The Ife South Breastfeeding Project: training community health extension workers to promote and manage breastfeeding in rural communities.

The Ife South Breastfeeding Project: training community health extension workers to promote and manage breastfeeding in rural communities A.A. Davies-Adetugbo1 & H.A. Adebawa2 Reported are the results of a project to promote exclusive breasifeeding in rural communities through the training of community health extension workers in rural Nigeria. A workshop for the trainers was organized for health workers in the study area; subsequently, these trainers ran district-level training workshops. In the study area perinatal facilities, early initiation of breasifeeding has increased compared with those in the control area (P < 0.001). Also, the trained health workers had significantly better knowledge about breastfeeding than their untrained colleagues in both the study (P < 0.00 1) and control areas (P < 0.001), and more often recommended timely initiation and exclusive breasifeeding than the controls (P < 0.001). A multivariate analysis showed that the training programme and the study area were the only significant variables that were predictors of breasifeeding knowledge (P < 0.001). Appropriate education of health extension workers can therefore contribute significantly to the promotion of breasifeeding in rural communities. Introduction The many nutritional, immunological, and contra- ceptive advantages of breast milk and breastfeeding have been well documented (1, 2). Breastfeeding is also being promoted for the prevention of diarrhoeal diseases (3), which claim the lives of over 300000 Nigerian children annually (4, 5). Although almost all Nigerian women breast- feed, there has been a decline in the duration of breastfeeding and less than 2% practice exclusive breastfeeding for the first 4 months (6). Also, initia- tion of breastfeeding is often delayed and many cul- tural perceptions and factors militate against effective breastfeeding practices in rural communi- ties (7). There is a need to promote, protect, and support breastfeeding as a child survival strategy (8) in rural communities in Nigeria. Reported here are results that were obtained as part of a larger study to pro- mote breastfeeding. This article describes the pro- motion programme, which included community education and the training of community health ex- ' Lecturer, Department of Community Health, College of Health Sciences, Obafemi Awolowo University, lie Ife, Nigeria. Requests for reprints should be sent to this author at the following address: 39 Invicta Grove, Northolt, Middlesex UB5 6RR, England. 2 Community Health Officer, Ife South Local Government, Ifetedo, Osun State, Nigeria. Reprint No. 5784 tension workers in breastfeeding and lactation man- agement, and evaluates its impact on breastfeeding knowledge and practice. Training of health workers makes an important contribution to breastfeeding promotion (8-10); however, while there is ample evi- dence that such training can improve hospital prac- tices, few reports have appeared on the effects of training extension health workers at the primary health care level in underserved rural communities. Methods Location of study The study was conducted in Ife South local govern- ment (ISLG) area, Osun State, Nigeria, as part of the Ife South Breastfeeding Project (ISBFP), a community-based promotion programme. ISLG is a predominantly rural area with an estimated popula- tion of 88900 persons who live in about 450 villages and smaller settlements, most of which are Yoruba farming communities. The largest of the villages are Ifetedo, the administrative headquarters of the local government, and Olode, each with a population of about 3000. Sanitation is generally poor. Only these two villages have electricity and piped water. Water for domestic use is drawn from shallow wells and streams. The study area consists of ten health districts, with a total of 30 health facilities: rural health cen- Bulletin of the World Health Organization, 1997, 75 (4): 323-332 © World Health Organization 1997 323 A.A. Davies-Adetugbo & H.A. Adebawa tres, maternity centres, and health posts. In addition, there is a primary health care (PHC) centre at Ifetedo. These health facilities are staffed by mid- level and lower primary health care workers: com- munity health extension workers, traditional birth attendants, and nurses. Village volunteer health workers are also active at the community level. There is one doctor at the primary health care facil- ity at Ifetedo, but none at any of the local govern- ment health facilities. Perinatal care is offered mainly by the PHC workers and traditional birth attendants. There are no secondary or tertiary health care centres in the study area, and Obafemi Awolowo University Teaching Hospital (OAUTH), where there is a baby-friendly hospital initiative pro- gramme, is 45km from it. Many of the health districts have health com- mittees. Ife South health department, including its 30 health facilities, has 103 health workers, eight of whom are trained nurses; 18 of the PHC workers were absent on various courses during the training component of the breastfeeding project. Atakumosa local government (ATLG) area served as