have reported the findings of a controlled trial of peer support for breastfeeding in a socially disadvantaged urban population in an area of Scotland known to have an extremely low prevalence of breastfeeding. Although crude cross-sectional analyses suggested that the impact of the intervention on breastfeeding prevalence was modest, there was a consistent tendency in favour of breastfeeding and a significant net gain in the proportion of women stating a (predelivery) intention to breastfeed in the intervention group. These positive findings in favour of breastfeeding support were strengthened when adjustment was made for possible confounders, including socio-economic status. By 6 weeks postnatally, however, no significant difference in breastfeeding prevalence was observed whatever method of analysis was used. In other words, although we found some evidence of an impact on the breastfeeding intention and behaviour, this was not sustained even for the modest duration of 6 weeks postnatally. These findings suggest that it would be premature to justify the widespread use of peer support programmes to increase the prevalence of breastfeeding in socially disadvantaged communities. We believe, however, that our results are sufficiently encouraging to support further work to develop and refine the peer support approach. Finally, the finding that only one in five of the study subjects stated an intention to breastfeed suggests that a major shift in

attitudes to infant feeding will have to occur in this population before pregnancy if breastfeeding incidence is to increase substantially in the future. That under 10 per cent of the subjects surveyed at 6 weeks were breastfeeding – around a fifth of the government target for 2005 in Scotland27 – indicates the enormous scale of the challenge facing public health professionals.

Acknowledgements The study was performed with the financial support of the Health Promotion Department, Greater Glasgow Health Board. We are grateful to the following for their advice and assistance: Drs P. Hanlon, C. Tannahill and their colleagues at the Greater Glasgow Health Board; Professor F. Cockburn, Professor L. Weaver and Dr T. Turner of the University of Glasgow’s Department of Child Health; Mrs P. Maxwell, Mrs J. Warren, Mrs J. Britten and Dr D. Tappin of the PEACH Unit; Professor C. Dungy of the University of Iowa’s Department of Paediatrics; the staff of the Easterhouse and Drumchapel health centres; the staff of the Queen Mother’s and Glasgow Royal Maternity Hospitals; and especially to the mothers and helpers, without whom the study would not have been possible.

References 1 WHO–UNICEF. Protecting, promoting and supporting breastfeeding: the special role of maternity services. Geneva: WHO, 1989. 2 Institute of Medicine, Subcommittee on nutrition during lactation. Nutrition during lactation. Washington, DC: National Academy Press, 1991. 3 Standing Committee on Nutrition of the British Paediatric Association. Is breastfeeding beneficial in the UK? Arch Dis Child 1994; 71: 376– 380. 4 Department of Health. Weaning and the weaning diet. Report of the Working Group on the Weaning Diet of the Committee of Medical Aspects of Food Policy. London: HMSO, 1994. 5 American Academy of Pediatrics. Work Group on Breastfeeding. Breastfeeding and the use of human milk. Pediatrics 1997; 100(6): 1035–1039. 6 Akre J. Infant feeding: the physiological basis. Bull WHO 1989; 67(Suppl.): 1–108. 7 White A, Freeth S, O’Brien M. Infant feeding 1990. London: OPCS, 1992. 8 Greater Glasgow Health Board. Annual report of the Director of Public Health 1994/95. Glasgow: Greater Glasgow Health Board, 1996. 9 Foster K, Lader D, Cheesbrough S. Infant feeding 1995. London: The Stationery Office, 1997. 10 Dix DN. Why women decide not to breastfeed. Birth 1991; 18: 222– 229. 11 Long DG, Funk-Archuleta MA, Geiger CJ, Mozar AJ, Heins JN. Peer counsellor program increases breastfeeding rates in Utah Native American WIC population. J Hum Lact 1995; 11: 279–284. 12 Morrow AL, Guerrero ML, Shults J, et al. Efficacy of home-based peer counselling to promote exclusive breastfeeding: a randomised controlled trial. Lancet 1999; 353: 1226–1231.

BREASTFE EDING PROMOTION

145

13 Schafer E, Vogel MK, Viegas S, Hausfus C. Volunteer peer counsellors increase breastfeeding duration among rural low-income women. Birth 1998; 25(2): 101–106.

21 Bloom K, Goldbloom RB, Robinson S, Stevens FE. Factors affecting the continuance of breastfeeding. Acta Paediatr Scand 1982; Suppl. 300: 9–14.

14 Carstairs V, Morris R. Deprivation and health in Scotland. Aberdeen: Aberdeen University Press, 1991.

22 Neyzi O, Gulecyuz M, Dincer Z, et al. An education intervention on promotion of breastfeeding complemented by continuing support. Paediat Perinat Epidemiol 1991; 5(3): 299–303.

15 Svenson GR, Hanson BS. Are peer and social influences important components to include in HIV–STD prevention models? Eur J Publ Hlth 1996; 6: 203–211. 16 Waterston T. Improving the duration of breastfeeding. Arch Dis Child 1997; 76: 503–504.

23 Lynch SA, Koch AM, Hislop TG, Coldman AJ. Evaluating the effect of a breastfeeding consultant on the duration of breastfeeding. Can J Publ Hlth 1986; 77(3): 190–195.

17 Bryant CA. The impact of kin, friend and neighbour networks on infant feeding practices. Social Sci Med 1982; 16: 1757–1765.

24 Loughlin HH, Clap-Channing NE, Gehlbach SH, Pollard JC, McCutchen TM. Early termination of breastfeeding: identifying those at risk. Pediatrics 1985; 75: 508–513.

18 Baranowski T, Bee DE, Rassin DK, et al. Social support, social influence, ethnicity and the breastfeeding decision. Social Sci Med 1983; 17(21): 1599–1611.

25 Buxton KE, Gielen AC, Faden RR. Women intending to breastfeed: predictors of early infant feeding experiences. Am J Prev Med 1991; 7(2): 101–106.

19 Bryant CA, Coreil J, D’Angelo SL, Bailey DF, Lazarov M. A strategy for promoting breastfeeding among economically disadvantaged women and adolescents. NAACOGS Clin Iss Perinat Womens Hlth Nurs 1992; 3(4): 723–730.

26 O’Campo P, Fader RR, Gielen AC, Wang MC. Prenatal factors associated with breastfeeding duration: recommendations for prenatal intervention. Birth 1992; 19(4): 195–201.

20 Houston MJ, Howie PW, Cook A, McNeilly AS. Do breastfeeding mothers get the home support they need? Health Bull (Edinburgh) 1981; 39: 166–172.

27 Management Executive. Local breast feeding targets. MEL(1994)110. Edinburgh: Scottish Office, 1994.

Accepted on 14 September 1999