the control area; no training activities were conducted there, and it is contiguous with the study area and similar to it. The control area has a popula- tion of 99200 living in about 200 small villages, with the predominant occupations being farming and petty trading; its health department has 86 PHC workers in 14 health districts. Focus group discussions Focus group discussions, using a discussion guide on breastfeeding issues, were conducted among a group of 8-10 PHC workers in each of the study and con- trol areas. The discussions were supplemented with in-depth interviews of two PHC workers in each area. These interviews were carried out prior to the training activities, and the results obtained contrib- uted to planning the training activities. Training workshops and programmes A one-day community mobilization workshop was held at Olode that was attended by traditional lead- ers, local government officials, community health workers, and mothers. The workshop was designed as a 6-hour introductory course on breastfeeding and child survival. It was conducted in Yoruba, the local language, and the following topics were included: anatomy and physiology of the breast; exclusive breastfeeding; expressed breast milk; lactation main- tenance; and practical demonstrations of attach- ment, suckling, expression of breast milk, and cup feeding. An intensive 2-day course for the workshop trainers was held in Ifetedo for health workers from each of the 10 health districts. Other participants included local government officials. The course con- sisted of 12 hours of instruction (including 3 hours of clinical practice). Its content had been adapted from the WHO/UNICEF 18-hour breastfeeding course manual (11), appropriately modified to suit nonhospital-based primary care and rural settings, and included lectures, demonstrations, practicals, as well as role-playing and case studies. The resource persons were master trainers from the OAUTH baby-friendly hospitals initiative programme. The workshop was also conducted in Yoruba. A total of 30 PHC workers from ISLG underwent training at this workshop. Training workshops were also held after the trainers' course in each of the 10 health districts. Prior to these, meetings were held in all the districts with health workers, nursing mothers, and com- munity leaders, at which the advantages of breast- feeding and its importance for child survival were discussed. Each health district was encouraged to form community breastfeeding support groups. These meetings also planned the district level train- ing workshops, whose participants, apart from PHC workers, included mothers, husbands, and commu- nity leaders. The district-level workshops consisted of 6 hours of instructions and 2 hours of demon- strations and practicals (scaled-down versions of the trainers' course); they were also conducted in Yoru- ba and the resource persons were PHC workers who had attended the trainers' course. A further 36 PHC workers and 569 community members received training at these district-level workshops. Study samples A pre-tested, self-administered questionnaire focus- ing on key issues - when breastfeeding is initiated, prelacteal feeding, exclusive breastfeeding, colos- trum, and the advantages of breastfeeding - was sent to all eligible PHC workers in the study area, and to the 56 PHC workers in 7 of the 14 health districts in the control area selected at random. The questionnaire study was conducted 4 months after the last of the training workshops. Sample size was calculated assuming that 40% of untrained health workers had acceptable breastfeeding knowl- edge scores, and that training would double this proportion. To achieve a power of detection of 95% and a 5% significance level, 42 respondents were needed from each of the control and study areas (12). Another questionnaire study was also con- ducted among the women who had been recruited WHO Bulletin OMS. Vol 75 1997324 The Ife South Breasffeeding Project from three health districts in each of the areas, chosen by convenience, into the ISBFP in their last trimester of pregnancy (7). This post-delivery ques- tionnaire included assessments of who delivered them and where, when breastfeeding was initiated, the use of prelacteal feeds, and other breastfeeding issues. Of the 126 women in the intervention area and the 130 in the control area, 98 and 108 women, respectively, completed the questionnaire. Data analysis Statistical analysis was performed by calculating means, standard error of the mean, 95% confidence intervals, x2 tests, and stepwise, multiple linear re- gression using EPI-Info software (13). Multiple logistic regression analyses were performed using LOGISTIC version 3.03e software (14). The questionnaire variables were defined as fol- lows: bfinit, knowledge of early initiation (within 30min of delivery) of breastfeeding; aprelact, knowl- edge about not using prelacteal feeds; bffull, knowl- edge that the breastfeeding baby does not need supplementary water; bfxclusv, response that exclu- sive breastfeeding, without supplementary water, should be practised for the first 4-6 months; hwcolstr, advchild, advmothr, ability to provide at least two valid advantages of feeding colostrum, breastfeeding to the infant, and breastfeeding to the mother, respectively. Results The age, sex, and marital status of the PHC workers are shown in Table 1. Table 1: Selected sociodemographic characteristics of primary health care workers in the study (ISLG) and control areas (ATLG) ISLG area ISBFP Characteristic trained Others Total ATLG area Age group <20 years 3 1 4 1 20-29 years 18 4 22 25 30-39 years 22 7 29 21 ¢40 years 23 2 25 4 Total 66 14 80 52 Males 17 5 22 2 Marital status Single 12 3 15 15 Married 51 11 62 35 Separated/widowed 1 - 1 1 Unspecified - - - 1 Focus group discussions The results of pre-training focus group discussions among groups of PHC workers in the control and study areas were the same. The health workers were aware that mothers in their communities did not feed colostrum to their babies because they perceived it as dirty ("like pus") and capable of transmitting dis- ease; however, they encouraged mothers to feed co- lostrum because it is good, "makes the baby grow and develop immunity", and "deepens love between mother and child". The health workers also held that for mothers "who lactate immediately", breast- feeding should commence 6 hours after birth or at the latest within the first day following delivery. The delay was stated to be necessary since "babies are unable to suck immediately after birth because they are tired". Mothers who were not able to lactate immediately, could take up to 2 days to commence breastfeeding. The opinions of the health workers on pre- lacteal feeds were divided. Most of them advocated giving clean water; some thought that the baby could wait for breast milk; most agreed, however, that it was not possible to breastfeed without giving supple- mentary water, because "the baby will become thirsty". Some of the health workers also stated that it was necessary to give other supplements because "today's babies cannot be fully satisfied with breast milk". According to this view, babies should there- fore start receiving supplements at 1-3 months of age; moreover, the health workers agreed that wean- ing should commence when the baby is 3-4 months of age. Thus, before the training programme, health workers in both the control and study areas did not seem to appreciate the importance of early initiation of or of exclusive breastfeeding. Training of health workers As a result of the preparatory meetings, district level leaders agreed to promote the formation of breast- feeding support groups in their communities. Where there were no functioning community development committees, these were put in place and it was agreed that they would promote breastfeeding. The PHC workers had appreciated that the training workshops were given in the vernacular and that they had gained new knowledge and learned new skills that they planned to employ in their work. This new knowledge included the importance of the early initiation of breastfeeding and of exclusive breastfeeding for the first 6 months of life, the benefits of colostrum, and the advantages of breast- feeding in reducing morbidity and mortality from WHO Bulletin OMS. Vol 75 1997 325 A.A. Davies-Adetugbo & H.A. Adebawa diarrhoeal and other diseases. The new skills in- cluded expression of breast milk, correct positioning and attachment of the baby to the breast, relactation, and counselling. The advantages of breastfeeding for the mother and the role of men in promoting and supporting breastfeeding were also new for many of the workers. A video recording of the trainers' workshop has been adopted for use by the teaching hospital's baby friendly hospitals' initiative for training PHC workers in other parts of Osun State. Early initiation of breasffeeding in primary perinatal facilities Table 2 shows the use of perinatal facilities by moth- ers in the study and control areas. About two-thirds of the deliveries occurred at the PHC facilities, while a fifth occurred at home, usually unassisted by any trained health worker. The proportions delivered in PHC facilities were not significantly different in the two study areas. In the intervention area, after training, 32% of the deliveries reported early initiation of breast- feeding (within 30min of delivery) compared with only 6% in the control area (Table 2). This differ- ence of 26 percentage points (95% confidence interval (CI): 16-36%) is statistically significant (X2 test = 21.97, P < 0.001) and is accounted for almost entirely by the practices in the PHC facilities; all the 31 early initiators in ISLG (the study area) were from the 66 deliveries in the PHC facilities (47%), Table 2: Use of primary health care (PHC) perinatal facilities in the study (ISLG) and control (ATLG) areas %in %in ISLG area ATLG area Where delivered: PHC facility 67 (66)a P = 0.26 63 (68) Home 21 (21) 25 (27) Hospital 5 (5) 2 (2) Other 6 (6) 10 (11) Who delivered: PHC worker 71 (70) 65 (70) Traditional birth 11 (11) 2 (2) attendant Doctor 3 (3) 5 (5) No health worker 14 (14) 27 (29) Early breastfeeding initiation: Delivered at PHC 47 (31) P < 0.001 4 (3) facility Delivered at other 0 (0) P = 0.25b 8 (3) facilities All deliveries 32 (31) P < 0.001 6 (6) a Figures in parentheses are the numbers of women. b Fisher's exact test values. while only 3 of the 68 deliveries (4%) at PHC facili- ties in the control area initiated breastfeeding early. This difference of 43 percentage points (95% CI: 29- 55%) between the study and control PHC facilities is statistically significant (X2 test = 29.83, P < 0.001). For the other non-PHC facilities, the differences in early initiation in the two areas were not significant (Fisher's exact test, P = 0.25, Table 2). There was therefore a marked and significant increase in early initiation of breastfeeding by mothers who delivered at perinatal facilities staffed by ISBFP-trained PHC workers. PHC workers' knowledge about and attitudes towards breastfeeding The questionnaire study was used to assess further any differences in knowledge about and attitudes towards breastfeeding conferred by the training pro- gramme that might be related to the differences in practice between the two groups of PHC workers. A total of 80 of the 83 available PHC workers in ISLG (response rate, 96%) completed the questionnaire, including 66 of the 68 (97%) trained by ISBFP. In the control area, 52 of 56 PHC workers returned the questionnaire (response rate, 93%). The results show that virtually all the PHC workers in both areas advocated prolonged breast- feeding (18-24 months: study area, 93%; control area, 85%) and demand breastfeeding (98% and 90%, resp.), and that all recommended the use of colostrum. Table 3 shows the proportions of the various categories of PHC workers (ISBFP-trained, un- trained, study area, and control area) who gave cor- rect responses to seven indicator questions, and their aggregate knowledge scores. In all instances, trained PHC workers had better knowledge of and attitudes towards breastfeeding than untrained controls. For example, 89% of those who had been trained recom- mended early initiation of breastfeeding compared with 46% of controls. The difference in these pro- portions (43 percentage points; 95% Cl: 28-58%) is statistically significant (X2 test = 26.06, P < 0.001). In addition, 65 of 66 trained workers (98.5%) recom- mended exclusive breastfeeding (i.e., the baby re- ceives no other foods or fluids other than breast milk) for the first 4-6 months of life compared with 19 of 52 (36.5%) controls. The difference in these proportions (62 percentage points; 95% CI: 49-75%) is also highly significant (X2 test = 51.43, P < 0.001). Thus, the trained PHC workers, significantly more often than controls, recommended correct breast- feeding practices early initiation of breastfeeding, avoidance of prelacteal feeds, and exclusive breast- feeding for the first 4-6 months of life. WHO Bulletin OMS. Vol 75 1997326 The Ife South Breasffeeding Project ISBFP-trained PHC workers had an average knowledge score of 9.4 (Table 3), while those with- out training scored 6.1. The 3.3 score difference is highly significant (Kruskal-Wallis H test - 75.10, P < 0.001). Also, the difference of 1.8 points between ISBFP-trained and untrained health workers in the ISLG study area is significant (Kruskal-Wallis H test = 18.17, P < 0.001), as is the score difference of 3.5 between the study and control areas (Kruskal- Wallis H test = 69.64, P < 0.001). Therefore, the trained health workers acquired significantly better knowledge of breastfeeding than their untrained colleagues. A total of 45 of the total sample gave correct responses to all the breastfeeding items on the ques- tionnaire. These included 44 of the 66 ISBFP-trained PHC workers, one other PHC worker from the study area, and none from the control area. Within the study area, this pattern of correct answers was sig- nificantly associated with the ISBFP training (X2 test = 14.3, P < 0.001), while the difference between the ISBFP-trained health workers and those from the control area was also highly significant (X2 test = 52.47, P < 0.001). Thus, the trained workers had significantly better knowledge about breastfeeding than their untrained counterparts. Table 3: Proportion of PHC work breastfeeding variables Table 4 compares the odds ratios for the breast- feeding variables for the various categories of PHC workers. Although the sample size of the non- ISBFP-trained workers in the intervention area was relatively small, their results are also shown. The trained PHC workers made the correct recommen- dations on all aspects of breastfeeding significantly more often and had better breastfeeding knowledge for each aspect assessed than their untrained coun- terparts (Table 4, columns B and C). In addition, the knowledge and attitudes of the study area PHC workers were significantly better than those of the controls; and they made correct breastfeeding rec- ommendations significantly more often than un- trained workers in the control area (column D). Untrained PHC workers in the study area made cor- rect recommendations on exclusive breastfeeding significantly more often than those in the control area and had significantly better knowledge of colostrum and of the mother-centred advantages of breastfeeding (column F). Within the study area itself, ISBFP-trained workers made the correct recommendations on breastfeeding initiation and prelacteal feeding significantly more often than those without such training, and had significantly better knowledge of the advantages of breastfeeding kers who gave correct responses to Study area ISBFPa Untrained All Control area (n 66) (n 14) (n = 80) (n 52) Recommended'l Putting babies to 89 43 81 46 breast within 30 min of delivery (bfinit) Against prelacteal 97 79 94 89 feeding (aprelaco Against supplementary 91 86 90 58 water (bffull) Exclusive breasffeeding 99 93 98 37 for the first 4-6 months (bfxclusv) Knew. Two advantages of 82 57 78 12 colostrum (hwcolstr) Two advantages of 90 57 84 44 breastfeeding for the child (advchild) Two advantages of 91 64 86 33 breastfeeding for the mother (advmothr) Aggregate knowledge 9.4 7.6 9.1 5.6 scoreb (9.1_9.7)C (6.6-8.6) (8.8-9.4) (4.7-6.5) a Ife South Breasifeeding Project. b Out of 10. c Figures in parentheses are the 95% confidence interval. WHO Bulletin OMS. Vol 75 1997 327 A.A. Davies-Adetugbo & H.A. Adebawa Table 4: Maximum likelihood estimates for odds ratios for breastfeeding variables for various categories of ISBFP-trained and untrained PHC workers Odds ratios for: Study area, ISBFP ISBFP Study area trained Study area, untrained vs vs vs vs vs control area untrained control area untrained control area Variablesa (B) (C) (D) (E) (F) bfinit 9.6 (3.5-29.8)b 9.9 (3.8-29.7) 5.0 (2.2-12.0) 10.7 (2.5-51.2) 0.9 (0.2-3.4) P < 0.001 P < 0.001 P < 0.001 P < 0.001 P = 0.54 aprelact 4.1 (0.7-43.6) 5.0 (1.0-49.4) 2.0 (0.5-8.6) 8.4 (0.9-110.9) 0.5 (0.1-3.5) P = 0.073 P = 0.027 P= 0.224 P = 0.0353 P = 0.287 bffull 22.9 (5.1-213.1) 16.8 (3.8-154.4) 18.4 (5.0-102.8) 2.4 (0.0-49.9) 9.3 (1.2-423.2) P < 0.001 P < 0.001 P < 0.001 P = 0.443 P = 0.012 bfxclusv 107.8 (16.0-4660) 67.1 (10.3-2849) 64.9 (14.6-605.4) 4.9 (0.1-398.3) 21.6 (2.8-987.7) P< 0.001 P < 0.001 P< 0.001 P = 0.3212 P < 0.001 hwcolstr 32.9 (11.0-117.3) 16.2 (6.6-43.4) 25.5 (9.0-85.3) 3.3 (0.8-13.4) 9.7 (2.2-49.1) P < 0.001 P < 0.001 P < 0.001 P = 0.054 P = 0.001 advchild 10.4 (3.8-32.2) 9.3 (3.6-27.9) 6.4 (2.7-15.9) 6.1 (1.3-28.2) 1.7 (0.4-6.7) P < 0.001 P < 0.001 P < 0.001 P = 0.0085 P = 0.288 advmothr 19.9 (6.9-67.9) 15.0 (5.5-48.8) 12.6 (5.1-33.7) 5.4 (1.1-26.8) 3.6 (0.9-16.1) P < 0.001 P < 0.001 P < 0.001 P = 0.020 P = 0.034 a See Table 3 for an explanation of these codes. b Figures in parentheses are the 95% confidence interval. for the mother (column E). The training programme therefore seems to have conferred significantly better breastfeeding knowledge and attitudes on its recipients, enabling them to make correct breast- feeding recommendations more frequently. In the univariate analyses, knowledge about early initiation of breastfeeding (bfinit) and ad- equate breastfeeding knowledge (a questionnaire test score of at least 9 out of 10) were associated with work location (study area), ISBFP training, and sex, but not marital status, work experience, skill category, or age. Multivariable analysis by logistic regression (Table 5) shows that ISBFP train- ing was the only significant predictor of knowledge about the early initiation of breastfeeding (odds ratio = 13.23; P < 0.001), suggesting the effective- ness of the ISBFP training. Also, work location in the study area (but not in the control area) and the ISBFP training were the only significant predictors of adequate breastfeeding knowledge (Table 5). This also suggests that the training programme has had a significant effect on changing workers' knowledge and attitudes about breastfeeding, even those in the study area who did not receive training. Multivariate analysis of knowledge scores by stepwise multiple linear regression shows that loca- tion (the study area) and ISBFP training were the only variables that made a significant contribution to Table 5: Logistic regression models of knowledge about breastfeeding and of its early initiation Multivariate Model Parameters included Odds ratio P-value bfinita Study (versus control) area 0.81 (0.23-2.88)b 0.747 (n = 130) ISBFP training versus no ISBFP training 13.23 (3.38-51.77) 0.0001 Male versus female 1.26 (0.34-4.68) 0.728 knowledge Study (versus control) area 10.89 (2.39-49.64) 0.0020 (n = 130) ISBFP training versus no ISBFP training 6.33 (1.76-22.68) 0.0046 Male versus female 1.36 (0.38-4.95) 0.6382 a Recommendation of early initiation (within 30min of delivery) of breastfeeding. b Figures in parentheses are the 95% confidence interval. WHO Bulletin OMS. Vol 75 1997328 The Ife South Breastfeeding Project Table 6: Stepwise multiple regression model of breastfeeding knowledge scores" 95% Confidence Partial Variable 13 SE interval F-value P-value (Constant) 5.72 0.22 5.29-6.16 Study area 2.06 0.47 1.12-3.00 18.86 0.0000 ISBFP training 1.78 0.46 0.86-2.69 14.68 0.0002 Nursing profession -0.99 0.50 -1.91 to -0.01 3.97 0.0485 a R = 0.76; F-value = 56.33, df = 3, P < 0.001. the model (Table 6). The mean contribution of the ISBFP training to knowledge scores was 18% (range: 9-27%), and that of the study area was 20% (range: 11-30%), again indicating the effectiveness of the training. Being a nurse (an indicator variable) had a marginal effect: being a nurse/midwife seems to have reduced the knowledge score by about 10% (range: 0-19%). Other occupational status indica- tor variables (age, sex, marital status, and length of experience as a PHC worker) had no significant effect on the model. Discussion The following variables play a role in shaping mothers' decisions about whether to continue or terminate breastfeeding: maternal knowledge and attitudes; ethnic and cultural background; socioeco- nomic and employment status of the mother; other maternal factors; the availability of breast-milk sub- stitutes; urbanization; and the health services. The work we have reported in this article has focused on changing, through training, health workers' knowl- edge, attitudes and practices about breastfeeding within the PHC services in rural communities. Before the training programmes, many health workers and community leaders did not see the need to promote breastfeeding in a community where all mothers breastfeed. The workshops also showed that most of the health workers did not know how to position correctly and attach a baby to the breast. Community health workers in the study area are now promoting breastfeeding in all the health facili- ties in the local government area, and the local government authority has endorsed the promotion of breastfeeding. The training of health workers can play an important role in the promotion, protection and support of breastfeeding (8, 10). Ife South is a "breastfeeding community"; however, the commu- nity did not understand the importance of early ini- tiation of breastfeeding, the use of colostrum, or of exclusive and sustained breastfeeding. Also, the community needed to be reintroduced to the skills of lactation management, correct positioning and at- tachment, and the expression of breast milk. A total of 68 of the 103 (66%) PHC workers in the study area underwent the training we have described. The results we have reported suggest that the training enhanced the health workers' knowledge about and attitudes towards breastfeeding, and that these workers have had a positive impact on at least one aspect of breastfeeding behaviour in the com- munity: mothers' timely initiation of breastfeeding. Focus group discussions before the training showed no differences in the knowledge of the health work- ers in the two areas. Thus, it appears that mothers' breastfeeding practices have improved as a result of changes in health workers' recommendations and practices brought about by the training. ISLG health workers who did not have the training had significantly better knowledge about exclusive breastfeeding than controls, probably be- cause of transfer of knowledge from their trained colleagues; there were, however, only 14 workers in this category. The impact of health worker training on the practice of exclusive breastfeeding by the mothers should be assessed; however, we were not able to carry this out. Exclusive breastfeeding is a relatively new and difficult concept to promote in the study communi- ties. Initially most health workers, including doctors, thought that the breastfeeding project would not succeed, and half of the health workers in the control area still think it impossible for a breastfeeding baby not to have supplementary water. The reasons most often given are that the baby must be thirsty. These findings are not new (15); nevertheless, in a situation where household water supply is unsafe, such supplementation can lead to diarrhoea (16, 17), de- crease breast milk intake by the baby, and reduce milk production by the mother (15). Thus it is impor- tant to promote exclusive breastfeeding and this may be the most sensitive variable that discriminates between trained and untrained health workers (see Table 4). WHO Bulletin OMS. Vol 75 1997 329 A.A. Davies-Adetugbo & H.A. Adebawa Early initiation of breastfeeding is a good pre- dictor of exclusive (18) and prolonged breastfeeding (19). In our study, knowledge about early breast- feeding was most closely associated with the ISBFP training; moreover, improved worker's knowledge seems to have been synonymous with improved mother's practice. The incidence of early initiation of breastfeeding was significantly higher in ISLG facili- ties (staffed by the PHC workers) than in those in the control area. Probably, the health workers in the ISLG area more frequently recommended early ini- tiation of breastfeeding and more often assisted their clients to do so than controls. These results accord with observations in Kenya that improved health worker knowledge and attitudes were associated with substantial improvements in breastfeeding- related hospital practices (10): as a corollary, the breastfeeding promotion campaign removed unde- sirable hospital practices such as separation of mother and baby, and use of formula feeding and of bottles. Exclusive breastfeeding is still very rare in the rural study communities. Previously, based on data from focus group discussions, we predicted that the early initiation of breastfeeding and the feeding of colostrum would be easier to accept in such commu- nities than changing the negative attitudes to ex- clusive breastfeeding (20). Further studies will be needed to demonstrate the impact of the health worker training on the acceptance and length of exclusive breastfeeding in the rural communities where the study was carried out. Nevertheless, our results suggest that the train- ing of extension health workers in rural communi- ties is feasible, can have a beneficial impact on breastfeeding promotion, and that such training should be continued and incorporated into the pri- mary health care system. In Nigeria, more than 75% of mothers live and receive perinatal care in rural communities and depend on the PHC workers for advice. Also, breastfeeding education should be given to all types of health professionals, including doctors and medical students, (21). In the rural study com- munities, the nurse/midwife is the highest trained health professional available; our results show, however, that being a nurse or midwife was as- sociated with low breastfeeding knowledge scores. If breastfeeding education were better integrated into the nursing curriculum, nurses would be better able to provide breastfeeding education in the community. One important factor in the success of the train- ing programme reported here was the administrative support given by the local government authorities, which is crucial for the success of breastfeeding pro- motion efforts (22). Such support was guaranteed in our study since it was implemented as a community linkage programme that was jointly planned and implemented by community representatives, local government, and researchers (23). The training pro- grammes were implemented by the investigators and the PHC workers themselves, and the district level training programmes were performed by the PHC workers who had attended the trainers' workshop. This assured the sustainability of breastfeeding pro- motion and training after the research project ended. As a result, the ISLG Health Department regards the training project as its own and its capacity has been enhanced to carry out such training. A partner- ship design, such as this, which has encouraged full community participation, should also enhance the sustainability of breastfeeding promotion in the rural study communities. In addition, the training and the resource materials developed from it were all in the local language, which is highly appropriate at the rural community level. Long-term, demand breastfeeding is highly prevalent in rural communities in Nigeria (24, 25). With increasing urbanization, it is important to en- sure that breastfeeding continues to be practised and to maximize its nutritional, immunological, and con- traceptive advantages for mother and child health. Health worker training at the community level can therefore be an important component of strategies to achieve these aims. Acknowledgement Financial support for this research was provided by the Applied Diarrheal Disease Research Project, Harvard Uni- versity, through a cooperative agreement with the United States Agency for International Development (USAID). Resume Projet sur I'allaitement a Ife South: formation d'agents de vulgarisation pour encourager et prendre en charge I'allaitement dans les communautes rurales Le pr6sent article rapporte les r6sultats de l'ISBFP (Ife South Breastfeeding Project: projet sur l'allaite- ment a Ife South), destin6 a favoriser de meilleures pratiques d'allaitement dans les communaut6s rurales par la formation d'agents de vulgarisation dans la zone d6pendant des autorit6s locales d'Ife South (Nig6ria). Un atelier a 616 tout d'abord organis6 pour les formateurs (agents de vulgarisa- tion aux niveaux 6l6mentaire et moyen) dans cette WHO Bulletin OMS. Vol 75 1997330 The Ife South Breasffeeding Project region, puis ceux-ci ont a leur tour organise des ateliers de formation dans chacun des dix districts sanitaires que compte cette zone. Aucune forma- tion de ce genre n'a ete organisee dans la zone t6moin. Les agents de sant6 ont estime que le programme de formation 6tait utile et valable et ont recommand6 que des programmes sur l'allaitement soient organis6s regulierement pour 1'ensemble du personnel. En cons6quence, les autorit6s locales ont adopte une politique en faveur des nourrissons pour les installations de soins perinatals primaires. Les discussions au sein de groupes de r6- flexion avant la formation ont montr6 que les agents de soins de sant6 primaires pensaient qu'il fallait commencer l'allaitement dans les 24 heures suivant la naissance mais pas dans les six premieres heures et qu'il etait impossible d'y avoir recours exclusivement (sans apports hydriques comple- mentaires). 11 n'y avait aucune diff6rence entre les deux zones. Au total, 68 des 85 agents de soins de sant6 primaires dans la zone 6tudi6e ont re,u la formation de l'ISBFP. L'6valuation aprbs la formation a montre que le d6but pr6coce de l'allaitement (dans les 30 minutes suivant la naissance) a eu lieu pour 32% des 98 naissances dans la zone etudi6e, contre 6% des 108 naissances survenues dans la zone t6moin (X2 = 21,97 - p < 0,001). Dans les installations de soins p6rinatals primaires ayant des agents de soins de sant6 primaires, le debut pr6coce de l'allaitement a eu lieu pour 47% des 67 naissances dans la zone 6tudiee, contre 4% des 68 naissances survenues dans la zone t6moin (X2 = 29,83 - p < 0,001). Les tests de connaissances sur l'allaite- ment ont donne des r6sultats significativement meilleurs pour les agents de sant6 form6s par l'ISBFP que pour ceux n'ayant pas ete form6s dans la zone t6moin (test H de Kruksal-Wallis = 75,10- p < 0,001) ou dans la zone d'6tude (test H de Kruksal-Wallis = 18,17 - p < 0,001). Beaucoup plus d'agents de soins de sant6 primaires form6s par l'ISBFP que d'agents non form6s ont recom- mand6 le d6but pr6coce de l'allaitement (89% contre 46%; X2 = 26,06 - p < 0,001) et l'allaite- ment exclusif (98% contre 37%; %2 = 51,43 - p < 0,001). L'analyse multivariable a indique que les resultats du test de connaissances 6taient li6s a la zone d'etude (p < 0,001) et au programme de formation (p < 0,001), mais pas a l'age, au sexe, a la situation matrimoniale ni a l'exp6rience de l'agent de soins de sante primaires. Is sont cependant li6s de fa,on marginale a la qualification d'infirmibre ou de sage-femme. Quarante-quatre des 66 agents formes par l'ISBFP ont donne 100% de reponses correctes au questionnaire contre 1 sur 14 de ceux n'ayant pas 6te formes dans la zone d'6tude (X2 = 14,3 - p < 0,001) et aucun des 52 agents de soins de sante primaires dans la zone t6moin (X2 = 52,47 - p < 0,001). Ces r6sultats indiquent que la formation de l'ISBFP a permis d'augmenter les connaissances des agents de soins de sant6 primaires et d'am6- liorer leurs conseils, et qu'il en est r6sulte de meil- leures pratiques d'allaitement maternel dans la zone etudi6e. L'education appropriee des agents de vulgarisation n'est donc pas seulement faisable; elle contribue 6galement de maniere significative a la promotion de l'allaitement dans les com- munaut6s rurales. Ces resultats indiquent aussi que l'allaitement devrait etre int6gr6 dans les pro- grammes de formation de toutes les categories de personnel soignant. References 1. Akre J, ed. Infant feeding: the physiological basis. Bulletin of the World Health Organization, 1989, 67 (suppl): 1-108. 2. Short R. The biological basis for the contraceptive effects of breastfeeding. Intemational journal of gy- naecology and obstetrics, 1987, 25 (suppl): 207-216. 3. De Zoysa I et al. Why promote breastfeeding in diarrhoeal disease control programmes? Health policy and planning, 1991, 6: 371-379. 4. Federal Ministry of Health/WHO/UNICEF/USAID/ CCCD. Nigerian Control of Diarrhoeal Diseases (CDD) Programme 1991-1995. Lagos, Federal Minis- try of Health, 1991. 5. Babaniyi OA. Oral rehydration of children with diar- rhoea in Nigeria: 12-year review of impact on mor- bidity and mortality from diarrhoeal diseases and diarrhoea treatment practices. Journal of tropical pediatrics, 1991, 37: 57-63. 6. Federal Office of Statistics. Nigerian Demographic and Health Survey, 1990. Columbia, MD, IRD/Macro International Inc., 1992. 7. Davies-Adetugbo AA, lrinoye 00. Perceptions of breastfeeding and diarrhoea in rural communities of Osun State, Nigeria. Nigerian joumal of paediatrics, 1994, 21 (suppl): 39-47. 8. WHO/UNICEF. Protecting, promoting and supporting breast-feeding: the special role of maternity services. Intemational journal of gynaecology and obstetrics, 1990, 31 (suppl. 1): 171-173. 9. Bradley JE, Meme J. Breastfeeding promotion in Kenya: changes in health worker knowledge, atti- tudes, and practices, 1982-89. Journal of tropical pediatrics, 1992, 38: 228-235. 10. Armstrong HC. Breastfeeding promotion: training of mid-level and outreach health workers. International journal of gynaecology and obstetrics, 1990, 31 (suppl. 1): 91-103. 11. Breastfeeding management and promotion in a baby- friendly hospital: an 18-hour course for maternity staff New York, UNICEF, 1993. WHO Bulletin OMS. Vol 75 1997 331 A.A. Davies-Adetugbo & H.A. Adebawa 12. Hassard TH. Understanding biostatistics. St Louis, MO, Mosby, 1991: 180. 13. Dean AG et al. Epi Info, version 6: a word processing, database, and statistics program for epidemiology on microcomputers. Atlanta, GA, Centers for Disease Control and Prevention, 1994. 14. Dallal GE. LOGISTIC: a logistic regression software for the IBM PC. American statistician, 1988, 42: 272. 15. Sachdev HPS et al. Water supplementation in exclu- sively breastfed infants during summer in the tropics. Lancet, 1997, 337: 929-933. 16. Badruddin SH, Kendricks K. Weaning and diarrhoeal disease. Journal of diarrhoeal disease re- search, 1994, 12: 4-13. 17. Henry FJ et al. Bacterial contamination of weaning foods and drinking water in rural Bangladesh. Epide- miology and infection, 1990, 104: 79-92. 18. Kurinj N, Shiono PH. Early formula supplementation of breast-feeding. Pediatrics, 1991, 88: 745-750. 19. Wright HJ, Walker PC. Prediction of duration of breast feeding in primiparas. Journal of epidemiology and community health, 1983, 37: 87-94. 20. Davies-Adetugbo AA. Promotion ofbreast-feeding in Ife South local government area. Dissertation. Lagos, West African College of Physicians, 1995. 21. Freed GL. Breastfeeding: time to teach what we preach. Journal of the American Medical Association, 1993, 269: 243-245. 22. Stockamer CL. Breastfeeding promotion efforts: why some do not work. International journal of gynaecol- ogy and obstetrics, 1990, 31 (suppl. 1): 61-65. 23. From health research to community development: an ADDR initiative in Nigeria. Ibadan, Applied Diarrheal Disease Research Project, 1994. 24. Rehan N, Abashiya AK. Breastfeeding and absti- nence among Hausa women. Studies in family plan- ning, 1981, 12: 233-237. 25. Dow T. Breastfeeding and abstinence among the Yoruba. Studies in family planning, 1977, 8: 208-214. 332 WHO Bulletin OMS. Vol 75 1997